Primosome assembly requirement for replication restart in the Escherichia coli holDG10 replication mutant

Authors


Summary

In this report, we study the role of pre-primosome proteins in a strain in which the frequency of replication arrest is increased because of a mutation in a replication protein. The holDG10 mutant was used, in which replication restart involves replication fork reversal. As expected, PriA primosome assembly function is essential for growth of the holDG10 mutant. The priA300 mutation, which inactivates only the helicase function of PriA in vitro, and priB inactivation strongly impair viability. In contrast, priC inactivation has no effect. Therefore, PriB is more important than PriC for PriA-dependent replication fork restart in vivo. The gain of function mutation dnaC809 restores the viability of holDG10 priA and holDG10 priB mutants only to some extent. The dnaC809 820 double mutation restores full viability to the holDG10 mutant lacking either PriA or PriB. Similarly to the holDG10 single mutant, the holDG10 priA dnaC809 820 strain is depend-ent on RecBC for viability, indicating that facilitating primosome assembly using the dnaC809 820 mutation does not allow bypass of replication fork reversal.

Introduction

Replication of bacterial chromosomes is not a continuous process, but is occasionally impaired by the encounter with obstacles such as DNA lesions, secondary structures or DNA bound proteins. When replication arrest causes the disassembly of the replication machinery, replication needs to be re-initiated from a non-origin sequence. Replication restart, in such cases, requires the loading of the DnaB helicase on the inactivated replication fork, which, in turn, triggers the binding of the DnaG primase and of the holoenzyme polymerase III (Pol III HE). DnaB is loaded by a specific protein called DnaC, which is targeted to the inactivated replication fork by the preprimosome proteins PriA, PriB, PriC and DnaT (reviewed in; Marians, 2000; Sandler and Marians, 2000). The assembly of the primosome involves (i) recognition of a forked structure or of a D-loop by the PriA protein; (ii) assembly of the pre-primosome proteins; and (iii) loading of DnaB by the DnaC–DnaB complex on the single-strand region of the lagging strand template, coated by single-strand DNA binding protein (SSB). PriA is the key protein for replication restart, presumably because of its targeting function. The absence of PriA leads to poor growth, SOS induction, cell filamentation and recombination deficiency (Kogoma et al., 1996; Sandler et al., 1996). In contrast, inactivation of either priB or priC genes causes no apparent deleterious phenotype. PriB and PriC proteins were proposed to fulfill redundant functions for PriA-dependent primosome assembly, based on the observation that although inactivation of either of them has no deleterious effect, inactivation of both is lethal (Sandler et al., 1999). However, purified PriB and PriC proteins were not interchangeable in vitro, as replication initiation from a recombination intermediate was strictly dependent on PriB (Liu and Marians, 1999; Liu et al., 1999). The reasons for the difference between in vivo and in vitro observations remained to be understood.

The residual viability of the priA null mutant relies on the presence of the PriC protein, suggesting that in the absence of PriA, DnaC can be targeted to arrested forks by a weak PriC-dependent pathway (Sandler, 2000). The efficiency of the PriC pathway was greatly enhanced by gain-of-function mutations in dnaC such as the dnaC809 mutation, which restores a wild-type phenotype to priA null mutants (Sandler et al., 1996). Finally, all preprimosome proteins were dispensable in strains carrying a further improved dnaC allele with two mutations, called dnaC809 820 (Sandler et al., 1999). In vitro studies confirmed that dnaC809 mutation conferred a gain of function. Indeed, the DnaC809 protein could promote in vitro DnaB loading and replication initiation from a recombination intermediate in the absence of all PriA, PriB and PriC proteins (Liu et al., 1999; Xu and Marians, 2000). In vivo, dnaC809 mutation only partially suppresses the growth defect of a triple priA priB priC mutant whereas dnaC809 820 mutation is able to suppress almost completely the need for PriA, PriB and PriC proteins (Sandler et al., 1999). The lethality of the priA priC double mutant suggested that replication restart is a process essential for viability (reviewed in Sandler and Marians, 2000). However, flow cytometry analysis in a dnaCts mutant suggested that replication arrest might be less common than previously thought, as 80% of the chromosomes can replicate to completion in cells deficient for replication restart (Maisnier-Patin et al., 2001). Consequently, in wild-type cells the importance of certain pre-primosome proteins may be underestimated because of a limited need for replication restart in the absence of a specific replication defect.

Here, we used the holDG10 mutation to analyse the requirement for pre-primosome proteins in a mutant cell in which the frequency of replication arrest is increased. holD encodes the Psi subunit of Pol III, a subunit of the Gamma complex involved in the loading of the processivity clamp at the onset of Okazaki fragment initiation (Naktinis et al., 1996; Yuzhakov et al., 1999). In the holDG10 mutant, replication is frequently arrested, leading to a specific reaction called replication fork reversal (Flores et al., 2001). This reaction involves the annealing of leading and lagging strand upon replication arrest, forming a double-strand end and a Holliday junction (Fig. 1, first step). The Holliday junction is recognized by RuvAB. The double-strand end is acted upon by RecBCD and is either re-incorporated into the chromosome by RecA- RecBCD-dependent homologous recombination (Fig. 1A), or degraded by the exonuclease V action of the enzyme (Fig. 1B). In the absence of RecBCD, resolution of the Holliday junction formed by replication fork reversal leads to chromosome breakage (Fig. 1C; Seigneur et al., 1998; reviewed in Michel, 2000; Michel et al., 2001). The replication fork reversal model is supported in the holDG10 mutant by two sets of data: (i) direct analysis of chromosomes showed RuvABC-dependent linearization in the absence of RecBCD (Fig. 1C) and (ii) genetic analysis showed that the holDG10 mutant requires RecBCD for viability, whereas it only requires RecA in the absence of exonuclease V and vice versa (Fig. 1A and B). In the present work, we tested the importance of PriA, PriB and PriC proteins for the viability of the holDG10 mutant. We show that PriA is essential, PriB important and PriC dispensable for replication restart in the holD strain. Therefore, in a strain with increased frequency of replication arrest, PriB is more important for replication restart than PriC, as in the in vitro reconstituted reaction. We also tested the effects of the dnaC809 and dnaC809 820 allele on cell viability. The dnaC809 allele only partly restored the viability of holD priA or holD priB mutant, whereas dnaC809 820 allele restored full viability. Finally, we show that the priA300 allele, which inactivates only the helicase function of PriA in vitro, affects the viability of the holDG10 mutant.

Figure 1.

Model for processing arrested replication forks using recombination proteins in the holD mutants (adapted from Flores et al., 2001). In the first step, the replication fork is blocked owing to a defect in the lagging strand polymerase in the holDG10 mutant. The two newly synthesized strands anneal, forming a Holliday junction that is stabilized by RuvAB binding. Pathway A: RecBCD initiates a genetic exchange mediated by RecA and resolved by RuvABC; replication restarts from a D-loop made by homologous recombination. Pathway B: RecBCD-mediated degradation of the double-strand tail progresses up to the RuvAB-bound Holliday junction; replication restarts from a forked structure after displacement of the RuvAB complex by RecBCD. Pathway C: in the absence of RecBCD, RuvC resolves the RuvAB-bound Holliday junction; resolution at both forks results in the formation of linear DNA. Continuous and discontinuous lines represent the template and the newly synthesized strands of the chromosome respectively; the arrow indicates the 3′-end of the growing strand.

Results

holD mutants require the primosome assembly function of PriA for viability

The Pol III HE defect in the holDG10 mutant causes frequent replication arrests, leading to replication fork reversal and presumably causing a need for the re-loading of a replisome. To analyse the proteins required for replication restart in the holDG10 mutant, this mutation was combined with different mutations that inactivate preprimosome proteins. We previously observed that strains carrying the holDG10 allele grow slowly and are sensitive to rich medium. During propagation, they acquire compensatory mutations that suppress the slow growth phenotype and the rich medium sensitivity. To avoid the appearance of suppressor mutations, strains were constructed in the presence of a plasmid that carries the holD wild-type gene and replicates with a conditional origin, pAM-holD (Flores et al., 2001). This plasmid replicates only in the presence of IPTG; consequently, cell propa-gation in the absence of IPTG leads to plasmid loss (Gil and Bouché, 1991; Flores et al., 2001). Propagation of holDG10 mutant cells containing this plasmid in the absence of IPTG leads to the recovery of holDG10 plasmidless cells. However, the presence of a mutation in-compatible with the holDG10 allele prevents plasmidless cells recovery (Flores et al., 2001). As pAM-holD requires PriA for its replication, for studies in priA mutants, a plasmid carrying both holD and priA wild-type genes was constructed (see Experimental procedures). Because priA mutation causes SOS induction, the viability of priA derivatives was tested in the presence of a sfiA mutation, to ascertain that any decrease in viability resulted from a replication restart defect and not from the inhibition of cell division by the SOS-induced SfiA protein (Nurse et al., 1991; Mukherjee et al., 1998).

The priA2::kan null mutation was introduced into the holDG10 cells carrying pAM-holD-priA and into wild-type cells used as a control. Cells were propagated for 8 h in the absence of IPTG and plated (i) on minimal medium (MM) plates to count the number of total viable cells in the culture and (ii) on MM plates supplemented in IPTG and spectinomycin to count the number of plasmid containing cells. The exact proportion of plasmidless cells obtained on MM plates was determined by picking colonies on plates supplemented or devoid of IPTG and spectinomycin. No plasmidless colonies were found with mutants carrying both holDG10 and priA2::kan mutations, whereas 2–4 × 108 plasmidless colonies per ml of culture were obtained for holDG10 or priA2::kan single mutants. This suggests that the holDG10 priA sfiA strain is not viable. To measure the viability of priA sfiA and holDG10 sfiA mutants, entire colonies were re-suspended in liquid MM and the number of colony-forming units (cfu) in the suspension was measured by plating appropriate dilutions on MM. After 3 days of incubation at 37°C, priA sfiA and holDG10 sfiA colonies contained 1% and 41% cfu compared with wild-type colonies respectively (Table 1). Independent lethality due to holDG10 and priA mutations would lead to the formation of clones containing 0.4% cfu compared with wild-type clones, i.e. 106 cfu, which is detectable by our assay. We conclude that the holD priA2::kan combination of mutations is lethal.

Table 1. priA300 mutation decreases viability of the holDG10 mutant.
Straina,bRelevant genotypecfu per colonycRatio of cfu per
colony versus wt
  1. a. Two holD G10 sfiA strains, three holDG10 and two holDG10 priA300 strains were used, which differ by the nature of the marker used to co-transduce the mutations of interest, or by the nature of the sfiA mutation. Results obtained with strains carrying the same relevant mutations were similar, therefore averages are presented.

  2. b. Strains were constructed and stored in the presence of pAM-holD, therefore the names indicated in the first column of the table are those of the parental strains carrying the plasmid ( Table 5). ‘Plasmidless’ indicates that the number of cfu per colony shown here was determined from isolated colonies cured of the plasmid.

  3. c. The numbers in parentheses indicate the number of colonies tested. For all strains carrying the holD G10 mutation, only rich medium-sensitive colonies were taken into account.

JJC520 wild type 2.4 × 108± 1.5 × 108 (4)1
JJC1536 priA sfiA 2.3 × 106± 1.4 × 106 (4)0.01
Plasmidless
JJC1477/JJC1654
holD sfiA 1 × 108± 0.46 × 108 (13)0.41
JJC1505 priA300 3.2 × 108± 0.53 × 108 (4)1.3
Plasmidless
JJC1307/JJC1157
/JJC1096
holD 2.4 × 107± 1.3 × 107 (20)0.1
Plasmidless
JJC1429/JJC1430
holD priA300 4.3 × 106± 3 × 106 (9)0.018

The priA300 mutation impairs growth of the holDG10 mutant

The priA300 mutation inactivates the ATPase, translocase and helicase functions of PriA and leaves intact the primosome assembly function in vitro (Zavitz and Marians, 1992). Nevertheless, none of the activities impaired by the priA300 mutation is essential for PriA action in vivo in otherwise wild-type cells, as priA300 mutation by itself does not cause any growth defect (Sandler et al., 2001). To test whether the activity of PriA300 protein is sufficient for holDG10 strain viability, the priA300 allele was co-transduced with an adjacent KanR marker in wild-type and holD strain carrying pAM-holD. After 8 h of propagation in the absence of IPTG, about 3 × 108 plasmidless cells were obtained with the priA300 holDG10 double mutant as with holDG10 single mutants. However, holDG10 priA300 colonies grew much more slowly than holD or priA300 single mutant clones and were highly heterogeneous, suggesting the appearance of mutation that suppress the poor growth of this double mutant. Plasmidless colonies were resuspended and plated on MM to count the number of cfu per colony and on Luria–Bertanin (LB), to check the rich medium sensitivity phenotype linked to the holDG10 mutation. Only nine out of the 17 holDG10 priA300 plasmidless colonies tested were sensitive to rich medium, indicating the appearance of holD suppressor mutations in the eight other clones. The priA300 mutation decreased more than fivefold the number of cfu per colonies of the holDG10 mutant, whereas it did not affect the growth of otherwise wild-type cells (Table 1). In conclusion, the recovery of viable priA300 holDG10 clones indicates that the priA300 mutation leaves intact some PriA activity essential for the growth of the holDG10 mutant. Nevertheless, the low viability of the holDG10 priA300 clones suggests that this mutation inactivates a function of PriA that is important in the holDG10 strain, possibly the helicase function.

The viability of holD mutants is affected by inactivation of PriB but not by inactivation of PriC

In wild-type Escherichia coli, the priB or priC single mutants show no growth defect whereas priB priC double mutants are lethal, leading to a model in which PriA-dependent primosome assembly requires either PriB or PriC (Sandler, 2000). In contrast with the apparent redundancy of PriB and PriC functions in vivo, PriB plays a more important role in primosome assembly than PriC in vitro (Liu et al., 1999). To test whether PriB is required for holDG10 mutant viability, the ΔpriB302 mutation was introduced in an holDG10 [pAM-holD] strain and cells were then propagated in medium devoid of IPTG. Whereas propagation of the holDG10 strain in the absence of IPTG yielded about 2 × 108 plasmidless cells per ml in 8 h, the amount of plasmidless cells with the holDG10 priB double mutants was 10-fold lower. The level of viability of the holDG10 priB double mutant was estimated by measure of the number of cfu per colony. This number was sixfold lower for holD priB colonies than for holD single mutant (Table 2). Furthermore, holD priB colonies were highly heterogeneous, containing mainly fast growing bacteria. In total, 72% and 50% of holD priB colonies were resistant to rich medium in sfiA+ and sfiA backgrounds respectively. Therefore, the holDG10 priB double mutant rapidly acquires mutation(s) that improve growth, most of which suppress the LBS conferred by the holDG10 mutation. These results indicate that the priB mutation seriously compromises the growth of holDG10 strains, causing the rapid accumulation of suppressor mutations.

Table 2. priB inactivation affects holD mutant viability whereas priC mutation does not.
Straina,b[link]Relevant genotypecfu per colonycRatio of cfu per
colony versus wt
  • a.

    Results obtained with strains carrying the same relevant mutations were similar, therefore averages are presented.

  • b. Strains were constructed and stored in the presence of pAM-holD, therefore the names indicated in the first column of the table are those of the parental strains carrying the plasmid ( Table 5). ‘Plasmidless’ indicates that the number of cfu per colony shown here were determined from isolated colonies cured of the plasmid.

  • c. The numbers in parentheses indicate the number of colonies tested. For all strains carrying the holD G10 mutation, only rich medium-sensitive colonies were taken into account.

JJC520 wild type 2.4 × 108± 1.5 × 108 (4)1
JJC1506 priB 2.2 × 108± 0.41 × 108 (4)0.9
JJC1507 priC 2.2 × 108± 0.86 × 108 (4)0.9
Plasmidless
JJC1307/JJC1096
/JJC1157
holD 2.4 × 107± 1.3 × 107 (20)0.1
Plasmidless
JJC1477/JJC1654
holD sfiA 1 × 108± 0.46 × 108 (13)0.41
Plasmidless
JJC1481/JJC1696
/JJC1482
holD priB 3.8 × 106± 2.6 × 106 (10)0.015
Plasmidless
JJC1695/JJC1697
holD priB sfiA 1.7 × 107± 0.96 × 107 (6)0.07
Plasmidless
JJC1483/JJC1484
holD priC 3.8 × 107± 1.4 × 107 (8)0.16

holDG10 priC303::kan double mutants were readily obtained after loss of pAM-holD from a holDG10 priC303::kan [pAM-holD] strain. Measurements of the number of cfu per colony showed that the priC mutation did not affect the viability the holDG10 mutant (Table 2). As expected, all priC holDG10 clones tested were as sensitive to rich medium as single holDG10 mutants (data not shown). We conclude that in contrast with PriA and PriB, the PriC protein is not required for replication restart in the holDG10 strain.

holDG10 priB priA300 combination of mutation is lethal

As we found that growth of the holDG10 mutant is impaired in the presence of either priA300 or priB mutation, we tested the effects of combining both mutations. A sfiA priA300 priB holDG10 [pAM-holD-priA] strain was constructed (JJC1841; see Table 5). Cells were grown in the absence of IPTG to cure pAM-holD-priA. No plasmidless sfiA priA300 priB holDG10 colonies could be recovered. The integrity of the PriA protein is therefore essential in the holDG10 priB sfiA strain and conversely, the PriB protein is essential in the holDG10 priA300 sfiA mutant.

Table 5. Strains.
StrainGenotypeOrigin or construction
JC19008DM4100 priA2::kan dnaC809 sfiA::MudAplacS. Sandler
JC19257DM4100 Δ(priB)302ΔpriC303::kan dnaC809 820 sfiA::MudAplacS. Sandler
JC19231DM4100 zjf-599::Tn10 purA46S. Sandler
SS97DM4100 priA300S. Sandler
JJC520 wild type
JJC1096 holDG10 mdoB::miniTn10kan [pAM-holD]P1 co-transduction of holDG10 with mdoB::miniTn10kan in
JJC520
JJC1157 holDG10 mdoB::miniTn10 [pAM-holD]P1 co-transduction of holDG10 with mdoB::Tn10 in JJC520
JJC1307 holDG10 [pAM-holD]P1 co-transduction of holDG10 with Thr+ in JJC520 thr::Tn10
JJC1429 holDG10 priA300 pflD::miniTn10kan [pAM-holD]P1 co-transduction of priA300 with pflD::miniTn10kan in
JJC1307
JJC1430 holDG10 mdoB::Tn10 priA300 pflD::miniTn10kan [pAM-holD]P1 co-transduction of priA300 with pflD::miniTn10kan in
JJC1157
JJC1477 holDG10 mdoB::miniTn10 sfiA::MudAplac [pAM-holD-priA]Transformation by pAM-holD-priA in a JJC520 holDG10
mdoB
::Tn10 sfiA::MudAplac strain
JJC1481 holDG10Δ(priB)302 zjf599::Tn10 [pAM-holD]P1 co-transduction of Δ(priB)302 with zjf599::Tn10 in
JJC1307
JJC1482 holDG10 mdoB::miniTn10KanΔ(priB)302 zjf599::Tn10 [pAM-holD]P1 co-transduction of Δ(priB)302 with zjf599::Tn10 in
JJC1096
JJC1483 holDG10ΔpriC303::kan [pAM-holD]P1 transduction of ΔpriC303::kan in JJC1307
JJC1484 holDG10 mdoB::Tn10ΔpriC303::kan [pAM-holD]P1 transduction of ΔpriC303::kan in JJC1157
JJC1505 priA300 pflD::miniTn10kanP1 co-transduction of priA300 with pflD::miniTn10kan in
JJC520
JJC1506Δ(priB)302 zjf599::Tn10P1 co-transduction of Δ(priB)302 with zjf599::Tn10 in JJC520
JJC1507ΔpriC303::kanP1 transduction of ΔpriC303::kan in JJC520
JJC1508 holDG10 mdoB::miniTn10 sfiA::MudAplac priA2::kan
[pAM-holD-priA]
P1 transduction of priA2::kan in JJC1477
JJC1534 holDG10 sfiA::MudAplac dnaC809 [pAM-holD]Transformation of pAM-holD in a JJC520 holDG10
sfiA
::MudAplac dnaC809 strain
JJC1535 holDG10 sfiA::MudAplac dnaC809 [pAM-holD-priA]Transformation of pAM-holD-priA in a JJC520 holDG10
sfiA
::MudAplac dnaC809 strain
JJC1536 priA2::kan sfiA::MudAplacP1 transduction of priA2::kan in JJC520 sfiA::MudAplac
JJC1649 holDG10 dnaC809 820 [pAM-holD]Transformation of pAM-holD in JJC520 holDG10 dnaC809 820
JJC1650 holDG10 dnaC809 820 [pAM-holD-priA]Transformation of pAM-holD-priA in JJC520 holDG10
dnaC809 820
JJC1651 holDG10 sfiA::MudAplac dnaC809 priA2::kan [pAM-holD-priA]P1 transduction of priA2::kan in JJC1535
JJC1652 holDG10 dnaC809 820 priA2::kan [pAM-holD-priA]P1 transduction of priA2::kan in JJC1650
JJC1654 holDG10 sfiA::MudAplac [pAM-holD]Transformation of pAM-holD in holDG10 sfiA::MudAplac
JJC1675 holDG10 sfiA::MudAplac dnaC809Δ(priB)302
zjf599
::Tn10 [pAM-holD]
P1 co-transduction of Δ(priB)302 with zjf599::Tn10
in JJC1534
JJC1688 holDG10 dnaC809 820Δ(priB)302 zjf599::Tn10 [pAM-holD]P1 co-transduction of Δ(priB)302 with zjf599::Tn10 in JJC1649
JJC1693 holDG10 sfiA::MudAplac dnaC809Δ(priB)302 [pAM-holD]P1 co-transduction of Δ(priB)302 in JJC1534 purA46
JJC1694 holDG10 dnaC809 820Δ(priB)302 [pAM-holD]P1 co-transduction of Δ(priB)302 in JJC1649 purA46
JJC1695 holDG10 sfiA::MudAplacΔ(priB)302 [pAM-holD]P1 co-transduction of Δ(priB)302 in JJC1654 purA46
JJC1696 holDG10Δ(priB)302 [pAM-holD]P1 co-transduction of Δ(priB)302 in JJC1307 purA46
JJC1697 holDG10Δ(priB)302 zjf599::Tn10 sfiA::kan [pAM-holD]P1 transduction of sfiA::kan in JJC1481
JJC1816 priA2::kan sfiA::MudAplac dnaC809P1 transduction of priA2::kan in JJC520 dnaC809 sfiA::MudA
JJC1817 priA2::kan dnaC809 820P1 transduction of priA2::kan in JJC520 dnaC809 820
JJC1841 holDG10 sfiA::MudAplacΔ(priB)302 zjf599::Tn10P1 co-transduction of Δ(priB)302 with zjf599::Tn10 in JJC520
 priA300 pflD::miniTn10kan [pAM-holD-priA] holDG10 sfiA::MudAplac priA300 pflD::miniTn10kan
[pAM-holD-priA]

The viability of holDG10 priA and holDG10 priB double mutants is partly restored by dnaC809 mutation

dnaC809 was isolated as a mutation suppressing the growth defect of a priA null mutant (Sandler et al., 1996). We tested the effects of dnaC809 mutation in the holDG10 priA sfiA mutant. Strains constructed in the presence of the conditional plasmid pAM-holD-priA or pAM-holD yielded similar results. About 3 × 108holDG10 priA dnaC809 sfiA mutant colonies were recovered after 8 h of culture in the absence of IPTG. The level of viability of this mutant was determined by comparison of the amount of cfu per colony with control strains. dnaC809 mutation restores the viability of priA sfiA mutant to a wild-type level (Table 3). Surprisingly, it decreased threefold the viability of the holDG10 sfiA mutant (Table 3). The viability of the holDG10 priA sfiA strain was only partly restored by the dnaC809 mutation, as holDG10 priA dnaC809 sfiA colonies contained only 8% of viable cells compared with holDG10 dnaC809 sfiA mutants (Table 3). Furthermore, one third of the colonies were resistant to rich medium, indicating a strong selection pressure for suppressor mutations. These results indicate that the dnaC809 allele restores only partial viability to the holDG10 priA sfiA strain.

Table 3. holD priA viability is partly restored by dnaC809 mutation and fully restored by dnaC809 820 mutations.
Straina,b[link]Genotypecfu per colonycRatio of cfu per colony
versus holD sfiA or holD
JJC1536 priA sfiA 2.3 × 106± 1.4 × 106 (4)
  • a.

    Results obtained with strains carrying the same relevant mutations were similar, therefore averages are presented.

  • b. Strains were constructed and stored in the presence of pAM-holD, therefore the names indicated in the first column of the table are those of the parental strains carrying the plasmid ( Table 5). ‘Plasmidless’ indicates that the number of cfu per colony shown here were determined from isolated colonies cured of the plasmid.

  • c. The numbers in parentheses indicate the number of colonies tested. For all strains carrying the holD G10 mutation, only rich medium-sensitive colonies were taken into account.

JJC1816 priA sfiA dnaC809 1.8 × 108± 4.2 × 107 (2)
Plasmidless
JJC1477/JJC1654
holD sfiA 1 × 108± 0.46 × 108 (13)1
Plasmidless
JJC1534/JJC1535
holD sfiA dnaC809 3.3 × 107± 1.4 × 107 (17)0.33
Plasmidless
JJC1651
holD sfiA dnaC809 priA 2.6 × 106± 1.3 × 106 (18)0.026
Plasmidless
JJC1675/JJC1693
holD sfiA dnaC809 priB 2 × 107± 1.8 × 107 (18)0.2
JJC1817 priA dnaC809 820 2.6 × 108± 1.1 × 108 (2)
Plasmidless
JJC1307/JJC1096
/JJC1157
holD 2.4 × 107± 1.3 × 107 (20)1
Plasmidless
JJC1649/JJC1650
holD dnaC809 820 1.3 × 108± 0.33 × 108 (14)5.4
Plasmidless
JJC1652
holD dnaC809 820 priA 8.2 × 107± 4.5 × 107 (32)3.4
Plasmidless
JJC1694/JJC1688
holD dnaC809 820 priB 1.1 × 108± 1 × 108 (8)4.5

holDG10 priB dnaC809 [pAM-holD] cells were constructed in a sfiA background. About 3 × 108 plasmidless colonies were obtained after 8 h of growth in the absence of IPTG. The presence of the dnaC809 allele did not modify the viability of the holDG10 priB double mutants, although the threefold decrease in viability associated with the dnaC809 mutation in the holDG10 single mutant was not observed in the holDG10 priB background [compare holD priB sfiA (Table 2) and holD priB sfiA dnaC809 (Table 3)]. One third of holDG10 priB sfiA dnaC809 clones were resistant to rich medium, indicat-ing a selection pressure for suppressor mutations. This experiment indicates that the dnaC809 allele only partly restores the viability of the holDG10 priB double mutant.

The viability of holD priA and holD priB double mutants is fully restored by the dnaC809 820 allele

The effect of the dnaC809 820 double mutation was tested in the holDG10 priA double mutant. A holDG10 priA dnaC809 820 strain was constructed in the presence of pAM-holD-priA or in the presence of pAM-holD, yielding similar results. About 3 × 108 plasmidless clones were obtained after 8 h of growth in the absence of IPTG. Counting of cfu in plasmidless colonies showed that the dnaC809 820 allele increased fivefold the viability of the holDG10 single mutant and fully restored the viability of holDG10 priA double mutant (Table 3). All clones tested were sensitive to rich medium, indicating no selection pressure for suppressor mutations. These results indicate that the DnaC809 820 double mutant protein allows wild-type levels of replication restart in the holDG10 priA mutant.

holDG10 priB dnaC809 820 clones were readily obtained by segregation of pAM-holD. Measures of cfu per colony showed that this allele fully restores the viability of the holDG10 priB double mutant (Table 3). All clones tested were as sensitive to rich medium as holDG10 single mutants, indicating no selection pressure for suppressor mutation (data not shown). Therefore, the dnaC809 820 allele fully suppresses the growth defect of the holDG10 priB double mutant.

The holDG10 mutation causes a requirement for RecBCD for growth, an observation which, in addition to other holDG10 properties, leads to the proposal of replication fork reversal occurrence in this mutant (Fig. 1; Flores et al., 2001). To test whether the holDG10 dnaC809 820 still requires the resetting of reversed forks by RecBC, a recB null mutation was introduced into holDG10 dnaC809 820 [pAM-holD] cells. No plasmidless clones could be ob-tained after propagation in the absence of IPTG, indi-cating that, as the holDG10 single mutant, the holDG10 dnaC809 820 mutant requires RecBC for viability. The observation that the dnaC809 820 mutation does not suppress the need for resetting of reversed fork in the holD strain suggests that the DnaC809 820 protein does not prevent replication fork reversal.

Discussion

In the holDG10 mutant, the frequency of replication fork reversal upon replication arrest is high enough to render RecBC essential for viability (Flores et al., 2001). This observation suggests frequent replication arrest, reversal and restart. Indeed, we find here that the key protein for pre-primosome assembly, PriA, is essential for holDG10 viability, which clearly indicates a requirement for re-assembly of disassembled replisomes. This allowed us to characterize the pre-primosome proteins needed for replication restart in a strain that suffers replication arrest at a high frequency.

PriB play an important role in holDG10 mutant

Three types of evidence indicate that PriB is important for holDG10 viability: (i) in priB holDG10 mutant, the rate of loss of the complementing plasmid pAM-holD is 10-fold lower than in holDG10 single mutants; (ii) priB holDG10 colonies contain 10-fold less viable bacteria than holDG10 clones; and (iii) priB holDG10 colonies often contain bacteria that have acquired a mutation(s) suppressing the rich medium sensitivity due to the holDG10 mutation. These results indicate that the PriA PriC pathway of primosome assembly, which is the active one in priB-defective cells (Sandler, 2000; Table 4), does not catalyse efficient primosome assembly in the holDG10 mutant. In contrast, the holDG10 priC double mutant does not exhibit any growth defect compared with the holDG10 single mutant, indicating that the PriA PriB pathway, which is functional in priC mutants (Sandler, 2000; Table 4), catalyses efficient primosome assembly. Similarly, the recombination-dependent restart reaction reconstituted in vitro required PriB, whereas adding PriC had little effect (Liu and Marians, 1999; Liu et al., 1999). Our observation is also reminiscent of the properties of the priA300 mutant, in which only priB inactivation affects viability (Sandler et al., 2001). However, although PriA PriB is the only pathway that allows efficient primosome assembly in both priA300 and holD mutants, this may be for different reasons as priA300 affects primosome assembly whereas holD affects polymerase progression. We propose that the PriA PriC pathway is not sufficient for full viability in the holD mutant because of the high frequency of replication arrest, hence the high requirement for replication restart. Such a model implies that the two PriA dependent pathways, PriA PriB and PriA PriC, are not equivalent, PriA PriB being more efficient. We suggest that PriC would only be able to replace PriB without causing a growth defect in wild-type cells, because of the relatively low frequency of replication arrest (Maisnier-Patin et al., 2001).

Table 4. Pathways for replication restart.
StrainPathwayViability in wild-type
backgrounda
Viability in holDG10
background
 PriA-dependent pathways  
wt PriA-PriB-DnaT-DnaC++
priC PriA-PriB-DnaT-DnaC++
priB PriA-PriC-DnaT-DnaC+±
priA300 PriA300-PriB-DnaT-DnaC+±
priA300 priB PriA300-PriC-DnaT-DnaC± (UVS)
 PriA-independent pathways  
priA PriC-DnaT-DnaC± (UVS)
priA dnaC809 PriC-(DnaT)-DnaC809+±
priA dnaC809 820 (DnaT)-DnaC809 820++
a. Phenotypes in wild-type backgrounds are from Sandler ( 2000) and Sandler and colleagues ( Sandler et al. 1999; 2001), and were also observed in this work.

The priA300 mutation impairs holDG10 mutant growth

In vitro, priA300 mutation eliminates ATPase, helicase and translocase activities of PriA without affecting its ability to assemble protein complexes on appropriate DNA substrates (Zavitz and Marians, 1992; Liu et al., 1999). In vivo, the main phenotype of the priA300 mutant is observed with the bacteriophage Mu. Primosome assembly is essential for Mu growth and the helicase function of PriA is crucial for primosome assembly at Mu transposition intermediate. Indeed, the ‘Mu-fork’ lacks single-stranded DNA on the lagging strand template, a potential problem for the loading of DnaB and the reason why priA300 mutation impairs Mu growth (Jones and Nakai, 1999; Jones and Nakai, 2000). We observed here that the priA300 mutation impairs growth of the holDG10 strain. priA300 single mutants have no deleterious phenotype whereas holDG10 priA300 double mutants grow slowly and accumulate suppressor mutations. Two interpretations can be proposed for this observation: (i) either the helicase function of PriA is important for restart in the holDG10 mutant, suggesting that the substrate for primosome assembly often lacks a single-stranded region on the lagging strand at the fork large enough for DnaB loading or (ii) the priA300 mutation impairs yet another function of PriA, in addition to the helicase activity.

The helicase function of PriA facilitates replication restart by unwinding the 5′-end of the lagging strand to expose a single-strand region for DnaB loading (Jones and Nakai, 2001). This function is therefore presumably more important for restart from an Y-structure than from a recombination intermediate. According to the replication fork reversal model, a Y-structure is formed when the reversed fork is degraded by exonuclease V action of RecBCD (Fig. 1B), whereas a recombination intermediate is formed when the reversed fork is reintegrated into the chromosome by homologous recombination (Fig. 1A). Degradation is the only viable pathway in a holD recA mutant, whereas homologous recombination is essential in a holD recD mutant. In the holD priA300 double mutant, we did not observe any significant effects of recA or recD mutations (data not shown), suggesting that the decreased viability of the holDG10 strain combined with the priA300 mutation does not depend on the substrate generated at blocked forks. This observation suggests that the priA300 mutation may not affect only the helicase function of PriA. Similarly, the recA and recD mutations did not modify significantly the low via-bility of holDG10 priB double mutants (data not shown), suggesting that the growth defect of the strain is not correlated with the binding of pre-primosome proteins to a specific substrate.

priA300 and priB mutations are synergistic in the holDG10 mutant

The priA300 mutation was described to decrease the viability of priB single mutants, as combination of these two mutations in wild-type cells induces the SOS response, causes cell filamentation and decreases plating efficiency (Sandler et al., 2001). We found that priA300 priB combination of mutations is lethal in the holDG10 mutant, indicating that efficient replication restart in the absence of PriB requires the integrity of the PriA protein. These observations lead to the conclusions summarized on Table 4: (i) the main pathway for pre-primosome as-sembly involves the successive binding of PriA and PriB, which accounts for the lack of phenotype of priC mutants regardless of the need for replication restart, and for the growth defect caused by PriB inactivation when replication restart is frequent and (ii) the priA300 mutation affects replication restart, however, the defect caused by the priA300 mutation is only detectable when the need for replication restart is high (holDG10 priA300 mutant) or when only the weak PriA PriC pathway is available (priB priA300 mutant). When replication arrests are frequent and the main PriA PriB pathway is inactivated, the priA300 mutation is lethal (holDG10 priB priA300 mutant).

PriA-independent pathways

Three PriA-independent pathways have been described (Sandler, 2000; Table 4, last three lines). The PriC-dependent pathway is poorly efficient in wild-type backgrounds. Its efficiency is increased by the dnaC809 mutation. The double mutant DnaC809 820 protein allows replication restart in the absence of all PriA, PriB, and PriC proteins. The PriC-dependent pathway could not support growth of the holDG10 priA double mutant. The viability of the holD priA double mutant is partly restored by the dnaC809 allele, indicating that the PriC DnaC809 pathway is sufficient for the viability of the holD mutant, even if it is not as efficient as the PriA pathway. Replication fork reversal was also shown to occur in the rep mutant, in which PriA is also essential for viability (Seigneur et al., 1998; Sandler, 2000). However, rep and holD mutants differ in respect with the dnaC809 mutation, which suppresses the lethality of holD priA strains (Table 3), whereas it does not suppress the lethality of rep priA double mutant (Sandler, 2000). This difference supports the view that the Rep protein may play a role in primosome assembly in addition to its role during replication fork progression (Sandler, 2000). Interestingly, dnaC809 single mutation slightly decreased the viability of the holDG10 sfiA mutant, suggesting that DnaC809 protein may compete with PriA and decrease the efficiency of replication restart in PriA+ strains. Such a hypothesis may explain why the dnaC809 allele did not suppress entirely the growth defect of priB holD mutants. The DnaC809 820 protein is the only way to bypass entirely the need for PriA or PriB in holDG10 mutant. This protein also increases the viability of the holDG10 single mutant, suggesting that part of the low viability of the holDG10 mutant is caused by a replication restart defect and that the DnaC809 820 protein provides the more efficient replication restart pathway. Nevertheless, the lethality of holDG10 dnaC809 820 recB triple mutant indicates that replication fork reversal still occurs in the presence of the DnaC809 820 protein. This observation suggests that either the activity that reverses forks has more affinity for blocked forks than the DnaC809 820 protein or DnaC809 820 is prevented from acting directly on inactivated forks in the holDG10 mutant and can only gain access to its target after resetting of reversed forks by RecBC.

Experimental procedures

Strain and plasmid constructions

All strains were constructed by P1 transduction. Strains used in this work are listed on Table 5. The holDG10 mutation is a change from Gln to amber mutation at position 10. It was isolated in a background that carries an uncharacterized amber suppressor and, consequently, the nature of the mutant HolD polypeptide synthesized remained unknown (Flores et al., 2001). The presence of the holDG10 mutation was verified by measuring sensitivity to rich medium. The presence of priA2::kan, Δ(priB)302, ΔpriC303::kan and sfiA mutations was verified by polymerase chain reaction (PCR). The presence of the dnaC809 and priA300 mutations were verified by digestion of PCR products with Hinf1 and BsiWI respectively (Sandler, 2000). In addition, the dnaC809 820 alleles were verified by sequencing of the dnaC gene in JJC1649 and JJC1650. The zjf-519::Tn10 insertion, used to co-transduce Δ(priB)302 in certain strains, was localized in the hypothetical yjfO gene of unknown function by sequencing. To construct pAM-holD-priA, an EcoR1 fragment carrying holD from pAM-holD was cloned in the EcoR1 site of a pAM34 derivative carrying the priA gene.

Segregation experiments

Segregation experiments were performed at 37°C. Cells containing pAM-holD or pAM-holD-priA were grown overnight in minimal medium (MM) in the presence of 50 μM IPTG and 60 μg ml−1 of spectinomycin. Cells were diluted 1000-fold in MM and grown for 8 h. Appropriate dilutions of these 8 h cultures were plated on MM plates to count the total amount of viable cells in the cultures, and on MM plates supple-mented with 500 μM IPTG and 60 μg ml−1 of spectinomycin to count the number of plasmid-containing clones. The proportion of plasmidless clones on MM plates was determined by picking colonies on 500 μM-IPTG/60 μg ml−1-spectinomycin plates. In addition, plasmidless clones were generally easily recognized due to their smaller size. For Δ(priB)302 derivative, probably due to an effect of the Δ(priB)302 in-frame deletion on the expression of the ribosomal genes present in the same operon, plasmidless clones exhibited a low level of resistance to spectinomycin and ampicillin, which prevented their identification by picking colonies on antibiotic-containing plates. Plasmidless clones were therefore identified by PCR using oligonucleotides specific for the pAM-holD plasmid. Loss of the plasmid led to no PCR amplification product.

Test of colony viability and rich medium sensitivity

Entire colonies were picked on MM plates after 3 days of incubation at 37°C, inoculated in 1 ml of MM devoid of glucose, and shaken for 1 h at 37°C for full suspension of the colony. Appropriate dilutions of these suspensions were plated on MM plates to count cfu and on Luria–Bertani (LB) plates to measure sensitivity to rich medium. The genotype of the re-suspended colonies and the absence of plasmid were verified by PCR and antibiotic resistance before and after plating.

Acknowledgements

We thank Dr Steve Sandler for communicating results before publication and for sending strains. We thank Drs Santanu Das Gupta and Sophie Maisnier-Patin for communicating results before publication. We are very grateful to Vladimir Bidnenko and Etienne Dervyn for careful reading of the manuscript and to Céline Costa for excellent technical assistance. B.M. is on the CNRS staff. This work is supported in part by the Programme de Recherche Fondamentale en Micro-biologie, Maladies Infectieuses et Parasitaire.

Ancillary