HOST-INDUCED MODIFICATIONS OF VIRUSES’ S. E. LURIA University of IHinois, Urbana, Illinois Viruses exhibit extensive adaptability to growth in various hosts or tissues. It was widely held in the past that virus adaptability reflected a peculiar plasticity of virus heredity, which allowed it to be directly influenced by its host cells. The alternative interpretation of virus adaptation to new host cells as due to spontaneous mutations, which provide a range of genotypes for the new hosts to select from, always had authoritative proponents (see Findlay, 1939). This viewpoint finally gained wide recogni- tion (see Burnet, 1946), partly as a consequence of the development of phage genetics and of the inter- penetration of various branches of virology. It is now recognized that most variation in virus proper- ties is caused by viral mutations, and that virus plasticity results from the variety of genotypes present in the large viral populations. It was, therefore, an unexpected development when a new type of virus variation was discovered in bacteriophages. This has been called host-in- duced or host-controlled variation (Luria and Hu- man, 1952; Ralston and Krueger, 1952; Anderson and Felix, 1952; Bertani and Weigle, 1953). Its outstanding characteristics are that it is strictly phenotypic, non-hereditary, and that it is deter- mined by the host cell in which a virus has been produced. In this paper, I shall summarize the fea- tures of this phenomenon; I shall compare it with other instances of nonhereditary phage variation; and I shall attempt to assess its possible bearing on certain problems in other areas of virology. Host-Inpucep MopiFIcaTIONs IN BACTERIOPHAGE Host-induced modifications have been described in coliphages, in salmonella phages, and in staphylo- coccus phages. The instances that have been recog- nized affect, by restricting or enlarging it, the host range of bacteriophages. There is no reason to assume that other phage properties cannot be af- fected by host-induced variation. Indeed, the phage- mediated “transduction” into a bacterial strain of some property of the host strain in which a phage has been formed (Zinder and Lederberg, 1952) is itself a host-controlled variation of phage, although it is recognized by changes of the host cell rather than of the phage itself. The common features of all host-induced modi- fication of host-range thus far recorded are, (a) a restriction of the ability of the phage to grow in 1 Supported by a grant from the American Cancer Soci- ety, recommended by the Committee on Growth. The assistance of Miss Doris Templin in some of the experi- mental work reported in this paper is gratefully .acknowl- edged, some host as a result of one cycle of growth in one type of cell; and (b) a release of this restriction following one cycle of growth in some other host. For example, phage P2 grown on Shigella dysen- teriae strain Sh can grow in every cell of Sh but only in about one of 10* cells of Escherichia coli strain B. The same phage P2 grown in B (for example, the phage liberated by the one cell in 10* above) can grow in all cells of B and in all cells of Sh. But, if grown in Sh, it is again restricted to one in 10 cells of B. We say that P2 grown on Sh is in the P2 Sh form, whereas P2 grown on B is in the P2 B form (Bertani and Weigle, 1952). The variation P2 Sh —> P2 B is adaptive, since it per- mits continued growth on the host B. The first instances of host-induced variation in- cluded only two alternative forms of each phage, but a more general situation may involve several such forms. The situation is conveniently described by a scheme proposed by Weigle and shown in Table 1. The restricted ability of a phage to grow in some host is characterized by a specific “yielder frequency” or “acceptance frequency,” that is, by the proportion of the cells of that host that, if in- fected, can support growth of that phage. The re- striction depends only on the strain in which the phage has undergone the last reproductive cycle. The full situation of Table 1 has been demon- Taste 1, THe Scueme or ApAptive Host-INDUCED MOopIFICATION The figures correspond to the yielder frequencies (= proportion of cells that liberate phage) for the various host-phage combinations. ; Hosts . Phage forms A B c PA Uw Ud 10-4 10-6 (= P grown on A) PB ...... pvnceececesesenssrereeneees OL 1 10-6 (= P grown on B) Cc _.... aececeeeseteceereeeeeeeee OL 10-4 1 (= P grown on C) PBC=PC 10-4 1 (= P grown on B, then on C P CB = PB _____.._.___- 1 1 10-6 (= P grown on C, then on B) PBA=PA 10-4 10-6 (= P grown on B, then on A) PGA=PA.,...WW. 1 10-4 10-6 (= P grown on C, then on A) Homologies for phage P2: E. coli B = A; Sh. dysen- terice — B Homologies for phage \: E. coli S = A; E. coli C= B Homologies for phage T1: E. coli B = A; E. coli 6 = B; E. coli F/50 = C. Homologies for staph phage Pl: Staphylococcus #145 = A; Staphylococcus Ki = B. [237] 238 S.E. LURIA | Taste 2, THe ScHeMe or Host-Rance Mutation The figures are yielder frequencies for hypothetical hosts infected with a phage or with its host-range mutants. Hosts Phage forms A B Cc PA Wo 1 10-8 10-7 P B (= Phg)-. 1 1 10-7 PC (= Phe)... 1 10-6 1 PB,A ___WW.... 1 1 10-7 P B,C (= Phghc) 1 1 1 PCA ow. 1 10—6 1 P CB (= Phehp) -... 1 1 1 The ability to grow on a given strain, once acquired, is permanently maintained (barring rare reverse mutations). strated with phage Tl by Mrs. N. Collins Bruce (personal communication). A, B, C are three un- related strains of E. coli (strains B, 0, and F/50). The situations with coli-dysentery phages P2 and A (Bertani and Weigle, 1953) and with staphylo- coccus phage Pl (Ralston and Krueger, 1952) fit the scheme of Table 1 with only two pairs of entries (A-P A and B-P B). In these cases there is one host (A) in which all phage forms grow with equal, maximal frequency. A situation might easily be encountered in which only two hosts like B and C are known, mutually restrictive in their effect on a phage. The relations in Table 1 differentiate the host- induced modifications from host-range mutations in phage, which for purposes of comparison are illustrated in the scheme of Table 2. Here, the characteristic feature is the persistence of the “adaptation” after return to the original host. The instance of host-induced modification that was recognized first (Luria and Human, 1952), and that led to the recognition and interpretation of most other instances, differs from the prototype of Table 1. It is illustrated in Table 3. Here, one growth cycle in a host strain E. coli B/4,, differing Taste 3. THe ScHemMe ofr Unapaptive Host-Inpucep MobIFIcATION Data from Luria and Human (1952) The figures are yielder frequencies for various host-phage combinations. Hosts A B Cc (= Sh. dysen- (= E.coliB) (= E.coli Phage forms teriae Sh) B/4.) PA (= T2) 1 1 1 PB (=T2) I 1 1 : PC (=T*2) 1 10-3 10-3 PC,A (=T2) 1 1 1 PC,B (= T2) 1 1 1 PC,C (= T*2) 1 10-38 10-3 by a single spontaneous mutation from another host, E. coli B, modifies phage T2 (or T6) to the form T*2 (or T*6), characterized by a restriction of growth ability to a small proportion of the cells of the modifying host and of its relatives (E. coli B and all its derivatives). Growth of both T2 and T*2 is unrestricted in the unrelated host S. dysen- teriae Sh, which liberates phage in the T2 form. The modification induced by B on the T*2 form is adaptive, but the modification induced by B/4, on T2 is unadaptive, since it restricts the growth abil- ity on B/4, itself, In summary, the known instances of host-con- trolled modification of phage involve, on the one hand, a restriction by one or more hosts of the growth ability of the phage on some host and, on the other hand, a release of the restriction by some other hosts. The latter hosts in turn may or may not impose other alternative restrictions. The modi- fications imposed by successive hosts are not addi- tive but mutually exclusive. Each host modifies the phage in a characteristic way, independent of the previous host history of the phage (see Table 1). The phage modification imposed by a given host ig the same, whether the phage is liberated by a lytic cycle or from lysogenic cells (Bertani and Weigle, 1953). As far as we know, a given modification is similar in all genetic mutants of a phage (for ex- ample, in P2 and its virulent mutants; in T2, T2A, T2r ... ). The modifications generally affect the | totality of the phage produced in the modifying host cells, GENERAL CHARACTERISTICS OF THE Host-INDUCED MopiFicaTions rv Host RANGE Host-induced modifications, as opposed to host- range mutations, are characterized, not only by their ready reversibility, but also by the determi- nation of the few successful particles of a restricted phage form that succeed in overcoming the restric- tion. The observations can be listed as follows: 1. The success of the few particles that manage to grow is not due to a difference in adsorbability. All alternative forms of a phage are equally well adsorbed by any given host, whether they grow in it or not. 2. The ability to overcome the growth restric- tion results from the attachment of particles of the restricted phage form to some exceptional cell of the host. The evidence for this statement is as follows: (a) The proportion of cells in which a restricted phage succeeds in growing (as for P2 Sh on B) can be altered by a variety of environmental fac- tors acting on the host before infection. (b) In mixed infection of bacteria with two mu- tants of a restricted phage (for example, P2 Sh and P2 vir Sh on B, or T*2 and T*2r on B) the fre- quency of mixed yielders may be, say 30 per cent when the total frequency of yielding complexes is only one per cent. If the yielders were bacteria infected with exceptional phage particles, the fre- HOST INDUCED MODIFICATIONS OF VIRUSES . 239 quency of mixed yielders should reflect the coinci- dence of two exceptional particles of different mu- tants infecting the same cell (less than 10~* in the experiment quoted above). (c) Phage mutant particles, when they first ap- pear in nonmutant populations, are present in char- acteristic clones of identical sibs, each clone deriv- ing from one mutation (Luria, 1951). Instead, when single bursts of a restricted phage are tested for the number of particles that succeed in over- coming the restriction, the rare yielders are dis- tributed at random (at least as long as the bacteria are in large excess, as in platings) (Bertani and Weigle, 1953). This is explained if we consider that a yielder is an exceptional bacterium infected with a nonexceptional phage particle. The facts listed above do not exclude completely that the particles of a restricted phage may be heterogeneous in their ability to grow in excep- tional bacteria. The fact that in single infection experiments the number of yielder bacteria is a linear function of the phage inoculum is explained by the linear increase of exceptional cells that be- come infected. In multiple infection, some compli- cations appear, which have not yet been adequately investigated. ‘In summary, the modification of a phage by growth in a host towards which it was restricted is due to the accident of acceptance of some particle of the restricted phage by some exceptional, “ac- tive” cell of the host. On the basis of experiments with phage Tl grown on various hosts, Fredericq (1950a, b) has questioned the hypothesis of a spon- taneous origin of host-range mutants. Prominent among Fredericq’s findings was the random, non- clonal distribution of the T1 particles with extended host range in platings on various hosts (strains E. coli B, 3, C.18). Apart from some inadequacies of methodology used in this work (small samples from each of six large cultures), most of the ob- servations are easily interpreted in terms of host- induced modifications. The critical test of reversi- bility of the modifications of Tl upon return to other hosts, lacking in the original work, was done for Tl B and Tl # by N. Collins Bruce (personal communication), who showed complete transitions between the two forms in single growth cycles on E. coli B and E. coli } respectively. Tue NATURE OF THE RESTRICTED GROWTH ABILITY OF MODIFIED PHAGE The stage of arrested development. This stage varies from phage to phage. It always follows ad- sorption; adsorption is equal for restricted and unrestricted forms of the same phage. In some instances, for example with P2 Sh on B, there is no killing of the host; after adsorbing the restricted phage, the host is not slowed down at all in its development. We have been unable to observe any gross nuclear changes in B cells that had adsorbed several particles of P2 Sh, With other systems, there is complete suppression of cell division (T*2 on B or on B/4,). The in- fected cells may elongate before dying. Desoxyribo- nucleic acid synthesis is stopped. The nuclvar changes are not those characteristic of the normal infection with the corresponding unrestricted phage forms. No infectious phage can be revealed in the infected bacteria by artificial lysis. It is possible that the differences between the T2 and the P2 situations are related to other differ- ences between the infection of E. coli B with phages of the T group (suppression of enzyme syntheses, rapid nuclear disintegration) and the infection of bacteria with phages that do not produce these changes. The difference is not simply between tem- perate and virulent phages (as defined with rela- tion to lysogenicity) since a highly virulent mutant of P2 Sh fails to kill the B cells in which it does not grow. Zinder (personal communication) observed that phage PLT-22, the agent of “transduction” in Sal- monella, is modified, by growth on S. gallinarum, to a form restricted in growth ability on S. typhi- murium (yielder frequency 10-5). The restricted form can still transduce genetic properties of galli- narum to typhimurium with about normal frequen- cy. This observation suggests that the interaction between restricted phage and host goes far enough as to permit introduction and acceptance of the accompanying host-genetic material. Because of technical reasons, we have as yet been unable to prepare any P®?_labeled, growth-restricted phage suitable for testing whether the phage DNA is inoculated into hosts in which the phage fails to grow. The nature of the- exceptional cells that allow a restricted phage to grow. In various situations, the exceptional yielder cells may be as many as one in 40 (Ralston and Krueger, 1952) or as few as one in 108 (N. Collins Bruce, unpublished). If even fewer, they might not be detected at all and the variation would probably remain undetected. The conditions that modify the frequency of ex- ceptional cells vary from system to system: 1. Age of cells. Old cells of E. coli B or of its mutants (from aerated cultures in buffered nutrient ‘ with exhausted food supply) accept T*2 or T*6 with a frequency of 1 to 4 X 10~? instead of 10~¢ to 10-3. Rejuvenation (that is, reduction of yielder frequency, or deactivation of “active” cells) occurs rapidly if the old cells are aerated in fresh nutrient broth or in solutions of glucose, lactate, or other oxidizable substrates. The temperature coefficient for this deactivation is high. These observations suggest the removal, by oxidation, of some metabo- lite that is accumulated in the old cells and is opera- tive in allowing the restricted phage to grow. The age of E. coli B cells has little effect on their accept- ance ability for phage P2 Sh. . 240 S. E. LURIA 2. Medium and growth conditions. When E. coli B is grown in a casein hydrolysate medium with at least one per cent glucose and not more than one per cent KzHPO,, growth stops as the pH of the medium reaches 4.9-5.0. The cells from the acid cultures are viable. After washing in buffer, these “acid cells” accept T*2, T*6, or P2 Sh with frequencies of 10 to 50 per cent. (Similar growth conditions barely alter the acceptance frequency of cells of E. coli S for XC.) The requirements for acceptance of T*2 are less strict than for P2 Sh. T*2 is well accepted by acid cells from media of a variety of compositions, synthetic or variously sup- plemented, whereas the acceptance frequency for P2 Sh is much higher with acid cells from complete media. The nutritional factors involved have not yet been worked out. The low pH is not itself re- sponsible for the increase in acceptance frequency, since young cells growing (slowly) in media at pH 5.0 are not active. Filtrates of old, low-pH cultures do not activate inactive cells, The active cells from the acid cultures are de- activated slowly by aeration in fresh media with various carbon sources; the physiology of this activation and deactivation needs further study. We cannot tell at present whether the activity of the “acid cells” depends on storage of phage-needed intermediates or on removal of phage-growth in- hibiting mechanisms. 3. Ultraviolet irradiation. The activation by ultraviolet irradiation, discovered first with E. coli S cells as acceptors of XC (Bertani and Weigle, 1953), occurs also with FE. coli B and its mutants as acceptors for T*2, but not for P2 Sh. In the ultraviolet activation of B for T*2 acceptance, the remarkable feature is the continued increase in acti- vation at very high doses of ultraviolet (see Figure 1). The activation is almost identical for B and for its radiation-resistant mutant B/r, in spite of the great difference in ultraviolet sensitivity of their colony-forming abilities. Activation by ultraviolet is partly eliminated by exposure to “photoreactivat- ing” light. Ultraviolet-activated cells of E. coli B are deacti- vated by aeration in fresh media at a rate similar to the rate of deactivation of acid cells, and much slower than the rate of deactivation of old, starved cells. Yet, ultraviolet and growth to high acidity cannot act by the same mechanism, since ultraviolet activates B only as acceptor for T*2 whereas acid growth activates B as acceptor for both T*2 and P2 In summary, a variety of agencies can activate or deactivate cells, that is, change their accepting ability for restricted phages. With a given host, activating agents are specific for a given phage. This emphasizes the differences in the development- al sequences of various phages in the same host and in the stages at which these sequences are blocked with different restricted phages. 10° .—-—+-—+-_;—+ ' al s 5 S $105 O'& So > Oo wn ul a 4 a a < rr e * g 104 116° g 1 1 800 1600 i 1 “B00 1600 2400 DOSE, ERGXMM® Ficure 1. The number of irradiated cells of E. coli B/r or of E. coli B that yield phage when infected with phage T*2, as a function of the dose of ultraviolet light. The values in the figures correspond to mixtures containing 1 X 108 bacteria and 7 < 108 phage per ml., with over 90 per cent of the phage adsorbed. The broken lines are the survival curves of uninfected bacteria. THE Causes oF THE Host-Inpucep MopIFICATIONS | Except for restrictions of growth ability, no other differences, serological or physiological, have , been detected between host-modified forms of the same phage. These alternative forms may be pro- duced either by bacteria that differ by unknown, probably multiple properties (Sh. dysenteriae Sh and E. coli B; E. coli C and E. coli S); or by bac- teria that differ by one spontaneous mutation (EZ. coli B/4, and. E. coli B). The latter example sug- gests that one-step genetic differences may also be involved in other situations. E. coli B/4) has the phage-resistance pattern B/3,4,7. It can be iso- lated from E. coli B by the selective action of either T4, or T3, or T7. It does not adsorb these phages and apparently does not carry them lysogenically. There is another mutant of E. coli B, called B/4o. ' (Luria and Human, 1952) which in young cultures behaves like B/4,, transforming T2 into T*2. In old, starved cultures most cells behave like B and liberate T2 instead of T*2. Here, not only the acceptance frequency, but also the modifying abil- ity of a host for a phage depend on the physiologi- cal conditions of the host cells. An important lead is provided by observations on the Vi-phages of Salmonella typhosa. The c rigi- nal Vi-phage II of Craigie and Yen (1938), plated on each of 30 types of Vi-positive S. typhosa, gave a few plaques, from which more or less specifically “adapted” Vi-phages were isolated. These differ from the nonadapted Vi-phage II because they can grow unrestrictedly on one or more of the Vi-posi- tive host strains. The different host strains are then HOST INDUCED MODIFICATIONS OF VIRUSES 241 recognized or “typed” by their pattern of sensitiv- ity to the adapted phages (see complete scheme of Vi-types in Felix and Anderson, 1951). The adapted phages were supposed to be host- range mutants (see Craigie, 1946). Recently, how- ever, many of them were shown to be due to host- induced modifications of Vi-phage II (Anderson and Felix, 1952, 1953a). Reversion to the A (= unadapted) form occurs in one single passage on host type A. (This was tested by a single plaque isolation; the occurrence of the modification in a single growth cycle has not yet been established). These facts are summarized in Table 4, which cor- Taste 4. Tue ReLation or SoME VaRIANTS oF VI-PHACE If wirn Vi-Srrains or Salmonella typhosa Data from Anderson and Felix (1953a). The — sign indi- cates a plaque count at least 1000 times lower than on host strain A. Phage forms Host strains T* D5 D6* a io] pa o =~ * So rg (TA ITC TIC,A TI El TELA IT TIT,A IT DS {1 DS,A IT D6 II D6,A IDI ITD1,A Tr D4 1D4,A * Strain T is lysogenic for phage t and identical to A(t) in Vi-type; likewise, D6 is like A(d6); Dl is like A(d1). From: Anderson and Felix (1953b). Phage forms II C, II El, II T are presumably host-in- duced modifications (see Table 1). Phage forms II D5, II D6, II D1 are presumably host- range mutants (see table 2). They might be designated: WIhps; Whpshne; ILhn. respectively. Phage form II D4 is apparently a host-induced modifica- tion of the host range mutant IIAp:. It might be desig- nated: Ap: D4. PEt tte Ph ledt | PEP Tt tl Pl Elid Pil td de PEE td] Peld dd It | Il dl Ltd | Plt | Ltd eet et mm] | Itt | I 1d | | | Itt | Leal | responds to a portion of the classic Vi-typing scheme homologized with the scheme of our Table 1. We introduce the symbol II to indicate the Vi- phage II, and to distinguish between phages and bacteria. Some adapted phage forms are presum- ably host-range mutants (e.g., II D5); others are host-induced modifications (e.g., II T); others are combinations of both (e.g., II D4 is a host range mutant II D1 modified by host D4, that is, Ap; D4). Thus far, the situation in host-induced modi- fications in phage II is analogous to the case of Table 1. Several mutually exclusive restrictions and releases are impressed by several host strains on phage II, and a common host is attacked by all phage forms and restricts them against growth in all other hosts. The additional important feature is that the dif- ferences among some of the Vi-strains of S. typhosa, which determine their susceptibility to adapted Vi- phages and their modifying ability for these phages, are due to latent phages carried by individual Vi- strains and completely unrelated to the Vi-phages (Craigie, 1946; Anderson and Felix, 1953b). Loss or gain of lysogenicity for one of these latent phages can transform one Vi-type bacterium into another type. In its pattern of sensitivity to the adapted Vi-phages, a transformed strain may either correspond to- one of the other known strains or may exhibit a new pattern (“untypable strains”). In at least one instance, that of the transforma- tion of host strain A into host strain T by lysogeni- zation with the phage t, the new lysogenic strain A (t) is indistinguishable from T and has presum- ably acquired the ability to impress onto the Vi- phage II A the phenotypic modification to the form II T (see Table 4.).2 Instances of such transforma- tions may become more numerous as the system is further explored. The point of importance is that a latent prophage can presumably impress upon the host, not only the inability to accept an unrelated phage (as in many other known examples), but also the ability to discriminate among modified forms of an un- related phage and, even more important, the ability to impress a specific modification upon that phage. It would be premature fo generalize as to the role of prophages in other instances of host-induced variation in phage. The difference between E. coli B and B/4,, for example, is not due to lysogeniza- tion. It might, however, correspond to a mutation in an undetected prophage. Indeed, the phenomena of transduction and the presumably close relations between the prophages and the genetic apparatus of their host cells make it difficult and possibly meaningless to distinguish between genotype-con- trolled and prophage-controlled properties of a bac- terium. The influence of the prophages carried by various Vi-strains of S. typhosa on the reaction of these strains to the phages of the Vi-group II might be due either to phage genes or to host genes transduced with the latent phages. The study of successive and multiple transformations induced in Vi-strains by lysogenization may provide some of the answers. THE MECHANISM OF PHAGE MODIFICATIONS BY THE. Host CELL We have tentatively concluded that host-induced modifications determine the ability or inability of a phage to perform some specific critical step of 2 The actual ability of an artificially produced strain T to modify phage II A into II T has not yet been tested (Anderson, personal communication). 242 S. E. LURIA interaction with one or more hosts. When different modifications can be impressed on a phage by dif- ferent hosts, these modifications are mutually ex- clusive rather than additive. These facts suggest that the same phage structure, needed for the re- stricted step, is altered in two or more alternative ways in different hosts. Several phages have been shown to consist of genetic and nongenetic mate- rial, the latter including at least the protein skin of the phage and possibly also some of its nucleic acid, Since the host-induced modifications are non- hereditary, we may incline to attribute them to changes in the nongenetic portions of the phage. This may be unjustified, however. As pointed out by Bertani (personal communication), the genetic portion of a phage might be so modified (although not intrinsically mutated) by its intimate relation with the genome of a host as to be unable to estab- lish successful connections with the genome of a different one. The fact that some modifications are adaptive (for example, the change P2 Sh —> P2 B, which extends the growth ability on B) and some unadaptive (for example, T2 —» T*2 on B/4,) might reflect different forms of nuclear inter- actions. It seems desirable to investigate thorough- ly the stages at which the development of restricted phages is arrested in a variety of cases, in order to , understand the role of various phage structures in phage development and to localize the modifying ability of the host on any one of the phage struc- tures. Tue Revation or Host-INDUCED VARIATION TO Otuer Forms or Non-HEREDITARY CHANGES IN BACTERIOPHAGE Apart from transitory changes in the particles that survive certain treatments, such as with ultra- violet light or antisera, two types of nonhereditary modifications have been described in phage, besides host-induced variation. Phenocopies of heat-stable mutants of phage T5. Lysates of T5 and of its relatives contain heat-re- sistant particles, which upon growth give rise to regular TS phage; these phenotypically heat-stable particles are phenocopies of stable mutants of the same phages (Adams and Lark, 1950). In phage T5, the mutants appear with a frequency of about 10-7, the phenocopies with a frequency of about 10-%. Adams (1953) reported that in the yield of in- dividual bacteria infected with phage T5 the pheno- copies are not distributed at random but are grouped clonally. When present, they constitute a minority portion of the yield. This suggested that. the phenocopies might be formed in response to “local conditions” or to a modified “template or pattern” (not a phage-genetic one) in some of the bacteria. The further suggestion was made that the local conditions may have to do with some bio- chemical irregularity within individual host cells. This suggestion would relate the production of the heat-stable phenocopies to host-induced modifica- tions. There is a basic difference, however, between. the production of heat-stable phenocopies and the established cases of host-induced variation. In these, the modifying influence of a host cell is ap- parently uniform on all the phage particles that cell liberates. As shown in Table 5, the heat-stable phenocopies of T5 in individual T5 bursts are grouped clonally TaBLe 5. THE DISTRIBUTION OF PHENOTYPICALLY Heat-Resistant Pace Partictes oF T5 Data from Adams (1953) analyzed according to Luria. (1951). Number of samples Expected according to the re- Resistant particles duplication hypothesis (from per sample Found the number with 1 or more) 1 or more 30 _ 2“ « 18 a) 3“ * 12 10 4“ * 8 15 5 ee oe 5 . 6 6 “ “ 3 5 10. “ce oe 1 3 20“ * 0 1 and the frequency distribution of clone sizes is very close to the one expected for groups of identical sibs arising -by reduplication of randomly mutated individuals (Luria, 1951). Thus, we are led to an: alternative hypothesis. We assume that the pheno- typic change in heat stability arises, like a muta- tion, spontaneously and randomly in individual phage particles during reproduction, and is trans- mitted from parent to daughter particles within the cell of origin, but is not transmitted to the progeny of the heat-stablé particles when they later repro- duce in other bacteria. The occurrence in the same phage of similar but permanent mutations (as much rarer events) strengthens this conclusion. We sug- gest that the phenocopies are due to a_ genetic change in a portion of phage material that, although “self-reduplicating” in the bacterium of origin,’ is not utilized as a model for reproduction in later cycles of multiplication in other bacteria. The identification for this “transitorily genetic” struc- ture in phages of the T5 group awaits further evi- dence. Phenotypic mixing in mixed infection. This con- sists of the production, in cells of E. coli B in- fected with’ phages T2 and T4, of particles that, like T4, can attack bacteria B/2 but that give rise to a pure yield of T2 (defined by heritable ability to grow on B/4 and inability to adsorb on B/2) (Novick and Szilard, 1951). Similar particles with mixed phenotype are probably formed also in mixed infection with T2 and T6 (Delbriick and Bailey, 1946). Hershey (unpublished) found HOST INDUCED MODIFICATIONS OF VIRUSES 243 phenotypic mixing between T2h and T2h--. Strei- singer (unpublished) found that in mixed infection with backcross strains of T2 and T4, especially selected to minimize mutual exclusion, all progeny particles are genetically either T2 or T4 in host range. Both groups include particles of three pheno- types: (a) adsorbed by B/4 only; (b) adsorbed by B/2 only; (c) adsorbed by B/2 and by B/4. All features of phenotypic mixing can be ac- counted for by modifications of the protein skin of the phage. The restricted step, when present, is adsorption; whenever adsorption occurs, growth follows. Genotypic T2 with T4 phenotype is neu- tralized equally well by anti-T2 and by anti-T4 sera, which act on the protein skin (Delbriick, unpub.). The mechanism of phenotypic mixing is un- known. The host-range specificity of the phage skin in particles from mixed infection may be de- termined by a complex between phage nucleus and some accessory genetic material of the other phage type that fails to appear in the progeny. Alterna- tively, the host range specificity of the phage skins might be influenced by many phage genomes through interactions in the host cell at the level of the synthesis of the “adsorption sites” of the phage. Formally, phenotypic mixing does not resemble host-induced modifications; the two types of altera- tion apparently modify different phage functions and probably also different phage structures, Phen- otypic mixing reflects interactions, at the phage phenotype level, among genetic materials of several phages. Host-induced modification reflects inter- actions, at the phage phenotype level, between phage and host genotype. The tie-up between the two phenomena, if any, may reside in some gen- eral pattern of interactions at the genetic level in phage-iniected cells. Host-InpuCED MopDIFICATIONS IN VIRUSES OTHER THAN PHAGE The essential characteristic of host-induced modi- fication is the complete transformation of one form of virus into another upon a single cycle of intra- cellular growth in a modifying host. Experiments on animal and plant viruses seldom permit observa- tion of virus after single cycles of growth; a search of the literature reveals no certain example of re- versible changes attributable to host-induced modi- fication. The numerous instances of host adapta- tion or tissue adaptation in animal viruses corre- spond probably to a selection of host-range mutants. The variation generally appears, gradually or sud- denly, after several passages in a new host and persists after return to the previous host. Rever- ston, when observed, has never been shown to occur by a single-cycle mass transformation. Variation detected after a single animal passage can be due to selection of mutants, since one passage Corre- sponds to many repeated intracellular growth cycles, especially if only a small fraction of the virus inoculated can multiply. Yet, the existence of host-induced modification in phage suggests that variation in other viruses should be reexamined experimentally in the light of this new phenomenon. In doing so, the following properties of host-induced modification, as observed in phage, should be remembered: 1. It affects the totality of the virus exposed. 2. It may either extend or restrict the host range of a virus, 3. The restriction in host range may concern either the same type of host cell which induces the modification or another type. ‘ 4. The modifying ability of the host cell can be affected by its growth stage (which, more gener- ally, corresponds to its developmental history). 5. The ability of a host to accept (hence, to reveal or unmask) a restricted virus may itself de- pend on the developmental stage of the host cells. 6. Host-induced modification of the adaptive type (see Table 1) can simulate selection of adapted mutants; host-induced modifications of the non- adaptive type (see Table 3) can simulate the pro- duction of noninfectious or masked virus. A single-cycle change in an animal virus is the formation of hemagglutinating, noninfectious par- ticles, following injection of nonneurotropic influ- enza virus into the mouse brain (Schlesinger, 1950) and possibly in other host tissues as well. This change is certainly a host-induced modification; formally, it can be homologized with the change T2 ——> T*2 induced by B/4,, assuming that only hosts B and B/4, are known (Table 3). That is, the “noninfectious” virus might be virus restricted in growth ability and might appear to be nonin- fectious only because no unrestricting host is avail- able. In the absence of any evidence, however, that the modified influenza virus from brain can grow in some other host, it seems more reasonable to consider it as incomplete, unfinished virus rather than as infectious virus with a restricted host range. Some cases of virus masking (Shope, 1950) may also seem to be formally analogous to nonadaptive host-induced modifications. Yet, in the classical cases of masking, for example, with rabbit papil- loma virus in domestic rabbit, virus particles seem to be few or absent; masking may reflect partly the small amount of virus present (Friedewald and Kidd, 1944), partly the presence of modified, non- infectious virus antigens. Rabbit papilloma virus has occasionally been maintained by repeated passages in the domestic rabbit (Shope, 1935). One such domestic rabbit- adapted strain lost its adaptation in a single passage in cottontail hare (Selbie et al., 1948). The condi- tions did exist for selection in the cottontail of a better growing variant, but the occurrence of a pair of host-controlled forms cannot be excluded. 244 S. E. LURIA Another field where host-induced modifications may play a role is exemplified by the determination of the tissue affinities of amphibian tumors derived from the virus carcinoma of the Vermont leopard frog (Rose and Rose, 1952). Passage of tumor cells from frogs to salamanders to frogs of different races, and from one organ to another, induces in the tumors (and presumably in their viral agent) a remarkable series of changes. The extreme speci- ficity of some of these changes is illustrated by the significant coincidence of exactly bilateral peri- osteal tumors. It is conceivable that by growth in. given host cells the tumor-causing agent may be- come, either specifically restricted to developmen- tally characteristic cells, or specifically prone to attack such cells. Unfortunately, methods for pre- cise quantitative work with these tumor agents still have to be developed. In summary, host-induced modification as ob- served in phage has not been demonstrated with animal or plant viruses. This does not constitute evidence against its occurrence since observations suitable for its detection have not been made. In view of the great importance that this type of vari- ation, if present would have in virus ecology and in the epidemiology and pathology of virus diseases, efforts to determine its range of existence appear desirable. REFERENCES Abas, M. H., 1953, The genotypically and phenotypically heat resistant forms in the T5 species of bacteriophage. Ann. Inst. Pasteur 84:164-174, Apams, M. H., and Lark, G., 1950, Mutation to heat re- sistance in coliphage T5. J. Immunol. 64:335-347. Anperson, E. S., and Feurx, A., 1952, Variation in Vi-phage II of Salmonella typhi. Nature, Lond. 170:492-493, 1953a, “Degraded Vi strains” and variation in Vi-phage II of Salmonella typhi. J. Gen. Microb. (in press). 1953b, The Vi-type determining phages carried by Sal- monella typhi. J. Gen. Microb. (in press). Bertani, G., and Weicte, J. J., 1953, Host controlled varia- tion in bacterial viruses. J. Bact. 65:113-121. Burner, F. M., 1946, Virus as organism. Haivard Univer- sity Press, Cambridge, Mass. Cratcie, J., 1946, The significance and applications of bac- teriophage in bacteriological and virus research. Bact. Rev. 10:73-88. Craiciz, J.. and Yen, C, H., 1938, The demonstration of types of B. typhosus by means of preparations of type II Vi phage. I. Principles and techniques. Canad. J. Pub. Hth. 29:448-463. Dexsriicx, M., and Battey, W. T., Jr., 1946, Induced muta- tions in bacterial viruses. Cold Spr. Harb. Symp. Quant. Biol. 11:33-37. Feiix, A., and Anverson, E, S., 1951, Bacteriophage, viru- lence and agglutination tests with a strain of Sal- monella typhi of low virulence. J. Hyg., Camb. 49:349- 364, Finptay, G. M., 1939, Variation in viruses. In: Doerr and Hallaver, “Handbuch der Virusforschung,’ vol. II. Springer, Vienna. Frepericg, P., 1950a, Une mutation de bactériophage in- téressant son spéctre d’activité. C. R. Soc. Biol. 144: 1140-1142, 1950b, Mutations intéressant le spéctre d'activité du bac- tériophage Tl. C. R. Soc. Biol. 144:1258-1260. Luria, S. E., 1951, The frequency distribution of spon- taneous bacterial mutants as evidence for the exponen- tial rate of phage reproduction, Cold Spr. Harb. Symp. Quant. Biol. 16:463-470. Luria, S. E., and Human, M.’L., 1952, A nonhereditary, host-induced variation of bacterial viruses. J. Bact. 64:557-569, Novicx, A., and Szitarp, L., 1951, Virus strains of identi- cal phenotype but different genotype. Science 113:34- 35. Ratsron, D, J., and Kruecer, A. P., 1952, Phage multipli- cation on two hosts. Isolation and activity of Staphylo- coccus phage Pl. Proc. Soc. Exp. Biol, N. Y. 80:217- 220. Rose, S. M., and Rose, F. C., 1952, Tumor agent transfor- mation in amphibia. Cancer Res. 12:1-12. ScHLesincer, R. W., 1950, Incomplete growth cycle of in- fluenza virus in mouse brain. Proc. Soc. Exp. Biol., N. Y. 74:541-548. Sexsig, F, R., Rozmson, R. H. M., and Snore, R. E., 1948, Shope papilloma virus: Reversion of adaptation to domestic rabbit by passage through cottontail. Brit. J. Cancer 2:375-380. Suope, R. E., 1935, Serial transmission of virus of infec- tious papillomatosis in domestic rabbits. Proc. Soc. Exp. Biol., N. Y. 32:830-832. 1950, “Masking,” transformation, and interepidemic sur- vival of animal viruses. In: “Viruses 1950,” M. Del- briick. California Institute of Technology Bookstore, Pasadena. Zinper, N. D., and Leperserc, J., 1952, Genetic exchange in Salmonella. J. Bact. 64:679-699.