Vol. 167, No. 2JOURNAL OF BACTERIOLOGY, Aug. 1986, p. 594-6030021-9193/86/080594-10$02.00/0Copyright © 1986, American Society for Microbiology

Identification and Characterization of recD, a Gene AffectingPlasmid Maintenance and Recombination in Escherichia coli

DONALD P. BIEK AND STANLEY N. COHEN*Departments of Genetics and Medicine, Stanford University School of Medicine, Stanford, California 94305

Received 12 February 1986/Accepted 15 May 1986

We isolated mutations that reduce plasmid stability in dividing cell populations and mapped these mutationsto a previously undescribed gene, recD, that affects recombination frequency and consequently the formationof plasmid concatemers. Insertions of the transposable element TnlO into recD resulted in increasedconcatemerization and loss of pSC101 and ColEl-like replicons during nonselective growth. Both concatemerformation and plasmid instability in recD mutants require a functional recA gene. Mutations in recD arerecessive to recD+ and map to a small region of the Escherichia coi chromosome located between recB andargA. Although the recD locus is distinct from loci encoding the two previously identified subunits of the RecBCenzyme, mutations in recD appear to affect the exonuclease activity of this enzyme.

The maintenance of plasmids as extrachromosomal ele-ments in bacterial cells is dependent on a number of complexprocesses. Of primary importance is the requirement forplasmid replication, which commonly involves at least somehost-encoded replication functions (see Scott [39] for areview). Replication of the pSC101 plasmid (7) requires theEscherichia coli dnaA gene product (21). Mutations in othergenes involved in chromosomal replication, including dnaB,dnaC, and dnaG (14, 22), also affect replication of thisplasmid. Stable maintenance of pSC101 in growing cellpopulations also requires a plasmid locus (named par forpartitioning) that is involved in the distribution of plasmidmolecules to daughter cells at the time of cell division (29,45). Earlier work has shown that the pSC101 par region doesnot encode a protein but instead includes partition-relatedsegments (30) that appear to be necessary for plasmids in theintracellular pool to be counted and partitioned as individualmolecules (45). To better understand the mechanism bywhich pSC101 segregates during cell division, we undertookto identify and characterize E. coli genes encoding productsinvolved in the plasmid maintenance (Pma) phenotype.We report here the isolation and characterization of mu-

tations that affect the stable maintenance of pSC101, as wellas that of certain other plasmids. We present evidence thatthese mutations, which can alter plasmid stability by increas-ing the frequency of multimer formation and are located in apreviously unidentified E. coli gene, affect the activity ofexonuclease V (ExoV).

MATERIALS AND METHODS

Strains, media, and general methodology. Relevant bacte-rial genotypes, plasmids, and phages are listed in Table 1.The media used for these studies have been previouslydescribed (31). Minimal medium was M9 or M63 supple-mented with 0.2% glucose (or another carbon source), 1 ,ugof thiamine per ml, and any required L-amino acids (SigmaChemical Co.) at 40 ,ug/ml. Solid media contained 1.5% agar(Difco Laboratories) or 0.7% agar for soft agar overlays.Lactose indicator plates contained 40 ,ug of X-Gal (5-bromo-4-chloro-3-indolyl-,-D-galactopyranoside; BoehringerMannheim) per ml in minimal medium containing 0.5%

* Corresponding author.

Casamino Acids (Difco). Antibiotics (Sigma) were used atthe following concentrations: kanamycin sulfate, 50 pLg/ml;tetracycline hydrochloride, 10 Fg/ml; chloramphenicol, 20,ug/ml; streptomycin sulfate, 150 ,ug/ml; and carbenicillin,300 ,ug/ml (used in place of ampicillin).Procedures for routine manipulation of strains were as

described by Miller (31) and Davis et al. (10). Generalizedtransductions with P1 vir phage were carried out as de-scribed by Miller (31), substituting 10 mM EGTA for sodiumcitrate. Lambda plate lysates were prepared by standardprocedures (28). Transformations of E. coli (8) were per-formed with cells made competent with CaCl2.

Isolation of TnlO insertions affecting maintenance ofpSC101. Insertions ofTnJO or mini-TnlO (mini-tet) insertionsin the E. coli chromosome were isolated by transpositionfrom defective A::TnlO transducing phages (X561, X1098[Table 1]) kindly supplied by Nancy Kleckner. Transposi-tions in strain DPB6 were isolated essentially as describedby Way et al. (46) with the exception that sodium pyrophos-phate was not used, since it appears to interfere with theX-Gal indicator. Prior to plating the X::TnlO-infected cells,unadsorbed phage was removed by three washes with Lbroth (LB) containing 20 mM sodium citrate. Platings wereon M63 medium containing tetracyline, X-Gal, glucose,Casamino Acids, and 2 ,ug of FeSO4 per ml (required foroptimal growth of this tonB mutant). Plates were incubatedat 37°C for 2 to 3 days. Colonies containing blue sectors werepicked and purified by two sequential single-colony isola-tions.

Isolation of srl::mini-kan insertions and construction of recAderivatives of strains. To facilitate construction of recAderivatives of strains carrying TnWO (Tc%, we isolated inser-tions of a mini-kan derivative of TnlO (element 9 in Way etal. [46]) in the sorbitol utilization (sri) genes located nearrecA. Transpositions from X1105 (46) in MG1655 were se-lected on MacConkey sorbitol (2%) plates containing kana-mycin. To make recA derivatives, P1 phage grown onDPB360 (srl-201: :mini-kan) was first used to transducestrains to Kmr. recA mutations were then introduced into thesrl-201 derivatives by using P1 phage grown on KL16-99(recAl) or PM191 (recA56) selecting Srl+ recombinants.About 50% of such Srl+ transductants had inherited the recAallele of the donor. Transductants carrying a recA mutation

594

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

MUTATIONS IN recS AFFECTING PLASMID MAINTENANCE

TABLE 1. Bacterial strains, phages, and plasmidsStrain, phage, Genotype phenotype Source or reference"or plasmid tAB1157 argE3 his4 leuB6 proA2 thi-l thr-l ara-14 galK2 lacYl mtl-l rpsL31 supE44 tsx-33 X- A. J. ClarkJC5519 As AB1157 but recB21 recC22 A. J. ClarkJC8679 As AB1157 but his-60 recB21 recC22 sbcA23 18; A. J. ClarkKL188 his-68 pyrD34 thi-I thyA25 trp45 galK2 malAl mtl-2 rpsL118 xyl-7 X- CGSC4211DPB353 As KL188 but recDl901::TnlO This studyPMl91 thi-l thr-l deoC deoB lacYl supE44JhuA2I X- 29AlacI5 ara A(lac proB) galE rpsL thi Valr [480 dlac with A(lacI tonB trp)S] X- 38; M. CalosDPB6 As AlacI5 but pZC1 This studyDPB28 As AlacIS but recD1901::TnJO/pZC1 This studyDPB115 As AlacIS but recDl901::TnlO This studyMG1655 A- 19; CGSC6300DPB267 X- recD1901::TnlO This studyDPB271 A- recD1903::mini-tet This study

X561 b221 c1171::TnlO c1857 Oam29 Pam8O 46; N. KlecknerX1098 b522 c1857 Pam8O nin-S mini-tet pTac transposase 46; N. Kleckner

pA Apr res (resolution site of Tn3); pBR322 replicon 3; N. CozzarellipCDK2 Apr argA +; pBR322 replicon 13; S. KushnerpCDK3 Apr thyA+ recC+ recB+ recD+; pBR325 replicon 13; S. KushnerpRDK41 Apr Tcs; dimer pBR322 replicon 11; R. KolodnerpTU4 Cmr; pACYC184 replicon 44; Laboratory collectionpZCl Kmr lacI+; pSC101 replicon This studypZC8 Cmr pTU4-210::Tn3 bla::tsr; pACYC184 replicon B. Jaurin and D. BiekpZC12 Cmr recD+; pBR325 replicon This studypZC13 Cmr recD+; pBR325 replicon This studya Coli Genetic Stock Center (CGSC) strains were obtained from B. Bachmann.

were distinguished from those that were recA+ by theirsensitivity to UV light.Assay of plasmid stability. Maintenance of plasmids in host

bacteria in the absence of selection was measured essentiallyas described by Meacock and Cohen (29), except that thezero-generation time point was made by plating cell suspen-sions of colonies picked from selective plates, and colonieson LB plates were scored for the presence of a plasmid byreplica plating to antibiotic-containing media.

Isolation and manipulation of plasmid DNA. Plasmid DNAwas isolated by alkaline lysis of bacteria (28) and bandedtwice in ethidium bromide-CsCl gradients. For isolation ofmultimeric forms, uncut plasmid DNA was subjected topreparative agarose gel electrophoresis, and electroelutedmultimers were used to transform PM191 (recA56). PlasmidDNA isolated from individual transformants served as thesource of multimers.

Restriction enzymes and DNA-modifying enzymes wereobtained from New England BioLabs or Bethesda ResearchLaboratories and used essentially as recommended by themanufacturers. Agarose gels contained 0.75 or 1% agarose inTBE (100 mM Tris-borate [pH 8.3], 2 mM EDTA) and wererun at 6 to 10 V/cm. DNA fragments separated in agarosegels were visualized after staining with ethidium bromide(0.5 mg/ml). Southern blot hybridizations (40) were per-formed as described by Davis et al. (10) by using nick-translated plasmid DNA as a probe.The pSC101 derivative pZC1 was constructed from

pDS101 (kindly provided by D. Stein), in which a 1.7-kilobase (kb) EcoRI fragment containing the lacI gene of E.coli (as well as lacOP and the 3' end of lacZ from pMC9 [32])was inserted into the EcoRI site of pSC101 (7). We replacedan HindIII-HpaI fragment containing the tetracycline resist-ance gene ofpDS101 with a 1.3-kb HindIII-SmaI fragment ofTnS containing the neo gene encoding resistance toneomycin and kanamycin.

The pZC12 and pZC13 plasmids, which contain the E. colirecD gene, were constructed by subcloning a 3.5-kb PstIfragment of pCDK3 (13) into pBR325 DNA that had beencleaved with PstI and then treated with alkaline phospha-tase. The 3.5-kb PstI fragment contains the region extendingthrough the 3' end of recB into argA (see Dykstra et al. [13]for a physical and genetic map of this region of the chromo-some). The identity of the PstI fragment was verified byrestriction digests with PstI, PvuII, and PstI + SalI, andadditionally by comparison with PstI-cleavedpCDK3::TnlOOO-59 (13), which contains a TnO000 insertionlocated between recB and argA. Plasmids pZC12 and pZC13differ only in the orientation of insertion of the 3.5-kb PstIrecD-containing fragment into pBR325.

Catenated and concatenated plasmid pA (3) species weredistinguished by the high-resolution gel electrophoresis sys-tem described by Sundin and Varshavsky (42). Catenationand decatenation of plasmids was performed with DNAgyrase (Bethesda Research Laboratories; isolated fromMicrococcus luteus) under conditions described byBenjamin et al. (3). After nicking the plasmids with DNase I(7.5 ,ug/ml) in the presence of 300 ,ug of ethidium bromide perml, the DNAs were separated by electrophoresis for 3 daysin a 30-cm, 0.8% agarose gel as described by Sundin andVarshavsky (42) at 1 V/cm. Transfer of DNA to nitrocellu-lose and hybridization with nick-translated pZC9(pBR322::neo) DNA were performed to enhance visualiza-tion of faint bands.

RESULTS

Isolation of Pma- TnlO insertion mutants. To identifynonessential chromosomal genes affecting plasmid stability(i.e., the Pma phenotype), we used the transposable elementTnlO as a mutagen. Insertions of TnJO or mini-TnlO into thechromosome were obtained after transposition of TnJO from

595VOL. 167, 1986

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

596 BIEK AND COHEN

defective X::TnlO phage. To detect mutants in which inser-tion of TnWO had inactivated a gene required for stablemaintenance of pSC101, DPB6 was used as a recipient forTnWO transpositions. This strain contained a Kmr derivativeof pSC101 (pZC1) in which the lacI gene of E. coli had beeninserted. Its chromosome carries a deletion of the lacI gene,such that all the lac repressor present is encoded by theplasmid-borne gene. The plasmid-encoded lac repressorinhibits transcription of the chromosomal lacZYA operon,resulting in colonies that are white on X-Gal indicator plates.Loss of the plasmid during colony growth leads to produc-tion of blue sectors in an otherwise white colony. For themutant search, TnlO transpositions were carried out onX-Gal plates, and colonies producing blue sectors werecollected for characterization.About 1 in 200 colonies containing a TnJO transposition

was found to contain blue sectors. When DPB6 lacking TnWOwas plated on X-Gal, spontaneously occurring blue-sector-containing colonies occurred at a frequency of about l0'.Whereas most of the Lac' cells from blue sectors observedafter TnWO mutagenesis resulted from spontaneous loss ofthe plasmid or from deletions that remove the lacI and neogenes (unpublished data), certain of the sector-containingcolonies contained TnWO insertions in bacterial genes thataffected the stability of pZC1. These were identified bytransducing TnWO, using P1 phage grown on the mutantsyielding sectored colonies, into nonmutagenized DPB6 andplating on tetracycline-containing X-Gal plates. Only strainsin which TnWO affected a gene involved in plasmid mainte-nance produced Tcr transductant colonies having blue sec-tors. This test demonstrated linkage between the TnlOelement in a mutant and the determinant affecting plasmidinstability.

In one mutant hunt involving the screening of about 50,000transposition events, 70 independent Lac' clones that werealso kanamycin sensitive and tetracycline resistant werefound. Of these, nine insertion mutants yielded instability ofpZC1 (i.e., the Pma- phenotype) when transduced intoDPB6. Characterization of the mutants obtained allowedthem to be divided into several classes. One class, whichincluded four of the nine TnWO insertion mutants, affected asingle genetic locus that we originally termed pmaA butwhich, as will be discussed, was renamed recD; an addi-tional insertion mutant in the same locus was subsequentlyisolated. These recD mutants are the subject of this study.Other classes of TnWO insertion mutants that yielded a Pma-phenotype will be described elsewhere.

Effect of recD::TnlO mutations on plasmid stability. recDmutants exhibited a moderate rate of plasmid loss as esti-mated by the number of sectors per colony grown onminimal agar plates containing X-Gal. Cells from blue sec-tors (i.e., Lac+ segregants) were Kms and lacked the pZC1plasmid (data not shown).

During nonselective growth in LB, the pSC101 derivative,pZC1, showed only limited instability in recD mutants (e.g.,recDl901::TnJO in DPB28 [Fig. 1A]). After 100 generationsof growth, about 10%o of the cells in a culture had lost theplasmid. This relatively low rate of loss was not expected,given the degree of sector formation observed when DPB28(AlacIS recDJ901::TnlO/pZC1) was plated on Casamino Ac-id-supplemented plates containing X-Gal. However, growthof DPB28 in supplemented liquid minimal media yielded ahigher frequency of plasmid-free cells than growth in liquidLB (Fig. 1A, dashed line), explaining the observed discrep-ancy between the degree of sector formation observed onminimal plates and the slow rate of plasmid loss in LB-grown

S 100

L 5_

0

z

0 1

C.)

cn [(B)10 0 7

50

1

20 40 60 80 100

NUMBER OF GENERATIONS

FIG. 1. Stability of plasmids during growth of recD and recD+strains. (A) The proportion of AlacI5-derived cells (/lacIS orDPB115) containing plasmid as a function of the number of gener-ations of growth in LB in the absence of selection. Symbols: 0,recD190/pZC1; *, recD19Ol/pBR322; A, recD1901/pSC166(ColE1::kan); *, recD1901/Fkan or Rl and recD+/pZC1, pBR322,pSC166, Fkan, or Rl. The dashed line corresponds torecD1901lpZCl grown in M63 with glucose and Casamino Acids. (B)The proportion of MG1655-derived cells [MG1655, DPB267(recDI901), DPB271 (recD1903)] containing plasmid as a function ofthe number of generations of growth in LB in the absence ofselection. Symbols are as in A; open symbols are recD1903 deriva-tives. Additionally, * denotes recD19031Fkan or Rl.

cultures. The media effects on plasmid stability observed inthe particular genetic background of DPB18 (i.e., AlacIS)were not studied further.To determine whether mutation of recD results in insta-

bility of plasmids other than pSC101 (pZC1), we introducedthe plasmid pBR322, pACYC184, pSC166 (a Kmr derivativeof ColEl), Fkan, or Rl into DPB115 (AlacIS recD1901). TheColEl-like plasmids pBR322 and pACYC184 (data notshown) were lost significantly faster than the pSC101 deriv-ative pZC1 (Fig. 1A). Plasmid pSC166 (ColEl::kan) was lostwith a rate intermediate between those of pBR322 and pZC1.The low-copy-number plasmids Fkan and Rl were stablymaintained in the recDl901 mutant.Because the strain in which the recD mutations were

isolated (the AlacIS strain DPB6) is very slow growing(possibly as a result of the deletion that removes tonB), wetransferred the recD mutations into strain MG1655 (19) bytransduction. Plasmid stability was tested in two of theresulting isolates [DPB267 (recDI901) and DPB271

J. BACTERIOL.

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

MUTATIONS IN recS AFFECTING PLASMID MAINTENANCE

mol wtoc ccC A B C

9.26-16.21 -14.17-12.14-

5.39- 10.10-

4.36 8.07-

7.05 -

6.03-

5.01

i[i 1 ccc

3.99FIG. 2. Agarose gel of plasmids isolated from pooled transformants of recD+ and recD strains. Positions of molecular weight (mol wt)

markers (not shown) are indicated on the left in kilobases for relaxed molecules (oc), which were plasmids pBR322, +X174, and pSC101, andfor supercoiled molecules (ccc), which were from a ladder of supercoiled plasmids (Bethesda Research Laboratories). Lanes A, B, and Ccontain pBR322; lanes D, E, and F contain pSC166 (ColE1::kan); and lanes G, H, and I contain pZC1. DNA was isolated from MG1655(recD+) (lanes A, D, and G); DPB267 (recD1901) (lanes B, E, and H); or DPB271 (recD1903) (lanes C, F, and I). To the right of each set ofplasmids are marks identifying the multimer state (1 = monomer, 2 = dimer, etc.) of the plasmid species observed (oc or ccc). Since pZC1plasmid multimers were not readily observable in ethidium bromide-stained gels, the gel was transferred to nitrocellulose and probed withnick-translated pZC9 (pBR322::neo), which has homology with pZC1. The autoradiogram is shown at the right side of the figure (also labeledlanes G, H, and I). Lanes J, K, and L, in which the pZC1 multimers are easier to see, represent the same DNAs shown in lanes G, H, andI run on a separate gel, transferred, and probed. (The molecular weight markers corresponding to this gel are not shown.)

(recD1903)], and similar results were obtained for each.Plasmids found to be unstable in the recDJ901::TnlO AlacISbackground were also unstable in the MG1655-derived recDstrains (Fig. 1B). However, the observed rate of plasmid lossusually was greater in the MG1655 background. We haveobserved strain-dependent differences in the degree of plas-mid instability conferred by recD mutations in other geneticbackgrounds as well.

Formation of plasmid multimers in recD mutants. To inves-tigate the molecular basis for the plasmid instability ob-served in recD mutants, we analyzed covalently closedcircular plasmid DNA isolated from recD+ and recD::TnJO(recDI901 recD1903) strains. Agarose gel electrophoresis ofDNA representing several different plasmids isolated frompooled transformants is shown in Fig. 2. The plasmid DNAisolated from the recD strains showed a significantly greaterfraction of multimeric species, and the presence of multi-mers was especially pronounced for ColEl-like plasmidswhich were very unstable in recD strains (compare lane Awith lanes B and C and compare lane D with lanes E and F).Multimers of the pSC101 derivative (pZC1 [Fig. 2, lanes G toLI) were also obtained in recD strains but less frequentlythan multimers of ColEl-like plasmids. Upon cleavage witha variety of restriction enzymes, plasmid DNA fromrecDl901 and recD+ strains appeared identical (data notshown), consistent with the interpretation that the higher-molecular-weight DNA species did not result from sequencerearrangement.To determine whether the DNA multimers formed in recD

mutants were catenated or concatenated molecules, we usedhigh-resolution gel electrophoresis as described by Sundinand Varshavsky (42) to analyze DNase I-treated plasmid

DNA isolated from recDJ901::TnlO or wild-type strains(Fig. 3). recD-derived multimers of plasmid pA, a derivativeof pBR322 (3), comigrated with concatemeric pA species(compare lane C with concatenated markers on the right) andnot with catenanes of pA generated with M. luteus DNAgyrase (lane B). Under suitable conditions, DNA gyrase isable to decatenate DNA molecules. Treatment of themultimers under conditions favoring decatenation did notchange the pattern of recD-promoted plasmid multimers(compare lanes C and D) but did convert gyrase-generatedcatenanes to monomers (compare lanes A and B). Similarresults were obtained in an experiment conducted in collab-oration with H. Benjamin and N. Cozzarelli in which E. coliDNA gyrase was used (data not shown). Together, theseresults indicate that the multimeric plasmid species from therecDI901 mutant consist of concatenated molecules.

recA dependence of multimer formation. To determinewhether the multimerization observed in recD strains re-quires the recA gene product, we introduced pZC1 orpBR322 into isogenic recA56 and recA+ derivatives of therecD+ and recDJ901::TnJO strains and plasmid stability wasmeasured. The concurrent presence of the recA56 mutationeliminated the plasmid instability associated withrecD901: :TnJO (Fig. 4A). Moreover, plasmid DNA isolatedfrom recA recD double mutants contained few concatemericmolecules (Fig. 5; unpublished results), suggesting thatconcatemer formation in recD strains requires RecA-dependent homologous recombination.Can recD mutants resolve plasmid multimers? To test

whether recD mutants lack the ability to resolve plasmidmultimers to monomeric forms, we transformed isogenicrecD+ and recDI901 or recDJ903 strains with the dimeric

r) F F a 14 I

-4 ccc

2 oc

- 3 ccc

-2 ccc

-1 oc

J K L

i ,.*.*

.4

-2 oc3 ccc

-2 ccc1 oc

-1 ccc

-2 oc3 ccc

- 2 ccc1 oc

- 1 ccc

VOL. 167, 1986 597

t.

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

598 BIEK AND COHEN

pBR322 derivative pRDK41 (11) and examined covalentlyclosed circular plasmid DNA isolated from these transform-ants by agarose gel electrophoresis. A small amount ofconversion of dimers to monomers occurred in the recD+strain (Fig. 5, lane B) and at least as much dimer-to-monomer conversion occurred in recD1901 'or recDI903strains (lanes C and D). In both instances the appearence ofmonomer species depended on a functional recA gene '(com-pare lanes B, C, and D with E, F, and G in Fig. 5), consistentwith the observed recA dependence of recD-promoted plas-mid instability as described above. Additionally, this recAcontrol demonstrated that the plasmid DNA monomersobserved in the recD mutants had not resulted from trans-formation by rare monomeric plasmids present in the trans-forming DNA.

a

A-j

a-z-

z00CA,-J-Jw

U-0

zwLLI

uL

A B C D E mol wtoc

-16.21

-14.17

-12.144 oc-

-10.10

-4 oc

-3 oc

3 oc-- 8.07

- 7.05

4._

2 oc- _

- 2 oc- 6.02

WV 5.01

_ - 3.99

1 oc- -1 oc

2.97

FIG. 3. Plasmid multimers isolated from a recD mutant are

concatemers. DNA was nicked and electrophoresed in a 0.8%agarose gel. To enhance identification ofDNA bands, the DNA wastransferred to nitrocellulose and then probed with nick-translatedpZC9 (pBR322::neo), which contains homology to the plasmid pAused in this study. An autoradiogram is shown. Lane B contains pAcatenanes produced by DNA gyrase, and lane A contains the samepreparation of pA catenanes as in lane B after decatenation bygyrase. (No catenated pA was seen in lane A even after overexpo-sure of the film.) The labels on the left side of the autoradiogramidentify the positions of catenanes of pA (lane B); catenanes largerthan'dimers are not readily observable in this exposure. oc, Relaxedmolecules. Lane C contains pA isolated from the recDI910 strainDPB267 (pooled transformants), and lane D is this DNA aftertreatment with gyrase under decatenation conditions. Molecularweight (mol wt) markers (lane E) were made by nicking a ladder ofsupercoiled plasmids (derived from pBR322; Bethesda ResearchLaboratories). The sizes of these markers in kilobases are indicated.The labels on the far right side of the figure correspond to thepositions of pA concatemers-monomer (1 oc), dimer (2'oc), andtrimer (3 oc); these DNAs had been isolated from recA strainscontaining individual species.

20 40 60 80 100

NUMBER OF GENERATIONS

FIG. 4. (A) Effect of recA mutation on the stability of plasmids ina recD mutant. Plasmid pBR322 or pZC1 was transformed- intoisogenic MG1655-derived strains, and the proportion of cells con-taining plasmid after growth in the absence of selection was deter-mined. Symbols: *, recDI901 recA+ (DPB373)/pZC1; *, recDl901recA+ (DPB373)/pBR322; *, recDI901 recA56 (DPB374)/pZC1 orpBR322; *, recD+ recA+ (DPB371)/pZC1 or pBR322; *, recD+recA56 (DPB372)/pZC1 or pBR322. (B) Effect of recBC sbcAmutations on plasmid stability. Plasmid pBR322 or pZC1 wastransformed into strain AB1157 and its derivatives JC5519 andJC8679, and plasmid stability was measured. Symbols: 0, recB+recC22 sbcA23/pZC1; *, recB21 recC22 sbcA23/pBR322; *, recB21recC+ sbcA+/pZC1 or pBR322; *, recB21 recC22 sbcA+/pZCl orpBR322.

Plasmid recombination in recD mutants. Since recD mu-tants contain a mechanism for converting plasmid multimersto monomers, we determined whether the concatemer ex-cess observed in recD strains was associated with an in-creased rate of plasmid recombination. In a test described byDoherty et al. (11) (and similar to the test used by Laban andCohen' [26]), a dimeric pBR322 derivative (pRDK41) con-taining two inactive copies of the tet gene (each copy havingan insertion of an XhoI linker at a different site) was used tomonitor plasmid recombination events. In this system, re-combination between two inactive tet genes on either thesame or different plasmids can generate an intact tet gene,conferring tetracycline resistance on the host cell. This testcannot be used directly for testing mutants that contain TnlOinsertions, since the transposable element itself encodes Tcr.We' circumvented this problem by allowing recombination tooccur in isogenic recD1901::TnlO and recD+ strains (intro-

J. BACTERIOL.

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

MUTATIONS IN recS AFFECTING PLASMID MAINTENANCE

mol wtoc ccC

9.26

5.399.26

4.36-

A R r n F F f.

ccc

2 oc

ccc

2 ccc

l

5.39 -

4.36 -1 ccc

FIG. 5. Conversion of dimer concatemers ^ monomers. Posi-tions of molecular weight (mol wt) markers are indicated on the left(relaxed circles = oc; supercoiled circles = ccc); sizes are inkilobases for pBR322, 4X174, and pSC101. Lane A containspRDK41 plasmid (isolated from recA strain RK1233) used to trans-form the isogenic MG1655-derived strains used in this study.pRKD41 is a dimer of pBR322 (11). Plasmid DNA was isolated frompooled transformants; however, plasmid DNA from individualtransformants gave the same results (data not shown). Lane Bcontains plasmid isolated from MG1655 (recD+); lane C containsplasmid from DPB267 (recDl901); and lane D contains plasmid fromDPB271 (recD1903). Strains (DPB302, DPB303, and DPB307) usedin lanes E, F, and G are recA::cat derivatives of those used in lanesB, C, and D, respectively. Labels on the right side correspond to thepositions of monomer or concatemer pBR322 molecules (oc or ccc,respectiyely). Resolution of pRDK41 converts it from the 2 ccc sizeto 1 ccc.

ducing the plasmid selecting for resistance to carbenicillin[Cbl) and then scoring for Tcr recombinants by isolatingplasmid DNA from individual transformants and retrans-forming an recAS6 (Tcs) recipient selecting for Cbl or CblTcr. The results (Table 2) indicate that Tcr recombinantswere detected at a sixfold higher frequency in plasmid DNAisolated from DPB373 (recDI901: :TnlO) compared withDPB371 (recD+), and the recD-mediated increase required a

functional recA gene. In the experiment shown, plasmidrecombination in the recDl901::TnlO recAS6 double mutant(DPB374) appeared to be higher than in the recD+ recAS6strain (DPB372), but the difference in absolute number ofrecombinants was small. In an experiment with anothergenetic background, the relative recombination levels be-tween recD1901 recA and recD+ recA differed by onlytwofold.To test whether recombination of chromosomal genes was

also increased in recD mutants, we performed two othertests. In one, we measured the frequency of replacement ofchromosomal markers (his or trp) by recombination occur-

ring during P1 transduction; in the second, we used thereplacement of chromosomal markers (his or trp) duringconjugational crosses as a measurement of recombinationalevents. The results of these tests showed that recombination

frequencies were elevated two- to threefold in DPB353(recD1901::TnJO) compared with an isogenic recD+ strain(KL188).

Stability of plasmids in other hyperrecombinogenic mu-tants. Work by others (11, 17, 26, 41)'has shown that plasmidrecombination and the formation of plasmid concatemers areelevated in recBC sbcA mutants, in which the recE pathwayof recombination is active (18). Using a series of isogenicderivatives of AB1157 (provided by A. J. Clark) that differedin their recBC sbcA genotypes, we introduced plasmidspBR322 and pZC1 and then measured plasmid stability. Bothplasmids were stably inherited over at least 100 generationsof nonselective growth in AB1157 (recB,+C+ sbcA+) andJC5519 (recB21 recC22 sbcA+) (Fig. 4B), a result consistentwith the findings of Bassett and Kushner (2). However, inagreement with the studies of Summers and Sherratt (41), wefound that these plasmids were not stably-maintained in therecB21 recC22 sbcA23 strain JC8679. Instability of pZC1was relatively slight; however, rapid loss of pBR322 wasobserved, paralleling the effects we obseped with our recDstrains.Plasmid DNA from pooled transformants of these strains

was examined for the presence of concatemeric species.Multimers of pBR322 were greatly increased in the recB21recC22 sbcA23 strain (Fig. 6, lane C) relative to either of thesbcA+ strains (lanes A and B). Consistent with the compar-atively slight instability observed for pZC1 in JC8679 (Fig.4B), pZC1 DNA isolated from this strain contained a com-paratively small percentage of multimers (data not shown).

Ability of the Tn3 site-specific recombination system tostabilize plasmids in recD strains. The transposable elementTn3 is known to encode a site-specific recombinationalsystem capable of resolving Tn3 cointegrate plasmid speciesinto monomers (see Heffron [23] for a review). To determinewhether accumulation of multimers in recD mutants couldbe reversed by the alternative resolution system encoded byTn3 and whether such monomerization would accomplishthe stable maintenance of plasmids, we performed the fol-lowing experiment. Concatemeric dimer pA DNA was iso-lated from DPB267 (recD1901::TnlO). Plasmid pA (3) is aderivative of pBR322 containing the resolution site (res) ofTn3. The dimeric pA DNA was transformed into isogenicrecD+ and recDI901: :TnlO strains that contained eitherpTU4 (44) or pZC8, a PTU4 derivative that carries a func-tional, resolvase-producing Tn3 element (i.e., pTU4::Tn3blas). We then examined both plasmid stability and conver-sion of pA dimers to monomers. The presence of a comple-menting Tn3 element located on pZC8 completely stabilizedpA dimers transformed into DPB395 (recD1901: :TnlOIpZC8) (less than 1% loss after 80 generations of nonselectivegrowth). This effect depended on the presence of Tn3, as no

TABLE 2. Effect of mutations in recD and recA on frequency ofplasmid recombination

Mean frequency Relative plasmidStrain Markers of Tcr pRDK41 recombination

recombinantsa frequency

DPB371 recD+ recA+ 2.6 x 1O-3 1.0DPB372 recD+ recA56 <3.5 x 10-5 <0.01DPB373 recDJ901 recA+ 1.45 x 10-2 5.6DPB374 recDI901 recA56 2.8 x 10-4 0.11

a Recombination frequencies represent means from 10 independent deter-minations in which plasmid DNAs from individual pRDK41 carbenicillin-resistant transformants of the indicated isogenic strains were used to trans-form PM191 (recA56). The fraction of carbenicillin PM191 transformants thatare Tcr is indicated.

599VOL. 167, 1986

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

600 BIEK AND COHEN

mol wt

oc ccc A B C

9.2616.21-14.17-12.14-

5.39- 10.10-

4.36- 8.07-

7.05-

6.03-

5.01

1 4 CCC2 oC

3 cCC

i2 CC>1 oc

3.99 -1 ccc

FIG. 6. Effect of mutations in recBC sbcA on formation ofplasmid pBR322 concatemers. Positions of molecular weight (molwt) markers (oc [relaxed circles] and ccc [supercoiled circles]) are

indicated in kilobases and correspond to those listed in the legend toFig. 2. Lane A contains plasmid pBR322 from pooled transformantsof AB1157; lane B contains pBR322 from JC5519 (recB21 recC22);and lane C contains pBR322 from JC8679 (recB21 recC22 sbcA23).The labels on the right side of the figure identify the multimer state(1 = monomer, 2 = dimer, etc.) of the plasmid species observed (ocor ccc).

stabilization was observed in its absence [97% loss of pAafter 80 generations in DPB399 (recD1901::TnlO/pTU4)].(We found that dimer pA was stably inherited in recD+ andrecDl901::TnlO recAl strains containing either pTU4 orpZC8.) Plasmid DNA isolated from these transformants was

examined (Fig. 7) and found to contain pA monomer speciesin the Tn3-containing recipients but only dimers in thestrains lacking Tn3. The recD1901::TnlO recA+/pTU4 strain(DPB399), which lacks Tn3, contained larger multimers inaddition to dimers, indicating that Tn3 resolvase can act intrans on recD-promoted multimers containing the res site ofTn3 to both resolve and stabilize the plasmid in recDJ901strains.Map location of recD. The location of the recDl901::TnJO

insertion was determined to be near 61 min on the E. colichromosome by Hfr mapping experiments (data not shown).Phage P1 cotransduction studies with markers in this regionestablished a very close linkage to argA. All five recD: :TnlOinsertions showed similiar linkage to argA52 (90 to 98%cotransduction) and lysA22 (29 to 40% cotransduction).Located on the E. coli chromosome in this region are therecBC genes (15), which encode subunits of ExoV (seeTelander-Muskavitch and Linn [43] for a review). WhereasTnlO insertion mutations in recD resulted in altered recom-bination levels (Table 2), the recD strains we isolated did nothave the properties of recB or recC null mutaitions (i.e.,abnormal sensitivity to UV light or a propensity for segre-gating inviable cells [4]).The recBC region of the chromosome has been well

characterized both genetically and physically (13, 37) and isdiagrammed in Fig. 8. By Southern blot analysis (40) usingcloned DNA from this region as a hybridization probe ofrestriction enzyme-digested chromosomal DNA from recDmutants or the wild type, we located the sites of the TnJOinsertions that yielded the Pma- (RecD-) phenotype. Theprobe used was plasmid pCDK2 (13), which included a 3-kbSall insert containing the argA gene and a portion of theregion between argA and recB. Digests of chromosomalDNA with PstI allowed localization of the TnJO insertions toa 3.5-kb PstI fragment spanning the region from the 3' end ofthe recB gene to the beginning of argA. Digests with Sallpositioned the insert in DPB267 (recDl901::TnlO) to theargA-proximal SalI-PstI fragment, whereas the other fourrecD::mini-tet insertions must have been located on the

A B C D E F G H I J K l M N O P Q R S T U

2 oc-

2 ccc-1 oc -

1 ccc-

-chr

-1 oc

-1 ccc

FIG. 7. Resolution of pA dimers to monomers in recD strains in the presence of Tn3. Dimeric pA DNA was isolated from DPB267(recDI901). On the left are indicated the positions of concatemeric pA species. oc, Relaxed circles; ccc, supercoiled circles. Lanes: A,monomer pA; B, dimer pA; C, pA from pooled DPB267 (recDI901) transformants; D, pZC8 (pTU4::Tn3b1as). (For the following, DNAs fromtwo transformants of each strain were isolated by a rapid NaOH-sodium dodecyl sulfate lysis protocol [28]; the DNAs were not CsCl banded.)Lanes: E to L, dimer pA transformed into MG1655-derived strains containing pZC8 (pTU4::Tn3blas): E and F, recD+ recA+ (DPB393); Gand H, recD+ recAl (DPB394); I and J, recDI901 recA+ (DPB395); K and L, recDI901 recAl (DPB396); M, pTU4; N to U, dimer pAtransformed into strains containing pTU4. Lanes: N and 0, recD+ recA+ (DPB397); P and Q, recD+ recAl (DPB398); R and S, recD1901recA+ (DPB399); T and U, recDI901 recAl (DPB400). On the right are indicated the positions of monomer pTU4 species (oc and ccc) andchromosomal DNA (chr).

J. BACTERIOL.

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

MUTATIONS IN recS AFFECTING PLASMID MAINTENANCE

Q c0kbQ

1 kb 1W-

t

I I I IC_Z IC ..

_ _ _ _

p22 thyA recC ptr recB argA

STABILIZATIONP50 OF pSC101

I- IpCDK2 -

pCDK3 +

I-~--- pZC12,13 +

FIG. 8. Genetic map of the region containing recD. The upperline shows some of the restriction sites in this region; beneath it areshown the positions of the genes known to be encoded. This figureis taken from Fig. 1 of Dykstra et al. (13). Above the physical mapare indicated the restriction endonuclease-generated fragments towhich we located the various recD::TnlO insertions. The positionsof TnlO insertions recD1902-1904 vis-a-vis each other are notknown. The lower portion of the figure indicates the extent of thisregion carried by various plasmids used in recD complementationtests and their abilities to allow stable maintenance of plasmid pZC1in recDI901 mutant DPB28 (i.e., to complement recDl90).

PstI-SalI fragment proximal to recB (Fig. 8). The insertionswere not mapped more precisely within this interval; how-ever, since none of the recD mutants we studied hadphenotypes associated with recB or argA insertion muta-tions, we believe that all five of the mutations were locatedin the intercistronic region between recB and argA. Nogenes have previously been ascribed to this region, which isestimated from the map of Dykstra et al. (13) to have a

maximum size of 1.5 kb.Complementation studies. In conjunction with our mapping

studies, we performed complementation tests for one recDmutation (recD1901: :TnJO). Plasmid pCDK2 (13), whichcontains a 3.0-kb Sall fragment that includes the site onwhich recD1901: :TnlO is located, did not complementrecD1901: :TnlO for stabilization of pZC1 maintenance.However, plasmid pCDK3, which carries an 18-kb BamHIfragment that partially overlapped the Sall fragment ofpCDK2 and contained the entire thyA-to-argA segment,allowed stable maintenance of pZC1 in recDJ901::TnJOstrains (i.e., no Lac' sector-containing colonies wereproduced). This finding localized rather precisely the siteof the recD gene and further indicated that mutationrecD1901::TnJO is recessive to the wild-type gene. To deter-mine whether a gene separate from recB is located in theregion in which we had localized the recD mutations, wesubcloned the 3.5-kb PstI fragment on which all the recDTnJO insertions mapped into the PstI site of pBR325. Thepresence of this fragment in either orientation within the blagene of pBR325 (yielding plasmid pZC12 or pZC13; seeMaterials and Methods) allowed stable maintenance of thecoexisting plasmid pZC1 in recD1901: :TnJO strains. The PstIfragment contains only the terminal portion of the recB gene

(not including the promoter for recB); moreover, neither ofthe plasmids carrying the fragment complemented eitherrecB21 or argA52.

Evidence that recD mutants are defective in ExoV. Notwith-standing evidence that the gene identified as recD is distinctfrom the recB and recC genes, our finding that recD islocated very near these known recombination genes and thatrecD mutants show altered recombination frequencies raisedthe possibility that a functional relationship exists betweenrecD and the recBC genes or their product, ExoV. For this

reason, we performed several tests of ExoV function in recDmutants. One of these utilized phage T4 gene 2 mutants,which are defective in a gene product that ordinarily protectsthe infected T4 ends from digestion by ExoV and thus fail tomake plaques on recBC+ strains (35). T42- mutants canform plaques on recBC strains which lack the exonucleaseactivity of RecBC (35). Additionally, bacteriophage lambdamutants defective in red-mediated recombination and lack-ing the gam product (which functions to block RecBCexonuclease activity [25]) make tiny plaques on recBC+strains because conversion to rolling-circle replication doesnot take place (16). On recA recBC+ strains, X red- gam-phage fail to make plaques; since packageable dimer mole-cules are not formed and rolling-circle replication is blocked,production of lambda concatemers for packaging is pre-vented. In our experiments, all five recD::TnJO mutants(recD1901-1905) showed the same efficiency of plating ofphage T42- as an isogenic recD+ strain, suggesting thatExoV was not active. Moreover, X red270 gam210 phage(XNK620) formed plaques on all five recD mutants whetheror not a functional recA gene was present and with the sameefficiency and appearance as on an isogenic recD+ strain.

DISCUSSION

We have described the isolation and characterization of aclass of E. coli mutations affecting the maintenance ofpSC101 and certain other plasmids in dividing populations ofcells. The basis for the plasmid maintenance defect imposedby recD mutations appears to be an increased rate offormation of plasmid concatemers rather than a defect inresolution; this view is supported by evidence that multimersintroduced into recD strains can be resolved as well as inrecD+ strains and that plasmid recombination is increased inrecD mutants. Earlier work has indicated that concatemerformation can also result from a defect in or loss of aplasmid-encoded resolution site, leading to failure to stablymaintain some plasmids, including ColEl (41), P1 (1), andClo DF13 (20). We speculate that either plasmid recombina-tion in recD strains is primarily interplasmidic rather thanintraplasmidic or, alternatively, that forward and backwardinterconversion of monomers and multimers occurs at sim-ilar frequencies but that, once formed, plasmid concatemersare quickly lost. Differences between intraplasmidic andinterplasmidic recombination frequencies have been re-ported for recBC sbcA mutants (26).We do not know why some plasmids are more unstable

than others in recD strains. Plasmids such as F, Ri, P1, andpSC101 encode partitioning systems that ensure the distri-bution of plasmid copies to each daughter cell at the time ofdivision (la, 29, 33, 34), mechanisms that accomplish selec-tive killing of cells that do not inherit the plasmid (17a, 24),or both. These systems may in part counteract the effects ofmultimerization resulting from mutations in recD. Addition-ally, the most stably maintained plasmids (i.e., the largeplasmids F and Rl) may contain site-specific recombinationmechanisms for resolving concatemers in recD mutants; theF plasmid carries an insertion of -yb (9), a Tn3-like elementthat is able to catalyze recombination between directlyrepeated -yb elements (23). Plasmid ColEl, which depends onmonomerization to increase its copy number to a level atwhich it can be maintained stably in cells (41), encodes a site(cer) involved in resolving multimers (41) and thus may havethe capacity to partially reverse the effect of mutations inrecD on plasmid multimerization. Such an effect mightaccount for observations indicating that ColEl itself is less

601VOL. 167, 1986

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

602 BIEK AND COHEN

unstable than ColEl-related plasmids pBR322 andpACYC184, which lack a cer region.The degree of instability might also be affected by the

mechanism by which the plasmid multimers are formed. IfrecD-promoted multimers are produced as a result of homol-ogous recombination between plasmid copies in the cell, thefrequency with which such recombination events take placemay be influenced by the number and size of plasmidmolecules available in the cell for recombination. Peterson etal. (36) have shown that the frequency of formation ofplasmid cointegrates between NR1 and a compatible plasmidcontaining fragments of NR1 is dependent on the length ofhomology. However, we have not excluded the possibilitythat the observed concatemers are formed as a replication-rather than recombination-product in recD mutants. Thispossibility has been suggested by recent experiments of A.Cohen and A. J. Clark (6a) in which greater-than-unit-length,linear plasmid DNAs were observed in recBC strains. It hasbeen postulated that these molecules result from rolling-circle plasmid replication that occurs in the absence ofExoV.TnJO insertions in recD have been localized to a small

region of the chromosome, between recB and argA. Resultsof complementation tests, mapping, and comparison ofmutant phenotypes indicate that the recD gene is separatefrom recB. Furthermore, expression of recD occurs even inthe absence of recB transcription. Chaudhury and Smith (5)have described a class of mutants called (recBCt) thatexhibit many of the phenotypes we found for recD mutants.The recBCt mutations abolish the exonuclease activity ofRecBC without yielding the other phenotypes associatedwith null mutations in recBC (i.e., decreased conjugationalrecombination proficiency, sensitivity to DNA-damagingagents, and segregation of inviable cells [4, 6]). We testedour recD mutants by several in vivo tests of ExoV activity,including plating of X red- gam- phage and T42- phage. Bythese criteria, recD mutants lack ExoV activity.During the course of our studies on chromosomal muta-

tions mapping near recB that result in the Pma- phenotype,we became aware of studies in progress by S. Amundsen, A.Taylor, A. M. Chaudhury, and G. R. Smith (Proc. Natl.Acad. Sci. USA, in press) which showed that some of therecBt mutations (5) actually affect a gene located betweenrecB and argA. Strains with mutations in this gene loseRecBC exonuclease activity while retaining other RecBCactivities (5). Amundsen et al. (in press) have demonstratedthat this gene encodes a third subunit of the RecBC enzymethat is required for the exonuclease activity of ExoV. Thepossibility of a third subunit had been suggested previouslyby the in vitro reconstitution experiments of Lieberman andOishi (27) and by studies of purified ExoV by Dykstra et al.(12). Our mutations affecting plasmid maintenance and themutations described by Amundsen et al. (in press) almostcertainly affect the same gene. To indicate the role this genehas in encoding a subunit of the recombination enzymeExoV, they have suggested that it be called recD, a choicewith which we concur. ExoV thus can alternatively bereferred to as RecBCD enzyme.The reason for plasmid instability in only some classes of

recBCD mutants (recD and possibly recCt) is not under-stood. The finding that null mutations in recB or recC do notyield the Pma- phenotype is consistent with our failure tofind recBC: :TnJO insertion mutations among the other Pma-mutants we isolated. This suggests that two conditions mustbe met for plasmids to be made unstable by this mechanism.First, the RecBCD exonuclease must be inactive and, in

addition, another activity of RecBCD (perhaps the helicase)must concurrently remain active (see Telander-Muskavitchand Linn [43] for a review of the activities associated withExoV). A possibility consistent with these findings is thatintermediate structures in plasmid recombination, producedas a result of unwinding by RecBCD and normally degradedby ExoV, are longer lived in recD mutants. The accumula-tion of such putative recombination intermediates mightexplain the higher recombination frequencies and conse-quently increased rate of concatemer formation observed inrecD strains.

ACKNOWLEDGMENTS

We thank Barbara Bachmann, Michele Calos, John Clark, NancyKleckner, Richard Kolodner, Sidney Kushner, and Gerald Smith forgenerously providing strains, helpful discussions, or both. HowardBenjamin and Nick Cozzarelli helped with experiments to distin-guish recD concatemers from catenanes and also contributed stim-ulating discussions during the course of this work. We especiallythank Susan Amundsen, Andrew Taylor, and Gerry Smith forcommunicating prior to publication their results concerning therelationship of recD and ExoV.

This work was supported by fellowship DRG-704 of the DamonRunyon-Walter Winchell Cancer Fund (D.P.B.), and Public HealthService grant GM 26355 from the National Institutes of Health(S.M.C.).

LITERATURE CITEDla.Austin, S., and A. Abeles. 1983. Partition of unit-copy

miniplasmids to daughter cells. I. P1 and F miniplasmids containdiscrete, interchangeable sequences sufficient to promoteequipartition. J. Mol. Biol. 169:353-372.

1. Austin, S., M. Ziese, and N. Sternberg. 1981. A novel role forsite-specific recombination in maintenance of bacterialreplicons. Cell 25:729-736.

2. Bassett, C. L., and S. R. Kushner. 1984. Exonucleases I, III, andV are required for stability of ColEl-related plasmids in Esch-erichia coli. J. Bacteriol. 157:661-664.

3. Benjamin, H. W., M. M. Matzuk, M. A. Krasnow, and N. R.Cozzarelli. 1985. Recombination site selection by Tn3 resolvase:topological tests of a tracking mechanism. Cell 40:147-158.

4. Capaldo-Kimball, F., and S. D. Barbour. 1971. Involvement ofrecombination genes in growth and viability of Escherichia coliK-12. J. Bacteriol. 106:204-212.

5. Chaudhury, A. M., and G. R. Smith. 1984. A new class ofEscherichia coli recBC mutants: implications for the role ofRecBC enzyme in homologous recombination. Proc. Natl.Acad. Sci. USA 81:7850-7854.

6. Clark, A. J. 1973. Recombination deficient mutants of E. coliand other bacteria. Annu. Rev. Genet. 7:67-86.

6a.Cohen, A., and A. J. Clark. 1986. Synthesis of linear plasmidmultimers in Escherichia coli K-12. J. Bacteriol. 167:327-335.

7. Cohen, S. N., and A. C. Y. Chang. 1977. Revised interpretationof the origin of the pSC101 plasmid. J. Bacteriol. 132:734-737.

8. Cohen, S. N., A. C. Y. Chang, and L. Hsu. 1972. Non-chromosomal antibiotic resistance in bacteria: genetic transfor-mation of Escherichia coli by R-factor DNA. Proc. Natl. Acad.Sci. USA 69:2110-2114.

9. Davidson, N., R. C. Deonier, S. Hu, and E. Ohtsubo. 1975.Electron microscope heteroduplex studies of sequence relationsamong plasmids of Escherichia coli. X. Deoxyribonucleic acidsequence organization of F and F-primes, and the sequencesinvolved in Hfr formation, p. 56-65. In D. Schlessinger (ed.),Microbiology-1974. American Society for Microbiology,Washington, D.C.

10. Davis, R. W., D. Botstein, and J. R. Roth. 1980. Advancedbacterial genetics. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.

11. Doherty, M. J., P. T. Morrison, and R. Kolodner. 1983. Geneticrecombination of bacterial plasmid DNA: physical and geneticanalysis of the products of plasmid recombination in Esche-

J. BACTERIOL.

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

MUTATIONS IN recS AFFECTING PLASMID MAINTENANCE

richia coli. J. Mol. Biol. 167:539-560.

12. Dykstra, C. C., K. M. Palas, and S. R. Kushner. 1984. Purifica-tion and characterization of exonuclease V from Escherichiacoli K-12. Cold Spring Harbor Symp. Quant. Biol. 49:463-467.

13. Dykstra, C. C., D. Prasher, and S. R. Kushner. 1984. Physicaland biochemical analysis of the cloned recB and recC genes ofEscherichia coli K-12. J. Bacteriol. 157:21-27.

14. Ely, S., and A. Wright. 1985. Maintenance of plasmid pSC101 inEscherichia coli requires the host primase. J. Bacteriol.164:484-486.

15. Emmerson, P. T., and P. Howard-Flanders. 1967. Cotransduc-tion with thy of a gene required for genetic recombination inEscherichia coli. J. Bacteriol. 93:1729-1731.

16. Enquist, L., and A. Skalka. 1973. Replication of bacteriophagelambda DNA dependent on the function of host and viral genes.I. Interaction of red, gam, and rec. J. Mol. Biol. 75:185-212.

17. Fishel, R. A., A. A. James, and R. Kolodner. 1981. recA-independent general genetic recombination of plasmids. Nature(London) 294:184-186.

17a.Gerdes, K., P. B. Rasmussen, and S. Molin. 1986. Unique type ofplasmid maintenance function: postsegregational killing of plas-mid-free cells. Proc. Natl. Acad. Sci. USA 83:3116-3120.

18. Gillen, J. R., D. K. Willis, and A. J. Chlrk. 1981. Geneticanalysis of the RecE pathway of genetic recombination inEscherichia coli K-12. J. Bacteriol. 145:521-532.

19. Guyer, M. S., R. R. Reed, J. A. Steitz, and K. B. Low. 1981.Identification of a sex-factor-affinity site in E. coli as -y. ColdSpring Harbor Symp. Quant. Biol. 45:135-139.

20. Hakkaart, M. J. J., P. J. M. van den Elzen, E. Veltkamp, andH. J. J. NUkamp. 1984. Maintenance of multicopy plasmid CloDF13 in E. coli cells: evidence for site-specific recombination atparB. Cell 36:203-209.

21. Hasunuma, K., and M. Sekiguchi. 1977. Replication of plasmidpSC101 in Escherichia coli K12: requirement for dnaA function.Mol. Gen. Genet. 154:225-230.

22. Hanuma, K., and M. Sekiguchi. 1979. Effect of dna mutationson the replication of plasmid pSC101 in Escherichia coli K-12. J.Bacteriol. 137:1095-1099.

23. Heffron, F. 1983. Tn3 and its relatives, p. 223-260. In J. A.Shapiro (ed.), Mobile genetic elements. Academic Press, Inc.,New York.

24. Jaffe, A., T. Ogura, and S. Hiraga. 1985. Effects of the ccdfunction of the F plasmid on bacterial growth. J. Bacteriol.163:841-849.

25. Karu, A. E., Y. Sakaki, H. Echols, and S. Linn. 1975. The Y

protein specified by bacteriophage X. Structure and inhibitoryactivity for the recBC enzyme of Escherichia coli. J. Biol.Chem. 250:7377-7387.

26. Laban, A., and A. Cohen. 1981. Interplasmidic and intraplas-midic recombination in Escherichia coli K-12. Mol. Gen. Genet.184:200-207.

27. Lieberman, R. P., and M. Oishi. 1974. The recBC deoxyribonu-clease of Escherichia coli: isolation and characterization of thesubunit proteins and reconstitution of the enzyme. Proc. Natl.Acad. Sci. USA 71:4816-4820.

28. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

29. Meacock, P. A., and S. N. Cohen. 1980. Partitioning of bacterialplasmids during cell division: a cis-acting locus that accom-plishes stable plasmid inheritance. Cell 20:529-542.

30. Miller, C. A., W. T. Tucker, P. A. Meacock, P. Gustafsson, andS. N. Cohen. 1983. Nucleotide sequence of the partition locus ofEscherichia coli plasmid pSC101. Gene 24:309-315.

31. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

32. Miller, J. H., J. S. Lebkowski, K. S. Greisen, and M. P. Calos.1984. Specificity of mutations induced in transfected DNA bymammalian cells. EMBO J. 3:3317-3321.

33. Nordstrom, K., S. Molin, and H. Aagard-Hansen. 1980. Parti-tioning of plasmid Rl in Escherichia coli I. Kinetics of loss ofplasmid derivatives deleted of the par region. Plasmid4:215-227.

34. Ogura, T., and S. Hiraga. 1983. Partition mechanism of Fplasmid: two plasmid gene-encoded products and a cis-actingregion are involved in partition. Cell 32:351-360.

35. Oliver, D. B., and E. B. Goldberg. 1977. Protection of parentalT4 DNA from a restriction exonuclease by the product of gene2. J. Mol. Biol. 116:877-881.

36. Peterson, B. C., H. Hashimoto, and R. H. Rownd. 1982.Cointegrate formation between homologous plasmids in Esche-richia coli. J. Bacteriol. 151:1086-1094.

37. Sasaki, M., T. Fujiyoshi, K. Shimada, and Y. Takagi. 1982. Finestructure of the recB and recC gene region of Escherichia coli.Biochem. Biophys. Res. Commun. 109:414-422.

38. Schmeissner, U., D. Ganem, and J. H. Miller. 1977. Geneticstudies of the lac repressor. II. Fine structure deletion map ofthe lacI gene, and its correlation with the physical map. J. Mol.Biol. 109:303-326.

39. Scott, J. R. 1984. Regulation of plasmid replication. Microbiol.Rev. 48:1-23.

40. Southern, E. M. 1975. Detection of specific sequences amongDNA fragments separated by gel electrophoresis. J. Mol. Biol.98:503-517.

41. Summers, D. K., and D. J. Sherratt. 1984. Multimerization ofhigh copy number plasmids causes instability: ColEl encodes adeterminant essential for plasmid monomerization and stability.Cell 36:1097-1103.

42. Sundin, O., and A. Varshavsky. 1981. Arrest of segregationleads to accumulation of highly intertwined catenated dimers:dissection of the final stages of SV40 DNA replication. Cell25:659-669.

43. Telander-Muskavitch, K. M., and S. Linn. 1981. recBC-likeenzymes: exonuclease V deoxyribonucleases, p. 234-250. InP. D. Boyer (ed.), The enzymes, vol. XIV. Academic Press,Inc., New York.

44. Tu, C.-P. D., and S. N. Cohen. 1980. Translocation specificity ofthe Tn3 element: characterization of sites of multiple insertions.Cell 19:151-160.

45. Tucker, W. T., C. A. Miller, and S. N. Cohen. 1984. Structuraland functional analysis of the par region of the pSC101 plasmid.Cell 38:191-201.

46. Way, J. C., M. A. Davis, D. Morisato, D. E. Roberts, and N.Kleckner. 1984. New TnJO derivatives for transposon mutagen-esis and for construction of lacZ operon fusions by transposi-tion. Gene 32:369-379.

603VOL. 167, 1986

on January 11, 2021 by guesthttp://jb.asm

.org/D

ownloaded from