HYBRIDIZATION BETWEEN ESCHERICHIA COLI AND SHIGELLA! 8. E. LURIA anp JEANNE W. BURROUS Department of Bacteriology, University of Illinois, Urbana, Illinois Received for publication April 18, 1957 Genetic recombination by mating has been demonstrated among a variety of strains classified as Escherichia coli (Lederberg and Tatum, 1946; Cavalli et al., 1953). The possibility of genetic transfers among bacteria currently classified in different species or genera has been proved by transformation and transduction experiments (Lederberg and Edwards, 1953; Schaeffer and Ritz, 1955; Lennox, 1955). These findings are hardly surprising since there is no reason to expect that current bacterial classification cor- relates closely with capacity for hybridization or even with evolutionary relations (Luria, 1947). Hybridization capacity, even if present, might indeed not be responsible for significant amounts of gene flow among natural bacterial populations, which can propagate indefinitely by vegetative reproduction alone. Hence, adaptive selection might lead to considerable diversity among po- tentially interfertile clones, both because of the rarity of fertilization events and because of other isolation mechanisms. In this paper we present evidence that places the majority of dysentery bacilli (genus Shigella) into the same fertility system as Escherichia coli. The results also indicate a possible role of genetic recombination in the origin of some Shigella serotypes and, more generally, in the evolution of natural populations of these bac- teria. They suggest some potential pitfalls of current bacteriological procedures for the identi- fication of pathogenic enteric bacteria. 1This work was supported by grants from the American Cancer Society (recommended by the Committee on Growth), from the National Insti- tutes of Health (E-1629), and from the Dazian Foundation for Medical Research. The authors wish to express their appreciation for the advice of Dr. W. H. Ewing and Dr. I. Saphra, who also generously supplied cultures and sera. The collaboration of Dr. E. 8. Lennox in the initial experiments is also gratefully acknowl- edged. 461 MATERIALS AND METHODS Cultures. E. coli strain K-12 and its derivatives were from our laboratory collection. Most of these strains were received from Drs. L. Cavalli, W. Hayes, E. M. and J. Lederberg, P. D. Skaar, and E. Wollman. #. coli strain C (Lieb et al., 1955) was also used. All the strains are motile. Shigella dysenteriae strain Sh is a rough strain, commonly used as indicator for the phages carried by £. colt strain Lisbonne-Carrére (Bertani, 1951). Type cultures of S. dysenteriae, S. flecneri, and S. boydii were received either from the Division of Laboratories, Illinois Department of Public Health, Chicago, thanks to Dr. H. J. Shaughnessy, or from the Com- municable Disease Center, U. S. Public Health Service, Atlanta, through the courtesy of Dr. W. H. Ewing. The Illinois strains are identified by the symbol I, the Atlanta strains by the symbol A (for example, 8. dysenteriae strain 11; S. flexneri strain 2aA). Streptomycin resistant mutants (S‘) were isolated from plates of solid nutrient media containing 100 ug/ml streptomycin that had been seeded with about 10'° streptomycin sensitive (S*) cells per plate. S™ cultures were maintained and grown with 100 ug of streptomycin/ml, except when this would interfere with the experi- mental procedure. Phage resistant mutants were isolated by standard methods. Lysogenic derivatives were obtained from plates with con- fluent incomplete lysis. Phage sensitivity tests were done by cross smearing. Media. L broth and L agar contain: tryptone, 10 g; yeast extract, 5 g; NaCl, 10 g; glucose, 1 g; water, 1,000 ml; (pH 7.0) with or without 10 g powdered agar, respectively. Minimal, EMB and EMS (= minimal EMB) media were prepared following Lederberg (1950a) with 15 g/L agar. Sugars were dissolved in water, sterilized and added to media to a concentration of 1 per cent. Media for gas production tests were made by adding 1 per cent sugar (sterilized 462 separately) to tubes of nutrient broth (pH 7.2) with bromthymol blue indicator and inverted vials. Assays of phage A were done on the medium recommended by Kaiser (1955). Phage P1. was assayed on L agar plus 107? m CaCh. Antisera. Sera against Shigella type cultures were obtained either commercially (Markham Laboratories, Inc.) or through the courtesy of Dr. W. H. Ewing and Dr. I. Saphra. Sera (anti-O) against E. coli strain K-12, its derivatives, and hybrid strains were prepared in rabbits by 1 or 2 injections of about 10° cells collected from nutrient agar, heated at 100 C for 2.5 hr and mixed with adjuvants (Freund’s method, see Cohn, 1952). Adsorbed sera were prepared by the double adsorption method according to Edwards and Ewing (1955) using cells from 10 plates of L agar/ml of serum. Agglutination tests were done in 9 mm tubes, using 0.5 ml vol of saline. The tubes were kept at 56 C for 2 hr, then at room temperature over- night. The antigens consisted of cells collected from agar slants or from heavy broth cultures, with or without washing. Whether used alive, or heated for 1 hr at 100 C, or treated with 75 per cent ethyl alcohol or with 0.5 per cent formalin overnight, the cells of a given organism gave O-agglutination endpoints agreeing within a factor of 2. In most tests, formolized antigens were used for nonmotile cultures and alcohol treated antigens for motile ones. A rapid diag- nostic test for serological changes in hybrids will be described in a later section. RESULTS Selection of recombinants. Strains to be crossed must exhibit stable differences in characters so that recombinants can be enriched and detected selectively. A convenient method proved to be selection for lactose utilization in streptomycin containing media (Lederberg, 1951; Hayes, 1953b). Recombinants LactS* can be obtained from mixtures of S* Shigellas and S* (strepto- mycin sensitive) E. coli parents plated on EMB lactose agar with 100 to 200 ug/ml streptomycin. All Shigella strains proved stably Lac; plated alone, they gave no Lact papillae on EMB- lactose agar and no detectable growth in minimal media with lactose as only carbon source (in addition to necessary growth factors); nor was any 6-galactosidase detectable in these strains by the o-nitrophenyl-$-p-galactoside test (Leder- LURIA AND BURROUS (vou. 74 berg, 19506). S* cells appear with a frequency lower than 10-* in S* cultures of #. colt strain K-12 derivatives. S* derivatives of the Shigella strains to be tested were isolated first. Heavy suspensions from stationary cultures of these Shigellas (Lac"S") and of E. coli (LactS*), grown in L broth to about 10° cells/ml, chilled and collected by cold centrifugation, were mixed directly on EMB-lactose streptomycin agar. Lact papillae, when present, appeared in 2 to 3 days. These papillae, upon isolation by streaking on EMB lactose, yielded stable Lact+S* clones, whose recombinant nature was confirmed by tests for other characters in which the parent strains differed (unselected markers, see below). The Lact recombinants gave a positive test for §-galactosidase. They produced acid from lactose in 18 to 24 hr (pH_ 4.9 + 0.2) and no gas. The parental combinations that were tested by this method are shown in table 1. The parental Shigella strains never yielded a single Lact mutant in any of the control plates, which con- tained altogether over 10" cells. Note that Lact recombinants were obtained also in crosses of Shigella with EF. coli strain C, a strain which differs from K-12 in many respects (Lieb e¢ al., 1955). Frequency of recombination and polarity. The frequency of recombinants LactS* (or Lact prototrophs, see below) as a function of the polarity (Hayes, 1953a; Cavalli ef al., 1953) of the Z. colt parents is shown in table 2. In these and in all other crosses, the Shigella parents be- have as typical F~ strains. All Hfr cultures of E. coli (see Hayes, 1953b) give higher frequencies of recombination than F+ cultures with Shigellas as well as with other strains of EH. coli. Thus, the mating polarity system observed in E. coli extends to Shigellas as well. The frequencies of recombinants are not much higher when the parents are incubated together before plating than when they are mixed on plates. The fre- quencies are uniformly lower for crosses of F+ or Hir E. colt with Shigellas than with F~ EZ. colt. It will be shown below that this difference is due to lower effectiveness of matings in producing recombination rather than to lower frequency of mating. Unselected markers and linkage. Recombinants isolated from crosses are expected to contain various assortments of the genetic traits, other 1957] HYBRIDIZATION BETWEEN £E. COLI AND SHIGELLA 463 TABLE 1 Fertile combinations of strains as teated by the LactSt selection Samples of 0.1 ml from each culture were mixed directly on EMB-lactose-streptomycin agar. Lact colonies were isolated by restreaking on the same medium. Organism Strain Shigella dysenteriae Sh X Escherichia coli K-12 Ft S. dysenteriae Sh X E. colt S. dysenteriae 1I* X E. col S. dysenteriae 2At X E. coli S. dysenteriae 31 xX E. cok S. dysenteriae 3A X E. coli S. dysenteriae 4l X E. coli S. dysenteriae 4A &X E. coli S. dysenteriae 5I &X E. coli S. dysenteriae 8A xX E. colt Shigella flexneri lal X E. coli S. flexneri 2al X E. colt S. flexneri -2al X E. colt S. flexneri 2aA X E. coli S. flexneri 2aA X E. colt S. flexneri 2bA X E. coli S. flexneri 2bA X E. colt S. flexneri 4al X E. coli S. flexneri 4aA X E. colt Shigella boydit 4A X E. colt S. boydit 4A xX E. coli S. boydit 6A X E. colt S. boydti 6A X E. coli Shigella flexneri 2al X EL. colt S. flexneri 2bA X LE, coli Organism Strain Recombinants K-12 Hfr K-12 Hfr K-12 Hfr K-12 Hfr K-12 Hfr K-12 Hfr K-12 Afr K-12 Hfr K-12 Hfr K-12 Hfr K-12 F+ K-12 Hfr K-12 F+ +e FHth FHFE+HH+HEH FHtHH+ HH tH * I = strains from the Division of Laboratories, Illinois Department of Public Health, Chicago, Tilinois. + A = strains from the Communicable Disease Center, U.S.P.H.8., Atlanta, Georgia. than the selected ones, in which the parent strains differed (‘unselected markers’), Gen- erally, more characters are contributed to the hybrids by the F- parent than by the F+ or Hfr parent (Hayes, 1953). The relative frequencies of various unselected markers and their depend- ence on the choice of selective markers make it possible to map the corresponding genetic factors in a linear linkage map (Lederberg, 1947). In the present work we have not attempted to construct a detailed genetic map of any one strain of Shigella. A number of characters were used either as selective or as unselected markers, mainly in crosses between #. coli strain K-12, S. dysenteriae strain Sh and S. flexneri strain Qal and their derivatives. The characters that were tested for presence or absence in hybrids are listed in table 3. As expected, the hybrids are predominantly like the Shigella F— parents. Note that some characters from the F*+ or Hfr coli parents never appeared in the hybrids. This is true for production of indole, for resistance to azide, and for utilization of xylose, maltose and mannitol with S. dysenteriae strain Sh; for mo- tility and gas production both with strain Sh and with S. flexneri strain 2al. Any character that appears in hybrids as a nonselected marker can also be used as a selective marker. For example, selection was successful for Ara+S* hybrids in crosses of S. dysenteriae strain Sh x E. coli, and for Xyl*S* (or DultS", or RhatS') with S. flexneri strain 2aI. All Arat and Xyl*+ recombinants produce acid in 24 hr 464 LURIA AND BURROUS [vou. 74 TABLE 2 Frequency of recombination and polarity of the Escherichia coli parent Examples of recombination frequencies obtained in crossing Shigella cultures with derivatives of E. coli strain K-12. Rati Experiment Shigella and Escherichia coli Parents: Organism, Strain Selection for Reconbinants/ Hfr or F* cells I* S. dysenteriae 57-1 prototroph X E. colt F-M- LactM+ <10 (none) 1-31-55 | S. dysenteriae 57-1 prototroph X E. coli F*C-H- LactCtH+ 107 S. dysenteriae 57-1 prototroph X 2. coli Hfr M-S* Lac*M+* 2x 10-5 S. dysenteriae 57-1 prototroph X E. coli Hfr M-S* LactMt 4x 10° It S. dysenteriae Sh St X FE. coli F+(\) prototroph S* Lac*8S* 1.4 X 107 1-8-57 S. flexneri 2aI S* (A) X EB. cola Ft) prototroph S* Lac*S: 6.6 X 107 E. coli F-T-L-S (A) X E. coli F+(a) prototroph S* TtLtS* 1x 10-5 JIIt S. dysenteriae Sh S* X E. coli Hfr 8* LactS' 6 x 10-4 12-13-56 | S. flexneri 2a S* X E. coli Hfr S* LactS* 3 X 10-4 E. coli F-T-L7*S* =X: E.. colt Hfr S* THLtS: 7x 107 * Cultures mixed directly on plates of EMS-lactose-agar. + Mixtures containing 3-12 < 10° F~ and 3.0 X 10 F* cells; incubated for 40 min with aeration be- fore plating. Assays after 30 min of aeration. { Mixtures containing 5-8 X 10° F- cells and7 X 10° Hfr cells; incubated for 40 min in broth without aeration before plating. In this experiment the number of Hfr cells was measured at the time of mixing. Platings in experiments II and III were done on EMB-lactose-streptomycin-agar for the Shigella X E. coli crosses; on minimal-thiamine-streptomycin-agar for the E. coli X E. coli crosses. The medium used in the Shigella xX Z. coli crosses may allow additional recombination on the plates. Symbols: T = threonine; L = leucine; M = methionine; C = cysteine; H = histidine; S = atrep- tomycin. TABLE 3 Transmission of characters from Escherichia coli parents to hybrids Characters from £. coli Parent Tested in Hybrids Shigella Parent: Organism, Strain Unselected. Present Absent S. dysenteriae, Sh (or 57-1): Lac* Vist; Vet; Arat; S*. Malt; Mtl+; Xyl*; Az; Lac- Ara~ Xyl- Mal- Mtl- Lac,” Vis. Ind+; Pit; A+; mot*; aert S* Vis* Ve* Az* Ind~ Pit AT Arat* Lact; S*. mot” ser” S. flexneri, 2al: Lact Vist. Ve; (Nie Git)t; mott; ser*; Lac” Xyl- Rha~ Dul” Vis* Xyl* Lact; Vis". -K -C Nie” Glt™ mot~ aer~ Rhat Lact. Dult* Code: Lac = lactose. Ara = arabinose. Xyl = xylose. Rha ~ rhamnose. Dul = dulcitol. Mal = maltose. Mtl = mannitol. Ind = indole production. mot = motility. aer = gas production. Az = azide. S = streptomycin. Nic = nicotinamide. Glt = glutamate. Vis = phages T1, T5. Ve = phage T6. Pl = Phage Pl. A-C = phage } produced like that from 2. colt strain C. \-K = phage 4 pro- duced like that from £. coli strain K-12. Laci” = cryptic. 6-D-galactosidase. + = utilized; or syn- thesized; or lysogenic; or present. ~ = not utilized; not synthesized; not lysogenic; or absent. * = sen- sitive. * = resistant. 1957} (and no gas) in liquid media with the correspond- ing sugars. S. flexneri strain 2aI grows in minimal medium supplemented with nicotinamide and glutamate (Nic"Glt). In crosses of S. flexneri strain 2al with auxotrophic F* or Hir strains of E. colt (M- or T-L~ or C-H-) no prototrophic re- combinants were observed. Crosses on minimal agar + nicotinamide + glutamate + lactose yielded only Lact Nic~ Glt~ recombinants. Thus, there was no evidence for transfer of Nict and Glt+ from E. colt to hybrids. S. dysenteriae strain Sh has ill-defined, non- specific requirements for a variety of amino acids, but it readily yields variants that grow fairly well on minimal medium. One such variant, strain 57-1, obtained from a S* derivative of strain Sh, has been used extensively in our work. In crosses of Shigella 57-1 x E. coli, we can select hybrids either for LactS' or for Lact prototrophy (on EMS minimal agar with lactose as only carbon source) as shown in table 2. The latter selection allows the isolation of recom- binants with the S* character from an £. coli parent. The frequency of unselected markers permits us to establish a few linkage relations. Crosses of Shigella Sh (or 57-1) x E. colt suggest the following linkage pattern: Ara—V,, s—-Lac—Vz (or possibly Ara—Vi,;—Ve—Lac). Here Ara and Lac represent the loci whose alleles are present in the Shigella parent. In crosses with S. flexneri, Lac is closely linked to Vi, 5 and only distantly linked to Rha and Xyl. The linkage relation between the loci V;,5, Lac, Vs and Ara is similar in Shigella and in Z. colt strain K-12 (Cavalli-Sforza and Jinks, 1956). This in- dicates extensive homology of genetic organiza- tion. It is impossible to decide, on the basis of our present data, whether the characters of the Ft (or Hfr) coli parents that fail to appear in hy- brids are excluded by the choice of selective markers (linkage with markers against which selection was made), or by incomplete homology between the linkage systems of EF. colt and Shi- gella, or by chromosomal rearrangements. Some characters, such as motility and aerogenesis, may involve multigene control. Failure of ex- pression of a gene may be due to interaction with other genetic factors. These problems may HYBRIDIZATION BETWEEN #. COLI AND SHIGELLA 465 be resolved more easily by use of transduction (Lennox, 1955). An interesting cross, included in table 3, is one in which the £. coli parent carries the muta- tion Lac; (cryptic 8-galactosidase) (Lederberg et al., 1951). Adaptive synthesis of 8-galactosidase can be induced more readily by alkyl-8-galacto- sides than by lactose. Crosses on EMB-lactose- streptomycin agar yield late papillae, from which one can isolate bacteria that form 8-galactosidase when exposed to methyl-8-galactoside. Thus, the hybrid has acquired the same Lac: allele present in the Z. colt parent. Behavior towards phages P1 and X. Strains of E. coli and of Shigella are hosts for many common phages (including the T1-T7 phages), with in- dividual differences in host range. A special inter- est applies to phages P1 and 4, the former as an agent of transduction (Lennox, 1955), the second as a phage whose prophage location in the linkage map of E. coli strain K-12 and whose behavior in bacterial crosses have been investigated (Wollman et al., 1956). The host range of these phages among Shigella cultures is shown in table 4. The widespread sensi- tivity to phage P1 opens the possibility of genetic analysis by means of transduction. Note also that some Shigella strains, like S. dysenteriae strain Sh, are susceptible to P1 with its original host range as first isolated from 2H. coli strain Lisbonne-Carrére (Bertani, 1951); other strains, like S. flexneri strain 2aI, are only lysed by the mutant P1k, isolated by its ability to lyse Z. colt strain K-12. This observation suggests fur- ther mutational homologies between Shigellas and £. colt. Phage \ attacks only S. flexneri strain 2al; the other Shigellas are neither lysed nor lysogenic for A. S. flernert strain 2al is readily lysogenized. When is liberated by this strain, either by induction of a lysogenic culture or by infection of a sensitive culture, it is modified in the same way as when grown on £. coli C (Bertani and Weigle, 1953); this \-C phage fails to grow in most cells of (nonlysogenic) FZ. coli strain K-12 derivatives. This indicates one additional genetic homology between a Shigella and LZ. coli strain C. Several hybrids from strain 2aI x JZ. coli strain K-12 were tested and all retained the C-like modifying property. Lysogenic derivatives were used in most 466 LURIA AND BURROUS {voL. 74 TABLE 4 Sensitivity of various strains to phages P! and Phage Pi Phage v Organism Wild type & Mutant, active on K-12 Kind of Ph: x Grown on Produ E.cahiC Grown on E. coli K-12 Escherichia coli K-12............ E.coli CO. oo... cece Shigella dysenteriae Sh.......... S. dysenteriae 1A. ............66. S. dysenteriae 2A ................ S. dysenteriae 5A. .....0.......5. S. dysenteriae 6A ...........0..5. S. dysenteriae 7A. .............5. S. dysenteriae 8A ..............5. S. flexneri 2aI and hybrids....... S. flexneri 2aA and hybrids...... S. flexneri 2bDA.........-....0005 S. fleanert 4aA........... 0.00000. B. boydit 4A... cece S. boydit 6A.... 6... eee eee DADMNDNRDADADHADDAAA rA-C DOARDRND MW HRD DODD DnNDADAHD BRN DWN ARDYNnAWe Dn S = sensitive; R = resistant by spot test; s = sensitive with low efficiency of plating; A = type strains from Atlanta; I = type strains from Chicago. crosses of S. flexneri strain 2aI with » lysogenic E. colt strain K-12 in order to avoid “zygotic induction” (Jacob and Wollman, 1956), that is, maturation of \ and lysis of the F~ parent when prophage X is transferred from a lysogenic F+ or Hfr mate. The phenomenon of zygotic induction of d made it possible (Wollman et al., 1956) to esti- mate the frequency of mating, as distinct from the frequency of recombination, which depends on the frequencies of mating, of transfer of the selective markers from the F+ or Hfr parent, and of postzygotic integration of these markers into the genome of the hybrids. Experiments on zygotic induction were done with EF. coli culture Hfr H By (A), a strain that was used extensively by Wollman e¢ al. (1956) in their analysis of the mechanism of recombination, details of which need not be given here. We need only recall that the order of gene transfer from this Hfr strain to a coli F- mate is: T-L-Lac-Gal-A. Under optimal conditions, mating can involve nearly 100 per cent of the Hfr cells. Zygotes that receive d are lysed; when mating is allowed to go to completion, this lysis eliminates about 50 per cent of the zygotes. This system made it possible to compare zygotic induction and recombination using either Shigellas or FE. coli as the F-. The same Hfr strain, but nonlysogenic, was used to estimate recombination frequencies undisturbed by zygotic induction of X (table 5). In the re- combination experiments selection was made for Lac*S" when using Shigellas F- and for (TL)+S* when using EF. colt F-. The results in table 5 show that, on the one hand, zygotic induction (and therefore mating and gene transfer) is about equally frequent in all cases. On the other hand, the recombination frequency is 100 to 1000 times lower in crosses with F~ Shigellas than with F~ £. colt. This is presumably not due to the different selec- tive markers since, at least in crosses with the F~ E. coli, the Lac locus is donated to re- combinants about 50 per cent as often as the TL segment (Wollman ef al., 1956). We conclude that the Jower recombination reflects lower post- zygotic integration of genes from the Hfr parent, rather than less frequent mating or less frequent transfer of genetic material. Test for a possible death of Shigella F- cells following mating with E, coli Hir gave negative results. A remarkable observation is that the phage X liberated upon zygotic induction in the mating of Hfr (A) with S. flexneri strain 2aI is only partly (~50 per cent) modified to the X-C form, although phage grown directly on strain 2aI is fully modified. In control experiments with E. coli strain C F— the phage produced by zygotic TABLE 5 Zygotic induction of phage \ and recombination frequencies > Plaques per ml of Mixtures Recombination Organisms, Strain and Titers (cells/ml) 40 Min 180 Min : Selection | Recombinants/ onc on K on C on K Shigella dysenteriae Sh (1.2 X 108) X E. coli Hfr (4) (3.9 X 107) | 1.2 X 107 7.7 X 10¢ 8 X 10° 4x 10° Lac*S" 5.1 X 10° 0.81 S. flexneri 2aI (2.6 X 10°) X E. coli Hfr (A) (3.9 X 107) 4.3 X 10% | 2.6 X 108 3.1 X 108 1.4 X 108 Lac*8* 7.2 X 10-4 0.1! S. fleanert 2al (A) (1.7 X 108) X £#. coli Hfr (A) (3.9 X 10°) 3.5 x 10%| 3.7 X 108* | 1.0 & 108 | 1.1 X 107* Lac*S* 1.2 X 10° S. flexneri 2al (A) alone 3.1 X 106 <10¢ 7.8 X 10° <5 X 105 _— _ E. colt F-(TL)- (2.6 X 108) X E. coli Hfr (A) (8.9 X 10°) 5.6 x 10* | 4.9 X 108 6 X 10° 5 X 108 (TL)*S* | 8.2 K 10°? 0.14 . E. coli Hfr (A) (3.9 X 10°) alone =7 X 104 =2 xX 10 =4X 10° =7 X 10° _ _ E. coli F-(TL)~ (2.5 X 108) & E. coli Hfr _ _ _ _ (TL)tS* | 5.6 X 10-* Zygotic induction. Growing cultures of S'F- bacteria were mixed with cells of Escherichia coli K-12 Hfr 8* (A) (derived from Hfr H). The cells had been taken from growing cultures, chilled, spun and resuspended in broth. The mixtures were aerated. After 40 min of contact, samples were treated with anti-A serum, diluted in broth and plated for ) plaques. A diluted sample was incubated for 180 min and plated again for \ plaques. All platings were done on )-sensitive, S' derivatives of both E. coli strain C and £. coli strain K-12, using agar with streptomycin. Genetic recombination. Suitable dilutions of the same mixtures as above, or of mixtures with nonlysogenic Hfr, were plated on selective media after 40 min of contact in broth. When selection was for (TL)*S*, the cells were washed and plated on minimal agar + thiamin + streptomycin. Assays and controls. Similar cultures of unmixed parent strains were plated for viable counts after 30 min of incubation. They were also plated for \ plaques and on selective media as controls for zygotic induction and recombination, respectively. The values in italics are the ratios “A plaques/Hfr cells” where zygotic induction was present. * Most plaques produced on C are accounted for by the lysogenic S. flexneri parent, and most of those on K by the lysogenic HZ. coli Hfr parent, without zygotic induction. [£961 VITADIHS ANY I700 ‘2 NAAMLAG NOLLVZIGIYdAH LOF 468 induction was more extensively modified to the d-C form, although not as completely as reported by Jacob and Wollman (1956). Experiments are in progress to clarify the reason for the partial modification. It seems possible that a relation exists between the low level of postzygotic in- tegration and the persistence of the A-K form in coli < Shigella matings. Note that zygotic induction occurs also in crosses with S. dysenteriae strain Sh, which can- not absorb phage A and which, apparently, does not become d-lysogenic upon mating. Growth of » transduced by phage Pl into S. dysenteriae strain Sh has also been observed by Lennox (éo be published). Zygotic induction of \ was absent, however, in a cross with S. flexneri strain 2aA EF. Unstable hybrid strains. Most hybrid strains are stable for all selected and unselected parental markers. Moreover, each recombinant colony appears to be pure (nonsegregating) for all markers. The one exception concerns some crosses of S. flexneri strain 2aI and its derivatives with one strain of EB. coli (Hfr M-V," Vi,s"). These crosses yield a certain proportion of unstable Lact colonies, which segregate out stable Lac”, unstable Lact and a few stable Lact. The Lact unstable hybrids are stable for other characters derived from the Hfr parent (except sometimes for V,,s, which is closely linked to Lac). It seems likely that in mutating from Vi,* to Vis® the parent Hr strain has acquired a factor resembling the Het factor described by Lederberg (1949). Further study of this factor is planned. The stable Lact segregants resemble other Lact recombinants from S. flerneri strain 2aI, except in a serological property to be discussed below. Changes in mating polarity, Most F~ strains of E. coli can be converted to F+ by infection upon cultivation in admixture with F+ strain (Cavalli et al., 1953; Hayes, 1953a; de Haan, 1954). This infection may be the expression of an incomplete mating (Wollman et al., 1956). The conversion of Shigellas to the F* state was accomplished and demonstrated by experiments of the type shown in table 6. Each of 3 Shigella strains was culti- vated in mixture with an F* strain of E. coli and reisolated at intervals by selective plating methods. Several of these isolates (and the original Shigella strains) were then subcultured with an F~ strain of EZ. coli (TLB,)~. The coli strain was in turn reisolated and tested for fer- tility with EF. coli F-M-. A positive test indicates LURIA AND BURROUS [vou. 74 that the Shigella strain used had become F* and a source of F+ contagion. The results were uni- formly negative for the original Shigellas and positive for the treated ones, confirming the F- nature of the former and their conversion to F+ by infection. Repeated attempts, however, to demonstrate recombination between these F+ Shigellas and any F~ strain, either Z. coli or Shigella, using a limited variety of selective methods, gave only negative results. Whether this is due to limita- tions of our tests or to an intrinsic property of F* Shigellas cannot be decided from the avail- able data. Serological properties of hybrids. In aggluti- nation tests for O-antigens, E. coli strain K-12 and its derivatives give only some slight cross reactions with the Shigella strains tested in our work.? When certain hybrids were tested with unadsorbed O-antisera against the parent strains, a new character appeared. Most Lact hybrids from crosses between E. coli strain K-12 de- rivatives and S. flexneri strain 2aI were found to give a reduced agglutinin titer with a com- mercial antiflexner “W’ serum. This finding prompted further serological investigation of these and other hybrids. Most of the Lact hybrids from crosses between E. coli (whether derivatives of strain K-12 or of strain C) and S. flexneri strain 2al show the altered serological behavior with antiflexner sera, which consists in a reduced agglutinating titer with unadsorbed sera antiflexner W and anti- flexner 2a, with increased agglutinability by several sera antiflexner 4a. These results are shown in table 7. Sera against other flexner prototypes were not tested. Two Lact hybrids with the abnormal behavior, labeled strains F21 and F22, one A-lysogenic, the other not, were selected and used to prepare antisera. These sera prove identical in range of reactivity. As shown in table 8 for the serum anti-F22, these sera gave high titers with the homologous hybrid strains and low titers with 2 Most cultures of Escherichia coli strain K-12 and its derivatives rapidly become rough under laboratory conditions. The sera were prepared with smooth cultures, which did not give self- agglutination in isotonic saline after overnight treatment with 70 per cent ethanol. In most agglutination tests the antigens were suspensions prepared from smooth colonies of an Hfr M~ derivative. 1957] HYBRIDIZATION BETWEEN £. COLI AND SHIGELLA 469 TABLE 6 Transfer of F+ property between Escherichia coli and Shigella Crosses (on Minimal Agar + Thiamin) Recombinants . colt F-(TLB,)- * E. colé F-(MB)- .coliy ¥-(TLB,)~ X £. coli F-(MB)- . colij+ F-(TLB,)~ X EZ. colt F-(MB)- . colixy F-(TLB,)~ X EZ. coli F-(MB)- . colitz+ F-(TLB,)~ X E. coli F-(MB)- . colint F-(TLB:)- X E. coli F-(MB)- . coligy7+ F-(TLB,)~ X E. colé F-(MB)- | by by by by Gy by by titi ti Initial cultures: I « S. dysenteriae 57-1 8"; II = S. flexneri 2al S?; TII = S. flexneri 2aA Sr. Mixed cultivation with Z. coli F+(TLB,)~S*. Inoculum: 10! E. coli + 104 Shigella. Two or three daily transfers of 0.1 ml into 50 ml broth. Reisolation of Shigellas I, II, III (F+?) on streptomycin agar; labeled I+, II*, ITIt. Mixed cultivation of Shigellas I, II, III and I+, II+, IIJ+ with EZ. cols F-(TLB,)~. Inoculum: 10¢ Shigella + 10 FE. coli. Two daily transfers as above. Reisolation of E. coli by washing and streaking on minimal agar plus thiamin. Labeled F. coli, E. colii+.... TABLE 7 Agglutination reactions of Shigella flexneri strain 2aI and of hybrids Unadsorbed Sera* Organism Anti-4a Anti-W Anti-2a Anti-2b Anti-K-12 Serum A Serum B S. flexneri strain 2al...............00065 25, 600 25,600 1,600 400 1,600 <25 Hybrid F22...... 00.060. 2 cece eee eee 400 6,400 1,600 3,200 6,400 <25 Escherichia coli strain K-12............. <50 50 100 100 _ 6,400 * Values are the highest serum dilutions with positive agglutination. TABLE 8 Agglutination reactions of Shigella flexneri strains with various sera Unadsorbed Sera Organism ; os Anti-w | Antize | Antizd | Antita, | Antihybrid | antix.s SG. flexneri 200 ........ 000 cc cece ees 25,600 | 25,600 1,600 1,600 400 <25 Hybrid F22.............. 20s cece eeeee 400 6,400 1,600 6,400 6,400 <25 S. fleanert 2A .... 0. ccc eee 25,600 25,600 6,400 6,400 800 <25 Hybrids from 2aA.......-..2-.20.0-- 1,600 6,400 3,200 6,400 | 12,800 200 S. fleaneri QA... 2. eee eee ee 25,600 12,800 12,800 <200 400 <25 Hybrids from 2bA.............-..4-- 1,600 6,400 3,200 12,800 12,800 50 S. flernert 4aA .... 0.0 cece eee 1,600 6,400 3,200 12,800 12,800 400 Hybrids from 4aA...........0-:00eee 1,600 6,400 3,200 3, 12,800 400 Escherichia coli K-12...............- <50 50 100 _ 400 6,400 cultures of S. flerneri strains 2aI, 2aA and 2bA. All Lact hybrids from S. flecnert strain 2al Instead, the antihybrid serum agglutinates that were tested gave either the typical “¥22- cultures of S. flerneri type 4a as much asitsown like” reactions shown in table 8 or normal 2a re- homologous antigen. It agglutinates Z. coli actions, without intermediates. It was possible, strain K-12 only with very low titer. therefore, to devise a simple test for extensive 470 screening of hybrids. This consists in inoculating each hybrid (from a pure restreaked colony) into 2 tubes with 1 ml of broth each, one containing serum anti-W diluted 1:3200, the other with serum anti-F22 1:800. Upon growth overnight at 37 C each hybrid is agglutinated in one or the other tube, but never in both. Several crosses be- tween cultures of S. flexneri strain 2aI and deriva- tives of E. coli strain K-12 gave 95 per cent F22- like Lact hybrids and 5 per cent serologically un- changed Lact hybrids. Out of 100 hybrids selected for markers other than Lact, the only two that were F22-like serologically were also Lact. Further crosses showed that most Lact hybrids between E£. coli and another strain of S. flerneri type 2a (= 2aA) were serologically similar (although not identical) to the F22-like hybrids. That is, they had lost some titer with anti-2a and often also with anti-2b sera, and had gen- erally become reactive at higher titer with serum anti-F22 and serum anti-4a. An unexpected finding was that very similar hybrid strains were obtained by crossing £. coli with S. flexneri type 2b. Also, hybrids with similar properties were obtained from crosses of EF. colt with S. flexneri type 4a. The data are included in table 8. These results suggested that all changed hy- brids belonged to a single serological group. To assess the meaning of this hybrid group, we con- sidered its serological relationship to E. colt strain K-12, to S. jflernert type 4a and to S. flexneri type Y. Tests on hybrids F21 and F22, kindly carried out by Dr. W. H. Ewing, classified these strains as S. flexneri type Y. We tested all the hybrids, the parent cultures, and a culture of S. flexneri type Y with serum against E. coli K-12 and also with a number of anti-flexner sera that had been cross-adsorbed in various ways. The results, shown in tables 8 and 9, can be summarized as follows: (1) The Y culture gives reactions similar to those given by the hybrids, although it is gen- erally more reactive than any of them. (2) E. coli strain K-12 cross-reacts to a limited but significant extent with S. flexneri type 4a; less with other type Shigellas. The hybrids vary somewhat in their reactions with anti-K-12 serum, but generally react with it less than cultures of flexner 4a. (3) The agglutinins of antihybrid F22 serum cannot be adsorbed completely by the parent LURIA AND BURROUS {vou. 74 strains of S. flerneri type 2a nor by 2b, but are completely removed by adsorption with any one of several 4a strains and by a strain of S. flexneri Y. (4) Adsorption of sera anti-2a or anti-2b with any one of the hybrids, as well as with the Y strain, removes heterologous reactions to the same extent. (5) Adsorption of several anti-4a sera with any of the hybrids, including those derived from S. flexneri type 4a itself, as well as with Y, removes all agglutinins for 2a, 2b, Y and all the hybrids, leaving various amounts of homologous ag- glutinins, depending on the serum. Results with one serum are shown in tables 8 and 9. Other anti-4a sera are even more radically adsorbed by the Y and hybrid strains. We conclude that the various hybrids, al- though probably not identical to one another, form a closely related group, similar to S. flexneri type Y, relatively closer to S. flexneri type 4a than to 2a or 2b, and whose antigenic structure includes the “group antigens” com- mon to 4a, 2a and 2b. Note that such Y-like hybrid strains are obtained also from S. flexneri type 2b, which, according to its known anti- genic composition, should, by Joss of the type antigen, give rise to X-like variants, not to Y-like ones (Edwards and Ewing, 1955). This obser- vation, and the fact that the agglutinins in anti- hybrid serum are only partially removed by ad- sorption with the parent S. flerneri type 2a (or 2b) prove that the serological change brought about by hybridization is not merely a loss of type antigen of the Shigella parent. The sig- nificance of these changes as to the possible origin and relations of various Shigella and Z. coli serotypes will be discussed later. The following points may be emphasized. First, the serological changes to the hybrid type are not due to the addition or substitution of a major E£. colt antigen for the Shigella antigens in the hybrids, since cross reactions of the hybrids with E. coli strain K-12 are only slight. Ap- parently, the genetic factor(s) responsible for the change does not reach full phenotypic expression in FE. coli, presumably because of the different genetic background. (Serological tests on Z. colt strain C were prevented by its roughness.) Second, the genetic factor(s) responsible for the change must be closely linked to the Lac locus, but are not located at the Lac locus, nor TABLE 9 Agglutination reactions with unadsorbed and adsorbed flexner sera Serum* Antigen aa | 7e/Hor | aayte | ae | 2H or | abyan | oth y | HOF | ayzaa | ren [EON raa/te | 22/201 Shigella flecneri strain 2alt..............| 25,600 | 6,400 | 6,400 | 1,600) 1,600 1,600 | 1,600 | <200 ; <200 400 | <200 | <200 | <200 Hybrid F22.............--.- 662 -ee eee 6,400 <200 400 | 1,600 <200 | <200 | 6,400 | <100 | <200} 6,400 | <200 | <200 | 1,600 Shigella fleaneri strain 2aAf.........--..| 25,600 | 12,800 | 6,400 | 6,400 3,200 800 | 6,400 | <100 | <200 800 | <200 | <200 | <200 Hybrid from 2aA.......--.-.--. 0s eee es 6,400 | <200 400 | 3,200 | <200 | <200] 6,400 | <100 400 || 12,800 | <200 | <200 | 3,200 Shigella flexneri atrain 2bAt.............| 12,800 | 6,400 | 3,200 | 12,800 | 12,800 | 6,400 | <200 | <100 | <200 400 | <200 | <200 | <200 Hybrid from 2bA...........-:. 202 e ees 6,400 | <200 400 | 3,200 200 200 | 12,800 | <100 400 || 12,800 | <200 | <200 | 3,200 Shigella flecnert strain 4aAf............. 6,400 | <200} <200/| 3,200| <200 | <200 | 12,800 | >800 | 6,400 | 12,300 | <200 | <200 | 1,600 Hybrid from 4aA.........-.....--- sees 6,400 <200 400 } 3,200 <200 | <200 } 3,200 | <100 200 || 12,800 | <200 | <200 | 3,200 Shigella flernert Y..........-.0 2.2000 es 25,600 | <200 400 | 12,800 | <200 400 | 12,800 | <100 400 || 12,800 | <200 | <200 | 6,400 * Adsorbed sera are indicated by the symbol/. For example, 2a/4a means: serum anti-2a adsorbed with 4a. /H means: serum adsorbed with any one of the four hybrids. { The same results are obtained with the type cultures and with their S' derivatives. ¢ The same results were obtained with two cultures received at several months interval from Dr. W. H. Ewing and with four single colony isolates from one of them. VITADIHS ANY £700 ‘7 NAAMLA NOLLVZIGTUdAH [2961 LP 472 are they another expression of the Lact property. In fact, about 5 per cent of the Lact recombin- ants are serologically unchanged. It is notable, although unexplained, that all stable Lact derivatives from the unstable Lact hybrids be- long in the serologically unchanged group. The linkage of the antigen-controlling factor to the Lac locus, used as a selective marker in our crosses, appears therefore to be a fortuitous coincidence. No other determinant of serological specificity has been detected yet in our rather limited range of crosses, using only a few strains of Shigella, a few selective markers, and rather crude serological tests. Further exploration is in order. Attempted crosses Salmonella x E. coli. The existence of coliform, Lact cultures which ferment lactose in 24 hr, have antigens identical to those of Salmonella newington (group E,, antigenic formula 3,15:e,h-1,6) and give rise to Lac variants (Seligmann and Saphra, 1946), suggested their possible origin as hybrids with E. colt. It seemed desirable to attempt crossing E. colt cultures F* or Hfr with some Salmonellas belonging to group E;. Cultures of S. cambridge, S. newington, S. new brunswick, S. kinshase, S. selandia, and of a Lac” variant of culture #3534 (Saphra and Seligmann, 1947) were made S™ and plated on EMB lactose streptomycin agar with strains of E. coli Lact S*, either Ft or Hfr. The results were uniformly negative? DISCUSSION Many bacteria included in the genera Eacher- ichia and Shigella have enough characteristics in common to suggest a close evolutionary relaticn- ship. They share, for example, susceptibility to certain phages and a common over-all pattern of catabolic reactions (although the fine details of catabolism and biosynthesis in Shigella remain largely unexplored). The main physiological dif- ferences, which are useful for practical purposes of classification because of stability and good cor- relation with presumed presence or absence of pathogenicity, include characters such as ability to utilize lactose. Yet, as stable as these characters * Equally negative results were obtained in a attempt to obtain recombinants between Esch- ertchia coli and a strain of Pasteurella pestis, kindly supplied by Dr. T. W. Burrows, who had observed that this strain and F! coli share sus- ceptibility to several phages. LURIA AND BURROUS [vou. 74 are in many strains, they are known to be con- trolled by genetic factors that are mutable in other strains. In the absence of hybridization tests, well established stable organisms can best be considered as “normotypes’”’ for purposes of classification. Our experiments have proved that Shigella can mate with Z. coli and shares with it a common system of mating polarities. By hybridization we can create hybrids that would be considered as monstrosities from the standpoint of traditional bacterial classification, such as strains of 8S. dysenteriae that promptly ferment lactose (or arabinose or both), of S. flexneri that ferment lactose (and xylose or rhamnose) and so on. Once the already suspected existence of extensive genetic homology between coli and dysentery bacilli has been confirmed by hybridization, these findings are not at all surprising. Indeed, it should be possible, and might be desirable for practical reasons, to decide by genetic analysis the reason for the apparent stability of char- acters such as Lac~ in most strains of Shigella. Such stability may reflect intrinsic gene proper- ties, or, more likely, the presence of multiple genetic blocks, or the absence of genetic loci (chromosomal deficiencies). The failure of certain characters of EF. colt parents to appear in hybrids suggests that the genetic homology is incomplete, and this may also be reflected in poor chromosomal pairing. Such poor pairing could underlie the low frequency of integration of EZ. coli genes into Shigella, in spite of the high frequency of mating revealed by zygotic induction. Similar observations have been reported in transformation experiments with Haemophilus; the frequency of integration of newly introduced genetic determinants is lower, the more distant the relationship between donor and recipient strains (Schaeffer, 1956). These ob- servations have also been interpreted as reflecting inadequate pairing between genetic structures. The availability of two methods of genetic recombination between FE. coli and Shigella, mating and transduction by phage Pl (Lennox, 1955), makes it certain that the taxonomic structure of this group could be placed on a sound genetic basis. In the present state of bacterial genetics, however, when the nature of the mating process has just begun to be clarified (Wollman et al., 1956) and the genetic structure of only one or two strains of EZ. coli is known in some detail, a 1957] detailed genetic analysis of any one Shigella would not seem too profitable. The establishment of a chromosomal basis for the evolution and taxonomy of this group of bacteria is a distant hope, although a definitely realizable one. We now know that strains of Shigella and Escherichia are potentially interfertile organisms. Therefore, three main questions arise: (1) How widespread is interfertility among other bacteria, at least among the Enterebac- tertaceae? (2) Does hybridization occur in nature, and if so, what role does the resulting gene flow play in the variability, survival, and evolution of these organisms? (3) What role, if any, does hybridization play in creating the range of practically important nor- motypes of Shigella that occur in nature? The answer to the first question, of the extent of potential interfertility among the bacterial groups, can only be guessed. Our few attempts to hybridize Salmonella and Z. coli have failed, and no positive results seem to have been reported from elsewhere. The criterion of cross-sensitivity to phages, which probably reveals genetic homology in view of the relations between pro- phages and bacterial chromosomes (Stocker, 1955), is of limited use since few phages affect bacteria belonging to different “genera” of Enterobacteriaceae (except for Escherichia, Shi- gella, and related groups). The existence of un- usual common properties (especially antigens) among strains classified in different groups may be a good guide to choice of cultures for further studies. These should take into account the possible existence of still unrecognized mating systems, with unsuspected polarities and re- strictions in hybridizing capacity. The question of the possible occurrence in nature of hybridization between Shigella and Escherichia, and possibly other groups, recalls suggestive observations by a number of workers. All experts in the diagnosis of enteric pathogens (Kauffmann, 1954; Weil and Saphra, 1953; Edwards and Ewing, 1955) have encountered cultures that appear to share properties of 2 (or more) well established normotypes and which could easily be explained by postulating a hybrid origin. To choose but a few examples, we may point first to entire groups of bacteria that fall into these “hybrid” classes: the Arizona group, which includes Suct, Lact organisms that might HYBRIDIZATION BETWEEN £. COLI AND SHIGELLA 473 be recombinants between Salmonella and E. coli; the Alkalescens-Dispar group, nonaerogenic, nonmotile organisms, including Lact ones, which might be hybrids between Z. coli and S. flerneri. Also some cultures (Seligmann and Saphra, 1946; Saphra and Seligmann, 1947) ferment lactose as promptly as Z. coli but are antigenically identical to known Salmonella strains and can, by a single mutation Lact — Lac, become indistinguishable from the latter organisms. Likewise, organisms of the Alkalescens-Dispar group are antigenically related to certain S. flexneri normotypes on the one hand, to typical Z. colt strains on the other hand. Individual strains with O-antigens in common with known coli, flexneri and Salmonella cultures have been described (Bernstein et al., 1941; Saphra and Wassermann, 1945). Coliform or- ganisms related to Shigellas (Ewing, 1953) have raised important questions concerning patho- genicity and diagnosis (Stuart et al., 1943). The experts have been unanimous in emphasizing the existence and importance of “intermediate’’ or- ganisms inbetween the major normotypes (Ed- wards and Ewing, 1955). So long as vegetative reproduction was thought to be the only re- productive process in bacteria, most authors concerned with bacterial evolution have inter- preted these intermediate organisms as variants fulfilling or recapitulating a mutational history of the Enterobacteriaceae. Yet, the possibility of a hybrid origin has not escaped some workers. For example, Saphra and Wassermann (1945) state: “. .. originally highly different forms might have a tendency of acquiring more and more similarity of antigenic properties. One might designate such a working hypothesis as a ‘hy- pothesis of convergent development’ in contrast to the ‘hypothesis of divergent development’ of White. Influences connected with the adjustment of life in the intestinal tract and possibly the in- fluence of one species upon another during their co-existence in the intestinal tract might favor such a development. The possibility that one species might impress its antigen upon another has been experimentally proven in the classical transformation experiments on pneumococci by Griffith and by Dawson.” Indeed, the normal habitat of Hnterobacteria- ceae, that is, the intestine of mammals and other animals, is quite conducive to hybridization, because of the presence of ubiquitous Z. coli 474 strains in populations of enormous size and of the chances for numerous collisions (and other in- teractions, such as phage transfer) among cells of these strains and of any newcomer organism. Our experiments establish the reality of the genetic interaction and justify the search for its occurrence in nature and for the role that the resulting gene flow may play in bacterial evolu- tion. This task has more than purely biological or genetic interest. It may hold the key to the whole epidemiology of enteric diseases; it adds many potential dimensions to the tremendously important work of tracing and identifying the enteric pathogens. Thus, for example, all our Lact hybrids, al- though they are predominantly “Shigellas” (and probably still pathogenic) would never be detected as potential pathogens by the routine diagnostic procedures. They would be discarded as “coliforms”? and might not even be tested for anaerogenesis, (which might be considered as evidence of a possible hybrid nature) except if isolated from cases of infantile diarrhea. Indeed, several of the “coliforms” supposedly responsible for diarrhea are related antigenically to some Shigellas (Edwards and Ewing, 1955) and may be hybrid strains. Also, for example, a coliform or- ganism antigenically related to S. flexneri type 2 was isolated from the stool of a patient who also yielded a typical, stable S. flexneri type 2 (Ewing, 1953, and personal communication). A systematic survey for hybridizable organisms and hybridi- zation phenomena in nature, carried out with controlled genetic methods, would be readily feasible and should be fertile of remarkable find- ings and applications. A question closely related to the above dis- cussion concerns the possible hybrid origin of some of the well known Shigella normotypes. Our serological findings establish that hybridiza- tion of S. flecnert type 2a or 2b with E. coli can bring about at least one major antigenic change, which is controlled by a genetic determinant near the Lac region of the coli-Shigella chromosome, and which produces a group of serotypes, the F22-like hybrids, similar to S. flexneri type Y. The association of the new serotype with the Lact property is clearly a fortuitous accident of genetic linkage. Lac~ hybrids with F22-like serotype would not have been detected in our crosses. Likewise, if Lact hybrids of such serotype occurred in nature, they would probably be identified as Shigellas only if they happened to LURIA AND BURROUS [vou. 74 mutate to Lac”. In view of the finding of one such major change in our limited range of experiments, it would be surprising if antigenic changes of some sort were not quite frequent following hybridiza- tion. Only a systematic study of the distribution of mating abilities among Shigella and Z. coli strains in nature can help evaluate this possi- bility. The similarity of the hybrid serotype to S. Jlexneri type Y raises interesting questions con- cerning the much debated significance of the Y organisms. According to Boyd (1938) and others (Edwards and Ewing, 1955) the Y strains should be considered as variants derived from various S. flexneri types by loss of the type anti- gens. Such variation has been observed in the laboratory, for example, in S. flernert types 4a (Boyd, 1938) and 2a (Ewing, 1954). Other authors (Weil and Saphra, 1953) consider these variants as different from S. flerneri type Y, which they report as possessing a type antigen of its own. - Our Y-like hybrids, derived from either 2a, 2b, or 4a type strains, cannot be simply loss variants, because they have antigenic relations qualita- tively different from those of the parent cultures; this is especially evident for the hybrid derived from S. flexneri type 2b. The genetic aspects of the phenomenon, namely its dependence on a genetic determinant linked with a specific region of the genome, indicate that the serological change reflects a new combination of antigen- controlling genes, probably due to the introduc- tion of coli genetic elements into Shigella. The relation of this variation by hybridization to similar spontaneous variations occurring in pure lines of S. flerneri remains unclarified. It is possible that both types of genetic changes reveal hidden antigenic potentialities of the strains of origin. There seems to be no definite evidence to show that the O-antigens of Shigellas consists of a mosaic of different antigenic determinants. The various antigens are defined only in terms of serological agglutinations and cross-adsorption tests. If the O-antigen (the carbohydrate-lipo- protein complex) of each organism owed its antigenic specificity to a single molecular species (Morgan and Partridge, 1940), then the various antigenic components, as defined by cross-re- actions, would be the expression of the differential affinities between the unique antigen of each organism, probably determined by several 1957] genes, and the various fractions of a population of antibody molecules that includes a whole spec- trum of configurations (Landsteiner, 1945). Each strain would react with, and could adsorb from, a serum those antibody molecules that fit it. The greater the proportion of such molecules in a serum, the stronger will be the cross-reaction and the closer the inferred chemical similarity between the antigens of the homologous and of the heterologous strain. Mutation or recombi- nation of genetic determinants could give rise to new antigens with new specificities. If, on the other hand, the O-antigens were a mosaic of different reactive sites, then mutation or recombination could act by changing one or more of these sites, while leaving the other un- affected. Such a mosaic structure cannot be proved by genetic tests, but only by the demon- stration of the existence of separable combining sites for different antibody molecules in the O-antigen of a given strain, either on the intact cell or in the extracted O-antigen complex. SUMMARY A number of Shigella strains were tested for mating ability with Escherichia coli. All the strains tested were fertile with F* or Hfr de- rivatives of E. coli strain K-12 and £. coli strain C. The Shigellas behave like F~ strains in the fertility system of Z. colt and, like E. colt F-, can be changed to the F* state by mixed cultivation with an F+ culture. Matings result in formation of hybrids, which exhibit new combinations of characters typical of Shigellas with characters typical of E. colt. The frequency of recombi- nation is lower than in similar crosses between strains of E. coli, although the frequency of mating, measured by zygotic induction of prophage A, is comparable in both types of crosses. Also, some of the characters of the coli parent fail to be transmitted to the EB. coli x Shigella hybrids. These results suggest an in- complete genetic homology between the two groups of organisms. Some hybrids between E. coli and type strains of Shigella flexneri possess somatic antigens closely related to those of S. flecneri type Y. This antigenic change is controlled by genetic factors closely linked to factors controlling lactose utilization. The results suggest a possible role of hybridization occurring in nature in the evolution of the Enterobacteriaceae and in the HYBRIDIZATION BETWEEN £. COLI AND SHIGELLA 475 origin of aberrant and intermediate strains of enteric bacteria. REFERENCES Berrani, G. 1951 Studies on lysogenesis. I. The mode of phage liberation by lysogenic Escherichia coli. J. Bacteriol., 62, 293-300. Bertani, G. anp Weie.e, J. 1953 Host- controlled variation in bacterial viruses. J. Bacteriol., 65, 113-121. Bernstein, S., Sappra, I., anp Daniets, J. B. 1941 The occurrence of Salmonella antigens in dysentery bacilli. J. Immunol., 42, 401-404. Boyp, J.S8. K. 1938 The antigenic structure of the mannitol-fermenting group of dysentery bacilli. J. Hyg., 38, 477-499. Cavatu, L, L., LEDERBERG, J., AND LEDERBERG, E. M. 1953 An infective factor controlling sex compatibility in Bacterium coli, J. Gen. Microbiol., 8, 89-103. CavaLii-Srorza, L. L. anp Jinxs, J. L. 1956 Studies on the genetic system of Escherichia coli K-12, J. Genetics, 54, 87-112. Conn, M. 1952 Production of antibodies in experimental animals, In Methods in medical research, Vol. 5. Year Book Publishers, Chicago, Il. Epwarps, P. R. anp Ewrna, W. H. fication of Enterobacteriaceae. lishing Co., Minneapolis, Minn. Ewine, W. H. 1953 Serological relationships between Shigella and coliform cultures. J. Bacteriol., 66, 333-340. Ewine, W.H. 1954 Mannitol negative varieties of Shigella flexneri serotypes. J. Immunol., 72, 404-410. pe Haan, P.G. 1954 Genetic recombination in Escherichia coli B. I. The transfer of the F agent to Z. coli B. Genetica, 27, 293-299. Hayzs, W. 1953a Observations on a transmis- sible agent determining sexual differentiation in Bacterium coli. J. Gen. Microbiol., 8, 1955 Identi- Burgess Pub- 72-88. Hares, W. 1953b The mechanism of genetic recombination in Escherichia coli. Cold Spring Harbor Symposia Quant. Biol., 18, 75-93. Jacos, F. anp Woutiman, E. L. 1956 Sur les processus de conjugaison et de recombinaison chez Escherichia coli. I. L’induction par conjugaison ou induction zygotique. Ann. inst. Pasteur, 91, 486-510. Karspr, A. D. 1955 A genetic study of the temperate coliphage A. Virology, 1, 424-443. KaurrMann, P. 1954 Enterobacteriaceae. 2nd ed. Ejnar Munksgaard, Copenhagen. 476 LANDSTEINER, K. logical reactions. Cambridge, Mass. Leperserc, J. 1947 Gene recombination and linked segregations in Escherichia colt. Genetics, 32, 505-525. LepERBERG, J. 1949 Aberrant heterozygotes in Escherichia coli. Proc. Natl. Acad. Sci. U.S., 85, 178-184. LEDERBERG, J. 1950a Isolation and characteri- zation of biochemical mutants of bacteria. In Methods in medical research, Vol. 3. Year Book Publishers, Chicago, Tl. LEDERBERG, J. 1950b The §-p-galactosidase of Escherichia coli strain K-12. J. Bacteriol., 60, 381-392. LeperBerG, J. 1951 Prevalence of Escherichia coli strains exhibiting genetic recombination. Science, 114, 68-69. LEDERBERG, J. AND Epwarps, P.R. 1953 Sero- typic recombination in Salmonella. J. Im- munol., 71, 232-240. Leversere, J. anp Tatum, E. L. 1946 Novel genotypes in mixed cultures of biochemical mutants of bacteria. Cold Spring Harbor Symposia Quant. Biol., 11, 113-114. LepERBERG, J., LepersenrG, B. M., ZinpEr, N. D., anp Livery, E. R. 1951 Recombination analysis of bacterial heredity. Cold Spring Harbor Symposia Quant. Biol., 16, 413-443. Lennox, E. S. 1955 Transduction of linked genetic characters of the host by bacterio- phage Pl. Virology, 1, 190-206. Lies, M., Weiae, J. J., aND KeLLENBERGER, E. 1955 A study of hybrids between two strains of Escherichia colt. J. Bacteriol., 69, 468-471. Luria, S. E. 1947 Recent advances in bacterial genetics. Bacteriol. Revs., 11, 1-40. 1945 The specificity of sero- Harvard Univ. Press, LURIA AND BURROUS [vou. 74 Morean, W. T. J. ano PartripGs, 8S. M. 1940 Studies in immunochemistry. 4. The frac- tionation and nature of antigenic material isolated from Bacterium dysenteriae (Shiga). Biochem. J. (London), 34, 169-191. Sapura, I. anp SetigMaNN, E. 1947 Coliforms with complete Salmonella antigens, or lactose- fermenting Salmonellas. J. Bacteriol., 54, 270-271. Sarura, I. AnD WassERMANN, M. 1945 Serologi- cal relationship amongst Salmonella and other Enterobacteriaceae. J. Immunol., 60, 221-227, ScuaErrer, P. 1956 Transformation inter- spécifique chez des bactéries du genre He- mophilus. Ann. inst, Pasteur, 91, 193-211. Scuazrrer, P. anp Ritz, M. E. 1955 Transfer interspécifique d’un charactére héréditaire chez des bactéries du genre Hemophilus. Compt. rend. Acad. Sci. 240, 1491-1493. Seviemann, E. anp Sapura, I. 1946 A coliform bacterium with the complete antigens of Sal- monella newington. J. Immunol., 54, 275-282. Stocker, B.A.D. 1955 Bacteriophage and bac- terial classification. J, Gen. Microbiol., 13, 375-381. Stuart, C. A., Rustraran, R., ZIMMERMAN, A., AND ConriGaNn, E. V. 1943 Pathogenicity, antigenic relationships and evolutionary trends of Shigella alkalescens. J. Immunol., 47, 425-437. Wert, A. J. anp Sapura, I. 1953 Salmonellae and Shigellae. Charles C Thomas, Spring- field, Ill. Wouiman, E. L., Jacos, F., anp Hayes, W. 1956 Conjugation and genetic recombination in Escherichia colt K-12. Cold Spring Harbor Symposia Quant. Biol., 21, 141-162.