MGi3

Mol Gen Genet (1984) 197:261-271

© Springer-Verlag 1984

Physical and genetic structure of the glpK-cpxA interval of the Escherichia coil K-12 chromosome Randi Albin and Philip M. Silverman Department of Molecular Biology, Division of Biological Sciences, Albert Einstein College of Medicine, Bronx, NY 10461, USA

Summary. Mutations at the cpxA locus of Escherichia coli K-12 affect cellular processes that are not otherwise related. We have now determined the physical and genetic structure of the E. coli chromosome in the region of cpxA (87.5 rain). Our results indicate that cpxA is a single gene. Previous studies showed cpxA to be linked to tpiA. We therefore isolated two tpiA + recombinant plasmids, pRA200 and pRA300, from EcoRI and BamHI digests of F'133, respectively. By genetic complementation or enzyme overproduction, the 9.5 kb EcoRI fragment in pRA200 was shown to include glpK, tpiA and cdh. The 13.6 kb BamHI fragment of pRA300 lacks glpK, but includes tpiA, pfkA and cpxA. Neither fragment complemented a deletion of the rha operon. These data indicate the chromosomal gone order: 87 min-rha-cpxA-pfkA-cdh-tpiA-glpK-88 min. The EcoRI and BamHI fragments overlap in an interval corresponding to about 8.2 kb of DNA. The total region of the E. coli K12 chromosome covered by the two fragments is about 15 kb. A terminal 2 kb EcoRI-BamHI fragment from pRA300 complemented the chromosomal cpxA2[Ts] allele with respect to isoleucine and valine synthesis, R N A bacteriophage sensitivity and surface exclusion in Hfr strains, and envelope protein composition. Complementation occurred when the fragment was subcloned in pBR325 but not when it was subcloned in pBR322, suggesting that the 2 kb fragment lacks expression sequences that are supplied by cat (chloramphenicol acetyltransferase gene) expression sequences of pBR325. The cpxA locus on the E. coli chromosome was established with respect to two chromosomal Tnl0 insertions by a combination of genetic and physical analyses. The locus established by those analyses was consistent with the location of the 2 kb EcoRI-BamHI fragment in the physical map of the region. Physical analyses of (rhapfkA) and (rha-tpiA) deletion strains showed that they lack cpxA and surrounding genes. Since these strains were viable, cpxA is not essential under all growth conditions.

Introduction Mutations at the cpxA and cpxB loci of the Escherichia coli K-12 chromosome affect two cellular processes that are not otherwise related. These are conjugal D N A donor and related activities in F' or Hfr strains (McEwen and Silverman 1980 a, b; Sambucetti et al. 1982) and the synthe-

Offprint requests to: P.M. Silverman

sis of isoleucine and valine (McEwen and Silverman 1980 a; Sutton et al. 1982). Accumulating evidence suggests that the cpx mutations affect the in vivo function, rather than the synthesis, of a protein specific for each process. We have identified these proteins as the F-plasmid TraJ protein (Sambucetti etal. 1982; Cuozzo et al. 1984), which is required for full expression of the remaining F-plasmid tra genes (Willetts 1977; Gaffney etal. 1983; Fowler etal. 1983), and acetohydroxyacid synthase I (Sutton etal. 1983; Eoyang and Silverman 1984), which catalyzes the first pair of homologous reactions of the parallel isoleucine and valine synthetic pathways (Umbarger 1978). The cpx mutations also lead to a complex but selective and characteristic pattern of alterations in envelope protein composition (McEwen and Silverman 1982; McEwen et al. 1983). On the basis of this fact, we proposed that the functions of both the TraJ protein and acetohydroxyacid synthase I depend on the composition or organization of the cell envelope. We further proposed that the cpx gene products are necessary at least under some conditions to maintain the envelope so as to support the cellular functions of the TraJ protein, acetohydroxyacid synthase I, and perhaps other proteins. A crucial assumption of our proposal is that each of the cpx mutations alters a single gene product. Otherwise, it could be argued that the observed effects of the cpx mutations merely reflect the complexity of the mutations themselves. That assumption is especially important with respect to the cpxA locus because a cpxA mutation alone leads to all of the cellular defects associated with mutations in both cpx genes, but the defects are quantitatively less severe in a cpxB + strain (McEwen and Silverman 1980b, c; 1982). In contrast, the cpxB1 allele by itself is cryptic. Previously, we mapped the cpxA locus between glpK and rha, at about 87.5 rain on the E. coli K-12 genetic map (Silverman 1982). Here, we extend those studies to locate cpxA with respect to neighboring genes and to a physical map of the glpK-cpxA interval. These and related experiments described in the accompanying communication (Albin and Silverman 1984) establish cpxA as a single gone.

Materials and methods

Bacterial strains, media, and growth conditions. The Escherichia coli K-12 strains used in these studies are described in Table 1. Cells were routinely grown aerobically

262 Table 1. Escheriehia coli K-12 strains Strain

Relevant genotype

Source or comment

AE2136

F- glpK1 cpxA2[Ts] cpxB1 recA1

AE2123

F- A(rha-tpiA)

This study. From AE2111 (Silverman, 1982) by conjugation with KL16-99 Silverman (1982); from ET2036

AE2074

F- cpxA + cpxB + recA1

J. McEwen; from AE2000 (McEwen and Silverman 1980a)

AE2072

F- cpxA2[Ts] epxB1 reeA1

ET2036

F- A(rha-tpiA)

J. McEwen; from AE2038 (Silverman t 982); isogenic with AE2074 except as indicated D. Fraenkel; Pahel et al. (1979)

ET2039

F- A(rha-pfka)

DF443

F- A(rha-tpiA)

AE2122 MWtl04

F- A(rha-pfkaA) cpxB1 HfrmetB + cdh-4::TnlO cpxA + MW1300 HfrmetB +zii-l::TnlO cpxA +

D. Fraenkel; Pahel et al. (1979) D. Fraenkel; from ET2036 (Babul 1978) Silverman (1982) C. Bulawa and C. Raetz C. Bulawa and C. Raetz

HfrH AEt031

Hfr leu + rpsL +

Laboratory stock

Hfr leu-6rpsLl04 cpxA + cpxB + zeb-1 :: TnlO

McEwen and Silverman (1980b)

AE1019

Hfr leu-6 rpsLl04 cpxA2[Ts] cpxB1

McEwen and Silverman (1980 a) ; isogenic with AE1031 except as indicated

in LB medium at 34 ° C; growth was monitored by the culture optical density at 600 nm or 660 nm. Antibiotic concentrations, where appropriate, were 25 gg/ml of ampicillin, 10 gg/ml of tetracycline, 25 gg/ml of chloramphenicol, and 100 gg/ml of streptomycin. Minimal medium contained Vogel-Bonner salts (Vogel and Bonnet 1956) supplemented with appropriate growth factors, each at 40 gg/ml, and a carbon source (0.2%), routinely glucose. Solid media were prepared with 1.5% Difco agar. Genetic methods. Complementation of glpK was determined by growth on glycerol; of tpiA by growth on fructose; and of pfkA by heavy growth on glucose (Silverman 1982). Complementation of rha was determined by colony color on MacConkey indicator plates containing 1% rhamnose. Complementation of the cpxA2[Ts] allele in a cpxB1 background was determined routinely by growth at 41°C in the absence of isoleucine and valine (McEwen and Silverman 1980b, c). In some instances, cpxA complementation was also determined by examination of outer membrane proteins (McEwen and Silverman 1982) and by the expression of F-plasmid-dependent surface exclusion and sensitivity to the donor-specific bacteriophage R17 (McEwen and Silverman 1980a).

PI transductions were carried out using Plvir, as previously described (McEwen and Silverman 1980b). Preparation and analysis o f outer membrane proteins. Cells (200 ml) were grown at 41 ° C to an optical density of 0.6 in minimal medium supplemented with ampicillin for strains containing a plasmid. Spheroplasts were prepared as previously described (McEwen and Silverman 1982) and disrupted by sonic oscillation at 4 ° C for three 0.5 min intervals, with intermittent periods for cooling, with a Branson sonifier set at maximum power (50-70 watts). Intact cells were removed by sedimentation at 1200×g for 20min. Crude envelopes were then isolated by sedimentation at 100,000 × g and 4 ° C for 1 h. Inner membrane proteins were solubilized by extraction for 10 rain at ambient temperature with a solution containing 10 m M Tris-HC1 (pH 8)/5 m M MgC12/2% (v/v) Triton X-100. Outer membranes were isolated by sedimentation as above, and the detergent extraction was repeated. The final outer membrane pellets were heated to 100 ° C for 3 min in electrophoresis sample buffer. The proteins were resolved by SDS-polyacrylamide gel electrophoresis, fixed for 30 min at 4°C in 20% TCA, and then stained (McEwen and Silverman 1982). D N A preparations. Chromosomal D N A was prepared as described by Womble et al. (1977). Small plasmid D N A was prepared as described by Humphreys et al. (1975); in some cases the D N A was prufied through two CsC1 gradients. F'133 D N A was isolated from AE2072 containing that plasmid, as described by Hansen and Olsen (1978). For rapid screening of clones, plasmid D N A was prepared by the alkaline lysis method (Maniatis et al. 1982). Restriction fragments smaller than 10 kb were purified by electrophoresis through a 3.5% polyacrylamide gel using a Tris/borate buffer system (Maniatis et al. 1982), followed by electroelution, phenol/chloroform extraction and two ethanol precipitations (McDonell et al. 1977). Restriction fragments larger than 10 kb were purified by zone sedimentation through a linear 10%-30% sucrose gradient containing 50 m M Tris-HC1 (pH 8)/0.1 m M EDTA/1 M NaC1/ 0.05% sarkosyl. Sedimentation was at 27,000 rev/min and 20 ° C for 23 h in the Beckman SW41 rotor. Gradient fractions were examined by agarose gel electrophoresis. Appropriate fractions were pooled and dialyzed at 4 ° C for 18 h against 10 m M against 10 m M Tris-HC1 (pH 8)/1 m M EDTA. The D N A was then concentrated by ethanol precipitation. Restriction endonuclease digestions. Digestion conditions were as recommended by New England Biolabs. Restriction fragments were analyzed by 0.8% or 1% agarose gel electrophoresis using a Tris-acetate buffer system (Maniatis et al. 1982). Fragment sizes were determined using HindIII fragments of bacteriophage lambda DNA, and HaeIII fragments of bacteriophage ~bX174 R F I D N A as standards. Comparisons were made by computer, using the program SIZER obtained from Dr. J.W. Chase. Cloning methods. Restriction fragments were cloned into pBR322 or pBR325 digested with appropriate restriction endonucleases and then with calf intestine alkaline phosphatase. Ligations were for 18 h at 15 ° C in the presence of bacteriophage T4 D N A ligase. Cells were transformed essentially as described by Cohen et al. (1982).

263 consistent with the most recent edition of the E. coli K-12 linkage map (Bachman 1983).

Filter (Southern blot) hybridizations. Chromosomal D N A (5 pg) was digested with the appropria~:e restriction endonuclease and the fragments separated by electrophoresis through a 0.8% agarose slab gel. Routinely, I ~tg of bacteriophage lambda D N A was included as an internal standard for complete digestion. Nitrocellulose blots were prepared as described by Maniatis et al. (1982), except that neutralization was carried out in 1 M Tris-HC1 (pH 6)/3 M NaC1. Radioactive probes were prepared by nick translation of restriction fragments in the presence of [~-32p]dCTP essentially as described by Rigby et al. (1977), except that DNase I was added to a final concentration of 0.067 pg/ml for fragments smaller than 10 kb. Hybridizations were for 48 h, using ] 06 cpm of radioactive probe. Autoradiographic exposures were made on D u p o n t Cronex medical x-ray film using a D u p o n t Cronex Xtra Life intensitfying screen. In the figures shown, contiguous lanes were derived from the same agarose gel, but in some instances the order of the lanes was rearranged for clarity.

Isolation of the tpiA + plasmids pRA200 and pRA300 Though chromosomal cpxA mutant alleles are recessive to the cpxA + allele carried on F' plasmids, and growth in the absence of isoleucine and valine in principle provides a positive selection for epxA complementation, the high reversion frequency of cpxA mutants to an Ilv + phenotype made a direct approach to the clonal isolation of cpxA unsuitable (McEwen and Silverman 1980b). Instead, we exploited the results of a previous study, which placed cpxA within the glpK-rha interval (87.2-87.9 rain) and linked to tpiA, the gene for triose phosphate isomerase (Silverman 1982). Accordingly, we digested F'133 D N A , which includes all of the 8 2 8 8 rain interval of the E. coli K-12 chromosome (Armstrong and Fan ] 976), with, respectively, BamHI, EcoRI, HindIII and PstI. The fragments were ligated into pBR322 or pBR325 digested with the appropriate enzyme, and the D N A was used to transform the tpiA deletion strain AE2123. We selected Tpi + transformants for growth on minimal plates with fructose as sole carbon source (Pahel et al. 1979; Thomson et al. 1979) and tested the tpiA + plasmids so obtained for their ability to complement mutations in other genes near tpiA. Tpi + transformants were obtained from BamHI, EcoRI and PstI digests of F'133. However, preliminary analyses showed the PstI fragment containing tpiA to be about I kb, a size we considered unlikely to encode more than one average-sized protein. The EcoRI fragment and the BamHI fragment were larger, and the respective plasmids containing these fragments, designated pRA200 and pRA300, were chosen for further analysis.

Materials. Restriction endonucleases, T4 D N A ligase, D N A polymerase I and bacteriophage D N A restriction fragments were obtained from New England Biolabs. Calf intestine alkaline phosphatase was obtained from Boehringer Mannheim. DNase I was obtained from Worthington Biochemicals and stored at - 8 0 ° C in 50 pl aliquots of a 1 rag/m1 solution in 0.01 N HC1. Nitrocellulose BA85 was obtained from Schleicher and Schuell. Unlabeled nucleotides were obtained from Boehringer Mannheim. [~-32p]dCTP ( > 3 0 0 0 Ci/mmol) was obtained from Amersham. Other materials were obtained from standard commercial sources. Results

For reference, Fig. 1 summarizes the physical and genetic map of the E. coli K-12 chromosome between glpK an cpxA. The map is based on genetic and physical analyses described below. It is unambiguous and except for cpxA and cdh,

tl

I

I

xd glp KIO0 E

B

B

,.o I \

glpK I

\

1~1.3

[

/

/

I

I

i

Analyses of pRA200 and pRA300 by complementation The tpiA deletion strain used to isolate pRA200 and pRA300 by tpiA complementation lacked a chromosomal I /

pLCl6-4

P

P

I,.o

I

t#iA I I

2.,

E

E

j

I

cdk-4...TnlO #fkA I PRA500 I ] pRA200___I

I I

z/7-D.TnlO

cpxA I

B

I

LpRA 310..-J

AE2122 ET2039

I /

Fig. 1. Physical and genetic map of the glpK-cpxA interval of the Escherichia coli K-12 chromosome. Plasmids pRA300, pRA200, and pRA3]0 are described in this coriamunication. Plasmid pLC16-44 (tpiA+ pfkA + ) was described in Thomson et al. (1979) and Shimosaka et al. (1982). Bacteriophage 2 dglpKlO0 was described in Conrad et al. (1984). The apparent deletion of 0.3 kb from the glpK gene of pRA200 is based on the analysis of 2dglpKt00 by Conrad et al. (1984) and by J.R. Johnson (personal communication). The cdh-4::TnlO and zii-l:: TnlO alleles were isolated by Bulawa and Raetz (personal communication). The insertion sites were located on the physical map as described in this communication. The deletion strains AE2122, ET2039, ET2036 and DF443 were described in Babul (1978), Pahel et al. (1979) and Silverman (i 982). The deletion endpoints in the glpK-cpxA inverval were determined as described in this communication. All of the deletions originated from a rha::2CI857 allele. The extent of the deletions in the 86-87 min interval was not determined. Genes glpK, pfkA and epxA are interrupted by BamHI or EcoRI restriction sites, as indicated. The locations of tpiA and cdh with respect to restriction sites is only approximate. B=BamHI site; E=EeoRI site; P=PstI site. The numbers between restriction sites are DNA fragment sizes in kilobase pairs

264 F'133

pRA300 pRA200

pBR325~pBR322

BamH1 [ EcoR1 E

P

P

B

pR200 A pRA 300

(fpi+)

(fpi+)

23.1 - -

rha

opxA -

+

pfkA

-

+

cdh

+

NT

fpiA glpK

+ +

+

Fig. 2. Construction and complementation properties of pRA200 and pRA300. See the text for experimental details of plasmid construction. The plasmids were introduced into relevant strains by transformation. Complementation of rha, tpiA, and pfkA were determined in strain AE2123 (A(rha-tpiA)).Complementation ofglpK and cpxA were determined in strain AE2136 (glpK1 cpxA2[Ts] epxB1). The presence of edh was inferred from enzyme overproduction (Bulawa and Raetz, personal communication)

9.4-6.6--

4.4

2,3

m

i

2.0--

pfkA gene as well. The pRA200 and pRA300 transformants of that strain were tested for pfkA complementation by their ability or to grow with glucose as carbon source (Pahel et al. 1979; Thomson et al. 1979). The untransformed strain and the strain transformed with pRA200 grew poorly under these conditions, indicating a P f k - phenotype; in contrast, the strain containing pRA300 grew well, indicating that pRA300 carries pfkA. Both plasmids were used to transform the glpK1 cpxA2[Ts] cpxBl strain AE2136. Ampicillin-resistant transformants were screened for glpK1 complementation by growth on glycerol and for cpxA2[Ts] complementation by growth at 41 ° C in the absence of exogenous isoleucine and valine (McEwen and Silverman 1980b; Silverman 1982). This analysis showed that pRA200, but not pRA300, included glpK. Conversely, pRA300, but not pRA200, complemented the cpxA2[Ts] allele. However, only about 5% of the antibiotic-resistant isolates transformed by pRA300 were Ilv ÷ at 41 ° C. Plasmid D N A isolated from these transformants and digested with BamHI yielded two fragments with the expected electrophoretic mobilities (see below), but the relative masses of the two fragments, judged by the intensity of ethidium bromide straining, indicated a considerable molar excess of vector DNA. Sequential transformations with plasmid D N A from the Ilv + transformants yielded essentially the same result; only a fraction of the ampicillin-resistant transformants wer Ilv + at 41 ° C, and plasmid D N A isolated from these transformants and digested with BamHI yielded a relative excess of vector DNA. We did not encounter this problem with either pRA200 or with cloned subfragments derived from pRA300 (see below), and have not further investigated its source. As judged by enzyme overproduction, pRA200 includes cdh, the structural gene for CDP-diglyceride hydrolase, which is located between tpiA and pfkA (Bulawa and Raetz, personal communication); pRA300 was not tested. Neither pRA200 nor pRA300 complemented the rha deletion of AE2123. These complementation data, summarized in Fig. 2, indicated that pRA200 and pRA300 contain overlapping D N A fragments from F'133. The gene order (87 min)-

--

1.35

--1.1)8 --

0.87

Fig. 3. Restriction endonuclease digestion of tpiA+ fragments. The BamHI fragment from pRA300 and the EcoRI fragment from pRA200 were isolated by zone sedimentation as described in Materials and methods. The isolated fragments were then digested to completion with PstI(P), BamHI (B) or EeoRI (R), as appropriate. Restriction subfragments were separated by 1% agarose gel electrophoresis and visualized after staining with ethidium bromide. Size standards indicated on the left were bacteriophage 2HindIII fragments; those on the right were bacteriophage ~X174 RFI HaeIII fragments

cpxA-pfl~A-cdh-tpiA-glpK-(88 min) revises that in the current version of the E. coli K-12 linkage map with respect to cdh and cpxA; further evidence for these revisions is presented below. Physical analyses of pRA200 and pRA300 As shown by studies of the tpiA +pfkA + plasmid pLC16-44 by Shimosaka et al. (1982), the D N A spanning tpiA and pfkA contains three PstI sites and one EcoRI site, defining PstI fragments of about 1 and 1.3 kb and a PstI-EcoRI fragment of about 2.3 kb. Moreover, those studies implied that the EcoRI site interrupts pfkA. Analysis of the cloned fragments from pRA200 and pRA300 yielded comparable data. Both fragments contained three PstI sites, yielding common subfragments of 1.0 and 1.4 kb (Fig. 3). These correspond to the 1 and 1.3 kb PstI fragments ofpLC-16-44 (Shimosaka et al. 1982). The EcoRI fragment from pRA200 yielded a 2.1 kb subfragment, which corresponds to the 2.3 kb EcoRI-PstI fragment pLC16-44 (Shimosaka et al. 1982). Since pRA200 failed to complement a pfkA deletion,

265 we suggest that the EcoRI site of the 2.1 kb subfragment lies within pfkA; this is consistent with the data of Shimosaka et al. (1982). The remaining 4.8 kb EcoRI-PstI subfragment from pRA200 presumably includes glpK. The EcoRI fragment of pRA200 also contains a single BamHI site, about 1.3 kb from one end (Fig. 3). Since the overlapping BamHI fragment of pRA300 failed to complement the glpK1 mutation, we infer that the BamHI site either interrupts glpK or lies between glpK and tpiA. The former seems more likely, since the g,~K coding sequence of 2dglpKlO0 contains a BamHI site (Conrad et al. 1984; J. Johnson, personal communication). One difference between our analysis of pRA200 and that of 2dglpKlO0 by Conrad et al. (1984) is the presence of a second BamHI site in or near the glpK promoter of 2dglpKlO0. According to Conrad et al. (1984), digestion of the EcoRI fragment of pRA200 should have yielded subfragments of 1.0 and 0.6 kb, rather than the single 1.3 kb subfragment we observed. Preliminary studies by J. Johnson (personal communication) suggest that the EcoRI fragment of pRA200 suffered a 300 bp deletion that included one of the BamHI sites. This deletion substantially reduced glpK promoter activity, as judged by glycerol kinase levels, but sufficient enzyme was synthesized to account for the ability of pRA200 to complement the glpK mutation. The apparent deletion in pRA200, with loss o f a BamHI site, is indicated in Fig. 1. The PstI-BamHI subfragments derived from the BamHI fragment of pRA300 were 3.7 and 8.6, kb (Fig. 3). One of these, extending towards glpK, should be 1.3 kb shorter than the 4.8 kb PstI-EcoRI subfragment of pRA200 that includes glpK; this corresponds to the subfragment we measure as 3.7 kb. The larger subfragment must therefore extend through pfkA to the other end of the BamHI fragment. Digestion of the BamHI fragment with EcoRI yielded three subfragments. The largest of these should be the BamHI-EcoRI subfragment extending from the BamHI site in glpK to the EcoRI site in pfkA. The order of the other two subfragments, 3.4 and 2.0 kb, cannot be inferred from the data in Fig. 3. However, as shown below, the 2.0 kb subfragment contains one EcoRI and one BamHI terminus. The 3.4 kb fragment must therefore be an internal EcoRI fragment. The genetic and physical structures of the tpiA ÷ EcoRI and BamHI fragments are summarized in Fig. 1. The data in Fig. 3 indicate that the Eco fragment is 9.5 kb and that the BamHI fragment is 13.6 kb.

Construction of the cpxA + plasmid pRA310 The BamHI fragment of pRA300 appeared to include cpxA, though the apparent instability of the plasmid made this conclusion tentative. However, the complementation properties of pRA200 and pRA300 implied that if pRA300 carries cpxA, the gene must lie to the right ofpfkA, as indicated in Fig. 1. We therefore digested the BamHI fragment of pRA300 with EcoRI to generate the three subfragments shown in Fig. 3. These were ligated into pBR325 digested either with EcoRI to accept EcoRI subfragments or with EcoRI and BamHI to accept EcoRI-BamHI subfragments (Fig. 4). After ligation the DNA was used to transform the cpxA2[Ts] cpxB1 strain AE2136. Ampicillin-resistant transformants, 500 from each DNA preparation, were screened for isoleucine and valine prototrophy at 41 ° C. Six such transformants were obtained from the EcoRI-

BamH1 ligation, and none from the EcoRI ligation. All six Ilv + transformants were sensitive to chloramphenicol and tetracycline, as expected, and all six contained a recombinant plasmid, designated pRA310, composed of the 4.4 kb EcoRI-BamHI vector fragment and a 2.0 kb EcoRIBamH1 fragment (Fig. 4). Each of the six plasmids was used to transform strain AE2136. Fifty ampicillin-resistant colonies from each transformation were tested for growth at 41 ° C in the absence of exogenous isoleucine and valine; all 300 colonies were Ilv +. Finally, the cloned 2.0 kb fragment hybridized to the 13.6 kb BamHI fragment ofpRA300 (not shown) and to a 13.6 kb BamHI genomic fragment (see below). We conclude that the 2.0 kb fragment of pRA310 corresponds to the terminal EcoRI-BamHI subfragment from pRA300, as indicated in Fig. 1. Moreover, this fragment complemented the cpxA2[Ts] allele for isoleucine and valine synthesis, but unlike pRA300, pRA310 appears stable, in that all of the ampicillin-resistant transformants were Ilv ÷ as well. Construction of pRA311 Since the EcoRI site of pBR325 interrupts the chloramphenicol acetyltransferase (cat) gene (Prentki et al. 1980, it remained possible that the 2.0 kb fragment of pRA310 lacks cpxA expression sequences, which could be supplied by vector cat sequences (see Fig. 4). This would have to occur if cpxA is part of an operon whose expression sequences were removed when the 2.0 kb EcoRI-BamHI fragment was separated from the adjacent bacterial DNA in pRA300. To examine this possibility, we isolated and cloned the 2.0 kb fragment from pRA310 into pBR322 digested with EcoRI and BamHI. pBR322, lacking the cat gene of pBR325, would be unable to supply expression sequences to cpxA in the orientation determined by the different restriction sites (Fig. 4). Using the screen described above, we were unable to identify any Ilv + derivatives of AE2136 among 250 ampicillin-resistant transformants. However, several of these were sensitive to tetracycline, and further analysis of the plasmids in these isolates showed them to contain the 2.0 kb EcoRI-BamHI fragment; these plasmids were designated pRA311 (Fig. 4). Thus, it appears that the 2.0 kb fragment does lack sequences required for cpxA expression, which are supplied in pRA310 by cat sequences. Experiments to be reported elsewhere (Albin and Silverman 1984) show that in fact the EcoRI site at one end of the 2.0 kb fragment lies within the cpxA coding sequence, and that pRA310 encodes a biologically-active chloramphenicol acetyltransferase-CpxA fusion protein.

Complementation of the cpxA2[ Ts] allele by pRA310 In addition to an Ilv phenotype at 41 ° C, the cpxA2[Ts] allele in cpxB1 strain affects F-plasmid-dependent DNA donor and related activities (McEwen and Silverman 1980 a, b), and the protein composition of the cell envelope (McEwen and Silverman 1982; McEwen et al. /983). The effects of pRA310 and pRA311 on these properties were examined. The cpx mutations reduce expression of the F-plasmid tra genes (Sambucetti et al. 1982). Thus, mutant cells are resistant at 41 ° C to donor-specific bacteriophage, like R17, reflecting the inability of mutant cells to elaborate F-pili (McEwen and Silverman 1980a, b). Mutant Hfr cells are also relatively good conjugal recipients, reflecting the effect

266

B

B Eco

RI~

Ca/,.,.,.~/f, i / t~ EcoR

j /

EcoR1

~tet

. . . . . . .

,~RA3~0 ......

BomH1

I

6.;8)b ~ r

J

÷,

~

Eco R1 + Barn H1

2

B

Eco RI+ BamH 1

bla//

EcoRI ~ c p x A

pRA31t

6kb Cpx-}

(AmpRTets

Barn

HI

te._Zt

of the cpx mutations on the expression of the traS and traT genes (McEwen and Silverman 1980a, b; Sambucetti et al. 1982). pRA310, but not pRA31J, restored R17 sensitivity and surface exclusion to the cpxA2[Ts] cpxBI Hfr strain AEJ019 (Table2). Hence, by these two criteria, pRA3J0 complements the cpxA2[Ts] allele with respect to F-plasmid transfer functions. We examined the effect of pRA310 on donor activity itself, but for reasons we do not understand the results of these experiments were variable. The cpx mutations also cause a complex but characteristic pattern of alterations in the protein composition of the

Fig. 4. Construction and properties of pRA310 and pRA3tl, pRA310 was derived from an EeoRI digest of the 13.6 kb BamHI fragment of pRA300. It consists of the 2 kb EeoRI-BamHI terminal subfragment cloned into pBR325 digested with EcoRI and BamHI. pRA3J1 consists of the 2 kb EcoRI-BamHI fragment from pRA3t0 cloned into pBR322 digested with EeoRI and BamHI. The cpxA complementing activity of pRA310 is attributable to a gene fusion between the chloramphenicol acetyltransferase gene of pBR325 and cpxA, as indicated (Albin and Silverman 1984)

inner and outer bacterial membranes. The outer membrane of mutant cells lacks detectable OmpF matrix porin, whereas the amount of OmpC porin remains unaffected. In addition, the amount of murein lipoprotein is reduced, and two large proteins absent or present at low levels in outer membrane from epx ÷ cells become more abundant in that from cpx mutant cells (McEwen and Silverman 1982; McEwen et al. 1983). As shown in Fig. 5, pRA3J0, but not pRA311, restored the pattern of mutant cell outer membrane proteins to that characteristic of cpx + cells. Thus, by the three criteria of isoleucine and valine prototrophy, expression of F-plasmid tra functions, and cell envelope protein composi-

267 Table 2. Complementation of the cpxA2[Ts] allele by pRA310 Hfr strain

Relevant genotype

Plasmid

Bacteriophage Rt 7 sensitivity"

Surface exclusion b

AE1019

leu-6 cpxA2[Ts] epxBl rpsLl04

pRA3t0

S

pRA311

R

77

AE1031

leu-6 epxA + cpxB + rpsLl04

pRA310

S

24,000

pRA31 l

S

15,000

7,700

13.6-

a Measured at 41 ° C. S=sensitive; R =resistant b Measured at 41 ° C. Values are the inverse of the number of Leu + Str R recombinants/recipient cell/ml with HfrH (leu + rpsL +) as donor 4.4 m

3.4-

ompF omp~'~ ompA

1

2

3

4

Fig. 6. Physical localization of the zii-l::TnlO insertion site by filter hybridization. Chromosomal DNA from the strains indicated at the top of lanes 2-4 was digested with EeoRI. The fragments were separated by agarose gel electrophoresis and blotted to nitrocellulose filter paper. Electrophoretically-purified 3.4 kb EcoRI fragment from pRA300 was labeled in vitro by nick-translation and used for hybridization. Lane 1 shows hybridization of the labeled fragment to the unlabeled fragment blotted from the agarose gel. The size standards were pBR322 digested with EeoRI and the 13.6 kb BamHI fragment from pRA300

LP-1

2

3

4

Fig. 5. Outer membrane proteins from CpxA + and CpxA strains. Proteins were isolated and separated by SDS-polyacrylamide gel electrophoresis as described in Materials and methods. Lane t, strain AE2074 (epxA + cpxB+); lane 2, AE2072 (cpxA2[Ts] cpxB1); lane 3, AE2072 transformed with pRA310; lane 4, AE2072 transformed with pRA311. The OmpF and OmpC matrix porins, the OmpA protein and the murein lipoprotein (LP) are indicated to the left. The arrowheads at the top left indicate two high molecular weight proteins that are characteristically more abundant in the outer membrane from cpx mutant cells. For comparison see McEwen and Silverman (1982) tion, p R A 3 1 0 fully complemented the cpxA2[Ts] allele, whereas pRA311 lacked any complementing activity.

Genetic and physical localization o f cpxA on the E. coli K-12 ehromosome To establish the c h r o m o s o m a l locus o f cpxA we first m a p p e d the c h r o m o s o m a l locus o f the cpxA2[Ts] allele with respect to two linked T n l 0 insertions by PI transduction, and then identified the physical location o f both insertions with respect to the restriction maps o f p R A 2 0 0 and pRA300. The TnlO insertions were isolated by Bulawa and Raetz (personal communication). One, cdh4: : T n l 0 , lies in the structural gene for CDP-diglyceride hydrolase. The

second, designated zii-l::YnlO on the basis o f the experiments reported here, is linked by P1 transduction to pfkA, but was not otherwise characterized. Three factor crosses with the insertion strains as PI donors and the metB1 cdh + zii + cpxA2[Ts] cpxB1 strain AE2038 as recipient are shown in Table 3; these crosses established the order metB-cdh (or zii)-cpxA. They also suggested that the zii insertion locus is closer to cpxA than the cdh insertion locus. To determine the physical location o f the two TnlO insertions, we exploited the fact that TnlO contains a single EcoRI site (Jorgensen et al. 1979). G e n o m i c D N A from the two insertion strains M W l I 0 4 (cdh-4::TnlO) and M W I 3 0 0 (zii-l::TnlO) and the cdh + zii + control strain AE2000 were digested with EcoRI. Southern blots o f the digested D N A were incubated with restriction fragments isolated from p R A 2 0 0 or p R A 3 0 0 and labeled in vitro. A c h r o m o s o m a l segment containing a TnlO insertion should yield two EcoRI fragments, and hence two radioactive bands in a Southern blot, whereas the same segment lacking the insertion should yield only one band. Moreover, the sum o f the two fragment lengths should exceed the single fragment length by 9.3 kb, corresponding to the length o f TnlO itself. By this analysis, the zii-l::TnlO insertion was located

268 Table 3. P1 transductional mapping of the cdh-4::TnlO and ziil : : T n l O alleles with respect to m e t B and cpxA Experiment a

Selected

Trans-

No. of

donor allele

ductant genotype b

transductants

A.

(%)

1. Donor: m e t B +

metB +

cdh-4:: TnlO cpxA+ Recipienti metB1 cdh+ cpxA2[Ts]

2. Donor: m e t B + zii-1 :: TnlO cpxA + Recipient: metB1 zii +

cdh

cpxA

D

D

53 (31)

D

R

45 (26)

R R

D R

0 (0) 75 (43)

9.5

9.5 m

metB +

zii

cpxA

D

D

52 (32)

D

R

12 (7)

R

D

0 (0)

R

R

lOl (61)

cpxA2[Ts]

a The PI donor strain in experiment 1 was M W t I 0 4 and in experiment 2, MW1300

1

2

3

1

2

3

Fig. 7A, B. Physical localization of the cdh-4:: TnlO insertion site.

b The cdh and zii alleles were scored by sensitivity or resistance to tetracycline. The cpxA alleles were scored as ability or inability to grow at 41° C in the absence of exogenous isoleucine and valine. D = donor allele; R = recipient allele

See the legend to Fig. 6 for details. A Chromosomal DNA was digested with EcoRI and hybridized to the 9.5 kb EcoRI fragment from pRA200 labeled in vitro. B Same as A, except hybridization was to the 2.1 kb PstI-EcoRI fragment from pRA200 (see Fig. 1)

within the 3.4 kb EcoRI subfragment contained in pRA300 (Figs. 1 and 6), and the cdh-4::TnlO insertion was located within the 9.5 kb EcoRI fragment of pRA200 (Figs. 1 and 7A). An additional experiment placed the cdh-4::TnlO insertion locus and hence cdh, between tpiA and pfkA. Thus, the terminal 2.1 kb PstI-EcoRI subfragment from pRA200, which contains part of pflcA, hybridized primarily to the genomic EcoRI fragment of MW1104 containing the lesser amount of chromosomal D N A (Fig. 7 B, lane 2) as judged by the relative intensities of the radioactive bands in Fig. 7 A, lane 2. However, we could detect a small amount of hybridization to the larger EcoRI fragment, suggesting that the TnlO insertion site in MW1104 lies within the D N A corresponding to the 2.1 kb PstI-EcoRI fragment. This is consistent with the observation that this fragment, subcloned from pRA200, leads to a 50-fold overproduction of hydrolase activity (Bulawa and Raetz, personal communication). This analysis, summarized in Fig. 1, is consistent with the three factor genetic crosses shown in Table 3, and with the localization of epxA, or most of it, within the 2.0 kb EcoRI-BamHI subfragment of pRA310. Furthermore, these data and additional hybridization experiments shown below indicate that the physical and genetic maps derived from analyses of pRA200 and pRA300 generally reflect those of the chromosome itself.

A(pfkA-tpiA) strains, could transduce a metB1 cpxA2[Ts] cpxB1 strain to a Met + Ilv + phenotype. These experiments suggested that pfkA deletion strains still contained cpxA which, with other genetic data, placed cpxA between glpK and tpiA (Silverman 1982). In contrast, the data reported above place cpxA between pfl~A and rha, in which case the pjT~A deletion strains should be cpxA deletions as well. To establish whether or not this is the case, we examined three deletion strains by Southern blot analysis. DF443 contains the A(tpiA-pfkA) allele (Babul 1978). ET2039 and AE2122 contain independently-isolated pfkA deletion alles. The former was isolated by Pahel et al. (1979). We isolated AE2122 as a primary deletion strain containing the cpxB1 allele (Silverman 1982). Our general approach was analogous to that described above to map the TnlO insertions. The absence of a radioactive band in a Southern blot of chromosomal D N A fragments was taken to mean that the chromosomal D N A segment corresponding to that used as the radioactive probe is entirely included within the deletion. A radioactive band with an altered electrophoretic mobility relative to a control strain with an intact chromosome was taken to mean that one deletion end point lay within the chromosomal segment corresponding to the probe fragment. By these criteria, all the deletions terminate at one end within the chromosomal segment corresponding to the BamHI fragment of pRA300 (Fig. 8A). One end point of the deletion allele in ET2039 lies within the D N A segment defined by the terminal 2.1 kb PstI-EcoRI subfragment of pRA200 (Fig. 8 B). This entire segment is deleted in DF443, whereas it appears to be intact in AE2122 (Fig. 8B). None of the deletion strains contains chromosomal D N A corresponding to the adjacent 3.4 kb EcoRI subfragment from pRA300 (Fig. 8 C) or to the 2.0 kb EcoRI-BamHI subfragment that in pRA310 complements the cpxA2[Ts] mutation (Fig. 8 D). We infer from these re-

Physical analyses of deletion strains Several pfkA deletion strains have been isolated by imprecise excision of 2CI857 inserted in the rha locus (Pahel et al. 1979; Silverman 1982). One of these deletion alleles extends through tpiA, as judged by genetic criteria. The others delete all or part ofpJkA, but leave tpiA intact. Previously, we showed that A(pfkA) tpiA + strains, but

not

269

A.

B.

D.

13.6--

13.6-9.5--

1

2

3

4

1

2

3

4

1

2

3

4

1

2

3

4

Fig. 8A-D. Physical localization of deletion endpoints in the glpK-cpxA interval. See the legend to Fig. 6 for details. A Chromosomal DNA from the deletion strains indicated at the top of the panel and control strain AE2000 were digested with BamHI. Hybridization was to the 13.6 kb BamHI fragment from pRA300. B Chromosomal DNA was digested with EcoRI and hybridized to the 2.l kb PstI-EcoRI fragment from pRA200. C Same as B, except hybridization was to the 3.4 kb EeoRI fragment from pRA300. D Same as A, except hybridization was to the 2 kb EcoRI-BamHI fragment from pRA310 sults that the deletion in DF443 extends through pfkA and tpiA towards glpK, as the genetic data suggest, and the deletion in AE2122 ends within pJ7cA, just before the EcoRI site in that gene (see Fig. 1). Thus, the physical and genetic properties of these deletions are unambiguous and consistent with each other and with the data presented above (Fig. 1). Moreover, all of the deletions in fact lack cpxA. Finally, the 2.0 kb EcoRI-BamHI fragment from pRA310 hybridized to indistinguishable BamHI or EcoRI genomic fragments from cpxA + and cpxA2[Ts]] strains (not shown), suggesting that the cpxA2[Ts] allele is not a major D N A rearrangement.

Discussion Mutations in the cpxA and cpxB genes of E. coli K-12 affect cellular processes that are otherwise unrelated. In previous studies we were unable to separate genetically the different effects of a cpxA mutation in a cpxB mutant background, and we assumed that mutations at the cpxA locus altered a single gene with pleiotropic effects on the cell (McEwen and Silverman 1980b). The results described above support this view. All of the defects associated with the cpxA2[Ts] allele in a cpxB1 background were remedied in the presence of a 2 kb restriction fragment isolated from F'133. Moreover, cpxA complementation by the 2 kb restriction fragment only occurred when the vector could supply expression sequences, as in pRA310, pRA311, which also contained the 2 kb restriction fragment but lacked appropriate vector expression sequences, had no cpxA complementing activity. As we show in the accompanying communication, a single fusion protein consisting of the N H 2terminal 73 codons of chloramphenicol acetyltransferase and 25 kDa of protein derived from the 2 kb restriction fragment immediately adjacent to the EcoRI site is responsible for the cpxA complementing activity of pRA310. Thus, the cpxA locus consists of a single gene whose coding sequence is interrupted by an EcoRI site. It remains to be determined whether or not the cpxA

gene is part of a larger transcriptional unit, since we have not localized its normal expression sequences. It is clear from the construction of pRA310 that transcription of the cpxA sequence it contains is from the EcoRI site towards the BamHI site, and consequently that the chromosomal cpxA gene is transcribed counter-clockwise with respect to the E. coli K-12 linkage map (Bachmann 1983). Hence, native cpxA expression sequences are probably located on the 3.4 kb EcoRI fragment of pRA300. In support of this assumption is the observation that the zii-i: :TnlO insertion allele did not inactivate cpxA expression (see Table 3). Since the zii-1 insertion locus lies within the genomic 3.4 kb EcoRI fragment, cpxA expression sequences presumably lie within that fragment, between the zii-1 insertion site and the EcoRI site. We have not determined the zii-1 locus precisely enough to estimate the maximum amount of D N A that can be assigned to the cpxA transcriptional unit. A previous analysis of pfkA and tpiA deletion strains by P1 transduction identified a locus between glpK and tpiA that restored to a cpxA2[Ts] cpxB1 strain the ability to grow at 41°C in the absence of exogenous isoleucine and valine (Silverman 1982). We identified this locus as cpxA itself. Since no other gene mapped to that interval, we concluded that cpxA is a new addition to the E. coli K-12 linkage map. That locus for cpxA is clearly incorrect. The experiments reported here establish by genetic, physical, and physiological criteria that cpxA is located between pfkA and rha, about 10 kb counter-clockwise from glpK and 3.4kb counter-clockwise from pfkA. Moreover, pRA200, which contains D N A corresponding to the entire interval between glpK and pfkA, failed to complement the chromosomal cpxA2[Ts] allele, which previous studies showed to be recessive to the cpxA + allele of F'105 (McEwen and Silverman 1980b). There are several possible explanations for our previous results (Silverman 1982), but additional experiments are necessary to determine which of these explanations is correct. In any case, the deletion strains clearly lack cpxA and surrounding genes. Since these strains, including the

270

A(cpxA) cpxB1 strain AE2122, are viable, cpxA and cpxB are not essential under all growth conditions. W e have considered the relation between cpxA and other genes placed within the pfkA-rha interval of the E. eoli K-12 linkage m a p (Bachmann 1983). These include sodA (Mn-dependent superoxide dismutase), kdgT (uptake o f 2-keto-3-deoxygluconate), and ssd (energy coupling factor [ecfB], u p t a k e o f aminoglycoside antibiotics, resistance to eolicin K, elevated levels o f serine dehydrase). Touati (1983) recently m a p p e d sodA at 87.5 min and to a 7 kb EcoRI fragment. T h a t fragment contained a single BamHI site that divided it into 4.85 and 2.15 kb subfragments. The sodA gene was located within the larger subfragmerit. The smaller subfragment m a y be the same as the 2 kb EcoRI-BamHI fragment that in pRA310 complements cpxA mutations because that fragment also hybridizeds to a c h r o m o s o m a l EcoRI fragment o f 7 kb (unpublished observation). If so, soda is located between cpxA and rha. Normally, E. coli K-12 strains are not able to use 2-keto3-deoxy-gluconate ( K D G ) as sole carbon and energy source. Spontaneous mutations at the kdgR locus (41 min) permit such growth by derepressing the expression o f genes required for K D G uptake (kdgT; 8%87.5 min) and metabolism (Pouyssegur and Lagarde 1973; Pouyssegur and Stoeber 1974). Genes kdgT and cpxA m a p in the same region, as do genes kdgR and cpxB. However, in preliminary experiments we found that cpxB1 strains cannot grow on K D G as c a r b o n and energy source; hence, cpxB1 is unlikely to be a kdgR allele (P. Silverman, unpublished experiments). Furthermore, spontaneous K D G + mutants o f a cpxA2[Ts] cpxB1 strain could be isolated, and these retained their Cpx m u t a n t phenotype; hence, the cpxA2[Ts] m u t a t i o n is unlikely to be a kdgT allele. Finally, using P1 transduction we found only 50% linkage between kdgT and cpxA, which also indicates that, while these two genes are fairly close to each other, they are n o t identical (P. Silverman, unpublished observations). A simple test for ssd mutations is inability to grow with succinate as c a r b o n source (Morris and N e w m a n 1980; N e w m a n et al. 1982). The cpxA2[Ts] allele does not have this effect (P. Silverman, unpublished observations). Thus, the conclusion that cpxA is a new addition to the E. coli K-12 genetic m a p appears correct (Silverman 1982). It is nevertheless interesting that cpxA, ssd and kdgT, all o f which lie in the pfkA-rha interval, each affects enveloperelated functions.

Acknowledgements. We thank C. Bulawa, C. Raetz and J. Johnson for providing us with materials and unpublished information. This work was supported by grants GMl1301, CA-13330 and 2T32GM07491 from the National Institutes of Health. P.M.S. acknowledges support as an Irma T. Hirschl Career Scientist awardee. References Albin R, Silverman P (1984) Identification of the Escherichia coli K-12 cpxA locus as a single gene: construction and analysis of biologically-active cpxA gene fusions. Mol Gen Genet 197 : 272-279 Armstrong K, Fan D (1976) Essential genes in the metB-malB region of Escheriehia eoli K-12. J Bacteriol 126:48-55 Babul J (1978) Phosphor fructokinases from Escherichia eoli. J Biol Chem 253 : 4350-4355 Bachmann B (1983) Linkage map of Eseherichia coli K-12, edition 7. Microbiol Rev 47:180-230 Cohen S, Chang A, Hsu L (1972) Non-chromosomal antibiotic

resistance in bacteria: genetic transformation of Escherichia coli by R-factor DNA. Proc Natl Acad Sci USA 69:2110-2114 Conrad C, Stearns G, Prater W, Rheiner J, Johnson J (1984) Characterization of a glpK transducing phage. Mol Gen Genet 193 : 376-378 Cuozzo M, Silverman P, Minkley E, Jr (1984) Overproduction in Escherichia coli K-12 and purification of the TraJ protein encoded by the conjugative plasmid F. J Biol Chem 258: 665%6666 Eoyang L, Silverman P (1984) Purification and subunit composition of acetohydroxyacid synthase I from Escherichia eoli K-12. J Bacteriol •57:184-189 Fowler T, Taylor L, Thompson R (1983) The control region of the F-plasmid transfer operon: DNA sequence of the traJ and traY genes and characterization of the traY-Z promoter. Gene 26 : 79-89 Gaffney D, Skurray R, Willetts N (1983) Regulation of the F connjugation genes studied by hybridization and tra-lacZ fusion. J Mol Biol 168:103-122 Hansen J, Olsen R (1978) Isolation of large bacterial plasmids and characterization of the P2 incompatibility group plasmids PMG1 and PMG5. J Bacteriol 135:227-238 Humphreys G, Willshaw G, Anderson E (1975) A simple method for the preparation of large quantities of pure plasmid DNA. Biochim Biophys Acta 383:457~463 Jorgensen R, Berg D, Allet B, Reznikoff W (1979) Restriction enzyme cleavage map of Tnl0, a transposon which encodes tetracycline resistance. J Bacteriol 137:681-685 Maniatis T, Fritsch E, Sambrook J (1982) Molecular cloning: a laboratory manual. Cold Spring Harbor Press, Cold Spring Harbor, New York, pp 366-369 McDonell M, Simon M, Stndier FW (1977) Analysis of restriction fragments of T7 DNA and determination of molecular weights by electrophoresis in neutral and alkaline gels. J Mol Biol

110:11%146 McEwen J, Silverman P (1980a) Chromosomal mutations of Escherichia coli K-12 that alter the expression of conjugative plasmid function. Proc Natl Acad Sci USA 77:513-517 McEwen J, Silverman P (1980b) Genetic analysis of Escherichia coli K-12 chromosomal mutants defective in the expression of F-plasmid functions: identification of genes cpxA and cpxB. J Bacteriol 144:60-67 McEwen J, Silverman P (1980c) Mutations in genes cpxA and cpxB of Escherichia coli K-12 cause a defect in isoleucine and valine synthesis. J Bacteriol •44:68-73 McEwen J, Silverman P (1982) Mutations in genes cpxA and cpxB alter the protein composition of the Escherichia coli K-12 inner and outer membranes. J Bacteriol 144 : 1553-1559 McEwen J, Sambucetti L, Silverman P (1983) Synthesis and translocation of outer membrane proteins in cpxA cpxB mutants of Escherichia coli K-12. J Bacteriol 154:375-382 Morris J, Newman E (1980) Map location of the ssd mutation in Escherichia coli K-12. J Bacteriol 143 : 1504-1505 Newman E, Malik N, Walker C (1982) L-serine degradation in Escherichia coli K-12: directly isolated ssd mutants and their intragenic revertants. J Bacteriol 150 : 710-715 Pahel G, Bloom F, Tyler B (1979) Deletion mapping of the polAmetB region of the Escherichia coli chromosome. J Bacteriol 138:653-656 Pouyssegur J, Lagarde A (1973) Systeine de transport du 2-ceto-3desoxygluconate chez E. coli K-12: localisation d'un gene du structure et de son operateur. Mol Gen Genet 121 : 163-180 Pouyssegur J, Stoeber F (1974) Genetic control of the 2-keto-3deoxy-D-gluconate metabolism in Escherichia coli K-12: kdgR regulon. J Bacteriol 117:641-651 Prentki P, Karch F, Iida S, Meyer J (1981) The plasmid cloning vector pBR325 contains a 482 base-pairqong inverted duplication. Gene 14:289-299 Rigby P, Dieckmann M, Rhodes C, Berg P (1977) Labeling deoxyribonucleic acid to high specific activity in vitro by nick-translation. J Mol Biol 113:232251

271 Sambucetti L, Eoyang L, Silverman P (1982) Cellular control of conjugation in Escherichia coli K-12: effect of chromosomal mutations on F-plasmid gene expression. J Mol Biol 161:13-31 Shimosaka M, Fukuda Y, Murata K, Kimura A (1982) Application of hybrid plasmids carrying glycolysis genes to ATP production by Escherichia coli. J Bacteriol 152:98-103 Silverman P (1982) Gene cpxA is a new addition to the linkage map of Eseherichia coli K-12, J Bacteriol 150:425-428 Sutton A, Newman T, McEwen J, Silverman P, Freundlich M (1982) Mutations in genes cpxA and cpxB of Eseheriehia coli K-12 cause a posttranslational defect in acetohydroxyacid synthase I function in vivo. J Bacteriol 151:976--982 Thomson J, Gerstenberger P, Goldberg D, Gociar E, Silva A, Fraenkel D (1979) ColEI hybrid plasmids for Escherichia coli genes of glycolysis and the hexose monophosphate shunt. J Bacteriol 137: 502-506 Touati D (1983) Cloning and mapping of the manganese superox-

idc dismutase gene (sodA) of Escherichia coli K-12. J Bacteriol 155:1078-1087 Umbarger HE (1978) Amino acid biosynthesis and its regulation. Annu Rev Biochem 47:533-606 Vogel H, Bonner D (1956) Acetylornithinase of Escherichia eoli: partial purification and some properties. J Biol Chem 218:97-106 Willetts N (1977) The transcriptional control of fertility in F-like plasmids. J Mol Biol 112:]41-148 Womble D, Taylor D, Rownd R (1977) Method for obtaining more accurate covalently closed circular plasmid-to-chromosome ratios from bacterial lysates by dye-buoyant density centrifugation. J Bacteriol 130:148-153 Communicated by W. Arber Received July 4, 1984