Reprinted from tho Proceedings of the NATIONAL ACADEMY OF SCIENCES, Vol. 43, No. 12, pp. 1060-1065. December, 1957. SIBLING RECOMBINANTS IN ZYGOTE PEDIGREES OF ESCHERICHIA coLI* BY JOSHUA LEDERBERG DEPARTMENT OF MEDICAL GENETICS, SCHOOL OF MEDICINE, AND DEPARTMENT OF GENETICS, COLLEGE OF AGRICULTURE, UNIVERSITY OF WISCONSIN, MADISON, WISCONSIN Communicated October 19, 19671 Until recently, the sexual process in Escherichiu coli was inferred only from the genetic analysis of recombination, without morphological corroboration. Cavalli's discovery' of a highly fertile mutant (Hfr) has made it possible to relate genetic exchange to a pairwise conjugation of the mating bacteria. Lederberg2 observed and isolated conjugal pairs under phase-contrast microscopy and showed a high incidence of recombinant genotypes among their progeny. Anderson3 has made striking electron micrographs of conjugal pairs, in which one parent was labeled by the pre-adsorption of bacteriophage. The significance of conjugal pairing has VOL. 43, 1957 GENETICS: J. LEDERBERG 1061 been underlined by recent experiments on the timed interruption of the sexual process by mechanical shearing or by chemical inhibition.3 The present report amplifies a preliminary account2 of the segregation of various markers in single-cell pedigrees of isolated zygotes. Particular interest attaches to the occurrence of complementary recombinants, that is, whether crossing over is reciprocal, as in Neurospora and Drosophila, or not,, as has been suggested for bacteriophage. Cultures and Methods.-The parental strains were chosen as a highly fertile com- bination, differing in a number of easily scored markers which had been introduced by a sequence of mutations. W-3011, an Hfr mating type derived from strain K-12, is motile and relatively slender; W-2401, an F- mating type derived from another line (wg28A) of E. co& is nonmotile and plumper. The conjugal pairs are thus quite characteristic under the microscope. The isolation of pairs and of cell lineages followed de Fonbrune's techniques of micromanipulation.2v 4 Aft'er isolation, the mating pair was left undisturbed for 1 or 2 hours, by when it usually separated spontaneously. The exconjugant's were then isolated to separate micro- droplets and allowed to form large clones either directly or after re-isolation of the daughters for one to six fissions to establish cell lineages. These clones were t'hen studied for their genotypic content by platings on various indicator media. Markers and Notation.-Nine markers were scored in this cross (Table 1). TABLE 1 CHARACTERIZATION OF 75 EXCONJUGANT CLONES A. Sequence of loci according to incidence of clones containing the 1 allele (from Hfr parent). ** The 1 allele is indicated below, the 0 allele above. - 000 0000 00 Fill 2+ 111 1111 11 F- Hfr B. x x x 0000 00 clones* (total 65) I. Clones with One II. Total Incidence Recombinant of This Type? 100* 37 45 C. x x x x x x x 00 clones/l (represented individually) 1. 1110010 2. 111 1000; 111 1111 3. 111 1000; 1110000 4. 100 0100: 000 0100 5. 100 0110; 100 OOOO; 000 0110; 600 0000~ 6. 111 1111; 111 0000; 000 1111; 000 OOOO!: 7. 100 0000; 100 1100; 000 1100; 000 OOOO$ 8. 100 1100; 100 0010; 000 1100; 000 0010 III. Clones with Rlorc than One Recombinant$ 111 & 011 111 & 100 110 & 010 100 & 010 001;110;100;010 001; 100; 010 110;100;010 D. 000 0000 XX clones 1. 000 0000 10 2. 000 0000 01 * The terminal group 0000 00 is to be understood. Thus 100 should be read 100 0000 00. t This item includes the occurrences in parts B and C. $ This class may not always have been ascertained. jJ The terminal group 00 is to be understood. Thus 111 1OOOshould be reed 111 1000 00. $ The 000 0000 00 class ( Q parent) was present in every exconjugant clone. It is written explicitly to complete possible tetrad& ** The sequence S Ma1 Xv1 is preferred &8 minimizing apparent multiple crossing-over. 1062 GENETICS: J. LEDERBERG PROC. N.A.S. Lac, Ara, Mal, Xyl, Mtl, and Gal refer to the fermentation of lact'ose, L-arabinose, maltose, n-xylose, mannitol, and galactose, respectively; VI and S refer to t,he reac- tions (sensitive versus resistant') to phage Tl and to strept,omycin. FZa refers to flagellation, scored by mot,ilit,y. However, for the purpose of this analysis, the phenotypic status of each recombinant is less important than the parent'al origin of the marker alleles. For simpler exposition, the allele from the F- parent is designat'ed 0, and its alternative from the Hfr parent, 1. For a marker not segre- gating or for another reason not specified in a given genotype, the designation will be x. Deleted segments are indicated as O O . The loci are written in the order specified in Table 1. This convention having been established, the genotypes are written as binary numbers; for example, 000 1111 00 represents Lac- VI8 Ara- 8" Mal+ Xyl+ Mtl+ Gal+ Fla-, the first three and the last two of the markers having stemmed from the F- parent, the middle group of four (8, Mal, Xyl, MtZ) from the Hfr. Specific segments may be marked off by underscoring; thus 000 0000 00 refers to the Ara-S segment. Hfr versus F- represents another marker difference but was not systematically scored. However, some dozens of recombinants were all F- with the single excep- tion of Dl (000 0000 10) which was Hfr, in accord with Cavalli and Jinks's demon- stration of linkage of Hfr to GaL5 Both parents were sensitive to the temperate phage lambda, except hhat the Mal- marker was associated with poor adsorption of the phage.6 Experimental Results and Discussion.-Table 1 exhibits the recombinant prog- eny from 75 conjugal pairs. About 125 other pairs that were isolated and gave a viable F- exconjugant produced no detect'able recombinants. The Hfr excon- jugant was usually viable as well; however, all the recombinants appeared in the progeny of the F- exconjugant. This asymmetry implies t'he unilateral transfer of genetic material from the Hfr to the F- cell and justifies the description of t,he two mating types as 3 and 9, respectively.' However, Hfr X Hfr crosses are also fertile, though F- X F- are not. The Hfr is therefore ambivalent, but in crosses with F- (obligate 0) it can function only as a 3. By analogy with the mating systems of other fungi, obligate 3 c7 may be supposed to exist but have not yet been notic,ed in E. coli. In addition to recombinants, each F- exconjugant contained a proportion, usually more than half, of the F- parental genotype, viz., 000 0000 00. In pedigreed clones, at the second fission, usually three of the cells were pure 000, etc., while the fourth was still generating both parental and recombinant genotypes, these be- coming pure in cells at the third or fourth fission. Figure 1 shows an instance of prolonged delay in segregation, with recombinants in three of the subclones at the four-cell stage. These results are readily understood by the regular assortment of t'hree unfertilized from one fert'ilized nucleus and its meiotic products in the lineage of the multinucleate F- exconjugant, complicated by occasional slippage and ir- regular assortment of nuclei. Ko persistent het,erokaryons were observed in this study, nor could they be expected in a short bacillary form, whereas they are ex- pected and are found in the filamentous streptomycetes.8 The various markers in Table 1 are arranged according to their segregation ratio, i.e., the relative incidence of the 1 allele among the exconjugant clones. As shown, this incidence varied from 1 each for Gal and Fla to 70 for Lat. This disparity con- VOL. 43, 1957 GENETICS: J. LEDERBERC 1063 tradicts the assumpt,ion of random segregation of unselected markers which was postulated in the initial analysis of segregation and linkage in E. c01i.~ The dis- parity does accord with the bias found in unselected progenies from persistent hetero- zygoteslO, l1 and is now explained by the loss of segments from the fl chromosome, either before fertilization,3 or after," or both. The 1 allele of each of the 9 markers of these parents has been recovered at least once, i.e., the 8 parent can transmit any element. The CT exconjugants of ex- ceptional zygotes such as Dl, D2, and others were examined for a persistent aher- ation in the segregation pattern of further crosses, but none was found. Within this series, there was no instance (barring one creditable to a possible contamination) of the reappearance of the whole 111 1111 11 genotype in the progeny of a 9 excon- jugant. Whether or not the entire 111, etc., genotype is transmitt,ed to the zygote, it therefore experiences either obligatory recombination or elimination of a chro- mosome segment at some stage, be this before or after fertilization. FIG. I.-Cell lineage from a conjugal pair ((28). The lineage is abbreviated: where a subclone was pure, only the pure parent cell is shown. Genotypes (see Table 1): fl = 111 1111 11. 0 = 0 = 000 oooo 00. 1 = 000 1100 00; 2 = 100 0010 00; S = 100 1100 00; 4 = 000 0010 00. t = died. Most of the clones contained only one detectable recombinant. However, as some pedigreed cells were inviable from technical mishaps or from no overt injury, this class may be overestimated. The incidence of these single-recombinant clones tells against the frequent occurrence of double fertilizations. The mating complexes chosen for isolation are, however, selected as simple pairs, and multiple fertilizations could still occur in the more complicated clumps which are also seen. The distribution of types in part BIII suggests considerable coincidence of inter- changes. Thus we find six clones with 111 alone, just one with 011, but seven with both together. Part C also points to an interaction between recovery of 111 and of further markers. That is, crossing over in the 000 0000 00 region is correlated - with crossovers in other regions. This type of coincidence has been stressed in earlier reports6 and can again be interpreted as a result of sporadic pairing, either 1064 GENETICS: J. LEDERBERG PROC. N.A.S. from poor synapsis in a complete heterozygote, or from the prezygotic loss of some chromosome segments. Twenty clones had from two to four recombinants, none more than four, which suggests a regular mechanism of meiosis with a four-strand stage. Clones C5 through C8 are especially striking, as they comprise four different tetratype tetrads with partly complementary elements. For example, C6 includes the complemen- tary set: 111 1111; 000 0000; 111 0000; and 000 1111. Another example of com- plementarity, viz., the reciprocal crossovers X" Mal+ Xyl- and S' Mal- Xyl+, is seen in C8. However, the suggestion that crossing over is inherently reciprocal must be qualified in view of the small numbers and the possible role of coincidence. The frequency of clones with three or four recombinants poses an additional problem. Of the four tetratypes, three show the same crossover, 100, etc., in two combinations. For example, C8 has 100 1100 and 100 0010. Three possibilit,ies are forwarded: a four-stranded heterozygote, with intense coincidence of crossing over; crossing over at a two-st'rand stage for some markers, followed by typical four-strand crossing over for others; or reiterated meiosis. As the last formulation also accounts for another anomaly, it will be set forth in more detail. The abbreviated notation of Table 1 should not obscure the fact that no one clone was segregating for every marker. Even C6 is pure Gal+ Flu-, while C5 is pure for Ara, VI, and MtZ as well; C7 for Ara, VI, Xyl, and Mtl; and so on. Some mechanism must be postulated to furnish 0 alleles for these recombinant genotypes. One scheme may be illustrated with C8 as follows: 1111 1 I 000 0000 00 primary heterozygote 111 1111 11 111 X + 000000000 *first meiosis + 100 1100 00 plus 000 0010 00 plus 1110011 Cl0 plus 011 1111 on. The last two are inviable on account of the terminal deficiency, which stems from breakage proximal to the two markers. The two viable products rejoin to give the secondary heterozygote 100 1100 00 000 0010 00' Further reduction, with crossing over at a four-strand stage in the region 100 1100 00, gives the four products: 100 1100 00; 000001000; 100001000;and0001~000. The secondary heterozygote can be formulated by an alternat,ive scheme, on the supposition of prezygotic loss of segments to give a fractional heterozygote (mero- zygote3) : 111 1111 11 000000000~ loo 1110 00 x __ 000 0000 00 - The merozygote would then be completed by partial replication from the 000, etc., template to give 100 1110 00 ooo 0000 00 followed, or accompanied, by crossing over to give 100 1100 00 000 0010 00 ' VOL. 43, 1957 GENETICS: J. LEDERBERG 1065 which is then reduced by a regular four-strand meiosis. However, as part'ial repli- cation can be posited as the mechanism of crossing over, the two versions are for- mally similar and cannot be distinguished in the present experiments. On both versions, crossing over takes place between Ma1 and Xyl during the primary cycle. The formation of secondary heterozygotes has been postulated previously'" to account for demonstrable homozygosity of Lac in some persist'ent diploids and for "twin prototroph" recombinants. Sporadic automixis is characteristic of persistent diploids'O and heterogenotes.12 Iterated meiosis is therefore a recurrent feature of recombination in E. coli. l3 Summary.-Individual mating pairs were isolated from highly fert,ile matings of E. coli, Hfr X F-. All the recombinants were found in the progeny of bhe F- exconjugant, which is therefore designated as 0. The segregation of markers in the cell lineage indicated (1) the loss of certain chromosome segments, before or after fertilization or both; (2) coincidence of crossing over, suggestive of sporadic pairing; (3) possible correlation of reciprocal crossovers; (4) recycling of meiosis; and (5) an interval of nuclear assortment from a temporary heterokaryon. * Paper No. 670 from the Department of Genetics. This work has been supported by grants from the National Science Foundation, the National Cancer Institute, Public Health Servire (C-2157), and the Research Committee of the Graduate School, University of Wisconsin, with funds allorated by the Wisconsin Alumni Research Foundation. t From Melbourne University, Victoria, Australia. I am indebted to Professors S. D. Rubbo and Sir Macfarlane Burnet for their hospitality during the tenure of a Fulbright lectureship. 1 L. L. Cavalli, Bol. Ist. Sieroter. Milan, 29,281-289, 1950. 2 J. Lederberg, J. Bact., 71,497-498, 1956; J. Cell. Comp. Physiol., 45, Suppl. 2, 75-107, 1955. 3 E.-L. Wollman, F. Jacob, and W. Hayes, Cold Spring Harbor Symposia Qua&. Biol., 21, 141-162,1956. 4 J. Lederberg, Genetics, 41,845871,1956. 6 L. L. Cavalli and J. L. Jinks, Genetics, 54,87-112,1956. 6 E. M. Lederberg, Genetics, 40,581-582,1955. 7 W. Hayes, Nature, 169, 118, 1952. 8 S. G. Bradley and J. Lederberg, J. Back, 72,219-225, 1956. 9 J. Lederberg, Genetics, 32,505-525, 1947. 10 J. Lederberg, E. M. Lederberg, N. D. Zinder, and E. R. Lively, Cold Spring Harbor Symposia Quant. Biol., 16,413-433, 1951. I1 T. C. Nelson and J. Lederberg, these PROCEEDINGS, 40,415-419,1954. I2 M. L. Morse, E. M. Lederberg, and J. Lederberg, Genetics, 41,758-779,1956. I3 T. F. Anderson and R. Maze, Ann. Inst. Pasteur, 93, 194-198, 1957, have reported on 11 zygote pedigrees from a similar analysis with results consistent with those presented here, in- cluding evidence of "successive recombination" or meiotic recycling. The absence of the other- wise frequent class 100 0000 from the progeny of C8 suggests that this recycling is confined to a specific synkaryonide and does not involve further matings with the remaining 0 nuclei in the same cell (figure 1).