Reprinted from Generics, Vol. 37, No. 5, September, 1952, pages 469-483. ALLELIC RELATIONSHIPS AND REVERSE MUTATION IN ESCHERICHIA COLI ESTHER M. LEDERBERG! Department of Genetics, University of Wisconsin, Madison, Wisconsin Received February 7, 1952 ANY aspects of mutation in bacteria have been studied by mutational patterns, biochemical diversity, and other phenotypic determinations without recourse to breeding test (Luria 1947). The recombination tech- niques that have been developed in Escherichia colt K-12 (Tatum and Leprr- BERG 1947) permit a more incisive analysis of the bacterial genotype. A series of lactose-negative (Lac~ ) mutants in strain K-12 which revert at different rates has provided experimental material for the study of reverse mutation, the genetic factors controlling the different mutation rates, and allelic multi- plicity at the Lac, locus. MATERIALS AND METHODS The intercrossable cultures used in these experiments were ultimately derived from the two auxotrophic lines of K-12: 58-161 (B-M~—, biotinless, methionineless) and Y-10 (7~L7-Th~, threonineless, leucineless, thiamine- less) in which each nutritional factor had been obtained in a discrete muta- tional step (Tatum 1945; Tartum and Leperserc 1947). Liquid cultures were grown in Difco penassay broth, while stocks were maintained on nutrient agar slants. The plating media used consisted primarily of EMB (a complete eosin-methylene blue agar) and EMS (a minimal modification of EMB), con- taining one percent sugar. The media were prepared from dry mixes to which water, the appropriate sugar, and other supplements were added before sterili- zation in the autoclave. On EMB and EMS agar fermenting colonies are opaque purple while nonfermenters are lightly tinted. Thiamine was often added to the minimal crossing media to relax selection for Th+ (see LEDER- BERG 1947). Details of media and various methods are given elsewhere (LEDERBERG 1950a; LEDERBERG et al. 1951). Crosses were usually carried out on EMS lactose agar for the direct scoring of fermentative characters by inspection of the colonies. After 48 hours incubation, the prototroph progeny were picked and streaked on EMB lactose to provide purified colonies for further tests. A 125 watt Hanovia high pressure mercury vapor arc emitting a variety of wave lengths was used for ultraviolet exposures because of its high in- tensity output. Most of the irradiations were made directly on EMB lactose agar plates prespread with a drop of a 20 hour broth culture containing about 1 Part of this study was conducted during the tenure of a National Cancer Institute Predoctorate Research Fellowship. Paper No. 488 from the Department of Genetics, University of Wisconsin. GENETICS 37: 469 September 1952. 470 ESTHER M. LEDERBERG 10° cells per ml. Each plate was exposed for 7-10 seconds at a distance of 15 em, resulting in approximately six decades of killing. The irradiations were conducted in this fashion for the empirical purpose of securing a variety of fermentation mutants. In earlier experiments many lactose-negative mutants were intercrossed. It soon became apparent that a number of distinct loci were implicated among the different mutants which shared the lactose-negative phenotype (LEDER- BERG 19482). The most common locus was designated Lac,~ and was selected for the present study. The principal stocks are summarized in table 1. The TABLE 1 List of stocks. Phenotype on TUL Th7 Genotype lactose Source Agent Y-10 Lact wild type 679-680 X-ray Y-53 Lac,~ semi-mutable Y-10 U-V on lactose agar Y-70 Lac, stable Y-53 U-V on lactose agar W-112 Lac. stable Y-10 U-V on lactose agar W-133 Lacy mutable Y-10 U-V on lactose agar W-716 Lact unstable Y-70 spontaneous W-844 Lac” mutable W-716 spontaneous W-888 Gal,~ Lac slow W-750 x W-588 segregation from Lac” Gal~/Lact Gal* diploid W-902 Gal,~ Lact Y-10 U-V on galactose W-1282 Lac,” semi-stable Y-53 spontaneous BU M— Y-40 LactVIF wild type 58-161 selection on Tl Y-87 Lacy mutable Y-40 N-mustard W-516 Lac,” Pur™ stable Y-87 U-V on lactose agar W-744 Lac,~ Mal#! stable Y¥-87 U-V on lactose agar W-750 Lac,” Gal,7 stable Y-87 U-V on lactose agar W-811 Lac,” Gal,~ stable Y-87 U-V on lactose agar W-842 Lacy stable Y-87 spontaneous W-1306 Lac’ V6r wild type Y-40 selection on T6 W-1307 Lac,” V6tF mutable Y-87 selection on T6 W-1435 Lac,” Mal~ Mth” V1FV6F_ semi-mutable sees segregation from H-1; then selection on T6 (see Lederberg, 1949) BtMtT+Lt W-946 Lac,” Gal,~ semi-mutable W-902 Y-87 recombination Lac,— designation means that preliminary tests revealed no lactose-positive recombinants in crosses of pairs of Lac,~ mutants. In addition to color differences of colonies on EMB or EMS lactose agar, Lact and Lac~ can be distinguished in Durham tubes containing a lactose- peptone broth with bromeresol purple indicator. The change from a purple to a yellow color and the evolution of gas as end products in fermentation are observed for Lac+ inocula. Lac~ cells give a corresponding turbid growth but the original alkaline color persists. A test of greater precision affords a measurement of the adaptive “ lactase,” 8-b-galactosidase (LEDERBERG 1950b). Nitrophenol is measured photometri- ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 471 cally after a short period of incubation of the chromogenic substrate, o-nitro- phenyl 8-p-galactoside with washed cells. The capacity to produce an adaptive lactase after growth in lactose broth has been lost in Lac~ mutants. However, a small residual activity is characteristic of Lac,~ cultures. The lactose analog, n-butyl-@-p-galactoside, can be fermented by this group of mutants and can evoke the adaptive formation of galactosidase in amounts comparable to that formed by Lact. It may be noted that this substrate appears to select for Figure |.—-Papillating colonies from EMB lactose agar streak plates. spontaneous lactase-negative mutants, presumably by virtue of the toxicity of enzymatically released butanol. PAPILLAE AND REVERSE MUTATION Several Lac~ strains develop outgrowths or papillae within the colonies, or marginal sectors, of the typical Lact color on EMB lactose agar (figure 1). Stable Lac+ cultures indistinguishable from type are obtained from these papillae. Such papillate forms (“ Bacterium coli mutabile”) provided one of the earliest recognized and most cited examples of mutation in bacteria (Massrnt 1907; Happow 1937; Monop and Aupcreav 1946; Luria 1947). 472 ESTHER M. LEDERBERG The Lac~ papillae may represent reverse mutation to the wild type allele or mutation at another “suppressor or “mimic” locus. To discriminate between these alternatives it is necessary to isolate the reversions and cross them with a wild type Lact strain. If the Lac+ derived from papillae are mimics of wild type, the progeny will include Lac recombinants, their num- ber depending upon the closeness of the linkage between the Lac; and the mimic locus. Accordingly, a single papilla was fished from each of 21 colonies of the mutable strain, Y-87 (Lac,—”). Each of the secondary colonies was carefully purified by successive streakings to rid them of adherent Lac— cells, and five more successive single colony transfers were made after homogeneity was established. Each reversion was crossed several times with the appropriate Lact (Y-10) on EMS lactose thiamine agar until about 1,000 recombinants TABLE 2 Papillation frequencies for Lac,” mutants, : No. of Mean No. of Chi-square* ays Strain _ colonies papillae and DF Probability W-1282 100 0.61** 0.05; 8 Y-53 151 Il 3.392 -18 95 2.4 3.334 “51 100 2.4** 10.44 -04 97 3.0 3.185 -68 289 4.0 18.475 -002 125 4.4 3.226 -90 Y-87 109 4.0 11.456 -76 204 6.1 5.498 -70 100 14.0** a wees “For goodness of fit to a theoretical Poisson distribution of the same mean with n—2 degrees of freedom. **Same experiment, had been examined. Eventually, a total of 31,523 recombinants had been ob- served. One stable Lac~ exception was found in the cross Y-87R5 x Y-10. The cross was repeated but no other Lac~ appeared, and the exception might well have been a spontaneous mutation. In view of the infrequency of Lac— among these prototrophs, the change of Lac~ to Lac+ in Y-87 most probably constitutes a reverse mutation at the Lac, locus. Similar isolations of papillae were made from single colonies of Y-53 but the genetic tests were not as exhaustive. None of the 4,784 recombinants of the derived Lact with Y-40 was distinguishable from wild type. In contrast to the mutable Lac~ strains, no reverse mutants have been demonstrated for the more stable strains, even after prolonged selection. Twenty derivatives isolated from old cultures of W-112 (Lac,—**) on lactose agar or broth were all slow lactose fermenters (/ac*'). Three classes of proto- trophs were recovered in crosses of Lac with Lact (Y-40): Lact, Lac*, and Lac~*'. The Lac*! types are therefore not reversions of Lac, but partial ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 473 mimics or suppressors. During this study several other suppressors of Lac,~ mutants encompassing a wide range of fermentative ability were discovered. The absence of mimic mutations among the mutable strains Y-87 and Y-53 is only apparent. The frequent and early production of reverse mutants in Y-87 and Y-53 would exclude the development and detection of the less fre- quent suppressors, especially when the reversions ferment lactose more vigor- ously. Where reverse mutation does not occur, the selection of Lac*! mimics is facilitated. On two separate occasions a colony of Y-70 and W-112 yielded a single normal Lact papilla at the 48 hour stage. One of these carried a coincidental mutation for parathiotrophy. This culture, W-716, proved to be an extremely unstable Lac+. When maintained on nutrient agar stock slants, it gave rise to a variety of Lac and Lac*' types. Lactose agar delayed but did not prevent the accumulation of Lac~ cells. The Lac~ colonies themselves were unstable, including a wider divergence of distinct patterns than had previously been encountered. Of 1300 colonies examined in one derived culture, W-719, five contained a single papilla representing probably the lowest perceptible muta- tion rate among these cultures. The colonies of a second derivative, W-844 (figure 1}, contained an average of well over 25 papillae so that counts were unreliable. The Lac+ papillae isolated from any of these forms resembled W-716 both in their instability and the constellation of Lac~ subtypes. The genetic position of W-716 has not yet been satisfactorily settled due to the extreme difficulty in separating mutant from recombinant Lac~. The standard test for reverse mutation was carried out with the 6 marker (re- sistance to phage T6, closely linked to Lacy, LEDERBERG 1947) in the wild type tester (W-1306) to aid in the identification of recombinant prototrophs. About 2 percent of the latter were Lac~. The stable Lact recombinants were 16"; the unstable 176", thus showing a close correlation with the parental coupling. Five single colonies of both these types were transferred to nutrient agar slants and sampled about a year and a half later. Lac~/Lac* mixtures were found only in the populations classified as 6%. This linkage locates the unstable mutant W-716 at or near the Lac, locus whose expression is modified. MUTABILITY PATTERNS The rate of formation of lactose-positive papillae has so far been the most characteristic index of subdivision among these otherwise phenotypically indis- tinguishable cultures. Representative examples of four established categories are, in order of revertibility : Y-70 and W-112, W-1282, Y-53, and Y-87 (fig- ure 1). The frequency distribution of Lac+ papillae was studied in this group of strains. Platings from appropriate dilutions of either 24-hour colonies on agar or broth cultures were made so as to give no more than 15 evenly dis- tributed Lac~ colonies on each plate. Lact papillae began to appear after 36-48 hours of incubation and continued to emerge as the colonies aged. The classification of colonies with 0, 1, 2, ...n papillae was recorded at the 48 474 ESTHER M. LEDERBERG TABLE 3 Test for recombination of mutability: Lact x Lac~ _ Lact x Lac™™ Lact x Lac™8t Cross Y-10 x Y-87 Y-40 x W-112 Y-40 x Y-70 Total number of prototrophs 3,188 1,097 3,905 Percent Lac” 23.2 61.3 71.3 Number Lac™ tested for papillation 551 305 105 Percent Lac™ stable 0 100 100 hour stage. Under these conditions, the observed numbers of papillae per colony often showed a good fit to a theoretical Poisson distribution with the same mean (table 3, figure 2). The data are consistent with the hypothesis of equally probable and independent mutations in every mutable colony. Deviations from the theoretical distribution were also observed, especially after 48 hours. An individual strain also showed fluctuation about the mean 60T 50} W- 1282 a*06 40h 30 FRE QUENCY 20F I 4 —_ ° 2 4 6 8 NUMBER OF PAPILLAE Figure 2.—Theoretical frequency distribution of papillae expressed as numbers of colonies with 0, 1, 2,...n Lac*mutants for the observed mean in mutable Lac” strains. The points represent the observed values. 4 ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 475 from experiment to experiment with occasional wide discrepancies from the Poisson distribution. These disparities probably reflect uncontrollable varia- tions in local environment, and correlations between changes in growth rates of the Lac~ mother cells, the mutation rate, and the capacity of the mutants to be manifested as papillae. Nevertheless, each strain retained its relative position in the mutability series. Intercrosses were made between mutable x stable and mutable x semistable to ascertain whether a clear-cut genetic difference was involved. Among the 220 Lac prototrophs from B-AI~Lac~ mutable (Y-87) x T—-L~Lac~ stable (W-112), there were only two sharply defined classes of segregants, resem- bling the papillation patterns of the parents. Sixty-six percent were stable, which is in fair agreement with the proportions expected from the established linkages of the nutritional markers and Lac,. Crosses of the intermediate l ! l ! J I 20 25 30 TWO- DAY SELECTION PERIODS Fictre 3.—Fluctuations in the maximum number of papillae per colony of Y-53. MAXIMUM NUMBER OF PAPILLAE ° w 3 a alleles, Y-53 or W-1282 x Y-87 types, yield results more difficult to interpret. Individual nonpapillate prototroph colonies right represent either a geneti- cally stable recombinant or the extreme class of a random distribution. Within this limitation, the recombinants all concorde1 with one or the other parent. Spontaneous changes in mutability patterns have all led to a reduction rather than enhancement of papillation. Stable, nonpapillate colonies have been de- ° rived from Y-53 and Y-87. W-1282 of spontaneous origin occupies a muta- bility level intermediate between the parental Y-53 and the completely stable types. In an attempt to augment the mutability of Y-53 by recurrent selection, serial platings were made using the colony with the most papillae among 50- 200 colonies at each two day selection interval. This selection was extended over thirty such stages, but no consistent trend in mutability was observed. Figure 3 illustrates the stability of the papillation pattern of this strain. Crosses of the various Lac— with Lact were carried out to establish whether the strains constitute a series of multiple alleles at the Lac, locus or 476 ESTHER M. LEDERBERG whether a mutability modifier is involved. The latter hypothesis predicts the occurrence of both stable and mutable recombinants among the Lac~ progeny. Because only parental types have been recovered in such crosses, the argu- ment that the differences in mutability reside within the locus is favored (table 3). Before the conclusion can be drawn that the alleles differ in their intrinsic revertibility one further possibility must be considered. Lact reversions may occur at a constant rate but the differences in papillation reflect differences in the selection and expression of Lac* mutants. This interpretation may be tested by determining whether large populations of Lac~*! mutants are able to suppress the appearance of small numbers of Lac? cells when inoculated together in lactose medium. In nutrient broth where there is no obvious basis for selective differences, the growth rates of Lac~** and Lac~* are indis- tinguishable. Six lactose indicator broth tubes were inoculated with a mixture of approximately 2 Lac~ and 3 x 10% Lac~** (W-842) cells. Each tube turned yellow after 48 hours as a result of the development of a lactose-fermenting population. Lac> cells therefore compete effectively against preponderant populations of Lac—*t in media favoring the growth of Lac~. Similar conclu- sions were reached by KrisTENSEN (1944) in a more elaborate study. INDUCED ALTERATIONS OF MUTABILITY The mutability of a gene may be controlled by other loci (RHoADEs 1941). To discover genic regulators of mutation rate, large populations of Lac,~ cells were subjected to ultraviolet irradiation. The surviving colonies were inspected for deviations in papillation pattern. Mutable strains were examined for non- papillating whole colonies or sectors, and similarly, Lac™ sectors or papillae were looked for among stable strains. The suspected mutants were isolated, purified, and checked for the genetic markers of the parent strain. Systematic study of over 33,000 survivors from a series of irradiations of the stable strains (Y-70, W-112, and W-842) yielded no examples of either mutable colonies or reverse mutants. On the other hand, stable substrains were obtained from treated populations of the mutable Y-53 and Y-87. The induced stable strains of Y-87 were studied in greater detail. Outcrosses with Lact (Y-10) separated the new strains into two groups: one involving allelic or pseudoallelic changes in mutability (1) and the other consisting of forms in “which a second locus was involved (II). The basis of the classification de- pended upon whether only parental types of prototrophs were recovered (I) or whether Lac mutable colonies also appeared among the progeny (IT). The majority of the mutants belonged to the first category. They were further characterized by the retention of fermentative activity on butyl galacto- side and by stability on lactose throughout the manipulations to be discussed below. Each member of group II (Lacy~X~) failed to ferment butyl galacto- side. The capacity for reverse mutation could be restored by replacing the modifier with ¥* by recombination or mutation. ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 477 In most of the derived stable forms of this group, the “ X ”-modifier exhibited its own phenotypic effects that facilitated its characterization. Alter- natively, the modifier could be extracted in combination with Lact. The re- synthesis of Lac~"N~, detected as stable recombinants in progeny tests of possible LactX~ x Lac~™X+ identify the extracted modifier. In order to accomplish the repeated crossings, advantage was taken of auxotrophic segre- gants from persistent diploids (LEDERBERG 1949). One type of mutability modifier was associated with a nutritional require- ment for purine plus thiamine in W-516 (Pur), The components of this combination were not separable by reverse mutation. Papillation was restored in Purt reversions. The growth of Pur~ colonies was restricted on EMB medium which is deficient in nucleic acid components. Moreover, glucose itself is less rapidly fermented by W-516 than Y-87 grown on such media. A study of the relationship of Lac,;-”™ and Pur~ disclosed that the effects on muta- bility were indirect. The LactPur- recombinants obtained in crosses of Y-10 x W-516 were fermentatively weak. The inability of a Lact Pur~ geno- type to express itself normally is a clue to the absence of Lact papillae in W-516 colonies. It is reasonable to conclude that the Pur— gene depresses glycolysis, consequently removing the selective advantage of Lact mutants as confirmed in reconstructed mixtures of Lact Pur~ and Lac~Pur—. There ts no evidence that the mutation rate itself has been altered by the introduction of Pur into Y-87. The Pur mutation has been noticed in many subsequent irradiations of Lact and Lac~ as a small, thin colony. A similar situation was found in W-744, where the second mutation showed reduced fermentation of maltose, mannitol, and galactose. The coincidence of a Gal mutation in a Lac~ stable strain derived from Y-87 was first recognized in W-750. Verified Galt+ reversions were recovered as papillae on EMB galactose agar after several days of incubation. These Lac~Gal* derivatives again produced papillate colonies on lactose, showing that the replacement of Gal+ for Gal restored the mutahility of Lae~ to Lact. The Lact papillae were isolated and confirmed as back-mutants of Lac,~. Other galactose fermenters were found which were stable on lactose. Backerosses to wild type proved these exceptions to be mutations at mimic loci rather than reverse mutants of Gal~, Alleles at this locus which did not interact with Lac,— were not found. All four combinations of Lact’- and Gal~’~ were recovered from the cross W-750 x Y-10. Among 41 Lac7 all the Gal~ were stable, while Gal* papillated on lactose. Preliminary linkage tests place this Gal locus between (BM) and Lacy. The Gal~ mutation was extracted from the double mutant by crossing W-750 with a suitable Het stock (J_EDERBERG 1949), Auxotrophic Lac* Gal~ segregauts for use in crosses were obtained via a persistent diploid. In the cross Y-87 x W-888 (Lac~Gal~ x LactGal—), which may be con- trasted with W-750 x Y-10 (Lac7Gal~ x Lac+Gal*), stable Lac~ reappeared as Gal~Lac~ recombinants. Reversion to Gal* restored the revertibility of 478 ESTHER M. LEDERBERG these prototrophs on lactose agar. The same types appeared in crosses of Y-53 x LactGal-. The Gal might influence the mutation rate of genetically susceptible alleles at the Lac, locus or it might affect the ability of Lac™ mutants to express themselves as visible clones in the mother colony. The recombination analysis cannot decide between these alternatives. The second interpretation, however, is compatible with physiological tests of the genotypes. The Gal~ allele slightly modified the appearance of Lac~ toward Lact and the Lact somewhat to- ward slow fermentation. Galactosidase assays were made on cells of various genotypes grown in a lactose-peptone broth. The values per unit dry weight, expressed in relation to wild type K-12 as 100 percent are: Y-87 (Lac, Galt ) 3 percent; W-750 (Lacy~Gal,—) 20 percent; W-946 (Lac,~Gal,—) 18 per- cent; W-811 (Laci~Galy— ) 25 percent; W-888 (Lact Gal,~) variable, 50- 100 percent ; W-902 (Lact Gal,— ) 90 percent. The galactosidase assays of the Lec combinations correspond reasonably well with their appearance on EMB lactose agar. The Lac+Gal~ types, however, appear much weaker than these assays would indicate. A loss of efficiency of lactose fermentation due to the nonfermentability of the galactose residue may explain why less acid is pro- duced from lactose in lactose broth indicator tubes, despite the retention of lactase activity. The accumulation of monosaccharide during the fermentation of lactose by LactGal~ but not Lac+Gal* cells has been demonstrated by Barfoed's monose assay method (J. LEDERBERG, unpublished). These results suggest that the competitive advantage on lactose of Lact over Lac— is modified in Gal— genotypes. This possibility was explored by reconstitution of the selective conditions under which Lact papillae were observed, LactGal~ introduced into lactose broth, although outnumbered 10 to 100 million-fold by Lac~Gal~ cells, were detected after 2448 hours, show- ing intense selection for Lact, Comparable inoculations with Gal~ strains failed to demonstrate any preponderance of Lact within that period unless they had been present initially in a ratio of 1 Lact: 1000 Lac~. The specificity of Gal~ genes on apparent mutability of Lac~ was studied with several nonallelic Ga/~ mutants. One of these was obtained originally as a semistable colony from ultraviolet irradiation of Y-87. This Gala~Lac~ had a papillation pattern resembling Y-53. Gal,~ and Gals~ gave stable Lac~ recombinants similar to Gal,— when crossed with Y-87. Gal,~ also suppressed the apparent mutability of a papillate stock carrying a mutation at another Lac locus (Lac7— ). Combinations of Lacy~ Gal,—~ or Lac,—Galy— strains show a characteristic interaction with Lac, on EMB lactose agar. A stronger fermentation reac- tion occurred at the junction of cross-streaked inocula after 1-2 days incuba- tion. Subculture from the lactose-positive area gave only the original Lac— types, and a new synergistic reaction at the juxtaposition of similar colonies. The Lac,~ mutant has already been diagnosed as producing lactase in response to butyl galactoside but not to lactose itself (LEDERBERG 1951). The influence ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 479 of the Gal~ mutation and the synergism of Gal~ with Lacy~Gal+ mutants reinforce the conclusion that the Lac;~ mutants retain a considerable lactase- forming competence not ordinarily realized in the presence of lactose. The disposition of modifiers that lack their own phenotypic effects is more difficult. The fact that all of the stable derivatives in this class lost the ability to respond to butyl galactoside suggests the possibility that a second Lac muta- tion, at another locus, had been introduced. Even though both mutations of a double Lac~ mutant, Lac,;~Lac,—, were potentially revertible, their combina- tion would be stable as far as detectible reversions to Lact is concerned. Reversion at one locus would be concealed by the mutant status of the second. This suspicion was verified by the production of a Lac;~Lac;— double mutant by recombination of two papillate single mutants. The double mutant was nonpapillate. Its composite nature was confirmed by crosses to wild type which separated and re-exposed the revertible components. The status of the stabilized Y-87 derivatives that fall into this category remains in doubt, pending systematic tests for the identity of the presumed Lac,~ by crosses to standard Lac~ mutants at various loci. Unfortunately, not all of the stocks needed for this enterprise have yet been developed in suitable form. Nevertheless, the modification of response to butyl galactoside as well as papillation of these derivatives argues against the participation of a true muta- bility modifier similar to Dotted in maize. PSEUDOALLELISM AT THE Lac, LOCUS Evidence for separating a group of four types of Lac, strains on the basis of their intrinsic mutation rate to a fifth allele, Lact has been offered. The observed effects appeared to reside in that locus alone. However, the identifi- cation of mutant factors as alleles is at best a negative inference, derived from the failure to detect nonparental classes in a large series oi tests. The scope of such experiments is subject to technical limitations inherent in the material handled (cf. STADLER 1951) and the spontaneous mutation rate of the crucial locus. Because of the applicability of selective techniques for screening large numbers, bacteria have special advantages for such studies. In an extension of the preliminary allelism tests already cited, the crosses of Lac~ mutants were repeated on a large scale. The presumed Lac, locus has been dissected into at least two components. Whereas no Lact recombinants were observed among 60,000 prototraphs from Y-53 x Y-87, about 1 in 1500 from Y-87 x W-112 was lactose-positive. To avoid confusion with spontaneous reversions in the Lac~ mutable parent, certain precautions were taken: the Lac~ parents were freshly reisolated from a nonlactose medium to minimize the accumulation of Lact mutants. Because of its close linkage to Lac, (LEDERBERG 1947), F’6 served as a use- ful marker. Of 14 Lac+ prototrophs from a total of 25,000 observed in a cross of W-1307 (B-M~-I6"Lac~" x T-L~16*Lae~**) 11 were '6* like the Lac— stable parent. In a parallel control experiment, of a reverted W-1307 x W-112, 480 ESTHER M, LEDERBERG all of the 24 Lact prototrophs tested were 176". It may be concluded that the exceptional lactose-positive prototrophs originated by recombination rather than spontaneous mutation. Y-87 and Y-53 may therefore be considered as Lacy,~ and W-112 as Lacy,~. The sequence BAL... V6...@...0...7L is indicated by these data. For further critical study, a second marker just to the right of Lacqa.x would be very useful, but has not vet been found. Studies on heterozygous compounds of Lacy, and Lac), suggest a very close relationship between these components. W-112 (Vo'atb-V1ISMil+ ) was crossed on EMS mannitol agar with a filial Y-53 derivative, W-1435 (V'6'a-b- VI’ Mtl-Het). Het is a factor that promotes nondisjunction (LEDERBERG 1949), Heterozygous diploid prototrophs were isolated on the basis of their segregation for A/#?, 1 and 16. Such diploids are presumably atb~/a-b+ at the Lac, locus. Although they showed little or no fermentation of lactose, this constitution was verified by the segregation of Lac~ stable (16'a+b—) and mutable (1°6"a—b+ ) haploids. Crossing over to yield atbt occurs relatively frequently and results in marked papillation. Heterozygotes which are almost certainly a+b+ /a—b~ occur among exceptional lactose-posi- tive prototrophs from the same cross. In distinction to the opposed “ repul- sion ” heterozygotes, these “ coupling ” diploids segregate lactose-positive and stable lactose-negative haploids. The latter are assumed to be a~b~, comple- mentary to the atb+ crossovers. Because the presumed atb*/a~b~ 1s lac- tose-positive while a+b—/a~b+ is lactose-negative, a “ position effect” is suggested. A group of mutants has been designated as Lacy~. These mutants differ from Lac;~ in showing little or no residual lactase when grown on butyl galactoside, and when crossed with Lac,~ give lactose-positive recombinants with a frequency of 0.1 to 0.2 percent (LEDERBERG 1951), Crosses of Lacy7 with W-112 (a+b—) or Y-53 (a~b+) have given identical results. In addi- tion to the rare Lact crossovers, lactose-positive nondisjunctional exceptions were found (LEDERBERG 1949). In contrast to compounds of a~bt/atb-, heterozygotes of the constitution Lacy-atb+/Lacyta~ bt and Lacg-atbt /Lacytatb~ approach Lac+ in their lactase activity. In spite of the close linkage of Lac, with Lacs, these loci appear to function independently of each other and are therefore assigned different symbols. DISCUSSION Multiple alleles can be characterized by differences in mutability as well as in physiological expression. For example, two inositol-dependent mutants have been described in Neurospora that represent allelic forms indistinguish- able from one another except for their capacity to revert to wild type upon suitable mutagenic treatment (GiLEs 1948). The mutability difference segre- gated in the ascospore progeny when the two mutants were crossed. Recurrent mutants with different rates of spontaneous reversion were identified at the Lac, locus in E. coli. Qualitative examination disclosed four ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 481 categories. Additional intergrading alleles were suspected but may be unclassi- fiable by ordinary means. The number of possible configurations at this locus may be very large, although empirical tests for precise separation of all the potential alleles have not been formulated. Rnoapves (1941) has already pro- posed the concept of a continuous spectrum of mutation frequency of which only a few points may be recognized. Recent genetic research has revealed the complexity of several genes. The components of the R (SrapLer 1951) and the 4 (Lavennan 1948) alleles in maize have been separated by their mutability and physiological action where recombination tests at first failed to subdivide the locus. Intensive studies of large numbers of progeny among allelic recurrences in E. colt (Leverserc 1951), Neurospora (BoNNER 1951; Gives 1951), and Dro- sophila (GREEN and Green 1949; Lewis 1951; Taxu-Komar 1950) have also revealed a fine structure in the loci involved. In E£. coli spontaneous or induced changes in mutability pattern at the Lac, complex invariably involved alterations to more stable types. The miniature alleles in Drosophila virilis which differ in their mutability, however, are able to mutate to either more stable or unstable forms (DEMEREC 1941). A similar situation exists for an unstable Afa/— mutant in £. colt (LEDERBERG ef al. 1951). The intrinsic factors determining the degree of mutability are often autono- mous, i.e., appear to reside within the locus. Other genotypic influences have been discovered, however. The existence of at least two mutability modifiers of the R locus has also been recently suspected (STADLER 1949). The most clear-cut specific intergenic regulator of mutability has been described by Ruoapes in Dotted maize (RHOADES 1938, 1941, 1945). Technical difficulties have precluded a biochemical analysis of the mutability interactions in maize. On the other hand, bacteria would be eminently suitable for exploring such effects. Modifiers that suppressed the formation of Lact reversions in unstable Lac~ strains were readily found. Closer examination revealed that the stable Lac~ types had remained potentially mutable, but the mutants could not be expressed in combination with certain alleles at other loci. In strains W-516, W-744, and W-750, a reduced selective advantage interfered with the expression of Lact mutants. In addition, Gal— modified the phenotype of the Lac~ cells by enhancing galactosidase activity. These genetic factors may he summarized as affecting the conditions under which the Lac+ mutants are favored. No gene-induced alteration of the intrinsic muta- tion rate of another gene was discovered in this study. ‘As far as could be determined, the Lac+ derivatives of each mutable Lac~ strain were reverse mutants. Suppressor loci were found, however, by select- ing lactose fermenters from stable strains. These loci are latent in the wild type Lact. They can be detected as mutations which mimic wild type by suppressing the effect of Lac,~ mutation. The two suppressors reported in Neurospora are each less efficient than the 482 ESTHER M. LEDERBERG comparable reversions (HovLawan and Mircnecy 1947; Gites 1950). This was also true for several Lacy+ mimic factors. The Lact of W-716 which most closely resembled Lac, + was rapidly eliminated in a lactose-free medium. The inefficiency of suppressor mutants may perhaps he ascribed to either genic inefficiency or to imbalance with the rest of the genotype. Duplication of a locus has been postulated as one mechanism for the origin of suppressors and pseudoalleles (ScuuLtz and Bripces 1932; HouLAHAN and MitcuHeit 1947; STEPHENS 1948). One may visualize the transfer of such a duplication to another location, followed by gradual modification of its original activity and eventual evolutionary divergence both in microorganisms (LEDERBERG 1948b) and in higher forms (STEPHENS 1951). SUMMARY A series of recurrent Lac;~ mutants was classifiable by the rates of rever- sion to Lact+. Four major types were recognized on the basis of the average number of Lact papillae per colony under standard conditions, All alterations of mutability, whether spontaneous or induced, were in the direction of greater stability. The papillae isolated from the mutable Lac— strains were back-muta- tions as proven by backcrosses to wild type. However, stable strains gave rise to mimic reversions (suppressors) phenotypically less competent than wild type. The primary differences in mutability were inseparable from the locus in recombination experiments. Other loci controlling apparent mutability were implicated in secondary stable types isolated from mutable strains. Further studies showed that the second mutation lowered the selective advantage of potential Lact mutants but did not necessarily modify the intrinsic muta- bility directly. The subdivision of the Lac,;— locus into two components, @ and b, was revealed by the rare occurrence of Lact recombinants among intercross progeny, a~b+/a~b~ heterozygotes were phenotypically lactose-negative. On the other hand, the at+6+ crossovers, both in haploids and in heterozygotes were phenotypically lactose-positive, indicating a * position effect.” LITERATURE CITED Boxner, D., 1951 Gene-enzyme relationships in Neurospora. Cold Spring Harbor Symp. Quant. Biol. 16: 143-158. Demerec, M., 1941 Unstable genes in Drosophila. Cold Spring Harbor Symp. Quant. Biol. 9: 145-150. Giues, N. H., Jr, 1948 Induced reversions of inositol-requiring mutants in Neurospora crassa. Abstract Book, 8th Int. Cong. Genetics, Stockholm, 48. 1950 The occurrence of suppressors in a methionineless mutant of Neurospora. Genetics 35: 108-109. 1951 Studies on the mechanism of reversion in biochemical mutants of Neurospora crassa. Cold Spring Harbor Symp. Quant. Biol. 16: 283-314. Green, M. M., and K. C. Green, 1949 Crossing-over between alleles at the lozenge locus in Drosophila melanogaster. Proc. Nat. Acad. Sci. 35: 586-591. ALLELIC RELATIONSHIPS IN ESCHERICHIA COLI 483 Happow, A., 1937 On secondary colony development in bacteria and an analogy with tumour production in higher forms. Acta Int. Union Against Cancer 2: 376401. Hovutanan, M. B., and H. K. MircHett, 1947 A suppressor in Neurospora and its use as evidence of allelism. Proc. Nat. Acad. Sci. 88: 223-229. KRISTENSEN, M. K., 1944 Recherches sur la fermentation mutative. XIII. Fermentation mutatives chez des bactéries coliformes. Acta Path. 21: 957-971. LaucHnan, J. R., 1948 The action of allelic forms of the gene 4 in maize. I. Studies of variability, dosage, and dominance relations. The divergent character of the series. Genetics 33: 488-517. LEpERBERG, J., 1947 Gene recombination and linked segregations in Escherichia colt. Genetics 82: 505-525. 1948a Gene control of Beta-D-galactosidase in Escherichia coli. Genetics 33: 617-618. 1948b Problems in microbial genetics. Heredity 2: 145-198. 1949 Aberrant heterozygotes in Escherichia coli. Proc. Nat. Acad. Sci. 85: 178-184. 1950a Isolation and characterization of biochemical mutants of bacteria. Methods in Medical Research 3: 5-22, . 1950b The Beta—D-galactosidase of Escherichia coli, strain K-12. J. Bact. 60: 381-392. 1951 Genetic studies with bacteria. Genetics in the 20th Century. viii + 634 pp. New York: Macmillan. / Leperserc, J., E. M. Leperserc, N. D, Zinper and E. R. Livery, 1951 Recombination analysis of bacterial heredity. Cold Spring Harbor Symp. Quant. Biol. 16: 413-444. Lewis, E. B., 1951 Pseudoallelism and gene evolution. Cold Spring Harbor Symp. Quant. Biol, 16: 159-174. ; Luria, S. E., 1947 Recent-advances in bacterial genetics. Bact. Rev. 11: 1-40.. Massinr, R., 1907 Uber einem in biologischer Beziehung interessanten Kolistamm (Bacterium coli mutabile)—ein Beitrag zur Variationen bei Bacterien. Arch. f. Hyg. 61-62: 250-292. Monon, J., and A. AupuREAU, 1946 Mutation et adaptation enzymatique chez Escherichia coli-mutabile. Ann. Inst. Past. 72: 868-878. Ruoapes, M. M., 1938 Effect of the Df gene on the mutability of a: in maize. Genetics 23; 377-354, 1941 The genetic control of mutability in maize. Cold Spring Harbor Symp. Quant. Biol. 9: 138-144. 1945 On the genetic control of mutability in maize. Proc. Nat. Acad. Sci. 31: 91-95. ScHuLTz, J., and C. B. Brinces, 1932 Methods for distinguishing between duplications and specific suppressors. Amer. Nat. 66: 323-334. Srapter, L. J., 1949 Spontaneous mutation at the R locus in maize. III. Genetic modi- fication of mutation rate. Amer. Nat. 83: 5-30. 1951 Spontaneous mutation in maize. Cold Spring Harbor Symp. Quant. Biol. 16: 49-64. STEPHENS, S. C., 1948 A biochemical basis of the pseudo-allelic anthocyanin series in Gossypium. Genetics 33: 191-214. 1951 Possible significance of duplication in evolution. Adv. in Genetics 4: 247-265. Taxu-Komat, K., 1950 Semi-allelic genes. Amer. Nat. 84: 381-392. Tatum, E. L., 1945 X-ray induced mutant strains of Escherichia coli. Proc. Nat. Acad. Sci. 31: 215-219. Tatum, E. L., and J. Lepersere, 1947 Gene recombination in the bacterium, Escherichia coli. J. Bact. 58: 673-684.