Reprinted from GENETICS Vol. 41, No. 5, September, 1956 Printed in U.S.A. PHASE VARIATION IN SALMONELLA! JOSHUA LEDERBERG axp TETSUO IINO Department of Genelics, Universily af Wisconsin, Madison, Wisconsin Received April 8, 1956 HE flagella of most Salmonella serotypes occur in two alternative, clonally stable phases. Earlier applications of transductional technique have shown that the phases correspond to two loci, 7; and 7; (LEDERBERG and Epwarps 1953). The present study is concerned with the internal control of the alternation. 1, HISTORICAL BACKGROUND 11 Serological analysis of Salmonella dates to the differentiation of antigenic components in motile and nonmotile strains of S. cholerae-suis by Siri and Rracu (1903). They and subsequent authors have classified these components as an “H’” thermolabile, flagellar and an “O” thermostable, somatic antigen respectively. Fur- ther analysis has proceeded by the now familiar techniques of cross-absorption of antiserums prepared against various strains, and has led to the elaborate tabulation known as the KAUFFMANN-WHITE diagnostic scheme, part of which is abstracted in table 1. The scheme is an important guide in epidemiological tracing of Salmonella infections, and newly discovered types are therefore promptly recorded and deposited with International Salmonella Typing Centers. The cooperative efforts of hundreds of bacteriologists therefore make a wealth of information and natural variants available for genetic study. 12 The most striking feature of the scheme is its combinatorial pattern— prima facie evidence for a mechanism of genetic recombination which encouraged even- tually successful efforts at its demonstration (LEDERBERG 1947, 1948; ZINDER and LEDERBERG 1952). The serotypes entered in the table can be construed as generated by the combination of three basic elements, the somatic antigen complex and the two flagellar phases. As we will not be immediately concerned with the former it will be symbolized only by the conventional groups, rather than detailed formulae. However, such abbreviations denote antigenic patterns whose genetic interest is no tess than that of the flagellar factors (cf. UETAKE ef al. 1955). 1.3 Flagellar phase variation was discovered by ANDREWES (1922) in S. lyphi- murium. Mass cultures of this serotype may agglutinate in either of two H-antisera, anti-i or anti-1.2. A reisolated clone might react only, say, with i. On further cul- tivation, the i clone will generate an occasional offshoot which reacts only with 1.2. In its turn, the 1.2 phase is liable to throw off an occasional i, an alternation which will be symbolized i:1.2. The variation from one phase to another may occur at any ‘Paper No, 618 of the Department of Genetics. This work has been supported by research grants from the National Science Foundation, [rom the National Cancer Institute (C-2157), Public Health Service, and from the Research Committee, Graduate School, University of Wisconsin with funds aocated by the Wisconsin Alumni Research Foundation, The numbering of paragraphs in this paper is intended to facilitate internal cross-reference and should otherwise be “ignored. 744 J. LEDERBERG AND T. INO time, with a small probability per bacterial division, which ranged from 107* to 1073 in the strains studied by Stocker (1949). In rapidly varying strains, phase variation can be detected by random colony tests; in more stable material, selection by the immobilizing action of antiserum against the existing phase is effectively practiced. 1.1 Whileits stochastic indeterminacy tends to signify phase variation as mutation, it is bounded by two fixed alternatives which are characteristic of the serotype. As shown in table 1, any phase-1 antigen may be associated with any phase-2, and the one cannot be predicted from the other. The table also shows that the flagellar antigens can be divided into two groups, viz., phase-1: a, b, c, d, eh, i, r, ete., and phase-2: 1.2, 1.5, 1.7, enx, etc. One antigenic alternative is chosen from one group, one from the second. This pattern of variation has no parallel in the mutation of other genes, and until recombinational analysis was applicable, genetic interpretations of phase variation could only be speculative (Dupos 1945; LEDERBERG 1948), is The genetic relationship of the phases in a wide variety of serotypes has been investigated by transductional techniques (LEDERBERG and Epwarps 1953). For example, the transduction from S. typhimurium, i11.2, as donor, to S. abeny, brenx as recipient gave the new combinations i:enx and b: 1.2, but not such types as itb or TABLE 1 The KAUFFMANN-WHITE scheme of Salmonella Seralypes, condensed from Kaurruanxn (1951 | : : | Flag- | Ao : Flag- . Salmonella serotype “antigen! Ree ' Antigens Salmonella serotype jantigen these ph ns I 1 a ar ae ij . po kisangani | B a | 4.2 | manhattan C2 d 1.5 arechavaleta ; Boja 17 newport » C2 eh 1.2 bispebjerg | B ; a i enx kottbus ! C2 + eh 1.5 tinda ' Bo' ai enas takoradi C2 i 1.5 abortus-equi . B | enx bonariensis C2 i cnx paratyphi B |B ib ft 1.2 bovis-morbificans C2 r 1.5 abony Boob ft ens hidalgo C2 r ena; wagenia ' B ho | enzys sencai 1D a 1.5 altendor! i; Boy c¢ y 4.7 4} loma-linda 1D a enx stanley [ B i a | 1.2.) || onarimon D h 1.2 saint-paul B eh | 1.2 typhi 1D d chester Boi eh | enx ndolo D d 1.5 essen Bo: gm a | eastbourne D ¢ ch 1.5 typhimurium | Boy t 1.2 enteritidis D gm heidelberg 7 Bot r i 4,2 shoreditch dD T eNZ45 san juan Ch} oa} 1.8) 4 butantan i h 1.5 oslo ' CL loa | enx — |} shangani i d 1.5 edinburg C1 b 1.5 | elisabethville El r 1.7 cholerae-suis Clic 1.5 | marseille F a 1.5 mission C1 od 1.5 |. chandans F d enx lomita Ct: eh 1.5 | aberdeen F i 1.2 montevideo Ct | gmsi -- | veneziana r i enx virchow Chior 3 2.2 3 Mississippi G b 1.5 narashino C2 a enx heves H d 1.5 gatuni C2 h enx 4) hvtttingfoss I h onx muenchen C2 d 1.2 PHASE VARIATION IN SALMONELLA 745 1.2:enx, which would have been as readily detectible by the selective method. These and many analogous results imply that the phase-1 antigens form one genetic homology group, the phase-2 a second. The factors within each group are mutually interchangeable by transduction, while factors of different groups are not. 16 Ina word, each group is the set of alleles of an H, and H; locus, respectively, so that the phenotype b:enx corresponds to the genotype H°H%$"". Furthermore the two loci are independently transduced, i.e., they are not closely linked to one another for the i:1.2 class was also not produced. If the two antigenic phases are thus deter- mined at separate loci, phase variation cannot be construed as the mutation from one specific allele to another. Instead, it is the alternative manifestation of each of the two antigenic specificities already inherent in the genotype. This concept was already provoked by the closed cycle of variation emphasized above. 1.7 To investigate this alternation, experiments were directed primarily at the inheritance of flagellar phase in recombinant clones generated by transduction. Where the phase of a culture was determined, it will be indicated by bold face: thus, bienx signifies a clone displaying the enx antigen that varies to the b:enx form, in which the b antigen is now manifest. is Recombinational analysis in Salmonella (so far) depends on a system of trans- duction (the transmission of hereditary fragments) mediated in this instance by bacteriophage particles (ZINDER and LepERBERG 1952). The leading work on this system involved a variety of nutritional, fermentative and resistance markers, and was followed by studies on mutants affecting the formation and action of flagella (STockER ef al. 1953) and on the flagellar antigens (LEDERBERG and Epwarps 1953), Every marker that was studied and was technically feasible has been subject to transduction at a more or less uniform incidence (cf. ZINDER 1955, for quantitative comparisons of transduction efficiency of different factors). The symbolism A- -x B or B x--- A will be used to indicate a transduction with A as donor, B as recipient, i.e., the application of a cell-free lysate of A to intact cells of B. 2. MATERIALS AND METHODS 21 Transduction was carried out with lysates of Salmonella phage PLT 22 grown on indicated host cultures as previously described, 1.8 22 Two modes of selection were used to isolate the minority of transduction clones from the excess background of unaltered recipient cells. Both methods take advantage of a soft nutrient gelatin agar (NGA) through which motile bacteria readily swarm, while nonmotile or immobilized cells grow in place. (A), Mixtures of bacteria and phage were inoculated on NGA plates to which antiserum for both phases of the recipient bacteria was added. These are therefore immobilized, but any bacteria which manifest a different H antigen, viz., one transduced from the donor, swarm from the point of inoculation. For example, outgrowing swarms of b:1.2 and i:enx from bienx x it 1.2 can be picked twelve to twenty-four hours after NGA plates containing 1 and 1.2 antiserum are inoculated. (B), Alternatively, a Fla~ (flagellaless, nonmotile) recipient was used in NGA without antiserum. Individual Flat swarms were then typed for their flagellar antigens. The application of this procedure to the 746 J. LEDERBERG AND T. ILNO present problem depends on the demonstrated linkage of Fla and H; factors (STOCKER ef al, 1953; 3.43.7), 23 Serotypes were characterized by routine methods (Epwarps and EwrncG 1955) of slide agglutination in 1:100 antiserum. Live cell suspensions were emulsified from colonies or used directly from penassay broth cultures. A number of the serotypes have been confirmed by tube agglutination tests in several laboratories. When the alternative phases were not both obvious on routine tests, they were ascertained after selection in NGA with antiserum against the existing phase. Most of the serums were standard reagents, ie., sterile cresolated preparations of rabbit antiserums that had been absorbed to remove somatic and heterologous antibodies, with a tube agglutination titer of at least 1:10,000. They were used at a dilution of 1:1000 for selection in NGA. 24 The principal markers for this study were the naturally occurring alleles of H, and H> in the following stock cultures (StockER e/ al. 1953; LEDERBERG and Epwarps 1933): S. typhimurium, TX12, i:1.2; S. abony, CDC-103, btenx; and S. heidelberg, SL-28, Fla~ r:1.2. Gal- (galactose-negative) mutanis of S. fyphimurium and of S. abony were obtained after UV-irradiation and scored on EMB galactose agar (LEDERBERG 1950). Additional //a~ mutants were obtained in S. typhimurium from UV-irradiated suspensions by plating them in NGA at a density of one to two hundred survivors per plate. The plates were inoculated three to six hours at 37°C to allow motile clones to initiate small swarms. They were then held an additional twelve to eighteen hours at 25°C at which temperature the gelatin would set and restrict further swarming. However, the clones could continue to grow to an easily visible density in situ. The Fia~ mutants were then readily distinguished as compact spherical colonies, in contrast to the diffuse nebulae of the parental #/a* clones. Eight Fla~ mutants were isolated, of which five were sufficiently stable to be tested further. 3. RESULTS AND CONCLUSIONS 3.1 Serum selection, mixed recipients. The leading precedent for alternative mani- festation of serotypic potentialities is undoubtedly the study of Paramecium aurelia by SoNNEBORN and his coworkers (BEATE 1954) whereby the cytoplasm was shown to be propagated in one of several inter-transformable states which determine the pattern of genic expression. If a principle of complete cytoplasmic determination were applicable to Salmonella phase variation, the phase should be inherited entirely from the recipient in transduction, in view of the disproportionate mass of the intact cell in comparison with that of the phage-borne fragment. The first experiments were therefore designed to establish whether the donor phase had any influence on the issue of transduction. Phage lysates were prepared from single colony cultures of S. typhimurium G:1.2) whose purity of phase was controlled by slide agglutination tesis on twenty or more colonies. All cultures used in these experiments consisted of at least ninety percent of cells in the indicated phase, and were more often entirely homogeneous over the sample of twenty subclones. For the recipient, an equal mixture of bienx and bienx cells of S. @bony was used so that the only variable would be the phase of the donor. The phage and cells were mixed and planted on NGA agar plates containing b and enx antiserum, and the resulting swarms isolated PHASE VARIATION IN SALMONELLA 747 and characterized as follows: Total swarms isenx b:1.2 tested a) i:1.2 —x btenx 19 0 19 b) i:1.2 —x btenx 14 14 28 For this 2 X 2 table, x? = 15, and P < .01. Similar results were obtained in repeti- tions of this design, and it was concluded that the donor phase does influence the results of transduction, contrary to the cytoplasmic hypothesis. 32 The absence of b:1.2 types in experiment a) above might be explained by the inherent nontransducibility of Hi” from phase-1 cells in which it is not expressed, or by the persistence of its inactivity in the recipient. On the latter view, the transduc- tion of H3° would result ina b:1.2 recombinant, which would not swarm, however, due to its inhibition by the anti-b serum. That is, in these experiments, a new geno- type will only be detected if it inherits the appropriate phase as well as antigenic potentiality. Some more direct substantiation for this assumption will be offered in the following sections. 33 Motility selection: H, linked to Fla. From the influence of the donor phase 31 it can be concluded that the control of phase is associated with the transduced fragments, i.e., that it is linked either to H; or Hz or both. The following experi- ments were designed as further tests of the assumptions of 3.2, by studying the inheri- tance of phase in transductions of H, as an unselected marker linked to a selected Flat factor. 34 The first of these experiments used SL 28, the naturally occurring Fla~ mutant of S. heidelberg, Fla~ r:1.2. Phase variation in this culture cannot be detected by phenotypic difference, as no flagella or flagellar antigens are produced, but can be determined after Fla+ has been restored either by (rare) reversions or transduction from another strain. Preliminary trials of S. abony —x SL28 were made on fresh single colony cultures, and two of them were saved when their motile progeny were preponderantly r:1.2 and r:1.2 respectively. Each phase was then used as a re- cipient with lysates from S. abowy, bienx and b:enx respectively, according to the procedure of 2.28 with results shown in table 2. 3s Columns 1 and 2 indicate the transductions of Flat uncoupled with HW? and verify the stability of the two phases in the recipient. Columns 3 and + show the transductions in which Flat was coupled with H? . This was immediately discernible TABLE 2 Transduction of Hr coupled with Fla* S. abony Vla* b:enx —x S. heidelberg Fla7 r:1.2 Phase of transductional parents Flat transductional clones { | | : 7 Donor , Recipient | 2), ff 2 | pe’s ove ; Total tested |_— _ | | b:enx rl.20 | 21 0 | 2 0 | 43 bienx 1.2 | 0 7 | 1 2 | 50 bienx rt. | 11 1) 30 0 AM bienx r:1.2 | 1 10 | 1 38 50 748 J. LEDERBERG AND T. IINO for b:1.2 types, and could be determined after the alternative phase was scored in b:1.2 vs. 1:1.2 types. The b:1.2 type is especially important as it verifies directly that a b phase can be transduced in an inactive form. In the experiments conducted by antiserum selections (31, 310), such recombinant clones would be suppressed. 36 Table 2 also shows that the recipient governs the yield of types, and the donor phase has no effect. This apparent conflict with the previous results can be resolved by recalling that in 31 both H, and i, transductions were enumerated, here only #,. There is also an unexplained difference in the frequency of linked and unlinked transductions in the different combinations. It is not known whether the discrepancy is systematically related to the phases of the donor and recipient, or is simply a casual peculiarity of a single lysate. No such effect was noted in the comparable experiment of table 4. 3.7 To complete an experimental design similar to 34 and table 2 in the more closely studied system S. abony —x S. typhimurium, a number of Fla~ mutants of the latter were examined for linkage to H, and H,. Each Fla~ mutant was tested x—- S. abory b:enx and b:enx, and Fla* progeny examined for the presence of the b and enx anti- gens. Three of the eight //a~ mutants reverted too readily to be useful in these experiments. Of the other five, one, SW-1157 showed occasional coupling of Flat with H7 (table 3). None showed linkage of Fla with H,. The Fla~ mutants were also tested against one another and with a number of other Fle~ stocks and found to transduce motility to one another in all combinations, i.e., they carry Fla~ mutations at distinctive loci (cf. Stocker ef al. 1953). TABLE 3 Linkage relations of Fla~ midants from S. typhimurium i:/.2 | Fla* selections Mutant | Fla~ x-~ Fiat brenx tests Fla-1, linkage . Fla~ x— Flat bienx tests Fla-H> linkage b:1.2 i: 1.2 | isenx 2 — - — _ . a 1 0 ; 39 | 0 16 2 0 | 48 : 0 28 3 0 12 | 0 7 4 0 | 35 | 0 15 5 (SW-1157) | 5 40 : 0 28 I _ _ _ _ TABLE 4 Transduction of BR coupled with Flat S. abony Flat b:enx —x S. typhimurium SI £1357 Fla® i:1.2 Phase of transductional parents Fla* transductional clones Danor Recipient hy ae 2) | bide | Total tested . ! : — poy bienx = t2.2 | 80 ee 0 | 95 bienx EB 3 | 86 0 27 / 16 )tenx i:1.2 i 68 4 18 0 : 90 bienx \ 1.2; 1 | 1 90 0 2 Ah PHASE VARIATION IN SALMONELLA 749 ss In a further search for technically useful linkages, a series of twenty-six auxo- trophic markers (DEmMeREC, BromMsrRAND, and Demerrc 1955) was tested for linkage to the Mla-H, segment. Flat selections from the various wx Flat —x lux? Fla~ transductions were tested, but each of 1352 isolates was still + (proto- trophic), implying that none of these markers is linked to Fia. 39 The experiment outlined 34 was repeated with S. abony —x SW-1157, with entirely analogous results, except that the proportion of linked /7?-Flat transductions was uniformly lower (table 4). 310 Serum selection, labelled recipients. The next experiments were a study of all four combinations of donor and recipient phase under serum selection. In order to improve the comparability of the experiments, the recipient was a mixture of phases, distinguished by a Gal marker. That is, from a Gal~ mutant of S. typhimurium, TABLE 5 Transductions of S. abony and S. typhimurium Serum-selection; Gal-labelled recipients i:1.2 —x brenx; b, enx serum | brenx — i:1.2; i, 1.2 serum Recipient ; Transduction clones | | Recipient | Transduction clones Donor en —_ —-—_______.. | Donor foo - Phase | Gal Gal | i:enx | bi2 | | Phase | Gal Gal | b:1.2 i:enx = “| —_ — ———_- | bienx + + | 15 | 0 i:1.2 + + 10 0 i:1.2 | benx bienx| — — 1' 0 rhea.) —- | —- | o | 9 oe ' ee J _ a a | brenx | + | + 6 | 8 fin) 4) 4 4 | 16 i125 brenx | | brenx| — | - 2 11 | 1.27 —- = 0 | 34 anne a eee - = brenx | — | = 10 0 | i:1.2 _ _ 16 | 1 is1.2 | | | bienx - | bienx,) + =; + 0 0 12.2 | + + 0 | 0 ee wel — aoe | 7 we no I brenx| — — |) 10 7 Fatal - | - 6) 19 1.2 | | | brenx brenx |p | + | 1 8 lite | 4 of 0 29 TABLE 6 Summary of all experiments similar to table 5 | Relative incidence of transduction clones | . - Donor | Recipient | Table 5 All experiments \ ; | | dyira ryidg | dy:r2 | ryidg . me I: i ee a _ dys Tits | 51 1 127 7 dy ids T1272 | 0 3 2 cide Tite | 26 | 50 108 155 aida nitz 3 82 3 185 750 J. LEDERBERG AND T. IINO colonies in either i: 1.2 or i:1.2 phase were picked, and the pure-phase culture mixed with the alternative phase from the Gal* strain. The mixed cells were then treated with phage from S. abowy and inoculated on NGA plates with i and 1.2 serum. The resulting transduction clones (either i:enx or b:1.2) were also scored for their Gal character. An analogous experiment was also conducted with the reciprocal trans- duction. Previous trials had shown no indication of linkage between Gal and the H markers, so that the Gal character of a new combination told the phase of the parental recipient for each transductional clone. The results are given in detail in table 5. In table 6, the corresponding results from a number of similar experiments are pooled, di:dz signifying the generalized donors, and rm:re the recipients. The qualita- tive features of table 6 may be summarized as follows: Manifest Probable missing Donor Recipient transductions classes a. diide Tits ditre Tiida . daids riz ride. dyirz c. diid2 ryiry ridz dyire d. d,:d2 rire rida dyir2 311 In this summary, the rare entries are disregarded, and may be considered to arise from a minor admixture of phase in the parents. The r1:de types seem also to outnumber the dy:rz2. This discrepancy might reflect a minor difference in the efh- ciency of transduction of the ZZ; and LH, loci by the lysates used, or in the recovery of the two types under the conditions of selection. As indicated earlier, the missing classes are presumed to occur but are suppressed because they retain the antigenic phase of the recipient parent, for which antiserum is present in the NGA. The excess unaltered parental cells restrict the total growth of the transduction clones at the site of inoculation so that they lack an opportunity to engender phase variants that can be detected. The occurrence of the missing di!rg type was verified in previous experiments, 3.5, 39; the rd» is hypothetical so far, owing to the lack of any means of selecting for it (e.g., another marker linked to 2). 3.12 Hypothetical expectations. A number of hypotheses may be considered for their concordance with the experimental results and are listed in table 7. The only hy- pothesis among those listed that agrees with the data is number 5, that the “‘state”’ of the H, locus (or one closely linked to it} governs the phase. It is difficult to estab- lish whether the list is exhaustive, but the following restatement may help to expose the implicit assumptions. Since the donor phase does influence the result of trans- duction of the H factors, we assume that some element associated with the factors themselves is involved. If the phage particle carries no other relevant material besides chromosome fragments, the state of the fragment itself is in question. Of the relevant possibilities, the experimental results are clearly in favor of a decisive role of the 4 locus. This hypothesis is amplified, and the possible coordinate role of cytoplasm and other factors is discussed in 41, 3.13 Consistency of the experiments. The earlier experiments can be considered as special cases of 3.10, 3.1 can be derived by lumping the two phases of the recipient, and is qualitatively consistent with it: compare the proportions 19: 0: 14: 14 with the corresponding 130: 9: 111: 340. Tables 2 and 3 can be correlated by considering PHASE VARIATION IN SALMONELLA 751 TABLE 7 Hypotheses of phase variation and their consequences Genotype of Types expected from Hypothesis dy:dg —x ry:rg | dy:d2 —x ry:rg | dy:dg —x ryirg | dy:dg —x ry:re Ty:r2 ryirg — a aenEnan _ 1 2 1 2 1 2 1 2 1. Cytoplasmic state) Hilfay!| HiWay?| ditrs | -ri- | -re- | mtda} dizm | -ri- | -re- | ritde 2. Controlling gene ALASG| WVH3G?| dasr. | -ri— | -re- | mid2| dgtre | -ri- | —r2- | miid2 3. Complementary | H4ls HiHS | ditt} —ri~ | dgire*| ritdat] dyzret! ritda*| -ro- | ritda states of HT, —T2- y -ry- | -re- | -ry- and > 4. State of H, Huw, | AiHs | datre| -ti— | dyin | mtd2| -re- | -ry- | -re- | ritda 3. State of M2 AYA} Hi’, dautre} -ri- | -re- | -ra- | datre | rida | -ra- | risd2 Iexperimental results Table 6 datre | ritdg | datre| rptd2! dgim | mid2] dyin | ritdz 127 7 3 2 108 | 155 3° 185 The symbols ~r,- in columns 2 and -r,- in columns 1 stand for r,:dz and di:r, respectively. In experiments like table 6, which are done in the presence of r, and rz antiserums, these types would not be recovered. The various hypotheses are defined by the genotypes which each assigns to the two recipient phases. The corresponding donor genotypes may be written, ex /irypethesi, by substituting the @ for the r allelic symbols. The expected types are inferred by substituting i? in the indicated state for Hi in columns 1, and 1% for JI} in columns 2. The phenotype of each transduction is then deter- mined from the postulated genotype. {* Auxiliary assumptions are required to make predictions from this hypothesis when two active* or two inactivef loci are combined by transduction. The table shows expectations when the choice of phase from this impasse is random. On other assumptions, the results of the two tests should be similar, unless the hypothesis is reduced to a case of No. 4 or 5 by assuming that one locus out- weighs the other. only the relative incidences of dy:rz in table 6, and ignoring the transductions of HM, listed in the ry: dz column; then compare 22: 1: 30: 1 and 12:0: 13: 0 with 127: 3: 108: 3, respectively, which constitute excellent agreement on the principal issue of the role of the recipient (viz. the ZZ, locus) in the activity of transduced H, factors. 3.14 Monophasic types. Although diphasic behavior is more prevalent, a number of Salmonella types are monophasic, e.g., S. typhi d:—; S. enteritidis gm: —; S. abortus- equi —:enx, Such types are especially valuable for the production of specific serums, and have been sought after for that reason (Epwarps and Bruner 1946). Curiously, the natural types carrying g and related antigens are almost always monophasic, despite the fact that diphasic combinations of these factors, e.g., gp: 1.2 are readily synthesized by transduction. The ecological selective factors that must be involved in this peculiarity are unknown. 313 Apparent monophasic behavior may be complicated by serological cross- reactions between the phases. For example, S. cholerae-suis, var. kunzendorf is recorded as ——!1.5 or (c):1.5, as selection with anti-kunzendorf 1.5 serum is occa- sionally but sporadically successful for the demonstration of the c phase (EDWARDS and Bruner 1946). However other 1.5 and 1.2 serums are more consistently effective. ~I wv bo J. LEDERBERG AND T. IINO The different activity of different serums might be taken as evidence that they have a specific inductive as well as a selective effect on phase variation. However, the c and 1.5 phases of this serotype are shown (LEDERBERG and Epwarps 1953, unpub- lished) to have a common antigen, agglutinins for which are present in the anti- kunzendorf 1.5 serum and which also tend to suppress the c as well as the 1.5 phase. This antigen (which might be a somatic rather than a flagellar component) had been disregarded previously because it is unusually labile to formalin, commonly used in setting up H-agglutination tests. 316 For other monophasic types, e.g., S. abor/us-equi (a):enx, this contingency probably does not apply, but the hidden phase can be demonstrated either by stringent selection for rare variations (Epwarps and Bruner 1939) or by trans- duction of the Hf allele to other recipients (LEDERBERG and EDWARDS 1953). This type therefore has the genotype H;H,”™ but some factor is present at these or other loci which stabilizes the existing phase, and is under current study. 3.17 S. typhi represents still a third variety of monophasic behavior which can be represented by the genotype H? H)- deficient, for no homology with H> factors of other strains can be detected by transductions in either direction, although the 7; factor is readily transduced and substituted for. S. enteritidis, Hf", and related types, and some varieties of “monophasic S. paratyphi B”, H, appear to belong in the same category. 318 A fourth category is represented by another strain of “monophasic S. para- typhi B”, CDC-137, which can be described as H? H,- where “~” implies a null allele always hypostatic to H, . In transductions to other strains, this type never transfers an effective fH» allele. However, it readily incorporates transduced ff; factors, e.g. H? I,- x— Hi Hs" to give a typically diphasic H! H.""'. This type may be the counterpart of S. aborius-equi (3.16) since an absolutely stabilized H, factor in the hy- postatic state could never be expressed, even if transduced elsewhere. 319 These types are enumerated to display some of the wealth of material now open for recombinational study in the Salmonella group. They also show how mono- phasic exceptions can be accomodated to the pattern of phase control which was deduced from typical diphasic forms. 320 Antigenic mutation. The technique of immunoselection can also be used for the selective isolation of new mutant alleles at a single locus, once the genetic struc- ture of diphasic types has been worked out. For some time, it has been known that some serotypes which carry b occasionally generate an “artificial phase” antigen, 233, when selected in b serum. Without further analysis, zas Might be interpreted either as a mutation H? -> H;*, or a phase variant in H H5, In the course of the present study, the following changes have been obtained as isolated events during selection in the homologous serums: S. paratyphi B, b:1.2 > 2321.2, and S. abony, db: enx > Z33:enx, The derived types showed typical phase variation between the indicated phases. They were also subjected to transductions to and from b:1.2, bienx, and i:1.2 in which z33 proved to be homologous with b and i respectively. This suggests that 23; corresponds to the allele H7°', and that the change from HW? to H** isa true mutation. The zs; shows, however, a strong antigenic relationship to the b antigen, and further studies on the differentiation of the b and 233 alleles should be preceded by a more detailed serological analysis of the stocks in question. “I TAL a PHASE VARIATION IN SALMONELLA 3.21 The following reports of antigenic variations should also probably be classified as mutations of the H, or H, locus as indicated, although they have not yet been analysed by transduction tests. H; changes: d — j; (see KAurrMann 1951) lv 3 lw (BRUNER 1954); anda complex series gm — gp, mt > gt, mt > gms, gmq — gm, gmq — gq, gmt — mt, gmt — gt, gmt — gm, and gst — gs (BruNER 1953). H, changes: 1.2 — 1.2.3, 1.2.3 — 1.2 (BRUNER and Epwarps 1948); 1.5 — 1.10, 1.5 — 1.11 (BrunER and Epwarps 1947), 3.22 Although the symbols for these antigens are more complex than those for b and z33, they are not fundamentally different, as the formulae primarily reflect a more detailed analysis. In general, the antigenic mutations have generated forms still related to the primitive form, and mutations of the scope, for example, of b > i have yet to be observed. 4, DISCUSSION 41 The preceding account has supported the conclusion that phase variation can be explained by an alternation of “states” of the H, locus. When this gene is in the active or epistatic state, its allele is phenotypically expressed, and the Z/, factor is functionally suppressed. Alternatively, when the JZ, gene is in the inactive or hypo- static state, the ZZ, factor is expressed. While the H, and If, factors are thus func- tionally complementary, the 7; factor does not vary in its Aerifable state, and its activity depends on the state of the MZ» locus. Therefore, when an inactive H, is transduced, from a donor in phase-2 where H, was epistatic, to a recipient in phase-1 wherein H, is hypostatic, the H, factor can be immediately expressed. The experi- ments have thus indicated that the “state” of the H» factor is inherited as such with the transduction of 77, specificity, while the functional state of 77, is not. The primacy (or stability of state) of the H> locus in transduction does not preclude a more inte- grated system during vegetative growth in which the cytoplasm or the H/, locus may still play an important role in the perpetuation of phase. For example, epistasis by Hz might depend on local saturation by the immediate products of its action, a steady state or feedback condition which would depend in turn on the functioning of the cytoplasm and the competitive inefficiency of a stripped H; locus. Unfor- tunately, we have almost no facts on which to build speculations on the material meaning of “‘state’’: one of the first unknowns to be pursued might well be the external regulation of phase. Phase variation (more accurately phase stability) evidently does not depend on the actual production of the antigenic end product, since it can be demonstrated in Fla~ mutants which are devoid of H antigens (STOCKER ef al, 1953; 34, 3.7). 42 The flagellar antigens are known to be proteins (Stocker 1956) but in Sal- monella they have not been studied in regard to structural features such as amino acid composition, or even their basic physicochemical properties, so there is no funda- mental basis for distinguishing the phases except by their serological reactivity. However, phase-2 flagella are differentially agglutinated by acridine dyes (BERN- STEIN and LEDERBERG 1955) which speaks for some overall difference in the action of H, and H,. In addition, some strains show differences in motility between the two phases (Epwarps ef al. 1954); whether this reflects a quantitative or qualitative difference in the flagellar proteins is a nice question. There is no precise evidence on 754 J. LEDERBERG AND T. IINO the regulation of phase variation by external influences, partly because of technical difficulties (cf. Stocker 1949), There has been considerable discussion of “induced phases” in Salmonella, but no evidence that antiserum plays other than a selective role in the detection of serotypic changes has been forwarded. 13 Throughout this paper we have referred to serotypes notwithstanding the use of binomial epithets and frequent allusions elsewhere to species of Salmonella. Bacterial taxonomy has been dominated by the principle that any clonally stable variety whose characters can be sharply differentiated constitutes a species, a definition which is at variance with the conventions for higher forms. Transductional analyses have shown that serotypes differ in regard to well defined unit markers, but the extent to which genic differences diffuse among bacterial populations in nature is unknown. 44 The characterization of serotypes in this work has depended on the Kavurr- MANN-WHITE scheme and on the reagents prepared to its specifications. Its authors have cautioned that the scheme is an abbreviation for diagnostic purposes, and not a complete exposition of the antigenic individuality of the serotypes. For example, the symbol “d”’ does not always betoken an identical antigenic constitution, and minor differences are often detectable by cross-absorption tests. The extent to which such differences are symbolized (e.g. in the distinction of enx from enzis in contrast to the lumping of distinguishable 1.5 phases) depends on two possibly gratuitous factors, as well as historical accidents: the apparent titers of the residual antiserums after cross-absorption, and the urgency of more detailed differentiation as an element of classification. Neither of these considerations stands out as a necessary correlate of the importance of ‘minor’ differences for genetic analysis, Unfortunately, the minor differences are not systematically catalogued, although they are usually mentioned in original publications of types. The sheer bulk of the present scheme has encouraged a more recent lumping of related serological phases, e.g., 1.2 with 1.2.3, rather than a finer analysis. The immediate practical consequence of these cautions is that the conventional instructions for the preparation of reagents should be followed ex- plicitly rather than be substituted for by seemingly plausible deductions from the diagnostic scheme (KAUFFMANN 1951; Epwarps and Ewinc 1955). 4.5 In the original classification of Fla— mutants (STOCKER ef ai. 1953), three were found to be closely linked to H, but were not identical. Many other Fla~ mutants were examined and found to be nonallelic with one another and not linked to ZZ,, Further studies have shown that S. heidelberg SL28 also carries an H,-linked Fla~ mutation, as do several additional isolates that were not available earlier, including the laboratory mutant of S. typhimurium SW-1157, table 3. These Fla~ mutants are all distinct from one another. Evidently, one cluster of Fla~ mutants occurs in the neighborhood of the H, locus. Unfortunately, there is no reliable direct method of demonstrating linkage of one Fla factor to another, owing to their identity of pheno- type. The non-//,-linked Fla~ mutants might therefore represent another compact cluster, or they might be widely dispersed (at least they are not closely linked to H; by a hypothetical analogy). It is impossible to judge whether the linkage of Fla factors to H; is fortuitous, or is somehow interrelated with their action on a common organelle (cf, DemERec ef al. 1955). The appearance of a clusters of closely linked, nonrecurrent mutants with the same phenotype is reminiscent of the cluster of “I on ant PHASE VARIATION IN SALMONELLA Lac, and of Gal mutants in Escherichia coli (E. M. LEDERBERG 1952; MORSE et al. 1956). Allelism tests on the more limited scale of conventional technique in, say, Drosophila would probably have described each of these clusters asa set of multiple alleles at a single locus. 46 For the moment, the 4 factors appear to be simple alleles, but these markers are not so readily screened for rare recombinants, and they may yet prove to be compound in the sense of the histocompatibility factors in mice (GoRER, unpubl., cited in ALLEN 1955). The persistent cross-reactivity of the original and variant in antigenic mutations such as b — z33 or 1.5 — 1.10 is the only hint in favor of this speculation, and the #7, factors so far have retained their integrity in recombinational analysis. Preliminary chemical studies encourage the view that antigenic components of a complex such as enx are manifestations of different aspects of the same flagellar protein (Nakayva ef al. 1952). 47 In a system of DNA-mediated transduction (transformation) in Hemophilus influenzae, Letpy, HAHN and ALEXANDER (1953) have described the separation of capsular determinants a and b in a fashion to suggest that these are nearly but not precisely allelic. In the pneumococcus, EpHRrussI-Tayior (1951) had likewise demon- strated a series of closely linked “subunits” which control the quantitative aspects of capsular polysaccharide synthesis. Her ‘‘autogenic” transformations are readily interpreted as linked transductions of the factor complex, the ‘‘allogenic”’ transforma- tions as transductions of a part only, ie., as crossing over within the complex. 48 The state of a locus can be considered as a parameter of genic variation inde- pendent of mutational changes in specificity, which recalls D. Lewis’ (1954) account of fractional changes in seli-compatibility genes in Oenothera. These mutant genes had lost only their function in the pollen, but retained their function and specificity in the style. The suppression of phenotypic effects of transposition of heterochromatin (E. B. Lewis 1950) or of the Ds and Mf elements (McCLInTock 1956; BRINK and NILan 1952) may also be viewed as local changes of state, which although they depend on structural modifications nevertheless suggest how gene action may be regulated. These may therefore be considered as conceptual, if not mechanical, models of phase variation. 49 The cytoplasm has attracted considerable favor as the place of morphogenetic differentiation, largely because of an argument which has been well rephrased by Epurusst (1953): “Unless development involves a rather unlikely process of orderly and directed gene mutation, the differential must have its seat in the cytoplasm”. Most experiences with genic changes have involved sporadic, irreversible changes in specificity that qualify them very poorly for a role in development. However, the -oncept of Jocal states may provide a more acceptable hypothetical basis for nuclear differentiation. This reconsideration of the nucleus has recently been bolstered by the caryonidal determination of mating type in Tetrahymena (NANNEY and CAUGHEY 1955), morphological diversification of chromosomal regions during the development of various tissues (see BREUER and Pavan 1955), and especially the direct demonstra- tion by Krxc and Brices (1955) with nuclear transplantations from cells in later to earlier stages in frog development. The interplay of coordinate nuclear and cyto- plasmic determiners has been emphasized by SONNEBORN (1954). 756 J. LEDERBERG AND T. IENO SUMMARY 1. The flagellar antigens of Salmonella types occur in two alternative phases, e.g., iand 1.2 in S. /yphimurium. Transduction analysis had shown that each phase corresponds to a distinct locus, so that the genotypic formula of S, ivphimurium would be HZ; Hj”. Phase variation therefore consists of the alternative manifestation of antigenic potentialities at these loci. 2, Transduction experiments were carried out with S. lyphimurium, S. abony b:enx, and S. heidelberg, r: 1.2, with donor and recipient in various phases. The expression of the 7; loci depended on the phase of the recipient (which retained the original ZT, locus), while the expression of a transduced H, locus depended on the phase of the donor cell. It is therefore concluded that antigenic phase is governed by H, locus, which occurs in either of two clonally stable (or nearly stable) states. The concept of “local state” presents a parameter of genic variation distinct from ordi- nary mutation, and may be applicable to other problems in genetics and morpho- genesis. ACKNOWLEDGMENTS This work would not have been possible without the cumulative labors which are represented by the KAuFFMANN-WHiItrE scheme. We are particularly indebted to Dr. C. C. Spicer for furnishing standard serums prepared by the Standards Labora- tory, Central Public Health Laboratory, Public Health Laboratory Service, Colin- dale Avenue, London, England, and to Dr. P. R. Epwarps, of the Enteric Bacteri- ology Laboratories, Communicable Disease Center, Public Health Service, Chamblee, Georgia, for serums, cultures, and innumerable bits of advice and Salmonella lore. We are also obliged to Dr. M. DEMEREC for furnishing lysates of a number of auxotrophic mutants. LITERATURE CITED ALLEN, S.1., 1955 Linkage relations of the genes Histocompatibility-2 and Fused tail, Brachyury and Kinky tail in the mouse, as determined by tumor transplantation. Genetics 40: 626-650, AsbREWES, F. W., 1922 Studies on group agglutination. J. Pathol. Bacteriol. 25: 515-521, Brace, G. H., 1954 The Genetics af Paramecium aurelia. Cambridge Univ. Press. Bernstein, A., and J. LeperBero, 1955 Agglutination of motile Salmonellas by acridines. J. Bacte- riol. 69: 142-146. BrevER, M.F., and C. Pavan, 1955 Behavior of polytene chromosomes of Rhyachosciara angelae at different stages of larval development. Chromosoma 7: 371-386, Brink, R. A., and R. A. NiLAN, 1952 The relation between light variegated and medium variegated pericarp in maize. Genctics 37: 519-544, Bruner, D.W., 1953 Induced variation in the g phases of somatic group B of the genus Salmonela. J. Bacteriol. 65: 223. 1954 Induced variation in the specific phase of Salmonella litchfeld. J. Bacteriol. 67; 127-128. Bruner, D. W., and P. R. Epwarps, 1947 Changes induced in the nonspecific antigens of Sal- monella. J. Bacteriol. 53: 359. 1948 Changes induced in the 1, 2, 3 antigens of Salmonella. J. Bacteriol. §5: 137. Drmerec, M., I. Bromsrranp, and Z. E. DemErEc, 1955 Evidence of complex loci in Salmonella. Proc. Natl. Acad, Sci. U.S. 41: 359-364. Dusos, R. J., 1945 The Bacterial Cell. Cambridge, Mass.: Harvard University Press. Epwarps, P. R., and D. W. Bruner, 1939 The demonstration of phase variation in Salmonella abortus-egul, J. Bacteriol. 38: 63-72. ~I ca ~T PHASE VARIATION IN SALMONELLA 1946 Notes on monophasic Salmonella cultures and their use in the production of diagnostic serums. J. Bacteriol. 52: 493-498. Epwarps, P. R., and W. H. Ewinc, 1955 /dentification ef Fnterebacteriuceae. Minneapolis, Minn.: Burgess Publ. Co. Epwarps, P. R., A. C. McWxHorter, and M. A. Fire, 1954 Studies on group G of the genus Sal- monella. J. Bacteriol. 67: 346-349. Epnrussi, B., 1933 Nucleo-evtoplasmic Relations in Micro-organisms. Oxford University Press. Eprrusst-Taytor, H., 1951 Genetic aspects of transformations of pneumococci. Cold Spring Harbor Symposia Quant. Biol. 16: 445-455. KAUFEMANN, F°., 1951 Entervbacteriaceae. Copenhagen: Munksgaard. Kine, T. J., and R. Brices, 1955 Changes in the nuclei of differentiating gastrula cells as demon- strated by nuclear transplantation. Proc. Natl. Acad. Sci. U.S. 41: 321-325. LEDERBERG, FE, M., 1952 Allelic relationships and reverse mutations in Ascherichia coli, Genetics 37; 469-483. LeEvERBERG, J., 1947 The nutrition of Salmonella. Arch. Biochem 13; 287-290. 1948 Problems in microbial genetics. Heredity 2: 145-198. 1950 Isolation and characterization of biochemical mutants of bacteria. Methods Med. Research 3: 5-22. LEDERBERG, J., and P. R. Epwarps, 1953 Serotypic recombination in Salmonella. J. Immunol. 71: 232-240. Liipy, G., E. Haun, and H. E. ALexanper, 1953 In vitro production of new types of Memo plilus influensae. J. Vexptl. Med. 97: 467-482. Lewis, D., 1954 Comparative incompatibility in angiosperms and fungi. Advances in Genet. 6: 235-285. Lewis, F. B., 1950 The phenomenon of position effect. Advances in Genet. 3: 73-115. MecCrintock, B., 1956 Intranuclear systems controlling gene action and mutation. Brookhaven Symposia Biol. 8: 58-71. Morsk, M. L., FE. M. LEDERBERG, and J. LEDERBERG, 1956 Transductional heterogenotes in Escherichia coli. Genetics 41: 758-779. Nakaya, R.. H. Ucuipa, and H. Fuxumt, 1952 Studies on the antigenic patterns of bacteria flagella. Japan. J. Med. Sci. Biol. 5: 467-473. Nanney, D. L., and P. A. Caucrry, 1955 An unstable nuclear condition in Yelrahymena pyri- formis. Genetics 40: 388-398. Sait, T., and A. L. Reacu, 1903 The non-identity of agglutinins acting upon the flagella and upon the body of bacteria. J. Med. Research 10: 89-100. SONNEBORN, ‘T. M., 1954 Patterns of nucleocytoplasmic integration in Paramecium. Proc. Intern. Congr. Genet. 91h Congr.: 307-325. SrockeEr, B. A. D., 1949 Measurements of rate of mutation of flagellar antigenic phase in Sa/- monella (vphimurinum, J. Hyg. 47: 398-413. 1956 Bacterial flagella: morphology, inheritance and constitution. Symposia Soc. Gen. Microbiol. 6: 19-40. Stocker, B.A. D., N. D. Zinper, and J. LEDERBERG, 1953 Transduction of flagellar characters in Salmonella. J. Gen. Microbiol. 9: 410-433. Urrake, H., T. Nakacawa, and T. AxrBpa, 1955 The relationship of bacteriophage to antigenic changes in group FE Salmonella. J. Bacteriol. 69: 571-579. ZNbER, N.D., 1955 Bacterial transduction. J. Cellular. Comp. Physiol. 45 Suppl. 2: 23-49. ZINvER, N. 1., and J. LeperpnerG, 1952 Genetic exchange in Salmonella. J. Bacteriol. 64: 679-699,