A dnaC Mutation in Escherichia coli That Affects Copy Number of ColE1-Like Plasmids and the PriA-PriB (but Not Rep-PriC)Pathway of Chromosomal Replication Restart

Escherichia coli nusG and rho mutants, which are defective in transcription termination, are killed following transformation with several ColE1-like plasmids that lack the plasmid-encoded copy-number regulator gene rom because of uncontrolled plasmid replication within the cells. In this study, a mutation [dna-C1331(A84T)] in the dnaC gene encoding the replicative helicase-loading protein was characterized as a suppressor of this plasmid-mediated lethality phenotype. The mutation also reduced the copy number of the plasmids in otherwise wild-type strains. In comparison with the isogenic dnaC+ strain, the dnaC mutant was largely unaffected for (i) growth on rich or minimal medium, (ii) tolerance to UV irradiation, or (iii) survival in the absence of the PriA, RecA, or RecB proteins. However, it was moderately SOS-induced and was absolutely dependent on both the Rep helicase and the PriC protein for its viability. A dnaC1331(A84T) dam mutant, but not its mutH derivative, exhibited sensitivity to growth on rich medium, suggestive of a reduced capacity in the dnaC1331(A84T) strains to survive chromosomal double-strand breaks. We propose that DnaC-A84T is proficient in the assembly of replication forks for both initiation of chromosome replication (at oriC) and replication restart via the Rep-PriC pathway, but that it is specifically defective for replication restart via the PriA-PriB pathway (and consequently also for replication of the Rom– ColE1-like plasmids).

T HE replication fork represents a complex of en-In the past few years, it has been recognized that replication forks initiated at oriC often do not progress zymes that act in concert to achieve unidirectionally processive semiconservative replication of the template to completion around the chromosome but instead collapse or disintegrate spontaneously at, for example, sites DNA, notwithstanding the opposing polarity of its two strands. In Escherichia coli, the replication fork that is of template DNA damage. In such cases, the fork may be reassembled at the site of its collapse in a process called involved in replication of the chromosome as well as of various plasmids and coliphages is composed of the replication restart, which in turn is intimately connected 5Ј-3Ј replicative helicase DnaB, the primase DnaG, and to the process of RecA-dependent recombinational repair two catalytic polymerase subassemblies of the DNA polyof the damaged DNA (reviewed in Kogoma 1997; Kuzmimerase III holoenzyme (one each for the leading and Although the mechanism of replication restart is not 1996). understood in all its detail, it is DnaA independent and Initiation of replication fork assembly on the chromoshares many features with phage φX174 replication. Biosome involves the delivery by DnaC protein of the DnaB chemical studies with purified proteins in vitro have helicase to a specific region oriC at which the initiator suggested that during replication restart, a restart priprotein DnaA is bound. This assembly is also known mosome (also called the φX-type primosome) is constias the ABC primosome (Marians 1996). DnaB then tuted at a three-way-junction DNA structure, the D-loop, recruits DnaG (Tougu and Marians 1996) and DNA which is an intermediate in recombinational repair polymerase III (Kim et al. 1996) to constitute the replica- (McGlynn et al. 1997;Liu and Marians 1999; Liu et tion fork. Two replication forks so constituted at oriC Xu and Marians 2003). The restart primosome then travel in opposite directions to achieve replication is formed by the sequential action of three proteins, of the entire chromosome.
PriA, PriB, and DnaT, at the D-loop. PriA is a bifunctional protein that both initiates φX-type primosome assembly and acts as a 3Ј-5Ј DNA helicase, but studies latter function is dispensable for replication restart at copy-number phenotype with Rom Ϫ ColE1-like plasmids and synthetic lethality with certain genes, such as D-loops; PriB is a structural component of the primosome whose precise role in replication restart is not rep and priC, implicated in replication restart. Our results suggest that DnaC-A84T is affected for the PriA-established, while DnaT apparently promotes the transfer of DnaB from the DnaB-DnaC complex in solution PriB pathway but not for the Rep-PriC pathway of replication restart or for replication fork assembly at oriC. to the PriA-PriB-DNA complex (Marians 1996). The subsequent steps of replication fork formation are similar to those at DnaA-bound oriC.

MATERIALS AND METHODS
Genetic evidence not only has provided corroboration for the involvement of PriA and PriB in replication Bacterial strains, growth conditions, and plasmids: Genorestart at D-loops (in the so-called PriA-PriB pathway), types of E. coli K-12 strains are listed in Table 1. The routinely employed rich and minimal growth media were, respectively, but also has suggested that two additional pathways exist Luria broth (LB) and glucose-minimal A (Miller 1992). Antifor this process (Sandler et al. 1999(Sandler et al. , 2001 plasmids (see Cesareni et al. 1991;Del Solar et al. 1996), and pCL1921 (Spec R ; Lerner and Inouye 1990). 1998 for reviews), an RNA primer generated by the Plasmid pHYD554 is a pCL1921 derivative carrying a 5-kb processing of the plasmid-specified transcript RNA-II is chromosomal fragment that includes the rep ϩ gene and was extended by DNA polymerase I to form a D-loop that obtained by subcloning of a KpnI-PstI fragment from the phage clone 556 of the ordered genomic library of Kohara et al. is then believed to serve for replication fork assembly (1987) into the corresponding sites of pCL1921.
in much the same way as that described above for repli-DNA methods: Standard protocols were followed for experication restart on the chromosome. RNA-II processing ments involving recombinant DNA (Sambrook et al. 1989). to generate the primer is modulated by an antisense Intracellular content of test plasmids in different strains was inhibitor RNA-I, and the interaction between RNA-II estimated following agarose gel electrophoresis of plasmid preparations that were made from constant numbers of cells and its inhibitor is stabilized by the Rom protein, enwithdrawn from the cultures; prior to the plasmid isolation coded by the plasmid-borne rom gene. Null mutants in procedure, a fixed volume of culture of DH5␣ carrying plas-priA do not support ColE1 plasmid replication, and mid pHYD762 (2.2 kb) was added as loading control to normalpoint mutations in priA as well as a triple substitution ize for efficiency of plasmid recovery (Harinarayanan and mutant allele of priB that confer a reduced plasmid- . The dnaTC locus (see Figure 1) was PCR amplified from chromosomal DNA preparations of iso-copy-number phenotype have been described (Lee and genic strains GJ1541 (spl-16) and GJ1556 (spl ϩ ) with the aid Kornberg 1991; Nurse et al. 1991;Sandler et al. 1996, of primers DNACFP (5Ј-GGCAATGGCACAGCACAG-3Ј) andBerges et al. 1997;Jaktaji and Lloyd 2003). DNATRP (5Ј-GTGCGTTATCCATCGGTC-3Ј); both primers Although it is apparent from the description above were used also for sequencing the PCR products (each 1.54 that DnaC-mediated delivery of DnaB is the first comkb long), along with a third internal primer, DNACIP (5Ј-CATACTGGCGCGCTTTGC-3Ј). mon step in all mechanisms of replication fork assembly, assembled at a D-loop). dnaC mutants that are temperaas described (Kleckner et al. 1991;Miller 1992). A P1 lysate prepared on a pool of 10 6 Kan R clones was used to transduce ture sensitive (Ts) for growth have been characterized the nusG spl ϩ strain GJ1504 to Kan R . The Kan R transductants (Wechsler 1975;Maisnier-Patin et al. 2001), consiswere pooled and transformed with plasmid pACYC184 to identify tent with the protein's essential role in replication fork derivatives of GJ1504 that, by virtue of their coinheriting the assembly at oriC (and perhaps also in replication respl-16 allele (from GJ1538) with a linked Tn10dKan insertion, start). Other dnaC mutations have been recovered as survived transformation with pACYC184. One clone (GJ1539) so identified was subsequently shown to have a Kan R insertion suppressors of sickness or inviability associated with priA (designated zji-901::Tn10dKan) ‫%09ف‬ cotransducible with disruption (Kogoma et al. 1996;Sandler et al. 1996Sandler et al. , the spl-16 locus. 1999Sandler et al. , 2001Gregg et al. 2002;Rangarajan et al. 2002; UV survival tests: Sensitivity of a strain to UV was determined Jaktaji and Lloyd 2003). In this article, we report the essentially as described by Miller (1992). Cells from a 50-ml identification and characterization of a novel dnaC muvolume of LB culture (grown at 37Њ to an A 600 of ‫)5.0ف‬ were chilled on ice for 30 min and resuspended in an equal volume tant [dnaC1331(A84T)] that exhibits both a reduced-  (Sambrook et al. 1989); MC4100 and GJ11 (Gowrishankar 1985); SS135 (Sandler et al. 2001); GJ1313 ( Jayashree and Gowrishankar 1995, wherein the nusG-G146D allele was designated as rpoB364); and GJ1504 and GJ3185 (Harinarayanan and Gowrishankar 2003). All other strains were obtained or constructed in this study.
b Genotype designations are as in Berlyn et al. (1996). All strains are F Ϫ . The following mutations were transduced from the parenthetically indicated strains described in SaiSree et al. c The dnaC1331(A84T) mutation has also been referred to as spl-16 in this study. d The mdoB::Tn10 insertion was previously identified as zji-202:: Tn10 (Berlyn et al. 1996;Nichols et al. 1998) and was transduced from strain CAG18430. Although the Tn10kan insertion in CAG18619 was reported to be at the same site as the Tn10 insertion in CAG18430 (Berlyn et al. 1996), our transductional mapping data suggest otherwise (see text and data not shown). e GJ5201 was recovered and maintained as a transformant derivative with the plasmid pHYD554 carrying rep ϩ . f GJ5205 is a derivative of strain GJ1992 (SaiSree et al. 2000. Not all mutations in the strain have been listed here.
g GJ5224 was constructed in several steps of P1 transduction from strain GJ1538.
of 0.1 m MgSO 4 . Five-milliliter aliquots of the suspension were Other methods: All strain constructions were by P1 transduction (Gowrishankar 1985); when the recipient strain carried placed in 85-mm petri dishes and exposed to UV (at 0.4 J/ M 2 /sec) for different lengths of time, after which they were an Amp R plasmid, transductants were recovered at limiting dilutions to minimize the extent of Amp degradation by the protected from visible light. Appropriate dilutions of each aliquot were then plated on LB for estimation of the viable recipient lawn on the selection plates. The mutator phenotype of a strain was scored by spotting ‫01ف‬ 7 cells of LB-grown count.
cultures on LB-Rif plates; typically, the mut ϩ and mut strains We have shown previously that a rho-A243E mutant yielded, respectively, Ͻ3 and 50-100 Rif R colonies per spot.
is also killed following transformation with pACYC184 Procedures for nitrosoguanidine mutagenesis and ␤-galactosi- (Harinarayanan and Gowrishankar 2003). With the dase assays were as described (Miller 1992). aid of the linked Tn10dKan marker, the dnaC1331 allele was introduced into the rho strain, and here, too, it was RESULTS able to suppress the plasmid-mediated lethality (data not shown). Identification of dnaC1331(A84T) as a suppressor of dnaC1331(A84T) reduces copy number of Rom Ϫ nusG-pACYC184 lethality: We have recently described a ColE1-like plasmids: Harinarayanan and Gowrisnovel lethality phenotype associated with runaway replihankar (2003) have shown that lethality in pACYC184 cation (that is, uncontrolled increase in copy number) transformants of the nusG and rho strains was accompaof ColE1-like plasmids such as pACYC184 in E. coli nusG nied by a substantial increase in plasmid copy number or rho mutants that are defective in factor-dependent that is most prominent in the early stationary phase of transcription termination (Harinarayanan and Gowgrowth; furthermore, mutations in several loci (includrishankar 2003). Following nitrosoguanidine mutaing hns, rpoB, pcnB, and polA), as well as in multicopy genesis of a nusG strain GJ1313, an extragenic suppres-recG ϩ or rom ϩ , were identified as suppressors of lethality sor mutant GJ1538 was identified that survived that were all also associated with reversal of the copy transformation with pACYC184. The mutation was desnumber alteration in the mutants. ignated spl-16 (for suppressor of plasmid lethality). To The dnaC1331(A84T) suppressor mutation, characfacilitate mapping and transfer of the suppressor mutaterized in the present study, also behaved like the suption into other strain backgrounds, it was tagged with pressors previously described in that it was able to rea transposon (Tn10dKan) marker as described in mateverse the increase in pACYC184 content otherwise rials and methods. Conjugational and transductional observed in cultures of the nusG strain in the earlymapping experiments (data not shown) demonstrated stationary growth phase ( Figure 2). Densitometric analythat the Tn10dKan insertion (zji-901::Tn10dKan) and sis of the band intensity of pACYC184 (relative to that the spl-16 mutation are at 98 min on the E. coli linkage of the loading control) in each of the lanes in Figure  map and that the latter is 42 and 97% cotransducible, 2 indicated that the pACYC184 content (per unit of respectively, with the previously described transposon biomass) in the nusG mutant in the exponential and insertions zji-3188::Tn10kan and mdoB::Tn10 (prestationary phases of growth was, respectively, two-and viously called zji-202::Tn10; Berlyn et al. 1996; Nichols ninefold higher than that in the corresponding culture et al. 1998).
aliquots of either the wild-type strain or the nusG On the basis of our finding (see below) that the muta-dnaC1331 double mutant. The latter two strains themtion affects copy number of ColE1-like plasmids in othselves exhibited an approximate doubling in pACYC184 erwise wild-type strains, we reasoned that it may be situcontent as the cultures progressed from the log phase ated in the dnaTC locus [which is 3 kb from mdoB to the stationary phase. (Blattner et al. 1997)] whose products are known to Two mechanisms have been identified by which the difbe involved in replication fork assembly. Accordingly, ferent suppressors act to suppress nusG-or rho-pACYC184 the dnaTC operon from a pair of isogenic spl ϩ and spllethality. The first is by reversing the transcription termi-16 strains was PCR amplified and sequenced. A single nation defect in the mutants, and the second is by a mutation was identified in the spl-16 strain that altered primary effect on plasmid copy number even in the codon 84 of dnaC from GCG to ACG (which is deduced nusG ϩ rho ϩ strain (Harinarayanan and Gowristo cause the change of a conserved Ala residue to Thr hankar 2003). The dnaC1331(A84T) mutation did not in the protein). The mutation has been designated bealleviate the defect in transcription termination in the low as dnaC1331(A84T).
nusG and rho strains, as judged by the extent of transcrip-The fact that the mutation also results in loss of a tional polarity relief at different loci such as galP3 (an BsmI site allowed us to test for cosegregation of the BsmI IS2 insertion in the interval between the promoter and polymorphism with suppression of pACYC184-mediated the first structural gene of the gal operon) or trpE (Oc; lethality in the nusG spl-16 strain. A P1 lysate prepared a nonsense mutation in the first gene of the trp operon; on strain CAG18619 (which has the Tn10kan marker data not shown). On the other hand, the mutation was 42% linked to spl ϩ ) was used to transduce GJ1538 (nusG associated with a pronounced decrease in plasmid pAspl-16) to Kan R , and a complete concordance between CYC184 content even in the nusG ϩ rho ϩ background (Figpresence (or absence) of the BsmI site and lethality ure 3). (or survival) following pACYC184 transformation was As mentioned in the Introduction, the plasmid-encoded observed among 20 transductants that were tested (Figure Rom protein serves to downregulate the copy number 1). These results established that the dnaC1331(A84T) of ColE1-like plasmids by stabilizing the interaction bemutation (or another closely linked mutation) is responsitween RNA-II and its antisense inhibitor RNA-I. When ble for suppression of pACYC184-mediated lethality in the nusG strain.
plasmids such as pUC19, pBR329, or pBR322 were ex-dnaC1331(A84T) Mutant of E. coli Figure 1.-Association of the spl-16 [dnaC1331(A84T)] mutation with loss of the BsmI site in dnaC. The dnaTC locus was PCR amplified from individual Kan R transductants of a cross [P1(CAG 18619) ϫ GJ1538] in which the spl ϩ and spl-16 alleles were segregating and the PCR products were analyzed by agarose gel electrophoresis following digestion with BsmI. Positions of DNA size markers (in base pairs) are marked beside the gels. (Top) Linear depictions of the PCR products from the dnaC ϩ and dnaC1331(A84T) strains with the expected sizes (in base pairs) of the BsmI fragments obtained therefrom marked (BsmI recognition sites shown as solid diamonds). Marked beneath the gel lanes are the phenotypes of the transductants with respect to lethality (ϩ, that is, spl ϩ ) or survival (Ϫ, that is, spl-16) following pACYC184 transformation and the deduced presence (ϩ) or absence (Ϫ) of the BsmI site overlapping the codon 84 in dnaC. Transductants in gel I were first classified as spl ϩ or spl-16 and subsequently tested for BsmI polymorphism, while those in gel II were first classified on the basis of BsmI polymorphism and then tested for pACYC184-mediated lethality. R, strain GJ1538 (spl-16) used as recipient in the transduction. amined in isogenic dnaC ϩ and dnaC1331(A84T) mu-et al. (1992) have shown that (i) the very-high-copytants, it was observed that the content of the first two number phenotype of plasmid pUC19 is the result of a (which, like pACY184, are Rom Ϫ ) is also markedly remutation in its gene encoding RNA-II, which destabiduced in dnaC relative to dnaC ϩ , whereas that of the lizes the latter's interaction with the inhibitor RNA-I; (ii) third (which is Rom ϩ ) is only marginally affected in the the interaction of pUC19-encoded RNA-II with RNA-I mutant (Figure 3). Densitometric analysis of the test is rendered even less stable at 42Њ, so that cultures that plasmid band intensities in Figure 3 (after normalizahave been shifted from 30Њ to 42Њ exhibit a further tion against the corresponding loading controls) indiincrease in plasmid copy number and a reduction in cated that the ratio of plasmid content in the dnac1331 growth rate; and (iii) Rom protein expression rescues mutant to that in dnaC ϩ was, respectively, 0.19, 0.06, the slow-growth phenotype at 42Њ. and 0.27 for the plasmids pACYC184, pBR329, and We confirmed in this study that pUC19 transformants pUC19 and that it was 0.77 for plasmid pBR322. These of the wild-type strain GJ3128 exhibit a pronounced results also established that there is no correlation beincrease in plasmid copy number upon temperature tween plasmids conferring lethality in the nusG or rho upshift to 42Њ ( Figure 4A).The transformants also exhibmutants and those whose copy number is reduced by ited a Ts growth phenotype on LB agar medium (Figure the dnaC1331(A84T) mutation [since pBR329 is an ex-4B). The isogenic dnaC1331(A84T) mutant GJ3127 example of a plasmid that does not kill the nusG/rho strains hibited neither the plasmid copy number increase at 42Њ (Harinarayanan and Gowrishankar 2003) and yet ( Figure 4A) nor the Ts growth phenotype ( Figure 4B). whose copy number is reduced in the dnaC strain].
As expected, the pUC19-associated Ts phenotype was Suppression by dnaC1331(A84T) or priB of the Ts also suppressed in a pcnB strain [in which plasmid copy growth phenotype of pUC19 transformants: Lin-Chao number is reduced because of an increased stability of the inhibitor RNA-I (Xu et al. 1993; Figure 4B)]. When    dard error of the mean as marked) are given in Miller units (Ϫ) derivatives GJ3128 and GJ3127, respectively, either upon (Miller 1992). Strain pairs employed were (in the order, continuous incubation at 30Њ or following upshift (at an A 600 dnaC ϩ and dnaC1331) wild type, GJ5205 and GJ5206; recA, of ‫)5.0ف‬ to 42Њ for 4 hr (30 → 42Њ). See materials and GJ5207 and GJ5208; and lexA3, GJ5209 and GJ5210. methods for description of loading control. Plasmid preparations were digested with EcoRI before gel electrophoresis. (B) Growth phenotypes on LB medium at 30Њ and 42Њ of wildcation restart pathways. In the first approach, double type (GJ3128), dnaC1331 (GJ3127), pcnB (GJ3185), priB mutants of dnaC1331(A84T) with each of the other mu-(SS135), and priC (GJ5204) strains with (ϩ) or without (Ϫ) pUC19, as indicated. Inocula were taken from LB medium tations were sought to be constructed by P1 transducplates that had been incubated at 30Њ. tion. As shown in Figure 6, viable double mutants were recovered with recA::Kan, recB::Tn10, or priA::Kan; likewise, double mutants were also obtained at the expected null mutations in priB or priC were tested, the former frequencies for three other loci: recG258::Kan, ⌬ruv but not the latter suppressed the Ts phenotype ( Figure  ABC::Cm, and ⌬priB302 (data not shown). [Consistent 4B). [In the absence of pUC19, the wild type and all with an earlier report (Gregg et al. 2002), the priA single the mutant strains were unaffected for growth at 42Њ mutant was itself very sick and gave rise to many faster-( Figure 4B).] A priB missense mutant with reduced copy growing suppressor clones (see Figure 6).] Growth of number of Rom Ϫ ColE1-like plasmids has previously each of the double mutants was in general no more been reported (Berges et al. 1997).
significantly impaired than that of the respective single-Chronic SOS induction in the dnaC1331(A84T) mulocus mutants, although the recB dnaC strain alone did tant and synthetic lethality with rep or priC mutations: exhibit a 40% slower growth rate than recB (see Figure 6). The growth rates of the wild-type strain GJ3128 and its On the other hand, no transductants were obtained when dnaC1331(A84T) derivative GJ3127 were indistinguishwe sought to introduce the rep::Kan, ⌬rep::Cm, or priC::Kan able in both LB (2.47 and 2.44 hr Ϫ1 , respectively, at 37Њ; alleles into the dnaC1331(A84T) strain (data not shown). see also Figure 4B and Figure 6) and glucose-minimal Two additional experiments were undertaken to es-A (0.87 and 0.88 hr Ϫ1 , respectively, at 37Њ), suggesting tablish that rep is synthetically lethal with dnaC-A84T. that the mutant protein is unaffected in its ability to The strategy for the first was analogous to that employed load DnaB helicase at replication forks initiated at oriC.
by Sandler (2000) for demonstrating rep-priA lethality At the same time, the mutation conferred a moderate and involved an attempt to introduce dnaC1331(A84T) level of chronic SOS induction with a threefold increase in a nonselective fashion into the ⌬rep strain. A P1 lysate of sulA-lac expression in the dnaC1331(A84T) strain; prepared on a strain (GJ3127) carrying the dnaC1331 the increase was abolished, as expected, upon introducmutation along with the closely linked mdoB::Tn10 tion of recA or lexA3 mutations that render the SOS marker was used to transduce a pair of isogenic rep ϩ regulon noninducible (Walker 1996; Figure 5). These and ⌬rep::Cm strains (MC4100 and GJ5203, respectively) observations suggested that the dnaC1331 mutant might to Tet R . The recipient strains also carried plasmid be affected for replication fork assembly at restart pri-pUC19 that rendered them Ts for growth (see above). mosomes.
As expected, the vast majority (39/40) of the Tet R trans-As a test of this possibility, we looked for synthetic ductants in the cross into the rep ϩ strain had become temperature resistant, indicating coinheritance of dna-lethal interactions between dnaC and other genes previously implicated in recombinational repair and repli-C1331(A84T), whereas none of 40 transductants in the former may be compromised for one or more pathways of replication restart on the chromosome. We reasoned that the mutant may also consequently be unable to tolerate conditions that increase the need for replication restart, and to test this possibility we introduced a dam mutation into the dnaC1331 strain. Mutants deficient in the Dam methylase suffer an increased frequency of double-strand breaks (DSBs) by a process that is dependent on a functional mismatch repair system encoded by general, the need for replication restart is also increased GJ3128 and GJ3127;recA,GJ5223 and GJ5222;recB,GJ5225 during growth of cultures in rich medium (when the and GJ5224; and priA, GJ5221 and GJ5220. time interval between successive replication forks is considerably reduced) when compared to that in minimal medium, and priA mutants that are defective in replica-⌬rep::Cm strain had become temperature resistant. In tion restart or priB gyrB strains exhibit rich-medium sena variation of this experiment, the recipient strains did sitivity (Kogoma 1997;Gregg et al. 2002; Grompone et not carry plasmid pUC19 and inheritance of the al. 2003). dnaC1331 mutation in the transductants was scored di-Transductants recovered from the cross in which a rectly by testing each of them (with the PCR-based dam::Tn9 mutation was introduced into the dnaC1331 method described above) for concomitant loss of the (A84T) strain were very sick on LB but not on glucose-BsmI site in the gene. Of 11 transductants tested from minimal agar plates (see Figure 8) and gave rise to sponthe cross into the rep ϩ strain, 10 had lost the BsmI site taneous fast-growing suppressors. Five of eight such in dnaC, whereas none of 12 from the cross into the suppressors that were tested had acquired a mutator ⌬rep::Cm strain had lost this site (Figure 7). phenotype, suggestive of the loss of mismatch repair In the second experiment, the frequency of loss of functions in them. dam::Tn9 transductants that were a Spec R plasmid carrying rep ϩ (pHYD554) in either a recovered in a dnaC1331(A84T) mutH strain GJ5212 also ⌬rep::Cm dnaC ϩ (GJ5202) or a ⌬rep::Cm dnaC1331 did not exhibit the rich-medium-sensitive phenotype (GJ5201) strain was scored following introduction of a ( Figure 8). second plasmid, pLG339 (Kan R ), that is incompatible These data provide strong support for the notion that with pHYD554. In the former strain, after 20 generations the dnaC1331 mutant is rendered growth sensitive by of growth in the absence of Spec, 26 of 30 Kan R transthe increased frequency of replication-directed DSBs formants had become Spec S even as they retained the that are generated in the absence of Dam methylase. ⌬rep::Cm marker. On the other hand, in the dnaC strain, On the other hand, the dnaC1331(A84T) single mutant only 5 of 30 were Spec S and in all these cases, the strains was no more sensitive to UV irradiation than was its had also become rep ϩ on the chromosome (that is, Cm S ) isogenic dnaC ϩ parent (Figure 9), suggesting that the presumably by recombination between the plasmid and mutant protein is able to function as effectively as the the chromosome. These results indicated that a ⌬rep wild type to satisfy the increased need for replication dnaC mutant cannot survive in the absence of the plasrestart on UV-damaged DNA templates. mid with rep ϩ .
That priC is synthetically lethal with dnaC1331 (A84T) was also established by the experiment of P1 DISCUSSION transduction with the lysate of the mdoB::Tn10 dnaC strain (GJ3127) into isogenic priC ϩ and priC::Kan recipi- The dnaC1331(A84T) mutation was identified in this ents (MC4100 and GJ5204, respectively) carrying plasstudy by its ability to suppress the lethality associated mid pUC19. As with rep, none of 50 Tet R transductants with runaway replication of Rom Ϫ ColE1-like plasmids in the priC mutant had lost the Ts phenotype (as comin transcription termination-defective nusG and rho mupared to 47/50 in the priC ϩ strain), indicating the abtants and was shown to do so by a primary effect on sence of inheritance of the dnaC mutation in any of plasmid copy number even in otherwise wild-type them. By the PCR-based BsmI polymorphism test as well, strains. The amino acid alteration in the mutant protein none of 12 Tet R transductants examined of GJ5204 has occurred in a short stretch of identified sequence (without pUC19) had inherited the dnaC1331 allele similarity with the P protein (which is functionally anal- (Figure 7). ogous to DnaC in phage replication; Nakayama et al. mutH-suppressible broth sensitivity of the dnaC1331 1987). The dnaC1331(A84T) allele was also synthetically (A84T) mutant: The synthetic lethality of dnaC1331 lethal with the rep or priC null mutations. These results are interpreted below in terms of a model in which the (A84T) with rep or priC mutations suggested that the Lanes in the three panels represent individual Tet R transductants obtained from crosses into the recipient strains wild type (MC4100), ⌬rep::Cm (GJ5203), or priC303::Kan (GJ5204), as indicated, with a P1 lysate grown on GJ3127 (dnaC1331 linked 95% to mdoB::Tn10). Chromosomal DNA from each transductant was subjected to PCR of and BsmI polymorphism testing at the dnaC locus, as described in the text and in the legend to Figure 1. R, chromosomal DNA from the corresponding recipient strain. Marked beneath each lane is the dnaC genotype (ϩ, dnaC ϩ ; Ϫ, dnaC1331) of the strain, as inferred from the BsmI polymorphism analysis.
dnaC mutant is specifically defective in the PriA-PriB supporting biochemical evidence is available at present (Xu and Marians 2003). pathway of replication restart.
Many phenotypes associated with the dnaC1331 DnaC and assembly of replication forks by different (A84T) mutation (for example, that it is synthetically pathways: As mentioned above, DnaC delivers the DnaB lethal with priC but not with either priA or priB) may be helicase for replication fork assembly during the proexplained by this multiple-pathway model of replication cesses of (i) initiation of chromosome replication at restart by making the additional assumptions that (i) oriC, (ii) replication restart following the collapse of the ability to restart replication at D-loops is important previously assembled forks, and (iii) plasmid replicafor viability and (ii) in the dnaC1331 strain, the PriAtion. The dnaC1331(A84T) mutant exhibits chronic PriB pathway of replication restart at D-loops is rendered SOS induction, which suggests some perturbation in largely, if not completely, nonfunctional; consequently, chromosomal DNA replication in the strain. The fact the mutant is dependent on the functionality of the that the mutant has a normal growth rate in both rich PriA-PriC and Rep-PriC pathways for its survival. The fact and minimal media argues that it is not affected for that the dnaC1331 mutant is sensitive to Dam methylase replication fork assembly at the ABC primosome (that deficiency (Figure 8), but not to UV irradiation (Figure is, DnaA-bound oriC). That, on the other hand, it may 9), would then suggest that there is a much greater be affected for replication restart (that is, at the φXneed for the PriA-PriB pathway of replication restart on type primosome at D-loops) is suggested by the findings DNA templates with DSBs than on those with singlethat (i) dnaC1331(A84T) confers rich-medium sensitivstrand lesions. McGlynn and Lloyd (2000) have also ity in combination with dam (that is, when the requireprovided evidence that replication-generated DSBs are ment for replication restart functions is expected to infrequent in UV-irradiated E. coli. be increased because of frequent DSBs) and (ii) this inviability is suppressed by mutation in the gene mutH for mismatch repair.
Sandler and co-workers have obtained genetic data to suggest that three alternative pathways exist for replication restart at D-loops in wild-type cells, which have been designated (on the basis of respective gene product requirements) as the PriA-PriB, PriA-PriC, and Rep-PriC pathways (Sandler et al. 1999(Sandler et al. , 2001Sandler 2000). Of these, it is only for the PriA-PriB pathway that Figure 9.-UV tolerance of dnaC ϩ (GJ3128, ᭝) and dnaC1331 (GJ3127, ᭹) strains. The percentage fraction of surviving cells, following exposure to UV for the indicated Figure 8.-Rich-medium sensitivity in dnaC1331(A84T) dam mutant and its suppression by mutH. Strains GJ3127 time periods, was determined. Two independent experiments yielded similar results, and the data from one of them are (dnaC), GJ5211 (dam dnaC), and GJ5213 (dam dnaC mutH) were scored for growth on glucose-minimal A (Min A) and plotted. The initial culture densities for the two strains were, respectively, 1.2 ϫ 10 8 and 9.2 ϫ 10 7 /ml. LB agar plates. Consistent with the model that the PriA-PriB pathway is rendered nonfunctional in the dnaC1331(A84T) mutant is the finding that the dnaC1331 strain shares several phenotypic features in common with the null priB mutant (Sandler et al. 1999;Sandler 2000;Grompone et al. 2003; this study): minimal or no growth phenotype as single mutants, no change in UV tolerance, reduced copy number of ColE1-like plasmids, viability in combination with priA, and severe sickness or lethality in combination with rep or priC. On the other hand, the dnaC1331 mutant, but not the priB mutant, is SOS induced, which suggests that the former might harbor 3Ј-5Ј DNA helicase and φX-type primosome assembly activities, respectively, of PriA.
One finding of ours that is not readily accounted for by the three-pathway replication restart model of Sandler and co-workers is that of rep-dnaC1331 synthetic priA300 strain is one that does not require PriC, which is lethality (because the PriA-PriC pathway would still be contrary to the proposal made by Sandler et al. (2001)]. expected to be functional in the double mutants). One On the basis of these arguments, therefore, we suggest possibility is that the PriA-PriC pathway requires the Rep that replication restart events occur primarily by the helicase (as has also been speculated earlier; Sandler PriA-PriB and Rep-PriC pathways at D-loops, as well as and Marians 2000); in such a case, however, rep would by the direct resetting of regressed replication forks be expected to exhibit synthetic lethality with priB, but ( Figure 10) and that, in the dnaC1331(A84T) mutant, the double mutants have been reported to be viable the PriA-PriB pathway is rendered nonfunctional. (Sandler 2000). The alternative possibility is that PriA-dnaC1331(A84T) and replication of ColE1-like plas-PriC may not represent a major pathway of replication mids: The dnaC1331(A84T) mutation was associated restart, as is further discussed below.
with a reduction in content of the Rom Ϫ ColE1-like Two lines of genetic evidence were earlier interpreted plasmids pACYC184, pBR329, or pUC19 (and conseas support for a PriA-PriC pathway of replication restart: quently with suppression of the Ts phenotype of pUC19 (i) priB or priC single mutants are reasonably healthy transformants), but not with that of the Rom ϩ plasmid whereas a priBC double mutant is even more sick than pBR322. One way to explain these results is to assume priA (Sandler et al. 1999) and (ii) the priA300 mutation, that DnaB loading by the mutant DnaC protein at the which specifies an altered PriA protein that is competent D-loop formed in the plasmid ori region occurs with for φX-type primosome assembly but is defective for the reduced efficiency and that it consequently becomes 3Ј-5Ј DNA helicase activity, behaves like the priA knockthe rate-limiting step for replication of Rom Ϫ plasmids, out when combined with a priB but not a priC null but that for Rom ϩ plasmids, it is the number of replicamutation; it was proposed that the mutant protein is tion primers that remains rate limiting for replication proficient for replication restart via the PriA-PriB but [since the Rom protein reduces the proportion of RNAnot via the PriA-PriC pathway (Sandler et al. 2001).
II molecules that are processed to generate them (Ces-However, both these findings may also be explained by areni et al. 1991;Del Solar et al. 1998)]. Consistent an alternative model in which the helicase activity of with this explanation, we have found that the dnaC1331 PriA participates in an additional, PriB-and PriC-indemutation also does not affect the copy number of Rom Ϫ pendent pathway of replication restart (see Figure 10). plasmids in a pcnB strain [in which the half-life of the Indeed, a pathway of direct resetting of regressed repliinhibitor RNA-I is increased (Xu et al. 1993), so that cation forks (that can occur following blockage of noronce again fewer replication primers are generated mal fork progression; Seigneur et al. 1998; McGlynn from RNA-II (data not shown)]. Our hypothesis is very and Lloyd 2000) that is mediated by the combined similar to that proposed by Berges et al. (1997), who actions of the RecG (or RecBCD) and PriA proteins showed that a missense priB mutant harbors reduced (including the latter's 3Ј-5Ј helicase activity) has been copy numbers of Rom Ϫ but not of Rom ϩ ColE1-like proposed by Lloyd and co-workers (McGlynn and plasmids. Lloyd 2000; Gregg et al. 2002;Jaktaji and Lloyd The effects on plasmid replication of dnaC1331(A84T), 2003). Furthermore, Flores et al. (2002) have shown priB null, or priC null mutations, as inferred from the that in a strain with increased frequency of replication data obtained in this study (Figure 4), are also in accord fork regression because of a DNA polymerase III -subwith a model in which replication fork assembly on the unit mutation, replication restart is compromised by plasmid template is served primarily by the components the priA300 mutation but not by a priC null mutation [suggesting that the pathway rendered defective in the of the PriA-PriB pathway of replication restart.
Need for and relative efficiency of the different repli- (2001) showed also that 100% of rep dnaC(Ts) cells failed to complete replication initiated at oriC following cation restart pathways: When one considers the multiple replication restart pathways (Figure 10), two major a temperature upshift, but whether this was because of increased fork collapse or decreased restart associated questions are: What fraction of replicating chromosomes in wild-type cells (in the absence of exogenous with the rep mutation is not clear. With regard to the relative efficiency of different path-DNA-damaging agents) suffers fork collapse necessitating replication restart? And what is the relative efficiency ways for restart, the extreme sickness of priA mutants has again been taken to mean that the PriA-independent of each of the pathways in the process? The estimated values, in answer to the first question, have varied from pathways are at best minor contributors to replication restart. On the other hand, our finding in this study 18 to 50 to 100% by different approaches, as discussed below.
that the dnaC1331(A84T) single mutant is quite normal for growth but is inviable in combination with priC i. The lower-end value was obtained in experiments [which, unlike priA and priB, has so far been implicated involving growth of dnaC(Ts) mutants at the restriconly in "minor" pathway(s) of replication restart (Santive temperature (Maisnier-Patin et al. 2001). Howdler 2000;Sandler et al. 2001;Flores et al. 2002; ever, our own observations with dnaC1331(A84T) Grompone et al. 2003)] strongly suggests that the Repsuggest that dnaC mutants may be differentially af-PriC pathway may indeed potentially be as efficient as fected for DnaB loading at different fork-precursor the "major" PriA-PriB pathway. structures, and it is therefore possible that, in the