Proceedings of the National Academy of SciencesVol. 65, No. 3, pp. 625-632, March 1970

Derepression of GDP-a-D-Mannose and UDP-GlucosePyrophosphorylases by a Regulator Gene Mutation;

Episomal Dominance in Partial Diploids*Michael M. Lieberman,t Christine E. Buchanan, and Alvin Markovitz

DEPARTMENT OF MICROBIOLOGY, UNIVERSITY OF CHICAGO, CHICAGO, iLLINOIS

Communicated by R. C. Lewontin, December 17, 1969

Abstract. 1\Jutants of Escherichia coli K12 at the capR locus are mucoid,overproduce capsular polysaccharide, and are derepressed for synthesis of severalenzymes involved in capsular polysaccharide synthesis'-' including GDP-man-nose pyrophosphorylase.8 UDP-glucose pyrophosphorylase is also derepressedin a haploid capR9 mucoid mutant. Heterozygous mucoid partial diploids withthe capR9 allele on the episome and the wild-type (capR+) allele on the chromo-some (F'capR9/capR+) are derepressed for UDP-glucose pyrophosphorylase andGDP-mannose pyrophosphorylase, while the reciprocal nonmucoid heterozygotes(F'capR+/capR9) are repressed for these enzymes. These results provide evi-dence that the episomal capR allele is dominant with respect to synthesis of thesetwo enzymes.

A regulator gene (designated capR; previously designated R,' 2) controls thesynthesis of capsular polysaccharide (mucoid) and several enzymes involved incapsular polysaccharide synthesis in Escherichia coli K12.1-3 The wild type isnonmucoid and contains low levels of the enzymes, while mutants at the capRlocus are mucoid and contain derepressed levels of several of the enzymes of cap-sular polysaccharide synthesis.2 3 8 The sugar components of the capsule areD-glucose, D-galactose, D-glucuronic acid, and L-fucose." 2

Independent experiments of Howard-Flanders, Simson, and Theriot,4 andAdler and Hardigree5 demonstrated that a mutation designated lon- causedE. coli K12 to become very sensitive to ultraviolet (UV) and ionizing radiation.These lon- strains are mucoid4 and after X-ray radiation form filaments thatdie.5 It is now clear that the phenotype of capR and lon mutants are identical,i.e., mucoid and UV sensitive," 2, 4-7 and arise from the same mutationalevent."1 4, The lon- or capR mutants are not defective in repair of irradiatedT14 or T7 bacteriophage (Uretz, unpublished data). The mechanism by whichcapR controls radiation sensitivity is unknown at the present time.

Genetic studies by Markovitz and Rosenbaum have demonstrated that a spe-cific mutation, capR9, is dominant when on an episome but recessive on achromosome. Thus a partial diploid strain with the genotype F'capR9/capR+ ismucoid, but a partial diploid strain with the genotype F'capR+/capR9 is non-mucoid. On the basis of this observation, a model for the capR gene product,was formulated.' The model, termed the active oligomer model, postulates that

625

626 GENETICS: LIEBERMAN ET AL. PROC. N. A. S.

the capR gene product is a repressor protein composed of subunits and that onlythe aggregate functions as a repressor.1 In the context of this model it seemedimportant to determine whether or not episomal dominance could be demon-strated with some of the enzymes that are derepressed by the capR9 mutation.The present communication describes the episomal dominance of the capR9

mutation on an enzymological level for the first time. In a separate report' itwas demonstrated that GDP-mannose pyrophosphorylase, an enzyme involvedin capsular polysaccharide synthesis, is derepressed in haploid capR9 strains.In this report it is shown that UDP-glucose pyrophosphorylase is derepressed inhaploid capR9 strains. Moreover, in heterozygous partial diploids with capR9on the episome, and capR+ on the chromosome, both of these enzymes are de-repressed as compared to the reciprocal partial diploid.

Materials and Methods. Bacteria: All strains of bacteria were streptomycin-resistant derivations of E. coli K12 (unless otherwise noted). Strain MC116 requiresleucine, proline, purine, and tryptophan for growth, is not able to grow on lactose, and isbacteriophage T6-resistant, bacteriophage P1-immune, streptomycin resistant, and F-.Its genotype is leu-1, proC, purE, trp-1, lac-2, tsx-, str-12. Strain MC116 is nonmucoidand contains the wild-type allele (capR+) that controls mucoidness (capsular poly-saccharide synthesis).'-' 9 Strain MC121 was constructed by transduction of the capR9allele into a histidine negative derivative of strain MCI 16 as previously described.3 Thedonor and recipient strains used to make the partial diploids by F-duction were describedpreviously.' Strain W3747 (streptomycin-sensitive) transfers the capR+ allele on itsepisome. Strain MC144 (previously designated S77-BM') and similar more-recentlyprepared strains were used to denote the capR9 allele. The leu+, trp+ transductants ofstrains MC100 and MC102 (previously designated strains X-156 and T72-C2)1 arehaploid strains containing the capR+ and capR9 alleles, respectively, and were usedas recipients.Media: M-9 minimal medium'0 was supplemented as described previously.'

Conditions and citations for transductions, F-ductions, protein determinations, andmethyl pentose analyses are recorded in earlier reports.2 9, 11Enzymatic analyses: GDP-mannose pyrophosphorylase was assayed by the

32P-pyrophosphate exchange method of Preiss.'2 UDP-glucose pyrophosphorylase wasassayed by the method of Hansen et al. (Method 1)" except for the following additionalingredients in a total volume of 0.5 ml: 0.001 jsmole glucose-1, 6-diphosphate, 2.5gmoles KF, and 5 Mmoles NaH2PO4-Na2HPO4, pH 7.0. All assays were run at roomtemperature. The rate of reaction was followed by a continuous recording of the pro-duction of TPNH from TPN by its increase in optical density at 340 m,4 with a Gilfordrecording spectrophotometer model 240. Enzymatic activity was directly proportionalto the amount of extract added and assay values are the average of two or three assayswith different quantities of extract. TDP-glucose pyrophosphorylase assay mixtureswere the same as for UDP-glucose pyrophosphorylase except that 1 /Imole of TDP-glucose was substituted for UDP-glucose. Details for the assays of phosphomannoseisomerase," UDP-galactose-4-epimerase,'4 GDP-mannose hydrolyase2 and GDP-ifucosesynthetase2 were described previously. Modifications for the latter two assays alongwith details of preparation of GDP-4-keto-D-rhamnose (GDP-KR) used in this studyhave been described.8 The GDP-KR concentration was 0.05 mM in the GDP-ifucosesynthetase assays reported here. All cultures for enzymatic analysis were grown to earlystationary phase in liquid M-9 minimal medium with 0.6% glucose as a carbon source.Cultures of haploids were grown at 23 or 350C while the partial diploids were grown at35 or 370C as indicated. The reason for differences in temperature will be discussed later.The cells were harvested by centrifugation, washed twice in 0.05 M NaH2-Na2HPO4,pH 7.0, and resuspended in the same buffer at approximately a 20-fold higher concen-tration of cells. The cells were disrupted by sonic oscillation with an MSE probe-type

VOL. 65, 1970 GENETICS: LIEBERMAN ET AL. 627

sonic oscillator using the small (1.0-cm diameter) probe. The optical density 600 mu of a1: 50 dilution of the cell suspension was measured before and after sonic disruption. Gen-erally, after 15-20 sec of disruption, 30-40%0/ of the original OD 600 mU remained. Thesonically disrupted cell suspension was centrifuged at 17,500 rpm in the Sorvall RC-2Bcentrifuge for 20 min and the supernatant used as the crude enzyme. Freshly preparedsupernatant fractions were used for all enzyme assays.

Results. Derepression of UDP-glucose pyrophosphorylase in a haploidmucoid strain: One of the enzymes involved in synthesis of three of the nucleo-tide sugar precursors of capsular polysaccharide is UDP-glucose pyrophos-phorylase. The results of Table 1 indicate that UDP-glucose pyrophosphorylaseis derepressed in a capR type mucoid mutant as compared to the wild type(capR+). TDP-glucose pyrophosphorylase was also measured in two differentcapR9 strains and was derepressed two- to threefold as compared to the capR+strain (data not shown).

TABLE 1. Derepression of UDP-glucose pyrophosphorylase in strains grown at 230C.UDP-glucose pyrophosphorylaset

Strain Genotype Polysaccharide* (miumoles/hr/mg protein)MC121 capR9 700 7140 i 1190MC116 capR+ 13 530 i 150

* Micrograms of nondialyzable methylpentose/ml/unit-cell turbidity (OD 600 mrA).t Results include data from three experiments with each strain.

Derepression of UDP-glucose pyrophosphorylase and GDP-mannose pyro-phosphorylase in heterozygous partial diploids: Previous studies showed thatGDP-mannose pyrophosphorylase was derepressed in halpoid capR9 type mucoidstrains as compared with haploid capR + strains. Therefore both GDP-mannoseand UDP-glucose pyrophosphorylases were measured in mucoid F'13capR9/-capR+ and nonmucoid F'13capR+/capR9 strains in order to determine whetherthe level of these enzymes was affected by the relative position of the capR9 andcapR+ alleles. The results, presented in Tables 2 and 3, demonstrate derepres-sion of UDP-glucose pyrophosphorylase (approximately 6-fold) and GDP-mannose pyrophosphorylase (40-fold) when the capR9 allele is on the episomerelative to when it is on the chromosome. These results demonstrate that theepisomal capR allele is dominant and determines the quantity of UDP-glucoseand GDP-mannose pyrophosphorylases synthesized.

It should be pointed out here that, due to the relative instability of thesepartial diploids,"5 it is essential to estimate quantitatively the proportion of cellsof a given genotype in a culture which is to be assayed for enzyme activity. Forthis reason all heterozygous partial-diploid cultures assayed for UDP-glucose andGDP-mannose pyrophosphorylases were dilution plated both on media selectivefor the episome and on nonselective media, and the per cent of mucoid or non-mucoid clones determined (Tables 2 and 3). The data for the plating on non-selective media is omitted from the tables since there were no significant differ-ences between the selective and nonselective platings in total plate counts andper cent mucoid clones. Thus, most of the cells had retained their episome in thelast generation. The results show that of those cells that had retained theirepisome, most were of the expected phenotype.

628 GENETICS: LIEBERMAN ET AL. PROC. N. A. S.

TABLE 2. UDP-glucose pyrophosphorylase activity in heterozygous partial diploids grownat 35 and 37'C.

UDP-glucosepyrophosphorylaset(mimoles/hr/mg

Genotype Phenotype on plating* Polysaccharidet protein)F'capR9/capR+ 84.3 + 5.8% mucoid 225 +t 90 5870 + 1670F'capR+/capR9 93.3 + 4.9% nonmucoid 1.8 i 1.1 930 :1 350

* The data are from plates selective for clones retaining episomal markers (lac + proC + purE + orlac + purE +) but there were no significant differences between these data and those obtained fromplatings on nonselective media.

t Micrograms of nondialyzable methylpentose/nil/unit-cell turbidity (OD 600 mAi).I Results for each partial diploid include data from two separate clones grown at 371C and two

other clones grown at 35'C. No corrections have been made for deviations from expected phenotypewhich would be 100% mucoid or 100% nonmucoid.

TABLE 3. GDP-mannose pyrophosphorylase activity in heterozygous partial diploids grownat 370C.

GDP-mannosepyrophosphorylaset(mgnoles/hr/mg

Genotype Phenotype on plating* Polysaccharidet protein)F'capR9/capR+ 75.3% i 3.2%mucoid 330 132 i 73F'capR+/capR9 95.5 i 2.7% nonmucoid 5 3.0 i 1.7

* The data are from plates selective for clones retaining episomal markers (lac + proC + purE + orsac+ purE +) but there were no significant differences between these data and those obtained fromplatings on nonselective media.

t Micrograms of nondialyzable methylpentose/ml/unit-cell turbidity (OD 600 mA).t Results include data from two separate sexductants representing five separate clones (F'capR +/

capR9) or four separate clones (F'capR9/capR+) and have not been corrected for deviations fromexpected phenotype which would be 100% nonmucoid or 100% mucoid.

Comparison of enzymes in strains homozygous diploid and haploid for capRalleles: In the course of performing these experiments with heterozygous partialdiploids, we also constructed homozygous partial diploids of the type with thecapR9 allele both on the episome and on the chromosome (F'capR9/capR9) andalso of the type F'capR+/F'capR+. These strains, as well as a haploid capR9and a capR+ strain were grown at 350C and the level of enzymes were measured.The results, presented in Table 4, demonstrate that GDP-mannose and UDP-glucose pyrophosphorylases, GDP-mannose hydrolyase and GDP-L-fucosesynthetase all behave similarly. They are derepressed in the mucoid strains(F'capR9/capR9 and capR9) and are repressed in the nonmucoid strains (F'-capR+/capR+ and capR+). There are no indications of greater derepression ofenzymes in the F'capR9/capR9 strain as compared to the capR9 strain. Similarresults were obtained with cultures grown at 370C concerning synthesis of phos-phomannose isomerase, UDP-galactose-4-epimerase, GDP-mannose hydrolyaseand GDP-L-fucose synthetase. However, GDP-mannose pyrophosphorylasewas lower in the haploid capR9 strain as compared to the F'capR9/capR9 strainin these experiments at 370C. Since we have noted much greater variability inthe specific enzymatic activity of GDP-mannose pyrophosphorylase betweenduplicate cultures of the F'capR9/capR9 strain and haploid capR9 strains whenthey were grown at 370C as compared to 350C, we have reported the data fromcultures grown at 350C.

VOL. 65, 1970 GENETICS: LIEBERMAN ET AL. 629

TABLE 4. Activity of enzymes in homozygous partial diploids and haploid strains grownat 35'C.

GDP-mannose UDP-iglucosepyrophosphorylase pyrophosphorylase(memoles/hr/mg (mpmoles/hr/mg

Genotype Polysaccharide* protein) protein)F'capR9/capR9 455 i 40t 1860 i 340 10,700 i 1,300capR9 240 2350 10,500F'capR+/capR+ 4 2.8 900capR+ 2 5.3 720

GDP-mannose GDP-L-fucosehydrolyase synthetase

Genotype Polysaccharide* (mpmoles/hr/mg protein) (mpmoles/hr/mg protein)F'capR9/capRl9 455 i 40t 280 i 46 192 4t 31capR9 240 462 197F'capR +/capR + 4 73 <50tcapR+ 2 93 <501

* Micrograms of nondialyzable methylpentose/ml/unit-cell turbidity (OD 600 mpA).t Where standard deviation is indicated results are from assays of four different clones (except for

GDP-mannose hydrolyase where three different clones were assayed). Assays for capR9 strains arethe average of two clones while those for F'capR +/capR + and haploid capR + are from single clones.

I No activity was detected.

Discussion. The demonstration of the derepression of UDP-glucose pyro-phosphorylase in haploid strains with capR mutations adds another enzyme tothe list of those that have previously been found to be derepressed in thesestrains.2 3 8 Our analyses indicated a much smaller derepression of TDP-glucose pyrophosphorylase. Nikaido has separated UDPG pyrophosphorylaseactivity from wild-type Salmonella into four peaks on DEAE-cellulose chroma-tography and they all have activity toward TDP-glucose. 6 Thus our smallincreases in TDP-glucose pyrophosphorylase may be a reflection of increasedUDP-glucose pyrophosphorylase synthesis. The fact that none of the sixenzymes derepressed in haploid capR9 strains (Table 4 and Results) was furtherderepressed in the F'capR9/capR9 strain provides no support for an activatortype control mechanism' but does not rule it out. The results presented inTable 4 also indicate that, in capR+ strains, the specific enzymatic activity ofUDP-glucose pyrophosphorylase is much higher than the three enzymes involvedin synthesis of GDP-L-fucose. This is consistent with the fact that UDP-glucosepyrophosphorylase is necessary for synthesis of cell wall glucose and galactosewhile L-fucose is only found in the capsular polysaccharide.The list of enzymes derepressed in capR mucoid mutants now includes UDP-

glucose pyrophosphorylase, UDP-glucose dehydrogenase (Lieberman, Shaparis,and AMarkovitz, submitted to J. Bacteriol.), UDP-galactose-4-epimerase,2 phos-phomannose isomerase,3 GDP-mannose pyrophosphorylase,8 GDP-mannosehydrolyase, and GDP-L-fucose synthetase.2 The structural genes for some ofthese enzymes have been mapped previously and are widely separated from eachother and from the capR gene which is linked by transduction to proC." 2, 7The probable structural gene for phosphomannose isomerase (man), was found tomap, by time of entry experiments, between the histidine and tryptophan regionsof the chromosome.9 The structural genes for UDP-galactose4-epimerase(galE) and UDP-glucose pyrophosphorylase (galU)'8 map at different loci than

630 GENETICS: LIEBERMAN ET AL. PROC. N. A. S.

either capR or man. Structural genes for the synthesis of GDP-L-fucose, UDP-glucose dehydrogenase, and polysaccharide polymerase have not been mapped.Thus, the capR regulator gene controls the expression of several widely separatedstructural genes which are repressed in the wild-type nonmucoid strains, butbecome derepressed in strains with a mutation in the capR gene. Therefore, itwas of particular interest to determine whether dominance of the episomal capRallele in partial diploids, which results in derepressed capsular polysaccharideproduction in the F'capR9/capR+ heterozygote but not in the F'capR+/capR9heterozygote, extended to enzymes involved in capsular polysaccharide syn-thesis.We must now discuss the UDP-glucose and GDP-mannose pyrophosphorylases

separately. UDP-glucose pyrophosphorylase is derepressed approximately 13-fold in a capR9 strain compared to a capR+ strain and does not vary amongcultures grown between 23 and 350C (Tables 1 and 4). In the F'capR9/capR+strain the derepression is approximately 6.5-fold compared to the reciprocalheterozygote (cultures grown at 35 and 370C, Table 2). The UDP-glucosepyrophosphorylase activity of the F'capR+/capR9 strain is repressed and verysimilar to the activity of either the F'capR +/capR + strain or the haploid capR +strain (Tables 2 and 4). Thus the dominance of the episomal capR allele isalmost complete with respect to control of UDP-glucose pyrophosphorylase.In capR+ strains GDP-mannose pyrophosphorylase synthesis is greatly affectedby the temperature of growth; the specific enzymatic activities are 110 at 250C8and 5 at 350C. In capR9 strains the activities are derepressed at 250C8 and350C and are very similar at both temperatures (ref. 8 and Table 4), approxi-mately 2400. Although more variability was found with capR9 and F'capR9/-capR9 strains grown at 370C, the specific enzymatic activity is approximately50 per cent that at 350C. The F'capR+/capR9 strain is repressed at 370 as isthe F'capR+/capR+ and the haploid capR+ strain (Tables 3 and 4). However,the F'capR9/capR+ strain is derepressed; while the derepression is 40-fold withrespect to the F'capR+/capR9 strain (Table 3) it is not maximally derepressed.We can estimate the percentage of maximum derepression assuming that thespecific enzymatic activity at 370C in F'capR9/capR9 strains is 50 per cent ofthat at 350C as mentioned above. Thus, the value for F'capR9/capR9 is 1860/2(Table 4) = 930. This figure is in good agreement with the actual measurementsat 370C on the F'capR9/capR9 strain. 19 Then [(F'capR9/capR +)/(F'capR9/-capR9)]100 = (130/930)100 = 14 per cent. Thus, the dominance of the epi-somal allele is not as complete with GDP-mannose pyrophorylase as it is withUDP-glucose pyrophosphorylase. In terms of the active oligomer model thismight result from the same mixed oligomer repressor acting differently at differentoperator sites. These results verify one of the predictions of the active oligomermodel (see below): that the control of polysaccharide synthesis in heterozygouspartial diploids containing capR+ and capR9 alleles extends to control of proteinsynthesis (UDP-glucose pyrophosphorylase and GDP-mannose pyrophosphoryl-ase). We have also measured the UV sensitivity of the partial diploids thatcontain capR+ and capR9. Surprisingly, both the mucoid F'capR9/capR+ andthe nonmucoid F'capR+/capR9 strains are UV resistant although the haploid

VOL. 65, 1970 GENETICS: LIEBERMAN ET AL. 631

capR9 and the F'capR9/capR9 strains are very sensitive to UV.21 Thus onlythe presence of a wild type (capR +) allele is necessary to produce a UV-resistantphenotype. The basis for UV sensitivity or resistance associated with capRalleles is not understood. We have discussed possible explanations for theseobservations. 20

In 1965 Markovitz and Rosenbaum suggested the active oligomer model toexplain the dominance of the capR allele on the episome with respect to poly-saccharide synthesis' and, as noted above, the dominance has been extended toUDP-glucose pyrophosphorylase and GDP-mannose pyrophosphorylase dere-pression. The model suggested that the repressor is an oligomer (monomers areinactive) and that certain monomers specified by mutant alleles (capR9) couldinactivate repressor made up mainly of wild-type (capR+) subunits.' Foley,Giles, and Roberts, also in 1965, described an in vivo negative complementationeffect between wild-type and certain mutant alleles of adenylosuccinase indiploids of Aspergillus nidulans." It is clear that both groups invoked the sameexplanation to provide a molecular model for the results of genetic experiments.Furthermore, both groups appreciated the predictive value of the model." 21More recently certain mutations have been reinvestigated in the i-gene that con-trols f3-galactosidase production. Davies has mapped an anomalous class of i-gene mutants originally considered to be deletions of the i-gene and of part of theoperator called "i-o c.*'22 Some of these are point mutants in the middle of the i-gene (designated i-d (for dominant)22) and are dominant to i+; i.e., the partial di-ploid F'i -dz-/i +Z+ makes $-galactosidase constitutively.23 Furthermore, a mu-tation designated iq (for quantity)23 was found to be dominant to is (pleiotropicnegative, not inducible) and the partial diploid F'is/iq has a lac+ phenotype.Mifller-Hill, Crapo, and Gilbert prefer negative complementation in an activeoligomer-type model involving subunit interactions to explain the above re-sults.23' 24 Other recent examples of negative complementation have come tolight in work with bacteriophage. Lieb suggested that the active oligomer modelcould explain the interaction of certain temperature-sensitive mutations and wild-type allele in the C1 gene, the immunity repressor, for bacteriophage X.25 Bern-stein and Fisher investigated dominance in bacteriophage T4 using conditionallethal mutants of gene 37 which specifies a tail fiber structural component.26They found that various temperature-sensitive and amber mutants show differentdegrees of dominance in mixed infections with wild-type phage. These mixedinfections resulted in a much smaller burst size than infection with wild typealone, and the progeny phage from the mixed infections had a slower rate ofadsorption to bacteria than did the wild-type phage. These results were inter-preted as indicating an interaction between wild-type and mutant polypeptidesresulting in the formation of hybrid tail fibers in the progeny phage.26 Similarreasoning was applied to explain complementation data for mutant and wild-typebacteriophage T4-induced DNA polymerase.27 It thus appears that the phe-nomenon of negative complementation in an active oligomer-type model isapplicable to other systems. It should be pointed out that there has been nodemonstration of negative complementation in vitro using an enzyme althoughone would predict that it should be observable. Attempts to demonstrate nega-

632 GENETICS: LIEBERMAN ET AL. PROC. N. A. S.

tive complementation in this laboratory with several complementing mutantsof alkaline phosphatase have been unsuccessful (S. Kang, unpublished results).

* Supported by grant A106966 from the National Institutes of Health. M. M. L. was aUSPHS predoctoral trainee (training grant 5T01-GM-00603) and C. E. B. was a NSF pre-doctoral trainee (GZ-833).

t Present address: Exobiology Division, Ames Research Center, National Aeronautics andSpace Administration, Moffett Field, Calif. 94035.

1 Markovitz, A., and N. Rosenbaum, these PROCEEDINGS, 54, 1084 (1965).2 Markovitz, A., these PROCEEDINGS, 51, 239 (1964).3 Markovitz, A., M. M. Lieberman, and N. Rosenbaum, J. Bacteriol., 94, 1497 (1967).4Howard-Flanders, P. E. Simson, and L. Theriot, Genetics, 49, 237 (1964).5 Adler, H. I., and A. A. Hardigree, J. Bacteriol., 87, 720 (1964).6 Adler, H. I., and A. A. Hardigree, Radiation Res., 25, 92 (1965).7 Donch, J., and J. Greenberg, Molec. Gen. Genetics, 103, 105 (1968).8 Lieberman, M. M., and A. Markovitz, J. Bacteriol., 101 (1970).9 Markovitz, A., R. J. Sydiskis, and M. M. Lieberman, J. Bacteriol., 94, 1492 (1967).

10 Adams, M. H., in Bacteriophage (New York: Interscience Publishers, Inc., 1959), pp.445-447.

'1 Kang, S., and A. Markovitz, J. Bacteriol., 93, 584 (1967).12 Preiss, J., in Methods in Enzymology (New York: Academic Press, 1966), vol. 8, pp. 271-

275.13 Hansen, R. G., G. J. Albrecht, S. T. Bass, and L. L. Seifert in Methods in Enzymology

(New York: Academic Press, 1966), vol. 8, pp. 248-253.14 Kalckar, H. M., K. Kurahashi, and E. Jordan, these PROCEEDINGS, 45, 1776 (1959).15 These experiments on the partial diploids were first tried at 230C, but the strains proved

too unstable at this temperature because most of the cells proved to be of the wrong phenotypeupon dilution plating.

16 Nikaido, H., M. Levinthal, K. Nikaido, and K. Nakane, these PROCEEDINGS, 57, 1825(1967).

17 Englesberg, E., C. Squires, and F. Meronk Jr., these PROCEEDINGS, 62, 1100 (1969).18 Taylor, A. L., and C. D. Trotter, Bact. Rev., 31, 332 (1967).19 Lieberman, M. M., Ph.D. thesis, University of Chicago (1969).20 Uretz, R. B., and A. Markovitz, J. Bacteriol., 100, 1118 (1969).21 Foley, J. M., N. H. Giles, and C. F. Roberts, Genetics, 52, 1247 (1965).22 Davies, J., and F. Jacob, J. Mol. Biol., 36, 413 (1968).23 MOiller-Hill, B., L. Crapo, and W. Gilbert, these PROCEEDINGS, 59, 1259 (1968).24 Epstein, W., and J. R. Beckwith, Ann. Rev. Biochem., 37, 411 (1968).25 Lieb, M., Virology, 29, 367 (1966).26 Bernstein, H., and K. M. Fisher, Genetics, 58, 307 (1968).27 Alikhanian, S. I., and V. Z. Pogosov, Genetics, 61, 773 (1969).