s^^^s Marine Biological Laboratory Library Woods Hole, Mass. Presented by Interscience Publishers Ea^E3E3E BACTERIOPHAGES BactcriophageTS, purified preparation, X 93,150. 2) ( (,UIQ BACTERIOPHAGES MARK H. ADAMS WITH CHAPTERS BY E. S. ANDERSON, Central Enteric Reference Laboratory, Central Public Health Laboratory, London J. S. GOTS, University of Pennsylvania, Philadelphia F. JACOB, Pasteur Institute, Paris E.-L. WOLLMAN, Pasteur Institute, Paris ELECTRON MICROGRAPHS BY E. KELLENBERGER, The University, Geneva 1959 INTERSCIENCE PUBLISHERS, INC., NEW YORK Interscience Publishers Ltd., London Copyright © 1959 by INTERSCIENCE PUBLISHERS, INC. Library of Congress Catalog Card Number 58-1 2722 INTERSCIENCE PUBLISHERS, INC. 250 Fifth Avenue, New York 1, N. Y. For Great Britain and Northern Ireland: INTERSCIENCE PUBLISHERS, LTD. 88-90 Chancery Lane, London, W.C. 2 PRINTED IN THE UNITED STATES OF AMERICA BY MACK PRINTING COMPANY, EASTON, PA. PREFACE Phage research, after a fitful history during its first twenty years, had all but died out in the middle 1930's. In the text- books of bacteriology, the bacteriophages, if they were men- tioned at all, figured as a curiosity item, unconnected with the rest and disposed of in a couple of pages at most. Today, phage research is vigorously pursued in many outstanding laboratories in this country, in France, and to a lesser extent in other coun- tries. There are perhaps 100 people directly engaged in basic research. In addition, the field has become well known to many scientists in other fields: genetics, biochemistry, virology, im- munology, to name the most important. Indeed, there is every reason for the popularity of phage research, and for the interest in its results on the part of other scientists. Phage research has become so interwoven with these other fields that it is quite difficult to extricate it from its entangling alliances, and to pre- sent it as a unit. The author of this book, Mark Adams, has a large share in recent developments, by his own research, by his teaching at New York University, and by two special contributions: the phage course at the Biological Laboratory in Cold Spring Har- bor, and the review on the methods in phage research first pub- lished in Methods in Medical Research, Vol. 2, 1950, and now re- printed in revised form as an appendix to the present volume. The phage course in Cold Spring Harbor was instituted in 1945. Mark Adams took it the second year, and taught it since then every summer, except two. In this course were trained many of those who are presently engaged in phage research, and in addition many who are interested in related fields acquired through it a critical understanding of the phage literature. It thus served to bring phage research out of its isolation, and to foster the many links to other parts of modern biology. The re- view on Methods is a classic, and easily the paper most often VI PREFACE referred to in the current phage Uterature. The sentence "the methods employed in this work are those described by Adams, etc.," is almost like a ritual invocation used by every phage worker, with a sigh of relief, relieving him of the necessity of an otherwise burdensome chore. Phage research has also been discussed in, and its knowledge disseminated through, numerous, perhaps too numerous, review articles, but there has been no book covering the whole field since 1926, long before the modern era of this field. The fact is that the literature on this subject has grown so complex as to deter everybody, except Mark Adams. I do not believe, in fact, that anybody, besides Adams, had the qualifications for this task. Adams started writing this book several years ago, and even he, in spite of his tremendous knowledge, critical ability, and superb expository gifts, found the going very hard. On October 17, 1956, Mark Adams died, of an acute infection, at the age of 44. At that time about three-quarters of the chapters of this book were in a semi-finished state, the rest not begun. Some of the chapters had been finished recently, and were up to date. Some had been written several years ago and needed additions and revision. All of them needed some editing. At the request of Mrs. Adams, a committee consisting of A. D. Hershey, R. D. Hotchkiss, A. M. Pappenheimer, Jr., and E. Racker, in consultation with the publisher, went over the manuscript and decided that several specialists should be invited to write the missing chapters, while Hershey volunteered to edit the others. After proceeding with this plan for some time, it became clear that the editorial job was greater than anticipated, and could not be completed by one man within a reasonable length of time. Accordingly, an inquiry was sent around asking for volunteers for the editing of individual chapters, the over-all editorial responsibility to remain in the hands of Hershey. This inquiry met with a wide response, and the plan was promptly executed. Accordingly we now have before us, for the first time in thirty years, a book on bacteriophages, reasonably complete, reason- PREFACE Vll ably up to date, and reasonably elementary; useful, we trust, to every student of modern biology in the widest sense. We also have before us a monument to Mark Adams, our unforgettable friend. Pasadena, California M. Delbruck May, 7958 EDITORS' PREFACE At the time of his death on October 17, 1956, Mark Adams had written all but four chapters of the present book. Messrs. Anderson, Gots, Jacob, and Wollman contributed the remainder, completing the text according to Adams' plan. The original chapters were written over a period of several years. They had to be revised, completed, and brought up to date, with a minimum of editorial changes. The necessary work was done by S. Benzer, G. Bertani, M. Delbriick, A. Garen, W. Harm, R. M. Herriott, A. D. Hershey, F. Lanni, R. G. E. Mur- ray, F. W. Stahl, G. S. Stent, G. Streisinger, J. D. Watson, and J. J. Weigle. Hershey is responsible for the form in which these chapters now appear. E. Kellenberger and his colleagues furnished many electron micrographs, and assisted in the selection and presentation of those appearing in this book. Unless otherwise stated, specimens were prepared by Kellenberger's agar filtration method (see Chapter XI), shadowed with uranium oxide, and translated to film in the RCA-EMU 2 microscope. Appreciation is due the Year Book Publishers, Inc., Chicago, Illinois, for permission to reprint, as an appendix to this volume, "Methods of Study of Bacterial Viruses," from Vol. 2 of Methods of Medical Research. The appendix was edited by Rollin D. Hotchkiss and Nancy Collins Bruce. Carnegie Institution of Washington A. D. Hershey Cold Spring Harbor, New York for the editors June, 7958 CONTENTS I Introduction 1 1 . Definition 1 2. Discovery of bacteriophages 2 3. Nature of bacteriophages 4 4. Origin of bacteriophages 6 5. Lysogeny 8 6. Classification of phages 11 II The Infective Process 13 1. The one-step growth experiment 14 2. Adsorption 15 3. Penetration 17 4. Multiphcation 18 5. Lysis of host cell and release of phage progeny. . 20 6. Lysis in fluid media 21 7. Lysis on soHd media 22 III Enumeration of Bacteriophage Particles 27 1 . Plaque counts 27 2. Dilution end-points 30 3. Measurement of lysis-time 31 4. Intracellular phage 32 IV Size and Morphology of Bacteriophages 35 1. Electron microscopy of bacteriophage particles. . 35 2. Ultrafiltration 41 3. Analytical centrifugation 42 4. Diff'usion constants 43 5. Electrophoresis 44 6. Ionizing radiations 44 7. Nonionizing radiations 46 8. Weight of the infectious particle 46 9. Summary 47 V Effects of Physical and Chemical Agents (Except Radiations) on Bacteriophage Particles 49 1. EflTectof pH 50 2. Urea and methane 50 3. Detergents 51 IX 75134 CONTENTS 4. Chelating agents 52 5. Mustard gas 52 6. Alcohols 53 7. Other chemicals 53 8. Sonic vibration 54 9. Surface denaturation 55 10. Heat inactivation 56 11. Hydrostatic pressure 60 12. Osmotic shock 60 13. Summary 61 VI Effects of Radiations on Phage Particles 63 1. Visible light 63 2. Ultraviolet light 65 a. Lethal effect 66 b. Growth-delaying effect 68 c. Host killing 69 d. Interference 70 e. Photoreactivation 70 f Multiplicity reactivation 73 3. Ionizing radiations 74 4. Decay of radiophosphorus 80 5. Summary 82 VII Chemical Composition 83 1. Concenti-ation and purification 83 2. Criteria of purity 85 3. Elementary composition 87 4. Gross chemical components 88 5. Amino acids, nucleotides, and bases 89 6. Extinction coefficients of phage preparations. ... 92 7. Phosphorus and nitrogen contents per infective particle 94 8. Summary 95 VIII Antigenic Properties 97 1 . Preparation of antiphage sera 98 2. Multiplicity of phage antigens 100 3. Neutralization of phage infectivity 101 4. Properties of incompletely neutralized phage preparations 102 5. Complement fixation 104 CONTENTS XI 6. Specific aggregation 105 7. Agglutination of phage-coated bacteria 106 8. Specific soluble substances 107 9. Kinetics of neutralization 109 10. Anomalous neutralization behavior 113 1 1 . Host dependence of serum-survivor assay 114 12. Inheritance of antigenic specificity 116 13. Mechanism of phage neutralization 117 IX Host Specificity 121 1. Microbial hosts for phages 121 2. Host ranges of bacteriophages 122 3. Host specificity in adsorption 123 4. Host specificity in infection 124 5. Acquisition of phage resistance by mutation 125 6. Acquisition of phage resistance by lysogenization . 128 7. Phage mutations affecting host range 129 8. Phenotypic modification of host range 132 a. Phenotypic mixing 132 b. Host-controlled variation 133 9. Summary 135 X Adsorption of Phage to Host Cell 137 1. Kinetics of adsorption 138 2. The ionic environment 141 3. Organic cofactors 143 4. The stepwise nature of adsorption 147 5. Chemical nature of receptor sites 150 6. Summary 159 XI Stages in Phage Multiplication 161 1. Initiation of infection 161 a. Penetration 161 b. Release of DNA from the phage particle. . . 163 c. Puncture of the bacterium 164 2. The latent period 169 3. Burst size 170 4. Premature lysis 172 5. Phage multiplication studied by irradiation of in- fected bacteria 175 6. Morphological stages in phage development 182 7. Conclusion 187 11 CONTENTS XII Cytological Changes in the Infected Host Cell 189 1. Observations on living cells 190 2. Observations on fixed and stained preparations. . 195 3. Electron microscope observations 202 4. Summary 205 XIII Isotopic Studies on the Fate of the Infecting Phage Particles 207 1. Morphological, functional, and chemical dif- ferentiation of the phage particle 208 a. Chemical composition 208 b. Physical fractionation — osmotic shock 208 c. Physiological fractionation- — ^the blendor ex- periment 209 d. Functional differentiation of phage protein and phage DNA 209 2. The breakdown of the infecting virus particle. . . 210 a. Absence of breakdown of phage proteins. . . 210 b. Breakdown of phage nucleic acid following superinfection 212 3. Material transfer from infecting phage to progeny 216 a. Lack of transfer of phage protein to progeny 216 b. Transfer of phage nucleic acid to progeny. . 218 c. Material transfer without genetic transfer. . 222 d. Physical state of parental isotope during transfer 226 e. Distribution of parental isotope among prog- eny particles 227 f. Efficiency of transfer 229 4. Summary 229 XIV Nutritional and Metabolic Requirements for Phage Production 233 1, Requirements for adsorption and penetration. . . . 234 2. Metabolic requirements for phage multiplication. 236 a. Lack of metabolic enzymes in mature phage 237 b. Enzymic activity associated with vegeta- tive phage 240 c. Enzymic activity of the host cell 242 d. Synthetic and energetic machinery of the host cell 244 CONTENTS Xlll 3. Partial metabolic requirements for vegetative replication 247 4. Sources of material for phage synthesis 248 a. Materials derived from the parental phage . 249 b. Protein materials assimilated before and after infection 249 c. Nucleic acid materials assimilated before and after infection 254 5. Kinetic studies of DNA metabolism 257 6. Summary 262 XV Chemical Interference with Phage Growth J.S.GoTS 265 1 . Prevention of adsorption 267 2. Prevention of penetration 270 3. Inhibition at the intracellular level 271 a. Interference with energy metabolism 272 b. Prevention of protein synthesis 273 c. Prevention of DNA synthesis as a conse- quence of protein deficiency 275 d. Prevention of DNA synthesis 276 e. Prevention of coenzyme activity 276 f. Prevention of maturation 278 4. Abortive infection 282 5. Noninfective phage particles containing structural analogues 283 6. Miscellaneous inhibitors 284 a. Antibiotics 284 b. Aromatic diamidines 285 c. Dyestuffs 286 7. Summary 286 XVI Mutation and Phenotypic Variation in Phages 289 1 . Definitions 290 2. Nonhereditary (phenotypic) variation 291 a. Host-induced modifications 291 b. Phenocopies 294 c. Phenotypic mixing 294 d. Other phenotypic modifications 296 3. Hereditary variation (mutation) 297 a. Mutations affecting specificity of adsorption 298 b. Mutations afi"ecting growth potential 299 XIV CONTENTS c. Mutations affecting requirements for ad- sorption 301 d. Mutations affecting lysis-inhibition 302 e. Mutations affecting plaque type 304 f. Mutations affecting stability 305 g. Mutations affecting antigenic specificity. . . , 305 h. Other mutations 306 4. Mutation frequency 308 5. Mutagenic agents 310 6. Summary 317 XVII Mixed Infection 319 1. Mixed infection with unrelated phages — mutual exclusion 320 2. Mixed infection with related phages 325 a. Absence of mutual exclusion 325 b. Partial exclusion 327 c. Limited participation 329 3. Summary 329 XVIII Bacteriophage Genetics 331 1. Mixed infection and recombination 331 2. The vegetative phage pool 333 3. The Visconti-Delbriick theory 335 a. Many genetic loci are available 335 b. Complementary recombinants are formed . . 335 c. Multiple recombinations 336 d. Linkage and negative interference 337 e. Multiplication after recombination 338 f. Drift toward genetic equilibrium 340 g. A problem in population genetics 342 4. Heterozygosis 344 5. Pseudoallelism 351 a. The gene as a functional unit 352 b. The mutational unit 354 c. The unit of recombination 356 6. Radiation genetics 357 a. Multiplicity reactivation of phage inacti- vated by ultraviolet light 357 b. Multiplicity reactivation of phage inacti- vated by ionizing radiations or by P^" decay 362 CONTENTS XV c. Cross reactivation 362 d. Effects of ultraviolet light on genetic recom- bination 363 XIX Lysogeny F. Jacob and E.-L. Wollman 365 1. Definition and occurrence 365 2. Prophage 366 3. Properties of temperate phages 369 4. Lysogenization 370 5. Production of phage by lysogenic bacteria 371 a. Spontaneous production 371 b. Induction 372 6. Immunity 374 7. Defective lysogenic bacteria 375 8. Lysogeny and bacterial genetics 376 a. Transduction 377 b. Changes in the bacterium resulting from lysogenization 378 9. Conclusion 379 XX Colicins and Other Bacteriocins F. Jacob and E.-L. Wollman 381 1 . Definition and criteria 381 a. Definition 381 b. Detection and titration 382 c. Physicochemical properties 384 2. Action of colicins on sensitive bacteria 385 a. Specificity 385 b. Mode of action 386 3. Colicinogenic bacteria 387 a. Production of colicin 388 b. Genetic determination of colicinogeny 389 4. Other bacteriocins 390 5. Bacteriocins and bacteriophages 392 XXI Use of Phages in Epidemiological Studies E. S. Anderson 395 1 . Historical 397 2. The Vi-phage typing scheme of Craigie and Yen. . 399 3. Technical considerations 402 a. Isolation of phages for typing purposes 402 b. The routine test dilution 403 c. Technique 404 XVI CONTENTS 4. Theoretical aspects of Vi-phage typing 405 a. The adaptation of Vi-phage II 405 b. Vi-type specificity in Salmonella typhi .... 408 5. The practical application of Vi-phage typing. ... 412 6. Phage- typing of Salmo?iella paratyphi B and S. typhimurium 413 7. Phage typing of other salmonellas 416 8. Phage typing oi Staphylococcus aureus 416 9. Phage typing of Corynebacterium diphtheriae 418 10. Bacterial typing by identification of carried phages 419 XXII Phage Taxonomy 421 1. Purpose and principles of taxonomy 421 2. Various proposals for phage classification 422 3. Taxonomic criteria at the species level 424 a. Serological relationship 424 b. Size and morphology 426 c. Chemical composition 426 d. Latent period 426 e. Susceptibility to inactivation 427 f. Distinctive physiological properties 428 g. Results of mixed infection experiments 429 4. Possible taxonomic criteria at levels above the species 431 5. Special taxonomic criteria applicable to temper- ate bacteriophages 432 6. Relationship of bacteriophages to other viruses. . 433 7. Importance of type specimens 435 8. Summary 437 Glossary 439 Appendix: Methods of Study of Bacterial Viruses 443 Introduction 444 Host organisms and viruses 445 Media 445 Isolation of bacterial viruses 447 Assay of host organisms 449 Assay of phage by agar layer method 450 Efficiency of plating 451 Stability of viruses and selection of diluents. . . 454 Preparation of high titer phage stocks 454 Filtration of phage stocks 456 CONTENTS XV 11 Concentration and purification of phage 457 Preparation and use of antiphage sera 461 Immunization cf animals 461 Assay of antiphage sera 463 Use of K value of antiserum to demonstrate serologic relatedness among phages 465 Rate of adsorption of virus to bacterium 466 1 . Assay of unadsorbed phage 467 2. Determination of number of infected bac- teria 469 3. Determination of proportion of surviving bacteria ^'^^ The single-step growth curve 473 Discussion ^'^ Lysis 481 Adsorption of phage without phage develop- ment ••• 483 Investigations of chemical action on virus growth 4o4 Single-cell method of studying virus multiplication ... 485 Host cell mutation to resistance to virus attack 490 Isolation of phage-resistant mutants 490 Mutational pattern of strain B of £". coli 492 Indicator strains and their properties 494 Plating of mixed viruses with mixed indicator 495 strains ^^-' Mutations of bacteriophages 501 1. Host range mutations 501 2. Adsorption cofactor mutations 503 3. Plaque morphology mutations 505 Mixed infection and genetic recombination 508 Mixed infections with r+ and r variants of same phage type 508 Recombination of genetic characters in mix- edly infected bacteria 509 Effects of ultraviolet irradiation of bacteriophage 511 Photoreactivation 513 Multiplicity reaction of ultraviolet-inactivated phages 514 XVlll CONTENTS Methods of studying intracellular growth of bacterio- phage 516 Biochemical methods 516 Radiation methods 518 Methods involving interruption of growth during the latent period followed by dis- ruption of host cells 519 Bibliography 523 Index 567 CHAPTER I INTRODUCTION 1. Definition The name bacteriophage was given by F. d'Herelle to a bacteriolytic substance that he isolated from feces. Now usually shortened to phage, it means "eater of bacteria" and refers to the remarkable ability of bacteriophages to bring about lysis of growing bacterial cultures. More remarkable still was d'Herelle's evidence that the lysis is accompanied by produc- tion of more phage ; the lytic agent is transmissible in series from culture to culture of susceptible bacteria. Today phages are universally recognized to form a group of bacteria-specific viruses, that is, ultramicrobes of diverse char- acter exhibiting all the signs of a long history of manifold varia- tion, adaptation, and specialization. D'Herelle himself sub- scribed to this view only in part: he considered all bacterio- phages to belong to a single variable species. In so doing he, and many of the opponents of the virus theory, ignored or de- nied the strongest evidence for it, with the unfortunate result that some of the ostensibly fundamental research on phages was for a decade or more directed along fruitless channels. The lytic cycle recognized by d'Herelle was such a striking- aspect of bacteriophage activity that other aspects were largely ignored. However, a nonlytic phase of reproduction has been rediscovered recently, causing phage research to branch out in new directions. Certain strains of bacteriophage when infect- ing a susceptible bacterial culture are able to enter into an in- timate symbiotic relationship in which the host cell continues to multiply, carrying the virus intracellularly in a noninfective condition for an indefinite number of cell divisions. This I BACTERIOPHAGES delicately balanced phase of viral growth in which the infected host cell and its carried "prophage" multiply at the same rate has been termed "lysogenesis" or "lysogeny." Infection leading to lysogeny is now thought to produce a modification of the genetic apparatus of the bacterial cell and often results in changes in bacterial properties, a striking example being the conversion of avirulent strains of diphtheria bacilli to toxigenic potency by in- fection with an appropriate phage. Under certain conditions some bacteriophages can attack and kill susceptible bacteria with no evidence of bacterial lysis or of phage multiplication. In these circumstances the phage par- ticle behaves as an antibiotic rather than as a virus. In fact certain antibiotics produced by bacteria, the colicins and pyo- cins, are rather similar to bacteriophages in certain of their properties. For these reasons the colicins as well as phages will be discussed in this book. 2. Discovery of Bacteriophages There is no doubt that numerous early bacterio ogists saw and described signs of phage action in bacterial cultures. How- ever, no intensive investigation of these phenomena was under- taken prior to the appearance of a brief but provocative paper by F. W. Twort (1915). This British bacteriologist described an acute infectious disease of staphylococci that produced marked changes in colonial morphology. The infective agent was filterable and could be passed indefinitely in series from colony to colony. Twort considered various hypotheses to explain this phenomenon; among others that it was a filterable virus analo- gous to the virus pathogens of animals and plants. Twort's remarkable paper contained in essence the present concept of the nature of bacteriophage, yet the paper remained unnoticed by scientists and Twort failed to pursue the matter further, perhaps because of his wartime duties in the British Army. In 1917, Felix d'Herelle, a Canadian bacteriologist working at the Pasteur Institute in Paris, published his independent dis- covery of "bacteriophage." In a noteworthy series of papers he INTRODUCTION J reported and correcdy interpreted many of the significant as- pects of bacteriophage action, and in 1921 published his classic book entitled Le bacteriophage: son role dans rimmiinite. D'Herelle, stimulated by the announcement that hog cholera was due to the synergistic action of a bacterium and a virus, was searching for evidence of such a mixed etiology for bacillary dysentery in man. He prepared Chamberland filtrates of stools of dysentery patients and added these filtrates to young cultures of Shiga's bacillus. He incubated the mixtures overnight intend- ing to inject the supposed mixture of bacterium and virus into experimental animals next day. To his surprise some of the culture fluids were clear and sterile. He filtered some of the lysed cultures and inoculated fresh cultures of Shiga's bacillus with the filtrates, keeping other cultures of the bacterium as un- inoculated controls. On incubation overnight he found the con- trol cultures to be turbid as expected while the bacterial cultures inoculated with the filtrates were completely clear. The "lytic principle" could be passed indefinitely from culture to culture using each time for phage inoculum a bacteria-free filtrate of the previous culture. This behavior suggested to him an ultravirus, pathogenic for bacteria, destroying its host cells as it multiplied. During the next few years bacteriophages were found for a variety of pathogenic bacteria such as staphylococci, cholera vibrio, and typhoid bacteria. Because of the susceptibility of pathogenic bacteria to phages, and because of the wide distribu- tion of phages in nature, d'Herelle suspected that they played a role in resistance to and recovery from disease. D'Herelle shares with Twort credit for the discovery of phages and did much of the basic research concerning them. In addi- tion his ideas relating phages to immunity to disease attracted the general attention of medical scientists, indirectly yielding both practical and theoretical benefits. The importance of the role of bacteriophages in natural control of infectious disease has not been assessed to this day, however. A vast amount of research was devoted to possible therapeutic ^applications of bacteriophages between 1920 and 1940. In 4 BACTERIOPHAGES most cases results were equivocal or disappointing but in certain diseases such as cholera there was some evidence for a favorable effect of phage therapy. Interest in this application has largely subsided since the introduction of the more convenient chemo- therapeutic agents. The early hopes of effective medical use stimulated much valuable work on host range, immunological properties, stability and variation, and other attributes of bac- teriophages. Many of the well-known names in bacteriology and immunology are found in the phage literature of this period: J. Bordet, Costa Cruz, Doerr, A. Fleming, Hauduroy, Levaditi, Prausnitz, and Topley. Also during this period a few men made phage research their life work. Included in this group besides d'Herelle are Asheshov, Bronfenbrenner, Flu, Gratia, and the elder Wollmans. These scientists were inter- ested in the biology of bacteriophages as well as in possible ap- plications and made many contributions of fundamental impor- tance. During the second decade of phage history appear a few men whose work combined biological "feeling" and quantitative methods in the modern manner: Burnet, M. Schlesinger, and C. H. Andrewes. The contributions of these older phage workers and of a very large contemporary school form the subject matter of this book. 3. Nature of Bacteriophages In his original publication, Twort (1915) considered the na- ture of his as yet unnamed lytic agent, asking if it was similar to bacteria, protozoa, or to the filterable viruses; whether it was a filterable stage in the life cycle of the micrococcus affected ; or whether perhaps it was a bacterial enzyme, produced autocatalyt- ically while destroying the bacterium that produced it. A much more elaborate series of hypotheses is discussed in d'Herelle's book (1926). Of these hypotheses two only have survived to the present: the "precursor" theory and the "virus" theory. An active controversy between the proponents of these two theories continued for many years and stimulated much experimental work as well as verbiage. INTRODUCTION i> According to the precursor theory bacteriophages are en- dogenous, existing in bacteria as precursors which either sponta- neously or after stimulation are transformed into characteristic lytic substances, much as trypsinogen can be converted into trypsin. This theory was supported by Gildemeister (1921), Bordet (1925), Northrop (1939a), Kreuger and Scribner (1939), and Felix (1953). Much of the experimental support for it was derived from studies on lysogenic bacteria, to be discussed pres- ently. According to the virus theory the bacteriophages are autono- mous microbes analogous to plant and animal viruses but obli- gately parasitic on bacteria. This hypothesis was adopted very early in his work by d'Herelle and seemed obvious to most virologists once something had been learned about the biology of viruses in general. In the light of current knowledge this theory amounts to very little, however, since both autonomy and parasitism as applied to viruses are terms difficult to define. Some current views are presented below in connection with theories concerning the origins of viruses. As is often the case in the historical development of a scientific field each of the rival hypotheses can now be construed as part of the truth; each was incomplete by itself. The current con- cept of the nature of bacteriophage is a synthesis of the old ideas plus a few novel ideas developed recently. Somewhat as a typical bacillus may exist in the spore state or the vegetative state, a typical bacteriophage may exist in three states, called prophage, vegetative phage, and mature phage. Phages exist outside of the host cell in mature form, metabolically inert and resembling very crudely the spore state of bacteria. Following adsorption to the host cell, part of the phage particle penetrates within and may begin to multiply. When it does, the multiply- ing, intracellular state of the phage particle diff'ers in many re- spects from mature phage and has been termed vegetative phage in recognition of its almost unlimited reproductive capacity. Some phages, called temperate, are distinguished by the fact t^at they can exist in a third state, prophage, as well. When a 6 BACTERIOPHAGES mature temperate phage particle infects a susceptible bac- terial cell it may be transformed either to the vegetative state leading to destruction of the host cell, or to the prophage state in which it enters into the hereditary symbiotic relationship with the host cell known as lysogeny. This relationship has been termed symbiotic rather than parasitic because it may be readily sur- mised that the lysogenic condition is of great survival value to both phage and host cell. The recognition that temperate phages could exist in three distinct states served to unify the precursor and virus theories. The endogenous precursor in lysogenic bacteria is simply the phage-specific prophage state of temperate phages. This synthesis of the opposing concepts was clearly foreseen by Burnet and McKie (1929). The status of precursor theories as applied to infection by exogenous phages is still unclarified, however. Recent research has demonstrated a marked similarity in physical, chemical, and biological properties between bacterio- phages and other viruses. The similarity has stimulated much research in which bacteriophages have been used as models for animal viruses, even in extensive screening programs for anti- biotics against viruses. To a limited extent this procedure has already justified itself, but it is doubtful whether the homology can be pursued very far. It is possible that bacteriophages and other viruses have evolved from quite distinct ancestors, as the following paragraphs suggest. The facts that the known bac- teriophages contain deoxyribonucleic acid whereas several typi- cal plant and animal viruses contain ribonucleic acid may prove more significant than the rather casual resemblances demon- strated so far. 4. Origin of Bacteriophages Theories of the origin of bacteriophages have been closely associated with theories of their nature. Many diverse ideas have been proposed, most of which fall into one of two cate- gories, the "creation" hypothesis or the "degeneration" hypothe- sis. The creation hypothesis considers that viruses are the direct INTRODUCTION / descendants of some very primitive form of life that originated prior to the evolution of organized cells (Oparin, 1938). This hypothesis is not very useful or satisfying, but it could be true for some or all phages, or for soine or all viruses. The degeneration hypothesis considers that viruses have gradually evolved from more complicated forms of life by the loss of all protoplasm un- necessary to the peculiar mode of existence bacteriophages have chosen. This hypothesis has a certain intellectual appeal in that it is easy to visualize intermediate steps in the degenerative proc- ess, and progressive loss by bacteria of synthetic abilities is sub- ject to experimental study. This hypothesis avoids the problem of the ultimate origin of life, which is not necessarily related to the problem of the origin of viruses, and it provides for the in- dependent evolution of different viruses. By this hypothesis there is no need to assume that animal viruses and bacterio- phages are derived from a common progenitor. Recent studies of the phenomenon of lysogeny have lent credence to a variant of the degeneration hypothesis that maV be called the "mutation" hypothesis (Lwoff, 1953). According to this hypothesis a series of mutations occurring in a segment of the genetic material of a bacterium has conferred on that segment a certain degree of autonomy. The mutations resulted in the synthesis of a specific kind of protein covering to protect the seg- ment of genetic material from damage by the extracellular en- vironment and to facilitate the transfer of the genetic material from one cell to another. It is not impossible that bacterio- phages may have evolved from a primitive mechanism of sexuality originally developed for the purpose of transfer of ge- netic materials between bacterial cells. At least certain bacterio- phages are able to perform that function at present. According to this hypothesis the phage genetic material would be a modi- fied form of the bacterial genetic material, yet sufficiently like the homologous segment of bacterial gene stuff to undergo ge- netic recombination with it or even to replace it. This hypothesis is consistent with the observed fact that the genetic material of certain temperate phages can associate more or less perma- « BACTERIOPHAGES nently with definite genetic loci in the hereditary apparatus of the host cell, and thus become part of that apparatus. Once temperate phages have evolved they could be irreversibly transformed into virulent phages by further mutations. It is suspected that certain virulent phages have retained the ability to undergo genetic recombination with host cells. Host range mutations of a virulent phage might permit it to attack hosts with which it had no genetic affinity whatever and so ultimately to give rise to hypervirulent phages that have lost all resemblance to any known bacterial strains. According to this hypothesis different phage strains may have completely independent phylog- enies and a given phage may be much more closely related to its host cell phylogenetically than to other phages. It is clear that if this hypothesis has any validity there may be no phylo- genetic relationship between bacteriophages and any other type of virus. The resemblances of phages to plant and animal viruses may be superficial and due to their filling of similar ecological niches. Referring to bacteriophages as bacterial viruses may convey some information about their way of life but it should carry no implications about their origins. 5. Lysogeny From what has been said above it is clear that the phenomenon of lysogeny has played a central role in the formulation of ideas about bacteriophages. Because of its significance to all aspects of phage research today, lysogeny must be discussed early in this book. Literally, the term means production of lysis and refers to the habit of bacterial strains that characteristically produce a bacteriophage capable of lysing other bacterial strains. Unfortunately the term has been applied to two en- tirely distinct phenomena with consequent confusion in thinking and definition. One of these phenomena results from the contamination of a partially susceptible bacterial strain with a bacteriophage. In such cases it is sometimes technically difficult to rid the bacteria INTRODUCTION V of the phage. D'Herelle (1930) referred to this phenomenon as "symbiose bacterie-bacteriophage," a name that reflects his interpretation of lysogeny. He assumed that the bacterial population was composed of individuals covering a considerable spectrum of susceptibility to phage action, while the phage population covered a considerable range of virulence. The most virulent phage particles would multiply at the expense of the most susceptible bacteria and in this way bacteria and phage particles would be perpetuated in mixed culture. D'Herelle re- fused to recognize any other explanation for lysogenic strains of bacteria. An essentially similar idea was proposed by Del- bruck (1946b) for a phenomenon he called "pseudolysogenesis," in which the contaminating phage reproduced at the expense of phage-susceptible bacteria produced by mutation during growth of a phage-resistant culture. Pseudolysogenic bacterial strains have been termed "carrier strains" by Lwoff (1953). In contrast to pseudolysogenic or carrier strains are those bacterial strains that exhibit true lysogeny. In such strains the prophage multiplies as an integral part of the genetic apparatus of each bacterial cell. During bacterial growth the conversion of prophage to vegetative phage occurs spontaneously with a small but characteristic frequency, for example in one cell per million cell generations, as evidenced by the lysis of an occasional cell to liberate infective phage particles. In some but not all lysogenic cultures the natural frequency of productive lysis may be greatly increased by exposure to ultraviolet light, X-rays, nitrogen mustard, peroxides, and certain other agents. The biochemical mechanism of such "induction" is unknown. It has been repeatedly demonstrated that each bacterial cell in a lysogenic culture is capable of transmitting the potentiality of phage production to its progeny in the absence of contact with exogenous phage. With inducible strains of bacteria, for in- stance, it is possible with suitable doses of ultraviolet light to cause at least 90 per cent of the cells to lyse and liberate viable phage particles (LwofF, Siminovitch, and Kjeldgaard, 1950). In the classic experiment of den Dooren de Jong (1936), spores 1 BACTERIOPHAGES of a lysogenic strain of Bacillus megalerium were heated above lethal temperatures for free phage particles without killing the spores. Each spore on germination gave rise to a lysogenic cul- ture. Similar conclusions had been reached from studies on a lysogenic strain of Salmonella enteritidis by Burnet and McKie (1929) who wrote that "the permanence of the lysogenic char- acter makes it necessary to assume the presence of bacteriophage or its anlage in every cell of the culture, i.e., it is a part of the hereditary constitution of the strain." For many years the manner of release of phage particles from lysogenic bacterial cells was a subject of dispute. One popular theory proposed that the phage was secreted by the growing bacterial culture in the manner of extracellular bacterial enzymes (Northrop, 1939b). An alternative theory held that the oc- casional disintegration of a bacterial cell liberated a number of phage particles in a way that was already familiar to students of the virulent phages (Burnet, 1929a). This question was finally settled by the work of Lwoff and Gutmann (1950) who demon- strated by the use of micromanipulator techniques that phage particles were not secreted by multiplying lysogenic bacteria, but appeared in groups simultaneously with the disintegration of single bacterial cells. The recent revival of interest in lysogeny begins with impor- tant reinvestigations of the phenomenon by Lwoff and Gut- mann (1950), Bertani (1951), and the Lederbergs (1953). These papers are all the more interesting because they record three quite different experimental approaches. Lwoff (1953) lists the characteristic signs of lysogeny as fol- lows: 7. In a lysogenic culture lysogenesis is a property of every cell and every spore. 2. Bacteria of a lysogenic culture generally can adsorb the mature phage produced by the culture, but are not damaged by it. 3. Lysis of lysogenic bacteria by enzymes, by other phages, or by mechanical means does not liberate mature phage particles. INTRODUCTION 1 1 The intracellular phage in lysogenic bacteria is noninfectious; it is prophage. 4. Infection of a susceptible bacterial culture by a temperate phage may result in the conversion of a considerable proportion of the bacterial cells to the lysogenic condition, potentially ca- pable of liberating the same kind of phage that was used to infect them. 5. Lysogenic bacteria can multiply without liberating mature phage and can undergo many cell divisions in the absence of ex- ternal phage without losing the lysogenic propensity. 6. Lysis of single lysogenic bacterial cells spontaneously or after a characteristic latent period following induction is ac- companied by the release of many mature phage particles. Lysogeny is potentially lethal to the bacterial cell. One may define lysogenic bacteria as bacteria in which the capacity to produce bacteriophage is a hereditary property perpetuated without intervention of mature phage particles. Lysogenic bacteria are very common in nature and probably constitute the principal reservoir of bacteriophages. Lysogeny resembles in some respects the phenomenon of latency in ani- mal and plant viruses, but work in the latter fields has not pro- gressed to the point that permits one to judge the validity of the analogy. The phenomenon of lysogenesis is intimately related to the problems of bacterial genetics and has already served as a powerful tool in the study of genetic structure and function in certain genera of bacteria. These and other aspects of lysogeny will be discussed in detail in later chapters. 6. Classification of Phages Phages are conveniently described as typhoid phages, staphy- lococcal phages, or coliphages, meaning phages attacking the in- dicated types of bacteria. In addition, phages are known by individual designations, such as lambda, Tl, or P8, which serve to identify the particular phage. These specific symbols usually have no meaning in themselves, except to indicate that the phage eame from a particular collection. The coliphages Tl, T2, 12 BACTERIOPHAGES . . . T7, for instance, were collected by Demerec and Fano (1945) and have in common only the ability to infect a strain of Escheri- chia coli known as B. In this instance, but in relatively few others, the phages have been classified by other means. They fall into four groups of "related" phages: Tl ; T2, T4, T6; T5; T3, T7. In this instance the phages within a single group are morphologically identical, and serologically related but not identical. In the following pages we shall frequently speak of re- lated phages, meaning two or more phages having a common host, common morphology, and usually also some common anti- gens. A given "wild-type" phage and its mutants are always related in this sense, as far as is known. In much of the earlier work on phages unclassified specimens, often no longer available, were used. These can only be described as "a staphylococcal phage," or "a coli-dysentery phage." When a phage is re- ferred to by a symbol without indication of host specificity, it is likely to be one of the better known coli-dysentery phages. These remarks will serve to indicate the meaning, or lack of it, of names of phages referred to in this book. General problems of taxonomy are discussed in the final chapter. CHAPTER II THE INFECTIVE PROCESS As d'Herelle found, the inoculation of a growing culture of susceptible bacteria with an appropriate bacteriophage results in a few hours in lysis of the bacterial cells and an increase in the amount of bacteriophage far above that added. D'Herelle conceived of this phenomenon as an infection of the bacterial host cell by a virulent virus particle which as a result of its multiplication brought about death and dissolution of the host cell. He noted that in order for lysis to occur the host cells must be growing in an appropriate physical and chemical en- vironment. It is not difficult to find environmental conditions of temperature, pH, salt concentration, or chemical composition in which the host cells will multiply but lysis will not occur. For convenience in discussion d'Herelle divided the infective process into arbitrary stages: (7) adsorption of the phage par- ticle to the host cell, (2) penetration of the phage particle into the bacterium, (3) the intracellular multiplication of the virus, and (4) the lysis of the host cell and release of phage progeny. This division is peculiarly appropriate from the experimental viewpoint and leads to the following interpretation of the lytic process. If to an actively growing fluid culture of susceptible bacteria one adds a single phage particle, it will adsorb to a bacterial cell, multiply therein, and eventually cause the cell to lyse. On lysis about 100 phage particles will be liberated thus completing one growth cycle. The 100 phage progeny from the first cycle will then infect 100 bacteria to initiate the second growth cycle. The progeny of the second cycle will initiate a third cycle and so the process will continue at an exponential rate until all susceptible bacteria are lysed. 13 14 BACTERIOPHAGES Similarly, if to a film of susceptible bacteria growing on an agar surface one adds a phage particle, it will adsorb to a bacterium and initiate the first lytic cycle. The phage progeny from the first cycle will infect neighboring bacteria to initiate the second cycle, thus producing a spreading lesion in the bacterial film. Eventually this process will result in a readily visible sterile area in the film of growing bacteria, the bacteria-free area being known as a plaque. Under ideal conditions each infective phage particle will produce a plaque, so that plaque counts give a simple and reproducible measure of the number of phage parti- cles present in a sample inoculated on the assay plate. 1. The One-Step Growth Experiment The over-all aspects of the infective cycle are most conveniently studied by means of the one-step growth experiment of Ellis and Delbriick (1939), which gives the minimum latent period of intracellular virus growth and the average burst size. The latent period is defined as the minimum time between the adsorption of phage to host cell and the lysis of the host cell with release of phage progeny. The burst size is the mean yield of phage par- ticles per infected bacterium. Each of these characteristics can be determined by other techniques, but the one-step growth ex- periment enables both to be determined at once. The tech- nique is as follows: 7. Phage and bacteria are mixed at concentrations that per- mit rapid adsorption. 2. After a suitable adsorption period the mixture is diluted into antiphage antibody to stop adsorption and inactivate unad- sorbed phage. 3. After sufficient time for antibody action the mixture is further diluted in growth medium so that each sample will con- tain about 100 infected bacteria. 4. The suspension of infected bacteria is sampled at intervals until lysis is completed and each sample is assayed by the plaque count method. Each infected bacterium if plated before lysis will produce one THE INFECTIVE PROCESS 15 plaque. If permitted to lyse before plating each bacterium liberates a considerable number of phage particles each of which will produce a plaque. Therefore the plaque count of succes- sive samples remains constant until the end of the latent period when lysis begins. From this time the plaque count increases with each sample until all infected bacteria have lysed, after which it remains constant at the new higher level. The final count of liberated phage particles divided by the initial count of infected bacteria gives the average burst size. The dilution of the adsorption mixture at the end of the ad- sorption period is an essential feature of the experiment because it prevents further adsorption. If the suspension of infected bacteria were not diluted, much of the phage released by the early lysing bacteria would be adsorbed onto yet unlysed bacteria and initiate secondary cycles of infection. The antiphage antibody inactivates unadsorbed phage without affecting the multiplication of adsorbed phage (Delbriick, 1945b). If unadsorbed phage were not inactivated the early plaque counts would include unadsorbed phage as well as in- fected bacteria and the estimate of the burst size would be too small. The one-step growth technique has been discussed in some detail because its significance has not been understood by all phage workers. This technique enables one to determine very simply the effect of changes in the physical and chemical en- vironment on the duration of the infectious cycle and on the yield of virus per infected host cell. 2. Adsorption Adsorption of phage to host cell is the stage in the infectious cycle which is most readily accessible to laboratory study and hence best known. If adsorption is prevented the infectious process cannot proceed and the host cells continue to multiply as if the phage were not present. One crucial factor in adsorption is the ionic environment, so one of the first things to consider in 16 BACTERIOPHAGES Studying any new bacterium-phage system is the effect of vary- ing th^salt concentration in the growth medium. An unfavor- able sah concentration can prevent phage adsorption. It is probable that antiphage antibody acts in part by interfering with adsorption, and antibacterial antibodies can also prevent adsorption by coating the bacterial receptor sites. The specificity of the phage-bacterium relationship is in most cases determined by the adsorption process. When a coliphage for instance fails to attack a particular strain of E. coli, the failure is usually due to lack of adsorption. Certain exceptions to this generalization will be noted in later chapters. The adsorption process is remarkably specific. A single mutation on the part of a host cell can result in loss of ability to adsorb a particular bacteriophage, without loss of susceptibility to other closely re- lated phages. The bacteriophage by a single mutation can gain the ability to adsorb to a particular host without other de- tectable change in viral properties. Another interesting aspect of adsorption specificity is the re- quirement for adsorption cofactors by certain phages. A particu- larly well studied example is a tryptophan requiring variant of coliphage T4 which fails to adsorb to its host cell unless trypto- phan or certain other substances are present in the medium. The tryptophan reacts reversibly with the phage, changing it from a particle incapable of adsorption to a particle which ad- sorbs with high efficiency (T. F. Anderson 1948a). This "'sen- sitization" of the phage particle to adsorption can be counter- acted by indole (Delbriick, 1 948) . Since E. coli can convert tryp- tophan into indole, we have a bacterium converting a sub- stance essential for the attack by a hostile virus into a substance which renders the virus innocuous. This is antibiosis on a primitive level. An important observation by T. F. Anderson (1951) gives further evidence for a high order of specificity in the adsorption process. Stereoscopic electronmicrographs show that the tailed bacteriophages T2, T4, and T6 attach to the host cell by the tips of their tails only. THE INFECTIVE PROCESS 17 3. Penetration Very little is known about penetration, the second step in the infectious process. It has been assumed that the phage particle must penetrate into the host cell interior, since microscopic stud- ies have quite clearly demonstrated that the phage progeny are formed intracellularly and released on rupture of the host cell membrane. Either the parental phage particle must penetrate the membrane or one must invoke biological action at a dis- tance as great as the membrane thickness. Convincing evidence of penetration was obtained by Hershey and Chase (1952) using T2 coliphage labeled with either radioactive phosphorus or sul- fur. The phosphorus labels the phage nucleic acid while the sulfur labels the sulfur-containing proteins. Hershey and Chase found that if the infected bacteria were subjected to violent agi- tation in a Waring Blendor shortly after infection, a remarkable fractionation of the labels took place. The high shearing forces of the Waring Blendor resulted in the liberation of 75 per cent of the radioactive sulfur but only 1 5 per cent of the phosphorus into the medium, the remaining radioactivity staying with the bac- teria. The treatment did not interfere with the infective proc- ess ; the cells remained capable of yielding phage progeny. The radioactive sulfur-labeled substance which was shaken loose by the Waring Blendor was precipitable by antiphage serum and could be readsorbed to sensitive bacteria, that is, it had the biological specificity of the phage particle. The reasonable con- clusion is that the sulfur-containing proteins of the infecting bacteriophage particles remain on the host cell surface from which they can subsequently be shaken loose, while the nucleic acid of the phage particle penetrates into the host cell. The ex- periment completely alters the problem of phage penetration since it is evident that the intact phage particle does not pene- trate and only certain parts of the particle participate in phage re- production. These parts may well gain entrance by some en- zymatic damage to the host cell membrane produced by viral enzymes, or by viral stimulation of host cell enzymes. Evidence for enzymatic activity on host cell materials is particularly good 18 BACTERIOPHAGES for the influenza group of animal viruses (Hirst, 1948), for phages T2 and T4 (Barrington and KozlofT, 1956; Koch and Weidel, 1956b), and for a Klebsiella phage (Adams and Park, 1956). This topic is discussed in Chapter XI. 4. Multiplication Of the basic mechanisms of intracellular phage multiplication very little is known. As usual where knowledge is meagre, hypothesis is rampant. We will not consider hypotheses but merely describe briefly those events that are known to occur dur- ing multiplication of phages related to T2. One type of information is obtained from the survival curves of infected bacteria as a function of dosage of ultraviolet or X-ray irradiation (Luria and Latarjet, 1947). The plaque-forming ability of infected bacteria is as susceptible to ultraviolet inac- tivation immediately after adsorption as is that of unadsorbed phage. However, within a few minutes after adsorption the infected bacteria become far more resistant to inactivation than they were earlier. It is probable that this marked increase in re- sistance is correlated with the first replication of the phage nucleic acid in the interior of the host cell. The inactivation curves re- main exponential until nearly half way through the latent period and then rapidly become multiple hit curves. Apparently there are few potential virus particles in the cell for the first half of the latent period, but then the mean number of potential virus particles per infected host cell rapidly increases. During the initial quiescent period the cytological appearance of the infected bacterium changes markedly, the "nuclear bodies" disintegrate, and the chromatin appears to migrate to the cell periphery. About the middle of the latent period the cell begins to fill with granular chromatin (Luria and Human, 1950). Chemical studies of infected bacteria during the latent period have also been of interest. Following phage adsorption cell division stops. The synthesis of certain adaptive enzymes is in- hibited but energy metabolism as judged by respiration con- THE INFECTIVE PROCESS 19 tinues (Cohen, 1949). Synthesis of ribose nucleic acid almost stops but protein synthesis continues. The synthesis of bacterial deoxyribose nucleic acid (DNA) stops but after a short lag the synthesis of phage DNA begins. The synthesis of phage protein and DNA then continues until bacterial lysis occurs (Hershey, Garen, Fraser, and Hudis, 1954). A third approach to a knowledge of intracellular multiplica- tion consists in stopping virus synthesis by chilling or by meta- bolic poisons such as cyanide at appropriate intervals in the latent period. The infected cells are then lysed by secondary means such as a second phage or sonic vibrations and the lysates are assayed to determine the mean number of mature phage par- ticles present per infected host cell. By such means it has been found that no mature, infectious phage particles are present intracellularly until half-way through the latent period. Then the number of mature phage particles increases as a linear func- tion of time until shortly before lysis (Doermann, 1952). In such circumstances it is useful to distinguish between two kinds of phage particles, mature or infective particles and im- mature, noninfective, vegetative particles. The mature phage particle is the typical extracellular stage which is assayed by its plaque forming ability. The immature or vegetative phage particle is the intracellular stage, undergoing multiplication, potentially capable of producing mature phage particles but not detectable by the plaque count method because it is noninfec- tious. The following picture of phage multiplication emerges from these observations: 7. The phage particle adsorbs to the host cell surface. 2. Certain reproductively essential components of the phage penetrate to the cell interior, expendable portions being dis- carded at the ceil surface. The phage particle thus changes from the mature to the vegetative condition. 3. The host cell stops dividing, certain synthetic reactions stop, and extensive cytological changes occur. 4\. Synthesis of phage protein and phage DNA start, and the 20 BACTERIOPHAGES irradiation inactivation curves become multiple hit, indicating that phage multiplication has started. 5. At about the mid-point of the latent period, conversion of vegetative phage particles into mature phage begins and the number of infective particles found per cell increases until host cell lysis occurs. It should be noted that this picture applies specifically to T2 and related coliphages. The details of the reproductive cycle may be quite different for other phages. Much additional in- formation about the reproduction of bacteriophages has been obtained from nutritional studies, tracer experiments, and genetic analysis, which will be reported later in appropriate chapters. 5. Lysis of Host Cell and Release of Phage Progeny Of the last stage in the infectious process, lysis of the host cell, almost nothing is known. Lysis is not dependent on the ac- cumulation of mature phage particles because lysis will occur even when phage multiplication is interrupted by chemicals such as cyanide (Cohen, 1949) or proflavine (Foster, 1948). No general method of temporarily interrupting the lytic process is known except chilling, which slows down all enzymatic reactions. Many chemical agents will interfere with phage growth but the application of such agents soon leads to irreversible changes and failure of lysis. The latent period of growth of certain phages can be greatly prolonged by the phenomenon of "lysis inhibition" (Doermann, 1948a) but the mechanism of this is not understood. Lysis may be induced before the end of the latent period by "lysis from without" by massive doses of phage (Delbriick, 1940b), by sonic vibration (Anderson and Doermann, 1 952a), or in certain cases by enzymes such as lysozyme (Lwoff, Siminovitch, and Kjeld- gaard, 1950). Towards the end of the normal latent period the infected bacteria become increasingly fragile and some, lysis will result from laboratory manipulations such as pipetting (Levin- thai and Visconti, 1953). This pre-burst fragility of the bac- terial membranes has led to the erroneous conclusion from the THE INFECTIVE PROCESS 21 examination of electronmicrographs that the bacterial cell wall ruptures before the end of the latent period and that the phage particles then mature in the extruded cell sap (Wyckoff, 1 949a) . Lysis of the host cell accompanied by release of virus particles is readily demonstrated by direct observation in the dark field microscope. The bacteria can be observed to disintegrate at a time corresponding to the end of the latent period. At the moment of disappearance of each bacterium a cloud of scintil- lating particles is released in numerical agreement with the known burst size (d'Herelle, 1926; Merling-Eisenberg, 1938; Pijper, 1945). Release of phage particles is coincident with lysis of the host cell also in lysogenic bacteria, as demonstrated by micromanipulation techniques (LwofT and Gutmann, 1950). 6. Lysis in Fluid Media We have described the sequence of ev^ents when a few phage particles are added to a growing culture of susceptible bacteria. The phage particles adsorb and in due course the infected bacteria lyse liberating the phage progeny. These then adsorb to unlysed bacteria initiating a second cycle of infection. The infectious cycles follow one another, the phage population in- creasing in each cycle by a factor equal to the burst size, until all susceptible bacteria have lysed. The final phage population is usually of the order of ten to several hundred times the maxi- mum bacterial population achieved. The theoretical phage yield would be the bacterial population multiplied by the aver- age burst size. The actual yield is often less because some of the liberated phage is lost by readsorption to yet unlysed bacteria or to bacterial fragments. The over-all time required for lysis depends principally on the latent period and the number of phage particles inoculated, and to a lesser extent on the ad- sorption rate and the bacterial concentration. If the entire bacterial population is susceptible to the phage, the culture after lysing will remain clear indefinitely. However, if the culture contains a few phage-resistant variants, the lysed culture will on further incubation become turbid again due to 22 BACTERIOPHAGES multiplication of the variant cells. The mutant bacterial population will be resistant to most of the phage particles, but it not infrequently happens that a large phage population contains a few variant phage particles with an extended host range, capable of attacking the phage resistant bacteria. These host range mutant phage particles will multiply at the expense of the mutant bacterial population and may cause a second clearing of the culture. The second lytic episode may be followed by a third turbidity if the bacterium has produced a mutant which is resistant to the host range mutant of the phage. Bacterial and phage mutations will be discussed in more detail in later chapters. The physiological state of the bacterial population is another important factor in the lysis of fluid cultures. The latent period is minimal and the burst size largest when the bacteria are grow- ing in the logarithmic phase in a nutritionally adequate me- dium. If the bacterial population has passed from the logarith- mic growth phase into the stationary phase, the cells become poor hosts for phage growth, and lysis becomes very slow or may not occur (Delbriick, 1940b). 7. Lysis on Solid Media If 10^ bacteria are spread on the surface of an agar plate, and then incubated, they will grow as a multitude of minute colonies which soon become confluent giving a uniform film or "lawn" of bacterial growth. The bacteria will continue to grow until ex- haustion of nutrients or accumulation of toxic products inter- feres. If a single phage particle is placed on the plate early in the growth of the bacterial film, it will adsorb to a bacterium and initiate an infectious cycle. The phage progeny remain localized at the site of lysis, and gradually radiate from this site by a slow diffusion through the agar. In this process neighbor- ing bacteria will be attacked and lysed, increasing the phage population further. Eventually the zone of lysis may become large enough in area to be readily visible to the naked eye as a circular zone devoid of bacteria. These cleared areas were ob- served by d'Herelle who called them "taches vierges" (virgin THE INFECTIVE PROCESS 23 spots), or "plages" (beaches). In German they are called "Locher" (holes), and in English "plaques" or "clearings." The plaque size and morphology are characteristic features for a particular phage-bacterium combination, and may vary for different phages grown on the same bacterial host or for the same phage grown on different bacterial hosts. Also the plaque size and morphology may change as a result of a single step mutation on the part of the phage. The plaque size increases during incubation of the inoculated plate, usually reaching a maximum after 8 to 12 hours, although there is sometimes a further slow increase in size on longer in- cubation. Plaque development does not continue indefinitely because it is dependent on active growth of the host bacteria. When bacterial growth slows because of exhaustion of nutrients, the lysis of infected bacteria is interfered with and the burst size is markedly reduced. Another factor affecting size of plaques is the size of the in- dividual virus particle. Elford and Andrewes (1932), who measured the particle size of a number of strains of bacteriophage by filtration through collodion membranes, noted that there was a rough inverse relationship between particle size and plaque size. They suggested that a small particle diffuses more rapidly than a large particle and consequently that, during the limited time available for plaque development, a population of small particles will radiate farther from a common center than will a population of large particles. In spite of the fact that this is a valid generalization and that its physical basis is readily under- stood it has been repeatedly criticized and does not invariably hold (Felix, 1953). In fact, plaque size may be limited by many factors other than diffusion. Such factors are the burst size, latent period, and adsorption rate. The meaning of the general correlation between plaque size and phage particle size may be questioned because there is also a tendency for small particle phages to have short latent periods. The role of adsorption in affecting size of plaques has been n6ted several times. If a phage particle does not adsorb to a 24 BACTERIOPHAGES bacterium to initiate the first infectious cycle until late in the de- velopment of the bacterial lawn the result will obviously be a small plaque or perhaps no visible plaque. It is typical of slowly adsorbing phage-bacterium systems that a single plate will show great diversity of plaque sizes, those particles ad- sorbing early producing much larger plaques than those ad- sorbing late (Wahl and Blum-Emerique, 1952a; Sagik, 1954). That this is the correct explanation can be demonstrated by adsorbing the phage particles to bacteria in fluid medium, in- activating or removing the unadsorbed phage, and then plating the infected bacteria. This results in a much greater uniformity of plaque size than when free phage is plated because all plaques start to develop at about the same time. Since the diflfusion rate is often a limiting factor, anything that interferes with diffusion will reduce the plaque size. The concentration of agar in the plating medium is irriportant for this reason, a more dilute agar permitting development of larger plaques. The agar layer method of Gratia (1936c) using a diluted agar in the upper layer is a useful means for increasing plaque size over that obtained by the classical spreading tech- nique. Another factor which may affect plaque size and morphology is the genetic constitution of the phage particle. Phage T2r"'' when plated on strain B of £". coli produces a small, fuzzy plaque. Phage T2r, derived from T2r''" by mutation, gives a larger plaque with a sharp outline. The difference is due to the phenomenon of "lysis inhibition" (Doermann, 1948a) caused by superinfection with phage during the latent period. These plaque type mutants have been of great value in the study of bacteriophage genetics (Chapter XVIII). The clearness or opacity of the plaque is a characteristic feature of the phage-host pair. If the bacterial culture is com- pletely susceptible, containing no resistant bacteria, the resultant plaques will be clear. If a small proportion of resistant bacteria is present the plaques will be clouded by scattered minute colonies. If large numbers of resistant bacteria are present the THE INFECTIVE PROCESS 25 plaques will be turbid and may be completely overgrown and in- visible on longer incubation. This may also occur if the phage particles are able to convert a considerable proportion of the sensitive bacteria into the lysogenic condition. In certain cases plaque development is accompanied by the release of a soluble lytic enzyme of much smaller particle size than the phage and which produces a spreading halo of lysis around the plaque. This lytic substance does not reproduce itself and in certain cases is a capsule hydrolyzing enzyme (Humphries, 1948; Adams and Park, 1956). It will diffuse across bacteria-free areas on the agar to alter the appearance of bacteria that are beyond reach of the more slowly diffusing phage particles (Sertic, 1929a). An additional morphological peculiarity of certain plaques should be noted. One sometimes observes on large plaques, which are turbid because of the growth of resistant bacteria, small subsidiary plaques developing on the secondary turbidity. These small plaques are due to host range mutants which are produced during phage growth in the parent plaque. These secondary plaques are quite analogous to papillae on bacterial colonies, and are illustrated by Wahl and Blum-Emerique (1952a). The importance of plaques in phage research is two-fold. They furnish a highly accurate and reproducible assay method for counting the number of viable phage particles present in a given sample of phage preparation, and they furnish most of the characters by which hereditary variation in phages is studied. CHAPTER III ENUMERATION OF BACTERIOPHAGE PARTICLES Quantitative work with bacteriophages depends on simple and accurate means of counting the particles. It is essential to be able to determine relative numbers of phage particles in different samples with precision and it is often important to be able to measure the absolute number of particles. Three principal assay methods have been used: (7) plaque counts on nutrient agar plates seeded with phage-susceptible bacteria, (2) dilution end-points using lysis of fluid bacterial cultures as an indicator for presence of phage, and (3) measure- ments of the length of time required for lysis of a standard fluid bacterial culture. 1. Plaque Counts Only the first method mentioned is generally useful. The method was originally described by d'Herelle in 1917. An ap- propriate dilution of a phage preparation is mixed with a con- centrated suspension of susceptible bacteria and an aliquot of the mixture is spread on the surface of an agar plate. On in- cubation the bacteria grow as a film, spotted with circular clear areas or plaques produced by the lytic action of the bacterio- phage. D'Herelle found that large amounts of phage could be recovered from the plaques, but that none was present in the unlysed areas of the plate. Since the number of plaques ob- tained was directly proportional to the amount of phage prepara- tion spread on the plate, the plaques could be understood as phage colonies containing the descendents of single phage particles. It does not follow, however, that every phage particle 27 28 BACTERIOPHAGES will produce a plaque. Although the plaque count is a good relative assay method, it will not give the absolute number of phage particles present in the inoculum. If the same phage preparation is assayed under different en- vironmental conditions it is immediately obvious that the plaque count has no absolute significance. A change in salt concentra- tion can change the assay of coliphage T2 1,000-fold, while omis- sion of tryptophan can reduce the plaque count of certain strains of phage T4 by a factor of 10"*. Assaying the same phage preparation on a series of different susceptible bacterial strains will often result in a different assay value with each host. Such observations led to the concept of efficiency of plating (Ellis and Delbruck, 1939). The relative efficiency of plating (EOP) can be defined as the plaque count obtained under a given set of conditions relative to the plaque count under standard conditions; for instance, the EOP of phage T2 on E. coli strain B/6 is 0.8 of that on strain B. For such comparative purposes it is reasonable to use as standard those conditions that are found to give the highest plaque count. However, it is unwise to assume that such standard conditions enable one to count all the potentially in- fective phage particles present in the inoculum. The absolute efficiency of plating may be defined as the plaque count relative to the total number of phage particles present in the sample. At present there are few practicable methods for determining the absolute efficiency of plating. One method compares the plaque count obtained per unit volume of phage preparation with an electron microscopic count. Luria, Wil- liams, and Backus (1951) made measurements of this kind. The phage was suspended in a solution containing a known con- centration of polystyrene latex particles, and the mixture was sprayed upon specimen screens for electron microscopy. All morphologically characteristic phage particles and all latex particles were counted in a number of fully visible droplet pat- terns. The latex particle count was used to determine the total volume of counted droplets, and the phage particle count then ENUMERATION OF BACTERIOPHAGE PARTICLES 29 gave the phage concentration within the sampHng errors of the two counts. The electron microscopic count was then compared with the plaque count obtained per unit volume of mixture. Such comparisons were made with a number of preparations of phages T2, T4, and T6. The ratio of plaque-forming units to visible particles varied from 0.4 to 1 .4, falling below unity for 8 preparations and rising above for 3. These results indicate that the assay methods for T2, T4, and T6 have an absolute EOP greater than 0.5. The low EOP for most of the preparations probably indicates that they contained noninfective particles. A mechanism of origin of noninfective particles of phage T2 has been identified by Sagik (1954), and it is likely that similar principles apply to other phages. A considerably less accurate way of measuring the absolute EOP involves determinations of the weight of the infective particle by two independent methods. For phage T2 the dry weight per infectious unit in purified preparations (Herriott and Barlow, 1952) is about twice the weight of a single particle as measured by hydrodynamic methods (Taylor, Epstein, and Lauffer, 1955). The correspondence indicates an absolute EOP of about 0.5 and also shows that the phage preparations are not grossly impure. A modification of d'Herelle's plaque counting method, called the agar layer method, is almost universally used by phage workers. This method was invented by Gratia (1936c) and independently by Hershey, Kalmanson, and Bronfenbrenner (1943a). About 2 ml. of melted 0.6 per cent agar is cooled to 45 ° C. and inocu- lated with a drop of a concentrated suspension of the host bacterium. A measured volume of phage suspension is then added and the entire mixture poured over the surface of a hardened layer of nutrient agar. After the upper agar layer has solidified the plate is incubated. The bacteria grow as a multi- tude of tiny colonies within the soft agar layer, fed by the layer underneath, forming an opaque background against which plaques are easily seen. Advantages of the agar layer method are: (7) phage samples up to 1 ml. in volume can be plated per 30 BACTERIOPHAGES petri dish; (2) the time required for plating each sample is less than 30 seconds; (3) the plaque size is larger than that given by the spreading method ; and (4) the efficiency of plating is often higher. The agar layer method has permitted accurate kinetic study of reactions that were too rapid to be followed by other methods of phage assay, and has been indispensable in genetic studies relying on recognition of different types of plaque. 2. Dilution End-Points The dilution end-point method consists in finding the smallest amount of the phage preparation that will bring about lysis of a growing culture of susceptible bacteria. Subject to the assump- tion that one phage particle will cause lysis, this method measures the number of particles. However, if the adsorption is poor or phage multiplication slow, lysis may occur only when a consider- able number of phage particles are inoculated, and so the phage population will be grossly underestimated. In such cases the estimate can be improved by testing for presence of phage in all tubes in which lysis failed. This method can be made quite precise by multiplying the number of tubes tested at the limiting dilution. The phage preparation is diluted to the point where each sample should contain about one phage particle. Then 50 young cultures of the host bacterium are each inoculated with a sample of the diluted phage preparation and incubated. Each tube is then checked for lysis or for presence of phage. If the test for infec- tive particles is reliable, the proportion of cultures in which phage multiplication did not occur will be equal to e~", from which n, the mean number of infective particles per sample, can be computed. The measurement fails unless roughly half but not more of the tubes tested prove to contain phage, which means that the method is too laborious for routine assays. It is in- dispensable, however, in certain types of experiment, and the student of phage should be familiar with the Poisson distribu- tion, of which the computation mentioned is an application. The accuracy of experimental results obtained by this method ENUMERATION OF BACTERIOPHAGE PARTICLES 31 can be computed (Haldane, 1939). Phage assays by this method should check closely with assays by the agar layer method if both media are optimal for phage multiplication. This is one way of defining the relative efficiency of plating. 3. Measurement of Lysis Time A kinetic method of phage assay was developed by Krueger (1930), for use with rapidly lysing phages, which gives precise and reproducible results within certain limits. The procedure is to add appropriate dilutions of phage to tubes containing standard suspensions of actively growing host cells, and to de- termine the length of time required for lysis to reduce the bac- terial turbidity to an arbitrary end-point. The time of lysis is inversely proportional to the logarithm of the initial phage con- centration over a considerable range. A standard phage prep- aration is assayed in parallel with the unknown and the activity of the unknown is expressed in terms of an arbitrary unitage de- fined for the standard. One defect of the method is that it does not give the assay in terms of infectious phage particles but only in amounts relative to the standard. This defect could be reme- died by determining the content of infective particles in the stand- ard phage preparation by one of the methods discussed above. Another defect is that the method is applicable only to free phage particles. A suspension of phage after adsorption to host cells will assay higher than the same amount of free phage because the time required for adsorption is not included in the observed lysis time for the adsorbed phage but is included in the case of the standard phage suspension. Therefore the application of the kinetic assay method to the study of phage adsorption (Krueger, 1931) and intracellular multiplication (Krueger and Northrop, 1930) has led to erroneous interpretations of the data. The method has been applied successfully to the assay of phages for anaerobic bacteria when for technical reasons the plaque count method did not giv^e satisfactory results (Gold and Wat- son, 1950). The method may well have other applications and 32 BACTERIOPHAGES should be considered whenever plating methods are unsatis- factory. One or another of these assay methods should suffice for most purposes. A few other methods have been used for special pur- poses. The use of the electron microscope to determine the total number of morphologically typical phage particles in a known volume of suspension has been mentioned above. A method that involves the ability of phage particles to kill the host cell on adsorption is particularly suitable to the assay of phage inac- tivated by ultraviolet light. On adsorption the phage particles are distributed over the host cell population in accordance with the Poisson distribution. The viable bacterial population is accurately determined before and after phage adsorption. The proportion of the bacterial population that survives is equal to ^~" where n is the mean number of lethal particles adsorbed per bacterium. Then n times the initial number of bacteria per unit volume gives the number of phage particles, per unit volume, adsorbed and capable of killing. Conditions should be such that most of the phage particles present are adsorbed to bacteria and the value of n should be greater than one (Luria and Dulbecco, 1949). 4. Intracellular Phage Mature phage particles present in yet unlysed bacteria present a special assay problem. An infected bacterium if plated be- fore bursting will form only a single plaque regardless of how many mature phage particles it may produce. However, the number of mature phage particles present in the bacterium at any time during the latent period may be determined by inter- rupting phage development by chilling or by chemical agents, and liberating the mature phage by breaking open the infected bacteria by biological, enzymatic, or physical means (Doermann, 1952; Anderson and l])oermann, 1952a). The average number of mature phage particles per bacterium liberated by premature lysis may then be determined by plaque counting methods. A peculiar biochemical feature of phage T2 and its relatives ENUMERATION OF BACTERIOPHAGE PARTICLES 33 has been exploited for the analysis of vegetative phage within infected bacteria. The nucleic acid of this serological group of phages contains the unique base hydroxymethylcytosine (Wyatt and Cohen, 1953). This base may be readily distinguished from the other bases present in bacterial or phage nucleic acids by paper chromatography of acid hydrolyzates of the nucleic acid fraction, and may be determined quantitatively. It is there- fore possible to distinguish the DNA of phage T2 from host cell DNA and to determine quantitatively the total amount of phage DNA present per infected cell. The amount of mature phage DNA per cell can be calculated from plaque counts on prematurely lysed infected bacteria. The amount of phage DNA present as vegetative phage in the infected bacteria can be calculated by subtracting the mature phage DNA from the total phage DNA. This method has been used by Hershcy, Dixon, and Chase (1953) to study the intracellular multiplication of phage T2. Another method for assay of intracellular phage substances is based on the high specificity of phage antigens. The antibody that neutralizes phage infectivity reacts with a protein situated on the phage tail but fails to react with any substances present in uninfected bacteria. The relative amount of neutralizing antibody present in a sample of serum can be accurately deter- mined (see Chapter VIII). Phage tail antigen will combine with phage neutralizing antibodies and so decrease the neutraliz- ing ability of the serum. This "serum-blocking power" of phage antigens can be calibrated in terms of phage particle equivalents and used as a measure of the total amount of phage tail antigen present in infected bacteria when they are prema- turely lysed at various times during the latent period. The amount of phage tail antigen already included in mature phage particles can be calculated from the mean number of mature phage particles present and the known antigen content per phage particle. The difference between the content of total tail antigen and mature phage tail antigen gives the phage tail antigen not yet built into mature phage at the time of prema- 34 BACTERIOPHAGES ture lysis (DeMars, 1955). Similar techniques have been used to determine the amount of phage-specific complement-fixing antigen present at various times in the latent period (Y. T. Lanni, 1954). It is evident that the speed and accuracy of assay methods for extracellular and intracellular mature phage and for various phage components have been major factors in the very rapid increase in knowledge of phage reproduction. The introduc- tion of plaque counting methods into the field of animal virology by Dulbecco (1952c) is permitting a parallel development in this field. CHAPTER IV SIZE AND MORPHOLOGY OF BACTERIOPHAGES The presence of phage particles inside the infected bacterial cell and the release of these particles on bursting of the cell were seen in the dark field microscope by d'Herelle (1926) and photographed by Merling-Eisenberg (1938). The agglutination of phage particles by antiphage serum was photographed in the ultraviolet microscope by Barnard (Burnet, 1933c). Although there is no difficulty in seeing light scattered by the larger phages under these conditions, they are not resolved by light micro- scopes. Consequently our knowledge of the morphology of the particles begins with their examination under the electron microscope. 1. Electron Microscopy of Bacteriophage Particles The electron microscope was first applied to the study of phages by Ruska (1940) and Pfankuch and Kausche (1940). It was soon clear that the particles are striking and characteristic in shape (Ruska, 1943; Luria, Delbriick, and Anderson, 1943). Most of the phages examined resemble tadpoles, with long tails attached to spherical, cylindrical, or polyhedral heads. Some phages at first thought to be tailless spheres revealed on closer examination polyhedral shapes and a rudimentary tail (Wil- liams and Fraser, 1953). In numerous instances the tail proves to be a specialized organ for attachment to bacteria ; very likely all phages possess such an organ (T. F. Anderson, 1953). Mi- crographs of several phages are shown in Figures 1, 2, and 3 and the frontispiece of this book. 35 36 BACTERIOPHAGES TABLE I Phage Morphology in Electron Microscope Subject Reference Book on Electron Microscopy Review on Bacteriophages Coli phage Tl Coli phage T2 Coli phage T4 Coli phage T6 Coh phage CI 6 Coli phage T7 Coli phage T5 Salmonella puUorum phage Erwinia carotovora phage Streptococcus lactis phage Streptomyces griseus phage Bacillus megaterium phage Streptococcus lactis phage Corynebacterium diphtheriae phage Mycobacterium smegmatis phage Streptococcus lactis phage Bacillus subtilis phage Salmonella paratyphoid B phage Staphylococcus phage (Twort) Staphylococcus phage K Pseudomonas aeruginosa phage Many phages Wyckoff (1949b) Ruska (1943) Williams and Eraser (1953) Williams and Eraser (1953) Williams and Eraser (1953) Williams and Eraser (1953) Giuntini, Lepine, NicoUe, and Croissant (1947) Williams and Eraser (1953) Williams and Eraser (1953) Baylor, Severens, and Clark (1944) Chapman, Hillier, and Johnson (1951) Cherry and Watson (1949) Koerber, Greenspan, and Langlykke (1950) McLauchlan, Clark, and Boswell (1947) Parmelee, Carr, and Nelson (1949) Toshach, S. (1950) Whittaker (1950) Williamson and Bertaud (1951) Giuntini, Lepine, Nicolle, and Croissant (1947) Giuntini, Lepine, Nicolle, and Croissant (1947) Giuntini, Lepine, Nicolle, and Croissant (1947) Hotchin (1954) Schultz, Thomassen, and Marton (1948) Terada (1956) Although many phages have now been examined (Table I), it is unfortunate that no morphological comparison has been made of a large group of phages classified also by other means. From the study of a small group, it appears that morphological fea- tures will aid in classification of phages much as they do in classi- fication of other organisms (Delbriick, 1946b). SIZE AND MORPHOLOGY OF BACTERIOPHAGES 37 Microscopic measurement of size of virus particles is com- plicated by shrinkage and distortion on drying, increase of size by shadowing with metals, and difficulties in calibration. These difficulties have been minimized in various ways (Anderson, 1951 ; Williams, 1953). Some very careful measurements have been published by Williams and Fraser (1953). Estimates of Figure 1. Phages of diverse morphology. Left, phage 1'3, unpurified lysate. X 46,000. UnpubHshed. Right, two phages carried by a lysogenic strain of Bacillus cereus (Kellenberger and Kellenberger, 1952). Unpurified lysate, X 40,000. Unpublished. particle size of several bacteriophages by electron microscopy and other methods are given in Table II. Further information about the structure of several phages has been obtained by degradation of the particles in various ways. Coliphage T2 and its relatives, and the staphylococcal phage K, are subject to osmotic shock (Anderson, 1949; Hotchin, 1954). To observe this, the phages are suspended in concentrated solu- tions of sodium chloride and, a few minutes later, water is dumped in. The phage particles lose their characteristic in- fectivity. Microscopic examination now reveals empty, but more or less intact, phage ghosts. Herriott (1951a) showed that the ghosts, after treatment with deoxyribonuclease, contained 38 BACTERIOPHAGES most of the phage proteins and retained abihty to adsorb to and kill bacteria. The phage nucleic acid presumably escapes through cracks in the phage head and may be seen in electron micrographs as long filaments about 20 A in thickness (Williams, 1953). These phage particles consist, therefore, of a nucleic acid core surrounded by a membrane responsible for the overall Figure 2. Phage T2 seen inside bacteria infected 40 minutes earlier. UJ- trathin sections prepared as described by Kellenberger, Ryter, and Schwab (1956). X 32,400. Unpublished. shape of the particle (T. F. Anderson, Rappaport, and Musca- tine, 1953). T5, as shown in a somewhat diff'erent manner, is similarly constructed (Lark and Adams, 1953). Some additional features of the tail structure of phages T2 and T4 have been resolved in considerable detail. Kellenberger and Arber (1955) observed stepwise changes in the morphology SIZE AND MORPHOLOGY OF BACTERIOPHAGES 39 of these phages during the action of oxidizing agents (Figure 3) . WilUams and Fraser (1956) described degradation products produced by freezing and thawing. Both groups reached es- sentially similar conclusions about the structure of the phage tail, which may be described as follows. The tail is composed of a rigid central core or pin running the full length, and an outer tubular sheath. The sheath itself is composed of two Figure 3. Left, phage T2, purified prepai ales. X31,- 500. Unpublished. Right, T2 particles treated with hydrogen peroxide showing alteration of tail structure. X 32,500. Reproduced from E. Kellen- berger and W. Arber, 1955, Z. Naturforsch., 10b, 698, with permission. parts. The half proximal to the head of the particle retains its integrity under the influences mentioned, and is occasionally obtained in detached form, when its hollow structure becomes evident. The distal half of the tail sheath is readily removed, and yields long slender fibrils, probably four per phage particle, in the process. These fibrils may be coiled about the tail pin, or may already exist partly as free ends, in the natural state. The central pin often shows a knob at one end, to which remnants of fibrils may or may not be adhering. Isolated pins were ob- 40 BACTERIOPHAGES - ^^ 3 o o ^ o o o 1 XXX U-l LO LO «N (N (N X X 1 X X o o I I I LO u-i Ln o X X X X LO LO LT) LO SO SO -O SO 1 g i I SO 00 ^- O O O o ti I LO LO lO lO ! i 00 rO so [^ •B I P B •2 ^ S i O LO CN I 1 I I ■^ O LO LO LO LO CS SO \0 ^ 1 1 I 1 :2 1 I I I I s I I so ^ -^ O (^j Tt SO r- M 2 LO ^ (M ro r- ^' "^ ro ro 5 c 2-Sg KJ o s-i — ' 3 ^ .2 ^ J E O S J>' ^ ^ ^ -rt •> ^ o H fe ii :^ ^ c Dh i/D SIZE AND MORPHOLOGY OF BACTERIOPHAGES 41 served earlier in unpurified lysates by Herriott and Barlow (1952). Avery similar structure for the tail of the staphylococ- cus phage K had been surmised by Hotchin (1954). The func- tional significance of this structure is discussed by Williams and Fraser (1956). Discontinuity of internal structure of the phage head is notice- able in certain types of preparation (Luria, Delbruck, and Anderson, 1943; Terada, 1956). Unlike the tail structure de- scribed above, which undoubtedly reflects functional differentia- tion, the appearance of the head contents may depend on acci- dents of drying (T. F. Anderson, Rappaport, and Muscatine, 1953). 2. Ultrafiltration Some of the earliest attempts to measure the size of phage particles sought to determine the largest pore diameter of col- lodion membranes through which the particles could not pass. The method was chiefly developed by Elford, whose review (El- ford, 1938) is the source of the information given here. The method is based on the following principles. A series of graded collodion membranes can be prepared by systematically varying the solvent mixture from which the mem- brane is precipitated. Measurements of the rate of flow of water through such membranes yield, with the aid of hydrodynamic theory, a nominal average pore diameter. Finally, an empirical factor, arrived at by testing substances of known particle size, relates the average pore diameter of the membrane to the size of particles it will hold back. This factor presumably corrects for heterogeneity of pore size, adsorption, electrostatic effects, and other imponderables. (Adsorption is minimized by suspending the phage in peptone solutions.) Elford multiplied the hydro- dynamic pore diameter of the coarsest membrane that would fail to pass phage by factors varying between 0.33 and 0.50, de- pending on the membrane, to obtain the diameter of the phage particle. The sizes he determined have proved to be somewhat too small. Approximately correct diameters are obtained from 42 BACTERIOPHAGES Elford's data by increasing his empirical factor to 0.83 (Lea, 1946). Nevertheless, Elford's data were for a long time the best information available as to size and diversity of size of dif- ferent phages. The relative sizes he determined were remark- ably exact. This is shown in Table II, which lists Lea's correc- tions to Elford's estimates. The method is still useful. Now that a number of phages of known size are available, it should be possible to estimate the comparative size of an unknown phage with considerable pre- cision. The usefulness of the method is all the greater since it does not require the phage to be purified or concentrated. For some of the smaller phages, not yet identifiable in the electron microscope, probably no better method is available. Ultrafiltration is also useful for concentrating and purifying phages (Elford, 1938) and for isolating phage-specific materials smaller in size than phage particles (Burnet, 1933b; DeMars, 1955). 3. Analytical Centrifugation Centrifugal measurements of particle size of phages were also attempted some years ago. Ingenious methods were developed to follow the sedimentation of infective particles by plaque counts under conditions in which no sedimentation boundary is es- tablished. This work is reviewed by Elford (1938). The ap- plication of the modern optical ultracentrifuge to the study of phages is described by Hook, Beard, Taylor, Sharp, Beard (1946), Putnam (1950), and Taylor, Epstein, and Lauffer (1955). Sizes of some phage particles calculated from sedimentation constants are given in Table II. These are computed without correction for water content and departure from spherical shape, which means that they are underestimates that have, in fact, no exact meaning. From combined sedimentation and dif- fusion measurements, the true particle weight can be calculated. In this way Taylor, Epstein, and Lauffer (1955) found the dry weight of a particle of T2 to be 3.3 X 10"^" grams. Putnam's SIZE AND MORPHOLOGY OF BACTERIOPHAGES 43 (1950) estimate for T6 is very similar. This corresponds to a particle diameter of 75 mju, which should be understood as the diameter of a phage particle dried and compressed to a sphere of density 1.5. These hydrodynamic measurements of the particle weight of T2 and T6 are the only satisfactory ones of their kind for any phage. Taylor, Epstein, and Lauffer (1955) also measured the wet density of T2 by sedimentation in a sucrose gradient. The density found, 1.27, calls for a water content of 37.5 per cent, a natural particle weight of 5.3 X 10"^*^ grams, and a spherical diameter of 92 m^u. If anything this water content is probably underestimated, because sucrose may extract water from phage (Anderson, Rappaport, and Muscatine, 1953). Phages T2 and T6 exist in two reversibly interconvertible forms. One, sedimenting and diffusing the more rapidly, pre- dominates at pH 5, the other at pH 7. According to Taylor, Epstein, and Lauffer (1955), these probably differ only in shape. It is possible that these difTerences reflect different states of the tail fibrils described earlier. 4. Diffusion Constants Measurements of diffusion coefficients are needed for the proper interpretation of sedimentation constants, and also lead directly to an equivalent particle diameter computed on the as- sumption of spherical shape (Putnam, 1950). Early applica- tions of diffusion measurements by the porous disk method to phages (Northrop, 1938; Kalmanson and Bronfenbrenner, 1939) yielded erroneous results and improbable theories. What was measured was not diffusion but flow of solvent through the disk (Hershey, Kimura, and Bronfenbrenner, 1 947) . Poison (1948) measured reasonable diffusion constants of phages T3 and T4 but Poison and Shcpard (1949) studying the same two phages again decided that T4 must be self-motile. In all these attempts, efforts were made to measure diffusion of infective particles in dilute solutions. None of them proved very suc- cessful. 44 BACTERIOPHAGES Diffusion of phage T6 across an optically visible boundary was carefully studied by Goldwasser and Putnam (1951). They found that the particles diffused like spheres 94 mfx in diameter, in excellent agreement with physical predictions. Similar re- sults were obtained by Taylor, Epstein, and Lauffer (1955) for T2. 5. Electrophoresis The mobility of particles in an electric field yields information about the sign and density of charges on their surfaces, but does not permit any estimate of size. Only phage T6 has been stud- ied in this way (Putnam, 1950). The particles were negatively charged down to pH 5.2, below which measurements were not attempted. Only one rather diffuse boundary was observed for the best preparations. Since reversal of the electric field did not sharpen the boundary, no evidence for electrophoretic in- homogeneity was obtained. The relatively slow mobility sug- gested a protein, rather than a nucleic acid, surface. 6. Ionizing Radiations Certain radiations such as electrons, protons, deuterons, neutrons, alpha particles, gamma rays, and X-rays are called ionizing radiations because in traversing matter they produce ion pairs by ejecting electrons from atoms. This ionization is often followed by chemical change, which if it occurs in an es- sential molecule can result in death of cells and viruses. All viruses and many single-celled organisms are inactivated by ionizing radiations as an exponential function of the dose of radiation. This has been interpreted as indicating that a single ionization is sufficient to inactivate (Lea, 1 946) . One may readily determine the dose of radiation which will produce an average of one lethal event per organism; that is the 37 per cent survival dose. Since the density and something about the spatial distribution of ionizations produced by a given dosage SIZE AND MORPHOLOGY OF BACTERIOPHAGES 45 of various radiations are known, it is possible to calculate from the 37 per cent survival dose the size of the target in which ioniza- tions must be produced in order that inactivation results. A description of the target theory and of the methods of calculating the target size from the 37 per cent survival dose of ionizing radiations is given by Lea (1946). For small viruses the calcu- lated target size is within experimental error the same as the virus size, indicating that a cluster of ionizations anywhere within the virus particle may be lethal. However for larger viruses the target size has been found to be smaller than the size of the virus particle. The larger viruses are heterogeneous in sensitivity containing considerable amounts of material that is relatively insensitive to irradiation. The target size will be a minimum estimate of the virus particle size. When the target size is de- termined using two different kinds of radiation such as X-rays and alpha particles, the results are usually in good agreement in the case of small viruses. However with larger viruses the target size determined from alpha particle data is larger than that calculated from X-ray data. This has been taken to mean that with the larger viruses the target is not a uniformly sensitive sphere as assumed in the theory but is structurally complex. That the particles of the larger viruses are structurally complex is evident from electron micrographs, and the inactivation data extend this complexity even into that portion of the particle which is radiation sensitive. Further applications of the target theory to the study of size and structure of viruses are discussed by Pol- lard (1953, 1954). Size of bacteriophages determined with the aid of target theory are included in Table H for comparison with sizes de- termined by other methods. These have been derived from graphs given by Lea (1946) and more recent data. It should be realized that the theory involves arbitrary assumptions that render interpretation doubtful (Watson, 1950). In particular it is assumed that two or more ion pairs are wasted for each lethal ionization within the phage particle, the number depend- ing on target size and kind of radiation. 46 BACTERIOPHAGES 7. Nonionizing Radiations Ultraviolet and visible light can also inactivate viruses. How- ever, the mechanism of inactivation is quite different from that of ionizing radiations. In the case of the relatively low energy ultraviolet and visible radiations, entire quanta of radiation are absorbed by specific chemical structures. Under suitable con- ditions the energy absorbed may be enough to break chemical bonds and inactivate viruses. Since the energy is absorbed only by certain structures, the rate of killing gives no information about the size of the virus particle. It was thought at one time that there was an inverse relationship between sensitivity to ultraviolet light and virus particle size. However, lambda, the phage carried by the lysogenic strain K12 of £". coli, is 13 times more resistant to ultraviolet radiation than is coliphage T5, which is about the same size (Weigle and Delbriick, 1951). The various physiological effects of ionizing and nonionizing radiations will be considered in detail in Chapter VI. 8. Weight of the Infectious Particle For those phages that have been isolated in highly purified form, the dry weight of the infectious particle has usually been determined. If the dry density is known the particle size can be calculated from the relation Particle weight = 47rRV3 X density where R is the radius of a sphere of mass and density equal to those of the dried phage particle. Sizes calculated in this way from available data are summarized in Table III, and may be compared with particle sizes of the same viruses obtained by other methods and summarized in Table II. As might be expected particle size estimated in this way is somewhat larger than the size obtained by other methods. Any impurity in the phage preparation would increase the dry weight per infective particle, and any inactivation of phage during purification, or an efficiency of plating less than unity, would have the same ef- SIZE AND MORPHOLOGY OF BACTERIOPHAGES 47 TABLE III Sizes of Phage Particles Estimated from Particle Weights Dry particle Particle weight diameter," Phage X 10-16 g. mfi Reference T2 10 110 Hook, Beard, Taylor, Sharp, and Beard (1946) T2 5 86 Herriott and Barlow (1952) T6 7.5 100 Putnam (1954) WLL 4.3 83 Schlesinger (1934) T7 4.0 80 Putnam (1954) ; Kerby, Gowdy, Dillon, Dillon, Csaky, Sharp, and Beard (1949) otapn. muscae 18 134 Northrop (1938) "■ Calculated for a measured or assumed density of 1.5 grams per cubic cm. feet. However, the purity of the best preparations is excellent, and at least half the particles form plaques (Luria, Williams, and Backus, 1951). Therefore the particle weight of T2 given by Herriott and Balrow (1952) is overestimated by a factor less than two, and the particle diameter by less than 26 per cent. 9. Summary The well-studied phage particles are sperm-shaped structures containing, in the head portion, mostly nucleic acid surrounded by a protein membrane, often hexagonal in cross section, to which the tail is attached. From some phages, the membrane and tail can be isolated as a functionally intact "ghost" following osmotic shock. Different phages range in size from approxi- mately 0.1 micron in diameter to a poorly defined minimum of perhaps 20 mju. They also show characteristic differences in shape and structure. Sizes of these particles can be measured in three more or less satisfactory ways. Electron microscopy is the most informative. It probably underestimates size in general, because of unavoid- 48 BACTERIOPHAGES able shrinkage on drying. Analytical centrifugation, supple- mented by measurements of diffusion, hydration, and wet density, also yields precise information. Weight measurements of puri- fied preparations, related directly or indirectly to electron- microscopic counts of particle number, together with independ- ent estimates of chemical purity, are now feasible. All the available methods have been applied only to T2. For this phage the particle "diameter" in m/x is 65 to 95 in electron micrographs, 75 by sedimentation and diffusion, and 86 to 110 computed from the weights per infective particle, all referring to dried particles. These estimates are consistent with an equiva- lent spherical diameter of about 90 irifx and other characteristics of the natural particles as determined by hydrodynamic methods. In addition, very simple methods, such as ultrafiltration or measurements of sensitivity to X-rays or beta rays, suffice to classify one phage in relation to others. These methods are available in instances where the other methods fail. CHAPTER V EFFECTS OF PHYSICAL AND CHEMICAL AGENTS (EXCEPT RADIATIONS) ON BACTERIOPHAGE PARTICLES The commonly studied bacteriophages are extraordinarily stable under appropriate conditions. The infective titer may fall at the rate of one per cent per day, or often much less. Needless to say, the stability depends on many environmental factors, some of which act differently on different phages. A knowledge of these factors is of immediate concern to anyone interested in the concentration, purification, or preservation of phage. Phages are generally stable in their own lysates provided these are free from specific inactivating agents (receptor substances) derived from the lysed bacteria, and contain suitable electro- lytes. The absence of general poisons is more or less guaranteed by the fact that the phage grew in the first place. Purified phages are generally stable in neutral, buffered solutions, pre- pared from glass-distilled water, containing 0.1 M sodium chloride, 0.001 M magnesium chloride, and, if the phage con- centration is low, a small amount of gelatin. The effects of receptor substances, if these are not completely removed, can be minimized by choosing conditions unfavorable for specific interaction, but these conditions vary greatly from phage to phage. Low temperatures, of course, favor stability, especially if other conditions are suboptimal. In addition to these purely practical considerations, the study of inactivation of phages under controlled conditions has yielded important information about the structure and function of phage particles. Much of the recent systematic work has dealt 49 50 BACTERIOPHAGES with the effects of radiations, which will be considered separately in Chapter VI. This chapter describes the effects of other physical and chemical agents on phage particles. Some of the agents that have been studied tend to be phage-specific in their action, and can be used as aids to the classification of phage types (Burnet, 1 933e) . This application is discussed in Chapter XXII. 1. Effect of pH Bacteriophages are usually stable over the pH range of 5 to 8. At low temperatures the range can often be extended to pH 4 and pH 9 or 10. T2 can be precipitated at pH 4 without loss of infectivity (Herriott and Barlow, 1952). Curves showing stability as a function of pH are given for purified preparations of coliphage T2 by Sharp, Hook, Taylor, Beard, and Beard (1946), of coliphage T6 by Putnam, Kozloff, and Neil (1949), and of coliphage T7 by Kerby, Gowdy, Dillon, Dillon, Csaky, Sharp, and Beard (1 949) . Some kinetic data on phages CI 6 and SI 3 are given by Wahl and Blum-Emerique (1947) over the pH range between 3 and 8. 2. Urea and Urethane Both urea and urethane denature proteins and inactivate enzymes. They also inactivate viruses. Burnet (1933e) in his studies on the classification of the dysentery-coli phages used susceptibility to inactivation by concentrated solutions of urea as a differential characteristic. He found large differences in rate of inactivation from one strain to another. Sato (1956) studied the action of urea on phage T4 and observed minimal in- activating effects at temperatures near 15°. Urea has been used also to extract nucleic acid from phage T2 (Cohen, 1947a). The effect of urethane on the thermal inactivation of coliphage T5 was investigated by Foster, Johnson, and Miller (1949). The thermal inactivation rate was increased by urethane at all concentrations above 0.05 M, the first-order velocity constant increasing as the 2.3 power of the urethane concentration. The EFFECT OF PHYSICAL AND CHEMICAL AGENTS 51 Arrhenius constant (see Glossary) was somewhat increased by the presence of urethane. The authors suggest that there may be more than one urethane-catalyzed reaction involved in phage inactivation. 3. Detergents Because of the marked bactericidal effects of soaps and other detergents, these agents were tested quite early for possible virucidal effects. Vaccinia virus and influenza virus are rapidly inactivated by detergents while poliomyelitis virus is resistant. The earlier literature was reviewed by Stock and Francis (1940) in connection with studies on the effects of soaps on influenza virus. These authors found that influenza virus was rapidly inactivated by oleic, linoleic, and linolenic acid soaps at pH 7.5 without impairment of antigenicity. Burnet and Lush (1940) tested dodecyl sodium sulfate, sodium deoxycholate, and saponin on a number of animal viruses and also on phages CI 6, D6, D44, two salmonella phages and two staphylococcus phages. All of the phages were resistant to inactivation by the three detergents although many of the animal viruses were rapidly inactivated. Klein, Kalter, and Mudd (1945) tested a number of cationic and anionic detergents on vaccinia virus, phage T2, a shigella phage, and a staphylococcus phage. Excepting phage T2 which was resistant to all the detergents tested, all the viruses were susceptible to inactivation by some of the detergents. Putnam, Miller, Palm, and Evans (1952) noted that phage T7 but not T6 was rapidly inactivated by the cationic detergent dodecyltrimethylammonium chloride. However, phage T6 is inactivated by A//1,000 benzyldodecyldimethylammonium chloride and less effectively by cetyltrimethylammonium bro- mide and dodecyl sodium sulfate (Putman, Kozloff, and Neil, 1949). Certain detergents cause the separation of protein from nu- cleic acid of tobacco mosiac virus, permitting the reconstitution of virus particles from the separate components (Fraenkel-Conrat, 1956). Dodecyl sodium sulfate has been used in a similar 52 BACTERIOPHAGES manner for the isolation of nucleic acid from T2 (Mayers and Spizizen, 1954). Bacteria are generally more sensitive to detergents than are bacteriophages, suggesting the use of detergents to rid ph age- containing materials of bacteria (Kalter, Mordaunt, and Chap- man, 1946). Detergents presumably act by denaturing proteins and cell membranes. Their properties are summarized in a symposium (Anson, 1946). 4. Chelating Agents Chelating agents are chemicals that form undissociated complexes with metal ions. Lark and Adams (1953) found that citrate, ethylenediaminetetraacetate and -triphosphate ac- celerated the inactivation of phage T5 at low salt concentrations. These substances may act by forming a complex with some cation bound to the phage particle. Stocks of phage T5 are heterogeneous with respect to susceptibility to the action of chelating agents, since inactivation curves are not exponential with respect to time. 5. Mustard Gas Mustard gas and nitrogen mustard are of biological interest not only because they inactivate enzymes and kill viruses and microorganisms but also because of their mutagenic activity. Mustard gas was tested on a variety of microorganisms, viruses, enzymes, and on the pneumococcus transforming principle by Herriott (1948). Among the test objects were coliphage T2 and a staphylococcus phage. The phages were inactivated in ac- cordance with first-order kinetics, the velocity constants being of the same order of magnitude as those found for microorganisms but larger than those for enzymes. Luria and Dulbecco (1949) reported that nitrogen mustard inactivated T2 and T6. An analysis of the effect of mustards on phages might help to eluci- date the nature of the so-called radiomimetic action of mustards. Herriott and Price (1948) found that mustard -killed bacterial EFFECT OF PHYSICAL AND CHEMICAL AGENTS 53 cells Still retain their ability to serve as hosts for phage multi- plication, another resemblance to the action of radiations. 6. Alcohols Cold aqueous solutions of glycerine and ethyl alcohol do not inactivate most phages, and in fact cold 30 per cent alcohol at pH 5.4 has been used to precipitate phage T6 as one step in the concentration and purification of this virus (Putnam, KozlofF, and Neil, 1949). Undiluted glycerine and ethanol cause rapid inactivation of phages (d'Herelle, 1926). Bronfenbrenner (1925) studied the effect of salts on the stability of phages during alcohol and acetone precipitation and concluded that suitable mixtures of monovalent and divalent cations were preferable to pure salts in stabilizing the phage. Wahl and Blum-Emerique (1949a) used alcohol precipitation to purify phage CI 6, and Hotchin (1954) used acetone to precipitate phage K. 7. Other Chemicals The enzyme poisons, cyanide and fluoride, do not damage phages in concentrations that are lethal to bacteria. Cyanide has been used to stop phage development without inactivating phage particles at desired times in the latent period (Doermann, 1952). Thymol and chloroform are bactericidal but do not inactivate phages. Any of these substances can be used to preserve phage lysates (Wahl and Blum-Emerique, 1949b; Fredericq, 1952a, Sechaud and Kellenberger, 1956). Formaldehyde inactivates bacteriophages, but Schultz and Gebhardt (1935) reported that formalin inactivated staphylococ- cus phage could be reactivated by dilution. Labaw, Mosely, and Wyckoff (1949) noticed a large difference between coliphage Tl and phages T2 and T4 in susceptibility to inactivation by formaldehyde. There are many observations to the effect that mercuric ion inactivates phages. Krueger and Baldwin (1934) found that the rate of inactivation of staphylococcus phage K by HgCl2 fol- 54 BACTERIOPHAGES lowed first-order kinetics to 99 per cent inactivation and then slowed. With 2.8 per cent HgCl2 a phage stock titering 5 X 10^ per ml. could be completely inactivated in a few days. Treatment of such a stock with hydrogen sulfide followed by centrifugation resulted in complete reactivation. Essentially similar results were obtained by Wahl (1939, 1946c) with six different phages, which differed, however, in rate of inactivation. Oxidizing agents such as peroxide, halogens, ozone, and permanganate rapidly inactivate viruses. In contrast mild re- ducing agents have no harmful effect and Wahl and Blum- Emerique (1946) have suggested the addition of hydrosulfite to protect phage preparations from oxidation during storage. Contrary to general expectation, however, Alper (1954) sug- gests that phage SI 3 is inactivated by the reducing action of hydrogen peroxide. In support of this he finds that the in- activation by peroxide is accelerated by exclusion of oxygen, that mild oxidizing agents like chlorate and iodate do not in- activate the phage, and that ascorbic acid does. The question calls for systematic work. 8. Sonic Vibration High intensity sonic vibrations imparted to a fluid medium by a vibrating crystal or diaphragm can denature proteins and destroy bacteria. They have been used to liberate enzymes from bacteria by rupture of the bacterial cell walls (Shropshire, 1947). The effect of intense sonic vibration on the seven coli- phages of the T group and on their common host, strain B of E. coll, was studied by T. F. Anderson, Boggs, and Winters (1948) using the Raytheon magneto-striction apparatus oper- ating at about 9,000 cycles per second. The inactivation fol- lowed the kinetics of a first order reaction and there was a large variation in sensitivity among the phage strains tested. The small particle phages Tl, T3, and T7 were relatively resistant (99 per cent inactivation in 60 minutes) whereas the large particle phages T2, T4, T5, and T6 were sensitive (99 per cent inactivation in about 5 minutes). The host bacteria were EFFECT OF PHYSICAL AND CHEMICAL AGENTS 55 somewhat more resistant than the large phages but much more sensitive than the small phages. T. F. Anderson and Doermann (1952a) applied this information in studies of the intracellular development of phage T3 (Chapter XI). Anderson and Doermann (1952b) also tested the effect of sonic vibration on phage that had been inactivated w^ith anti- body. Phage T3 was treated with anti-T3 serum to a phage survival of 10~^. After dilution, treatment of the neutralized phage with sonic vibrations resulted in a 40-fold increase in phage titer, presumably due to breaking of the bonds linking antibody molecules to phage particles. 9. Surface Denaturation Proteins are rapidly denatured when subjected to the un- balanced forces existing at gas-liquid or liquid-liquid interfaces, and physiologically active proteins such as enzymes and toxins are thus inactivated. The first quantitative study of this phenomenon with phages was made by Campbell-Renton (1942). She found that all the phages tested were rather rapidly in- activated when aqueous solutions were vigorously shaken in air. In one experiment for which detailed data are given the in- activation was first order to a survival of less than lO""*, except for an initial lag of 25 minutes due to the presence of 100 /xg. per ml. of peptone in the medium. The inactivation could be greatly retarded by increasing the concentration of peptone, and ac- celerated by omitting peptone. The same phenomenon was studied by Adams (1948) with the seven coliphages of the T group. Surface inactivation occurred very rapidly in a chemically defined diluent at the gas-liquid interfaces formed by bubbling or shaking with air or nitrogen. Under conditions of vigorous shaking the inactivation kinetics were first order with a low temperature coefficient, as might be expected of a reaction in which the rate-limiting step is diffusion of the virus particles into the region of the interface. The re- action rate was nearly independent of pH over most of the stability range of the phages but was greatly accelerated near the 56 BACTERIOPHAGES acid and alkaline ends of this range. The first-order velocity constants differed from one phage type to another, ranging from 0.05 to 1.2 per minute. The velocity constant calculated from the data of Campbell-Renton is 0.06 per minute. Surface inactivation can be completely prevented by adding enough protein to the diluent to saturate the gas-liquid interface and prevent access of the virus to the surface. For this purpose 10 to 100 ixg. of gelatin per ml. of solution are adequate. Sur- face inactivation must be considered whenever virus prepa- rations are diluted in buffer solutions free of protein or peptones, especially when surface-volume ratios become large. Similar losses can be expected at solid-liquid interfaces, and call for similar precautions. The effect of denaturation at liquid-liquid interfaces has not been studied systematically. Garen (personal communication) isolated DNA from phage T2 by shaking aqueous suspensions with chloroform. Hotchin (1954), on the other hand, reported no inactivation of phage K on shaking with isobutanol-chloro- form mixtures; he used this method as a step in purification of the phage. From what has been said above the effects should depend very much on total concentration of phage and other proteins. 10. Heat Inactivation D'Herelle (1926) noted that several phages were inactivated by heating at 75 ° C. for 30 minutes, whereas some survived and some did not after heating at 70 ° C. Observations of this type led to the notion of a qualitatively defined inactivation tempera- ture, characteristic for each phage, analogous to the thermal death point used to report the results of similar tests with bac- teria. Such qualitative information sufficed for practical purposes, and led to the use of heat treatment in lieu of filtration to eliminate bacteria from phage preparations. Quantitative study of the effects of heating showed that phages are inactivated in accordance with first-order kinetics. The temperature coefficient of inactivation is very high, the Ar- EFFECT OF PHYSICAL AND CHEMICAL AGENTS 57 rhenius constant being of the order of 100 kilocalories (Krueger, 1932). The only known reaction with such a high coefficient at laboratory temperatures is the heat denaturation of proteins. Presumably heat inactivation of phages is the result of denatura- tion of this kind. Some data from the literature on Arrhenius constants and the temperature range tested are given in Table IV. TABLE IV Arrhenius Constants for Heat Inactivation of Phages Activa- Temper- tion ature energy, range, Phage cal. °C. Reference In broth Staph K 101,000 51-62 Krueger (1932) Coli T5 170,000 above 65 Foster, Johnson, and Miller (1949) Coli T5 5,000 below 65 Foster, Johnson, and Miller (1949) Coli T5 86,000 63-73 Adams (1949a) Coli T5st 100,000 65-75 Lark and Adams (1953) Coli Tl 106,000 65-70 Adams (1949a) Coli Tl 73,600 60-70 Pollard and Forro (1949) CoU Tl 95,000 60-75 Pollard and Reaume (1951) Coli T7 77,000 55-60 Adams (1949a) Coli T7 60,700 45-60 Pollard and Reaume (1951) Coli T3 105,000 47-60 Pollard and Reaume (1951) Coli T4 131,000 65-70 Adams (1949a) Coli T2 71,700 60-74 Pollard and Reaume (1951) Staph phage 137,000 above 65 Chang, Willner, and Tegarden (1950) Staph phage 14,000 below 65 Chang, Willner, and Tegarden (1950) In dry state Coli Tl 27,500 95-145 Pollard and Reaume (1951) Coli T3 19,100 20-47 Pollard and Reaume (1951) Coli T7 12,700 25-60 Pollard and Reaume (1951) From the table it may be noted that two values for the Ar- rhenius constant are given for phage T5 by Foster, Johnson and Miller (1949) and for a staphylococcus phage by Chang, Willner, 58 BACTERIOPHAGES and Tegarden (1950). In both cases the value below 65° C. is exceedingly low, 5,000 and 14,000 calories. These low values are probably due to chemical inactivation of the phages by some substance in the broth diluent, which becomes inappreciably slow in comparison with thermal inactivation above 65 ° C. With these two exceptions, all Arrhenius constants recorded for wet phage are above 70,000 calories. The inactivation constants for dry phages listed in Table IV are difficult to interpret because drying itself inactivates most phages and subsequent heating may merely continue the drying process. Phage Tl is exceptionally resistant to drying, however, and the dried phage is notably resistant to heat. The low Ar- rhenius constant in this instance may, therefore, reflect the ef- fect of drying on thermal denaturation. These results suggest further experiments, as discussed by the authors (Pollard and Reaume, 1951). It is obvious that different determinations of the Arrhenius constant for the same phage do not always agree. A signifi- cant clue to these discrepancies is found in the paper of Foster, Johnson, and Miller (1949). These authors failed to obtain a first-order inactivation rate for phage T5 in broth. However, the addition of 0.01 A^ MgClg to the broth resulted in typical first- order curves with a velocity constant about half that found in the absence of added magnesium salt. The addition of phosphate accelerated the inactivation. Since broth is not a chemically defined medium it is not likely to give reproducible results from one lot to another. In chemically defined media the heat susceptibility of phages depends very markedly on the chemical composition of the medium. Nanavutty (1930) found a coliphage to be inactivated 10 times faster in saline than in broth. Burnet and McKie (1930) found several phages to be far more susceptible to heat inactivation in 0.1 N solutions of sodium or potassium salts than in broth. The addition of magnesium or calcium salts made the salt solutions equivalent to broth as far as phage stability was concerned. Similar results were obtained by Gratia (1940). EFFECT OF PHYSICAL AND CHEMICAL AGENTS 59 The coliphage T5 belongs to the same class, and its response to heating has been studied in detail by Lark and Adams (1953). T5 is maximally resistant to heat at cation concentrations exceeding Af/ 1,000 for magnesium or M/X for sodium, and maximally sensitive at concentrations less than M/10^ for magne- sium or M/10 for sodium. The span of inactivation rates at 50 ° C. (measured indirectly) between these two extremes is about 1 X 10'''. These and other characteristics of the dependence on cations presumably indicate complex formation between cations and some heat-labile phage protein. In its stable form, the complex resists heating to 70 ° C. ; in its least stable form the phage is rapidly inactivated at 30 ° C, or even below in the presence of chelating agents. Inactivation of the phage by heat is accom- panied by release of nucleic acid into the solution. The result- ing nucleic acid-free ghosts do not adsorb to bacteria. Further details of the physical and biological characteristics of heat in- activation are given by Lark and Adams (1 953). Coliphage T5 mutates to a form, T5st, that is about 1,000 times more heat resistant in 0.1 iVNaCl than is the parental wild type (Lark and Adams, 1953). The two forms do not differ in heat sensitivity at high salt concentrations. They differ, there- fore, in the manner of interaction with cations. The mutant occurs with a frequency of about 1 0~^ in wild type stocks initiated from single plaques, but on repeated subculture accumulates until it may amount to 10 per cent of the population. Wild type stocks of phage T5 also contain a phenotypically heat-resist- ant form composing about 0.1 per cent of the population. The phenotypically heat-resistant particles have the same stability as the heat-resistant mutants but differ in that the property of heat resistance is lost on a single growth cycle in the host bac- terium. Similar phenotypically heat-resistant particles have been found in stocks of phages BG3, 29 alpha, and PB which are serologically related to phage T5 (Adams, 1953a). No satis- factory explanation for the origin of these phenotypically heat- resistant particles is available at present. Another example of nonheritable heat resistance was observed 60 BACTERIOPHAGES in a coliphage by Fischer (1950). The heat resistance increased on rapid subcuhure and decreased on storage at room tempera- ture. It may be related to the cofactor independence of nascent phage T4 (E.-L. Wollman and Stent, 1952). 11. Hydrostatic Pressure The effect of high pressure on the heat stability of phages was investigated by Foster, Johnson, and Miller (1949). Pressures of 10,000 pounds per square inch were found to increase the stability of phages Tl , T2, and T5, but to accelerate the inactivation of phage T7. The change of velocity constant for inactivation of phage T5 with changing pressure indicated that thermal inacti- vation of this phage was characterized by a volume increase of activation of 113 cubic centimeters per mole. This is the type of effect usually found in protein denaturation. 12. Osmotic Shock The production of ghosts from phages T2, T4, T6, and K by osmotic shock, first described by T. F. Anderson, has already been noted in Chapter IV. The phenomenon shows that phages are rapidly penetrated by simple electrolytes, glycerol, sometimes by sucrose, and even more rapidly by water. Sudden reduction of salt concentration presumably causes water to enter the particle until the membrane ruptures, permitting the internal contents to leak out more or less completely. Gradual changes in salt concentration do not have this effect (T. F. Anderson, Rappaport, and Muscatine, 1953). Purely mechanical considerations suggest that the principal damage should occur to the membrane of the head and this seems to be so, since the tail structure of the ghost remains morphologically and functionally intact (Herriott, 1951a; Kellenberger and Arber, 1955). The same considerations probably explain in part the resistance of the smaller phages Tl, T3, T5, and T7 to osmotic shock (Anderson, 1949). Other EFFECT OF PHYSICAL AND CHEMICAL AGENTS 61 factors are probably involved as well, however. Anderson, Rappaport, and Muscatine (1953) describe variants of T6 differently sensitive to osmotic shock, and an unexpected tem- perature-dependence in some of them. The discovery of osmotic shock was of the greatest importance in contributing to an understanding of the structure (Chapter IV) and function (Chapter XIV) of phage particles. The properties of the ghosts produced by osmotic shock of T2 have been studied in some detail. All observers agree that they kill bacteria with a low efficiency varying between 10 and 50 per cent. That ghosts can adsorb to bacteria without killing them is especially clear from the microscopic observation of Bon if as and Kellenberger (1955) and the metabolic studies of French and Siminovitch (1955), which show that the surviving bacteria are damaged. Multiplication and protein synthesis are interrupted for about 80 minutes, and ability to support growth of superinfecting phage is temporarily lost. Since in- fection with live T2 does not interrupt protein synthesis, it ap- pears that ghosts and phage damage the cell in different ways. For this reason it is incorrect to say that the killing power of T2 resides in its protein coat ; the death of the cell penetrated by a ghost could be due, for example, to leakage of material through fissures in the ghost. This consideration, as well as the observed low efficiency of killing by ghosts, does not support the idea some- times expressed that phage T2 belongs to a necessarily virulent species. 13. Summary Substances or conditions that denature proteins or react chemi- cally with proteins or nucleic acids inactivate phages. The effects tend to be phage-specific, and sometimes help in classi- fication of phages. Substances that react reversibly such as formaldehyde and mercuric ions cause reversible inactivation under appropriate conditions. Specific enzyme poisons such as cyanide, fluoride, and dinitrophenol have no effect on phage 62 BACTERIOPHAGES particles. These substances, as well as thymol and chloroform, can be used to kill or inhibit bacteria and molds in the prepa- ration and preservation of phage stocks. Several protein denaturants, as well as osmotic shock, can be used to liberate nucleic acid from phage particles. Osmotic shock also permits isolation of the proteinaceous ghost of certain phages in relatively undamaged form. CHAPTER VI EFFECTS OF RADIATIONS ON PHAGE PARTICLES The discussion of the action of various physical and chemical agents on the viability of bacteriophages will now be continued by a consideration of the effects produced in phage particles which have been exposed to various sorts of radiations. The irradiation of bacteriophages has been a very useful tool in the elucidation of their structure and physiology, and it may be said that with bacteriophages radiobiological methods have found one of their most profitable applications. Diverse types of bacterio- phages have at some time or other been subjected to radiations of almost all parts of the electromagnetic spectrum, ranging in wavelength from the infrared (3,000,000 to 7,600 A) through the visible (7,600 to 4,000 A) and ultraviolet (4,000 to 130 A) to the X-ray regions (100 to 0.1 A), and extending even to the most highly energetic radiations of extremely short wavelength emitted by radioactive substances and those produced in the ac- celerators of atomic physics. The body of literature concerned with this subject is already so large that not all of the results can be discussed here. The reader may find further information in a number of specialized reviews (Lea, 1946; Bowen, 1953; Luria, 1955; Pollard, 1954; Kleczkowski, 1957; Stent, 1958). 1. Visible Light Wahl and Guelin (1942) observed that the small phage SI 3 is particularly sensitive to inactivation by light, being killed by radiations from the visible part of the spectrum at a much greater rate than the large phage CI 6. Wahl (1946b) found by use of appropriate filters that the wavelength of the active 63 64 BACTERIOPHAGES radiations is in the vicinity of 4,500 A, or in the blue region. He also observed that, like SI 3, a large subtilis phage is sensi- tive, while the phages coli 36 and streptococcus B563, like CI 6, are resistant to inactivation by visible light. This differential action of visible radiations is presumably due to some difference in the chemical composition of these phage strains, in that only the light-sensitive phage types contain some pigment which absorbs blue light. Wahl and Latarjet (1947) examined the inacti- vation spectrum of SI 3 more critically and found that light of wavelengths greater than 5,550 A is ineffective, the efficiency of inactivation increasing with decreasing wavelength down through 3,650 A. Throughout this part of the spectrum phage SI 3 is more radiosensitive than phage CI 6. Between 3,130 and 3,650 A, an inversion of the relative sensitivities of these two phage strains occurs, so that at shorter wave lengths in the ul- traviolet CI 6 becomes more radiosensitive than SI 3. The ki- netics of inactivation of SI 3 by light at both 3,650 and 4,500 A are complex in that the first 75 per cent of the phage particles are killed according to an exponential survival curve (see Glossary) of greater slope than the rate of inactivation of the remaining 25 per cent. A change of temperature from 17 to 37° C. does not change the rate of inactivation of phage SI 3 at 4,500 A, indicating that the lethal action results from a direct photochemical effect. The effectiveness of the incident energy in the visible region at 4,500 A is only 10~'* of that in the ultraviolet region at 2,537 A, although no data are available for such a comparison in terms of the absorbed energy. The dif- ferential sensitivity of phages to visible light may be an im- portant clue to their chemJcal structure, although no informa- tion appears to be as yet at hand concerning the nature of the material which absorbs the blue light. The photodynamic action of dyes on a bacteriophage was inves- tigated by Clifton (1931) who observed that addition of 0.01 per cent methylene blue to a suspension of a staphylococcus phage has no effect on the viability of the phage particles as long as the suspension is kept in the dark but results in complete inactivation EFFECT OF RADIATIONS ON PHAGE PARTICLES 65 within 5 minutes after the suspension is placed in light. The presence of 0.01 per cent cysteine gives complete protection against such photodynamic action, for which also oxygen ap- pears to be essential, since no inactivation takes place in the light in a nitrogen atmosphere. Similar results were reported by Perdrau and Todd (1933). Burnet (1933e) investigated the photodynamic action of methylene blue on a series of coli-dys- entery phages. He found marked differences in sensitivity, differences which were correlated not with particle size but rather with serological classification. Hence photodynamic sensitivity, like sensitivity to visible light, is of taxonomic signifi- cance, although here too the chemical basis for such differential behavior is as yet unknown. Krueger Scribner, and Mc- cracken (1940) reported the photodynamic inactivation of a "phage precursor." In these experiments, a reduced yield of phage was encountered after illumination prior to infection of the host bacteria in the presence of 10~^ M methylene blue, a treatment which was said not to reduce the bacterial colony count. Since the assays in this experiment were made by Krueger's kinetic method, it is difficult to interpret such results. Illumination of bacteria in the absence of methylene blue sup- presses their capacity to produce phage without killing them (Dulbecco and Weigle, 1952; Latarjet and Miletic, 1953; Hill, 1956). Yamamoto (1956) studied the relation between structure of dyes and photodynamic inactivation of the T series of coliphages, and observed competitive interactions between effective and in- effective dyes. Welsh and Adams (1954) found that photo- dynamic action on T2 destroys infectivity three times faster than it destroys the host-killing property of the particles. No photo- reactivation or multiplicity reactivation could be demonstrated. 2. Ultraviolet Light Of all the agents capable of inactivating bacteriophages, ul- traviolet light has been studied most extensively and has shown the most diverse and interesting effects. Besides simply in- 66 BACTERIOPHAGES activating phage particles, ultraviolet light produces important physiological and genetic effects, such as inducing phage de- velopment in lysogenic bacteria, causing a growth-delay in the phage particles surviving irradiation, stimulating genetic re- combination, and exerting a mutagenic action. Some of the effects of ultraviolet light, furthermore, are reversible under ap- propriate conditions. Ultraviolet light has also been used to study the course of intracellular phage development. The pos- sibilities of ultraviolet light as a tool in phage research appear to be far from exhausted. a. Lethal Effect Repeated observations of the lethal effect of ultraviolet light on bacteriophages have been made ever since 1922 (see d'- Herelle, 1926). The first quantitative study appears to have been carried out by Baker and Nanavutty (1929). These authors found that the inactivation of a shiga phage is an exponential function of the dose of ultraviolet light to a survival of 10"\ By use of filters it was possible to determine that wavelengths above 3,000 A are harmless and that the wavelength of maximum effectiveness is somewhere below 2,650 A. Gates (1934), using a monochromator to isolate single emission lines, studied the inactivation of a staphylococcus phage and found that the kinetics of inactivation follow a simple exponential at all wave- lengths studied from 2,300 to 2,970 A. The energy output of the radiation source was determined at each wave length by means of a thermopile, thus permitting a determination of the action spectrum for the ultraviolet inactivation of this phage. It was found that the incident energy required to inactivate a given fraction of the phage population decreases continuously between 2,950 and 2,600 A, then increases to a peak at 2,400 A and finally decreases again toward 2,300 A. Northrop (1938) determined the ultraviolet absorption spectrum of purified staphylococcus phage and found it to be very similar in shape to Gates' action spectrum. Analogous investigations carried out subsequently to determine the action spectra of ultraviolet inactivation of Tl, T2, EFFECT OF RADIATIONS ON PHAGE PARTICLES 67 and the megaterium phage M5 (Fluke and Pollard, 1949; Zelle and Hollaender, 1954; Franklin, Friedman, and Sedow, 1953) and of the ultraviolet absorption spectra of suspensions of purified T6 (Putnam, KozlofT, and Neil, 1949), T2 (Cohen and Arbogast, 1950b), and T7 (Putnam, Miller, Palm, and Evans, 1952), showed that in all these cases both action and absorption spectra are very similar to the typical absorption spectrum of purified nucleic acids, which displays a maximum absorption near the wavelength of 2,600 A. It seems probable, therefore, that the inactivating photons of ultraviolet light are absorbed by the phage nucleic acid. The quantum yield, i.e., the number of phage particles inactivated per quantum absorbed, has been estimated for Tl and T2 at various wavelengths by Zelle and Hollaender (1954). These authors found that the yield for either phage is only about 3 X lO^^ throughout the range of wavelengths from 2,200 to 3,000 A. Campbell-Renton (1937) compared the efl^ects of ultraviolet irradiation on five strains of phage and found the inactivation to be an exponential function of dose. The ultraviolet sensitivity of these strains, measured as the slope of the exponential survival curve, varied over a ten-fold range. Latarjet and Wahl (1945) compared the ultraviolet inactivation of phages CI 6 and SI 3 and found that while the survival curves for both types are exponen- tial, phage SI 3 has only ^/g the ultraviolet sensitivity of phage CI 6, although SI 3 contains only about V30 as much nucleic acid as CI 6 (Sinsheimer, 1957). The nucleic acid of SI 3, therefore, appears to be intrinsically more sensitive to ultraviolet light than that of CI 6. It is possible that the same difference in chemical composition which is responsible for the very different sensi- tivities of these two phage strains to visible light is also respon- sible for the greater intrinsic ultraviolet sensitivity of SI 3. In this connection it may be noted that although T4 is indistin- guishable from T2 and T6 in size and morphology, it has only one half the ultraviolet light sensitivity of its two relatives. Streisinger (1956a) traced the difference in ultraviolet sensi- tivity of these three strains to a single genetic locus. Phage 68 BACTERIOPHAGES lambda is inactivated by ultraviolet light at only V13 the rate of phage T5, although these two phages have almost the same size (Weigle and Delbriick, 1951). As we will see later, there is much evidence that different strains of phages react very dif- ferently towards ultraviolet light. Latarjet and Morenne (1951) found that phage T2 is in- activated slowly when irradiated by Mazda fluorescent lamps of the daylight type. Examination of the radiations emitted by these lamps showed that appreciable amounts of radiant energy emanate at wavelengths from 3,130 A down to 2,890 A, and calculations indicate that these radiations can account for the observed rate of phage inactivation. The kinetics of inacti- vation by fluorescent lamps are not strictly exponential but follow a "3-hit" curve, which suggests that the mechanism of ultraviolet inactivation of these phages may be more complex than had been imagined previously. A downward concavity of the exponential survival curve has also been noted at low doses of ultraviolet light for phages T2, T4, T5, and T6, but the sig- nificance of these deviations from first-order kinetics is not known. In contrast, the ultraviolet light survival curves of phages Tl, T3, and T7 show an upward concavity at low survival values, as if a fraction of the phage is inactivated at a slower rate than the bulk of the population (Dulbecco, 1950). This break in the survival curve of such phages, as well as the low intrinsic ul- traviolet sensitivity of the salmonella phage P22, led Garen and Zinder (1955) to propose that some of the particles in ultraviolet- irradiated populations of such phage types are reactivated by homologous genetic material present in the nuclear structures of the bacterial host cell. A comprehensive discussion of this hypothesis is given by Stent (1958). b. Growth- Delaying Effect Luria (1944) noted that the latent period of bacteriophage particles surviving moderate doses of ultraviolet light is con- siderably increased. This observation indicates that phage particles can be damaged by the action of absorbed ultraviolet EFFECT OF RADIATIONS ON PHAGE PARTICLES 69 quanta without being killed, and hence that ultraviolet light can produce more than one kind of lesion. The growth delay in- creases with increasing dosage of ultraviolet light and is not hereditary, in the sense that progeny of affected phage particles reproduce themselves with normal latent period. The magni- tude of the effect varies from one phage type to another, the growth delay for a given dose of ultraviolet being much greater with phages Tl and T7 than with T2. Setlow, Robbins, and Pollard (1955) determined the ultraviolet action spectrum of the growth delaying effect in Tl and found that wavelengths near 2,600 A, the absorption peak of nucleic acids, have the maximum effect. This suggests that extension of latent period, like in- activation, results from the absorption of radiation by nucleic acid. c. Host Killing Luria and Delbriick (1942) discovered that ultraviolet in- activated T2 phage particles, though no longer able to reproduce themselves, are still capable of killing phage-susceptible bacterial cells. By determining the proportion of surviving bacteria after infection with various amounts of ultraviolet-inactivated phage particles, these authors demonstrated that the adsorption of a single irradiated phage particle suffices to kill the host cell. A dose of ultraviolet light that reduces the number of plaque- forming particles to lO"'' of their initial number leaves Va of the original phages still able to kill susceptible bacteria. The host- killing property of phage T2, therefore, is enormously more resistant to ultraviolet light than the viability of the phage. Heat, in contrast to ultraviolet light, destroys both of these activities at the same rate. Microscopic examination of bac- teria infected with ultraviolet inactivated T2 particles shows that such cells do not multiply and do not lyse, Cohen and Arbogast (1950c) demonstrated that infection of bacteria with heavily irradiated T2 phages arrests synthesis of both ribonucleic acid and deoxyribonucleic acid. Cytological methods were employed by Luria and Human 70 BACTERIOPHAGES (1950) to study the effects of phage infection on the chromatinic structures of the bacterial host cells. These authors also ob- served the effects of infection with ultraviolet-inactivated Tl, T2, and T7 phages. It was found that ultraviolet-inactivated phages, like normal unirradiated particles, cause marked al- terations of cell structure. The details of this work will be discussed in a later chapter. Luria and Human also followed the synthesis of material absorbing ultraviolet light at 2,600 A in bacteria infected with normal and with irradiated phages. They found that infection with irradiated Tl or T2 phages causes some increase in optical density at 2,600 A, but at a much slower rate than in uninfected bacteria. It may be concluded, there- fore, that although ultraviolet-inactivated phage particles are unable to reproduce themselves, they can nonetheless cause ex- tensive changes in the infected bacterium, changes which ulti- mately result in the death of the host cell. d. Interference Luria and Delbriick (1942) found that infection of bacteria with ultraviolet-inactivated T2 phages makes it impossible for unirradiated particles of Tl to multiply in these cells. In this respect, ultraviolet-inactivated T2 phages behave exactly as do active T2 particles. In contrast, ultraviolet-inactivated Tl phages do not interfere with the multiplication of active T2 phage. Infection of bacteria with ultraviolet-inactivated phages also results in interference with induced enzyme synthesis under the same conditions under which active phages produce such interference (Luria, 1950). These results demonstrate that ultraviolet-inactivated phage particles still possess some of the physiological properties of active phages. e. Photoreactivation Kelner (1949) discovered while working with ultraviolet- inactivated conidia of Streptomyces grisens that visible light has the remarkable ability of restoring viability in this material. Since this observation, there have been numerous reports of the photo- EFFECT OF RADIATIONS ON PHAGE PARTICLES 71 reversal of radiation lesions in a number of different organisms. Photoreactivation of bacteriophages was discovered independ- ently by Dulbecco (1950), who observed that ultraviolet- inactivated phages can be reactivated by exposure to visible light, but only after their adsorption to susceptible bacteria. Visible light does not reactivate unadsorbed ultraviolet-ir- radiated phages, nor induce phage development in bacteria il- luminated just before infection with such phages. The rate of photoreactivation of adsorbed phage particles is a function of the temperature, increasing by a factor of two in the interval from 25 to 37 ° C. These observations suggested that photoreacti- vation depends on some enzyme system present in the bacterial host cell but absent from the phage particles. Attempts to obtain photoreactivation of free phages in cell-free extracts of bacteria have thus far been unsuccessful, although photoreacti- vation of ultraviolet-inactivated transforming DNA has recently been demonstrated in extracts of E. coli (Goodgal, Rupert, and Herriott, 1957). Photoreactivation of adsorbed phages is not inhibited by cyanide or anaerobiosis. The rate of photoreacti- vation is a linear function of the intensity of the reactivating light at low intensities and approaches a maximum value at high in- tensities, at which point some factor other than absorption of light quanta becomes rate limiting. Light of wavelength near 3,650 A possesses the maximum efficiency of photoreactivation, although the entire wavelength range from 3,000 to 5,000 A is effective. If ultraviolet-inactivated phages are adsorbed to rapidly growing bacteria and the infected cells are incubated for various lengths of time before exposure to photoreactivating light, it is found that the maximum amount of photoreactivation at- tainable decreases rapidly. If such phages are adsorbed to resting bacteria, however, the complexes remain photoreactivable at 37 ° C. for at least 70 minutes. The rate of photoreactivation at a given light intensity is less in resting than in rapidly me- tabolizing bacteria. Only a portion of ultraviolet-inactivated phage particles is photoreactivable. The fraction of the phage population capable 72 BACTERIOPHAGES of forming plaques after maximum photoreactivation decreases as an exponential function of the dose of ultraviolet light, but less rapidly than the fraction which survives in the dark, i. e., in the absence of photoreactivation. If the absorption of ultraviolet light in a given type of phage has a probability a of producing a photoreactivable lethal lesion and probability \-a of producing a nonphotoreactivable lethal lesion, then a is said to be the photoreactivable sector of ultraviolet damage in this phage strain. The ratio of the slope of the dose-survival curve after maximum photoreactivation to the slope of the dose-survival in the dark, or in the absence of photoreactivation, is then evidently 1 -a. The photoreactivable sector a may vary from unity (complete photoreactivability) to zero (no photoreactivability) in different phage strains. Experimental values for the photo- reactivable sectors of the 7 T phages are : Tl = 0.68 ; T2 = 0.56 ; T4 = 0.20; T6 = 0.44; T3 = 0.39; T7 = 0.35; T5 = 0.20. These values show no simple correlation with any of the other properties of the phage concerned ; for instance, the three closely related phages T2, T4, and T6 are seen to differ widely in their photoreactivable sectors. In spite of this, the photoreactivable sector is a remarkably reproducible characteristic of a particular phage strain, although it depends on the physical conditions under which the ultraviolet light has been administered (Hill and Rossi, 1954). The photoreactivable sector also depends somewhat on the type of host cell to which the ultraviolet-in- activated phage particles have been adsorbed (Dulbecco, 1955). By means of experiments in which phage-infected bacteria were exposed to flashes of photoreactivating light rather than to continuous illumination it could be shown by Bowen (1953) that photoreactivation consists of at least two steps. One, a temperature-sensitive dark reaction (i. e., one for which no light is required) appears to precede a second, temperature-insensi- tive light reaction (for which light is required). Bowen pro- posed that the function of the dark reaction is to generate those substances which absorb the visible radiations and thus become EFFECT OF RADIATIONS ON PHAGE PARTICLES 73 "activated" for the reactivation process. Lennox, Luria, and Benzer (1954) carried out experiments in which phage-bacterium complexes were inactivated by ultraviolet light, photoreactivated and then subjected to a second ultraviolet irradiation. They observed that photoreactivated complexes have the same ul- traviolet sensitivity as nonreactivated complexes and inferred from this fact that photoreactivation probably constitutes a direct reversal of the primary ultraviolet damage, rather than a by-pass mechanism. /. Multiplicity Reactivation When assaying phage stocks inactivated by ultraviolet ir- radiation, Delbriick and Bailey noticed that the plaque counts of surviving phage particles appear to depend on the relative pro- portions of phage particles and bacteria in the plating mixture. Luria (1947) investigated this titration anomaly further and discovered the phenomenon of multiplicity reactivation. An ultraviolet-inactivated phage particle, though unable to re- produce itself, is, as we have already seen, far from physiologically inert. It may kill the host cell, interfere with the multiplication of other phages, or even regain its ability to reproduce after exposure to visible light. Now if two or more such ultraviolet- inactivated phage particles are adsorbed to the same host bac- terium, then there exists a good chance that they may cooperate in some way so that this multi-infected cell lyses and liberates viable progeny phage particles. This is multiplicity reactivation, which can be demonstrated readily with phages T2, T4, T6, and T5. It is, however, barely or not at all detectable with Tl, T3, T7, and lambda. Luria formulated a theory of multi- plicity reactivation in terms of genetic exchange of undamaged parts between irradiated phage particles, a theory whose essence, though not its particulars, later work has shown to be indeed the most likely explanation of this phenomenon. The detailed results of studies on multiplicity reactivation, as well as those of an associated phenomenon, cross reactivation, will be considered in Chapter XVIII. 74 BACTERIOPHAGES 3. Ionizing Radiations Most investigators of the effects of ionizing radiations, i.e., radiations belonging to the shortest wavelength sector of the electromagnetic spectrum, have been interested in the interpre- tation of inactivation data in terms of the "target theory." It was hoped that development of the target theory would yield in- formation about the size and shape of the radiation-sensitive structures in viruses (Latarjet, 1 946 ; Lea, 1 946) . Some of the calculated target sizes have already been presented in Chapter IV, and their relation to phage particle sizes determined by other means has been discussed. Recently, an increased em- phasis has also been placed on the study of the physiological effects of ionizing radiations on phage particles, stimulated, no doubt, by the amazing variety of effects produced by ultraviolet (nonionizing) radiation. Extensive discussions of the chemical effects of ionizing radia- tions and possible mechanisms for their biological action can be found in Lea (1946) and in Hollaender (1954). We may state briefly here that the effects of ionizing radiations may be classed under two categories: direct and indirect. Direct effects are produced when an ionization occurs directly in the substance under investigation, whereas indirect effects are caused by chemi- cal agents produced by the action of ionizing radiations on the solvent molecules. Irradiation of pure water produces highly reactive but unstable free radicals, such as OH, H, and OoH, as well as more stable peroxides. The presence of dissolved oxygen increases the toxic and mutagenic effects of ionizing radiations, while the presence of sulfhydryl compounds and certain simple organic substances reduces these effects. The irradiation of dry substances will result in direct effects only, since there is no solvent. Irradiation of dilute solutions will result in both direct and indirect effects, while irradiation of concentrated solutions will result in predominance of direct effects and a relative decrease in indirect effects. The addition of certain kinds of solutes will protect against the indirect effects because the added solute molecules will react preferentially with EFFECT OF RADIATION ON PHAGE PARTICLES 75 the radiation-activated solvent molecules — free radicals and peroxides^in competition with the material under study. Since target volumes are calculated on the assumption that the lethal action has resulted from energy absorbed within the physical domain of the biological structure whose death is being followed, it is important to remember that target-size estimates of bacteriophages can be based only on experiments in which exclusively direct effects come into play. Indirect effects must be excluded by the addition of protective substances. Luria and Exner (1941) found, while studying the inactiva- tion of phages with X-rays, that the inactivation rate in distilled water or buffers is very much greater than the rate in nutrient broth. The addition of gelatin to water, however, gives maxi- mum protection, so that the rate of X-ray inactivation of phages suspended in this medium is no greater than that in broth. Egg albumin and serum albumin also protect against this indirect effect, while sugars do not. Alper (1948) made similar obser- vations of the protective effect of proteins and attributed the indirect effect to the lethal action of hydrogen peroxide. Latar- jet (1942) had considered this possibility but his experiments indicated that not enough peroxide is produced to account for the indirect effect on phage CI 6. It seems to be generally agreed that although some peroxides are produced by ionizing radiations, the major portion of the indirect effect is due to short- lived radicals generated in the water (Alper, 1954). Latarjet and Ephrati (1948) investigated a number of simple chemical substances for protective action against the indirect effect. They found that glucose, sucrose, alanine, and histidine confer little or no protection, while thioglycolic acid, tryptophan, cysteine and cystine, glutathione, ascorbic acid, and phenylal- anine possess considerable protective power. The presence or absence of oxygen has no effect on the inactivation rate either in the presence or absence of protective substances. The properties of X-ray-inactivated phages may be compared with those of ultraviolet-inactivated phages discussed in the previous section of this chapter, principally on the basis of an 76 BACTERIOPHAGES extensive investigation by Watson (1950, 1952) of the physiologi- cal modifications produced in bacteriophages by both direct and indirect effects of X-irradiation. To study the direct effect, phage preparations were irradiated in broth. It was found that T2 phage particles inactivated by the direct effect are still adsorbed to bacteria as efficiently as active T2 particles. Not all the inactivated phage particles, however, are able to kill the host cells on adsorption, the host-killing property being abolished by X-irradiation at Vs the rate at which plaque-forming ability is lost. X-ray-inactivated T2 particles are also able to interfere with the multiplication of active Tl phages, the interfering power being inactivated by X-rays at the same rate as the host-killing property. X-ray-inactivated T2 phages retain their ability to cause "lysis from without," and are even more effective in this regard than unirradiated phage particles. Cytological obser- vations on bacteria infected with X-ray-inactivated T2 phages show nuclear disintegration and chromatinic change, analogous to those observed with ultraviolet-inactivated T2 phages. The proportion of inactivated phage particles capable of inducing these cytological changes is the same as the proportion capable of killing the host cell, so that host cell death appears to be the result of the phage-induced destruction of the nuclear apparatus. The ability of T2r"'" phage particles to cause "lysis inhibition" is retained after X-irradiation even after the host-killing property has been abolished. Photoreactivation can also be detected in X-ray-inactivated T2 phages, but in a much smaller proportion of the phage particles than after ultraviolet inactivation. Both multiplicity reactivation and cross reactivation of X-ray-inacti- vated T2 and T4 phages exists and will be considered in detail in Chapter XVIII. It is evident that X-irradiation permits one to isolate several physiological properties of the phage particles. It is also evident that the effects of X-rays are quite distinct from those of ultraviolet light, suggesting that the nature of the lesions produced by these two types of radiation are different. This is also indicated by the fact that a population of T2 bacteriophages inactivated by ionizing radiations is no longer able to inject all EFFECT OF RADIATION ON PHAGE PARTICLES 77 of its DNA into the bacterial host cells, whereas a similar popula- tion inactivated by ultraviolet light to the same extent is still able to do so (Hershey, Garen, Fraser, and Hudis, 1954). It would appear that T2 phage is a rather complex organism, which can be functionally dissected by means of radiations. The indirect inactivation of viruses by X-irradiation in the absence of protective substances is in reality an inactivation by chemical agents generated by the radiant energy. These chemi- cal agents, as stated above, are of two kinds : the short-lived free radicals and the long-lived "peroxides." Because of the brief life of the free radicals, their range of action is short, and only those radicals produced in the water of hydration of the phage particles can have an appreciable probability of causing physio- logical damage. The properties of phage particles damaged by the short-lived radicals (indirect effect) can be studied by ir- TABLE V The Properties of Bacteriophage T2 Inactivated by X-rays and Ultraviolet Light" Multi- Inacti- Rate of Bacterial Photo- plicity vating inacti- Adsorp- killing reacti- reacti- agent vation tion ability vation vation Direct Exponential Normal Lost at a + + effect of rate Va X-rays the rate of inacti- vation Indirect ef- Cumulative Greatly Uncertain — Uncertain fect of X- damage reduced because because rays of poor adsorp- tion of poor adsorp- tion After effect Cumulative Slightly Lost at slow - + + of X-rays damage reduced rate Uhraviolet Nearly ex- Normal Normal + + + + + + + + light ponential From J. D. Watson, J. BacterioL, 63, 473 (1952), with permission. 78 BACTERIOPHAGES TABLE VI The Sensitivity of Various Bacteriophages to Ionizing Radiations En- Inacti- viron- vation ment dose« Bacteriophage during irradi- ation Radi- ation X 10-6 roent- gen Host Strain Reference Coli-dysen- C16 Broth X-ray 0.054 VVollman and Lacas- tery sagne (1940) C16 Broth X-ray 0.040 Luria and Exner (1941) C16 Broth X-ray 0.039 Holweck, Luria, and Wollman (1940) C16 Broth X-ray 0.075 Latarjet (1942) C16 Broth a-ray 0.3 Holweck, Luria, and Wollman (1940) T2, T4, T6 Broth X-ray 0.04 Watson (1950) T2 Broth X-ray 0.034 Latarjet (1948) T5 Broth X-ray 0.035 Buzzell and Lauffer (1952) TI(P28) Broth X-ray 0.09 Luria and Exner (1941) Tl Broth X-ray 0.085 Watson (1950) C13 Broth X-ray 0.12 Luria and Exner (1941) T7 Broth X-ray 0.085 Watson (1950) lambda Broth X-ray 0.11 Epstein and Englander C36 Broth X-ray 0.10 C36 Dry 7- ray 0.21 C36 Dry X-ray 0.43 C36 Dry a-ray 0.94 S13 Broth X-ray 0.39 S13 Drv 7-ray 0.58 S13 Dry X-ray 0.99 813 Dry a-ray 3.50 T105a Broth X-ray 0.059 Salmonella P22 Broth X-ray 0.11 Pyocyanea PS Broth X-ray 0.083 (1954) Wollman and Lacas- sagne (1940) Lea and Salaman (1946) Lea and Salaman (1946) Lea and Salaman (1946) Wollman and Lacas- sagne (1940) Lea and Salaman (1946) Lea and Salaman (1946) Lea and Salaman (1946) Wollman and Lacas- sagne (1940) Garen and Zinder (1955) Tobin (1953) EFFECT OF RADIATIONS ON PHAGE PARTICLES 79 TABLE VI {Continued) En- Inacti- viron- vation ment dose'^ Bacteriophage during . irradi- Radi- X io-« roent- Host Strain ation ation gen Reference Megaterium Broth X-ray 0.09 Wollman and Lacas- sagne (1940) Subtilis Broth X-ray 0.044 Wollman and Lacas- sagne (1940) Staphylococ- K Broth X-ray 0.054 Wollman and Lacas- cus sagne (1940) K Broth X-ray 0.045 Luria and Exner (1941) K Dry 7-ray 0.079 Lea and Salaman (1946) K Dry X-ray 0.109 Lea and Salaman (1946) K Dry a-ray 0.45 Lea and Salaman (1946) Streptococ- B Broth X-ray 0.20 Exner and Luria (1941) cus C Broth X-ray 0.11 Exner and Luria (1941) D Broth X-ray 0.06 Exner and Luria (1941) " Inactivation dose = dose required to reduce titer of phage population to the fraction \/e. radiating dilute phage suspensions in buffer, though some fraction of the population must also be inactivated by the direct effect under these conditions. The long-lived peroxides, in contrast, may continue to cause damage long after the irradiation has ceased. Watson (1952) called this second kind of indirect action after effect. The properties of phage particles damaged by after effect can be studied by diluting phages into medium previously heavily X-irradiated and allowing the after effect to occur in the absence of any direct irradiation of the particles themselves. The results of such studies are summarized in Table V, where the physiological properties of phages inactivated by ultraviolet light and by direct, indirect, and after effects of X-rays are presented. It is evident from this summary that the damage caused by in- 80 BACTERIOPHAGES direct and after effects is quite different from that caused by direct effect or ultraviolet light. Buzzel and Lauffer (1952) studied the effect of X-rays on phage T5 under conditions in which either the direct effect or the indirect effect was predominant, and then investigated the sen- sitivity to heat inactivation of the surviving phage population. It was found that the survivors of the direct effect irradiation have the same sensitivity to heat inactivation as an unirradiated control population, while the survivors of the indirect effect ir- radiation are far more heat-sensitive than the control. The chemical agents responsible for the indirect effect thus increase the heat-sensitivity of the survivors. The fact that the survivors of the indirect effect are already modified in some way was also shown by Watson (1952) who found that such survivors have an increased X-ray sensitivity upon a second exposure to indirect effects, i.e., that the indirect effect is cumulative (see also Alper, 1954). In contrast, Watson found that direct damage is not cumulative, i.e., that there is no evidence for "sublethal" direct lesions. For purposes of reference, the available data for the sensi- tivity of various phages to ionizing radiations are presented in Table VI. 4. Decay of Radiophosphorus It was discovered by Hershey, Kamen, Kennedy, and Gest (1951) that T2 and T4 bacteriophages lose their infectivity upon the decay (half-life of two weeks) of radiophosphorus P^^ atoms incorporated into their DNA. The kinetics of this inactivation proceed in such a manner that the surviving fraction of the population is an exponential function of the proportion of P^- atoms which have decayed by the time of assay. The rate of inactivation is proportional to the specific radioactivity of the medium in which the phage has been grown and is independent of the phage concentration during decay, provided that the particles are stored in sufficiently great dilution in a medium which contains protective substances against any indirect effects EFFECT OF RADIATIONS ON PHAGE PARTICLES 81 produced by radiations attending the radioactive decay. These observations indicate that the death of each radioactive phage particle is the consequence of the disintegration of one of its own atoms of P^-. It is possible to calculate the fraction of all radio- active disintegrations which are actually lethal to the phage particles in which they occur, on the basis of the inactivation rate and the number of P^^ atoms per phage known to have disinte- grated at various times. This fraction is approximately 0.1 in the case of T2 and T4, an efficiency of killing so high that it is possible to show by means of reconstruction experiments that the ionizations produced on the way out of the virus particles by the hard /3 electrons emitted at each P^^ disintegration cannot be the principal cause of death. Instead, a short range consequence of the radioactive disintegration, like the transmutation of phos- phorus to sulphur or the recoil energy sustained by the decaying nucleus, must be responsible for the loss of infectivity. When the lethal effects of P^- decay were similarly studied in phages Tl, T3, T5, T7, and lambda (Stent and Fuerst, 1955) and in the salmonella phage P22 (Garen and Zinder, 1955), it was found that the efficiency of killing per P^^ disintegration is very nearly the same in all of these strains, i.e., that approximately one out of every ten P^^ disintegrations kills any phage particle in which it occurs. This efficiency depends on the temperature at which radioactive decay is allowed to take place, rising from a minimum value of 0.04 at -196° C. to 0.3 at +65° C. (the value of 0.1 holding only at +4° C.) (Stent and Fuerst, 1955; Castagnoh, Donini, and Graziosi, 1955a, 1955b). On the basis of these findings, Stent and Fuerst inferred that the efficiency of killing reflects the double-stranded helical Watson-Crick structure of the DNA which harbors the decaying P^^ atoms. They proposed that the high proportion of nonlethal decays derives from the possibility that the physiological function of the DNA macro- molecules can be preserved even after radioactive decay has interrupted one of the two single polynucleotide strands and that the small proportion of lethal decays represent disintegrations which have resulted in a complete cut of the DNA double helix. 82 BACTERIOPHAGES Phage particles inactivated by decay of P^^ can still participate in cross reactivation (Stent, 1953; Stahl, 1956), indicative that the lethal radioactive disintegrations destroy only part of the viral genome. The details of these experiments will be con- sidered in Chapter XVIII. It has not been possible, however, to detect any multiplicity reactivation of phage particles in- activated by decay of P^^ xhe ability to give the lytic and lysogenic responses are both destroyed by P^^ decay at the same rate in the temperate phages lambda and P22. 5. Summary The group of physical agents whose effects on bacteriophages have been considered in this chapter includes various forms of radiant energy, such as visible and ultraviolet light and ionizing radiations, as well as the energy released by radioactive decay. In the case of radiant energy, the inactivation of phage particles is seen to result from the chemical reactions caused by the ab- sorption of the electromagnetic quanta, whereas in the case of radioactive decay the local energy suddenly generated in the vicinity of the disintegrating atom appears to be responsible for the lethal effect. Inactivation by these agents is often highly specific, involving only certain properties of the phage particle and leaving other properties unharmed. A thorough investi- gation of the mechanisms by which bacteriophage particles are inactivated may reveal much information about the complexities of structure and function in viruses. The results already ob- tained and discussed here serve as an indication of what may be accomplished. CHAPTER VII CHEMICAL COMPOSITION In order to determine the chemical composition of a virus one must be able to isolate and purify adequate amounts of it. Ap- propriate criteria of purity must be applied if the analyses are to have any validity. Therefore, these subjects will be discussed before presentation of the available data on chemical composi- tion. An extensive review, "Bacteriophages," by Putnam (1953) includes methods of concentration and purification, cri- teria of purity, and determination of size, density, and chemical composition of purified bacteriophages. A similar review by Beard (1948), "Purified Animal Viruses," includes some ma- terial on bacteriophages. 1. Concentration and Purification In general, it is unusual to obtain more than 10^" phage par- ticles per ml. in lysates prepared on a large scale. However, lysates of the even-numbered T phages are now easily obtainable which titer up to 5 X 10^^ per ml., or 250 mg. of phage per liter. The first successful purification of phage was achieved by Schlesinger (1933a) using coliphage WLL related to the T2-G16 serological group. The host bacteria were grown in a chemi- cally defined medium to avoid contamination of the product with peptones and proteins from the starting material. Bacteria and large particle debris were rem.oved by bacteriological filters. The phage was concentrated by means of collodion membranes of a pore size sufl^ciently small to retain the phage particles. The concentrated phage was further purified by high speed centrifugation. Modern criteria of purity were not yet availa- ble, but comparison of Schlesinger's results with recent data 83 84 BACTERIOPHAGES indicates that his phage preparations were probably as pure as any up to date. Northrop (1938) concentrated and purified a staphylococcus phage by applying the techniques so successfully used for enzyme purification. The host bacteria were grown in a yeast extract medium, and after phage lysis was complete much nonphage material was removed by precipitation with lead acetate. The supernatant fluid was concentrated in vacuo to 1 /20 of its volume and digested with trypsin, which does not harm the phage. The phage was then precipitated with ammonium sulfate and further purified by ammonium sulfate fractionation. Kalmanson and Bronfenbrenner (1939) concentrated and purified a strain of coliphage T2, using collodion membranes. Hershey, Kalmanson, and Bronfenbrenner (1943a) prepared highly concentrated phage suspensions by simply washing the phage from the surface of agar plates on which large populations of bacteria had been lysed. From 10^^ to 10^^ phage particles per ml. were obtained in this way. This method was further developed and applied to the seven coliphages of the T series by Swanstrom and Adams (1951), with resultant phage concen- trations about ten-fold greater than those obtained with broth lysates. This method was applied by Cohen and Arbogast (1950a) and Marshak (1951) and is useful in laboratories with limited centrifuge facilities. Fraser (1951a) developed an ap- paratus for growing bacteria in batches up to 6 liters under conditions of vigorous aeration. He obtained yields of phage T3 of better than 10^^ per ml. in large scale lots. Coliphage T2 had been concentrated on a large scale from fluid lysates by means of the Sharpies continuous flow centrifuge (Hook, Beard, Taylor, Sharp, and Beard, 1946). The concen- trated phage suspensions were then further purified by sedi- mentation in a high speed vacuum centrifuge. Similar methods were used by Kerby, Gowdy, Dillon, Dillon, Csaky, Sharp, and Beard (1949) for the purification of coliphage T7 and by Putnam, Kozloff, and Neil (1949) for the purification of coliphage T6. In the latter paper are preliminary reports of the fractionation CHEMICAL COMPOSITION 85 of phage concentrates with alcohol in the cold and by acid pre- cipitation at pH 4.2. These methods were also applied to the purification of phage T7 by Putnam, Miller, Palm, and Evans (1952). Herriott and Barlow (1952) reported yields of 2 to 5 X lO'^ per ml. of phage T2 in a chemically defined medium. Ap- parently, adequate aeration brought about by vigorous shaking is the explanation of the high yields, as in the case of the results with phage T3 of Fraser (1951a) noted above. The phage stocks were filtered through a calcined diatomaceous earth (Celite) to remove bacterial debris, and the phage was pre- cipitated in the cold by acid at pH 4.0. The precipitated phage was resuspended at pH 6.5 and treated with deoxyribo- nuclease, after which it was purified by differential centrifuga- tion. A fairly simple method of preparing T2 phage lysates in 10 liter quantities was described by Siegel and Singer (1953). A defined medium was aerated by forcing air through a sintered glass tube and lysates which titered 2 to 5 X 10^^ per ml. were produced. Wyatt and Cohen (1953) obtained lysates titer- ing up to 1 X 10^2 per ml. through the use of a thin film of en- riched medium in which cells at 3 X 10^ per ml. were multi- infected. Wahl and Blum-Emerique (1949a) used alcohol precipitation for concentration of phage CI 6 with good recovery of activity. Wahl, Terrade, and Monceaux (1950) used successive adsorp- tion and elutions from calcium phosphate for concentration and purification of a salmonella phage. Northrop (1955b) again applied salting out and enzymatic treatment to purify a mega- terium phage, which is inactivated by repeated centrifugation. 2. Criteria of Purity Certain criteria of purity have been discussed from another viewpoint in Chapter IV. From its dry weight and density one can calculate the diameter of the infective particle. If the size calculated in this way is larger than the true size but the effi- 86 BACTERIOPHAGES ciency of plating is high, the discrepancy must be attributed to aggregation or gross impurity in the phage preparation. This is a rather crude criterion of purity but it suffices to discredit many claims in the literature that pure phage has been pre- pared. Direct examination in the electron microscope can de- tect aggregates of phage as well as contaminating particles with sedimentation properties similar to those of phage particles. The analytical centrifuge is useful for detecting impurities with a slower rate of sedimentation than the phage particles. It also gives a measure of the homogeneity of the preparation with re- spect to particle size and density (Putnam, 1953). Electro- phoresis is a useful technique for detecting impurities which have a different mobility in the electric field (Putnam, 1953). Opti- cal analysis of diflE'usion at a phage-buffer boundary has been used also as a test for homogeneity of preparations of phage T6 (Putnam, 1953). The specific infectivity of preparations will be low if a fraction of the phage particles are not infective for any reason. Such a preparation might well react as though it were homogeneous by any other test. An inhibitor of phage T2 which produces a low specific infectivity has been observed in cer- tain phage lysates but fortunately Sagik (1954) and others have found methods for separating it from the phage. An immunological criterion of purity has been used by Cohen and Arbogast (1950a). The most probable impurity in phage purified by diff"erential centrifugation is particulate bacterial debris which should react with antibacterial antibodies to give a precipitate. Cohen obtained such precipitates with purified phage preparations and demonstrated that ribonucleic acid was removed in the precipitate, so that treatment with antiserum was a method of purification as well as a test of purity. Serum proteins could then be removed by centrifugation. The im- munological criterion was used by Kozlofl^ and Putnam (1949) and by Herriott and Barlow (1952). The constancy of certain properties of the phage preparation such as infectivity per mg. nitrogen, infectivity per unit of turbidity (extinction at 4,000 A) and infectivity per unit of ab- CHEMICAL COMPOSITION 87 sorbancy (extinction at 2,600 A) was used as a criterion of homo- geneity of purified T2 phage by Herriott and Barlow (1952). The phage was assayed for these properties before and after fractionation by alcohol precipitation, adsorption with alumina or copper hydroxide, centrifugation, acid precipitation, adsorp- tion to host cells, and treatment with antibacterial sera. The criteria of purity offered by an investigator for his ana- lyzed preparations are an essential part of his evidence for any unusual analytical results claimed. This point is too often dis- regarded. TABLE VII Elementary Composition of Purified Bacteriophages (per cent of dry weight) Phage C N P H Reference WLL 42 13.2 3.7 6.4 Schlesinger (1934) T2(PC) 49.3 14.0 0.07 7.9 Kalmanson and Bronfenbrenner (1939) T2 — 11.8 3.7 — Cohen and Anderson (1946) T2 42 13.4 5.0 — Taylor (1946) T2 — 16 5.2 — Herriott and Barlow (1952) T6 — 13.3 3.95 — Kozloff and Putnam (1949) T7 — 14.7 4.2 — Csaky, Beard, Dillon, and Beard (1950) Staphylo- coccus 41 14.3 4.8 5.3 Northrop (1938) 3. Elementary Composition Elementary analyses are available for only three serologically distinct groups of phages, the staphylococcus phage of Northrop, coliphage T7, and the group of phages closely related to T2. Although these analyses (Table VII) are in good agreement among themselves it is probably unwise to make generalizations about phage composition from such a restricted sample. The only seriously discrepant result is the low phosphorus content reported by Kalmanson and Bronfenbrenner, which can prob- ably be discarded as an error. It is specially noteworthy that the assays on coliphage WLL by Schlesinger in 1934 are in such excellent agreement with later results on T2 and T6. BACTERIOPHAGES TABLE VIII Gross Chemical Components of Phages (per cent by weight) Car- Pro- bohy- Phage tein DNA" RNA" drate Lipid Referer T2 — 37 — — Cohen and Anderson (1946) T2 51 42 4 12 2 Taylor (1946) T2 40 50 <1 — <1 Herriott and Barlow (1952) T6 — 42 3 — — Kozloffand Putnam (1949) T7 49 41 " 17 <1 Csaky, Beard, Dillon, and Beard (1950) " DNA = deoxyribonucleic acid. * RNA = ribonucleic acid. 4. Gross Chemical Components The few analyses available on the protein, carbohydrate, lipid, and nucleic acid components of bacteriophages are sum- arized in Table VIIL There seems to be general agreement that the phages analyzed are about half protein and half nucleic acid of the deoxyribose type. The intrinsic lipid and ribo- nucleic acid of phage T2 appears to be very small, if any. On the basis of the radioactivity in the mononucleotide fraction of phage T2 produced from a P^--labeled medium, Volkin and Astrachan (1956c) placed the upper limit of the ribonucleic acid content of this phage at 0.025 per cent and argued that the dis- tribution of radioactivity was such that even this quantity is probably an impurity. The protein of phage T2 is physically separable into two kinds (Hershey, 1955). One of these is the coat or "ghost" of the phage particle and the other is a relatively small soluble protein. The latter is probably mixed with the DNA in the phage head for it is injected into the cell, and it is readily lib- erated into the medium following osmotic shock which releases the DNA. The amino acid analyses of these two protein com- ponents are not distinctively different. CHEMICAL COMPOSITION 89 The carbohydrate of T2 is part of the nucleic acid, but that of T7 may be in part a constituent of the phage particle or an im- purity. It is desirable to have additional analyses of this type, since only two serologically distinct groups of phages have been analyzed and this is an inadequate sample from which to draw general conclusions about the chemical nature of phages. The gross chemical composition of the few phages analyzed is quite different from the composition of any plant or animal virus which has been analyzed (Beard, 1948). The known bac- teriophages are characterized by a large content of deoxyribo- nucleic acid (DNA). The well known plant viruses contain ribonucleic acid (RNA) exclusively, usually in relatively small amounts. Two, and perhaps three, animal viruses contain pre- dominantly RNA. The latter include the carefully studied PR8(A) and Lee(B) strains of influenza virus (Ada, 1957) which are reported as having 0.9 per cent RNA and 0.1 per cent DNA, and MEF-1 poliomyelitis virus (Schwerdt and Schaffer, 1955) which analyzed 24 per cent RNA and 1 per cent DNA. Nearly 10 per cent RNA was found in Eastern equine encephalitis virus by Taylor, Sharp, Beard, and Beard (1943). Wyatt (1952) reported DNA in some insect viruses. 5. Amino Acids, Nucleotides, and Bases Poison and Wyckoff (1948) published the first amino acid analyses for phage T4. The phage was purified by centrifuga- tion, hydrolyzed, and the amino acids separated by paper chro- matography. No criteria of purity were presented and the authors seem to have reservations about the significance of the results. Luria (1953a) reported amino acid analyses of T2 and T4 bacteriophages from his laboratory. Fraser and Jerrell (1953) described in more detail the amino acids in preparations of T3. Hershey (1955) analyzed several protein fractions from T2. The results are summarized in Table IX. The phage pro- teins do not diflfer markedly in composition from plant and ani- mal virus proteins but are quite diflferent from the histones and 90 BACTERIOPHAGES protamines that form part of the nucleoproteins of higher organ- TABLE IX Amino Acid Analyses of Some Bacteriophages Amino acid T2" T2* T4« T4^ j3d Aspartic acid Glutamic acid 10.5 10-14 13-15 — 12 12 11.7 10.5 Glycine Alanine 17.2 4-6 6-10 12 7.3 9.4 9.8 9.7 Valine 6.5 6-8 6.1 6.5 6.6 Leucine Isoleucine 5.7 7.4 12-15 5.6 5.5 6.5 3.9 9.8 4.9 Serine 5.7 4.5-6 2.3 4.8 2.9 Threonine 5.7 4.5-6 6.1 7.0 5.3 Cystine 0.2 0.1-0.4 — — — Methionine 3.0 1-2 2.35 < 1.3 1.7 Proline 3.7 3.6-5 — 5.0 4.2 Phenylalanine Tyrosine 5.0 4.5 7.5-9 3-9 5.3 4.6 4.2 3.7 3.5 4.2 Tryptophan Histidine 0.9 1.0 0.6-2.5 0.64 <2.6 1.7 Lysine 7.4 4.7-7 4.2 8.5 6.3 Arginine 5.6 3.6-5.3 4.0 6.5 7.2 ° Luria (1953a), results of microbiological assays expressed as per cent of phage protein. * Hershey (1955), results for C"-labeled acid-insoluble protein by radio- chemical analysis after separation on paper chromatograms, showing distribu- tion of carbon as per cent of total recovered. '^ Poison and WyckofT (1948), separation by paper chromatography and ninhydrin assay, expressed as per cent of recovered amino acids. ^ Fraser and Terrell (1953), separation by ion exchange chromatography, analysis by ninhydrin reaction expressed as per cent of recovered amino acids. Early analyses of the constituents of phage nucleic acids yielded conflicting results. Kozloff, Knowlton, Putnam, and Evans (1951) reported the isolation from phage T6 of adenine, guanine, thymine, and cytosine, but the relative proportions of the bases were not determined. Weed and Cohen (1951), while CHEMICAL COMPOSITION 91 Studying transfer of host cell pyrimidines to phage T6, isolated thymidylic acid, deoxycytidylic acid, and adenine from the phage. Smith and Wyatt (1951) reported the presence of ade- nine, guanine, cytosine, and thymine in both T5 and T2. Marshak (1951) found adenine, guanine, and thymine in phage T2 but made the rather startling observation that neither cytosine nor 5-methylcytosine could be found, although cytosine was present in phage T3. Marshak also noted that the absorption spectrum for the guanine isolated from phage T2 was abnormal suggesting the possibility of contamination of the guanine with some other base. The confusion was finally resolved by Wyatt and Cohen (1953) who reported that phages T2, T4, and T6 contain neither cytosine nor 5-methylcytosine, but contain in- stead a new base, 5-(hydroxymethyl)cytosine. Some analyses on the nucleic acids of phages are given in Table X. A few addi- tional data are given by Lwoff (1953). It appears that only T2, T4, and T6 contain 5-(hydroxymethyl)cytosine. TABLE X The Molar Proportions of Nucleic Acid Bases in Phages T2, T4, Base TS" T2* T6'^ TG'' T2« T7/ Adenine 34 33 33 33 32 26 Guanine 21 17 18 24 17 24 Cytosine 11 — — — — 24 HMC — — 17 — 16 — Thymine 35 31 33 — 35 26 « Smith and Wyatt (1951). * Marshak (1951), absolute values arbitrary. 'Wyatt and Cohen (1953). ^ Koch, Putnam, and Evans (1952), absolute values arbitrary. ' Hershey, Dixon, and Chase (1953). •'^ Lunan and Sinsheimer (1956). " 5-(Hydroxymethyl) cytosine. The presence of glucose in glucosidic linkage with 5-(hydroxy- methyl) cytosine (HMC) of T2, T4, and T6 is clearly indicated 92 BACTERIOPHAGES by work from several laboratories. The presence of a hexose in the material released from T4 by a specific lipocarbohydrate was first mentioned by Jesaitis and Goebel (1953). Sinsheimer (1954) independently found glucose attached to some but not all of the HMC released from T2 phage DNA by the combined action of pancreatic deoxyribonuclease and purified venom phos- phodiesterase. Volkin (1954b) reported similar results for T2, T4, and T6 phages, except that in T4 he found one mole of glu- cose for each mole of HMC. In addition, he found no glucose in Tl phage. Sinsheimer (1956) confirmed the difl?"erence between T2 and T4, and found that glucose content is one of several characters showing unusual inheritance in crosses between T2 and T4 (Streisinger and Weigle, 1956). Cohen (1956) reported that the r mutants of T2, T4, and T6 contain higher glucose/deoxyribose ratios than do the wild type phages. Sinsheimer's (1956) analytical results do not support this finding so that clarification must await further studies. 6. Extinction Coefficients of Phage Preparations An analytical quantity of some interest to phage workers is the light absorption at 2,600 A since this is a measure of the nucleic acid content of the virus particle. However, there has been no uniformity in the presentation of data. The absorbancy has been expressed as extinction coefficient per infective phage particle, per morphologically typical phage particle, or per mg. of phage phosphorus per ml. In some cases the extinction coefficient has been corrected for light scattering, in some cases not. This lack of uniformity makes comparison of data difficult. The expres- sion of analytical results in terms of the infective particle is very convenient for the analyst but is only of comparative value be- cause of unknown contamination of phage preparations with inactive phage particles and uncertainties in the efficiency of plating. Luria, Williams, and Backus (1951) avoided these difficulties by giving the absorbancy per morphologically typical phage particle counted in the electron microscope. Examples of the available data are given in Table XI. They express optical CHEMICAL COMPOSITION 93 densities per cm. divided by phage concentrations in visible particles per ml., plaque-forming particles per ml., and mg. of phosphorus per ml., insofar as the data given permit. In many instances values are given for the best of numerous preparations that vary considerably. These variations depend in part on variable content of noninfective phage particles in the prepara- tions but, as the data of Luria, Williams, and Backus (1951) show, depend also on other impurities that contribute to the absorbancy or light scattering by the preparation. TABLE XI Extinction Coefficients of Phage Preparations cm. VI 01'^ cm.VlO^" cm.Vmg. Phage particles plaques P Tl" — 8.3 300 T2* 4.5' 8. 8-^ — Tl" — 6.0 310 TA" 2.r 7.3' — TV — 6.7 270 Te" 5.0<^ 7.4^ — T6^ — — 260 T6^ — — 363 T5« — 5.0<= — TS" — 32 300 T7/ — 2.7'^ 255 DNA" — — 278 ° Herriott and Barlow (1952). ^ Luria, Williams, and Backus (1951). ' These figures are about 20 per cent low relative to the others owing to cor- rection for scattering. ^ Hershey, Dixon, and Chase (1953). * Hershey, Kamen, Kennedy, and Gest (1951). •^Putnam, Miller, Palm, and Evans (1952). " Cohen and Arbogast (1950b). " Smith and Wyatt (1951). Measurements of extinction provide a rapid and convenient method of assay of phage during purification, and the extinction per plaque-forming particle of the final product is an excellent 94 BACTERIOPHAGES measure of contamination by impurities that contain nucleic acid, on the one hand, or of the efficiency of the plaque count on the other. 7. Phosphorus and Nitrogen Contents per Infective Particle The content in phosphorus and nitrogen per plaque-forming particle has been frequently measured, and a few data are given in Table XII. The phosphorus content of some additional phages is cited by Lwoff (1953) and Stent (1958). The results are only moderately consistent among themselves and with the expectations based on size measurements, composition, and absolute infectivity. The results for T5 given by Smith and Wyatt (1951) illustrate the uncertainties that are inseparable TABLE XII Phosphorus and Nitrogen per Plaque-Forming Particle (yUg./lO" particles) Phage Phosphorus Nitrogen Reference T2 2.0 — Hershey, Dixon, and Chase (1953) T2 3.7 10 Hook, Beard, Taylor, Sharp, and Beard (1946) T2 2.8 8.3 Herriott and Barlow (1952) T4 2.3 — Cohen and Arbogast (1950a) T6 3.3 10 Kozloff and Putnam (1949) T6 3.4 — Cohen and Arbogast (1950a) WLL 1.6 5.7 Schlesinger (1934) Tl 1.7 — Labaw (1953) T7 1.3 — Labaw (1953) T7 0.82 — Lunan and Sinsheimer (1956) T7 1.5 5.3 Putnam, Miller, Palm, and Evans (1952) T7 1.5 5.0 Csaky, Beard, Dillon, and Beard (1950) T5 3.8 — Labaw (1953) T5 10 36 Smith and Wyatt (1951) Staphylococcus 6.7 20 Northrop (1938) Megaterium — 17 Northrop (1955b) CHEMICAL COMPOSITION 95 from measurements of this type. The ratio of nitrogen to phos- phorus indicates that the material consists of phage particles but the plaque count is improbably low. The phosphorus (or nucleic acid) content per particle is important for several reasons (Stent, 1958) and reliable ways to determine it need to be de- veloped. 8. Summary The few phages that have been analyzed contain deoxyribo- nucleic acid and protein in approximately equal amounts and very little else. This distinguishes them from most living things including most of the other viruses, excepting perhaps certain animal sperm and some insect viruses. A few phages, notably T2, contain a chemically unique nucleic acid. Phage proteins have been relatively little studied. Most of the protein of T2 is membrane material. It is not unusual with respect to amino acid composition. CHAPTER VIII ANTIGENIC PROPERTIES Early studies on the antigenic properties of bacteriophages have been reviewed by d'Herelle (1926) and Burnet, Keogh, and Lush (1937). The latter reference gives an excellent dis- cussion of the mechanism of the antigen-antibody reaction from the kinetic viewpoint and, in addition, contains much experimental material which has not been published elsewhere. Bordet and Ciuca (1921) first demonstrated that injection of rabbits with phage lysates stimulates the production of phage- neutralizing antibody. Their antisera also contained aggluti- nins for the host bacterium. The injection of host bacteria, however, failed to stimulate the production of phage-neutraliz- ing antibody. Otto and Winkler (1922) were able to remove all bacterial agglutinins by absorption with host bacteria with- out affecting the antiphage properties of the serum. These findings demonstrated that phage is antigenically distinct from its host. It was early recognized that antiphage sera had considerable specificity, inactivating the homologous and some heterologous strains of phage but not others. The fact that serological re- lationship was correlated with other properties of bacterio- phages was not generally appreciated, however, until the taxo- nomic work of Burnet and Asheshov. Since these early experiments it has been found that the ac- tivity of antiphage antibodies is not confined to neutralization. The addition of a suflSciently concentrated suspension of phage particles to homologous antiserum results in visible aggregation of the virus particles followed by settling of the clumps as a precipitate. By suitable ultrafiltration methods one can sep- 97 98 BACTERIOPHAGES arate from phage lysates soluble substances possessing the antigenic specificity of the phage particles. When antiphage antibody is allowed to react with such substances it loses its ability to neutralize phage. Some, if not all, phage-antiphage systems fix complement. These and other properties of anti- phage sera will now be discussed in more detail. 1. Preparation of Antiphage Sera Rabbits are the most convenient animals to use for antibody production, and most of the serological work with phage has been done with rabbit sera. The phages themselves are non- toxic and nonpathogenic for animals and can usually be in- jected in large amounts without damage to the recipient. Crude phage lysates, however, may contain toxic bacterial sub- stances, such as the endotoxins of the enteric bacteria and the much more potent exotoxins of the corynebacteria and Clos- tridia. In such cases the toxins must first be removed by frac- tionation or inactivated. For many purposes it is desirable to purify the phages before using them for immunization, since adequate purification will result in the absence of host-cell antibodies from the resultant antisera (Kalmanson, Hershey, and Bronfenbrenner, 1942). If crude phage preparations are used it may be necessary to remove host-cell antibodies by absorption with bacterial suspensions. This is essential, for instance, in studies involving complement fixation. Stimulation of antibody production requires the administra- tion of adequate quantities of antigenic material. The daily injection of about 10^'' virus particles (about 10 jjig.) for five days will usually result in the presence of considerable neutralizing antibody in serum obtained one week after the last injection. Repetition of this course of immunization will further increase the antibody titer. The final result does not seem to depend greatly on the route (intravenous, intraperitoneal, or subcu- taneous) by which antigen in administered. Use of the sub- cutaneous route, however, is least likely to lead to toxic or ana- phylactic reactions. Since there are large differences in anti- ANTIGENIC PROPERTIES 99 body response from one rabbit to another, it may be desirable to immunize at least three rabbits with the same phage prep- aration. Judged by the neutralization test, antisera prepared against certain phages, such as T2, T3, T4, T6, and T7, usually have very high homologous antibody titers in comparison with antisera against others, such as Tl and T5, which may be rela- tively weak. Such differences do not necessarily reflect corre- sponding differences in antibody concentration, since the rate of neutralization may easily depend on other factors as well. Actually, a very few determinations have been made of the absolute antibody content of antiphage sera. The effect of adjuvants on phage antigenicity has been very little studied, probably because most phages are already ex- cellent antigens. Wahl and Lewi (1939) obtained a marked improvement in the antigenicity of a low-titer B. suhtilis phage by adsorption to alumina. Intraperitoneal injection of 3 X 10^ particles of unadsorbed phage gave little or no antibody response. Injection of the same amount of absorbed phage re- sulted two weeks later in an antiserum giving 90 per cent in- activation at a dilution of Vsq. The titer increased to V200 in a month. This technique may be useful to workers dealing with low-titer temperate phages. Phages retain antigenicity, in some instances without appre- ciable loss, following careful inactivation of their infectivity by heat, formaldehyde, ultraviolet irradiation (Muckenfuss, 1928; Schultz, Quigiey, and Bullock, 1929; Nagano and Takeuti, 1951; Pollard and Setlow, 1953; Miller and Goebel, 1954); phenol, photodynamic action of methylene blue (Burnet, Keogh, and Lush, 1 937) ; osmotic shock, and sonic vibration (F. Lanni and Lanni, 1953; Nagano and Oda, 1954). Phage which has been "overneutralized" by homologous antibody is, however, nonantigenic in guinea pigs, even after treatment with papain (Kalmanson and Bronfenbrenner, 1943; Nagano and Takeuti, 1951). The antigenicity of premature lysates of a staphylococcal phage has been studied by Rountree (1952); of purified T2 100 BACTERIOPHAGES "doughnuts" by Lanni and Lanni (1953); of proflavine lysates of T4, and of T4- and T3-infected bacteria by Barry (1954); and of infected and lysogenic B. megaterium by Miller and Goebel (1954). The antigenicity of ultrafiltrable phage-related ma- terials is discussed below. 2. Multiplicity of Phage Antigens Until a few years ago it was believed, simply because there was no contradictory evidence, that each strain of phage pos- sessed a single kind of antigen. Accordingly, all of the specific eflfects of antiphage serum were interpreted in terms of the reac- tion of a single antibody (neutralizing antibody) with the virus particles. In an extensive study, Lanni and Lanni (1953) (see also Rountree, 1952; De Mars, Luria, Fisher, and Levin- thai, 1953) presented clear evidence for the existence of at least two distinct antigens in phage T2. One of the antigens, which appears to be localized in the phage tail, reacts with neutralizing antibody. The primary measurable eff'ect of this reaction is neutralization of infectivity; under appropriate conditions virus aggregation and complement fixation can also be demon- strated. The second antigen, which appears to be localized in the phage head, participates in specific aggregation and com- plement fixation, but does not react measurably with neutraliz- ing antibody. The two antigens are confined to the surface structures (protein) of the virus ; thus far, the internal contents, liberated by osmotic shock, have given no sign of serological activity (see also Hershey and Chase, 1952; Hershey, 1955). The available evidence indicates that staphylococcal phage 3A (Rountree, 1952) and phage T5 (Fodor and Adams, 1955) likewise possess two distinct antigens, only one of which reacts with neutralizing antibody. Evidence bearing on the sero- logical heterogeneity of the phage tail will be discussed in later sections. The failure until recently to recognize the antigenic complexity of phage attaches considerable doubt to many of the structural interpretations given to early serological observations. ANTIGENIC PROPERTIES 101 3. Neutralization of Phage Infectivity When a phage preparation is mixed with homologous anti- serum there is a progressive decrease in the number of plaque- forming particles. The inactivation process can be interrupted at any time by diluting the phage-antibody mixture below the antibody concentration at which collisions between the reactants occur at a significant rate. Following this dilution there is, in general, no detectable reversal of the inactivation process; the titer of surviving phage does not increase. Apparently, dissociation of the phage-antibody complex does not occur at a measurable rate (Burnet, Keogh, and Lush, 1937; Hershey, 1943; see, however, Andrewes and Elford, 1933b; Jerne and Avegno, 1956). Attempts to facilitate dissociation by the addi- tion of formalin-inactivated phage to bind the antibody were unsuccessful (Hershey, 1943). The addition of soluble phage antigen, however, appeared to have a slight reactivating effect (Burnet, 1933b). Although in most instances antibody does not appear to dissociate from the phage particles spontaneously, there is evidence that it can be removed by certain methods. Kalmanson and Bronfenbrenner (1943) found that neutralized phage could be completely reactivated by digestion of the anti- body with papain providing the antibody had not been too concentrated or the serum treatment too prolonged. Appar- ently, if a sufficiently large number of antibody molecules had reacted with the phage particle, the inactivation became irre- versible. Anderson and Doermann (1952b) succeeded in ob- taining more than a 30-fold increase in titer of a neutralized preparation of phage T3 by subjecting it to sonic vibration. In the cited reactivation experiments with papain and sonic vibration the concentration of phage during preliminary treat- ment with antiserum was sufficiently high in most instances as to admit the possibility that part, at least, of the plaque-count decrease was due to specific aggregation. It is not certain, therefore, to what extent the observed reactivation can be ascribed to reversal of neutralization, as opposed to disaggrega- tion of virus clumps. However, Hershey (1943) found that 102 BACTERIOPHAGES papain can reactivate phage which has been treated with serum under conditions minimizing the possibility of specific aggrega- tion. These successful attempts to reactivate neutralized phage particles indicate that antibody does not kill or destroy phage particles. Presumably, it interferes with some step in the inter- action of the intact virus particle with the host cell. The nature of this interference will be discussed in a later section. From the foregoing discussion, it is clear that neutralizing antibody acts on the phage particles themselves, rather than on the host cell. Pretreatment of host cells with phage anti- bodies has no effect on the infectious process. Delbriick (1945b) found that, following phage adsorption to the host cell, the infected bacteria are immune to antiphage serum through- out the latent period (see, however, Lieb, 1953; Gross, 1954a; Nagano and Mutai, 1954a). The resistance of infected bacteria to antiphage serum is of great importance in phage technology because it permits the inactivation of unadsorbed phage with- out affecting the multiplication of phage which is already ad- sorbed. Although complement is not required for specific neutraliza- tion, it can modify the neutralization of Tl and T2 (Hershey and Bronfenbrenner, 1952). Complement and properdin to- gether, but not separately, inactivate T2 in the absence of specific antibody (Van Vunakis, Barlow, and Levine, 1956). 4. Properties of Incompletely Neutralized Phage Preparations Following the treatment of phage preparations with anti- serum, the surviving infectious particles are not normal. An- drewes and Elford (1933b) reported that the survivors of anti- body treatment produced smaller plaques than normal. They interpreted this as indicating a delay in the initiation of the infectious process. The surviving particles also failed to pass through bacteriological filters and collodion n^embranes which were freely permeable to the same phage particles prior to serum treatment. This might be ascribed to increase of par- ticle size or change of surface charge due to an antibody coating, ANTIGENIC PROPERTIES 103 or to aggregation of phage particles. Frequency distribution curves for plaque sizes of phage CI 6 before and after treatment with antiserum are given by Burnet, Keogh, and Lush (1937). The distribution for the incompletely neutrahzed preparation was bimodal with one maximum at less than Vg the peak plaque diameter of the unimodal curve for normal phage particles. Surprisingly, Delbriick (1945b) failed to find any difference between normal T2 phage and serum-survivors at the 1 per cent level with respect to adsorption, latent period, or burst size (cf. Adams and Wassermann, 1956). The reason for the discrepant observations wdth these closely related phages is not obvious. Gough and Burnet (1934) extracted from suspensions of the host bacterium a phage-inhibiting agent (PIA) which was able to inactivate phage particles in m.uch the same manner as anti- phage sera. They regarded the extract as a solution of the bacterial receptors for phage adsorption. Burnet and Freeman (1937) investigated the effect of PIA on phage preparations which had been inactivated to the extent of 90 per cent with dilute antiphage sera. The survivors of serum inactivation proved to be very resistant to the inhibiting action of PIA whereas the untreated phage population was 99 per cent in- activated by PIA. This is further evidence that phage anti- body can modify the properties of phage particles without in- activating them. It also suggests that the isolated PIA is not identical to the receptor substance on the bacterial surface, since the survivors of the serum action are still able to infect the host cell although they are resistant to the action of PIA. Another very striking property of partially neutralized phage preparations was demonstrated by Hershey, Kalmanson, and Bronfenbrenner (1944) using a variant of phage T2. The var- iant adsorbed very poorly to bacteria in a salt-free medium and had a relative efficiency of plating of 10~^. In a medium containing 0.1 N' NaCl the phage adsorbed well and the effi- ciency of plating was about half of that under optimal condi- tions. The phage was treated with very dilute antiphage 104 BACTERIOPHAGES serum (dilution 10~-^ )to a survival of about 0.2. The plaque count in a salt-free medium of the survivors of serum treatment was increased 1,000-fold as compared with that of the untreated phage preparation. Treatment with a small amount of anti- serum had the same effect on phage infectivity as an increase in salt concentration. It is probable that these effects of salt and of antiserum on phage infectivity are due to effects on electro- static charge (see Cann and Clark, 1955), although this explana- tion was discarded by the authors. This problem will be dis- cussed further in the chapter on phage adsorption. It is not certain at present which of the several antibodies in antiphage serum is responsible for the effects described above. A peculiar antibody which stabilizes the activated state of a tryptophan-requiring strain of T4 has recently been described (Jerne, 1956). 5. Complement Fixation Early results (d'Herelle, 1926) left considerable doubt as to the ability of phage-antiphage systems to fix complement. These doubts have been dispelled in recent years by studies showing not only that such systems do, in fact, fix complement but also that complement fixation offers a valuable quantitative tool for the study of those phage antigens and phage-related materials that do not react with neutralizing antibody. An example is the work of Lanni and Lanni (1953) in their demon- stration of a serological relationship between T2 doughnuts and T2 phage and their use of antidoughnut sera for studying the structure of the infective virus particle. The scope of the procedure is not restricted, however, to such antigens. Other recent studies illustrate further the broad applicability of complement fixation to problems of phage structure and growth. In confirmation of the discovery of Hershey and Chase (1952) with T2, Lanni (1954) showed that the bulk of the com- plement-fixing antigens of infecting particles of T5 remained outside the infected host cells. Previous work had suggested instead that the complement-fixing antigens of T5 (Rountree, ANTIGENIC PROPERTIES 105 1951b) and of staphylococcal phage 3A (Rountree, 1952) dis- appeared during the early stages of infection. Both workers observed that newly synthesized complement-fixing antigens appeared and increased intracellularly a few minutes before mature virus. Miller and Goebel (1954) injected lysogenic cells of B. megalerium into rabbits and failed to demonstrate the production of phage-specific, complement-fixing, or neutralizing antibodies. Since nucleic acids have not, in general, been found antigenic, this negative result, which extended an earlier failure by F. Lanni (cited by LwofF, 1953), accords with the current conception that prophage consists of DNA. Fodor and Adams (1955) used complement fixation in analyzing the antigenic relationship of T5, the related phage PB, and T5 X PB hybrids. Details of a photometric complement titration are given by Lanni (1954). 6. Specific Aggregation Schlesinger (1933a) observed the aggregation of phage WLL particles by antiphage serum using the dark field microscope. Burnet (1933c) demonstrated that phage CI 6 gave a visible precipitate when mixed with antiserum. Photographs of the aggregated phage particles, taken by Barnard with the ultra- violet microscope, are included in Burnet's paper. A visible precipitate was formed when antiserum was mixed with phage CI 6 at all concentrations above about 2X10^ particles per ml. For smaller virus particles a considerably higher concentration is needed to give a visible precipitate, since the amount of precipitate formed is dependent on the total mass of antigen added. The relationship between the size of an antigenic particle and the concentration of particles needed to give a visible precipitate has been discussed by Merrill (1936). This author also discussed the relationships between particle size and the minimal concentration required for complement fixation and for the stimulation of antibody production. These relation- ships should be of considerable value to virologists. Ignorance of them has been responsible for much wasted eflfort. 106 BACTERIOPHAGES The phage-antibody system could be readily investigated by the techniques of the quantitative precipitation reaction, but the only attempt at such a study was made by Hershey, Kal- manson, and Bronfenbrenner (1943a) using coliphage T2. The authors concluded that the nitrogen of phage lysates which was specifically precipitable by phage antibodies corresponded to less than 10~^^ mg. N per lytic unit. Chemical analyses of purified T2 phage preparations by various investigators agree with a figure for the nitrogen content of the lytic unit which is close to 10~i3 mg. (Chapter VII). The amount of antibody nitrogen capable of precipitating with phage T2 was found to be 0.6 mg. per ml. for a serum with a K value of about 360 min- utes ~^ It would be of interest to know the ratio between anti- body content and A' value for various other antiphage sera. With sera of low neutralizing antibody content, it is possible to follow specific aggregation of phage by observing the con- sequent reduction in plaque count (Lanni and Lanni, 1953; Fodor and Adams, 1955). 7. Agglutination of Phage-Coated Bacteria Individual cells of susceptible bacterial strains are able to adsorb a considerable number of phage particles. Delbriick (1940a) found that cells of E. coli became saturated after ad- sorption of from 20 to 250 phage particles depending on the physiological condition of the cells. Burnet (1933a) investi- gated the serological properties of such phage-coated cells using formalin-killed bacteria to avoid lysis. The formalinized bacteria were suspended in a high-titer phage stock for an hour, then centrifuged and washed to remove unadsorbed phage. These phage-coated bacteria were less susceptible than uncoated bacteria to agglutination by antibacterial sera, but were readily agglutinated by antiphage serum which had no effect on un- coated bacteria. The agglutination was phage-specific. Ab- sorption of the serum with phage-coated bacteria depressed both the agglutinating and the phage-neutralizing activities. The result is interesting since it suggests that antigens capable ANTIGENIC PROPERTIES 107 of binding neutralizing antibody exist elsewhere than at the tip of the phage tail. More work is needed, however, before this interpretation is accepted. Burnet also performed cross-absorption experiments using serologically related phages. Absorption with bacteria coated with homologous phage reduced the neutralization titer for homologous and heterologous phages alike, whereas bacteria coated with heterologous phage absorbed antibody against this phage but had little effect on the titer against homologous phage. This type of experiment permits one to make an anti- genic analysis of a series of related phages. Antiphage sera can also be absorbed with phage concentrates, the precipitated phage particles being removed by low-speed centrifugation (Hershey, Kalmanson, and Bronfenbrenner, 1943a, 1943b; Hershey, 1946a). An advantage of the use of phage-coated bacteria is that the adsorption of the phage particles to the bac- teria serves to concentrate and purify them in one step. It should be kept in mind, however, that those phage antigens which are located at or near the site of adsorption to host cells may not be accessible to antibody. 8. Specific Soluble Substances The first evidence for a soluble substance having the antigenic specificity of phage particles was obtained by Asheshov (1925). He filtered a phage preparation through a collodion membrane of a pore size small enough to retain the phage particles. The immunization of rabbits with the phage-free filtrate resulted in the production of specific phage-neutralizing antibody. Burnet (1933b) prepared sterile ultrafiltrates of high-titer stocks of coliphage CI 6 using Gradocol membranes of a suitable per- meability to retain all the phage particles. The ultrafiltrates immunized rabbits with the production of neutralizing antibody for phage CI 6. The ability of the ultrafiltrate factor to react with antibody was most conveniently detected by mixing it with antiphage CI 6 serum and observing a decrease or blocking 108 BACTERIOPHAGES of the phage-neutralizing activity of the serum. The antibody- blocking effect showed the serological specificity of phage CI 6. The ultrafiltrate factor was not adsorbed by phage-sensitive bacteria and had no bacteriolytic activity. It was relatively heat resistant, requiring treatment at 90 ° C. for 30 minutes for inactivation. On this basis Burnet suggested that it might be a polysaccharide. DeMars, Luria, Fisher, and Levinthal (1953) demonstrated the presence of similar specific soluble substances in bacteria infected with coliphages T2, T4, or T5. The substances were liberated from the infected bacteria during the latent period by premature lysis, and were separated from phage particles and bacteria by ultrafiltration through collodion filters. The quan- tity of ultrafiltrable specific antigen increased during the latent period. What is probably the same type of substance was lib- erated from bacteria infected with phage T2 or T6, in which normal phage development had been prevented by the addition of proflavine (see also DeMars, 1955; Barry, 1954). The ultra- filtrable fraction had the same serological specificity as the in- fecting phage. At equivalent antibody-blocking activity the ultrafiltrable fraction from T2 lysates had roughly the same complement-fixing activity as intact virus (Lanni and Lanni, 1953). Crude T2 lysates contain a second noninfectious antibody- blocking component, which appears to be intermediate in size between the ultrafiltrable component and the infectious virus particle (Lanni and Lanni, 1953; DeMars, 1955). The latter reference should be consulted for details of the measurement of antibody-blocking agents and for a description of their intra- cellular development. Actually, the antibody-blocking meas- urement may be regarded as an abbreviated absorption experi- ment, in which the absorbing agent and its attached antibody are allowed to remain in the system during subsequent measure- ment of residual neutralizing activity. Lanni (1954) has found that at least one-half of the total phagc-specific complement-fixing activity of T5 lysates is ANTIGENIC PROPERTIES 109 associated with a noninfectious fraction which sediments more slowly than the virus. Efforts to isolate phage antigens by degradation of the virus ])article have met with partial success (Lanni and Lanni, 1953; DeMars, 1955; see also Hershey, 1955). 9. Kinetics of Neutralization It had been recognized since the work of Prausnitz (1922) that the neutralization of phage activity by antiserum was a relatively slow process, and that a small proportion of the phage population resisted inactivation by antibody. There seems to have been no further attempt to study the kinetics of phage neutralization until the work of Andrewes and Elford (1933a). These authors systematically varied the concentrations of antiserum and of phage in the reaction mixture, and determined the decrease in the number of infectious phage particles as a function of time. The "percentage law" is a summary of their experimental results: "over a very wide range a given dilu- tion of serum neutralized in a given time an approximately constant percentage of phage, however much phage there was present." This relationship held from concentrations of a few hundred phage particles per ml. up to at least 10^ per ml., and for serum concentrations from 10~^ to undiluted. Although this relationship between phage and antibody is taken for granted by phage workers today, it was very startling at the time of its discovery, and its implications are still not understood or ap- preciated by many immunologists and virologists. Burnet, Keogh, and Lush (1937) presented kinetic measure- ments for the inactivation of various phages by antiserum. In most cases the proportion of active virus decreased logarith- mically with time after a slight initial lag. The rate of in- activation was directly proportional to the concentration of antiserum. The inactivation was assumed to be the result of a collision between a phage particle and an antibody molecule. Since the reaction is usually carried out in the presence of a large excess of antibody, phage neutralization does not decrease 110 BACTERIOPHAGES the concentration of antibody perceptibly. Therefore, the kinetics is that of a first-order reaction, represented by the equa- tion: -dPidt = KP/D Integration gives the very useful result: 23D K= -y- logio (Po/P) in which A' is the velocity constant with the dimension minute ~\ D is the dilution of serum (100 for serum diluted 1/100, etc.), / is the time in minutes, Pq is the phage assay at zero time, and P is the phage assay at time t. Burnet found that the kinetics of neutralization of phage CI 6 agreed satisfactorily with this equation at serum concentrations from V30 to ^/3o,ooo• Phage neutralization was logarithmic to an inactivation of more than 99 per cent, but then the rate slowed, the survivors appearing to be more resistant. Most of the phages studied by Burnet behaved in this manner, including a streptococcal phage and a staphylococcal phage as well as various immunological types of enterophages. However, he noted that the neutralization of coliphages in serological group 3 deviated markedly from a first-order reaction, and that a relatively large proportion of the phage population was resist- ant to inactivation by even high concentrations of antiserum. Delbriick (1945b) found that the neutralization of T2 and T7 obeyed the equations above, but that neutralization of Tl did not. The rate of inactivation of Tl decreased progressively throughout the course of the reaction. Similar anomalous behavior has been observed with phages related to T5 (Adams, 1952), with the M phages of B. megaterium (Friedman and Cowles, 1953), and with D20, a relative of Tl (Adams and Wade, 1955). The value of K, the velocity constant, is a characteristic of a particular lot of serum and may vary from one rabbit to another or from one bleeding to another from the same animal. The K value is a convenient measure of the neutralizing potency of ANTIGENIC PROPERTIES 111 a serum, a high A' value indicating a high titer. Once the A' value of a sample has been determined, it can be substituted in the equation and used to calculate the dilution, D, of this serum required to inactivate any desired fraction of the phage in any given time. The range of inactivation over which the equation is applicable must be determined by experiment. For most purposes an inactivation of 90 to 99 per cent is ade- quate and within the range of first-order kinetics. Any phage-antibody system which follows first-order kinetics will necessarily obey the percentage law of Andrewes and Elford. The law is more general, however, since it applies also to phages in serological group 3, which do not follow first-order kinetics. Hershey, Kalmanson, and Bronfenbrenner (1943a) found that the inactivation of phage T2 by antiserum followed first- order kinetics, the K value being independent of D, t, and Po up to a concentration of about 10^ phage particles per ml. Above this concentration, aggregation of phage particles by antiserum resulted in an increased A' value. Kalmanson, Hershey, and Bronfenbrenner (1942) studied some of the variables influencing the rate of neutralization of phage T2 by antibody. A particularly careful study of the reaction using highly diluted antiserum confirmed the initial lag in the neutralization that had been noted by Burnet. Analy- sis of the convexity of the inactivation curve suggested that between 2 and 3 antibody molecules must react with each phage particle on the average for neutralization to occur. Neutralization of T2 by antibody thus appeared to be a multi- hit process. However, it is now known (Sagik, 1954) that a large fraction of the phage particles in certain stocks of T2 are unable to form plaques when plated directly, but become in- fective during treatment with antiserum. The inhibited par- ticles may also be activated in other ways, for example, by heating or by lowering the salt concentration. Activated stocks showed no sign of deviation from first-order neutralization kinetics (Cann and Clark, 1954). Since allowance must be made for genetic differences in the lines of T2 employed by 112 BACTERIOPHAGES different workers and for differences in antisera, these findings do not necessarily explain all examples of multi-hit neutraliza- tion kinetics (see Park, 1956). The temperature coefl^cient for the velocity constant A' was determined by Burnet, Keogh, and Lush (1937) for phage CI 6. The K value at 0, 37, and 45° C. was 45, 930, and 1350 min- utes"'^ respectively, the Qio thus being about 2.1. Similar studies were made by Kalmanson, Hershey, and Bronfenbrenner (1942) with the serologically related phage T2. The A' value was found to be 0.05 seconds"^ at 0° C. and 0.7 seconds"^ at 37° C, corresponding to a Qjo of about 2. The Arrhenius constant was determined by Hershey (1941) to be about 11,000 cal. per mole. The meaning of these estimates has been placed in doubt by the finding of Cann and Clark (1954) that the neu- tralization of activated T2 (see above) shows a Qio of only 1 .4, suggesting that the process is limited solely by diffusion. How- ever, their measurements, unlike earlier ones, were made using media of low salt concentration. Essentially all serological work with viruses has been done with antiserum diluted either in physiological saline or in broth containing added salt because such diluents are conventional in serology. Recently Jerne (1952) and Jerne and Skovsted (1953) reported that physiological diluents actually inhibited phage neutralization, monovalent cations above 10"^ M and divalent cations above 10~^ M being inhibitory. Optimum rates of inactivation were obtained in \Q~^ M NaCl containing a few jug. per ml. of certain proteins which appeared to act as cofactors. Highly diluted normal serum, ^gg white, and crys- talline lysozyme were effective as cofactors. An anti-T4 serum with a K value of 600 minute"^ in broth gave under optimal conditions a A value of 100,000 or more. The kinetics were first order to a phage survival of 10^^ to 10~^. It has long been recognized that salt concentration has a marked effect on reactions involving colloidal particles because of surface charges, but no one had anticipated such an astonishing effect on the rate of the antigen-antibody reaction. Additional ANTIGENIC PROPERTIES 113 data and discussion of these effects are given by Cann and Clark (1954, 1955). Hershey (1941) discussed the absolute rate of the phage- antiphage reaction, giving theoretical equations from which the maximum rate could be calculated. The calculations and some of the discussion in this paper are incorrect because of erroneous concepts of the nature and size of phage T2 current at that time. However, recalculation using Hershey's methods and more modern data indicates that the actual inactivation rates observed by Jerne are of the same order of magnitude as the maximum rate attainable assuming that every collision resulted in antibody attachment. Control of the ionic environ- ment and the use of cofactors in Jerne's experiments resulted in increasing the number of effective collisions between anti- body molecules and susceptible sites on the phage particle to nearly the theoretical limit. A theoretical analysis of the neutralization of T4 is given by Mandell (1955). 10. Anomalous Neutralization Behavior A fact which puzzled Andrewes and Elford (1933a) was that neutralization of their phages proceeded rapidly until 90 to 99 per cent of the phage was inactivated and then slowed down rather abruptly, the residual phage activity appearing to be very resistant to neutralization. They demonstrated that this could not be due to exhaustion of antibody and hence must be due to inhomogeneity of the phage population with respect to susceptibility to inactivation. That this inhomogeneity was not genetic in origin was demonstrated by subculturing plaques of phage particles which had survived treatment with antiserum. The descendants proved to be normally susceptible to neutraliza- tion by antiserum. The possibility that the inhomogeneity was caused by the action of antiserum itself, combining with the phage particles in such a way as to make them resistant to neutralization, was next considered. This possibility was discarded because the 114 BACTERIOPHAGES resistant survivors of the action of Vioo serum were inactivated by Vio serum at the same rate as untreated phage particles; however, the description of this experiment leaves much to be desired. Attempts to correlate resistance to antiserum with other properties such as heat resistance, abihty to adsorb to host cells, age of phage stock, and plating efficiency on different bacterial hosts were unsuccessful. Delbriick (1945b), dealing with the anomalous neutralization behavior of Tl, suggested that the inhomogeneity might be inherent in the virus population or, more likely, that it might develop during the course of the reaction as a result of attach- ment of antibody at noncritical sites on the phage. Although many samples of anti-Tl serum are like that described by Del- briick in failing to follow first-order kinetics, Hershey (personal communication) found that anti-Ti sera from two rabbits gave excellent first-order inactivation curves. Similarly, Fodor and Adams (1955) found only one antiserum, of some dozen pre- pared against various strains of the T5 serological group, which gave good first-order neutralization curves. It is not clear, therefore, whether the observed neutralization anomalies reflect an inherent heterogeneity of phage preparations, to which not all rabbits respond, or a heterogeneity which is ac- quired during interaction with certain samples of antiserum. Nor is it clear that the responsible serum factor is necessarily specific antibody. According to Hershey and Bronfenbrenner (1952) anomalous neutralization behavior may be an uncon- trolled effect of complement. Recent studies (Van Vunakis, Barlow, and Levine, 1956) show that serum properdin, together with complement, have nonspecific effects which could confuse the study of specific neutralization. Unfortunately, nonspecific serum efl?"ects have not received explicit attention in most studies of anomalous neutralization behavior. 11. Host Dependence of Serum-Survivor Assay It has been known for many years (d'Herelle, 1926) that difl^er- ent assay hosts may give diflfcrent estimates of the fraction of ANTIGENIC PROPERTIES 115 unneutralized phage in a given phage-antiserum mixture. Kalmanson and Bronfenbrenner (1942) analyzed this host effect with a strain of T2 that plated on certain strains of Shiga and Flexner dysentery bacteria with very nearly the same effi- ciency as on E. coli. The host-range properties and suscepti- bility to neutralization by antiserum were independent of the bacterial strain used as host for production of phage stocks. When the kinetics of neutralization were studied by plating replicate samples of serum-phage mixtures on each of the three host strains, it was found that the rate of neutralization was decidedly greater when measured with the Shiga and Flex- ner hosts than when measured with E. coli (see also Friedman, 1954; Tanami and Miyajima, 1956). This was interpreted to mean that a certain fraction of the initial phage population lost its ability to produce plaques on the dysentery strains while retaining its infectivity for E. coli. Subculture of these "mono- valent" survivors on E. coli resulted in a "polyvalent" phage stock with full infectivity for all three host species. Reaction with antiserum had seemingly produced a phenotypic modifica- tion of host range. Bronfenbrenner had previously been able to produce the same kind of modification of host range in this phage by mild heat treatment. The authors suggested (see also Tanami and Miyajima, 1956) that their experiments provide evidence for a serological hetero- geneity inherent in individual phage particles. In the present conceptual framework, this heterogeneity would be a feature of the phage tail. It is not necessary, however, to postulate het- erogeneity, either of antibody-attachment or of host-attach- ment sites, in order to explain the results. The differential host effects could easily be due to minor differences in the phage receptor sites on the several bacterial hosts, such that there would be differences in bonding strength between phage and the different hosts. In the random reaction of antibody mole- cules with uniform antigen sites on the phage particles, the virus surface could be so altered that certain of the particles would adsorb abortively, or even be unable to adsorb, to the 116 BACTERIOPHAGES dysentery host cells, whereas stronger bonds with E. coli would still allow productive interaction. Both of these interpretations attribute the observed effects to antibody-induced phenotypic modification of host range, as opposed to selection from a genet- ically ■ uniform, phenotypically heterogeneous population (see Lanni and Lanni, 1956). Direct evidence that "monovalent" survivors have indeed reacted with antibody in a significant fashion is contained in a brief note by Lanni and Lanni (1957). 12. Inheritance of Antigenic Specificity The perpetuation of serological character of phages during continued subculture indicates, of course, that antigenic struc- ture is genetically stable. On the other hand, the existence of phage strains which show partial serological cross-reaction implies the possibility of mutations affecting serological behavior. Efforts to demonstrate serological mutation in phage, by examining either the survivors of prolonged serum action or mutants selected for nonserological markers, have generally been unconvincing (d'Herelle and Rakieten, 1934, 1935; Burnet and Freeman, 1937) or unsuccessful (Luria, 1945a; Hershey, 1946a, 1946b). Recent reports, however, indicate that some success may have been achieved in isolating serolog- ical mutants of a Xanthomonas pruni phage (Eisenstark, Goldberg, and Bernstein, 1955) and of T5 (Lanni and Lanni, 1956, 1957). Detailed verification of these briefly reported observations could make the antigenic structure of these phages amenable to or- dinary genetic analysis. Several workers, proceeding along a different line, have ob- tained valuable information by examining hybrids obtained from crosses of serologically related phages. Fodor and Adams (1955) have shown that certain T5 X PB hybrids resemble one or the other parent serologically, while others resemble both parents to some extent. By contrast, Streisinger (1956b) found that the genetic determinants of serological specificity in T2 and T4 are allelic and inseparable from the host- range determinants. Progeny from T2 X T4 crosses pos- ANTIGENIC PROPERTIES 117 sessed the serological genotype of one or the other parent, but never of both. Interestingly, the first-cycle T2 X T4 progeny were found to fall into three phenotypic classes : the two parental classes, and a class with mixed serological phenotype (Streisinger, 1956c). The quantitative distribution among these classes was independent of genotype (Chapter XVI). At the moment, it is not possible to give a unique structural interpretation to the particles, obtained in either the T5 X PB or the T2 X T4 crosses, which resemble both parents serolog- ically. It is tempting to draw the conclusion that they con- tain two distinct tail antigens, corresponding to the distinct parental antigens. They could, however, contain neither of these but a third antigen, which cross-reacts with both anti- bodies. In this respect, serological data obtained with cross progeny are no less ambiguous than data obtained by cross- absorption analysis of related phage strains. This ambiguity, clearly stated for phage workers by Hershey and Bronfenbrenner (1952) and Luria (1953a), is conveniently overlooked by most serologists. But while the heterogeneity of the tail antigen has proved elusive, there can be no question of the heterogeneity of neutralizing antibody. The simple demonstration that ex- haustive absorption of an antiserum with heterologous phage removes part but not all of the antibody for the homologous phage proves that the neutralizing antibody molecules in the serum are not all alike. Unfortunately for structural interpre- tations, it is well known that a single antigen can stimulate the production of a variety of antibody molecules. 13. Mechanism of Phage Neutralization We shall briefly review the observations that seem especially relevant to the mechanism of neutralization by antibody. 7. Neutralized phage can be reactivated by treatment with proteolytic enzymes or high frequency sound. Hence, antibody molecules do not destroy the phage, but interfere mechanically with infection by their presence at the phage surface. 2. Although a particle of phage T2 can accommodate a total 118 BACTERIOPHAGES of about 4,600 antibody molecules (Hershey, Kalmanson, and Bronfenbrenner, 1943b, Hershey, Kimura, and Bronfenbrenner 1947), not more than two or three molecules, and possibly only one, actually participate in neutralization, regardless of how many attach. 3. Depending on the conditions of preparation, neutralized phage may absorb to host cells normally, poorly, or not at all (Burnet, Keogh, and Lush, 1 937 ; Hershey and Bronfenbrenner, 1952; Nagano and Mutai, 1954b; Nagano and Oda, 1955). It may even adsorb to one host, which it proceeds to infect, but not to another, for which it has been neutralized (Lanni and Lanni, 1957). A normal particle which has already adsorbed to a host cell may or may not be resistant to neutralization. 4. The tip of the phage tail is the site of adsorption to the host (T. F. Anderson, 1952). 5. Following adsorption of a normal phage particle, the phage DNA enters the host cell, while the phage protein (ghost) remains outside and can be stripped off without interfering with phage reproduction (Hershey and Chase, 1952). Hence, infec- tion consists in the injection of phage nucleic acid into the bac- terium, this process being mediated by the phage tail in a way which is not yet clear. Neutralized phage that succeeds in adsorbing to a host cell does not inject its DNA (Nagano and Oda, 1955; Tolmach, 1957). 6. Neutralizing antibody is just one of several species of anti- phage antibody. There is evidence that its action is confined to the phage tail, while other antibodies react mainly with the phage head (Lanni and Lanni, 1953). These observations clearly indicate that the phage tail is critically important for neutralization. It would appear that attachment of one or a few antibody molecules at appropriate sites on the tail can render the virus noninfectious either by pre- venting adsorption, or by interfering with injection if the virus remains capable of adsorbing. Reaction of adsorbed virus with antibody, after injection of the DNA, would have no consequence for the infection. How antibody manages to prevent injection ANTIGENIC PROPERTIES 119 without preventing adsorption is not known, but mechanisms are not difficult to imagine. Nor is it difficult to understand, in oudine, how attachment of antibody at other sites on the phage might even, by altering the distribution of surface charges or in other ways, facilitate adsorption rather than inactivate the phage (Hershey, Kalmanson, and Bronfenbrenner, 1944- Sao-ik 1954- Jerne, 1956). ' ^ ' CHAPTER IX HOST SPECIFICITY Because the bacteriophages are obUgate intracellular parasites of bacteria, the properties of the host cell must be of major inter- est to the virologist. Of particular concern are those properties which permit a bacterium to serve as a host for phage multiplica- tion or which render it immune to phage attack. As is true of plant and animal viruses, phage strains vary in their host specificity. This specificity was obvious to early phage workers and led to the practical classification of these viruses into such categories as coliphages, staphylophages, streptococcal phages, and cholera phages. Although the utility of such a classification is evident, it has no phylogenetic significance, any more than has a classification of mammalian viruses based on tissue tropisms or host pathology. The role of host specificity in the classification of phages will be discussed in Chapter XXII on phage taxonomy. 1. Microbial Hosts for Phages It would be a difficult task to list all the bacterial hosts for phages that have been recorded in the literature. Such a list would include bacterial strains belonging to the following genera : Pseudomonas, Xanthomonas, Vibrio, Rhizobium, Chromobaclerium, Micrococcus, Gaffkya, Neisseria, Streptococcus, Lactobacillus, Coryne- bacterium, all genera of Enter obacteriaceae, Pasteurella, Brucella, Bacillus, Clostridium, Mycobacterium, Streptomyces, Nocardia, and probably others. There are some bacteria such as the pneumo- cocci and the spirochaetes for which phages have never been found. More highly organized microbial forms such as yeasts, molds, algae, and protoza also seem to be free of phages, unless the kappa substance of paramecia (see Sonneborn, 1955) might be considered a relative of the viruses. 121 122 BACTERIOPHAGES 2. Host Ranges of Bacteriophages D'Herelle (1926) noted that some phage strains were highly specific attacking only certain strains in a single species of bac- terium, while other phage strains possessed "multiple virulences" enabling them to attack bacteria in different genera. He noted one pure phage strain which attacked several species of shigella, several strains of E. coli, and some strains of paratyphoid B. Phage strains capable of attacking only a single bacterial species were termed "monovalent," while phages capable of attacking two or more bacterial species were called "polyvalent" (Kalman- son and Bronfenbrenner, 1 942) . The difficulty with this termi- nology is that the term polyvalent phage was also applied to mixtures of phages prepared for therapeutic use, and it is often difl[icult to tell in the early literature whether a "polyvalent phage" was a "pure line phage" or a mixture of phages. It is evident that the host ranges of some phages cut across the lines of bacterial classification. A Pasteurella peslis phage, for instance, also lyses some salmonella and some shigella species (Lazarus and Gunnison, 1947). Such extensive host ranges are also observed with certain strains of plant and animal viruses. The host range is a useful characteristic for the identification of phage strains. For instance the closely related coliphage strains T2, T4, and T6 are similar in most respects but may be readily distinguished by their host ranges on appropriate mutants of their common host, strain B of E. coli (Demerec and Fano, 1945). Mutant B/6 is resistant to T6 but susceptible to T2 and T4, whereas mutant B/3,4 is resistant to T4 but suscepti- ble to T2 and T6. Similarly B '2 is resistant only to T2. The host range is not an invariant characteristic of a phage strain, however, because the host range can alter as a result of phage mutation, or as a result of phenotypic modification. Also the phage susceptibility of the indicator bacterial strains may be changed by several methods. These modifications of host range will be discussed later. HOST SPECIFICITY 123 3. Host Specificity in Adsorption D'Herelle (1926) demonstrated that centrifugation of a mix- ture of phage and host bacteria, following a brief incubation, re- sulted in sedimentation of 98 per cent of the phage with the bacteria. Because of the generality of this phenomenon, d'Herelle concluded that adsorption is an essential first step in phage multiplication. He also showed that bacteria which are not able to adsorb a given phage cannot serve as hosts for its multiplication. The reverse, however, is not necessarily true. Burnet (1927) in an extensive study of the host ranges of a group of salmonella phages noted a high correlation between the possession of certain heat stable agglutinogens (the O or somatic antigens) and susceptibility to certain bacteriophages. Smooth to rough mutation in these salmonella strains was correlated with a loss of the heat stable antigens and with loss of susceptibility to the phages. At the same time the bacteria acquired sensitivity to a new group of phages, specific for rough salmonellas (Burnet, 1929b). Certain of these rough salmonella strains were able to undergo the reverse mutation to smooth, simultaneously becom- ing resistant to the "rough-specific" phages and reacquiring susceptibility to the "smooth-specific" phages. Although the correlation between antigenic structure and phage susceptibility was not complete, it was evident that susceptibility or resistance to certain groups of phage was dependent on the presence or ab- sence of certain host cell antigens (Burnet, 1930). The suscepti- bility to other phages was dependent on surface substances which apparently were not agglutinogens. An interesting example of phage specificity is seen in the typhoid phage reported by Sertic and Boulgakov (1936b) which attacks H (flagellated) strains but not O (nonflagellated) strains. Phage-resistant mutants are invariably nonmotile, nonflagel- lated, and lacking in H-antigen. The growth of phage-suscepti- ble, motile strains on agar containing phenol results in the loss of flagella and confers phage resistance on the population. The phage is also inhibited in typhoid strains possessing Vi antigen. 124 BACTERIOPHAGES Mutation with loss of Vi antigen makes such strains susceptible to phage. Further examples of a relationship between host antigenic structure and phage susceptibility are found in the dysentery bacteria (Burnet and McKie, 1930), the typhoid bacterium (Craigie and Yen, 1937; Sertic and Boulgakov, 1936a), Vibrio cholerae (Asheshov, Asheshov, Khan, and Lahiri, 1933) and E. coli (Sertic and Boulgakov, 1937). A good review and discussion of the role of bacterial antigens in protecting bacteria against the attack of certain phages is given by Nicolle, Jude, and Diverneau (1953). In most cases the protective antigens seem to act by preventing the phage from adsorbing to the receptor sites by mechanical interference. This may be, however, an over- simplified concept. The evidence for a higher order of chemical specificity in the adsorption of phage to host cell will be pre- sented in Chapter X. Bacterial cultures are not necessarily homogeneous in their ability to adsorb a given phage. Wahl (1953) has called "semi- resistant" those bacterial strains which appear to be heterogene- ous in receptivity. Such strains contain bacteria that are able to adsorb the phage and bacteria that are not. The cell types do not breed true and must therefore be considered either as phenotypic modifications or as genetic changes recurring at high rates, similar perhaps to the diphasic variation of salmonella strains. 4. Host Specificity in Infection Some bacteria are able to adsorb phages very well but fail to serve as host cells for phage multiplication. Lysogenic bacteria, for example, usually adsorb the phage they carry in the latent (prophage) state, but are not affected by it. Lysogenic bacteria are therefore said to be immune to superinfection. Lysogenic bacteria are susceptible to lysis by other phages however. Some phage-resistant bacterial mutants retain the ability to adsorb the phage to which they are resistant. For instance, a mutant of £". coli called B/1 adsorbs coliphage Tl in a reversible HOST SPECIFICITY 125 manner. Its resistance to infection is not due solely to failure of attachment. A different mutant of the same strain of E. coli, called B/1,5, is resistant to phage Tl because of failure of adsorp- tion (Garen and Puck, 1951). There are numerous other refer- ences in the literature to phage-resistant variants that adsorb the phage to which they are resistant. In such cases the mechanism of resistance is not known. Some bacteria adsorb a phage, are killed by it, but fail to liberate viable phage progeny. In such cases the phage simu- lates an antibiotic and the bacterium cannot be considered a host for the phage. Although the individual bacterial cell is killed, the ultimate result may be protection of the bacterial culture because the adsorbed phage particle is destroyed. In some cases a bacterium may either respond in this way to phage attack or may respond normally with phage multiplication depending on environmental conditions such as nutrients or temperature. Literature references to the antibiotic action of phages have been collected by Fredericq (1952b), and "abortive infection" is dis- cussed in Chapter XV. 5. Acquisition of Phage Resistance by Mutation Bacteria may become resistant to phage attack either by mutation or by becoming lysogenic. Since the mechanisms of these processes are distinct, they will be discussed separately. A mutation is a permanent, heritable, modification in some property of a cell. Mutations occur spontaneously during cell reproduction. Since mutation is typically a rare event, m.utants are usually detected by the use of appropriate selective agents in the environment. For instance the presence of phage-resistant mutants in a population of phage-susceptible cells may be de- tected by plating the population with an excess of phage. The phage-susceptible bacteria are destroyed and the phage-resistant mutant cells produce colonies which may be readily subcultured and free of phage. That the mutation occurs spontaneously prior to the addition of the selecting agent has been demonstrated by the fluctuation 126 BACTERIOPHAGES test of Luria and Delbruck (1943), the replica plating technique of Lederberg and Lederberg (1952), and the redistribution test of Newcombe (1949). Methods for the determination of mutation rates have been discussed by Newcombe (1948), Lea and Coulson (1949), and Armitage (1953). Measured mutation rates to phage resistance are of the order of 10 ~^ to 10~^° muta- tions per bacterial division. Mutation to phage resistance may involve any of a number of distinct physiological mechanisms. Resistance to infection cor- related with the failure of phage adsorption has been noted above. Lack of adsorption may be due to the absence of the re- ceptor substance to which the phage particle normally becomes attached. When this receptor substance happens to be a major antigen of the bacterial cell, its loss by mutation may be readily detected by immunological techniques. If the receptor substance is nonantigenic or is a minor antigen, it may be more difficult to discover the reason for failure of adsorption. Failure of adsorp- tion may also be due to the covering of the receptor substance by some other bacterial layer, as in the mutation from rough to smooth in the salmonellas. In this case the acquisition of the O antigen results in loss of susceptibility to the "rough specific" salmonella phages (Burnet, 1930). Mucoid variants of phage- susceptible bacteria are usually phage-resistant because the masses of slime material interfere with phage adsorption (Gratia, 1922). Doubtless other mechanisms may play a role in the acquisition of phage-resistance by mutation, but these have not been studied. Phage-resistant mutants of bacteria very often retain a normal susceptibility to some phages while becoming resistant to others. Such mutant strains are of considerable value in phage research since they can be used as specific indicators to demonstrate the presence of a particular strain of phage in a mixture. The nomenclature used for phage-resistant mutants was origi- nated by Burnet and McKie (1933) and modified by Demerec and Fano (1945). Strain B of E. coli is susceptible to phage-Tl through T7. A phage-resistant mutant isolated by means of HOST SPECIFICITY 127 phage T3 is designated B/3. If this mutant is found to be resist- ant also to phages T4 and T7 it is called B/3, 4, 7. If a subsequent mutation confers resistance also to T6, it is named B/3, 4,7/6, each mutational step being distingished by a bar. Mutations to phage resistance have been extremely important in the development of bacterial genetics, because of the ease with which they can be selected, and because of their great variety. Certain mutations to phage resistance are accompanied by the loss of the ability to synthesize growth factors such as tryptophan (E. H. Anderson, 1946) or proline (E.-L. Wollman, 1947). Selecting for phage resistance in these cases is thus an easy way to obtain biochemical requirements in bacterial strains. Since rates of mutation to phage resistance are easily measured, this type of mutation has been extensively used to test for the ability of various physical and chemical agents to induce mutations (see, for instance, Novick and Szilard, 1951b). Because of the variety of properties which may be associated with resistance to a given phage (resistance to certain other phages, biochemical require- ments, colony type, etc.), often several mutant types can be easily distinguished among all the resistant mutants that are selected with a given phage. One obtains thus a "mutation pattern" which can be used as a test of the specificity of various agents capable of inducing mutations (Bryson and Davidson, 1951). More information on this general subject will be found in the book by Braun (1953). Bacterial mutations involving a change from resistance to sensitivity are more difficult to study because of lack of selective agents for isolation of the mutant strains. An exception is found in the phages specific for rough and smooth strains of salmonella studied by Burnet (1929b), which can be used for selection in both directions. In other cases phage-sensitive variants can be selected because of morphological differences from the phage- resistant parent stocks; for example, in E. coli (Nelson, 1927), Sh. dysenteriae (Arkwright, 1924), and Sh. sonnei (Miller and Goebel, 1949). 128 BACTERIOPHAGES 6. Acquisition of Phage Resistance by Lysogenization The adsorption of a temperate phage to a susceptible bacterial cell may have either of two results; the bacterium may be lysed with release of phage progeny, or the bacterium may be con- verted to the lysogenic condition, in which the cell and its progeny are permanently endowed with the potentiality of pro- ducing phage. This potentiality is controlled by a latent, non- infectious form of the infecting phage, the prophage, which is carried by lysogenic cells and inherited as part of their genetic make-up. To unify the terminology of lysogenic cultures it has been agreed to use the notation of Williams Smith (1951a). In this system the symbol designating the latent phage is placed in parentheses after the sym.bol designating the bacterial culture. For instance if Pseudomonas aeruginosa strain 13 is lysogenic for phage strain 8, the notation is 13(8). Lysogeny is discussed in Chapter XIX. Only a few facts concerning the effects of lyso- genization on host specificity will be mentioned here. It has been noted before that lysogenic bacteria are immune to superinfection by the phage of the type they carry as prophage, although adsorption occurs normally. Burnet and Lush (1936) first realized that the acquisition of phage resistance by lyso- genization is a very different process from the acquisition of resistance by mutation. These authors reported that from 10 to 20 per cent of effective contacts between phage and susceptible bacteria resulted in conversion of the bacteria to the lysogenic condition. The conversion was not attributable to genetic heterogeneity in the bacterial population. Bertani (1953a), using the terrperate enterophage P2, was able to convert up to 40 per cent of the infected cells of strain Sh oi Shigella dysenteriae to the lysogenic form, Sh(P2). The suscepti- ble strain, Sh, is subject to lysis by the 7 coliphages of the T group, and, of course, by P2. The lysogenic strain, Sh(P2), adsorbs all these phages but is resistant to lysis by P2, T2, T4, T5, and T6. Strain Sh(P2) is lysed by phages Tl, T3, and T7 only. In this case conversion of the bacterium to the lysogenic condition results in the acquisition of resistance not only to the HOST SPECIFICITY 129 carried phage, but also to a number of unrelated virulent phages. Several other examples of this phenomenon are known and will be discussed later on. Infection with a latent phage would ap- pear to be a valuable protective mechanism to the bacterium. The intimate details of this acquired resistance to virulent phages are unknown, but must be rather specific since the bac- terium remains susceptible to lysis by certain other phages. Lysogenization may produce phage-resistance in still another way. Salmonella strains of group E2 (somatic antigens 3,15) are lysogenic for a phage which can be demonstrated by means of indicator strains from group Ei (somatic antigens 3,10). Strains of group Ej, which become lysogenic for such a phage, lose at the same time the ability to adsorb the phage (Uetake, Nakagawa, and Akiba, 1955). In this case lysogenization is accompanied by a change in the surface of the host cell, such that adsorption is blocked. The change can be demonstrated serologically, since the lysogenized strains have also lost antigen 10 and gained antigen 15. If such lysogenic strains are exposed to antiserum specific for antigen 15, antigenic variants can be obtained which have lost antigen 15, and show now antigen 10. These variants are not lysogenic any more, and are susceptible to the phage, thus confirming the close association of the lysogenic condition with the change in the cell surface. 7. Phage Mutations Affecting Host Range D'Herelle (1926) considered the extension of host range of a phage preparation to be an adaptive process which he called "acquisition of virulence." His procedure of adaptation was to grow two bacterial strains in mixed culture, add a phage active for one strain and watch for lysis. If lysis was not evident the culture was filtered and the filtrate used to inoculate a fresh mixed culture, this process being repeated until lysis occurred. Then the "adapted" phage was isolated by repeated subculture on its new bacterial host. A better method is to spread the same mixture of bacterial strains and phage on the surface of agar 130 BACTERIOPHAGES plates, because the presence of a host range mutant is promptly made evident as a plaque. Unfortunately the earlier experimenters did not examine critically the properties of the "adapted" phage strains and con- sequently certain of their claims should probably be discounted. For instance d'Herelle claimed to have adapted a staphylococcus phage to lyse a strain of Shiga's bacillus. It is probable that this "adaptation" was the result of phage contamination, especially as the fully adapted phage had lost its lytic power for the staphy- lococcus strain. No such radical alteration of host range has been reported in recent years. What appears to be the first clear description of the isolation of host range variants of a phage was given by Sertic (1929b). His experiments will be briefly summarized using his nomencla- ture for the phage and host cell variants. Cells of strain Fb of E. coli were lysed by strain beta of phage Fez except for a few cells of a phage resistant bacterial mutant called Fbrl. If a lawn of variant Fbrl was inoculated with phage strain beta, iso- lated plaques were obtained with a frequency about 1/1,000 that of the plaques obtained on the parent host strain Fb. By suc- cessive single plaque isolations on strain Fbrl, a pure culture of host range variant beta/rl was obtained, which was able to lyse both bacterial strains, Fb and Fbrl. By inoculating host strain Fbrl with phage variant beta/rl a second phage resistant mutant Fbr2 was obtained. Mutant bacterial strain Fbr2 was com- pletely resistant to phage strain beta, but when inoculated with phage beta/rl yielded a few plaques of a new variant, beta/r2. Sertic pointed out that by this two stage process he was able to obtain a variant phage virulent for a bacterial strain which was completely resistant to the original phage. Many reports of host range mutants of bacteriophages have since appeared, many of them dealing with adaptation of phages to attack new host strains in the development of phage typing methods. In most cases the authors have presented no evidence to prove that the "adapted" phage was actually derived from the supposed parent. HOST SPECIFICITY 131 The significance of this omission was demonstrated by Roun- tree (1949b) in an investigation of the serological relationships of a number of staphylococcal phages. She found that typing phages 42C and 42D were serologically unrelated to phage 42B although all were supposed to have been derived from a common source, strain 42. Strain 42 had been adapted to host strain 36 to give phage strain 42A ; 42A was adapted to host strain 1 307 to give 42C, which on adaptation to host strain 1363 became 42D. On investigation Rountree found that host strain 36 was lysogenic, carrying a latent phage which was capable of lysing host strain 1307. This latent phage from host strain 36 proved to be serologically related to phage strain 42C and 42D. There- fore the phage strains 42 C and 42D were derived from a con- taminating phage from a lysogenic host strain used for passage and were not host range mutants of strain 42. Since the fre- quency of lysogenic strains is high in many bacterial species, this source of confusion must be kept in mind. Serological rela- tionship is the minimum evidence required to demonstrate derivation by mutation, and resemblance in other properties should be looked for as well. These requirements were clearly stated by Craigie and Felix (1947) in discussing the standardiza- tion of the Vi typing phages for S. typhi. Luria (1945a) in a detailed study of host range variants of phages Tl and T2 applied the fluctuation test to demonstrate the spontaneous mutational origin of these variants. The mutant phage particles proved to be indistinguishable from their parents in serological specificity and in growth characteristics on the common host strain B of E. coli. They appeared to differ only in host range. The frequency of these host range mutant virus particles in small independent culture lysates was of the order of 10-8. Hershey (1946b) demonstrated the existence of sev^eral inde- pendent host range mutations of phage T2 and that in mixed infection experiments genetic recombination between host range and plaque type mutants could occur (Hershey and Rotman, 1949). In further studies Hershey and Davidson (1951) found 132 BACTERIOPHAGES that host range mutations of phage T2 occurred at two different gene loci, and that three different mutant alleles might occur at one of these loci. The host range mutants of phages Tl and T2 just mentioned have gained the ability to adsorb to indicator strains of bacteria. Phage mutants are also known that are able to overcome the immunity produced by the lysogenic condition, which has a different basis. The temperate phage P2, mentioned in the preceding section, does not lyse lysogenic cells Sh(P2), although it does absorb onto them. Rare mutants of phage P2 are found, which are able to lyse Sh(P2) cells. This mutation is commonly associated with the loss of the ability to establish lysogeny in susceptible cells (Bertani, 1953a). In a similar case involving the temperate coliphage lambda, it is possible to demonstrate genetic recombination between the factor imparting virulence and other mutations affecting the plaque type (Jacob and Woll- man, 1954). These studies have demonstrated conclusively the mutational nature of some examples of host range variation in bacteri- ophages. It has become evident in recent years that host range modifications may occur by mechanisms other than mutation of the virus. These phenotypic modifications of host range will be discussed next. 8. Phenotypic Modification of Host Range A phenotypic modification of the properties of an organism implies a nonhereditary change, usually attributable to some environmental influence. A phenotypic modification is re- versible in the immediate descendants of the organism in the absence of the causal environmental factor. Numerous examples of phenotypic alterations in phages have been discovered in recent years. Some examples involving host range are described below. a. Phenotypic Mixing The first clear example of a phenotypic modification in host range was brought to light in consequence of an observation re- HOST SPECIFICITY 133 ported by Delbruck and Bailey (1946). When bacteria were mixedly infected with the closely related phages T2 and T4, up to 90 per cent of the bacterial population liberated both T2 and T4 phage particles when plated on the mixed indicator bacteria B/2 and B/4. However when plated on B/4, the indicator for phage T2, only a small proportion seemed to yield T2. The authors concluded that the liberated T2 particles had a low efficiency of plating on B/4 as compared with that on mixed indicator. This paradoxical result was explained by Novick and Szilard (1951a) who found that bacteria mixedly infected with T2 and T4 might liberate three kinds of phage particles, typical T2. typical T4, and particles with the genotype of T2 but the host range phenotype of T4. The latter particles adsorbed to host strains B and B/2 but not to B/4. After one passage through either B or B/2, the phage reverted to its T2 phenotype and henceforth would multiply on B/4 but not on B/2. This ex- plains why such particles produced plaques on mixed indicators or on strain B, but not on either B/2 or B/4 by itself. In this case the T4 phenotype was the result of the presence of particles having the genotype of T4 in the mixedly infected bacteria. Since the T4 host range characteristic is replaced by the T2 phenotype on subculture, these particles must be genotypically T2. b. Host-Controlled Variation Another type of phenotypic host range modification was dis- covered by Luria and Human (1952). This involves a phage resistant mutant of E. coli called B/3,4,7 (2,6). This mutant may be isolated from cultures of E. coli strain B by the selective action of T3, T4, or T7 but not by T2 or T6. It is resistant to T3, T4, and T7 because of failure of these phages to adsorb, but it's resist- ance to T2 and T6 is of a different kind. When concentrated T2 and T6 phage stocks are plated with this bacterial mutant, a clear, sterile area results; but if the phage stocks are progres- 134 BACTERIOPHAGES sively diluted, this effect is lost without ever passing through a level of dilution at which discrete plaques are formed. The phage particles adsorb normally to the mutant bacteria, and the infected bacteria lyse after the usual latent period. They also liberate phage particles which, however, cannot multiply fur- ther. After considerable search it was found that strain Sh of Sh. dysenteriae was a suitable indicator for the phage particles liberated on lysis of B/3,4,7(2,6). After passage through strain Sh the phage progeny had the properties of normal T2 or T6 phage. In summary, phage T2 infects bacterial strain B '3,4,7(2,6) to produce a phenotypically modified phage progeny symbolized as T2*. Phage T2* does not produce plaques on B/3,4,7(2,6), or on B, but does produce plaques on strain Sh. Phage T2* after passage through strain Sh reverts to phage T2. Phage T2* adsorbs to strains B, B/4, and B/6 and kills them, but the infec- tion is abortive. These experiments indicate that the phenotypic change in T2* is different from that found in the phage liberated from the bacteria mixedly infected with T2 and T4. In the latter case the modification is in the specificity of adsorption. In the case of T2* the adsorption specificity is the same as that of T2, so the phenotypic modification must involve some stage in multi- plication subsequent to adsorption. It may be noted that the changes in this instance are not adaptive. An adaptive type of host-induced modification was reported by E. S. Anderson and Felix (1952), involving the Vi typing phages of S. typhi. The original Vi II phage strain, phage A, when tested at the limiting test dilution gives confluent lysis with Type A typhoid strains only. When tested at higher concentra- tions with a bacterial lawn of a Type C strain, a few plaques may be found. When these plaques are picked and subcultured on Type C bacteria, an "adapted" typing strain is obtained, which at the limiting test dilution will give confluent lysis with both Type C and Type A typhoid strains but only partial or no lysis with all other strains of typhoid bacteria. So far, the behavior is HOST SPECIFICITY 135 typical of that of host range mutants and these adapted typing phage strains were referred to as mutants by Craigie (1946). The significant observation of Anderson and Felix is that when the "adapted" phage C is plated on a lawn of Type A typhoid, it reverts completely to the characteristics of phage A. For this and other reasons, Anderson and Felix designated this adaptive re- sponse "phenotypic modification of host range." The general phenomenon is very common and exhibits other remarkable features that are discussed in Chapters XVI and XXI. 9. Summary Most taxonomic groups of bacteria include strains which are susceptible to some bacteriophage. The host range of a given phage is often restricted to a single bacterial genus but exceptions to this rule are common. Changes in the surface of a bacterium may block the adsorption of a phage otherwise able to multiply in such a host. These changes may occur by mutation, and are often associated with antigenic changes. Multiplication of a phage in a given host may be blocked at a stage past adsorption. This is often a consequence of the lysogenic condition of the host. Both types of blocks may be overcome by mutation of the phage. The host range of the phage may be affected by the conditions of its growth in a way not involving mutation and selection. One class of such effects is called phenotypic mixing. It is observed when two closely related phages differing in adsorp- tion-specificity multiply in the same bacterium, and affects pri- marily adsorption-specificity. The effects are promptly reversed when the phage multiplies in pure culture. A second class of such effects is called host-induced modification, and affects host range independently of adsorption-specificity. It is conditioned by the genotype (including carried prophages) of the bacterial host. This class of effects is likewise reversed by transfer of the phage to another host. CHAPTER X ADSORPTION OF PHAGE TO HOST CELL The first step in the infectious cycle is the adsorption of the phage particle to the host cell. Adsorption may be defined as attachment of phage particles to bacterial surfaces so that phage and bacteria will sediment together. As described in Chapter IV, phages adsorb to their host cells by the tips of their tails. The attachment is made to specific receptor substances on the cell surface, often parts of the cell wall (Chapter IX). Adsorption is usually measured by centrifuging a mixture of bacteria and phage, and counting the phage in sediment or supernatant fractions, or both, by means of plaque counts. The reduction in plaque count in the supernatant fluid can be used to measure adsorption in excess of 30 per cent ; for smaller frac- tions the measurement by difference is too inaccurate, and the sedimented phage must be counted directly. If the fraction adsorbed is very small, antiserum must be used as well (Hershey and Davidson, 1951). Other methods of measurement do not require centrifugation. One such method depends on the fact that infected bacteria can be destroyed by agents such as chloro- form that do not inactivate free phage particles (Fredericq, 1952a). Another makes use of antiserum to inactivate un- adsorbed phage (Delbriick, 1945b). Adsorption can also be measured in terms of the fraction of bacteria killed by the phage. Many of these methods, however, depend on steps in the infective cycle taking place subsequent to adsorption as defined above. It cannot be assumed, without test, that any other measurement will yield information equivalent to the reduction in plaque count of the supernatant fluid on centrifugation of the adsorption mixture. Moreover, the results of this measurement may vary, 137 1 38 BACTERIOPHAGES under conditions in which adsorption is reversible, depending on whether the mixture is diluted or not before centrifugation. 1. Kinetics of Adsorption Krueger (1931) investigated the adsorption of a phage to liv- ing and heat-killed staphylococci. He found that with an excess of bacteria, the adsorption follows the kinetics of a first-order reaction. The rate of adsorption is -dP/dt = kBP (1) in which k is the velocity constant, B is the bacterial concentra- tion, and P is the phage concentration. The appropriate solu- tion yields ^ = ^log^ (2) in which Pq is the concentration of unadsorbed phage at the beginning, and P at the end, of the time interval t. Krueger found k to be 2.4 X 10"!" ml. per minute either with living or heat-killed bacteria. Equations (1) and (2) are generally applicable to adsorption of phage, as indicated by further work described below. In them B can be considered constant provided the ratio of phage to bac- teria is kept sufficiently low so that the bacterial surfaces remain effectively unaltered. This explains why the two-body inter- action can be described by first-order kinetics. Krueger (1931) also attempted to study adsorption-equilib- rium with results that are not now interpretable owing to ambiguities of his assay method (Chapter III). A very extensive study of phage adsorption was made by Schlesinger (1932a) using coliphage WLL. This phage was re- ported by Burnet (1934a) to be related to phage CI 6 and hence is also related to T2. The bacterium used was E. coli 88, either as a broth culture of living cells, or as a cell suspension killed by heating at 70 ° C. for one hour. Adsorption was carried out at 37 ° C. In these experiments the amount of unadsorbed phage ADSORPTION OF PHAGE TO HOST CELL 139 decreases as an exponential function of time until about 95 per cent of the phage is adsorbed. The adsorption rate then de- creases and, when killed bacteria are used, an unadsorbed residue is obtained amounting lo 10~^ of the initial phage popu- lation. This unadsorbed fraction does not adsorb if additional heat-killed bacteria are added, indicating that failure of adsorp- tion is not due to saturation of bacterial receptors, nor to a re- versible equilibrium. The unadsorbed fraction adsorbs slowly to living bacteria. After adsorption of 90 to 99 per cent of the phage to heat- killed bacteria, the residual phage was titrated before and after removal of the bacteria by centrifugation. The plaque counts are the same in either case, indicating that adsorption is essen- tially irreversible. However, after prolonged periods of adsorp- tion it is found that a small fraction (10~^) of the phage popula- tion can be eluted by washing. The phage population therefore consists of three kinds of phage particles: those that adsorb irreversibly, those that adsorb reversibly, and those that do not adsorb at all to heat-killed bacteria. In his next paper Schlesinger (1932b) demonstrated that the adsorption of at least 95 per cent of the phage population obeys the kinetics described by equations (1) and (2) above. The observed velocity constant k is independent of B, P, and t over ranges of B from 3 X 10*^ per ml. to 7 X 10^ per ml., of Pq from 6 X 10^ per ml. to 6 X 10^ per ml., and of ^from 2 minutes to 12 hours. The adsorption rate on living bacteria is 2.6 times faster than on heat-killed bacteria. Schlesinger interpreted this difference as being due to destruction of somewhat more than half of the receptor sites during the heat-killing of the bacterial culture. Schlesinger determined the adsorption capacity of heat-killed bacteria for phage by increasing the phage to bacterium ratio to several hundred. The maximum adsorption capacity is 140 phage particles per bacterium. If Schlesinger's explanation for the reduced adsorption rate is correct, the adsorption capacity of living bacteria might be 140 X 2.6 or 360 phage per bac- 140 BACTERIOPHAGES terium. Delbriick (1940a) using a different coliphage and a different strain of E. coli found that the saturation value for a culture in the logarithmic growth phase was 250 phage per bac- terium, and Watson (1950) found a maximum value of 300 phage per bacterium for the adsorption of T2 to heat-killed E. coli strain B. Schlesinger (1932c) adapted the coagulation theory of von Smoluchowski to the kinetics of phage adsorption, and derived the equation : -dP/dt = AttDRBP (3) in which —dP/dt is the rate of adsorption, D is the diffusion con- stant of the phage particle, R is the radius of a sphere of the same surface area as the bacterium, and the other symbols are as in equation (1). By combining equations (1) and (3) he ohtained the equation k = AttDR (4) Equation (4) contains the assumptions that every collision between phage and bacterium results in irreversible adsorption, that the bacteria and surrounding fluid are stationary, and that the phage particle itself is dimensionless. In spite of obvious defects, it gives a rough estimate of what the maximum possible rate of adsorption ought to be. Schlesinger calculated the diffusion constant from his observed adsorption-rate constant by means of equation (4) : 3.4 X 10-" 6.3 X 10-'* cm.- sec. -1 At X 4.3 X 10-^ According to Taylor, Epstein, and Lauffer (1955), the diffusion coefficient of phages T2 and T6, related to WLL, is about 3.5 X 10-^ cm. 2 sec.-^ at 20° C, in reasonable agreement with Schlesinger's indirect measurement at 37 ° C. This means that the observed adsorption-rate constant of equation (2) is in agree- ment with the theory represented by equation (4), as found also by Delbriick (1940a), Puck, Garen, and Cline (1951), and Stent ADSORPTION OF PHAGE TO HOST CELL 141 and Wollman (1952). These results seem to show that a large fraction of the collisions between phage and bacterium lead to successful attachment. Tolmach (1957) and Hershey (1957) question this interpretation, and we conclude here only that adsorption-rate constants in the range 10"^ to 10~^ per minute can be expected under optimal conditions. The optimal condi- tions depend on size and condition of the bacterial cells (Del- briick, 1940a; Hershey and Davidson, 1951), and numerous environmental factors to be discussed below. 2. The Ionic Environment Many reports in the early literature of the effect of salts on phage lysis have been summarized by d'Herelle (1926) and by Sertic (1937). For example, da Costa Cruz (1923) reported that bacteria grown in salt-free Witte's peptone were not lysed by phage, but that the addition of sodium chloride or calcium chlo- ride resulted in lysis. Lisbonne and Carrere (1923) confirmed this salt effect with three different phages. The addition of sugars had no effect. These authors demonstrated that the effect of the electrolytes was on the adsorption of phage to host cell ; if phage and bacteria were mixed in salt-free peptone, they could be readily separated by centrifugation. In the presence of salts, the phage speedily became attached to the bacteria and could not be separated by centrifugation. An interesting paper by Hershey, Kalmanson, and Bronfen- brenner (1944) gave a preview of later developments in this field. These authors found that the relative efficiency of plating (EOP) of phage T2 varies markedly with the concentration of electro- lyte added to the agar medium. The EOP is 0.001 or less in the absence of added electrolyte, 0.1 in 0.03 M sodium ion, and 1.0 in 0.2 M sodium ion. Similar results were obtained with potas- sium, lithium, and ammonium ions. Divalent cations were not tested at comparable concentrations. The EOP was inde- pendent of the nature of the anions present. In an attempt to determine whether the cation effect was on adsorption, phage adsorption was studied with and without 142 BACTERIOPHAGES added salt, by means of the centrifugation technique. The sedi- ments and supernatants were assayed on both salt-free agar and on agar containing the optimal amount of salt. The fraction of phage adsorbed was calculated from the plaque counts on agar containing salt. It was found that appreciable adsorption of phage occurs in salt-free tryptose broth. The striking observa- tion is that phage adsorbed to its host cell in the presence of salt has a high efficiency of plating on salt-free agar, while phage absorbed in the absence of salt has a very low EOP on salt-free agar, as does unadsorbed phage. It would appear from these experiments that phage adsorbed in salt-free broth does not initiate infection of the host cell and may be reversibly adsorbed. The possibility of reversing adsorption was tested by adsorbing the phage in the absence of salt, sedimenting the bacteria, and washing the sediment with broth to remove free phage. The sediment was then diluted in salt-free broth, incubated for 20 minutes, and centrifuged again. Assays on both supernatant and sediment indicated that about 50 per cent of the adsorbed phage had been eluted. Phage adsorption in salt-free broth is partly reversible and infection of the host cell by the adsorbed phage does not occur in the absence of salt. Hershey, Kalmonson, and Bronfenbrenner (1944) also found that phage T2 treated in the presence of salt with very dilute anti-T2 serum showed an increased rate of adsorption under suboptimal conditions, and a greatly increased EOP on low-salt agar. This effect may be related to the activation of cofactor- requiring strains of T4 by one of the anti-T4 antibodies (Jerne, 1956). The observations with T2 might be well worth reinvesti- gating but are now of doubtful meaning owing to possible com- plications caused by inhibitors of adsorption often present in phage stocks (Sagik, 1954). The same could be said, in fact, of much of the work reported in this chapter. The experiments on reversible adsorption described above were abandoned by the authors because it appeared that the transition from reversible to irreversible adsorption could not occur on addition of salt, and therefore that reversible adsorption ADSORPTION OF PHAGE TO HOST CELL 143 could not be a step in the infective process (Hershey, personal communication). Further work described below has led to the contrary interpretation. There were no further quantitative studies of the effect of salts on phage adsorption until the work of Puck, Garen, and Cline (1951). These authors studied the rates of adsorption of various phages to E. coli under a variety of environmental conditions, and reached the conclusion that the initial step in phage adsorp- tion is the establishment of electrostatic bonds between appro- priate configurations of ionic charges on the two bodies. In these studies the first-order velocity constants are used to characterize the adsorption rates as the environmental conditions are changed. With phage Tl in buffered salt solutions at 37 ° G. the optimum salt concentration for phage adsorption is about 0.01 M for salts of univalent cations and 0.001 M for salts of divalent cations. The rate of adsorption is depressed markedly by higher or lower concentrations. The maximum velocity constant is the same in simple salt solutions and in broth, sug- gesting that no cofactors other than salts are required. Puck, Garen, and Gline (1951) adopted the hypothesis, consistent with their results, that the efficiency with which collisions between phage particles and cells lead to attachment is controlled by electrostatic charges on the colliding bodies, and that cations in the medium afTect adsorption bv influencing these charges. 3. Organic Cofactors T. F. Anderson (1945a) discovered that some phages are un- able to adsorb to their host cells unless certain organic compounds are present in the environment. He compared plaque counts of phage stocks on the usual nutrient broth agar with counts on agar prepared from ammonium lactate and other salts. Coli- phage T2 gives the same counts on the two media, whereas counts of T4 and T6 are much lower on lactate agar than on broth agar. By testing various substances that might be present in broth he found that the addition of tryptophan or phenyl- anlaine to the svnthetic medium results in a large increase in the 144 BACTERIOPHAGES efficiency of plating. Tryptophan is more effective. Anthranilic acid, indole, and indole-3-propionic acid are inactive. Anderson showed that the role of cofactor is to allow adsorption to occur. In a later paper T. F. Anderson (1945b) reported that Bz-3- methyl tryptophan, which inhibits growth of bacteria and phage, can substitute very efficiently for tryptophan as an adsorption cofactor. T. F. Anderson (1946) listed a number of organic sub- stances which had varying degrees of cofactor activity. Substi- tutions or changes in the amino or carboxyl groups of tryptophan destroy all cofactor activity, whereas quite extensive changes are permissible in the remainder of the molecule. Replacement of the indole structure by pyridine, benzene, or thiophene results in a cofactor activity about 1 per cent of that of tryptophan. Delbriick (1948) reported that DL-norleucine had about 0.3 per cent of the activity of L-tryptophan. The remarkable structural specificity of the cofactor is indicated by the fact that L-trypto- phan is the most highly active substance yet tested but o-trypto- phan is inactive. By kinetic methods, T. F. Anderson (1948a) showed that cofactor reacts with the phage particles rather than with the host bacteria. This process is termed "activation." Anderson esti- mated that about six molecules of tryptophan are involved in the activation of each phage particle. There is a rather high tem- perature coefficient for activation with a temperatu»'e optimum at about 35 ° G. When activated T4 is diluted in a tryptophan- free medium, the phage very rapidly become "deactivated." It is evident that phage T4 reacts in a reversible manner with tryptophan to form an activated complex capable of adsorbing to the host cell. A small portion of the particles of phage T4 in certain stocks are able to form plaques when plated on tryptophan-free agar. When these plaques are picked and subcultured they give rise to strains which can adsorb in the absence of tryptophan. Thus the ability to adsorb lo the host cell in the absence of cofactors is a genetically determined characteristic of ihc j^hage (T. F. Ander- son, 1948b). ADSORPTION OF PHAGE TO HOST CELL 145 Delbriick (1948) isolated additional cofactor-requiring vari- ants of T4. One of these requires calcium ion in addition to tryptophan for adsorption to take place. The calcium require- ment can not be met by magnesium ion. A second strain ad- sorbs well in the presence of tryptophan without either Mg or Ca ions. The adsorption of both tryptophan-requiring strains is inhibited by the presence of indole in the medium. Anderson's tryptophan-requiring stock, however, is indifferent to indole. The sensitivity to indole also varies among different lines of T2, none of which requires tryptophan. The reversible activation of phage T4 by tryptophan and the adsorption of the activated complex to the host cell make an interesting kinetic problem which was analyzed in a series of beautiful papers by Wollman and Stent (1950) and Stent and Wollman (1950, 1951), confirming and extending the previous work ofT. F. Anderson. The deactivation of the T4-tryptophan complex is a first-order reaction. The rate of activation by tryptophan is proportional to the fifth power of the tryptophan concentration at low concentrations, but is independent of tryptophan concentration at high concentrations. This indi- cates that five tryptophan molecules must react at a single site to form an activated phage particle. The limiting rate of activa- tion at high concentrations is attributed to a step involving re- arrangement of the five adsorbed tryptophan molecules into a specific and relatively stable configuration. The rate of deactivation is markedly sensitive to tryptophan concentrations below the level at which measurable activation occurs. This is explained by the assumption that deactivation involves loss of a single tryptophan molecule from the organized site leaving the remaining four molecules in a specific configura- tion. The active configuration can then be restored by the addi- tion of a single molecule at a rate much faster than that charac- teristic of the primary activation. The fraction of the phage population activated by a given con- centration of tryptophan decreases very rapidly with decreasing temperature. This effect can be overcome by increasing the 146 BACTERIOPHAGES tryptophan concentration, so that all phage particles are acti- vated l)y concentrations above 100 ;ug. per ml. independently of temperature. The rate of activation is also markedly tempera- ture-dependent at low tryptophan concentrations but not at high concentrations. These temperature characteristics are consistent with the model proposed. The work of Stent and Wollman yielded a satisfactory kinetic model for activation by tryptophan but failed to clarify the nature of the activation process in relation to adsorption. Sato (1956) found that activation could also be brought about by urea under circumstances suggesting that the underlying process resembled protein denaturation. By extension of his idea it might be guessed that all or many phages require a similar acti- vation, differing only with respect to the nature of the requisite cofactors, organic and inorganic. Williams and Fraser (1956) showed that the cementing substance for adsorption is composed of a number of fibrillar structures in the tail of T2, which can be induced to "unwind" under conditions not yet clarified. Pos- sibly activation involves a reversible orientation of these fibrils, and possibly the pH-dependent transformation of T2 described by Taylor, Epstein, and Lauffer (1955) (Chapter IV) is a re- lated phenomenon. Jerne (1956) found that one of several antibodies produced in response to immunization with T4 can permanently activate cofactor-requiring strains of this phage. His discussion of this phenomenon is very suggestive in connection with the ideas discussed above. Probably a little more work on requirements for adsorption will elucidate the nature of the activation of T2 and related phages by salts and other cofactors. A paradoxical fact discovered by Anderson (1945a), that in- fected bacteria form plaques on synthetic agar although the free phage T4 particles do not, was explained by Wollman and Stent (1952). They found that T4, immediately after release from lysed bacteria, is able to adsorb to bacteria without added co- factor. This activated state is transient but deactivation is slow. The natural cofactor responsible for "nascent" activity has not been identified. ADSORPTION OF PHAGE TO HOST CELL 147 In Studying the adsorption of various phages to sintered-glass filters, Puck, Garen, and Cline (1951) noted that cofactor-requir- ing strains of phage T4 adsorb to glass in the presence but not in the absence of tryptophan. A variant of T4 which does not require cofactors for adsorption to cells adsorbs to glass in the absence of tryptophan. This observation discouraged the view that tryptophan might serve to activate an enzyme responsible for adsorption. 4. The Stepwise Nature of Adsorption Some experiments by Hershey, Kalmanson, and Bronfen- brenner (1944) discussed above demonstrate that phage T2 can adsorb to its host cell in salt-free broth, but that under these con- ditions it fails to infect. On dilution in salt-free broth some of the adsorbed phage elutes. These experiments demonstrate that adsorption of phage to susceptible bacteria is not equivalent to infection. A similar experiment was performed by Puck, Garen, and Cline (1951) with phage T2. They also obtained clear-cut evidence for the reversibihty of phage adsorption under certain environmental conditions. Under other conditions, adsorption is irreversible and Garen and Puck (1951) investigated the hypothesis that irreversible ad- sorption is a two-step reaction passing through a reversible phase. Several methods were devised for separating the steps, mostly in experiments with phage Tl . One method depends on effects of temperature. The rate of attachment of Tl to its host cell was determined by centrifuging the undiluted adsorption- mixture (so that elution would not occur) and assaying the super- natant fluid for unadsorbed phage after various periods of ad- sorption. Under these conditions, the first-order velocity con- stants were 2.7 X 10~^ ml. per minute at 2° C. This result showed that the rate of adsorption was indifferent to temperature except for small effects that might be attributed to the change in viscosity. This result was unexpected since Puck, Garen, and Cline (1951) had previously reported a 40-fold increase in rate of adsorption with a rise in temperature from to 37 ° C. How- 148 BACTERIOPHAGES ever, in the earlier experiments the adsorption mixture had been diluted 100-fold before centrifugation, thus permitting elution of any reversibly adsorbed phage. That this accounts for the dis- crepancy was demonstrated by Garen and Puck (1951) who showed that phage Tl adsorbed at 3 ° C. could be largely eluted on dilution into cold broth, but that phage adsorbed at 37 ° C. could not. Thus the temperature coefficient is low for reversible adsorption and high for irreversible adsorption. Other methods devised for separating the two kinds of adsorp- tion are the following : 7. Zinc ions specifically prevent irreversible adsorption of Tl (but not of T2). Phage Tl adsorbed to B at 37° C. in the presence of zinc ions is largely eluted on dilution, whereas Tl adsorbed in the presence of calcium ions is not. Zinc competes with calcium and magnesium ions for sites on the surface of the bacterial cell, thereby protecting the bacterium against in- vasion by phage Tl . 2. Suspensions of bacteria heavily irradiated with ultra- violet light still adsorb phage Tl quite rapidly at 37 ° C, but on dilution into cold broth all the phage is eluted. With unir- radiated bacteria, the adsorption is irreversible under the same conditions. Irradiation must destrov some bacterial substance essential for irreversible adsorption. 3. Strain B of E. coli has at least two different spontaneous mutants that are resistant to phage Tl, B/1,5 resistant to both Tl and T5, and B/1 resistant to Tl only. Phage Tl fails to ad- sorb to B/1,5 under any condition tested, but adsorbs reversibly to B/1. Thus, mutation to B/1 results in loss of some factor essential for irreversible adsorption without seriously affecting reversible adsorption. In B/1,5 both kinds of adsorption fail. Similarly, B/2 does not adsorb T2. There is also no reversible adsorption of T4 to B/4 (Stent and WoUman, 1952). Similar experiments on the adsorption of phage T4 to its host cells in +he presence and absence of tryptophan show that no attachment of the phage to the host cell takes place unless tryptophan is present. This indicates that trytophan is essen- ADSORPTION OF PHAGE TO HOST CELL 149 tial for the reversible adsorption to host cells (Garen and Puck, 1951). As mentioned previously, the same is true of the ad- sorption to glass. Stent and Wollman (1952) found that the rate of irreversible adsorption of T4 decreases with decreasing temperature in the range from 25 to 5 ° C. The rate constant also depends on bac- terial concentration, decreasing at concentrations in excess of 10^ per ml. They considered three hypotheses to account for these characteristics. 7. A reversible equilibrium exists between two states of the phage particle, an active state capable of adsorption, and an inacUve state which cannot adsorb. The equilibrium would be temperature dependent. 2. Adsorption may occur by two competing reactions, revers- ible and irreversible. At high bacterial concentrations, most of the phage is adsorbed reversibly and the rate of desorption controls the rate of irreversible adsorption. .3. Adsoiption necessarily involves an initial, reversible at- tachment. Reversibly adsorbed phage may desorb, or become irreversibly attached, by competing reactions. At high bacterial concentrations and especially at low temperatures the conversion from reversible to irreversible attachment, dependent on tem- perature, becomes rate-limiting. The third hypothesis is the same as that of Garen and Puck (1951). The infection of host cell (B) by phage (P) may be ex- pressed as: k^ k3 P + B . PB > X ki where X represents the irreversibly infected host cell. The ve- locity constant ki is the rate of primary attachment and has a low temperature coefficient. The constant k^ is the rate at which reversibly attached phage becomes irreversibly fixed to the host cell. This rate has the relatively high temperature coefficient discussed above. The constant Ajo is the rate at which reversibly attached phage is liberated again into the medium. Stent and Wollman using T4, and Garen (1954) using Tl, demonstrated 150 BACTERIOPHAGES that ko must have a low temperature coefficient, as might be ex- pected for the rupture of ionic bonds. Garen and Puck (1951) have reported evidence suggesting that /..-j may be an cnzymically catalyzed reaction. Garen (1954) made a detailed study of the reversible interaction between Tl and B/1, showing that the rate of primary attachment of Tl to B and to B/1 is about the same. This relatively simple conception of a two-step mechanism of adsorption has been widely accepted and seems to be the most reasonable explanation for the kinetics of adsorption of phages Tl, T2, and T4 to their host cells. Tolmach (1957) thoroughly reviewed the evidence for this view. Hershey (1957), however, pointed out that the experimental evidence leaves much to be de- sired, notably because no test has been made of the postulate that the transition from reversible to irreversible attachment is pos- sible and rapid. 5. Chemical Nature of Receptor Sites There have been two principal approaches to the problem of the chemical nature of the bacterial receptor sites for phage ad- sorption. One approach involves a studv of the kinetics of phage adsorption to bacterial cells treated with various agents which might alter the cell surface. The second approach in- volves a study of the chemical properties of the receptor sub- stances after their isolation from the bacterial cells. The dominant role of salt concentration in the adsorption process led Puck, Garen, and Cline (1951) to postulate that the primary attachment of phage particle to host cell is by electro- static bonds. The chemical nature of the charged groups in- volved in adsorption was studied by Tolmach and Puck (1952) and Puck and Tolmach (1954). They used phage labeled with P'^- and studied the distribution of radioactivitv between super- natant and sediment in centrifuged samples of adsorption mix- tures. This technique permits the study of adsorption under environmental conditions which would destroy the viability of phage or host cell. The effect of pH was studied over the range ADSORPTION OF PHAGE TO HOST CELL 151 from 5 to 12 using phage T2 and E. coli B. Maximum adsorp- tion occurred at pH 6 to 8. The pH dependence suggested to Tolmach and Puck that carboxyl and amino groups are pri- marily involved in phage T2 adsorption and that phosphoric, sulfhydryl, and phenolic groups are unimportant. This notion was tested by studying the adsorption of phage to host cells which had been treated with various reagents designed to eliminate specific chemical groupings. The results are given in Table XIII. They were interpreted by the authors as indi- cating that adsorption of phage T2 involved primarily carboxyl groups of the bacterial surface while adsorption of Tl involved amino groups. TABLE XIII The Binding of Phages Tl and T2 to Chemically Modified Host Cells" Adsorption to modified cells" Reagent used to modify bacteria T2 phage Tl phage Carboxyl reagents Propylene oxide (sat'd.) + CH3OH (80%) + HCl (0.1 M) + + + Amino reagents Nitrous acid (0.5 M, pH 4) + + + + Formaldeliyde (1.2 M) + + + + + Acetic anhydride (0 . 3 M) + + + + + Other reagents 12(5 X 10-3 A/) 4- 0.5 A/KI + + + + + + + + /?-Chloromercuribenzoate (5 X lO^* A/) + + + + + + + + HCl (0.1 A/) + + + + + + + + CH3OH (80%) + + + + + + + + " From L. J. Tohiiach and T. T. Puck, J. Am. Chem. Soc, 74, bbbl (1952). " signifies 0-30%o adsorption; +, 30-45%o; + + , 45-60%; + + + ,60- 80%; + + + + ,80-100%. Barry and Goebel (1951) studied the adsorption of phage to cultures of phase II Sh. sonnei which had been treated in various ways. The untreated bacteria adsorb T3, T4, and T7 at similar rates. Different culture samples were treated with phenol, 1 52 BACTERIOPHAGES formaldehyde, ethylene oxide, ethanol, periodate, thymol, chloroform, heat, sonic vibration, and osmotic shock, and were then tested for their ability to adsorb phage. In all cases the adsorption of phage T3 is very quickly lost whereas the ability to adsorb T7 is scarcely impaired. The adsorption of T4 is markedly decreased following treatment with heat and ethvlene oxide. Killing of the bacteria with moderate doses of ultra- violet light does not affect the adsorption of any of the phages but heavier doses selectively inactivate the T3 receptors. It is evident that the receptor for T3 is markedly labile to all the agents tested and that these experiments give no clue to the chemical nature of the receptor sites other than to emphasize that the sites for various phages are different. Similar experiments on strain B of E. coli were performed by Weidel (1953a). Formaldehyde decreases the activity of the receptors for Tl, T3, T5, and T7 but not for T2, T4, and T6. Periodic, nitrous, and other weak acids destroy the receptors for Tl, T3, T4, T5, and T7 but not for T2. Sodium hydroxide (A^/60) removed the T5 receptor from the bacterial surface with- out affecting the T2 receptor. Weidel notes the unusual stabil- itv of the T2 receptor; "means to destroy it specifically have not yet been found." However, Tolmach and Puck found that T2 receptors are destroyed by CH3OH -}- HCl without inactiva- tion of the Tl receptors. Further work along these lines with more specific chemical reagents may be well worth while. If phage particles attach to antigenic substances on the bac- terial surface (Chapter IX), one might expect that bacterial ex- tracts containing soluble antigens should react with the phage particles to interfere with adsorption. Early attempts to test this possibility met with failure because of faulty concepts of the nature of the bacterial antigens. Because in many cases the immunological specificity is due to a polysaccharide hapten, the efforts of immunochemists were devoted to the isolation of the polysaccharides in pure form. When these were tested for phage neutralizing activity they were usually found to be inert (Burnet, 1934b). It was later found that the phage-inactivating ADSORPTION OF PHAGE TO HOST CELL 153 activity was dependent in many cases on the intact structure of the bacterial antigen, and could be destroyed by relatively mild reagents. Apparently the first demonstration of the inactivation of bacteriophages by bacterial extracts was made by Levine and Frisch (1934). Saline extracts of salmonella and shigella species were found to protect the homologous organisms from attack by phage. The active substance could be precipitated by alcohol and partially purified . These results were confirmed and extended by Burnet (1934b) who referred to the active sub- stance as "phage-inhibiting agent" (PIA). The PIA was pre- pared by autolyzing the bacteria for two days at 55 ° C, after which bacterial debris was removed by centrifugation and the solution filtered through gradacol membranes of one micron pore size. Filtration through Seitz or Berkefeld filters resulted in loss of activity. The activity was measured by incubating appropriate dilutions of PIA with diluted phage preparations and determining the surviving phage by plaque count assays. Ex- tracts were prepared from five variant strains of Flexner Y dysen- tery bacilli, and tested for inhibiting activity against eight differ- ent phages. In general it was found that phages attacking a given bacterial strain were inhibited by extracts from that strain, while these extracts had no effect on phages to which the bac- terial strain was resistant. Occasional extracts that failed to inactivate the expected phages are not surprising in view of the lability of many receptor substances. The active extracts invariably gave precipitates when mixed with homologous antibacterial serum, the amount of precipitate being proportional to the phage inhibiting activity. After re- moval of the specific precipitate, the supernatant fluid was devoid of phage-inhibiting activity. The neutralization of PIA by anti- bacterial sera is parallel to the agglutination by these sera of the bacteria from which the extracts were prepared. An antiserum which agglutinated several bacterial strains also neutrahzed the PIA extracted from these strains. The action of the most potent preparation of PIA was studied 154 BACTERIOPHAGES kinctically, and it was found lliat j)hagc inaclivalion by this bac- terial extract was remarkal)ly similar to phage neutralization by antiphage serum. The inactivation rate was initially first order and later decreased. The fraction of the phage inactivated was independent of the initial phage concentration. The rate was directly proportional to the PIA concentration and had a tem- perature optimum at about 37 ° C. The phage particles which survived incubation with PIA produced plaques which were smaller than normal and widely variable in size suggesting a de- creased rate of adsorption to the host cell. Burnet concluded that PIA is intimately associated with the somatic bacterial antigen, and that this substance blocks phage infection by combining with receptor sites on the phage particle which in the normal course of events make contact with the bacterial surface. Further interesting experiments were reported by Burnet and Freeman (1937) using phage H (related to T2) and PIA from a Flexner Y strain of shigella. A variant of phage H was isolated which adsorbed at a slower rate to its host cell. The variant also differed from its parent in a number of other characteris- tics : larger plaque size, higher titer stocks, greater susceptibility to heat inactivation, and more rapid inactivation by antiphage serum. The variant was also extremely resistant to a prepara- tion of PIA which rapidly inactivated the parent phage stock. A similar variant was isolated from stocks of phage CI 6. Successive interaction of antiphage serum and PIA with phage H was also examined. Phage which had been treated with dilute antiphage serum to a survival of 50 per cent proved to be a great deal more resistant to inactivation by PIA than was un- treated phage. In contrast treatment of phage with PIA had no effect on its subsequent neutralization with antiphage serum. These facts suggest that combination of phage with PIA is dif- ferent from combination of phage with antiphage serum or with host cells. Phage H which had been inactivated to better than 95 per cent by PIA was still able to adsorb to the host cell as judged by ADSORPTION OF PHAGE TO HOST CELL 155 the fact that the host cells became agglutinable by antiphage serum. PI A did not prevent adsorption of phage to host cell but did interfere with the next step in the infectious process. As we have seen in Chapter VIII phage particles treated with anti- phage serum are also able to adsorb to the host cell ahhough they are noninfectious. Burnet and Freeman concluded from these experiments that phage antibodies and PIA probably do not re- act with the same site on the phage surface but rather at adjacent sites. The interaction of phage, PIA, antiphage serum, and host eel) is discussed further by Burnet, Keogh, and Lush (1937) : The first stage in the lysis of susceptible bacteria requires tiie mutual specific union of certain elements of complementary molecular configuration on the surfaces of the phage particle and bacterium respectively. The specific ele- ment on the bacterial surface is intimately related to the polysaccharide hapten which determines the antigenic character of the bacterium. The hapten, as finally isolated by chemical methods, is not present as such in the living bac- terial surface, but forms an essential part of a more complex molecular pattern. It is to certain aspects of this pattern, not necessarily the same for each phage lysing the organism, that specific phage adsorption occurs. Any process by which the bacterial surface components are brought into solution necessarily destroys to some extent their specific molecular pattern. The fact that reaction of phage with its antibody makes the phage resistant to inactivation by PIA must mean that neutralizing antibody and PIA react with receptors that are spatially contiguous and parts of a single com- plex. These detailed suggestions have been shown by later work to be remarkably accurate. The kinetic experiments of Burnet were confirmed by Ellis and Spizizen (1941) using coliphage PI. The inactivation by PIA was markedly influenced by salt concentration, being opti- mal at 0.5 per cent NaCl and negligible in 25 per cent NaCl or in the absence of salt. Phage inactivated by PIA could not be reactivated by changing the salt concentration. About 5 per cent of the phage population was very resistant to inactivation. 1 56 BACTERIOPHAGES Similar observations were made by Goebel (1950) using phage T3 and the purified somatic antigen of phase II Sh. sonnei. Phage T3 was rather slowly inactivated by the somatic antigen and a residue of the population, about 1 in 10^, proved to be completely resistant to inactivation. The original T3 phage population plated with equal efficiency on E. coli strain B and on Sh. sonnei, phase II. However, the PIA-resistant survivors showed a 40 times greater plaque count on E. coli than on Sh. sonnei. PIA-resistance and the property of plating only on E. coli were not hereditary. There are many references to qualitative observations on the phage-inhibiting activities of bacterial extracts among which are Freeman (1937), Rountree (1947b), Beumer (1947, 1953), Weidel (1953a, b), and Mondolfo and Hounie (1948). There have not been many experimental studies of the physical and chemical properties of the phage-inhibiting agents of bacterial extracts. Gough and Burnet (1934) reported on the properties of PIA isolated from Sh. fiexneri. The material was soluble in half-saturated ammonium sulfate but was precipitated by saturated ammonium sulfate, and by 2.5 volumes of alcohol. Material repeatedly precipitated by these agents was active serologically and as PIA. It gave a strong Molisch test for carbohydrate, a weak biuret test, and no pentose color reaction. Treatment with hot one per cent NaOH liberated the poly- saccharide hapten which was serologically active but had no phage-inhibiting activity. Heat treatment at 90° C. and pH 9 caused a rapid loss of PIA with no change in serological activity. The inhibitory activity toward different phages was lost at widely different rates suggesting that not all phages reacted with the same part of the PIA. Attempts at further purification in- variably resulted in decreased PIA activity. Meanwhile a great deal of chemical work on the somatic anti- gens of the enteric pathogens demonstrated that these antigens are not simple polysaccharides, but are, rather, a complex as- sociation of polysaccharide, phospholipid, and protein. The phospholipid can be separated by use of polar solvents without ADSORPTION OF PHAGE TO HOST CELL 157 affecting the antigenicity of the mucoprotein, but the polysac- charide and protein separately are nonantigenic. The most intensive study of the phage-inhibiting activity of purified bacterial antigens was made by Goebel and collabora- tors using Sh. sonnei and the T series of coli-dysentery phages. Phase I of Sh. sonnei is susceptible to phages T2 and T6 but resist- ant to the other T phages. Phase I organisms mutate to phase II, which is serologically distinct from phase I, and concomi- tantly become susceptible to phages T3, T4, and T7 in addition to T2 and T6. The somatic antigens of phase I and phase II cultures have been shown to be lipocarbohydrate-protein com- plexes (Baker, Goebel, and Perlman, 1949). The highly puri- fied and electrophoretically homogeneous antigens were tested for phage-inhibiting activity by Miller and Goebel (1949). Phages T2 and T6 were inhibited by neither antigen, and phages T3, T4, and T7 were inhibited by the phase II antigen only. The diluent used had a marked effect on the extent of inhibition. T3 and T7 were strongly inhibited in broth but not in buffer. A tryptophan-requiring strain of T4 was inhibited in broth but not in buffer unless tryptophan was added. A variant of T4 which did not require tryptophan was inhibited equally well in buffer and broth. These results again suggest that the role of tryptophan is in the attachment of phage T4 to the bacterial re- ceptor substance. They also suggest the possibility of a cofactor requirement for T3 and T7. Heat treatment of the phase II antigen resulted in rapid loss of phage-inhibitory activity. Digestion with pancreatin followed by dialysis and deproteinization by shaking with CHCI3 and octyl alcohol yielded a lipocarbohydrate with negative tests for pro- tein. This material was fully active serologically and essentially as active as the undegraded somatic antigen in its ability to in- hibit plaque formation with phage T4. The material is similar to that isolated by Gough and Burnet (1934) from autolyzed cul- tures oiSh.flexneri. By using somewhat different isolation procedures, Jesaitis and Goebel (1952) isolated a phase II antigen which inactivated 158 BACTERIOPHAGES phages T2, T3, T4, T6, and T7. The difference between this antigen and that described above which inactivated only T3, T4, and T7 was traced to the use of pyridine in the extraction proce- dure. Incubation in 50 per cent pyridine at 37 ° C resulted in a rapid destruction of the T2 and T6 neutralizing activity of the purified antigen without affecting its activity against T3, T4, and T7. The purified antigen was an electrophoretically homo- geneous complex of protein, lipid, and polysaccharide. Digestion with pancreatin resulted in removal of most of the protein, with a concomitant increase in serological activity and in phage-neutralizing activity. Extraction with phenol, which removed all detectable protein, resulted in loss of T2 and T6 activity. This lipocarbohydrate contained glucose, galactose, glucosamine, and heptose. Its composition was 45 per cent re- ducing sugars, 29 per cent lipids, 4 per cent acetyl, 4 per cent phosphorus, and 1 3 per cent ash. In an extension of this work, Goebel and Jesaitis (1953) studied the properties of a somatic antigen isolated from a phage-resist- ant variant of phase II Sh. sonnei, designated as 11/3,4,7, sus- ceptible only to T2 and T6. The somatic antigen was isolated from strain 11/3,4,7 and its chemical, serological, and antiviral properties were examined. The purified lipocarbohydrate protein antigen inhibited T2 but had no inhibiting effect on T3, T4, T7, or T6. The purified antigen was digested with pan- creatin to remove much of the protein. This degraded antigen was much more active against T2 but still was without effect on the remaining T phages. The variant antigen is serologically distinct from the parent phase II antigen, but is related to it. The enzymatically degraded antigen of the variant differs from the corresponding antigen of the parent in electrophoretic mobility, optical activity, and chemical composition. The lipocarbohydrate was separated from the protein by treatment of the degraded antigen with phenol. Paper chromatography of the hydrolyzed lipocarbohydrate revealed only one saccharide, an amino sugar which was neither glucosamine nor chondrosa- mine. In the case of phase II Sh. sonnei, mutation to phage re- ADSORPTION OF PHAGE TO HOST CELL 159 sistance is marked by a change in the chemical constitution of the polysaccharide moiety of the lipocarbohydrate-protcin antigen of the organism. Whether changes occur also in Hpid or protein portions is not known. The mechanism of phage inactivation by PIA is in most cases unknown. However, Jesaitis and Goebel (1955) studied the inactivation of phage T4 by the specific lipocarbohydrate which had been isolated from the antigenic complex of phase II Sh. sonnei. The addition of this lipocarbohydrate to a concen- trated suspension of T4 phage results in an immediate large in- crease in viscosity (see also Jesaitis and Goebel, 1953). Elec- tron microscope examination reveals phage ghosts and long filaments, probably of deoxyribonucleic acid (DNA). The phage DNA is susceptible to degradation by deoxyribonuclease. These results show that attachment of T4 to bacterial receptor substance causes the release of DNA from the phage particles. Weidel and his collaborators studied receptor substance for T5, isolated from E. coli B, with results paralleling those described above in many respects (Weidel, Koch, and Bobosch, 1954; Weidel and Koch, 1955; Weidel and Kellenberger, 1955; Koch and Weidel, 1956a). 6. Summary Adsorption is defined as attachment of the phage particle to the host cell surface. The particles attach by the tips of their tails. The kinetics of adsorption is that of a first-order reaction, the rate being proportional to the varying phage concentration and to the constant amount of bacterial surface. The adsorp- tion rate is markedly aff'ected by environmental conditions such as salt concentration, pH, and temperature. Some phages re- quire the presence of organic molecules which must become fixed to the phage before adsorption occurs. This cofactor require- ment for adsorption is subject to modification by mutations of the phage. Adsorption probably involves at least two successive steps, the first of which is reversible. The rate of the first step is little affected by temperature, whereas the rate of the second 160 BACTERIOPHAGES Step is temperature-dependent. The first step involves the for- mation of salt linkages between charged groups on the phage and host cell surfaces. Carboxyl and amino groups play the prin- cipal role in a few systems which have been studied. In some cases the phage receptor substance on the bacterial surface is a major bacterial antigen. Bacterial extracts containing the somatic antigen have the property of inactivating some phages. Highly purified antigens in the form of complexes of lipid, pro- tein, and polysaccharide retain this property. Treatment with various chemical reagents has a diff'erential effect on the inhibi- tory activity toward various phages. CHAPTER XI STAGES IN PHAGE MULTIPLICATION An over-all view of the lytic cycle of phage reproduction was given in Chapter II. In this chapter we shall examine in more detail some of the biological and physical methods used to elucidate the sequence of events. Strictly chemical experiments will be discussed in Chapter XIV. 1. Initiation of Infection a. Penetration The electron microscope furnished convincing evidence that phage multiplies within the host cell and is liberated by rupture of the cell membrane (DeMars, Luria, Fisher, and Levinthal, 1953). A section of an infected bacterium illustrating in- tracellular particles of phage T2 is shown in Figure 2 (Chapter TV). Evidently the infecting phage particle must penetrate through the cell wall to reach the interior. Chemical evidence of penetration was furnished by the observation that phosphorus and nitrogen from the infecting phage were contained in the phage progeny (Chapter XIII). However, these observations did not at first yield information about the chemical composition or biological organization of the phage material which pene- trated the host cell wall. A direct attack on this problem was made by Hershey and Chase (1952) by using isotopic labels. The proteins of phage T2 were labeled with S^^ and the nucleic acid was labeled with P^2. The labeled phage was purified by differential centrifu- gation and its properties studied by following its radioactivity. Neither the phosphorus nor sulfur was acid-soluble and neither became soluble on exposure of the phage particles to deoxy- 161 162 BACTERIOPHAGES ribonuclease. About 90 per cent of both the phosphorus and sulfur was specifically adsorbable to sensitive bacteria and precipitable by anti-T2 serum, suggesting that the phage was relatively free of extraneous protein and nucleic acid. The labeled phage was subjected to osmotic shock, which separates the phage nucleic acid from the phage membrane or ghost (Chapter V). In the resulting shocked phage neither the phosphorus nor the sulfur was acid-soluble. However, treat- ment with deoxyribonuclease made 80 per cent of the phos- phorus acid-soluble, the sulfur remaining insoluble. About 90 per cent of the sulfur was still adsorbable to sensitive bac- teria but only 2 per cent of the phosphorus could be adsorbed. About 97 per cent of the sulfur was precipitable by anti-T2 serum but only 5 per cent of the phosphorus. These results confirm the previous observations of Anderson and Herriott that osmotic shock permits the separation of phage T2 into a protein membrane and soluble DNA. They indicate fur- ther that almost all of the phage sulfur and almost none of the phage phosphorus is in the membrane which adsorbs to the host cell and precipitates with antiserum. Conversely, most of the phosphorus and little of the sulfur is in the DNA which is liberated from the membrane and thereby made susceptible to attack by deoxyribonuclease. Electron micrographs of infected bacteria had demonstrated that phage T2 adsorbs to its host cell by the tip of its tail (Chap- ter IV). This suggested the possibility that the adsorbed phage might be broken off from the bacterial surface by the strong shearing forces in a Waring Blendor. In fact, Anderson (1949) had found that phage does not adsorb to the host cell while the adsorption mixture is agitated in the Blendor. Her- shey and Chase (1952) performed the appropriate experiments using isotope-labeled phage. The phage was adsorbed to the host cell at low multiplicity of infection and unadsorbed phage was eliminated by centrifugalion. The infected bacteria were agitated in the Blendor for various times and the cHs- tributions of sulfur and phosphorus between bacteria and STAGES IN PHAGE MULTIPLICATION 163 extracellular fluid were dclcrniined by ccntrifunation. The results of this signihcant experiment are as follows. 7, The plaque-forining potentiality of infected bacteria is not afifected by agitation in the Waring Blendor. 2. Seventy-five to 80 per cent of the phage sulfur can be stripped from the infected cells in the Blendor. This liberated sulfur is precipitabie by antiphage serum. 3. Only 20 to 35 per cent of the phage phosphorus is liberated into the medium, half of it without any agitation. These results demonstrate that, following adsorption of phage T2, the phage membrane remains on the bacterial surface, from which it can be stripped in the blendor without affecting the course of infection. In contrast, most of the phage DNA enters the host cell soon after phage adsorption. A similar situation holds for the T5 serological group of phages. T5 injects its DNA into the host cell, while the complement-fixing antigen remains on the host cell surface and can be liberated in the Waring Blendor (Y. T. Lanni, 1954). Considered as "injection" of DNA, the penetration mech- anism presents at least three problems : the opening of a hole in the tail of the phage particle, the opening of a hole in the cell wall, and the passage of DNA through these holes. Ore and Pollard (1956) suggested that the passage of DNA into the cell can be explained by purely physical mechanisms. The other two problems have been partly clarified by several types of ex- periments summarized below. h. Release of DNA from the Phage Particle Hershey and Chase (1952) noted that partial release of DNA into solution, and complete exposure of viral DNA to deoxyri- bonuclease, followed the attachment of T2 to the bacterial debris left after lysis of infected cells. They also confirmed important experiments by Graham (1953), showing that adsorption of T2 to heat-killed (but not living) bacteria resulted in exposure of the viral DNA to enzyme action. These findings were extended to the interaction between phage and isolated receptor substances. 164 BACTERIOPHAGES for T4 by Jesaitis and Goebel (1953, 1955) and for T5 by Weidel, Koch, and Bobosch (1954). These results seem to show that the opening of a hole in the phage particle and release of DNA does not call for the action of a bacterial enzyme. DNA can also be released from phage T2 following interaction with ion exchange resins (Puck and Sagik, 1953) or with cad- mium cyanide (Kozloff and Henderson, 1955), and from T5 by interaction with citrate or merely by removal of calcium (Lark and Adams, 1953). In the last two cases, the reaction is ac- companied by loss of adsorbability of the phage ghosts and al- terations at the tip of the tail of the phage particle. The relation of these facts to normal injection is obscure, however, especially since the injection by T5 requires calcium and is presumably in- hibited by citrate (Luria and Steiner, 1954). If a phage enzyme is involved in the release of DNA, it remains to be demonstrated. The morphological aspects of penetration by T2 were clarified by Kellenberger and Arber (1955) and Williams and Fraser (1956) (Chapter IV). Fibers at the tail-tip of the phage particle attach to the bacterial surface, exposing the central pin which actually penetrates the cell wall. This penetration could be purely mechanical (Williams and Fraser, 1956) or might be aided by enzymic processes. Phage T2 inactivated by formaldehyde still adsorbs to the host cell, and kills it with low efficiency. However, the DNA of the inactivated particles is not injected into the bacteria but remains within the phage membrane in a form resistant to deoxyribonu- clease. It could be supposed that formaldehyde inactivates an enzyme in the phage particle but in view of the complicated nature of the penetration process, the action of formaldehyde can be explained in other ways. c. Puncture of the Bacterium Infection of bacteria by T2 initiates a complex series of cellular reactions many of which have to be considered in any discussion of mechanisms of penetration, especially because no satisfactory conclusions can be reached at this time. STAGES IN PHAGE MULTIPLICATION 165 It will be recalled that infection with large numbers of T2 (but not of many other phages) is abortive and causes prompt "lysis from without." Smaller numbers of T2 ghosts (Barlow and Herriott, 1954) or of phage particles inactivated by ultra- violet light (Watson, 1950) produce the same effect, as do also phage particles inactivated by X-rays until bacteria-killing ability is lost (Watson, 1950). These facts may be summarized by say- ing that phage particles can damage cells severely at the time of infection, and that this damage is independent of injection of DNA, appearing, in fact, to be especially severe when one or another early step in the normal sequence of events is blocked. Thus sensitivity to lysis from without is also increased in the presence of metabolic inhibitors or by deprivation of food (Heagy, 1950; Watson, 1950), but is greatly reduced in the presence of high concentrations of magnesium that do not pre- vent infection (Barlow and Herriott, 1954). Many of these facts can now be partly understood in terms of two competing processes, one lytic and one antilytic, set in train by the normal process of infection. Doermann (1948a) first noticed that normal infection pro- duces a rapid change in cellular properties. He observed a sudden fall in turbidity of bacterial cultures following infection. The turbidity passed through a minimum at about 10 minutes after infection, and then rose again. The initial drop was caused by T2, T4, T6, and T5, but not by Tl, T3, or T7. It was seen also following infection with T5 in the absence of calcium (no injection), in which case there was no subsequent rise. The effect was independent of multiplicity of infection in the range between 3 and 10 or 15, above which lysis from without was ob- served. Doermann suggested a spreading alteration of the bac- terial surface followed by "recovery." Subsequent experiments confirm his interpretation, except that his "recovery" is much too slow to reflect the development of resistance to superinfection and probably should be ascribed to the general biosyntheses accompanying phage growth. Two phenomena, the chemical breakdown of superinfecting 166 BACTERIOPHAGES phage and their genetic exclusion, led to the recognition that in- fected cells rapidly develop refractoriness to penetration by phage in a reaction completed after about two minutes (Chapter XVII). Visconti (1953) showed that a reaction of similar swift- ness produces almost complete resistance to lysis from without by superinfecting phage. The resistance to lysis cannot reflect merely the failure of normal penetration, because cells infected at multiplicities sufficient to cause lysis in the presence of cyanide do not lyse if the cyanide is added a few minutes later. Evi- dently the properties of the cell surface as a whole are very quickly altered by metabolic processes occurring after infection. This alteration must account for faulty penetration, genetic exclusion, and breakdown of superinfecting phage, as well as resistance to lysis by numerous agencies. Puck and Lee (1954, 1955) described important experiments that clarified considerably the nature of these processes. They measured leakage of cell constituents, labeled with radioactive phosphorus or sulfur, into the culture fluid during the first few minutes following infection. Most of the experiments were per- formed with T2, although a slight leakage occurred also with several other phages tested. Like lysis from without, leakage is prevented by magnesium ions and, unexpectedly, also by po- tassium ions. The chief difliculty in these experiments is to discriminate be- tween leakage accompanying normal infection, and frank lysis from without. The difficulty is increased because in most of the experiments of Puck and Lee infection was allowed to occur at low temperatures, which causes abortive infection (Chapter XV). In one experim.ent (Figure 3 of Puck and Lee, 1955) infection at moderate multiplicities liberated considerable amounts of acid-soluble phosphorus but practically no ribonucleic acid or galactosidase. Since all these substances are released during lysis from without, it seems clear that the leakage of acid-soluble phosphorus is something different. In other experiments galac- tosidase was released at moderate multiplicities and the authors STAGES IN PHAGE MULTIPLICATION 167 draw conclusions from these experiments based on the assumption that there was no lysis from without, but this seems doubtful. In still other experiments it appears that homologous super- infecting phage does not cause a second leakage. This could be explained by exhaustion of acid-soluble phosphorus during the first leakage coupled with resistance to lysis from without. The authors postulate instead a "resealing reaction." Since this re- action is supposed to explain resistance to lysis from without and, in fact, faulty penetration by superinfecting phage, the two interpretations are not very different. It seem.s reasonable to conclude, in agreement with Puck and Lee, that normal infection causes leakage of some cellular con- stituents, incidentally to penetration of the cell wall by phage, and that the holes have to be repaired before viral growth can proceed. All of the facts discussed above show that the repair does not merely restore the cell surface to its initial condition, but confers on it several new properties. The ideas of Puck and Lee serve to unify numerous observa- tions, as already discussed. Penetration of the cell wall causes damage detected as leakage of low molecular weight constituents and possibly others, and causes a slight fall in the turbidity of the bacterial suspension. This damage is limited in some way, since it is independent of multiplicity of infection in the absence of lysis from without. The damage is accompanied by an overcompen- sating repair process that explains resistance to lysis from without and exclusion of superinfecting phage. The two processes are in competition : lysis from without results when the repair mecha- nism is overwhelmed by excessive damage. The repair mecha- nism, at least, depends on cellular metabolism. Metabolic in- hibitors favor lysis and magnesium ions oppose it. Puck and Lee (1954) attributed leakage to holes produced by bacterial enzymes, largely by analogy between lysis from without and effects of certain synthetic polyamino compounds that cause lysis. The following work does not contradict this view but sug- gests the participation of phage enzymes as well. Barrington and Kozloff (1956) and Koch and Weidel (1956b) 168 BACTERIOPHAGES described rather similar experiments in which solubiHzation of nitrogenous constituents of isolated cell walls by adsorbed phage is measured. The effect is seen with phages T2, T4, T6, and T5, but not with T7. Their results may be summarized as fol- lows. The amount of material solubilized corresponds to about 1 per cent of the cell wall nitrogen per phage, increasing linearly with multiplicity of adsorption up to about 1 5 per cent solubili- zation, after which attachment of additional phage particles has no effect. This maximal solubilization can also be achieved with fewer phage particles if the mixture is repeatedly centri- fuged, an effect attributed to an increased opportunity of contacts between insoluble phage enzyme and insoluble substrate. In either case, the maximum solubilization produced by T2 is not increased by adsorption of additional particles of either T2 or T4. This result shows clearly that the materials solubilized are not the phage-specific receptor substances. Phosphorus and amino acids are both solubilized. The reaction fails at low tem- peratures, but is not inhibited by pretreatment of the cell walls with heat or with phenol. The role of this reaction in penetra- tion is not clear, especially because of the limited effects observed, but an important role in infection may be surmised. It seems very likely that it prepares the cell for the leakage of cell constit- uents observed by Puck and Lee. It also helps to explain their "sealing reaction," since cell wall materials are very rapidly syn- thesized after infection (Hershey, Garen, Fraser, and Hudis, 1954). Park (1956) and Adams and Park (1956) described an enzyme present in phage lysates of Klebsiella pneumoniae that acts on the capsular material of these bacteria. The enzyme is found both free in the lysate and attached to phage particles, and the amount produced seems to depend on the hereditary constitution of both bacterium and phage. It is not found in uninfected cells. The role of the enzyme in adsorption and penetration has not yet been studied. STAGES IN PHAGE MULTIPLICATION 169 2. The Latent Period After adsorption of phage to the host cell, a period of time elapses before the cell lyses, liberating phage progeny. This time interval is known as the latent period and is measured by a one-step growth experiment (Chapter II). In such an experiment, not all the infected cells lyse at the same time. However, if the plaque count is plotted on a logarithmic scale as a function of the time of sampling, the rising portion of the curve is linear (Adams, 1949b; Maal0e, 1950; Barry and Goebel, 1951; Doermann, 1952; Bentzon, Maal0e, and Rasch, 1952). On such a plot, the intersection of the line with the baseline count of infected bac- teria provides a suitable working definition of the miniinum la- tent period. The latent period depends on the phage and host strains, and also on the environmental conditions. Different phages growing on the same bacterial strain may have widely different latent periods (Delbriick, 1946b). Even closely related phages such as T2, T4, and T6 (Delbriick, 1946b) and T5 and PB (Adams, 1951b), may have quite distinct latent periods on the same host. In the case of T5 and PB, Adams (1951b) showed that the latent period of each phage may be separated from other characteris- tics by genetic recombination. The influence of different hosts on the latent period of the same phage has been reported by Barry and Goebel (1951). The latent period is strongly affected by the physiological state of the host cell, as demonstrated by Delbruck (1940a) and by Heden (1951). Delbruck (1946b) re- ported that the latent periods for phages Tl, T2, and T7 were the same in a synthetic medium as in broth, even though the bacterial growth was much slower in the synthetic medium. This is not true of all phage-host systems; Adams (1949b) found a latent period for phage T5 of 55 minutes in synthetic medium as compared with 40 minutes in broth. Temperature generally affects the latent period in a similar manner to its effect on the generation time of the bacteria (Ellis and Delbruck, 1939); at lower temperatures, both increase. Metabolic poisons are known to alter the latent period. For 170 BACTERIOPHAGES example, the addition of penicillin (Krueger, Cohn, Smith, and McGuire, 1948) or aureomycin (Altenbern, 1953) shortens the latent period. Damage to phage particles caused by ultraviolet light can result in a prolonged latent period for those phage par- ticles surviving the irradiation (Luria, 1944). A phenomenon which has a marked effect on the duration of the latent period is lysis inhibition (Doermann, 1948a), which occurs with phages in the T2-C16 serological group. If, throughout the latent period, an infected bacterium is con- tinuously reinfected with phage, bacterial lysis may be delayed for an hour or longer. The m_echanism of lysis inhibition is not known. The ability to produce this phenomenon may be lost by mutation of the phage to the so-called r (rapid lysis) form. However, a mutant which is r type on one host may still produce lysis inhibition on a different host (Benzer, 1957). Thus, lysis inhibition is controlled by both phage and host. The latent period is insensitive to the number of phage par- ticles with which the host cell has been infected (Delbriick and Luria, 1942). This result was originally interpreted as indicat- ing that only one of the adsorbed phage particles participated in the infectious process. However, later work with genetically marked phage particles indicated that as many as 10 particles might successfully infect a single bacterial cell (Dulbecco, 1949b). Although the minimum latent period is a reproducible charac- teristic of a given phage-host system, the precise times of lysis of the individual cells are far from identical. There may be a two- fold difference between the latent periods of individual cells. Adams and Wasserman (1956) described a probit method for analyzing the variation in latent periods for individual cells. They show that the method permits one to recognize heteroge- neous populations of bacteria or phage arising from experimental treatments of either one. 3. Burst Size The burst size is the number of phage particles released per in- fected bacterium. Its average value is determined by the one STAGES IN PHAGE MULTIPLICATION 171 Step growth experiment. The average burst size depends on the phage strain, the host cell, and the environmental conditions. J3ifTerent phages growing in the same host strain may have quite different burst sizes (Delbriick, 1 946b) . The same phage growing in different host strains may also have different burst sizes (Barry and Goebel, 1951). An effect of the physiological state of the host cells was demonstrated by Delbriick (1940b) who ob- tained an average burst size of 20 with resting bacteria and 170 with growing bacteria. Heden (1951) studied the burst size during the period of transition from the resting phase to the phase of logarithmic growth of the host cell. He found the maximum yield per cell at about the time of onset of bacterial division. This was also the time when the bacterial cell size and the con- tent of ribonucleic acid per cell were maximal. This applied to bacteria grown in broth or in synthetic iriedium. The burst size is generally much larger in media such as nutrient broth than it is in synthetic media. Under conditions of lysis inhibition, where the time of lysis is delayed, the burst size may be more than doubled (Doermann, 1 948a) . Since delaying lysis of the cell permits inore phage par- ticles to be produced, it would appear that lysis rather than ex- haustion of materials interrupts phage growth. The phage yield from individual bacteria can be determined by the single burst technique. A suspension of infected bac- teria is diluted sufficiently so that, when samples are distributed into separate tubes, only a small proportion of the tubes will con- tain infected cells, mostly only one. The samples are incubated until the bacteria have lysed, and then each is plated to deter- mine its phage content. In experiments of this type with phage Tl, Delbriick (1945a) found that burst sizes ranged from below 20 to over 1,000. Since the distribution of burst sizes was much broader than either the distribution of host cell dimensions or the distribution of latent periods, the enormous range in burst sizes was probably not due to either of these factors. Similar observa- tions have been reported by Delbriick (1945c) and by Hershey and Rotman (1949). There is no adequate explanation at present for the wide range of burst sizes in a culture. 172 BACTERIOPHAGES 4. Premature Lysis The one-step growth experiment reveals nothing about the im- portant intracellular events during the latent period. In order to penetrate behind the scenes, various methods for rupturing the host cell before the end of the normal latent period have been .0= lO'- Ar*^ A LYSIS BY KCN -t- T6 • CONTROL LYSIS ■ SONIC DISRUPTION OF CELLS O 10 20 30 40 MINUTES OF INCUBATION IN GROWTH MEDIUM AT 30* Figure 4. Maturation of phage T3 in infected bacteria. The curve marked "control lysis" is the ordinary one-step growth of this phage. The other two curves show the yields of infective particles released by two different methods of artificial lysis. Reproduced from T. F. Anderson and A. H. Doermann, 1952a, ./. Gen. Physiol., 35, 657, with permission. STAGES IN PHAGE MULTIPLICATION 173 devised. Several methods for inducing "premature lysis" are described below. One method utilizes intense sonic vibration to disrupt the host cell and liberate its contents (Anderson and Doermann, 1952a). This method is limited to phages that are more resistant to sonic vibrations than are their host cells. Phage T3 is suitable for this purpose. The results of a typical experiment are presented in Figure 4. The curve marked "control lysis" is an ordinary one- step growth curve for phage T3 ; the latent period is 20 minutes at 30° C. and the period of lysis covers the next 10 minutes. The curve marked "sonic disruption of cells" gives the number of infective centers found after sonic treatment of the samples. The count of infective centers remains constant at about 10 per cent of the number of infected bacteria until the 12th minute. This count probably represents infected bacteria which were not dis- rupted. The samples taken near the end of the latent period (at 16 and 18 minutes) show a marked rise in phage count due to the premature release of mature phage particles. Sonic treat- ment at 24 minutes or later liberates the normal yield of phage particles. A second method of premature lysis developed by Doermann (1952) is based on the phenomenon of "lysis from without" described by Delbruck (1940b). If bacteria are attacked by a very large number of phage particles of the T2 species, they may respond by lysing without liberating phage progeny. The phenomenon is due to damage to the host cell membrane by the phage particles rather than to infection in the usual sense. In some cases, as with phage T2 itself, single infection may confer resistance to lysis from without by a secondary challenge with many T2 particles (Visconti, 1953). In the experiments of Anderson and Doermann (1952a), cells infected with T3 were "lysed from without" by exposure to large concentrations of phage T6 (plus cyanide added to "freeze" the metabolism). By using B/6 as the plating bacterium, it is possible to assay T3 without interference from the T6 present. In Figure 4 the curve marked "lysis by KCN plus T6" gives the 174 BACTERIOPHAGES results of such an experiment using the same culture of E. coli infected with phage T3 that was used for the other two curves in the figure. The base line for the cyanide-T6 lysis curve is very low, corresponding to less than one per cent of the infected bac- teria, showing that this method of lysis is very efficient. Other- wise the two methods of premature lysis yield remarkably similar curves. This suggests that either method accurately measures the intracellular content of infective phage particles at the time of sampling. Two important conclusions about the multiplication of phage T3 may be drawn from these results: (7) during the first half of the latent period there are no mature phage particles inside the infected cell (this interval is called the "eclipse period"), and (2) during the second half of the latent period the number of mature phage particles increases rapidly, reaching an average of about 50 per infected bacterium at the tim.e when spontaneous lysis begins. Doermann (1948b, 1952) also studied the growth of phage T4 by premature lysis, and many other phages have been studied since with similar results. Doermann also noted that cyanide or 5-methyltryptophan alone could cause premature lysis when added after the end of the eclipse period. Other lysing agents have since been used, notably proflavine (Foster, 1948; De- Mars, 1955), dinitrophenol (Heagy, 1950), shaking with glass beads (Joklik, 1952), decompression under nitrous oxide (Fraser, 1951b; Levinthal and Fisher, 1952), glycine at high concentra- tions (Kay, 1952; DeMars, 1955), chloroform (Sechaud and Kellenberger, 1956), and chloramphenicol (Tomizawa and Sunakawa, 1956) (see also Chapter XV). In all of these methods some means must be used to prevent loss of the liberated phage by adsorption to bacterial debris. One may use dilution as in the usual one-step growth experiment, a culture medium unfavorable to adsorption, an inhibitor of adsorption (French, Graham, Lesley, and Van Rooyen, 1952), or addition of a large excess of homologous irradiated phage (Maal0e and Watson, 1951). In the latter case the irradiated STAGES IN PHAGE MULTIPLICATION 175 phage serves both as a lysing agent and as an inhibitor of adsorp- tion but the specificities are difl'erent: T2, T4, and T6 cause lysis fi'om without but only T2 can compete with the adsorption of T2 (Hershey and Chase, 1952). The lysing agents mentioned above doubtless act in many dif- ferent ways. Two extremes may be noted. Cyanide promptly stops biosynthesis generally by blocking certain steps in respira- tory metabolism, and lysis ensues shortly. Proflavine does not stop synthesis of phage-specific materials, but blocks some final step in maturation of phage particles (DeMars, 1 955) . Lysis fol- lows only at the end of the normal latent period. Cyanide and chloroform are perhaps the most generally useful lysing agents. To summarize, premature lysis of infected cells may be induced by various means. During an eclipse period lasting half way through the latent period no intact phage particles can be de- tected within the cell. The noninfective form of the phage be- lieved to multiply in the cells during this time is called" vegeta- tive phage" (Doermann, 1953). Its nature was a mystery until the discoveries of Hershey and Chase (1952) and subsequent work pointing to the identity of vegetative phage and phage-pre- cursor DNA (Chapter XIV). 5. Phage Multiplication Studied by Irradiation of Infected Bacteria The theoretical basis for this method of studying phage multi- plication is quite simple, but as often happens in biological sys- tems the interpretation of the experiments is unexpectedly com- plicated. The inactivation of a suspension of phage particles by ultraviolet light or X-rays is a close approximation to a "one-hit" phenomenon, that is, the surviving fraction is an exponential function of the dose of radiation (Chapter VI). The survival fraction jy equals e~''^ where D is the dose of radiation and the coeflficient k depends on the dosage unit and on the sensitivity of the phage. If a bacterium contains n phage particles and the survival of any one of these is sufficient for survival of the plaque- 176 BACTERIOPHAGES forming potential of the infected bacterium, the survival should be given by the expression ;; = 1 - (1 ■kD\n Such theoretical survival curves for bacteria containing various numbers of phage particles are given in Figure 5, taken from the paper of Luria and Latarjet (1947). ^^ ^i ^ \ V A V \ \\ \ \ V A \ \ V A \ \ \ \^ ^\ \» A \20 y \5o\ 100 \ \ \\ \ \ \ \ \" \ \ ^ \ A V A A \ ^ \ \ \ \^ \ ^ \, A \ \^ i) A \ \ .7 .5 .3 .2 "5 > E .1 W .07 .05 .03 .02 •°' 123456789 dose Figure 5. The relation between dose of irradiation and survival of infected bacteria according to the target theory. The numbers on the curves corre- spond to different numbers of identical targets per bacterium. Reproduced from S. E. Luria and R. Latarjet, 1947, J. BacterioL, 53, 149, with permission. STAGES IN PHAGE MULTIPLICATION 177 In these curves the length of the initial plateau is related to the number of particles per cell, and the ultimate slope of the linear portion is determined by k, characterizing the sensitivity of the individual particles to the radiation. Therefore the survival curves obtained at various times during the latent period should give a clue to the kinetics of phage reproduction. This approach to the study of phage multiplication was made by a number of investigators including T. F. Anderson (1944), Luria and Latar- jet (1947), Latarjet (1948), Benzer (1952), and Benzer and Jacob (1953). The subject has been reviewed by Latarjet (1953). An assumption underlying this method is that the loss of the plaque-forming ability of infected bacteria is due to the effect of the radiation on the intracellular phage particles, rather than on the host cell mechanisms required for phage multiplication. Anderson (1944, 1948d) and Benzer (1952) demonstrated that E. coli B treated with many lethal doses of ultraviolet light re- tained the capacity to support multiplication of bacteriophage T2. The same result for bacteria killed by X-rays was observed by Rouyer and Latarjet (1946) and Labaw, Mosley, and Wyck- ofF (1953). Very heavy doses could render bacteria incapable of serving as hosts for phage growth, but such doses were much larger than those used in the inactivation of intracellular phage. In the case of some phage-bacterium systems, however, the "capacity" may be very sensitive (Benzer and Jacob, 1953). The essential steps in a so-called "Luria-Latarjet experiment" are the following: (7) phage is mixed with bacteria and ad- sorption permitted to occur for a short period of time; (2) un- adsorbed phage is eliminated, and the number of infected bac- teria is determined; {3) phage multiplication is permitted to take place, and samples are removed at intervals during the latent period; (4) each sample is exposed to several doses of radiation; and (5) the surviving fraction of infective centers is determined for each dose of radiation, and a survival curve is constructed for each time of sampling. In experiments of this type it is important to have phage 178 BACTERIOPHAGES growth in the population of infected bacteria synchronized as much as possible, and to be able to arrest phage development during the irradiation. Benzcr (1952) devised techniques which solved these problems for T2 and 17. The bacteria were grown in broth, then aerated in buffer at 37 ° C. for one hour to starve them. Adsorption of phage added to such bacteria is rapid but the infective process is arrested at a very early stage. If broth at 37 ° C. is subsequently added, phage growth starts and pro- gresses normally. This technique permits the separation of adsorption from growth, and presumably phage growth starts in all bacteria simultaneously when broth is added. A defect of this method is that a considerable proportion of the adsorbed phage particles may be inactivated during the adsorption period, a phenomenon known as "abortive infection." Another tech- nique for arresting development without interfering with adsorp- tion is the use of cyanide. When the cyanide is removed by dilu- tion, development starts promptly (Benzer and Jacob, 1953). Phage growth can be stopped at chosen times during the latent period by dilution in chilled buffer. Chilling halts phage de- velopm^ent and the samples can then be irradiated at conven- ience; dilution serves to reduce the ultraviolet absorptivity of the medium, which is prohibitive with broth. The results for phage T7 (Benzer, 1952) come close to the theoretical expectations as may be seen in Figure 6. For the first 3 minutes of the latent period the survival curves are similar to that for free phage T7. At later times, the curves become multiple hit in character, in- dicating that phage multiplication has started. By the sixth minute, the curve indicates a relatively high multiplicity but the ultimate slope is only slightly less steep than for free phage. For the remainder of the latent period the multiplicity increases further and the ultimate slope decreases som.ewhat. These sur- vival curves do not have the flat initial plateau seen in the theoretical curves of Figure 6, but have a steeper initial gradient. Benzer (1952) suggested that this may reflect lack of synchroni- zation of phage growth in different bacterial cells. Similar experiments with T2-infected bacteria were carried STAGES IN PHAGE MULTIPLICATION 179 out by Luria and Latarjet (1947) and by Benzer (1952). The results with this phage are strikingly different from those ob- tained with T7. Immediately after infection, the infective cen- ters have the same sensitivity to ultraviolet inactivation as free phage, but there follows an increase in resistance to inactivation until, at 5 or 6 minutes after infection, the resistance has doubled. 100 200 300 UV DOSE IN SECONDS Figure 6. Survival of phage-producing capacity of bacteria infected with T7. The bacteria were irradiated at times after infection indicated on the curves in minutes. Latent period (at 37° C.) 12 minutes. Reproduced from S. Benzer, 1952, J. BacterioL, 63, 59, with permission. Thereafter the resistance increases with great rapidity until maximum resistance, 20-fold greater than that of free phage, is reached at about 10 minutes after infection. At this time the in- activation curves become definitely multiple hit in character. From 10 minutes to the end of the latent period the multiplicity of the inactivation curves increases while the slope of the curve 180 BACTERIOPHAGES becomes steeper, the sensitivity approaching again that of free phage. This remarkable change in sensitivity to ultraviolet in- activation during the first half of the latent period was com- pletely unexpected and entirely different from the behavior of phage T7. It cannot be explained on the basis of screening by ultraviolet absorbing materials (Benzer, 1952). Latarjet (1948) studied the sensitivity of T2-infected bacteria to inactivation by X-rays at various times during the latent pe- riod. During the first 7 minutes, the sensitivity remained essen- tially the same as that of extracellular virus. Then, during the 8th and 9th minutes there was an increase in resistance to X-rays without a change in the multiplicity of the inactivation curves. From the 9th to the 13th minute the multiplicity increased very rapidly to a value of over 100 with little change in the sensitivity to inactivation. For the remainder of the 21 -minute latent pe- riod there was little change in the multiplicity of the inactivation curves, but there was a gradual increase in X-ray sensitivity, tending toward that of extracellular phage. An interpretation of the results obtained with T2 (Benzer, 1952) is that infection may be followed by a series of steps, each step having a certain cross section for interference by ultraviolet light. For free phage, or immediately after infection, the total cross section is the sum of the individual cross sections. As de- v^elopment proceeds, the completed steps drop out and the cross section progressively decreases. This process could be related in some way to multiplicity reactivation. Both the rapid change in ultraviolet sensitivity during the latent period and the phenom- enon of multiplicity reactivation occur with phages of the T2 serological group; neither efi'ect occurs in the T7 serological group. The remarkable resistance of the phage-producing mechanism to agents supposedly acting on the intrabacterial DNA can be shown in another way. Phage particles containing P^^ of high specific radioactivity are subject to "suicide" owing to the decay of radiophosphorus, which presumably produces local damage in the viral DNA. Stent (1955) applied this principle in experi- S7AGES IN PHAGE MULTIPLICATION 181 ments designed to explore the role of DNA during infection. He infected bacteria with phage in various combinations such that the parental phage DNA, the newly synthesized intrabacterial DNA, or both, contained ?■'- of high specific radioactivity. Phage growth was allowed to proceed for varying time intervals, and samples of the infected culture were frozen. Decay of radiophosphorus occurred for various times in the frozen state, after which the samples were thawed and titrated to measure the number of infected bacteria still capable of producing phage. The results may be summarized as follows. /. When only the parental viral DNA contained P'^-, the ability of the infected cells to produce phage was initially about as sensitive to radioactive decay as are free phage particles. As phage development progressed, however, this sensitivity was gradually lost. By the ninth minute it had disappeared com- pletely. 2. When the newly synthesized viral DNA (and indeed all the intrabacterial phosphorus excepting the parental viral DNA) contained P'-, the ability of the infected cells did not at any time reach a stage at which the ability to produce phage was ap- preciably sensitive to radioactive decay. 3. When all the intrabacterial DNA contained P^-, the results were virtually the same as when only the parental viral DNA contained P^^. These results show in an interesting way the dependence of viral growth during its early stages on the integrity of the paren- tal viral DNA. Otherwise they show the same remarkable fact revealed by the experiments with ultraviolet light, namely, that the T2 phage-producing mechanism becomes highly resistant to damage to intrabacterial DNA near the mid-point of the latent period. Stent suggests three ways in which such stabilization might be brought about: (/) a stage in viral growth may be reached after which damage to DNA can be repaired, as in the phenomenon of multiplicity reactivation (Chapter XVIII) ; (2) a stage may be reached in which some or all the intrabac- terial DNA becomes insensitive to radiochemical damage; or 1 82 BACTERIOPHAGES (3) a stage may be reached after which the intrabacterial DNA is dispensable, presumably owing to a transfer of DNA function to other substances not containing phosphorus. The tliird alternative was also suggested by Tomizawa and Sunakawa (1956) on the basis of experiments showing that chloramphenicol, added to cultures infected with phage T2 under conditions permitting synthesis of DNA but not protein, blocked those processes resulting in stabilization of the infected bacteria toward the destructive effects of ultraviolet light. At the present time no decisive choice can be made among the three hypotheses suggested by Stent. It should be recalled, however, that the experiments with T7 do not show the same phenomenon, and one can ask which result reflects basic features of viral growth least complicated by side eff'ects. These and other related experiments are discussed in somewhat different contexts by Hershey (1957) and Stent (1958). 6. Morphological Stages in Phage Development Objects the size of bacteriophages can be studied by means of the electron microscope, which has revealed structures that may be intermediate stages in the maturation of infective particles. One of these is known by the colloquial namie "doughnut." It is a crumpled disk with a central depression which makes it look like a torus in shadowed electron micrographs. Beautiful pic- tures of doughnuts were published by Wyckoff' (1949b). They are also illustrated in papers by Hercik (1955), Heden (1951), Levinthal and Fisher (1952), DeMars, Luria, Fisher, and Levin- thai (1953), and T. F. Anderson, Rappaport, and Muscatine (1953). The latter authors prepared specimens by the critical point method and show that doughnuts are really empty mem- branes of the same size and shape as the phage head but without any tail. They have been found in association with the large, tailed phage particles T2, T4, T6, and T5. The size, shape, and occurrence of these particles suggested that they might be an intermediate stage in reproduction. Evidence in support of this view was obtained by Levinthal and STAGES IN PHAGE MULTIPLICATION 183 Fisher (1952) in quantitative studies of T2-infected bacteria which were prematurely lysed by the decompression technique at different times during the latent period. The lysates were mixed with polystyrene latex particles and sprayed on collodion films for electron micrography, thus permitting counts of the number of particles per infected bacterium. Doughnuts are first detected at about the 9th minute of the latent period. By the 12th minute, when mature phage first appears, the doughnuts number about 15 per bacterium. At the 21st minute their num- ber reaches a inaximum of 30 per cell, while mature phage par- ticles continue to increase, numbering 60 per cell at 25 minutes. It seems reasonable to conclude that the doughnut represents a stage in the assembly of phage particles. The relationship of phage nucleic acid to these structures is not clear. It is possible that the incomplete forms contain nucleic acid that is lost during the preparative manipulations. Relatively pure preparations of doughnuts can be obtained by the use of proflavine, which specifically inhibits final steps in the maturation of phage particles (DeMars, Luria, Fisher, and Levinthal, 1953). The incomplete particles can be readily con- centrated and purified by high speed centrifugation. On ex- amination in the electron microscope they are indistinguishable from doughnuts observed by other means and are essentially free of tailed particles. The formation of doughnuts during early stages of viral growth occurs at the same rate with or without proflavine. The preparation of purified concentrates of doughnuts ob- tained by the proflavine technique permitted a study of their serological properties. They fix complement with antiphage serum with about half the efficiency per particle of mature phage. They are precipitated by antiphage serum but do not adsorb phage-neutralizing antibodies from the serum. Purified dough- nuts were used for the immunization of rabbits by Lanni and Lanni (1953). The resulting sera gave very strong complement fixation reactions with both doughnuts and matui^e phage par- ticles, but contained very little neutralizing antibody. From 184 BACTERIOPHAGES this and other evidence these authors concluded that the dough- nuts are serologically equivalent to the heads of phage particles (Chapter VIII). Proflavine lysates were prepared ft'om T2-infected bacteria labeled with either P'^- or S^'". The isolated doughnuts were found to contain about -/s as much sulfur per particle as mature phage, but less than Ve as much phosphorus. These observa- tions are in agreement with the electron microscope studies in suggesting that the doughnuts are empty phage heads, made of protein and containing little or no nucleic acid. The dough- nuts do not adsorb to bacteria, which is consistent with the absence of tails. Further morphological study was facilitated by the develop- ment of the "agar filtration method" of quantitative electron microscopy (Kellenberger and Kellenberger, 1952; Kellenber- ger and Arber, 1957). In this method a sample of crude lysate is placed on a collodian membrane supported on a specially pre- pared agar surface. Water and crystalloids pass into the agar, while larger particles are deposited on the membrane. The material is fixed with formaldehyde vapor and the membrane is transferred to a specimen grid for electron microscopy. Count- ing of particles per unit volume of lysate is achieved by calibra- tion with latex spheres. The advantage of this method over others is that no fractionation of the lysate is necessary before preparation of specimens. Kellenberger and Sechaud (1957) applied this method to pre- mature lysates of both normal and proflavine cultures of bac- teria infected with T2 and T4. They found, in confirmation of earlier work, that doughnuts appeared a few minutes before phage particles in the lysates, and soon reached a constant num- ber of about 50 per bacterium, whereas phage particles increased for a longer period, to 150 per bacterium. Proflavine greatly suppressed the formation of phage particles but did not afTect the formation of doughnuts. This result is unexpected, because, if proflavine acts mainly to prevent the conversion of doughnuts into phage particles, some additional mechanism must be invoked STAGES IN PHAGE MULTIPLICATION 185 to explain the failure of doughnuts to accumulate in response to the drug. However, the effect of proflavine may be more com- plicated (Chapter XV). Kellenberger and Sechaud were able to count tail pins as well as doughnuts (Chapter IV). Tail pins did not appear until phage particles had already become very numerous, and finally Figure 7. Products of lysis by T2 in the presence of proflavine. Intact phage particles, empty heads, tail pins and, in the upper part of the micro- graph, fibrils that may be DNA. X22,125. Reproduced from E. Kellen- berger and J. Sechaud, 1957, Virology, 3, 256, with permission. numbered about 30 per bacterium. Their formation was not affected by proflavine. The delayed appearance of tail pins might suggest that they are products of defective synthesis or breakdown of phage particles, rather than precursors, in which case some of the doughnuts might originate in a similar way. Kellenberger and Sechaud found that tail pins of T2 adsorb in- completely and mostly reversibly to bacteria. They do not react with the neutralizing antibody of antiphage serum. 1 86 BACTERIOPHAGES Figure 7 shows an electron micrograph of the products re- leased by lysis of bacteria infected with T2 in the presence of pro- flavine. The same and other intermediate structures related to viral growth have been detected by serological methods (Chapter VIII) and by chemical methods (Chapter XIV). The total in- formation may be summarized as follows. Infected cells con- tain at least four phage-specific materials in addition to phage particles: free DNA, doughnuts, tail pins, and free tail antigen, all separable from each other in lysates. Only free tail antigen and free DNA are caused to accumulate by proflavine, that is, proflavine does not aff'ect the total amount of these substances formed, but suppresses their incorporation into phage particles (DeMars, 1955; Kellenberger and Sechaud, 1957). Doughnuts, tail antigen, and tail pins are probably made of protein. Tail antigen might be expected to correspond to the tail fibers of phage particles. When prepared by degradation of phage particles, the fibers can adsorb to bacteria (Williams and Fraser, 1956). Free tail antigen found in premature lysates, however, cannot. The significance of this difference cannot now be assessed, because present information about the adsorp- tion characteristics of materials found in premature lysates seems to be unreliable. Thus doughnuts never adsorb, but Maal0e and Symonds (1953) reported adsorption of an S^Mabeled fraction in T4 lysates that should have been chiefly doughnuts. Kellen- berger and Sechaud (1957) observed adsorption of tail pins found in T2 lysates, but none of the materials, including phage particles, found in their T4 lysates would adsorb. Unfortu- nately little of the work with premature lysates has been carried out with adequate precautions to prevent interactions between phage-specific materials and bacterial debris in the lysates, partly because these interactions are not themselves understood (Sagik, 1954) (Chapter XVI). The free DNA found in premature lysates undoubtedly is a phage precursor (Hershey, 1953a). Some of the antigenic pro- tein is, too (Maal0e and Symonds, 1953; Hershey, 1956b). SIAGES IN PHAGE MULTIPLICATION 187 The phage-specific protein precursors include materials that sediment like doughnuts as well as smaller fragments, in approxi- mately equal amounts as measured by sulfur content (Hershey, 1956b). 7. Conclusion Phage growth is initiated by the injection of the viral DNA into the cell. Completed phage particles do not reappear until the mid-point of the latent period. Morphologically recognizable elements are formed somewhat earlier. Radiobiological analy- sis of phage growth also shows that important processes occur during the eclipse period of viral growth. These and other lines of evidence suggest that noninfective phage particles, called vegetative phage, are multiplying during the period of eclipse. Attempts to enumerate vegetative phage particles in infected cells by radiobiological methods have generally failed. Instead the experiments show that vegetative phage T2 is remarkably resistant to ultraviolet light, X-rays, and decay of constituent radiophosphorus. Vegetative phage T7 does not show this property. For this and other reasons the proper interpretation of the radiobiological experiments is not clear. CHAPTER Xll GYTOLOGIGAL CHANGES IN THE INFEGTED HOST GELL This chapter is devoted to studies of visible changes in bacteria caused by phage infection, and to the sequence of cytological events during the latent period of intracellular phage multipli- cation. These studies may be conveniently classified into cate- gories on the basis of the technical methods used : ( 7) direct ob- servation of living cells by dark field, ultraviolet, and phase con- trast microscopy ; (2) observation of fixed and stained prepara- tions to determine the succession of cytochemical changes; and (3) observation in the electron microscope of fixed and dried whole cells and of embedded and sectioned preparations to ob- tain higher resolution than is possible in the first two techniques. All methods agree in demonstrating that marked changes oc- cur in bacterial cells as a result of phage infection. A major problem is the correlation in time of the events observed by the various techniques, and almost as difficult is the elimination or control of artefacts caused by the technical methods employed in the preparation of specimens for observation. Other complicat- ing factors include variation in the sequence of events with dif- ferent phages attacking the same host strain, and variation in the time required for intracellular virus growth under different en- vironmental conditions, which make it more difficult to compare the experiments of difTerent investigators. These technical hurdles as well as the highly developed imaginations of the inves- tigators are responsible for the great diversity of interpretations which have been published. 189 190 BACTERIOPHAGES 1. Observations on Living Cells The lysis of bacteria by phage may be readily observed under the ordinary light microscope. For such studies a culture of the susceptible bacterium in the logarithmic growth phase is exposed to an excess of bacteriophage under conditions in which at least 90 per cent of the bacteria are infected within a few minutes. The infected bacteria may then be observed under oil immersion by hanging drop methods or on the surface of agar under a cover glass. Bacterial lysis begins after the usual latent period for the conditions of temperature and bacterial nutrition used. The time course for the number of bacteria lysed corresponds to that for phage liberation as observed in the one-step growth experi- ment, which suggests that the lysis of a bacterium coincides with the liberation of the mature phage particles contained in that bacterium. With most of the phage-host cell systems studied, infection in- hibits further cell division. An exception occurs with certain temperate phages studied by Lwoff, Siminovitch, and Kjeld- gaard (1950) in which the infected cell undergoes one or two divisions before lysis. In many cases no gross change in cell morphology occurs up to the moment of lysis, at which time the refractive index of the cell suddenly changes, the cell bursts and becomes invisible. Studies of this event by cinematography have demonstrated that cell disappearance can occur within 3 seconds, the time interval between successive photographs (Bronfen- brenner, Muckenfuss, and Hetler, 1927; Bayne- Jones and Sand- holzer, 1933). Many observers have reported morphological changes occur- ring in cells after infection. In some cases the cells swelled markedly and became nearly spherical in shape before bursting. In other cases greatly elongated, filamentous forms developed which might or might not lyse (Burnet, 1925; Bronfenbrenner, 1928). This variety of changes in cell morphology gave rise to some controversy in the early literature (see Bronfenbrenner, 1 928) and may have been due to differences in phage and host strains or differences in environmental conditions in use in CYTOLOOIGAI, CHANCES IN THE INFECTED HOST CELL 191 different laboratories. An important development was the demonstration by Delbrtick (1940b) that two distinct types of ly- sis could be observed with the same phage-host system depending on the multiplicity of infection. Under conditions of normal in- fection there was no change in cell shape up to the moment of lysis. In contrast, when the multiplicity of infection was of the order of 200 phage particles per cell, there was a relatively rapid swelling of the rod-shaped bacteria into an oval or spherical shape which then slowly faded out. The latter process, termed "lysis from without," was never accompanied by liberation of viable phage particles. At intermediate multiplicities of infec- tion part of the bacterial population would lyse by the normal, productive bursting, while the remainder of the population would suffer the unproductive "lysis from without." There has been relatively little study in recent years of modes of lysis during bacteriophagy and a comparative study of various phage-host cell systems might well be rewarding. Observations of the lytic process in the dark field microscope were first reported by d'Herelle (1921, 1926). "Au debut, I'as- pect des bacilles est normal; apres quarante-cinq a soixante minutes on voit des fins granules, de plus en plus nombreux, dans I'interieur des bacilles, le nombre des bacilles avec granules inclus augmente rapidement avec diminution parallele du nombre des bacilles normaux." As the number of intracellular granules increased the bacteria swelled, became spherical, and burst, liberating the granules into the medium. The bacterial debris, however, quickly became invisible in the ultramicroscope. D'Herelle concluded that the highly refractile granules that he observed inside the infected bacteria were actually the bacterio- phage particles. He based this conclusion on two kinds of evi- dence: there was always a parallelism in lysates between the number of granules seen in the ultramicroscope and the number of phage corpuscles determined by plaque count; and the num- ber of refractile bodies counted per bacterium infected with the Shiga phage under study varied from 15 to 25, corresponding to an average: yield of 1 8 phage corpuscles for one particle inocu- 1 92 BACTERIOPHAGES lated as determined by plaque counts. The quantitative rela- tionship between the number of highly refractile particles seen in the dark field microscope and the number of infectious par- ticles determined by plaque count was confirmed much later by Schlesinger (1933b). Dark field observations of phage lysis of E. coli in hanging drop cultures, made by Merling-Eisenberg (1938), may be summa- rized as follows. He found that the motile organisms became motionless following infection. Then fine internal granules be- gan to appear, which were in motion in some cases, while the bacterium increased in size. After a variable period of time lysis occurred. The disruption might occur without any fur- ther change in the size or shape of the bacillus as if the whole cell membrane suddenly became permeable. On other occasions an opening appeared at one end or at the side of the bacterium, through which the granules were spewed. The explosion re- leased a very distinct cloud of particles with a blue diffraction color showing marked Brownian movement and soon leaving their "birthplace." They could be readily distinguished from bacterial debris and unspecific particles by this blue color and their fairly uniform size. The phage bodies (1 50 to 350 per cell) were expelled by the force of the explosion to a distance esti- mated at 5 to 6 times the length of the bacterium. During lysis the number of phage bodies free in the medium soon reached astro- nomical figures. This very interesting description of lysis is illus- trated with photographs which clearly demonstrate the refractile granules both inside the bacteria and in the medium. Merling- Eisenberg (1941) made similar observations of phage lysis of staphylococci, in which 1 to 4 granules per cell were released. Very nice motion pictures of phage lysis were made with dark field illumination by Pijper (1945), which are in agreement with the earlier observations described above. Similar dark field observations were reported again bv Weigle (see Benzer, Del- briick, Dulbecco, Hudson, Stent, Watson, Weidel, Weigle, and Wollman, 1950). The possible use of the dark field ultraviolet microscope for CYTOLOGIGAL CHANGES IN THE INFECTED HOST CELL 193 Studying phage multiplication was suggested by the work of Barn- ard who published very fine pictures of some of the larger viruses. Pictures of phage CI 6 before and after agglutination by anti- phage serum were taken by Barnard and published by Burnet (1933c). The phage particles were very easily visible, could be counted, and were barely below the limits of resolution of the microscope. This beautiful optical system does not seem to have been applied to the study of phage-infected bacteria. The use of the light field ultraviolet microscope for pho- tography of phage infected bacteria by F. L. Gates was described by Bronfenbrenner (1928). He wrote: "In unstained cells photographed at this stage (swollen but not lysed) by means of ultraviolet light, the cytoplasm appears to be of uneven density, quite unlike that of normal bacteria." Studies of E. coli infected with phage T2 were made by ultra- violet microscopy by Heden (1951) and compared with observa- tions by phase contrast microscopy, electron microscopy, and light microscopy of stained preparations. This permitted a direct comparison of developmental changes in unfixed and living cells (ultraviolet and phase contrast observations) with corre- sponding stages in fixed cells (electron microscope and stained preparations). Uninfected cultures of £". coli in the logarithmic growth phase were uniformly opaque in the ultraviolet micro- scope, presumably because of the cytoplasmic RNA which ob- scured the DNA-containing bodies, which are so prominent in Feulgen stained cells. After infection with T2 and incubation the bacterial poles gradually became much more light-absorbent at 2,570 A than the equatorial region. After several hours of in- cubation under conditions of lysis inhibition the polar bodies con- tained several times more light-absorbing material than did the rest of the cytoplasm. This contrast between polar and equato- rial regions was not evident at a wave length of 3,300 A, suggesting that nucleic acid is in reality the light-absorbing material. In phase contrast studies of living bacteria, the cells appeared uniform until infected. Within 15 minutes after infection the bipolar appearance became evident, and the contrast between 1 94 BACTERIOPHAGES polar and equatorial regions increased until lysis. In dried preparations fixed with osmium the bipolar arrangement had disappeared. Polar bodies were not seen in fixed and stained preparations, which suggests that the procedures of fixation and dehydration may create artefacts having little resemblance to the situation in living infected cells. Similar phase contrast observations of living, phage-infected cultures were made with Salmonella typhimurium (Boyd, 1949a) and E. coli (Boyd, 1949b). Again the uniformly dense appearance of the uninfected bacteria quickly changed after infection to a bi- polar distribution of dense areas. For reasons that are not clear, Boyd assumed that the less dense areas are due to bacteriophage and the darker polar areas are clue to bacterial cytoplasm dis- placed toward the poles by the developing phage. The obser- vations of Heden with the ultraviolet microscope, referred to above, suggest that it is the DNA and, presumably, bacteriophage synthesis which is concentrated at the poles rather than the bac- terial cytoplasm. This suggestion of Heden's is further sup- ported by the dark field photographs of phage-infected bacteria taken by Merling-Eisenberg (1938) which clearly show a polar distribution of the phage particles. Boyd (1949b) studied all seven T phages by phase contrast microscopy using strain B of E. coli as host. He found charac- teristic changes in cellular appearance which were similar for serologically related phages but differed from one serological group to another. These observations are significant in demon- strating that the physiological response of the bacterium to in- fection depends on the genetic constitution of the phage and sug- gest a rather remarkable control by the phage of enzymatic processes in the bacterium. Such characteristic phage-specific changes in morphology have been defined with greater clarity in stained preparations as we will see. Other observations of phage action using phase contrast have been reported by Hofer (1947), Delaporte (1949), Rice, McCoy, and Knight (1954), and Murray and Whitfield (1953). CYTOLOGICAL CHANGES IN THE INFECTED HOST CELL 195 2. Observations on Fixed and Stained Preparations Stained smears, made after the fashion of diagnostic bacte- riology, were the basis of the earlier observations and they were not very informative. More recently, however, cytological methods appropriate for the study of bacteria have been de- vised and these have provided a more rewarding approach to the study of the effects of phage infection on the structure of the host cell. These involve the fixation in situ of the cells in or on the culture medium with cytological fixatives (such as osmium tetr- oxide and other, more complex, solutions) ; the application of selective staining methods allowing the differentiation of the nu- clear chromatin bodies and the cytoplasm, which is strongly basophilic due in large part to RNA (see Murray, Gillen, and Heagy, 1950); and suitable mounting of the preparation to minimize distortion of the cell and to provide for optical needs. D'Herelle (1926) described briefly the effects of phage in- fection on the staining properties of the bacterial host. The affinity of the infected bacteria for basic dyes gradually decreased. Weakly staining, amorphous debris gradually increased in amount as the number of intact bacteria decreased. He never saw corroded or broken bacterial cells. The intact cells often became spherical and then burst, becoming amorphous im- mediately. The spherical forms were extremely fragile and were always less numerous in stained specimens than in dark field preparations made at the same time. The phage particles them- selves were below the limits of visibility in the stained prepara- tions. Rather similar observations were recorded by Bronfenbrenner (1928). The cytoplasm showed marked changes during the swelling of the bacteria, staining less intensely and more un- evenly so that the bacteria appeared segmented or beaded. In highly swollen bacteria the chromatin was distributed through- out the bacteria in the form of dustlike particles. Borrel (1928, 1932) used an iron-tannate-fuchsin mordant which was effective in making visible bacterial flagellae and the larger animal viruses such as vaccinia and moUuscum contagio- 196 BACTERIOPHAGES sum. When applied to phage-infected bacterial cultures during lysis it enabled him to see a halo of adsorbed phage particles sur- rounding the yet unlysed bacteria. Although this procedure may have permitted visualization of extracellular phage particles, it did not contribute to knowledge of the infectious process. Hofer (1947) also used a flagellum stain and the Ziehl-Nielsen stain in an attempt to color extracellular phage particles but again without yielding much information about phage multipli- cation. His claim to have seen phage particles under these con- ditions is not convincing. A remarkable paper by MacNeal, Frisbee, and Krumwiede (1937) described the application of the Casteneda rickettsial stain to the study of the lysis of the cholera vibrio by phage and by antiserum plus complement. The phage-infected bacteria were pleomorphic, enlarged and distorted, and filled with as many as 50 tiny blue-staining granules per cell. Such granules were never seen in normal bacteria, nor in bacteria undergoing lysis by complement and antibody. It seems very probable that the granules were intracellular phage particles. This work does not seem to have been followed up nor to have attracted much notice. The first applications of chromatin-staining procedures to the study of phage-infected bacteria seem to have been made by Luria and Palmer (1946) and by Beumer and Quersin (1947). Luria and Palmer used Feulgen and Giemsa stains following the general procedures of Robinow (1944). Infection with phage T2 resulted in the disorganization of the nuclear bodies and mi- gration of the chromatin to the periphery of the cell. The cells then gradually filled up with granular chromatin. Phage Tl caused relatively little change in the appearance of most cells. Infection with phage T7 caused great swelling of some nuclear bodies and shrinking of others. The different phages caused characteristically distinct cytological changes in appearance of the host cell nuclear apparatus. Beumer and Quersin (1947) and Quersin (1948) used Robi- now's staining techniques to study the cytological changes pro- CYTOLOGICAL CHANGES IN THE INFECTED HOST CELL 197 duced bv several different phage strains on a number of different host bacteria. They found that in general the bacterial nucleus was distorted or destroyed and replaced with a mass of chro- ma tinic material, and the bacterium swelled. The type of lesion produced was characteristic of the phage and not of the host cell because the same phage produced essentially the same sequence of changes in different bacteria, whereas different phages acting on the same bacterial strain produced characteris- tically different lesions. Delaporte (1949), using similar meth- ods, studied the action of phage T4 on normal E. coll cells and on ultraviolet-treated bacteria. Lysing cells liberated large numbers of tiny, stained particles which may have been the phage particles. The action of a phage on B. cereus was also de- scribed briefly. Rita and Silvestri (1 950) gave a brief description of modifications in the host cell nucleus resulting from infection of £". coli with phages of the T series, including T5. P'an, Tchan, and Pochon (1949) studied the action of phage on Pasteur ellapestis. The cell and its nuclei increased in size after infection, and the nuclei fused into a convoluted axial filament. At lysis the cell outline became less distinct and the axial filament disappeared. A rather detailed description of progressive cytological changes produced in E. coli following infection with phages Tl, T2, T4, T6, and T7 was published by Luria and Human (1950). Sam- ples of the infected bacteria were taken at intervals during the latent period and added to a formalin-dichromate fixative. Af- ter fixing the bacteria were washed, and spread on agar, and im- pression films were stained by the Robinow technique. Again the sequence of cytological changes was found to be characteris- tic of the phage type ; the effect of the closely related phages T2, T4, and T6 being very similar but quite distinct from the changes caused by Tl and T7. With ultraviolet-inactivated T2 phage, infection was followed by the same initial destruction of the bac- terial nucleus as occurs with active T2 phage. However, these cells did not subsequently fill up with granular chromatin as do cells infected with active phage. With inactivated Tl phage 1 98 BACTERIOPHAGES the cytological changes were the same as those observed with active Tl except that lysis did not occur. With inactivated T7 phage the initial changes were the same as with active T7 phage, but then the cells filled up with chromatin like Tl -infected cells instead of forming a large central chromatinic body as was seen with active T7 phage. In the case of infection with irradiated Tl phage, synthesis of material absorbing at 2,600 A proceeded at the same rate as with active phage, but with irradiated T2 phage the synthesis of light-absorbing material proceeded at a relatively slow rate compared with synthesis in bacteria infected with active phage T2. Heden (1951) reported that his cytolog- ical studies with T2-infected E. coli cells were in essential agree- ment with the results of Luria and Human. An intensive study of the cytological changes produced in E. coli by infection with phage T2 was made by Murray, Gillen, and Heagy (1950). The staining methods of Robinow were used, and replicate samples of infected bacteria taken at intervals dur- ing the latent period were stained with Giemsa (after treatment with HCl) for chromatin, with thionine for cytoplasm, and with tannic acid and gentian violet for cell walls. The Giemsa and thionine staining methods complemented each other and the cell wall stain permitted the visualization of bacterial "ghosts" after lysis had occurred. The observations with the Giemsa stain were in agreement with those of Luria and Human (1950). During the first few minutes the bacterial nucleus disintegrated and the chromatin migrated to the periphery of the cell forming "marginated cells." Then during the second half of the latent period, this chromatin became granular. In lysis-inhibited cells the granules increased in size and coalesced, giving a banded ap- pearance. The staining capacity of the cytoplasm with thionine decreased and was alm.ost lost at the time of lysis. The authors suggested that late in the latent period the chromatin is distrib- uted cylindrically near the cell wall. This paper is illustrated with beautifully reproduced photographs of stained preparations. Studies of T2-infected bacteria by Beutner, Hartman, Mudd, and Hillier (1953) using DeLamatcr's staining technique are in agree- ment with the results obtained by the previous investigators. CYTOLOGICAL CHANGES IN THE INFECTED HOST CELL 199 Murray and Whitfield (1953) made a detailed study of the effects of phage strains of the T5 species on various host cells. Again it was found that the cytological changes were characteris- tic of the phage and independent of the particular host strain, whether Escherichia, Shigella, or Salmonella. The sequence of changes was similar to that observed with the T-even phages: cessation of cell division, loss of stainable nuclear chromatin and disruption of nuclear sites, followed during the second half of the latent period by a new synthesis of finely granular chromatin. However, unlike the T-even phage infections, the loss of stain- able chromatin was virtually complete and was not accompanied by migration to the periphery of the cell. In the case of three of the phages (T5, BG3, and 29 alpha) lysis occurred at this stage; with three other strains (PB, PBl, and poona) there was an ac- cumulation of spherical masses of chromatin during" the last 10 minutes of the latent period preceding lysis. Five hybrids of T5 with PB isolated and characterized by Adams (1951b) were studied by Murray and Whitfield. Three were found to behave like the T5 parent and two like the PB parent. The cytological pattern segregated independently of host range, latent period, and heat resistance, and thus behaved as an additional genetic marker for this group of phages. Adsorption of ultraviolet- inactivated phage to the host cell caused the initial nuclear disin- tegration, but the secondary synthesis of granular chromatin did not occur. Adams (1949b) had previously demonstrated that calcium ion was essential for some step in the reproduction of phage T5 subsequent to adsorption. Murray and Whitfield found that when infection occurred in the absence of calcium ion there was nuclear disintegration as usual, but the new synthesis of granular chromatin did not follow. Later Luria and Steiner (1 954) demonstrated that calcium was essential for the penetration of phage T5 nucleic acid into the host cell. These results show that adsorption of phage T5 to the host cell is of itself enough to cause nuclear disintegration, even though the phage nucleic acid does not penetrate to initiate phage replication. Adsorption of phage T5 in the absence of calcium may be analogous in its ef- fects to the adsorption of ghosts of phage T2 (Herriott, 1951a). 200 BACTERIOPHAGES Bonifas and Kellenberger (1955) observed that adsorption of ghosts of phage T2 caused an aggregation of the host cell chroma- tin rather than the peripheral migration observed when either active or ultraviolet-inactivated phage was adsorbed. This was confirmed by Whitfield and Murray (1957) who further ob- served that if the multiplicity of infection with the active phage was increased to 70 or 100 phage per cell chromatin aggrega- tion preceded the premature "lysis from without." Multiple infection with T5 phage likewise caused aggregation to precede the usual loss of stain able chromatin. The above observations suggest that alterations of the cell boundary, more severe follow- ing infection with ghosts than with phage, contribute to cytologi- cal effects. Puck and Lee (1954, 1955) suggested that infection with any of the T phages initiates a cycle of permeability changes in the cell boundary during the period following the attachment of the phage to the cell. Direct evidence of visible as well as chemical disorganization of cell membranes by phage adsorption has also been noted (Chapter XI). The studies of Whitfield and Murray (1956, 1957) showed that some of the cytological effects of infec- tion may be secondary to permeability changes. They found that the chromatin aggregation, so commonly observed as a first cytological stage of infection (e.g., with phage T7), occurred at low multiplicities of infection if a sufficient quantity of sodium chloride was present in the medium; on salt-deficient medium this effect was replaced by chromatin fragmentation. With phages T2 and T5 the same was true, provided that there was a sufficient multiplicity of infection. The results suggest that the conformation of chromatin is dependent on the ionic balance maintained in the living cell; disturbance of this balance, whether by phage adsorption to the cell surface or by metabolic interference, is reflected in the subsequent behavior of chroma- tin, which varies according to the cationic environment provided for the experiment. These considerations underline the cau- tions expressed at the beginning of this chapter. Lysogenic systems have not been subjected to much cytological CYTOLOGIGAL CHANGES IN THE INFECTED HOST CELL 201 Study. Observations of cultures of lysogenic B. megaterium 899 after induction by ultraviolet light have been reported by Dela- porte and Siminovitch (1952). For the first 17 minutes after irradiation the bacterial cells continue to grow and divide in a manner that is not distinguishable from nonirradiated cultures. After this time the nucleus is disposed as an axial filament nearly as long as the cell. This then begins to swell at some points and the chromatin seems to move to form the surface of a cylinder, considerably larger in diameter than the original filament. Dur- ing this time the bacterial cell increases in both length and diam- eter. The central nuclear cylinder is about Vs the diameter of the bacterium at the end of the latent period. At the time of lysis the bacterial membrane loses its rigidity and the cytoplasm disperses, followed immediately by the disappearance of the axial cylinder. This sequence of events diflfers from that seen with previously studied lytic systems. The authors suggest that the phage particles are synthesized at the surface of the hyper- trophied nuclear filament. Whitfield and Murray (1954) studied the sequence of cytolog- ical events during the lysogenization of Shigella dysenteriae with the temperate phages PI and P2. The first stage of infection with both phages consisted of the condensation of the chromatin into annular aggregates which tended to fuse and form into an axial filament. In PI infections this stage occupied about half of the latent period. The axial filament then expanded within the cell to form a complex reticulum which, in the case of lyso- genization, separated into discrete normal nuclear elements. The resulting cells were indistinguishable from those of the sensi- tive strain. The same stages were observed when a virulent mutant of PI was used, but lysis occurred while the chromatin was in the stage of expansion. In P2 infections, on the other hand, the axial filament of chromatin persisted up to lysis or, in the event of lysogenization, the chromatinic filament showed ex- pansion and separation into discrete normal elements, starting from the poles of the growing elongating cells. The first normal- appearing cells were nipped off" from the terminal portion of the cell and division was not equatorial. 202 BACTERIOPHAGES As an interesting comparison with the previous studies, Dela- porte (1952) described the cytological changes in an induced colicinogenic strain of £■. coli. Strain ML after ukraviolet induc- tion was incubated at 37 ° C. and sampled at intervals. Intra- cellular colicin is detectable within 15 minutes, and appears in the medium after 75 minutes, increasing in amount until about 2 hours after irradiation. During this tim.e the bacteria swell and gradually fill up with Feulgen positive chromatin which seems to have no definite arrangement in the cells. The lysis which liberates the colicin is accompanied by a gradual de- crease in cell size and amount of chromatin per cell. However, it should be mentioned that Kellenberger and Kellenberger (1956) published an electron microscope study of lysates of cohcinogenic strains (including ML). They were unable to identify a colicin particle, possibly because the size of these would be of the same order as the many granules liberated from normal bacteria. In any case, the colicins m.ay not correspond in structure to the phages and there are not enough studies to al- low a clear statement of cytological events. 3. Electron Microscope Observations There is a very extensive literature dealing with electron micro- scope studies of phage- infected bacteria. The pertinent refer- ences have been listed by Beutner, Hartman, Mudd, and Hillier (1953). There are a number of defects which severely limit the value of much of this work. The ability to distinguish phage par- ticles and other minute structures within the relatively large, whole bacteria is limited by electron scattering and by lack of contrast, and the situation is made worse by shadowing. It is un- likely that studies of whole bacteria by electron microscopy can be very rewarding; however, various means of circumventing the difficulty have been devised and these will be discussed. The techniques of studying biological materials with the electron microscope are still in embryonic state. One approach is breaking open the bacteria at intervals dur- ing the latent period and studying the liberated material. The CYTOLOGICAL CHANGES IN THE INFECTED HOST CELL 203 rapid decompression technique of Fraser (1951b) was used by Levinthal and Fisher (1952, 1953) (Chapler XI). Phage-in- fected bacteria are unusually fragile and may be easily ruptured (Levinthal and Visconti, 1953). This undoubtedly accounts for the claim by Wyckoff (1949a) that phage-infected bacteria lyse, in the sense of cell rupture, within 5 minutes after infection. Wyckoff concluded "that the membranes of young bacteria rup- ture soon after contact with phage which then proliferates in and at the expense of the bacterial protoplasm by a process that re- sembles the division shown by bacteria and other microorgan- isms." This conclusion is at variance with all other evidence about phage multiplication. Useful information can be obtained from experiments with iso- lated host cell components, as was the case in Weidel's (1951) study of the effect of the T-even phages upon isolated cell mem- branes of E. coli B (see also Chapter XI). He showed that a local damage to the cell wall occurred after adsorption had taken place. The local loosening of the wall fabric was recognizable as visibly altered patches of the wall and, if a sufficient number of particles were adsorbed, could result in complete destruction. His description of the phases of destruction is very reminiscent of the previously noted descriptions of "lysis from without." Another method for studying whole phage-infected bacteria in the electron microscope employs a high contrast lens system, which permits some differentiation of structure to be observed even in uninfected cells (Mudd, Hillier, Beutner, and Hartman, 1953). In cultures of E. coli typical cells have three opaque areas, at the two poles and in the center of the cell. These opaque, cytoplasmic regions are separated by areas of lesser density corresponding to the two nuclear bodies as seen in stained cells and to the "vacuoles" seen in sectioned bacteria by Birch- Andersen, Maal0e, and Sjostrand (1953). Some minutes after infection with phage T2, the less dense areas move to the margins of the cells, much as does the chromatin seen in stained prepara- tions. For the first half of the latent period there is no evidence of any intracellular objects resembling mature phage. During 204 BACTERIOPHAGES the second half of the latent period there is a decrease in the opacity of the bacteria and they become markedly granular. Mature phage particles appear within the bacterial cell, more numerous as time goes on. Also one may see objects, slightly larger and less opaque than mature phage, which are probably homologous to the "doughnuts" described by Levinthal and Fisher (1952). At a still later time the infected bacteria are disrupted and exhibit a ruptured membrane, a dense pile of phage particles, and relatively little other bacterial substance. Prominent features of the cells throughout this course of events are large, opaque bodies, several hundred m^ in diameter, which are referred to by the authors as bacterial mitochondria. The properties of these large granules were reported by Hartman, Mudd, Hillier, and Beutner (1953), When triphenyltetrazolium chloride was added to a growing culture of E. coli, it was reduced to formazan which stained these granules red. They remained stainable and unchanged in location or size throughout the pe- riod of phage multiplication, even while the bacterial nuclei were disintegrating and the bacterial cells were filling up with phage particles. Quantitative studies indicated that the tetrazolium- reducing activity of an infected bacterial culture remained con- stant throughout the latent period and only decreased when lysis began. Even after the bacterial cells had lysed, the red- stained granules were readily visible in the microscope. These granules are, according to these workers, functionally analogous to mitochondria in the cells of higher organisms, and much cir- cumstantial evidence suggests that they play an essential role in phage multiplication. A technique likely to have extensive and useful application in the future is the sectioning of fixed bacteria embedded inmethac- rylate (Birch-Andersen, Maal0e and Sjostrand, 1953; Kellen- berger, Ryter, and Schwab, 1956). Preliminary results with T4 and E. coli B indicate that this may become a powerful tool for following the course of intracellular events in infected bac- teria (Maal0e, Birch-Andersen, and Sjostrand, 1954). Log phase cells of E. coli contain two or more large vacuoles, each CYTOLOGICAL CHANGES IN THE INFECTED HOST CELL 205 vacuole in turn containing a dense central object which is inter- preted to be a coil of tightly wound threads. The vacuoles oc- cupy positions in the cell corresponding to the nuclear bodies seen in stained cells. Cells fixed 4 to 8 minutes after infection showed considerable rearrangement. The nuclear areas, in particular, were replaced by similar foci (containing elongated dark bodies) at the periphery of the cells. The uncertainties of fixation and damage during polymerization of the embedding plastic must be considered as major difficulties to be solved be- fore really useful preparations and interpretations can be ex- pected (see Maal0e and Birch-Andersen, 1956). A defect in the interpretation of much of the work with the electron microscope is the tendency of the technologist to arrange his photographs in what seems to him to be a logical sequence. Even when samples for examination are taken serially during the latent period, it may be impossible to arrange them in true tem- poral order. This is in part due to difficulties in starting and stopping the infectious process simultaneously in all cells and in part because, even when trouble is taken to synchronize phage development in the population of infected bacteria, the develop- mental process soon gets out of phase in the individual cells (Benzer, 1952). However, probably none of the difficulties is insuperable, and current technical developments lead one to hope that micrographic work will contribute greatly toward an understanding of viral growth. 4. Summary Phage multiplication is accompanied by major changes in the cytology of the host bacteria. Bacterial multiplication stops, in some cases immediately after phage adsorption but in other sys- tems only after one or two cell divisions. In most systems there is an enlargement of the bacterial cells during the latent period. The release of phage which terminates the latent period is usually accompanied by an abrupt disintegration of the cells. In a few cases where the bacterial cell walls seem to be unusually sturdy, lysis involves a slow collapse of the cell with leakage of the cyto- 206 BACTERIOPHAGES plasm through a hole in the wall. Phage multiplication in- variably involves marked changes in the appearance of the bac- terial nucleus and, usually, a marked decrease in the stainability of the cytoplasm; however, the sequence of changes varies greatly from one phage to another. The cytological changes ob- served are characteristic of the phage rather than of the host cell, are under genetic control, and are of taxonomic significance be- cause phylogenetically related phages produce similar cytological patterns. A careful comparison of the appearance of living cells by dark field or phase contrast microscopy with the appearance of fixed and dried cells by electron microscopy or in stained preparations suggests that the procedures of fixing and drying are likely to re- sult in the production of artefacts. This places limitations on the interpretation of some of the cytological evidence that have not always been evident to workers in this field. Any interpr-e- tation should take into consideration the large amount of infor- mation about the mechanisms of phage multiplication which has been obtained by other means. Up to the present the nuclear staining techniques have been the most productive of new information about the response of the host cell to invasion by bacteriophage. However, it is probable that recent improvements in electron microscope technology will permit studies of the phage-infected bacterium at an order of resolution approaching that of biochemical techniques. As is true of most recent phage research, a tremendous amount of effort has been devoted to the cytological examination of a very few phage strains. It would be desirable to apply the techniques now available to a broader range of phages and host strains and conditions. CHAPTER XIII ISOTOPIG STUDIES ON THE FATE OF THE INFECTING PHAGE PARTIGLES The first use of the expression, "fate of the infecting virus par- ticle," occurs in the title of a paper by Putnam and Kozloff (1950). The expression is usually understood in the sense of the fate of the chemical constituents of which the infecting virus particle is composed. However it could with equal justification be used to mean the fate of the genetic factors with which the in- fecting virus particle is endowed. In the present chapter we will be concerned primarily with the fate of the material substance. The principal experimental approach to this problem calls for the use of isotopes, because it is only by the use of such specific labels that one is able to distinguish phage material from host material. By using appropriate differential labels it is possible to trace phage protein and phage nucleic acid during the infec- tious process. The bacteriophage particles are labeled by grow- ing the host bacteria in a nutrient medium containing the ap- propriate labeled ingredient, infecting with phage, and per- mitting lysis to take place in the labeled medium. The phage is then isolated and purified by differential centrifugation until the isotope content becomes constant. Putnam and Kozloff (1950) found no evidence that extracellular phage can exchange isotope labeled compounds with the environment. This is in agreement with the generally accepted principle that bacterio- phages have no metabolic activity of their own. The isotope composition of a labeled phage preparation which has been prop- erly purified remains constant until infection, except for changes due to radioactive decay. 207 208 BACTERIOPHAGES 1. Morphological, Functional, and Chemical Differentiation of the Phage Particle A most important discovery relative to the fate of the infecting phage particle was reported by Hershey and Chase (1952) and further elaborated by Hershey (1955). These experiments have been described in detail in Chapter XI. The essential features together with other relevant details will be briefly summarized here. Most of the information has been obtained with the T2 serological group and with T5, and it may be unwise to ex- trapolate at present to other kinds of phage. a. Chemical Composition Phage T2 consists almost exclusively of protein and deoxy- pentose nucleic acid (DNA). Few attempts have been made as yet to fractionate the proteins of phage but serological evidence suggests that at least two different proteins are present, one in the phage head and a second in the phage tail. The protein con- tains a considerable amount of methionine and so can be con- veniently labeled with S^^. The DNA may be labeled with P^- and both protein and DNA can be labeled with C and N isotopes. h. Physical Fractionation — Osmotic Shock Phage T2 consists of a protein membrane and a DNA core. By means of osmotic shock followed by high speed centrifugation it is possible to separate the sedimentable membranes from the DNA in the supernatant fluid. Not more than 5 per cent of the total phage protein remains with the DNA fraction. Even this protein has the same amino acid composition as the sedi- ment. About 3 per cent of the total sulfur-containing protein of the phage may enter the host cell with the phage DNA. The function of this protein is not known. This does not leave much room for a hypothetical, sulfur-free, basic protein which might be in intimate association with the liberated DNA. If such a protein exists it must amount to less than 5 per cent by weight of the phage DNA (Hershey, 1953c). Therefore phage T2 must FATE OF INFECTING PHAGE PARTICLES 209 consist almost exclusively of a sulfur-containing protein mem- brane and a DNA core, which can be separated from each other by osmotic shock. The membrane is undamaged by this pro- cedure as far as can be observed in the electron microscope and retains important physiological properties (Chapter V). c. Physiological Fractionation — the Blendor Experiment Phage T2, difTerentially labeled with S^'^ and P^', is adsorbed to the host cells in a salt medium which insures a high efficiency of attachment. The infected bacteria are freed of unadsorbed isotopic material by centrifugation and resuspension in an appro- priate diluent. The infected bacteria are now vigorously agi- tated in a Waring Blendor, centrifuged, and both bacteria and supernatant fluid assayed for S^^ and P^^. The stirring in the Blendor has essentially no effect on the plaque-forming potential of the infected bacteria. After this treatment about 80 per cent of the phage S^^ is found in the supernatant in the form of protein which is sedimentable after two hours at 12,000 X g and is precipitable by anti-T2 serum. From 20 to 35 per cent of the phage P^2 is left in the supernatant, the rest being sedimented with the bacteria. This means that 80 per cent of the phage pro- tein can be separated from 65 to 80 per cent of the phage DNA phosphorus by the physiological procedure of infection of the host cells. The Blendor serves to tear the phage membranes loose from the bacterial surface, but the separation of phage DNA from phage protein has already occurred as a stage of infection. d. Functional Differentiation of Phage Protein and Phage DNA These experiments demonstrate that at least 80 per cent of the sulfur-containing proteins of the phage remain on the surface of the infected bacteria, from which they can be detached by agita- tion in the Blendor. Because this treatment has no efTect on the course of the infection, it may be concluded that the bulk of the phage protein is expendable after infection and plays no role in the intracellular multiplication of phage. Similar conclusions 210 BACTERIOPHAGES were reached in the case of T5-infectecl bacteria by Y. T. Lanni (1954) who found that the complement-fixing antigens of the in- fecting phage could be removed from the bacterial surface by- treatment in the Blendor. Conversely 65 to 80 per cent of phage P^^ [^ j^ot removable by treatment in the Blendor, presumably because the phage DNA has penetrated into the interior of the bacterial cell where it is in- volved in intracellular phage multiplication. The general con- clusion from these experiments is that following adsorption the protein membrane of the phage serves as a syringe to "inject" the phage DNA into the host bacterium. It may be assumed, but it has not yet been proved conclusively, that nucleic acid is the sole agent of genetic continuity during multiplication of the phage (Hershey, 1956a) and that the phage protein, after injection of the DNA has been completed, plays no further role in phage replication. If these assumptions are cor- rect it greatly simplifies the problem of the fate of the infecting phage particle. The phage protein after infection is discarded, and so the fate of the phage DNA is all that remains to be con- sidered. Remembering that these are tentative assumptions rather than proved conclusions we can see what other evidence can be brought to bear on this problem. 2. The Breakdown of the Infecting Virus Particle We have seen that the infection of the host cell results in a physiological fractionation of the phage particle into an external protein portion and an internal nucleic acid portion. We may now consider whether these two major segments remain intact during infection, or whether there is a chemical degradation of either portion into smaller fragments. The evidence in the case of the phage protein seems to be more clear cut and will be presented first. a. Absence of Breakdown of Phage Proteins In the experiments of Hershey and Chase (1952) with S^'^- labeled phage T2, about 80 per cent of the phage methionine FATE OF INFECTING PHAGE PARTICLES 211 could be liberated from the infected bacteria by violent agitation, together with about 80 per cent of the carbon of other amino acids (Hershey, 1955). All of the liberated S^^ is sedimentable at high speed and specifically precipitable by antiphage serum, which in- dicates that the fragments must be more or less intact phage mem- branes. Examination of the liberated phage membranes in the electron microscope shows them to have damaged tails, suggesting that part of the missing 20 per cent of phage S'^ may consist of phage tail tips which remain fixed to the bacterial surface when the remainder of the phage is torn away in the Blendor. Much of the residual S^^ remains attached to the bacterial debris after bacterial lysis and when the final phage lysate is analyzed only about 2 per cent of the initial phage S^^ is found in the superna- tant after centrifugation at 12,000 X g for 30 minutes (Hershey and Chase, 1952). These experiments indicate that essentially none of the proteins of the infecting T2 phage are converted into low molecular weight fragments during the infectious process. This is in agreement with the basic assumption that once the in- jection of phage DNA into the host cell has been completed the phage protein plays no further role in the infectious process, but remains physiologically inert, outside of the bacterial cell wall. Experiments by French (1954) are in essential agreement with these conclusions. French prepared stocks of phage T2 which were labeled with 2-G^^ lysine. He was able to demonstrate by fractionation procedures that in the purified phage more than 90 per cent of the C^'^ label was in the lysine of the phage protein. This labeled phage was then used to infect host bacteria at a multiplicity of 10 and after lysis the lysate was fractionated by centrifugation and the distribution of the C^^ label was deter- mined. The bacterial debris, sedimented at 5000 X g, con- tained 83 per cent of the C^^, and the high speed supernatant con- tained 1 1 per cent. Unfortunately this nonsedimentable mate- rial was not further investigated so it is not known whether it con- sisted of phage protein or of smaller brciikdown products. Only 1.2 per cent of the label was found in the final yield of purified phage. One may conclude from these experiments that there is 212 BACTERIOPHAGES little or no degradation of the protein of infecting phage particles during the infectious process. In contrast to these conclusions are those of Kozloff (1952a,b) which are based on experiments with N^Mabeled T2 phage. Kozloff concluded that "up to 80 per cent of the parent virus DNA and protein was extensively broken down to a variety of small fragments during virus reproduction." Actually these papers give no experimental evidence that the virus protein was broken down. b. Breakdown of Phage Nucleic Acid Following Superinfection The work of Hershey and Chase (1952) demonstrated that shortly after phage adsorption the bulk of the phage nucleic acid enters the bacterial cell. There seems to be general agreement that some of this nucleic acid is broken down to very small frag- ments during intracellular phage growth. The first indication of the extent of this degradation is in a paper of Putnam and Kozloff (1950). Purified labeled T6 phage was used to infect bacteria at a multiplicity of 3, and after bacterial lysis the lysate vvas fractionated to determine the distribution of the P^- label. About 10 per cent of the P^^ was found in the bacterial debris, 40 per cent in the purified phage progeny, and 50 per cent failed to sediment at 20,000 X g for 2 hours. In another experiment in which the lysate was fractionated by means of a Sharpies supercentrifuge 57 per cent of the P^^ was found in the effluent. On further fractionation of the effluent, 26 per cent of the initial phage P^^ was recovered in acid-soluble form. Of this about half, or 13 per cent of the initial phage P^^, was inorganic phosphate. Essentially the same distribution of P^- was found whether the multiplicity of infection with labeled phage was 1 or 3. The next important development was the investigation of the kinetics of release from infected bacteria of trichloroacetic acid- soluble phosphorus derived from the infecting phage. This work led to the discovery that a considerable part of the P*- of phage T2 may be liberated within a few minutes of the time of FATE OF INFECTING PHAGE PARTICLES 213 adsorption, and that the amount liberated depended on the multiplicity and timing of the infection (Lesley, French, and Graham, 1950). This phenomenon, which is called "superin- fection breakdown," has been studied further with results sum- marized by Graham (1953). The essential features of the phenomenon may be summarized as follows. 7. At a multiplicity of 0.2 T2r+ phage particles per bacterium, about 5 per cent of the phage P^^ is liberated within 10 minutes of adsorption. No more P^^ is liberated until bacterial lysis when readsorption of the first phage liberated occuis. During the ensuing period of lysis inhibition about 20 per cent more of the phage P^2 is liberated. 2. At multiplicities of 3 to 10, about 15 per cent of the P^^ is liberated in the first 10 minutes, followed by a slow liberation of P'^2 starting at 30 minutes. 3. If bacteria are infected with unlabeled phage, followed 5 minutes later by labeled phage, about 50 per cent of the P^^ of the superinfecting phage is liberated within the next 10 minutes. This phenomenon of superinfection breakdown is in part responsible for the small amount of P^^ released after primary infection at low multiplicity. 4. If the time interval between primary infection and super- infection is varied, it is found that the amount of P'^^ liberated is 30 per cent after a 1 minute interval and nearly maximal after a 2 minute interval. The physiological condition responsible for superinfection breakdown is established within 2 minutes after primary infection. 5. Primary infection with one phage particle per cell is enough to establish the conditions for superinfection breakdown and in- creasing the multiplicity to 10 has no further effect. 6. The multiplicity of the superinfecting phage is also without effect. 7. The phenomenon can be demonstrated repeatedly with the same batch of infected bacteria under conditions of lysis inhibi- tion. Superinfection with a multiplicity of 3 labeled phage at 5 minutes, 23 minutes, and 43 minutes after primary infection re- 214 BACTERIOPHAGES suited in liberation of 56 per cent, 53 per cent, and 74 per cent, respectively, of the added P'''-. 8. Superinfection breakdown of 'r2/'+ phage occurs after pri- mary infection with T2, T4, and T6 phage, r or r+, and with T5 but does not occur after primary infection with Tl, T3, or T7. Superinfection with r phages follows the same pattern as with r+ phages but the amount of breakdown is less (30 per cent). Primary infection with T3 followed by superinfection with T2r+, followed still later by superinfection with labeled T2r+ does not lead to superinfection breakdown. Primary infection with T2 or T7 does not result in breakdown of superinfecting T7. 9. Superinfection breakdown does not occur in heat-killed bacteria but does occur in bacteria killed by ultraviolet light. Superinfection breakdown is prevented if cyanide is added to the bacteria before the primary infection, but is not prevented if cyanide is added 5 minutes after primary infection and before superinfection. One may recall that in cyanide-poisoned bac- teria the development of phage T2 is arrested at an early stage (Benzer and Jacob, 1953). Streptomycin added either before or after primary infection prevents superinfection breakdown, even in streptomycin-resistant bacteria in which phage develop- ment is normal. Deoxyribonuclease isolated from E. coli is in- hibited by streptomycin. 10. Heat-killed phage T2 is inactive both in primary infection and superinfection, whereas ultraviolet-inactivated T2 performs either role in superinfection breakdown. 7 7. In all experiments recorded above the bacteria were grown in tryptose broth. KozlofF (1952a) records a personal commu- nication from Graham that superinfection breakdown does not occur with bacteria which have been grown in ammonium lac- tate medium. It likewise does not occur when magnesium is de- ficient. Hershey, Garen, Fraser, and Hudis (1954) report that reduction of the magnesium level in their cultures to 10~° M blocks superinfection breakdown, presumably by inhibiting bac- terial deoxyribonuclease. They also state that the limitation of breakdown at about 50 per cent is due to incomplete injection by superinfecting phage. FATE OF INFECTING PHAGE PARTICLES 215 As discussed in Chapter XVII, superinfection breakdown is probably related to but does not cause genetic exclusion of superinfecting phage. Similarly, it seems clear that superin- fection breakdown results from the action of bacterial deoxyri- bonuclease, but it is not likely that it depends simply on the activation of this enzyme after infection, because T3 activates enzyme but does not activate the breakdown mechanism. Pardee and Williams (1952, 1953) found that the activity of bacterial deoxyribonuclease increases after infection with T2 and T3. KozlofF (1953) showed that the effect is due to the destruc- tion of a specific inhibitor of the enzyme, probably part of the bacterial ribonucleic acid. The role of the enzyme in superinfection breakdown is indi- cated by the fact that streptomycin and low magnesium concen- trations interfere with breakdown and with the action of the enzyme (Graham, 1953). Superinfecting phage is treated differently from the primary infecting phage in three ways. The superinfecting phage, only, is subject to a powerful genetic exclusion, to partial interference with injection, and to complete breakdown of the injected DNA. Inhibition of deoxyribonuclease prevents breakdown but does not affect exclusion or injection. The exclusion may mean that injection is not only incomplete but also abnormal in other, un- known, ways. The fact that the breakdown products do not re- main in the cells also suggests this conclusion. If so breakdown, though requiring active deoxyribonuclease, probably depends primarily on the hypothetical, abnormal kind of injection. In any case the breakdown as such does not influence the outcome of the infection in any discernible way (Graham, 1953; Hershey, Garen, Fraser, and Hudis, 1954). Also following primary infection, a small amount of the phage DNA, not exceeding 5 per cent under optimal conditions, is broken down to low molecular weight material by the intrabac- terial deoxyribonuclease. This may reflect some side reaction to which only a few of the phage particles are subject, rather than a necessary consequence of the infectious process itself. Further 216 BACTERIOPHAGES investigation might show that the breakdown of both primary and secondary infecting phage is the resuk of a similar failure at some early step in the infectious process, differing only with respect to the proportion of particles involved. In general, some particles are excluded from participation in viral growth by failure to in- ject, and some by subsequent unspecified failures, under all con- ditions (Chapter XVII). Genetic exclusion of superinfecting phage is thus the consequence of two distinct but probably re- lated kinds of physical exclusion, one measurable by the Blendor experiment, the other by the breakdown effect. Both may be re- garded as consequences of a pathological exaggeration of a nor- mally inefficient excluding mechanism. 3. Material Transfer from Infecting Phage to Progeny An interesting aspect of the fate of the infecting phage particle is the transfer of its chemical substance to the phage progeny. One may propose such questions as: Does the DNA of the in- fecting phage particle or a major portion of it appear as a unit in the phage progeny? Is the entire phage particle expendable once it has served its function as a pattern for the synthesis of new phage? Answers to these questions have been sought by the use of isotopic labels with a detailed analysis of the distribu- tion of the isotope in the various fractions of the lysate. The most interesting information has been derived from experiments in which the infecting phage has been labeled with different iso- topes so that differential transfer of the various parts of the in- fecting phage can be detected. This type of experiment has already yielded much information which in relation to genetic experiments has suggested mechanisms of phage replication. a. Lack of Transfer of Phage Protein to Progeny As discussed previously, the protein membranes of the infecting phage particles do not participate in the intracellular replication of phage. Once penetration of the phage nucleic acid into the host cell is accomplished, the membranes can be detached from FATE OF INFECTING PHAGE PARTICLES 217 the infected bacteria by vigorous shaking without affecting phage development. These facts in themselves imply that little or no transfer of phage protein from parent to progeny should be expected. However, prior to this discovery there were several papers which indicated that phage protein was indeed transferred to progeny. It is worth while to discuss these experi- ments briefly as an indication of the type of technical error which can result in false conclusions. In a brief report Hershey, Roe- sel, Chase, and Forman (1951) stated that about 35 per cent of the sulfur of infecting T2 phage was found in the phage progeny, and that when this phage progeny was used in a second cycle of infection, 40 per cent of its sulfur was found in the second cycle progeny. Since only the phage protein is labeled in sulfur, these experiments were interpreted as transfer of parental pro- tein to progeny. The transfer of about 18 per cent of parental T6 phage N'^ to progeny was reported by Kozloff (1952b). The apparent transfer of protein N was not equal to the transfer of nucleic acid N, and the ratio of the two was not constant from one experiment to another, varying from 0.4 to 0.9. Kozloff (1952c) concluded that nucleoprotein was not transferred from parent to progeny as a unit, but that there was extensive rear- rangement of the contributed parental material which suggested the use of breakdown products of the infecting virus for synthesis of protein and DNA of the progeny virus. The earlier reports of the transfer of parental protein to progeny were demonstrated by Hershey and Chase (1952) to be the result of unsuspected contamination of progeny phage with protein remnants of the infecting phage. If, after infection with sulfur-labeled phage, the bacteria were agitated in the Waring Blendor, centrifuged to eliminate the dislodged membranes, and then permitted to lyse in a fresh medium, the isolated phage progeny was found to contain only one per cent of parental sul- fur. Therefore there is no obligatory transfer of sulfur-contain- ing proteins from parent to offspring. Under the usual condi- tions of multiple infection there is a considerable spontaneous elution of phage membranes which cannot be distinguished from 218 BACTERIOPHAGES or separated from the phage progeny and which therefore simu- late a material transfer from parent to progeny. The treatment of the infected bacteria in the Waring Blendor followed by cen- trifugation seems to be a useful way of preventing this contami- nation. These conclusions were confirmed by Kozloff (1953) and by French (1954). Experiments by Volkin (1954a) suggested that phage T4 con- tains about 30 per cent of its total protein content in tl>e form of a nucleoprotein. Hershey (1955) failed to confirm this in care- ful studies of phage T2 labeled with either S'''' or C^^ He did find evidence for a "nonsedimentable" protein fraction in osmot- ically shocked T2 preparations, amounting to about 3 per cent of the total protein. This nonsedimentable protein was chemi- cally similar to phage ghosts but serologically distinct and ap- parently was injected into the host cell along with the phage DNA. There was no evidence that this protein contributed materials to the phage progeny. b. Transfer of Phage Nucleic Acid to Progeny In marked contrast to the situation with phage protein, there is conclusive evidence that substances from the infecting phage con- tribute to the nucleic acid of phage progeny. In experiments of Hershey and Chase (1952) with P^Mabeled phage T2 there was a transfer of 30 per cent of the parental phosphorus to the progeny phage after treatment of the infected bacteria in the Blendor. Therefore this cannot be due to residual DNA in contaminating phage membranes. The results of a number of other transfer experiments are recorded in Table XIV. The transfer of parental nucleic acid substance to progeny varies from 1 5 to 60 per cent in diff'erent experiments, the efficiency of transfer depending largely on experimental conditions. The percentage transfer is of the same order of magnitude for T2, T3, T4, T6, and T7; for P'^^^ for DNA-N^^ and for C^^-purines and pyrimidines. Much of this work has been briefly summarized by Kozloff (1953), and more recent work by Hershey and FATE OF INFECTING PHAGE PARTICLES 219 Burgi (1956). Some of the factors which affect the efficiency of transfer will now be considered. TABLE XIV Transfer of DNA from Parent to Progeny Phage Isotopic Phage marker Per cent transfer Reference T6 P32 22-42 T2 p32 15-25 T2 p32 20-40 T2 p32 35 T6 p32 44 T6 N'5 27 T2 p32 30 T7 p32 20-30 T3 p32 38-46 T4 p32 40-50 T2 C"-purine 48-55 T3 C>''-purine 32-38 T4 C'^-purine 40-44 T2 P32 40-45 T2 p32 28 Putnam and KozlofT (1 950) Lesley, French, Graham, and van Rooyen (1951) Maal0e and Watson (1951) French, Graham, Lesley, and van Rooyen (1952) KozlofT (1952b) KozlofT (1952b) Hershey and Chase (1952) Mackal and KozlofT (1954) Watson and Maal0e (1953) Watson and Maal0e (1953) Watson and Maal0e (1953) Watson and Maal0e (1953) Watson and Maal0e (1953) Hershey (1953a) French (1954) 7. Superinfection breakdown. In view of the phenomenon of superinfection breakdown described above, one might antici- pate that transfer of P^^ from superinfecting phage to progeny would fail. This point was checked by French, Graham, Lesley, and van Rooyen (1952) who infected bacteria with un- labeled T2 phage and then superinfected with P^^.i^beled T2 phage after various time intervals. The transfer of P^^ decreased from 30 per cent for simultaneous infection to 18 per cent after one minute, 7 per cent after 2 minutes, and 2 per cent after 5 minutes. This experiment indicates quite clearly that superin- fection breakdown will decrease the efficiency of transfer under conditions of multiple infection unless the adsorption period is made very short. These experiments were independently con- 220 BACTERIOPHAGES firmed with P'' '^-labeled phage T4 by Watson and Maal0e (1953). This cause of reduced efficiency of transfer can be elimi- nated by making the adsorption period less than one minute. 2. Readsorption of progeny phage. If some of the labeled prog- eny phage is lost either by readsorption to yet unlysed bacteria or to bacterial debris, this fraction of the transferred label will be lost from the progeny. This will cause an apparent decrease in the efficiency of transfer of parental materials to progeny phage. This loss of daughter phage was prevented in the experiments of Watson and Maal0e (1953) by treating the infected bacteria with antibacterial serum toward the end of the latent period. Antibacterial serum effectively prevents adsorption of phage to otherwise susceptible bacteria (Delbriick, 1945b). In earlier studies Maal0e and Watson (1951) prevented loss of progeny T2 phage by saturating the bacterial receptors with ultraviolet- killed, unlabeled T2 phage. These or similar devices have been used in most subsequent experiments. 3. Latent period and burst size. In infections with P"'--labeled T4r phage, lysis occurs at the end of the usual latent period with a burst size of about 140 and a transfer of P^^ ^q progeny of 40 to 50 per cent (Watson and Maal0e, 1 953). About half of the parental P^2 is liberated into the medium in a nonsedimentable form. A natural question is whether or not some of this wasted P^^ would be converted into mature phage if the infectious process had not been interrupted by lysis. To answer this question Watson and Maal0e (1953) infected bacteria with labeled T4r+ and then produced lysis inhibition by reinfection with unlabeled phage. Although the burst size was increased to 350 the transfer of parental P^^ was still only 50 per cent indicating that essentially no parental P^^ was transferred to phage particles formed after the end of the normal latent period. The converse experiment, shortening the latent period by pre- mature lysis, has been done several times. Maal0e and Watson (1951) using P'^--labeled T2/-+ for infection, lysed the infected culture prematurely at 1 1 minutes, before there was one mature phage per cell. With an excess of ultraviolet-inactivated T2 FATE OF INFECTING PHAGE PARTICLES 221 phage as a carrier, the lysate was fractionated in the usual way and only 1 .6 per cent of the parental P^^ was found in (he phage fraction. In a second sample lysed at 34 minutes, the burst size high speed pellet ow speed pellet ocid-soluble 10 20 30 40 50 60 70 80 90 MINUTES AFTER INFECTION Figure 8. Transfer of radiophosphorus from parental to offspring phage T2. The curves show distributions of P^^ measured by centrifugal fractionation of lysates prepared at different times after infection with 3 labeled phage particles per bacterium. The vertical lines span the observed range of variation in three independent experiments. Acid-soluble P^^ is measured from separate aliquots of unlysed culture. P^^ not accounted for is DNA nonsedimentable after lysis. The dashed line shows infective phage particles per bacterium released during lysis. These are distributed, independently of time of lysis, in the ratio 14 : 86 between low and high speed pellets, and account for the rise in both fractions after 10 minutes. The observed efficiency of transfer is therefore about 60/0.86 = 70%. The rise in acid-soluble P^- is probably the combined result of superinfection breakdown of readsorbed offspring particles and enzymatic destruction of phage-precursor DNA, both occurring after lysis of some bacteria at about 25 minutes. Experiments by A. D. Her- shey, unpublished. 222 BACTERIOPHAGES was 248 and the P" transfer was 29 per cent. This experiment served two purposes: it demonstrated that contamination of the progeny with parental P"^^, as contrasted with incorporation of the isotope in the progeny, was not a significant factor in such experiments. It also demonstrated that just prior to the appear- ance of mature phage in the infected bacteria there is no pre- cursor DNA in a sedimentable form that might simulate mature phage in its behavior. Kinetic experiments, in which the incorporation of parental phosphorus into offspring phage particles is measured at various times after infection, were reported by French, Graham, Lesley, and van Rooyen (1952), Watson and Maal0e (1953), and Hershey (1953a). The experiments of the last-named author are especially detailed. The results of subsequent experiments of the same kind, embodying minor improvements in technique, are illustrated in Figure 8. They show all the features discussed above. Shortly after infection, the parental DNA in prema- turely lysed cultures does not sediment at centrifugal speeds sufficient to throw down phage particles. It remains largely in- soluble in acid, howev^er. Beginning at the end of the eclipse period, the parental DNA is reincorporated into phage particles, most of it entering the first ones to mature. By 50 minutes after infection the maximal incorporation has been achieved, al- though phage particles continue to be formed after this time. The observ^ed transfer is about 70 per cent. c. Material Transfer Without Genetic Transfer In previous sections of this chapter we hav^e considered the fate of the substance of the infecting phage particles when these were the direct ancestors of the final phage yield. We will now con- sider the available evidence with respect to transfer of substance from phage particles which are not the parents in a genetic sense of the final phage yield. These experiments call for mixed in- fection with two phages in which matters are arranged so that only one of them is capable of multiplying. FATE OF INFECTING PHAGE PARTICLES 223 /. Material transfer between unrelated phage strains. The first ex- periment of this type was performed by Kozloff (1952b). Host bacteria were infected with unlabeled T7 at a multiplicity of 6 and after incubation for 3 minutes were infected with P^Mabeled T6 at a multiplicity of 1.5. The infected bacteria were centri- fuged to remove unadsorbed phage and then permitted to lyse. The final yields of the phages were 4.4 X 10^ T7 and 3.3 X 10^ T6 per ml. The T6 and T7 phages were liberated from dif- ferent bacteria because plating on mixed indicators gave no clear plaques, confirming the occurrence of mutual exclusion be- tween this pair of phages. The phage yield was purified in the usual way by differential centrifugation and the T6 progeny were then removed by repeated absorption of the phage mixture with B/3,4,7. Of the original P^- present in the parent T6 phage, 37 per cent was found in the T6 progeny and 4.6 per cent in the T7 phage. This experiment suggests the transfer of P^^ from T6 to T7 in mixedly infected cells, although the amount transferred is far less than in the homologous system. Unfor- tunately the incorporation of the P'^^ \^^q ^j^^ 'py phage was not further checked by determining the distribution of the isotope after adsorption of the T7 phage to B/6, for instance, or after precipitation with anti-T7 serum, so that the transfer cannot be accepted as proved by these experiments. In experiments by French, Graham, Lesley, and van Rooyen (1952) the transfer of V^"- from T2 to T7 and to Tl was tested. The experimental conditions were such that T2 was almost com- pletely excluded, the T2 yield being at the most 10 per cent of the total phage. The amount of P^- which was found in the isolated phage progeny varied from 3 to 6 per cent. However this repre- sents maximum possible transfer values to heterologous phage rather than actual transfer values, because in these experiments neither parental labeled T2 phage which failed to absorb, nor progeny T2 phage which should be relatively rich in P^^ label, were removed from the phage yield. Therefore these experi- ments do not demonstrate actual heterologous transfer of phage materials, as was realized by the authors who concluded "that 224 BACTERIOPHAGES little of the phosphorus of T2 phage is incorporated into Tl or T7 progeny." A definitive experiment involving heterologous transfer of phage label was reported by Watson and Maal0e (1953). Bac- teria were mixedly infected with 5 particles of unlabeled T4r and 1 particle of P*--labeled T3 per bacterium. Under condi- tions of simultaneous infection with this pair of phages, T3 is ex- cluded from multiplication in essentially all bacteria. In one such experiment 24 per cent of the P^^ label was in the bacterial debris, 49 per cent was not sedimentable, and 27 per cent was in the purified phage yield. Of the P"*'- in the phage yield only 4 per cent was precipitated by anti-T3 serum, whereas 92 per cent was precipitated by anti-T4 serum. The authors concluded that about 25 per cent of the P^^ label of the excluded T3 phage was incorporated into the yield of T4. This experiment seems to demonstrate quite conclusively that material transfer can take place between unrelated phages in the absence of genetic transfer and under conditions in which the donor phage does not multiply. The reasonable assumption is that the excluded donor phage was extensively degraded inside the host cell and part of its chemical substance used as raw material for multiplication of the dominant phage. In different experiments, V2 to Vs of the P'^^ of the excluded phage was liberated on host cell lysis in a form not sedimentable at 12,000 X g, further evidence that extensive degradation of the excluded phage occurs. 2. Material transfer from inactivated phages. In the experiments of KozlofT (1952b), N^Mabeled T6 was inactivated with various doses of ultraviolet light or X-rays, mixed with unlabeled active T6, and the mixture used to infect bacteria. The T6 progeny was isolated and the amount of label incorporated into progeny nucleic acid and protein determined. Between 5 and 18 per cent of the parental N^'' was found in the nucleic acid of the phage yield and Kozloff^ concluded that transfer had occurred. This interpretation, however, is weakened by the parallel finding that from 4 to 15 per cent of the parental N^^ was found in progeny protein. This is evidence for contamination of progeny FATE OF INFECTING PHAGE PARTICLES 225 phage with the protein membranes of parental phage as dis- cussed previously and raises the possibility of a similar contamina- tion with parental phage nucleic acid. Similar experimental findings were reported by French, Graham, Lesley, and van Rooyen (1952) and by Watson and Maal0e (1953) but their results are likewise ambiguous due to the possibility of contamination of progeny particles with damaged parental phage. This danger is well illustrated by the following experiment of Hershey, Garen, Fraser, and Hudis (1954). T2 was suspended in a 2 per cent peptone solution containing P^^ When the radiation damage had reduced infec- tivity to about one per cent of the initial value, the inactivated phage was tested in a Blendor experiment. The damaged phage adsorbed normally but injected less than 20 per cent of its DNA. A mixed infection was made with P^^.i^beled, /3-ray- inactivated phage and unlabeled active phage. The progeny were then examined as in Kozloff's experiment but it proved im- possible to interpret the results. A large proportion of the in- activated phage do not inject and at the time of cellular lysis be- come detached from the bacterial cells. Apparently many particles killed by ionizing radiation not only fail to inject but also make a weak attachment to bacteria. Ultraviolet radiation has less effect upon injection and here it may be possible to follow the fate of the injected damaged DNA. Hershey and Burgi (1956) studied this question. They repeated Kozloff's experiments in which bacteria are infected with ultra- violet-killed, P^Mabeled T2 and unirradiated, unlabeled T2. They confirmed Kozloff's conclusion that there is normal transfer of isotope from the irradiated particles to the issue of the mixed infection. However, by special techniques they could show that about half of the reincorporated P^^ is in noninfective par- ticles among the offspring. Apparently such particles are non- infective because of the radiation-damaged DNA they receive. If so, the results tend to support the idea of direct transfer of large pieces of DNA from parental to offspring particles — quite the contrary of conclusions suggested by the early experiments of this type. 226 BACTERIOPHAGES d. Physical Slate of Parental Isotope during Transfer The above experiments demonstrate that constituents of parental nucleic acid appear in progeny particles. They do not directly tell us, however, whether the transfer occurs by way of genetically specific macromolecules or whether the parental DNA is first degraded to low molecular weight nucleic acid precursors. KozlofT favored the latter route, largely because of his observa- tions of transfer unconnected with genetic transfer. Hershey and Burgi's experiments cited above destroy the force of this argument. Another possible way of deciding between these alternatives is to examine the physical state of the parental isotope during various stages of the latent period. This was done by Watanabe, Stent, and Schachman (1954) who broke open infected bacteria by decompression and examined the sensitivity to deoxyribonuclease and sedimentation characteristics of the intracellular isotope. At all times over 50 per cent of the P^^ was in high molecular weight form, either as fibrous DNA or as part of mature phage particles. In all their experiments some isotope was observed in low molecular weight material, partly owing to suboptimal conditions of viral growth. Their experi- ments (which had another purpose) thus failed to answer the question asked here. All observations of this type are com- patible with breakdown of parental DNA if we allow for the possibility of very rapid resynthesis into genetically specific material. Hershey, Garen, Fraser, and Hudis (1954) made sev^eral at- tempts to find evidence for possible short-lived intermediates. By growing infected bacteria in the presence of large amounts of nucleic acid precursors, they tried to influence the outcome of transfer experiments with C^Mabeled T2. Even though free bases and nucleosides compete efficiently with CO2 and glucose as a source of viral DNA carbon, they are without effect on parent to progeny transfer when added to cultures infected with labeled phage. The same amount of parental G^^ is transferred to the progeny irrespective of the amount of precursors added to the infected system. Not only is the total transfer of QV^ the FATE OF INFECTING PHAGE PARTICLES 227 the intermediates must be nucleotides or larger fragments. Kozloff (1953) summarizes early experiments in which unequal transfer of parental phosphorus and DNA nitrogen was ob- served. This result is inconsistent with all other comparable observations, summarized by Hershey and Burgi (1956). Kozloff's results remain unexplained, but it should be pointed out that the transfer experiments he describes were performed under conditions rather unfavorable to phage growth and ef- ficient transfer. It must be concluded that there is no decisive evidence for or against the possibility that some of the transfer occurs by way of small, functionally unspecific, fragments of DNA. e. Distribution of Parental Isotope among Progeny Particles All the previous work has dealt with populations of virus particles multiplying within a very large number of bacteria. The measured efficiency of transfer is an average summed over very many individual infections. By themselves, the incomplete transfer values are compatible with the assumption that the DNA of the infecting phage remains intact during replication and with a certain probability becomes infective again and reappears among the progeny. This possibility, however, is ruled out by the P^2_suicide experiments of Hershey, Kamen, Kennedy, and Gest (1951). In their experiments phage containing P^^.i^^gj^j-j DNA loses infectivity upon radioactive decay (Chapter VI). Uniformly P'^Mabeled bacteria growing in P^'-labeled medium were infected with nonradioactive parental T4 particles. Fol- lowing lysis and the release of about 100 progeny particles per bacterium, the progeny were isolated and all proved to be sub- ject to killing by radioactive decay. The failure to observe any stable progeny indicates that none of them contained the intact DNA from an unlabeled parental particle. It is clear that the phage DNA does not remain in one piece but is dispersed among progeny particles. 228 BACTERIOPHAGES The DNA is not randomly dispersed among a large fraction of the progeny particles. This was first suggested by the transfer experiments already discussed which showed that most of the parental isotope is incorporated into the earlier formed progeny. More recently Stent and Jerne (1955) confirmed this point in some remarkable experiments employing the P^^-decay principle. In their experiments nonradioactive bacteria were infected with heavily P^2_i3i3eig(^ particles of phage T2, and a progeny of early formed particles secured. These particles proved to be stable insofar as repeated plaque-counts could determine, although they contained large amounts of P^^. Stent and Jerne realized that this must reflect an extreme concentration of P^- in very few particles, so few that their loss by suicide could not be detected by reductions in plaque count. To prove their inference, they performed the following experiment. Unlabeled bacteria were infected with a highly radioactive parental generation of phage particles as before, to obtain a. first generation of particles that was likewise radioactive, but contained only P^^ derived from its parents. As a test of viability of the labeled first generation particles, they measured the transfer of P^^ from particles of the first generation to those of a second. They found about 50 per cent transfer as expected. However, if a sample of the first generation progeny was tested again, a few days later, the eflficiency of transfer to a second generation had markedly decreased. Thus the first generation progeny must contain much of its P^^ concentrated in a few particles that are subject to radioactive decay. Other particles also contain P^^^ but are not subject to decay at appreciable rate. Stent and Jerne concluded that part of the transfer of DNA from parental to off'spring particles preserves large pieces of parental DNA among the offspring particles. Levinthal (1956) developed an autoradiographic method for determining the amount of P*^ in individual phage particles. This furnishes a powerful technique for investigating the distribu- tion of transferred P^^ among offspring phage particles. By it he shows that the phage DNA is indeed transmitted to off'spring FATE OF INFECTING PHAGE PARTICLES 229 particles partly in the form of large pieces each containing about 20 per cent of the DNA from a single parental particle. The possible implications of these and other experiments in progress make exciting reading and are discussed by Levinthal (1956), Hershey and Burgi (1956), and Delbruck and Stent (1957) . The experimental results and interpretations are still too fluid, however, to discuss here in detail. /. Efficiency of Transfer As previously described, the efficiency of transfer of labeled parental DNA to offspring seldom exceeds 50 per cent even under the best attainable experimental conditions. This fact occasionally prom^pted speculation that the limited transfer signified something fundamental about replication of DNA. One such speculation was testable. Maal0e and Watson (1951) performed the first two-cycle transfer experiment, which had been suggested by S. S. Cohen. They found that the efficiency of transfer during first and second cycles of growth from P^Mabeled parents was the same. This result disposed of the possibility that phage particles contain two kinds of DNA, one transferable and one not. Hershey and Burgi (1956) present evidence that the incomplete transfer should be attributed to random losses that have nothing to do with the mechanism of transfer (Figure 8). Other developments support their contention to this extent: it now seems clear that an explanation of the low eflficiency of transfer will not contribute much to an understanding of the mechanism of transfer. The experiments of Levinthal (1956), Hershey and Burgi (1956), and Stent, Jerne, and Sato (1957) revive the idea that the DNA of T2 is composed of functionally distinct parts. The merits of this idea are still debatable. However, it is clear that such parts, if they exist, are transmitted from parents to progeny with similar efficiency and probably without intercon version. 4. Summary Here we will summarize the contributions which the transfer experiments give to our knowledge of phage reproduction. Al- 230 BACTERIOPHAGES most all these experiments have been done with the T series of coli phages, and most of them with T2 and its relatives. First of all the transfer experiments confirm the differentiation of the phage particle into a DNA core and a protein membrane. The components are structurally different and functionally distinct. Upon adsorption to a host cell, most of the protein membrane remains on the bacterial surface while most if not all the nucleic acid core penetrates to the interior of the bacterium. Accordingly little phage protein, unlike the injected DNA, is incorporated into progeny particles. The Blendor and transfer experiments thus establish the physical basis for the disappearance of the in- fective virus particle upon infection. In contrast to the rather static role of phage protein, the nucleic acid is functionally and materially active. The Blendor experiment of Hershey and Chase suggested a primary genetic role for phage DNA and a major aim of present day transfer ex- periments is to establish a connection between genetic function and material behavior of DNA. Early experiments showed that approximately 50 per cent of the parental DNA reappeared within progeny particles but they left open the question whether the phage DNA remains intact during replication or whether it is dispersed and fragmented. After much early confusion and ambiguity this problem now seems capable of experimental reso- lution. The entire DNA content of the phage particle certainly does not remain intact during replication. This was shown by the initial P^^-suicide experiments. It was not apparent, how- ever, whether this dispersal of DNA involved fragmentation of a single molecule, the distribution of several intact molecules among several progeny particles, or possibly both processes. This question can now be attacked by methods capable of analyzing the isotopic composition of single progeny particles. Independently, Levinthal and Stent and Jerne have achieved this objective, the former by the development of an elegant auto- radiographic technique, the latter by subtle manipulation of the P^2-suicide experiment. Both agree that the parental isotope is not randomly distributed among progeny particles. Approxi- FATE OF INFECTING PHAGE PARTICLES 231 mately half of it is transmitted as pieces containing perhaps 20 per cent each of the DNA in one particle, the remainder in con- siderably smaller pieces. The transfer as large and small pieces without apparent inter- conversion suggests to Levinthal a functional division into two kinds of DNA which, however, must be transmitted with equal efficiency. The large pieces, at least, may have genetic function since they seem, in the experiments of Hershey and Burgi, to be transmitted in association with radiation damage and authentic genetic markers. At present, however, these ideas are rather precariously based on preliminary results that are by no means free of inconsistencies. An unexpected finding of the transfer experiments is the phe- nomenon of superinfection breakdown. Shortly after infection with T2, the receptivity of the cell to virus is changed in several important ways. Superinfecting T2 injects only half its DNA, and this half is quickly broken down to low molecular weight material and excreted into the culture medium. If the cell is superinfected with some other phages no breakdown occurs, but the superinfecting phage particles are nevertheless excluded from participation in growth. It is clear that the cellular deoxyribo- nuclease brings about the breakdown of superinfecting T2, and that an intracellular inhibitor of this enzyme is partly destroyed shortly after infection. The fact remains, however, that the DNA of the primary infecting phage is subject to little or no breakdown at the time of infection, and remains immune afterwards. It must be protected either by a change in state, or by a favored location in the cell that the DNA of superinfecting phage fails to reach. The second alternative is perhaps favored since injection by super- infecting T2 is clearly abnormal, and may fail entirely for other phages whose DNA is not subject to breakdown. In any case it is clear that the breakdown by deoxyribonuclease is not the primary cause of genetic exclusion of superinfecting phages. CHAPTER XIV NUTRITIONAL AND METABOLIC REQUIREMENTS FOR PHAGE PRODUCTION The nutritional requirements for phage production are con- siderably more complex than are, for instance, the requirements for bacterial multiplication. How complex they seem depends on our frame of reference and on the operational methods used. The nutritional requirements for phage reproduction will be quite different whether one considers the intracellular vegetative phage, the infected bacterium, or the mixture of free phage and uninfected bacteria to be the reproducing unit. In the latter case one has to consider requirements for adsorption as well as sources of food. If the infected bacterium is the unit under con- sideration, the nutritional requirements are in general the same as those of the host cell alone in the same environment be- cause the infected cell is synthesizing chemically similar proto- plasm using the same enzymes as are used by the uninfected cell. A few exceptions to this generalization will be noted below. If the intracellular phage particle is considered as the reproducing unit, the nutritional environment must include all the chemical elements, amino acids, and nucleotides that go to make up the particle because the phage particle is incompetent to synthesize these substances for itself. If one uses isotopic labels to determine the ultimate source of nutrients, further refinement is possible. One can, for instance, allocate the source of some of the phage phosphorus to the infecting phage particle, some to the bacterial host before infection, and some to the extracellular environment after infection. These various aspects of phage nutrition will be considered in what is hoped will be a logical fashion. Certain points of view with respect to phage nutrition have been pre- 233 234 BACTERIOPHAGES sented in reviews by Cohen (1949, 1953b, 1956) and by Putnam (1952) and the topic has been briefly considered in a number of general reviews of phage work such as that by Benzer, Delbriick, Dulbecco, Hudson, Stent, Watson, Weidel, Weigle, and Woll- man (1950) and one by Putnam (1953). The role of the enzy- matic constitution of the host cell as a factor in the nutritional environment of the reproducing phage has been discussed by Cohen (1952) and by E. A. Evans, Jr. (1954). 1. Requirements for Adsorption and Penetration The role of the ionic environment in phage adsorption has been discussed extensively in Chapter X and will be briefly summa- rized here. Adsorption of phage to host cell involves at least two separate steps, the first step being a reversible attachment of phage particle to host cell and the second step being irreversible. The second step involves liberation of phage DNA from the pro- tein membrane and under appropriate conditions this DNA penetrates into the host cell interior. For a particular phage- host cell system there is an optimum salt concentration at which the rate of attachment is maximal. The optimum salt concentra- tion for monovalent cations diff'ers from that for divalent cat- ions in instances where either is effective. The cationic environ- ment also controls the rate of the second step and for some phages the cationic requirement is rather specific. A surprisingly large proportion of the phages which have been examined have a more or less specific requirement for calcium ion. In many cases this requirement is known to involve a step subsequent to adsorption and in the case of phage T5 the step is definitely penetration of the phage DNA into the host cell. In addition to the ionic re- quirements just mentioned, there is one case of a relatively specific requirement for an organic cofactor for adsorption, the tryptophan-requiring strains of phage T4 which have been dis- cussed in Chapter X. It would be surprising if this were a unique case. The very extensive literature dealing with the ionic cnxiron- ment required for adsorption or penetration would make pretty REQUIREMENTS FOR PHAGE PRODUCTION 235 TABLE XV Salt ElTccts or Requirements Phage Comment Reference Megaterium 899 Calcium requirement Megaterium Megaterium 899 Diphtheria, B and Pasteurella Streptococcus Streptococcus Streptococcus Streptococcus Streptococcus Streptomyces Staphylococcus Staphylococcus Staphylococcus Staphylococcus Coli Lisbonne Coli Lisbonne Shiga phage Coli phage Coli phages Coli phages Coli phage SI 3 Typhoid phage Enterophages Staphylococcus Coli phages Various phages Various phages Loss of calcium require- ment Calcium after adsorption Calcium for adsorption Calcium requirement Citrate inhibition Calcium after adsorption Calcium required Calcium required Calcium for penetration Citrate inhibition Citrate inhibition Citrate inhibition Calcium after adsorption Ca or Mg for adsorption Oxalate inhibition Calcium for adsorption Citrate inhibition Calcium after adsorption Citrate inhibition Calcium after adsorption Calcium requirement and phosphate inhibition Calcium after adsorption Divalent cations required Salt effect Salt effects Salt effects Salt effects WoUman and Wollman (1936a) Wollman and Wollman (1938) Clarke (1952) Barksdale and Pappen- heimer(1954) Rifkind and Pickett (1954) Cherry and Watson (1949) E. B. Collins, Nelson, and Parmelee (1950) Shew (1949) Reiter (1949) Potter and Nelson (1953) Perlman, Langlykke, and Rothberg (1951) Asheshov (1926) Burnet and Lush (1935) Rountree (1955) Rountree (1951a) Bordet and Renaux (1928) Beumer and Beumer- Jochmans (1951) Stassano and de Beaufort (1925) Andrewes and Elford (1932) Burnet (1933e) Wahl (1946a) Wahl and Blum-Emerique (1951,1952b) KayandFildes (1950) Fildes, Kay, and Joklik (1953) Scribner and Krueger (1937) Gest (1943) Gratia (1940) Sertic (1937) 236 BACTERIOPHAGES dull reading. Much of it consists of some such statement as "inhibited by 0.01 M citrate" with no attempt to study the mechanism by which citrate inhibits phage reproduction. Therefore much of this literature is summarized in Table XV. There are undoubtedly many additional references which have been overlooked, but the table does suggest how broad is the range of phage types that require calcium. For a discussion of the more fundamental papers dealing with mechanisms see Chapters X and XI. 2. Metabolic Requirements for Phage Multiplication Phage multiplication involves the synthesis of about 100 copies of the infecting phage particle in a period of time commensurate with the generation time of the host cell. The total amount of new phage protoplasm synthesized in a single host cell, per- haps 10~^^ gram, is considerably less than the weight of the host cell which is of the order of 10"^^ gram, so the effort involved in producing the usual yield of phage is certainly no greater than that required for duplicating the host cell. The raw materials, the energy supply, and the anabolic enzymes required for phage synthesis should be well within the capabilities of the host cell insofar as phage protoplasm contains the same ingredients as the host cell. The only building block so far found in phages but not in the host cell is the hydroxymethylcytosine of the DNA of phages related to T2 (Wyatt and Cohen, 1953). With this single known exception, one might anticipate that the host cell would be able to furnish everything needed for phage multiplication ex- cept the unique patterns such as those which control the antigenic specificity of the phage proteins. These patterns are presumably furnished by the DNA of the invading phage particle. Despite this apparent simplicity of the problem of phage nutrition, a considerable amount of effort and ingenuity has been devoted to work which is directly or indirectly related to the requirements for phage multiplication. A question which has often been asked is, does the phage particle contain any enzymes, or is it entirely dependent on the metabolic activity of the host cell? REQUIREMENTS FOR PHAGE PRODUCTION 237 This question should be divided into two parts, one concerning the mature extracellular phage particle, the other involving the intracellular, vegetative phage. The answer may be quite dif- ferent in the two cases. a. Lack of Metabolic Enzymes in Mature Phage As discussed previously in Chapters X and XI, it is quite possible that phage enzymes are involved in the release of phage nucleic acid from its protein shell, and in the penetration of phage DNA into the host cell. Yet even here there is no certain evidence for the existence of phage enzymes. Far less evidence is available for the participation of phage enzymes in the later stages of phage reproduction. The important work of Hershey and Chase (1952) demonstrated that, following adsorption of phage to the host cell, most of the phage protein remains on the host cell surface while the phage nucleic acid penetrates through the cell wall into the cell interior. Other experiments indicate that about 3 per cent of the total phage protein penetrates into the host cell (Hershey, 1955). The function of this protein is not known but it might be enzymatic. Prior to this work it was commonly assumed that phage infection involved the penetration of the entire phage particle into the host cell where it multiplied in the host cell juices like a miniature bacterium. It is un- doubtedly this concept which induced various workers to search for oxidative and fermentative activities in phage preparations. Much of the earlier work on possible metabolic activities of phage particles was reviewed by Bronfenbrenner (1928) who con- cluded that although there was no evidence for an independent metabolism of phage, the question was still open. Bronfenbren- ner attempted to measure COo production by phage in a micro- respirometer using 10^^ particles (about 1 mg.) of phage. Al- though the method was sensitive to 5 /il. of CO2 he was unable to detect any COo production by this relatively large amount of phage. A very careful study of the respiratory and fermentative ac- tivities of a resting phage suspension was made by Schiller (1935) 238 BACTERIOPHAGES using phage WLL of Schlesinger, which is related to T2. This phage was concentrated and purified by high speed centrifuga- tion and contained 2.3 X 10^^ particles per mg. dry weight. Its respiration was measured in the Warburg apparatus in various nutrient media, including ammonium lactate-phosphate me- dium and glucose broth, both in the presence and absence of heat-killed bacteria. With amounts of phage varying from 1 to 17 mg. per vessel the respiration was negligible. For in- stance, with 1 7 mg. of phage in glucose broth containing 30 mg. of boiled £". co// cells the uptake of oxygen was 0.28 jul. per mg. per hour, in contrast to about 100 for bacteria. Glucose fer- mentation, measured as displacement of CO2 from a bicarbonate buffer, was likewise negligible in comparison with the fermenta- tive activities of E. coli. In an attempt to study the respiration of multiplying phage, Schiller measured the oxygen uptake of living host cells in lac- tate-phosphate medium in the presence of an excess of ac- tive or heat-killed phage. There was no detectable difference in respiratory activity whether the phage was active or dead. Schiller also tested 1 mg. amounts of active phage for the pres- ence of the following types of enzymic activity, trypsin, papain, lipase, amylase, maltase, nucleosidase, catalase, urease, arginase, and phosphatase. In tests lasting 24 hours, all results were negative except for phosphatase. The phage preparations were very active in splitting hexosediphosphate and also in hydrolyzing hexosemonophosphate, glycerophosphate, and nucleic acid. The phosphatase activity was inhibited by Af/10 fluoride. It could be greatly decreased by washing the phage with distilled water, and then largely restored by addition of a bacterial filtrate which had little activity itself, suggesting a requirement for some kind of cofactor. It is quite possible that this is bacterial phosphatase which is adsorbed to the phage particles, but if so the adsorption seems to be rather specific. A similar study was reported by Ajl (1950) using centrifu- gally concentrated T2 phage. Measurements in the Warburg respirometer using 5 X lO^'^ T2 particles per vessel showed no REQUIREMENTS FOR PHAGE PRODUCTION 239 detectable oxygen uptake with succinate, fumarate, malate, pyruvate, acetate, or glucose as substrates. An equal amount (4.5 mg. dry weight) of E. coli cells catalyzed the uptake of 100 to 200 ^il. of oxygen in 2 hours with these substrates under the same conditions. Similar experiments using 8.6 mg. of phage with glutamic or aspartic acids resulted in no detectable oxygen uptake, CO2 evolution, or NH3 liberation. Dehydrogenase ac- tivity was measured in Thunberg tubes using methylene blue at 1 :5,000 to 1 :50,000 and phage at 5 X 10^2 particles per tube. There was no detectable reduction of the dye in 6 hours using glucose, succinate, malate, or boiled E. co/z juice as substrates. No endogenous CO2 production and no CO2 evolution from glucose could be detected, but pyruvate and oxalactate as sub- strates yielded considerable amounts of CO2. However, this was readily demonstrated to be nonenzymatic in nature because boiled phage was more active than unhealed phage. Prelimi- nary experiments indicated that concentrated phage preparations did contain some ATPase activity, but no quantitative results were given and it is not clear whether this is a true property of the phage or due to a contaminating bacterial enzyme. With this single exception, all experiments were in agreement in that they failed to yield evidence for metabolic activity in mature phage particles. Similar studies on concentrated preparations (5 X 10^^ ^^ly- ticles/ml.) of staphylococcus phage have been reported by Price (1952) but without details. "Tests for all the reactions in gly- colysis and those in the Krebs cycle were negative." Price also stated that there was no oxidation of gluconic acid, amino acids, or fatty acids Putnam (1953) reported that Kozloff (private communication) was unable to detect the presence in phage T6 of glycerol phos- phatase, phenolphthalein phosphatase, DNAase, ATPase, or protease. These results are in conflict with those of Schiiler and Ajl, who used closely related phages. One may safely conclude that up to the present there is no unequivocal evidence for the existence of any enzymic activity 240 BACTERIOPHAGES in mature, extracellular phage, and there is strong evidence against the existence of the usual enzymes of intermediary metabolism. This implies that the multiplying phage is de- pendent on the host cell for energy yielding enzyme systems and for the synthesis of the common organic compounds which are constituents of mature phage. In addition the infected host cell may synthesize certain new enzymes which are required for phage replication and which are found in neither the uninfected host cell nor in mature phage. Such hypothetical enzymes would then be uniquely associated with the vegetative state of bacterio- phage. b. Enzymic Activity Associated with Vegetative Phage The infected bacterium has a number of physiological proper- ties not shared by the normal bacterium. It synthesizes several antigenically specific proteins that are unique to the phage particle, as well as a unique kind of nucleic acid. These proteins and nucleic acid molecules are assembled into the peculiar morphological structure characterizing the mature phage particle. Nothing is yet known about the enzymes required for the syn- thesis of proteins and nucleic acids, but if any of the specificity of these substances is dependent on specific enzymes, then vegetative phage must have associated with it some phage-specific en- zymes. A search for such enzymes is beyond the reach of present biochemical techniques. In the meantime, several pieces of evidence suggest the possibility that there may be enzymes of intermediary metabo- lism specifically associated with vegetative phage. One piece of evidence stems from the discovery of the pyrimidine base 5- (hydroxymethyl)cytosine (HMC) in phages T2, T4, and T6 by Wyatt and Cohen. This base is not present in detectable amounts in the uninfected host bacterium and has not yet been found elsewhere in nature. Infection of the host cell with phage T2 results in a prompt halt in the synthesis of cytosine and initia- tion of the synthesis of hydroxymethylcytosine (Hershey, Dixon, and Chase, 1953). It has been suggested by Cohen (1953b) that REQUIREMENTS FOR PHAGE PRODUCTION 241 this shift in the synthetic activities of the bacterium following in- fection with T2 phage is sufficient explanation for the long- recognized fact that the net synthesis of bacterial DNA and RNA stops following infection. If this suggestion is correct there still remains the major problem of the mechanism for the shift from cytosine synthesis to the synthesis of HMC. Various possible mechanisms have been discussed in detail by Cohen (1953b), among them the possibility that the infecting phage supplies the essential enzymes for HMC synthesis. A possible example of enzymic activity supplied by the infect- ing phage was reported by Earner and Cohen (1954). A mutant strain of E. coli, 15T~, is unable to grow unless thymine or thymidine is supplied. When this strain is infected with phage T2 in the absence of thymine, the phage multiplies and its thymine content ultimately exceeds that of the cells prior to in- fection. The authors suggested two possible explanations, that the infecting virus supplies an enzyme or coenzyme essential for thymine synthesis, or that virus infection releases an inhibition of an enzyme system already present in the uninfected cell. A later paper furnished further information on the metabolism of strain 15T~ (Cohen and Earner, 1954). The bacteria were grown in uniformly labeled C^'*-glucose and the nucleic acid bases were iso- lated and analyzed for C^'^ content. The experiments indicate that strain 15T~is able to synthesize thymine at about 4 per cent of the normal rate but much too slowly to permit growth of the bacteria. This finding suggests that phage infection releases an inhibition rather than furnishes a lacking enzyme. These studies demonstrate the difficulty in drawing valid conclusions without a thorough investigation of the system. A purine requiring mutant of E. coli strain E which permitted production of phages Tl and T5 in the absence of added purines was reported by Gots in a discussion of the paper of Cohen (1953b). This strain did not permit production of phages T2, T3, and T4 unless a purine was added to the medium. Similar experiments with amino acid-requiring strains of bac- teria yielded different results. The infecting phage does not 242 BACTERIOPHAGES compensate for the bacterial deficiencies (Cohen, 1949; Gots and Hunt, 1953; Burton, 1955). Although there is abundant evidence that an infecting bac- terial virus can seriously alter the host cell metabolism, there is still no clear cut evidence that the invading bacteriophage in- troduces metabolically significant enyzme systems into the host cells. Phages can also introduce bacterial genetic substances controlling enzyme synthesis by means of the process known as transduction, and prophages in lysogenic bacteria may have profound effects on the metabolic activities of the host cell (Chap- ter XIX). Indeed, the evidence furnished by work with tem- perate phages makes it seem quite probable that enzymatic activities uniquely associated with infection will be demon- strated eventually. c. Enzymic Activity of the Host Cell It is now generally accepted that the energy requirement for phage multiplication must be obtained through the functioning of the enzymes of the host cell. Much of this evidence, obtained by the use of enzyme inhibitors, will be described in Chapter XV. However, one piece of evidence derived directly from an interest- ing property of the phages is worth discussing here. The infec- tion of a susceptible bacterium by certain virulent phages pre- vents the synthesis of adaptive enzymes which are readily formed in uninfected bacteria (Monod and Wollman, 1947). A strain of E. coli infected with a virulent bacteriophage lysed and liberated mature phage if glucose were present as an energy source. How- ever, in the presence of lactose, phage growth and cell lysis occurred only if the bacteria had been adapted to lactose utiliza- tion before infection. Synthesis of the required enzyme took place in uninfected cells within an hour after addition of lactose, but did not occur at all in infected cells. In a further refinement of this technique Benzer (1953) demonstrated that the rate limiting factor in phage multiplication in a lactose medium is the amount of induced enzyme synthesized before phage infec- tion. These experiments show clearly that the lactose-hy- REQUIREMENTS FOR PHAGE PRODUCTION 243 drolyzing enzyme is a host cell enzyme and is not replaceable by phage enzymes. Rather similar evidence is available with respect to the enzymes involved in the synthesis of amino acids. If a culture of E. coli is grown in a medium containing ammonium lactate and salts it will support the growth of a number of phage strains without the addition of amino acids or growth factors. How- ever, if broth-grown bacteria are transferred to the chemically defined medium and then infected with T2 phage the latent period is greatly prolonged and the burst size decreased. An appropriate mixture of amino acids and purine and pyrimidine bases restores phage production by broth grown bacteria (Fowler and Cohen, 1948; Cohen and Fowler, 1948). These experiments show that the rates at which the bacterium can synthesize the raw materials for phage protein and nucleic acid depend on the environment in which the bacteria were grown before infection. Very similar results were observed by Gots and Hunt (1953) in studying the requirements for growth of phage lambda after ultraviolet induction of lysogenic cultures. It was found that for lambda production in broth-grown cells, isoleucine, leucine, and valine are essential. Yet when strain K12 is grown in a glucose- ammonium chloride medium, induced lysis and phage produc- tion occur without the necessity of added amino acids (Borek, 1952). Again the nutritional requirements for phage produc- tion are seen to depend on the synthetic capabilities of the host cells as conditioned by their previous environment. In bacteria genetically incompetent to synthesize certain amino acids, phage production does not occur unless the amino acids required for bac- terial growth are included in the medium (Borek, 1952; Burton, 1955). Adsorption and invasion by T2 occurs in the absence of the amino acid required for cellular multiplication. However, no phage DNA synthesis occurs, nor does the complex become resist- ant to ultraviolet light (see Chapter XI), until the required amino acid is added. A similar block in development is caused by chlo- ramphenicol (Chapter XV). In washed and starved bacteria unknown physiological 244 BACTERIOPHAGES changes take place which interfere with phage production. Such studies were reported by Gross (1954a, b) using phage T2 and strain K12 of E. coli gro-wn in a glucose ammonium chloride medium. If such cultures were infected with phage T2, they lysed normally with an average burst size of 10 to 20 phage particles per cell. If the bacteria were washed in buffer, starved by aeration, infected, and then returned to glucose medium there was no phage yield. However, if the starved infected bacteria were incubated in broth there was an average burst size of 20. Results similar to those observed in broth were obtained by re- suspending the bacteria in a mixture of amino acids. No single amino acid would suffice for phage production. Evidently starvation of the bacteria causes some physiological damage which renders the cells unfit for phage production in unsupple- mented media. The effect of the damage can be reversed by a suitable mixture of amino acids. One may conclude from what information is available that phage production is dependent on the host cell metabolism for energy and for the synthesis of the raw materials of protoplasm such as amino acids and nucleic acid bases if these are not fur- nished in the medium. Therefore, any interference with the energy metabolism or the synthetic enzyme systems of the host cell may be expected to have an effect on phage production. However, it is possible to interfere with bacterial multiplication without affecting phage growth and vice versa, showing that the nutritional requirements for the two processes are not identical. d. Synthetic and Energetic Machinery of the Host Cell The immediate effects of infection by the virulent phage T2 are numerous and dramatic. Cell division is halted and the synthesis of new respiratory enzymes is stopped, but the func- tioning of the existing respiratory apparatus remains unim- paired. The net synthesis of ribonucleic acid and of bacterial deoxyribonucleic acid (l)NA) stops, and the synthesis of phage DNA soon starts (Cohen, 1949). The bacterial nuclei are de- stroyed (Luria and Human, 1950), but the tetrazolium reductase REQUIREMENTS FOR PHAGE PRODUCTION 245 activity of the infected cells remains unimpaired (Hartman, Mudd, Hillier, and Beutner, 1953). Synthesis of at least one enzyme is abruptly terminated (Benzer, 1953), but protein synthesis as a whole continues unabated (Cohen, 1949; Hershey, Garen, Fraser, and Hudis, 1954). Cohen (1947a) summed up these metabolic effects as follows: "The virus appears to be synthesized by the cell according to the models (templates) which it provides for the host's enzymes." Luria (1950) refers to "parasitism at the genetic level." Both authors express what is now the common view: after infection with T2 specific bacterial functions, such as synthesis of bacterial DNA and bacterial enzymes, can be dispensed with. General- ized processes, such as formation of amino acids and nucleotides, and doubtless oxidative phosphorylation, remain essential to viral growth. The bacterium, as we have seen, must furnish these generalized working systems out of its past activity; the phage cannot create them and indeed renders the bacterium un- fit to do so ; it supplies mainly a new detailed plan for coordinated action of existing metabolic systems. The following observations are consistent with this view in showing that deliberate inter- ference with specific bacterial syntheses does not prevent growth of phage. Bacteria treated with as many as 30 lethal doses of ultraviolet light are still able to support the multiplication of phage T2 (T. F. Anderson, 1944, 1948d; Luria and Latarjet, 1947; La- baw, Mosley, and Wyckoff, 1950a, b). Jacob, Torriani, and Monod (1951) exposed E. coli to sufficient ultraviolet light to re- duce the survivors by a factor of 1 0\ As a result of this treatment the bacteria lost ability to synthesize the inducible enzyme j8- galactosidase in detectable amounts, yet still were able to produce T2. Phage can also multiply in bacteria treated with many lethal doses of X-rays, as demonstrated by Rouyer and Latarjet (1946), by Latarjet (1948), and by Labaw, Mosley, and WyckofT (1953). Similarly, bacteria rendered nonviable by treatment with mustard gas are able to support growth of phage T2 (Herriott, 1951b). Irradiation and mustard treatment probably 246 BACTERIOPHAGES kill bacteria by specific inhibition of DNA synthesis (Kelner, 1953; Herriott, 1951b). Evidently phage infection reverses this inhibition, which suggests that the original damage was con- fined to terminal steps in DNA synthesis. Similar conclusions may be drawn from numerous papers on the induction of phage multiplication in lysogenic bacteria by agents normally lethal for bacteria. For instance, a dose of ultraviolet light that kills 70 per cent of a nonlysogenic variant of K12 gives 95 per cent induction of phage growth in the lysogenic strain (Weigle and Delbriick, 1951). Penicillin is supposed to interfere specifically with formation of cell wall material in E. coli (Hahn and Ciak, 1957). Concentra- tions of penicillin which prevent bacterial multiplication do not prevent phage growth, although the phage yield is somewhat reduced because of premature lysis (Price, 1947a; Elford, 1948; Krueger, Cohn, Smith, and McGuire, 1948). Such lethal con- centrations of penicillin do not interfere with most of the synthetic abilities of the bacteria because the cells increase markedly in size and in content of protoplasm after treatment with penicillin, much as they do after treatment with ultraviolet light, X-rays, or mustard gas. We may conclude that viability of the host cell is not important to the lytic cycle of phage growth provided generalized metabolic processes continue to function. However, interference with energy metabolism or with the source of supply of building blocks results in failure of phage growth. These conclusions reached from the study of phage T2 have been informative, but must be regarded as an example of the extreme case. Less severe metabolic effects of infection are seen with other phages (Siminovitch, 1953). Synthesis of DNA and cellular multiplication may be interrupted, but other biosyn- theses, including enzymic adaptation, may continue at a reduced rate. Even synthesis of DNA does not stop completely (Lwoff, 1 953) . Evidently if the bacterium is to survive, as in the event of lysogenization by a temperate phage, practically all metabolic effects of infection would have to be reversible. Tl, though little REQUIREMENTS FOR PHAGE PRODUCTION 247 Studied from this point of view, may be a case in point. Accord- ing to Luria and Delbriick (1942), Tl inactivated by ultraviolet light does not kill bacteria. It nevertheless produces striking cytological changes in the cells (Luria and Human, 1950). In no instance is it clear which effects of infection should be ascribed to action localized at the cell surface and which to the injected materials. Thus ultraviolet-inactivated T2 and T5 (and a few other phages, perhaps) kill bacteria with high efficiency. Ghosts of T2 kill with low efficiency, and adsorbed ghosts that fail to kill produce transient cellular changes (French and Siminovitch, 1955). Systematic elucidation of these questions will be a neces- sary part of the clarification of current ideas about possible phage-host relationship (Stent, 1958). 3. Partial Metabolic Requirements for Vegetative Replication We have just seen that multiplication of phage T2 is inde- pendent of specific bacterial biosyntheses such as that of DNA, certain enzymes, and probably bacterial ribonucleic acid. Even more remarkable is recent evidence that synthesis of phage pro- tein and DNA are to some extent independent. Watanabe (1957) showed that bacteria heavily irradiated with ultraviolet light some time after infection with phage T2 showed a markedly reduced capacity to synthesize DNA, but continued to form serologically specific phage protein at an appreciable rate. Their results suggest that the role of DNA in protein synthesis is a passive one, as do the corresponding experiments with uninfected bacteria (Kelner, 1953). Burton (1955) first showed that synthesis of DNA is independ- ent of synthesis of protein in bacteria infected with T2. To test this, he used amino acid-requiring strains of E. coli, which were grown in supplemented medium, infected with phage, and transferred to deficient medium at various times after infection. He found that deprivation of an amino acid during the first few minutes after infection could prevent phage DNA .synthesis from starting, but had little eff'ect if the transfer to a deficient medium was postponed until a later time. The same phenomenon was 248 BACTERIOPHAGES discovered independently by Tomizawa and Sunakawa (1956), who used the antibiotic chloramphenicol to inhibit protein syn- thesis (Wisseman, Smadel, Hahn, and Hopps, 1954). Hershey and Melechen (1957) made a thorough study of the action of chloramphenicol in T2-infected bacteria. Suitable time schedules of chloramphenicol inhibition and reversal, to- gether with labeling of phage precursors by means of radioactive isotopes, showed that large amounts of typical phage precursor DNA could be accumulated in cells that contained virtually no phage precursor protein. When such cells were transferred to a medium free from chloramphenicol, phage particles were formed that contained principally the phosphorus of DNA formed be- fore, and protein formed after, the removal of the antibiotic. These findings constitute the chief evidence for the view (Her- shey, 1953b) expressed frequently in this book, that vegetative reproduction and maturation in phage T2 can be equated, re- spectively, with synthesis of phage DNA and protein. According to Hershey (1957) the evidence is still incomplete. Another line of evidence pointing in the same direction may be cited here. Watanabe, Stent, and Schachman (1954) in- fected bacteria with P^^-labeled particles of phage T2 and subse- quently broke open the infected cells and measured the sedimen- tation constant of the P^Mabeled DNA in the extracts. Depend- ing on the time of breakage of the cells, the labeled material sedimented like free DNA or like mature phage particles. No evidence was found for phosphorus-containing structures of intermediate complexity. The authors interpreted their results to mean that phage precursor DNA does not form stable attach- ments to complex particles in the cell until incorporated into a finished phage particle. 4. Sources of Material for Phage Synthesis Three sources of the raw inaterials used for the formation of phage particles can be distinguished by the appropriate use of isotopic tracers: the substance of the parental phage particles, the bacterial contents at the time of infection, and the culture REQUIREMENTS FOR PHAGE PRODUCTION 249 mediuin in which phage [growth occurs. Much of the work along this line was done by Putnam, Kozloff, and Evans, summarized in several reviews (E. A. Evans, Jr., 1952; Putnam, 1952; Kozloff, 1953). a. Materials Derived from the Parental Phage The infection of one bacterial cell by a single phage particle can result in the production of 100 or more progeny phage particles. It is evident that quantitatively the contribution of parental substance to phage progeny must be small. Nonethe- less, this topic has been very actively studied because of the evi- dence it may furnish about the mechanisms of phage reproduc- tion. This work was discussed in detail in Chapter XIII. It is sufficient here to recall that only constituents of the parental DNA are significantly transferred to the offspring, with an efficiency of about 50 per cent. There is no reason to suppose that this transfer is obligatory in a nutritional sense, indeed, only the first offspring particles to be formed contain measurable amounts of parental phosphorus. In point of fact, however, the transfer cannot be prevented by experimental means. b. Protein Materials Assimilated before and after Infection The contributions from these two sources can be measured separately and should add up to the total phage protein. By growing the bacteria in a medium appropriately labeled with isotopes one can label the bacterial proteins, nucleic acids, or both. Then by transferring to an unlabeled growth medium one can measure the bacterial contribution to the phage substance. The dilution suffered by a given isotope in any compound in going from the bacterium to the phage is a measure of the bac- terial contribution providing exchange reactions are ruled out. Similarly by growing the phage on unlabeled bacteria in a labeled growth medium one can determine the fractional contribution to phage substance made by the medium. One may study the process kinetically by introducing an appropriate isotopic label 250 BACTERIOPHAGES at any time before or after bacterial infection and similarly one can dilute out the label at any time. This provides a tremendous flexibility to the experiments and permits one to study the dy- namics of phage synthesis. Relatively less information is available as to the source of phage proteins because the major effort has been expended on the currently more interesting problem of nucleic acid metabo- lism. Kozloff, Knowlton, Putnam, and Evans (1951) studied the source of phage T6 nitrogen compounds by labeling either the bacterium or the growth medium with N^\ They concluded that from 10 to 20 per cent of the phage protein nitrogen is derived from bacterial substances assimilated before infection and the remainder is obtained from the medium after infection. The contribution to phage amounts to about 2 per cent of the bacterial protein indicating that most of the bacterial protein is unavailable for phage synthesis. The experiments were extended by Siddiqi, Kozloff, Putnam, and Evans (1952). Using bacteria labeled with N^^ and with C^*-lysine, they concluded that about 1 5 per cent of the phage T6 protein could have been derived from the host cell. The source of this material was host cell proteins rather than acid-soluble metabolites. Only 1 to 2 per cent of the bacterial lysine was available for synthesis of phage protein. Most of the virus pro- tein must be synthesized de novo from materials in the growth medium. Putnam, Miller, Palm, and Evans (1952) isolated the protein from phage T7 grown in bacterial cells labeled with N^^ and concluded that about 40 per cent of the phage protein nitrogen was derived from bacterial substances assimilated before infec- tion and the remainder was derived from the medium after in- fection. The increased fraction of host protein in phage T7 as compared with phage T6 is probably a reflection of the smaller size of phage T7, since only a small amount of bacterial sub- stance is utilized in either case. There is one major difficulty with all these experiments, and that is the difficulty in determining when the phage preparation REQUIREMENTS FOR PHAGE PRODUCTION 251 is adequately pure. In all of the previous papers the final phage yield used for analysis was purified only by differential centrifu- gation. Physical and chemical criteria of purity applied to such preparations are virtually useless as tests of radiochemical purity in experiments of the type under consideration. This point was tested in experiments by Hershey, Garen, Fraser, and Hudis (1954) using phage T2 grown in bacteria labeled with preassimi- lated S^\ The radiochemical purity of the centrifugally isolated phage preparations was measured by following the adsorption of the sulfur label to strain B of E. coli and failure of adsorption to B/2. These tests indicated that indeed most of the radioactivity of the final phage yield was due to contaminating bacterial pro- teins and was not built into the phage particles. Only 2 to 3 per cent of the total phage protein could have been derived from labeled bacterial protein, the latter contribution amounting to only 0.4 per cent of the total labeled bacterial protein. Even this small amount may have been derived from bacterial gluta- thione rather than from bacterial protein. These experiments demonstrate quite conclusively that the host cell makes a negligi- ble contribution to the proteins of phage T2. This finding that contaminating bacterial protein is responsible for the host cell isotopic label found in the phage yield is sufficient explanation for the observation by Kozloff', Knowdton, Putnam, and Evans (1951) of an inverse relationship between the phage yield and the apparent host contribution to the phage. If the amount of con- taminating bacterial particles per cell present in the phage yield is approximately constant, as indicated by the results of Hershey, Garen, Fraser, and Hudis (1954), it would constitute a smaller fraction of the yield, as the yield of phage particles per cell in- creased. This might also be the explanation for the apparently large host cell contribution made to the small phage T7 as ob- served by Putnam, Miller, Palm, and Evans (1952). One may conclude that at present there is no definitive evidence that aiiy phage materials are derived from host cell proteins although in the case of phage T2 it is possible that 2 to 3 per cent of the phage protein might be derived from this source. Evidently 252 BACTERIOPHAGES most if not all of the phage protein is derived from materials as- similated from the growth medium after phage infection. The kinetics of protein synthesis in phage-infected bacteria has been investigated in several laboratories. Cohen (1947 a, 1948) found that protein synthesis in T2-infected bacteria proceeded without interruption, although the net synthesis of RNA and DNA was stopped. The protein increment was not characterized to determine whether it was of phage or bacterial specificity. Levin thai and Fisher (1952, 1953), by breaking open T2-in- fected bacteria at intervals during the latent period, observed toroid-shaped objects ("doughnuts") which appeared before mature phage, increased in number during early stages of phage maturation, and then decreased toward the end of the latent period. These objects were apparently phage precursors, re- sembling empty phage heads in shape and size and being agglu- tinated by antibodies to phage heads. They did not adsorb to the host cells because the organ for attachment, the tail, was lacking. A small number of doughnuts with tails were also ob- served in these premature lysates. Similarly DeMars, Luria, Fisher, and Levinthal (1953) and DeMars (1955) reported the detection of soluble phage antigens in T2-infected bacteria which were not part of either mature phage particles or "doughnuts." These soluble antigens were characterized by their ability to block phage-neutralizing antibodies. These experiments dem- onstrate that proteins with the specificity of phage but not yet built into mature phage are present in phage-infected bacteria during the latent period. Similar experiments with T5-infected bacteria were reported by Y. T. Lanni (1954). A different kind of phage precursor was detected in T2-in- fected bacteria by Maal0e and Symonds (1953) by the use of S^^ from the medium as a label. These precursors contained protein but no nucleic acid, sedimented more slowly than mature phage, adsorbed to sensitive bacteria but did not kill them, and were agglutinated by antiphage serum, lliese properties resemble those of "osmotic ghosts" of phage T2. Their number was ap- proximately constant from 15 minutes after infection until lysis REQUIREMENTS FOR PHAGE PRODUCTION 253 and measured 30 to 50 per infected bacterium (assuming the same sulfur content as mature phage) . By introducing S '^•'-sulfate into the medium at various times during the latent period it was possible to determine the time required for a newly assimilated S^'' atom to appear in a mature phage particle. The average elapsed time was found to be 6