|Home | About | Journals | Submit | Contact Us | Français|
We observed that cells lacking Rep and UvrD, two replication accessory helicases, and the recombination protein RecF are cryo-sensitive on rich medium. We isolated five mutations that suppress this Luria–Bertani (LB)-cryo-sensitivity and show that they map in the genes encoding the RNA polymerase subunits RpoB and RpoC. These rpoB (D444G, H447R and N518D) and rpoC mutants (H113R and P451L) were characterized. rpoBH447R and rpoBD444G prevent activation of the Prrn core promoter in rich medium, but only rpoBH447R also suppresses the auxotrophy of a relA spoT mutant (stringent-like phenotype). rpoCH113R suppresses the thermo-sensitivity of a greA greB mutant, suggesting that it destabilizes stalled elongation complexes. All mutations but rpoCP451L prevent R-loop formation. We propose that these rpo mutations allow replication in the absence of Rep and UvrD by destabilizing RNA Pol upon replication–transcription collisions. In a RecF+ context, they improve growth of rep uvrD cells only if DinG is present, supporting the hypothesis that Rep, UvrD and DinG facilitate progression of the replication fork across transcribed sequences. They rescue rep uvrD dinG recF cells, indicating that in a recF mutant replication forks arrested by unstable transcription complexes can restart without any of the three known replication accessory helicases Rep, UvrD and DinG.
Replication forks are susceptible to be arrested by a variety of obstacles, including DNA-bound proteins such as RNA polymerases (RNA Pol) (Mirkin and Mirkin, 2007; Rudolph et al., 2007). DNA instability is associated with replication fork arrest (reviewed in Aguilera and Gomez-Gonzalez, 2008), and in order to limit the deleterious consequences of replication–transcription collisions, cells encode enzymes that facilitate replication through transcription units. In yeast, the Rrm3 helicase travels with the replication fork machinery and dislodges RNA Pols from tRNA and rRNA genes as well as tightly bound proteins from heterochromatin (Azvolinsky et al., 2009 and references therein). In Escherichia coli, this function was originally ascribed to the Rep helicase; first, based on the observations that in the rep mutant chromosome replication is slowed down and requires specific functions (Lane and Denhardt, 1975; Seigneur et al., 1998; Petit and Ehrlich, 2002), and second, because the purified Rep helicase is specifically capable of dislodging DNA-bound proteins (Yancey-Wrona and Matson, 1992). More recently, evidence was provided that transcribed sequences are indeed a major obstacle to replication in rapidly growing cells, and that Rep is the main, although not the only, helicase that assists replication progression across highly transcribed sequences (Guy et al., 2009; Boubakri et al., 2010).
Rep and UvrD are two 3′ to 5′ helicases that share 40% homology. The existence of a redundant function for these paralogues has been suspected since the original observation that the rep uvrD double mutant shows severe growth defects (Taucher-Scholtz et al., 1983). However, the situation turned out to be more complex and UvrD is now known to play two different roles in rep mutants, revealed by the two classes of mutations that suppress the growth defects of the rep uvrD double mutant. The first class of mutations that was identified inactivates the RecFOR recombination pathway (Petit and Ehrlich, 2002). Combined with the observation that UvrD can remove RecA from ssDNA in vitro, the rescue of rep uvrD cells by the inactivation of recombination proteins led to the proposal that UvrD is essential in rep mutants because it removes deleterious RecFOR-dependent RecA-filaments that assemble at blocked replication forks (Veaute et al., 2005; Lestini and Michel, 2008). However, the nature of the obstacles arresting replication in the first place, thus allowing RecFOR-RecA binding to forks in rep mutants, remained unknown. Moreover, the suppression of rep uvrD co-lethality by recFOR inactivation was shown to be only partial, confirming the existence of an obstacle to replication restart other than RecFOR-RecA bound to DNA in rep uvrD cells (Guy et al., 2009; Boubakri et al., 2010). Two lines of evidence identified RNA polymerases as the original cause of replication arrest in rep uvrD cells: first, mutations that map in the RNA Pol genes rpoB and rpoC were shown to suppress the growth defects of this double mutant (Guy et al., 2009; Boubakri et al., 2010); second, replication arrest sites were directly visualized in rep and rep uvrD recF mutants at inverted ribosomal operons (rrn), provided that these operons were facing replication and were highly transcribed (Boubakri et al., 2010). Rep was confirmed to be the major accessory helicase and was shown to be attracted to replication forks by a direct interaction with the replicative helicase DnaB (Guy et al., 2009). In addition to UvrD, a third player was identified, the 5′ to 3′ helicase DinG (Boubakri et al., 2010). DinG is essential for the viability of the rep uvrD recF mutant and it was proposed that, provided that recF is inactivated, DinG can clear RNA Pols from blocked replication forks in the absence of both Rep and UvrD. In addition, DinG is essential for the viability of rep and uvrD single mutants when replication collides with RNA Pol at highly expressed inverted rrn, indicating that the presence of two of these three helicases is required when replication–transcription collisions are increased. Finally, DinG has the specific function of unwinding RNA–DNA hybrids in vivo, as it is also essential for viability when replication forks are arrested by R-loops (Boubakri et al., 2010).
The described RNA Pol mutations that suppress rep uvrD growth defects are mutations that constitutively confer phenotypes akin to the induction of the stringent response (Guy et al., 2009; Boubakri et al., 2010). The stringent response is an adaptation to amino acid starvation through the induction of the alarmone ppGpp (reviewed in Potrykus and Cashel, 2008). Binding of ppGpp to RNA Pol, as well as mutations that mimic this binding, affect transcription initiation from specific promoters, which decreases the expression of ribosomal operons (rrn) and activates the expression of amino acids biosynthetic genes in relA spoT double mutants, allowing their growth in minimal medium without amino acids (Bartlett et al., 1998; 2000; Zhou and Jin, 1998; Barker et al., 2001a,b;). The so-called ‘stringent mutations’ in RNA Pol mimic the presence of ppGpp in decreasing the half-life of open complexes. Although we have not measured the half-life of transcription open complexes, we will call here ‘stringent-like’ the phenotype conferred by RNA Pol mutations that both decrease rrn expression in rich medium and allow growth of relA spoT double mutants on minimal medium. RNA Pol mutations that increase transcription of amino acid biosynthetic promoters also destabilize transcription elongation complexes (TEC) (Trautinger and Lloyd, 2002; Trautinger et al., 2005). Such mutations were thought to suppress the lethality of rep uvrD double mutants owing to the destabilization of RNA Pol–DNA complexes (Guy et al., 2009; Boubakri et al., 2010).
In this work, we report that rep uvrD recF cells grow poorly on rich medium at low temperature and we isolated five mutations suppressing this LB-cryo-sensitivity. Similarly to the mutations that suppress the growth defects of rep uvrD mutants at 37°C, the suppressor mutations isolated here in rep uvrD recF mutants at 30°C map in rpoB and rpoC. One of these mutations is close to the active site; the others are in, or close to, the primary DNA–RNA binding channel, suggesting that they affect the stability of transcription complexes on DNA. Only one of these mutations exhibits a stringent-like phenotype, showing that our assay provides a new way of isolating RNA Pol mutants that are weakly bound to DNA, in transcription initiation or elongation complexes. Furthermore, these RNA Pol mutants allow us to extend our study of helicases that assist replication progression across transcription obstacles.
We constructed rep uvrD recF triple mutants in the presence of a conditional plasmid that carries the rep wild-type gene (pAM-rep; Lestini and Michel, 2008). This plasmid only replicates in the presence of the Lac promoter inducer, IPTG and is cured upon cell propagation in the absence of IPTG. We analysed the properties of rep uvrD recF cells cured of pAM-rep. For historical reasons the work was realized part in an AB1157 background (classically used for homologous recombination studies) and part in an MG1655 background (the more generally used, sequenced wild-type strain). In the AB1157 background, colony formation was delayed on LB at 37°C and 30°C; at 25°C only a variable subpopulation of cells formed colonies (JJC4048, Table 1). In the MG1655 background, results were similar except for a partial defect of plating efficiency on LB at 30°C (JJC5136 and JJC5166 Table 1). Finally, we constructed an Hfr strain, which allowed the co-introduction of the three rep uvrD recF mutations by conjugation, and observed that in this Hfr-PK3-PO131 background the growth defect was more pronounced than in other backgrounds, as plasmid-less rep uvrD recF cells were not recovered with the expected efficiency even on minimal medium (MM) at 37°C (not shown). In order to understand the reasons for the cryo-sensitivity of rep uvrD recF mutants, we studied five AB1157 rep uvrD recF suppressed clones that are able to form large colonies on LB at 30°C in 2 days. The rpoCΔ215–220 mutation, previously shown to restore the viability of rep uvrD cells at 37°C (Boubakri et al., 2010), and a suppressor mutation identified in the Hfr rep uvrD recF context at 37°C were included in this study.
Analysis of the subpopulation of Hfr plasmid-less colonies formed in 2 days at 37°C on MM supplemented with casamino acids revealed that one of them was resistant to rifampicin. RifR mutations map in rpoB, the gene coding for the β subunit of RNA Pol, which suggested that this particular rep uvrD recF clone carries an rpoB mutation that suppresses the rep uvrD recF growth defect (JJC4100, Table S1). To ascertain that the RifR mutation was responsible for the improved growth of the rep uvrD recF RifR clone, pAM-rep was reintroduced and the suppressed clone was P1 transduced with thiC::Tn10, a locus close to the rpoB rpoC genes. As expected, the Rif resistance phenotype was 90% linked with thiC::Tn10 (43/48 TetR transductants were RifS). Curing of pAM-rep showed that rep uvrD recF thiC::Tn10 clones that had remained RifR had kept the capacity to form colonies overnight at 37°C on LB whereas the thiC::Tn10 clones that had lost the RifR phenotype grew as poorly as the original Hfr rep uvrD recF mutant. This mutation is therefore necessary and sufficient for the improved viability of rep uvrD recF cells. Sequencing of the rpoB gene revealed the presence of a mutation, N518D, in the rpoB RifR cluster 1.
Overnight cultures of AB1157 rep uvrD recF cells were plated on LB at 30°C. Five colonies isolated in three independent experiments were kept for further studies (called S1, S2, and S3a, S3b, S3m). Replacement in the five suppressed strains of the rep uvrD recF region of AB1157 by that of the Hfr JJC4100, or of the rep::ApR allele by a rep::CmR allele, did not modify their growth properties (Table S1, data not shown). To determine whether the isolated mutations map in the rpoBC genes, a TetS derivative of each suppressed clones was P1 transduced with thiC::Tn10. For each mutant 7–11 out of 12 thiC::Tn10 transductants became cryo-sensitive, indicating that these five suppressor mutations are linked to thiC.
A plasmid carrying an IPTG-inducible rpoB+ gene was introduced in ApS derivatives of the suppressed clones and viability was measured in the presence of IPTG (Table S2). Expression of RpoB rendered S3m cryo-sensitive on LB, suggesting that this mutant carries a recessive rpoB mutation; sequencing rpoB showed the presence of a H447R mutation. In S3a, the control vector pUC19 could not be introduced whereas the plasmid pUC-rpoB+ transformed with a normal efficiency (not shown); nevertheless, transformants remained cryo-resistant (Table S2). Sequencing rpoB revealed the presence of a D444G mutation. These observations suggest that, in addition to suppressing the rep uvrD recF growth defect, the rpoBD444G mutation somehow prevents pUC propagation. This defect in plasmid propagation is recessive (as it is complemented in cis by the wild-type rpoB+ allele) whereas the suppression of rep uvrD recF cryo-sensitivity is dominant over the wild-type allele. In S1 and S3b expression of rpoB+ did not prevent growth at 25°C (Table S2), so we sequenced rpoC in these two mutants and found a P451L mutation (S1), and a H113R mutation (S3b). Finally, both rpoB and rpoC genes were sequenced in S2. rpoB was intact and rpoC carried the H113R mutation, which was thus obtained twice independently. To our knowledge, the three rpoBD444G, rpoCH113R and rpoCP451L mutations have not been described previously, whereas the rpoBN518D allele has been already isolated in a screen for RifR clones (Garibyan et al., 2003), and the rpoBH447R allele in a screen for mutations that increase the expression of amino acid biosynthetic genes in a relA spoT context (Trautinger and Lloyd, 2002).
To determine whether the rpo mutations isolated here (called rposup thereafter) are necessary and sufficient for the suppressor phenotype, these mutations were transferred to a rep uvrD recO strain by P1 co-transduction with thiC::Tn10[RecF is known to act in conjunction with two other proteins, RecO and RecR (Kuzminov, 1999), and inactivation of either the RecF, RecO or RecR protein allows colony formation of rep uvrD cells at 37°C on MM with casamino acids (Petit and Ehrlich, 2002)]. rep uvrD recO cells harbouring the plasmid pBGts-rep were used for strain construction; this plasmid carries the wild-type rep gene and can be cured by growing cells at 42°C (Petit and Ehrlich, 2002). The phenotype of plasmid-less cells obtained after propagation at 42°C was analysed. The five rep uvrD recO rposup mutants formed about 100% colonies on LB at 37°C, 30°C and 25°C, as the original rep uvrD recF rposup cells (Table 1). We conclude that the rposup mutations are necessary and sufficient to restore full viability to rep uvrD recF (recO) cells at low temperatures.
RecQ acts in concert with RecFOR to promote RecA binding to blocked forks in the rep uvrD mutant (Lestini and Michel, 2008). We observed that the rep uvrD recQ mutant is more sensitive to LB than rep uvrD recF (recO) cells (Table 1), suggesting that the RecFOR proteins still bind to arrested forks in a rep uvrD recQ mutant. The rposup alleles suppressed the LB-sensitivity of rep uvrD recQ cells, with the notable exception of the RifR (rpoBN518D) mutation that suppressed only at 37°C (Table 1). The residual cryo-LB-sensitivity of the rep uvrD recQ rpoBN518D indicates that RecFOR bind blocked replication forks in the absence of RecQ in this particular RNA Pol mutant, and that this mutation is a less efficient suppressor than the others at low temperature, in agreement with its original isolation at 37°C. Nevertheless, all other rposup mutations and the rpoCΔ215–220 mutation suppress the residual growth defects of rep uvrD recQ cells.
The rpoCΔ215–220 mutation alters the kinetic properties of transcription complexes, reducing rRNA transcription and increasing transcription from some amino acid biosynthetic genes (Bartlett et al., 1998). On the other hand, it behaves as most of the mutations isolated here as it suppresses the growth defects of rep uvrD recF, recO and recQ cells at all temperatures (Boubakri et al., 2010) (Table 1). In addition, one of the mutations isolated here (rpoBH447R) was also previously isolated in a screen for mutations that confer a stringent-like phenotype (Trautinger and Lloyd, 2002). These observations prompted us to test whether the rposup mutations isolated here prevent the stimulation of the rrnB core promoter after a shift to rich medium. The rposup mutations were co-transduced with thiC::Tn10 into a strain carrying a P1rrnB-lacZ fusion, which was used to compare the expression of the P1rrnB core promoter in MM and in LB (Bartlett et al., 1998). As expected, in wild-type cells the expression of lacZ from P1rrnB promoter was higher in LB than in MM, whereas P1rrnB activity remained low in both media in the presence of the rpoCΔ215–220 or rpoBH447R mutations (Fig. 1). This experiment revealed that rpoBD444G also reduces rRNA expression in LB (Fig. 1). In contrast, P1rrnB expression remained higher in LB than in minimal medium in the presence of the rpoBN518D, rpoCP451L and rpoCH113R alleles as in wild-type cells, showing that none of these three mutations affects the activity of rRNA promoter.
RelA and SpoT proteins are implicated in ppGpp alarmone synthesis. relA spoT double mutants do not induce the stringent response and are thus unable to grow on MM unless amino acids are provided. RNA Pol mutations that mimic the stringent response are classically isolated as suppressing the auxotrophy of relA spoT mutants. Although this has not been actually demonstrated, it was suggested that the inability of relA spoT mutants to grow on MM and the ability of the rpoBC suppressor mutations to suppress this defect results from effects of the mutations on transcription of amino acid biosynthetic operons (Paul et al., 2005; see Rutherford et al., 2009, for discussion). In order to analyse the capacity of the rposup mutations to suppress the auxotrophy of relA spoT double mutants, we constructed rposupΔrelA::KanR mutants (strains JJC4553 to JJC4559 Table S1) and P1-transduced them with a ΔspoT::CmR null mutation, plating half of the transduction mixture on LB and half on MM. All rposupΔrelA::KanR mutants could be transduced by the ΔspoT::CmR null mutation on LB while only one mutant, rpoBH447RΔrelA::KanR, provided transductants on MM (Table S3). As expected from this result, none of the rposupΔrelA::KanRΔspoT::CmR mutants obtained on LB could grow on MM except for the rpoBH447RΔrelA::KanRΔspoT::CmR strain (Table S3). The growth defect on MM was specific for a relA spoT context, because none of the RNA Pol mutations prevented growth on MM in RelA+ SpoT+ cells (see plating efficiencies on MM in Figs 3 and and4).4). This result shows that only the rpoBH447R mutation prevents the increase of rrn expression in rich medium and allows growth of a relA spoT mutant on MM. The rpoBD444G mutation only affects rrn expression and the other rposup mutations exhibit none of these stringent-like phenotypes. This result shows that four of the five mutations isolated in this study would not be obtained in a classical screen for mutations that restore growth of a relA spoT mutant on MM. Although they do not confer a stringent-like phenotype, the rposup mutations bypass the need for accessory replication helicases, suggesting that they affect the stability of transcription complexes without affecting their kinetic properties on rrn and amino acid biosynthetic gene promoters. The possible instability of the mutant RNA Pol–DNA complexes was tested by two different genetic approaches.
RuvABC is a recombination complex that acts at the last step of homologous recombination by resolving recombination intermediates called Holliday junctions (Kuzminov, 1999). ruv mutants are hypersensitive to UV irradiation and rpoB or rpoC mutations that exhibit a stringent-like phenotype partially relieve this hypersensitivity. It was proposed that they increase the intrinsic instability of RNA Pol–DNA complexes when RNA Pol is blocked by a DNA lesion (Trautinger and Lloyd, 2002). A ruvABC deletion was introduced in all rposup mutants and the UV resistance of the resulting strains was measured. With the exception of the rpoCH113R allele, which was poorly viable in a ruv mutant context and yielded variable results, all mutations improved the UV resistance of the ruvABC mutant (Fig. 2A). This result supports the idea that the mutations isolated here affect the stability of RNA Pol–DNA complexes, either at promoters or in TEC.
When RNA Pol encounters a block during elongation and backtracks, the transcription factors GreA and GreB suppress pausing by stimulating the intrinsic nucleolytic activity of RNA Pol (reviewed in Borukhov et al., 2005). greA greB double mutants are non-viable at high temperature, presumably because prolonged RNA polymerase pausing prevents replication and/or transcription. A mutation isolated through its stringent-like phenotype was previously reported to suppress the thermo-sensitivity of greA greB mutants (Trautinger and Lloyd, 2002). greA greB rposup mutants were constructed and tested for growth at 42°C. Only the rpoCH113R mutation allowed the growth of greA greB mutant at 42°C (Fig. 2B). This result presumably reflects the destabilization of backtracked RNA Pol by the rpoCH113R, but this level or type of destabilization is not essential for the growth of rep uvrD recF cells at 30°C as it is observed for only one of the suppressor mutations. This result also indicates that mutations that do not affect rrn or amino acid biosynthetic gene expression can nevertheless decrease the stability of stalled TEC enough to rescue a greA greB mutant. Altogether, we infer from the phenotypes conferred by the RNA Pol mutations described here that these mutations compromise the stability of transcription complexes; the consequences of this putative destabilization were further investigated in backgrounds that lack different accessory helicases.
If the viability of rep uvrD recF rposup mutants results from less replication arrest, or from facilitated replication restart, the inactivation of recF might not be needed for viability. Actually, certain RNA Pol mutations including the rpoCΔ215–220 allele were reported to suppress the growth defect of rep uvrD cells in a RecF+ context at 37°C (Guy et al., 2009; Boubakri et al., 2010). To construct rep uvrD rposup mutants, rep and uvrD were introduced in rposup single mutants containing pGBts-rep, and/or the rposup mutation was co-transduced with thiC::Tn10 to a rep uvrD (pAM-rep) mutant. The viability of plasmid-less clones was measured after curing cells by propagation at 42°C (pGBts-rep) or in the absence of IPTG (pAM-rep) (Fig. 3A). The rpoCH113R mutation allowed formation or rep uvrD colonies on LB at all temperatures. The rpoCP451L mutant yielded slow-growing colonies that were heterogeneous in size at low temperature. Two of the mutations that affect rrn expression (rpoCΔ215–220, rpoBH447R) restored rep uvrD cells viability on LB, but colony formation was delayed at low temperature and also slightly decreased for rpoCΔ215–220 at 25°C. In the rpoBD444G context, colony formation on LB was delayed at all temperatures and significantly decreased at 25°C. Finally, the rpoBN518D mutation had only a partial effect at 37°C and did not allow colony formation at low temperatures. These observations indicate that in one rep uvrD rposup mutant (rpoCH113R) RecFOR does not bind replication forks at any temperature. In the other rep uvrD rposup mutants, although RecFOR is not lethal any more as in rep uvrD Rpo+ cells, it remains partly deleterious, slowing down and/or preventing growth, mainly at low temperature. Therefore, the modifications of RNA Pol activity caused by the different mutations determine the accessibility of replication forks to RecFOR recombination proteins.
The viability of the rep uvrD recF mutant relies on the presence of a third helicase called DinG (Boubakri et al., 2010). We previously reported that the rpoCΔ215–220 mutation suppresses the lethality of rep uvrD recF dinG mutants at 37°C on MM and on LB, but not that of rep uvrD dinG mutants, indicating that it facilitates replication restart in the absence of all three accessory helicases only if RecF does not poison arrested forks (Boubakri et al., 2010). We tested whether the rposup mutations also suppress rep uvrD recF dinG lethality by constructing rep uvrD recF dinG rposup mutants in two ways, first a dinG deletion was introduced in the original rep uvrD recF rposup clones (JJC4043 to JJC4047, context AB1157, Table S1), and second the rposup mutations were introduced in a rep uvrD recF dinG mutant (JJC5405 to JJC5426, context MG1655, Table S1). Results were similar in both backgrounds (data not shown and Fig. 3B). rpoCP451L was the only mutation that did not suppress the lethality of rep uvrD recF dinG cells as no plasmid-less cells could be obtained (not shown). The best suppressor mutation was again the rpoCH113R, which restored 100% rep uvrD dinG recF plating efficiency at 37°C (although colony formation was delayed) and at low temperature (Fig. 3B). The three mutations rpoCΔ215–220, rpoBH447R and rpoBD444G also restored 100% plating efficiency but colony formation was delayed on LB at 37°C for rpoBD444G, on LB at all temperatures for rpoCΔ215–220, and in all growth conditions for rpoBH447R. Finally, rpoBN518D allowed normal colony formation on MM but formed only about 10% of heterogeneous slow-growing colonies on LB at 37°C and did not suppress the rep uvrD recF dinG lethality on LB at low temperature.
It was previously reported that rpoCΔ215–220 only rescues a rep uvrD dinG mutant that lacks RecF (Boubakri et al., 2010). To address this question for the rposup mutants, rep uvrD dinG rposup RecF+ mutants were constructed, first by introducing the dinG deletion in the original rep uvrD rposup mutants previously made RecF+ (JJC5253 to JJC5257 context AB1157 and JJC5258 context JJC4100, Table S1) and second by introducing successively all three helicase deletions in the rposup mutants (JJC4911 to JJC4918, JJC5310 and JJC5311, context AB1157, Table S1). All strains yielded similar results: the rposup mutations did not suppress the lethality of rep uvrD dinG mutants in a RecF+ context, as either no plasmid-less colonies could be recovered after growth under non-permissive conditions for plasmid replication (rpoBN518D), or only a few plasmid-less colonies were obtained that could not be propagated and eventually acquired additional suppressive mutations, possibly because the uvrD context is mutator (data not shown). In conclusion, in the absence of all three helicases the rposup mutations restore cell viability only if recF is inactivated, as previously observed for the rpoCΔ215–220 mutation in a MG1655 background at 37°C (Boubakri et al., 2010). This result indicates that, at least in the absence of the three Rep, UvrD and DinG helicases, replication forks are still arrested by the encounter of the rposup mutated RNA polymerases, allowing RecFOR to gain access to DNA. This result also indicates that DinG is responsible of the viability of rep uvrD rposup DinG+ mutants shown in Fig. 3A, as the inactivation of dinG in this context is lethal.
To further test the effects of the rposup mutations on replication–transcription collisions, we introduced these mutations in cells where such collisions are increased by a chromosome rearrangement.
Inversion of an rrn operon creates a region of increased head-on collisions between replication and transcription. Such inversions render the dinG mutant sensitive to rich medium because of R-loop formation, and strongly impair growth of the rep dinG double mutant, even on MM, because of DNA Pol–RNA Pol collisions (Boubakri et al., 2010) (Fig. 4A). In a strain that carries an inverted rrnA operon (InvA) the rpoCΔ215–220 mutation suppresses the growth defects of both dinG and dinG rep mutants at 37°C, on MM and on LB, presumably by decreasing the transcription efficiency of rrnA (Boubakri et al., 2010) (Fig. 4A).
We tested whether the rposup mutations decrease the formation of R-loops in inverted rrn by introducing these mutations in InvA dinG cells and measuring plating efficiencies of the resulting combination of mutations on LB. rpoCΔ215–220, previously tested at 37°C only, was tested here at 30°C. The rpoCH113R mutation and the three mutations that affect rrn expression (rpoBH447RrpoBD444G and rpoCΔ215–220) restored full viability of the InvA dinG mutant at 37°C and 30°C (Fig. 4A). Therefore, they decrease R-loop formation in this context at both temperatures. In contrast the RifR mutation (rpoBN518D) allowed InvA dinG colony formation on LB at 37°C but not at 30°C, and the rpoCP451L mutations had no effect. The phenotype of this last mutant suggests that suppression of the rep uvrD recF LB-cryo-sensitivity by rposup mutations is independent of R-loop removal. To ascertain directly that the LB-cryo-sensitivity of the rep uvrD recF mutant does not result from R-loop formation, a plasmid that overexpresses RnaseH (pEM001) was introduced in this mutant. As shown in Table 1, this plasmid did not improve plating efficiency. In conclusion, most of the rposup mutations suppress or decrease R-loop formation in InvA cells, allowing InvA dinG viability, but conversely suppression of R-loops is neither necessary nor sufficient to improve growth of rep uvrD recF cells at 30°C on LB.
rposup mutations were also introduced in an InvA dinG rep mutant, where replication is arrested in the inverted rrnA by collisions with RNA Pol (Boubakri et al., 2010). As previously reported the mutation rpoCΔ215–220 restored a normal plating efficiency on MM and LB (Boubakri et al., 2010), and as expected the rpoBD444G mutation, which decreases Prrn efficiency (Fig. 1), exhibited a similar phenotype (Fig. 4B). Surprisingly, the rpoBH447R mutation improved colony formation only on MM. It is conceivable that the decrease of rrn expression measured in Fig. 1 with a promoter deprived of its FIS-binding sites is compensated by the presence of these sites at inverted rrnA (Bartlett et al., 2000). Finally, rpoCH113R also improved viability on MM whereas the rpoCP451L and rpoBN518D mutations had no effect. Altogether, these results indicate that in rich medium only two mutations, rpoCΔ215–220 and rpoBD444G, destabilize RNA Pol enough to allow replication across an inverted rrn in the absence of Rep and DinG, i.e. when only the UvrD helicase is active. Among the suppressor mutations that do not affect rrn expression, rpoCH113R is the only one that rescues an InvA dinG rep mutant, and only on MM.
A consequence of replication arrest in the rep mutant is a reaction called replication fork reversal, in which the two ends of the newly synthesized strands at a blocked replication fork anneal to form a DNA double-strand end and a Holliday junction. This reaction renders RecBC (the recombination enzyme specific for DNA double-strand ends) essential for viability (Seigneur et al., 1998; Michel et al., 2007). Interestingly, although this study and other recent studies suggest that in the rep mutant replication forks are arrested by transcribed sequences, none of the mutations isolated here rescued the viability of rep recB cells, as no plasmid-less colonies could be recovered after propagation of rep recB rposup (pAM-rep) mutants in the absence of IPTG (data not shown, strains JJC5532 to JJC5536 in Table S1). This indicates that the rpoB mutation does not prevent replication fork reversal in a rep mutant, and that, even when the RNA Pol is mutated and weakly bound to DNA, its removal from DNA by UvrD and/or DinG takes place within the context of restarting reverted replication forks.
We show here that transcription is a stronger obstacle to replication at lower temperatures than at 37°C and we isolated mutations that suppress the rich medium cryo-sensitivity of rep uvrD recF cells. The genetic properties of these mutants suggest that the rep uvrD and rep uvrD recF strains provide a new way of isolating mutations that decrease the stability of RNA Pol on DNA, at promoters and/or in TEC. Furthermore, the analysis of these RNA Pol mutants reinforces the notion that Rep, UvrD and DinG have redundant functions in E. coli to facilitate the progression of replication forks across transcribed sequences.
The atomic structure of Thermus thermophilus and Termus aquaticus RNA Pol have led to structural models of TEC (Zhang et al., 1999; Korzheva et al., 2000; Vassylyev et al., 2002, 2007a,b). The similarity of these RNA Pols with the E. coli enzyme allows us to map the residues affected in our rpoB and rpoC mutants on these structures. The position of the mutated amino acids on a representation of a wild-type transcription complex is shown in Fig. 5. The three rpoB mutations, D444G, H447R and N518D (D324, H327 and T398 in T. thermophilus and T. aquaticus) lie in the major channel for the DNA/RNA hybrid, in, or close to the β-lobe 1, in a region previously described to be important for the stability of RNA Pol–DNA complexes (Trautinger and Lloyd, 2002; Rutherford et al., 2009). The rpoCH113R mutation (H90 in T. thermophilus and H101 in T. aquaticus) is on the other side of the major channel and although the mutated residue is buried in the enzyme, it may affect interactions of RpoC with the hybrid DNA-RNA. rpoCP451L (P719 in T. thermophilus and P730 in T. aquaticus) is in a different location in the complex, in a highly conserved region around the active site. This mutation is adjacent to a mutation isolated as a ΔdksA suppressor (H450R) and, as the RNA Pol cofactor DksA is also involved in the stability of transcription complexes, mutations in this region could affect the enzyme in another way to those which flank the major channel (Rutherford et al., 2009). In spite of the proximity of this mutation to the active site, we did not detect any deleterious effect of the rpoCP451L allele on E. coli growth (data not shown). Biochemical analysis of these mutant enzymes would tell whether the in vivo instability of transcription complexes is related to an increased propensity to pause and/or to terminate transcription.
The phenotypes of the isolated RNA Pol mutations do not allow their classification in specific groups but allow us to rank them according to the advantage that they confer to cells that lack Rep, UvrD and/or DinG helicases (summarized in Table 2). The best suppressor of growth defects in the absence of these helicases is the rpoCH113R mutation, which was isolated twice independently. This RNA Pol mutant is the only one that forms TEC unstable enough to suppress the temperature sensitivity of a greA greB double mutant. It confers LB resistance to rep uvrD, rep uvrD dinG recF and InvA dinG cells at all temperatures, and improves growth of InvA dinG rep cells at 37°C on MM. Of all mutants isolated here, the RpoCH113R RNA Pol is probably the one that forms the weakest complex on DNA. It is also the only one that slows down the growth of otherwise wild-type cells (data not shown), and could not be reliably combined with a ruvABC deletion. Slightly less efficient than rpoCH113R, the mutations rpoBH447R and rpoBD444G confer an intermediate phenotype, as the previously described rpoCΔ215–220 allele. The destabilization of transcription complexes caused by these mutations, which is deduced from their stringent-like phenotype and from the increased UV resistance conferred to ruv mutants, allows rep uvrD dinG recF and rep uvrD cells to grow on LB. They rescue InvA dinG and InvA rep dinG mutants, with the exception of the LB sensitivity of InvA rep dinG rpoBH447R cells (discussed below). Finally, the two less efficient suppressors are rpoCP451L and rpoBN518D. rpoCP451L is the only mutation that is close to the active site, while rpoBN518D was isolated at 37°C. The only sign that these mutated RNA Pols might be weakly bound to DNA, beside the rescue of rep uvrD recF cells, is the increase of UV resistance in ruv mutants. In their presence, rep uvrD cells form slow-growing colonies only at 37°C, and InvA dinG rep cells remain sick on MM. Furthermore, rpoCP451L fails to rescue InvA dinG cells on LB, while rpoBN518D only rescues it at 37°C and does not rescue rep uvrD recQ cells at 30°C.
Several mutations that mimic the effects of ppGpp have been isolated by different means, most often because they activate amino acid biosynthetic genes in a relA spoT double mutant, enabling it to grow on minimal medium (Bartlett et al., 1998; Zhou and Jin, 1998; Trautinger and Lloyd, 2002; Broccoli et al., 2004; Szalewska-Palasz et al., 2007). Some of these mutations decrease expression from the rrn promoter regardless of the presence of the upstream FIS-binding sites (Szalewska-Palasz et al., 2007), while the FIS sites compensate for the rpoCΔ215–220 mutation effects on the rrnB promoter (Bartlett et al., 2000). One mutation conferring a stringent-like phenotype was shown to reduce R-loop formation in a non-translated sequence by causing premature transcription arrest in vivo (Broccoli et al., 2004), and two others were shown to decrease the stability of TEC in vitro (Trautinger et al., 2005). Therefore, the rpoCΔ215–220, rpoBD444G and rpoBH447R mutations may be affected for the stability of transcription initiation complexes, or for the stability of TEC, or for both. Because the FIS sites compensate for the rpoCΔ215–220 mutation effects on the rrnB promoter, this mutation is likely to rescue helicase mutants carrying an inverted rrn by destabilizing TEC, as proposed in Boubakri et al., (2010). rpoBD444G may rescue InvA dinG rep cells because the promoter remains weakly expressed on LB in the presence of the FIS Sites, or because TEC are unstable. The RpoBH447R RNA Pol may fail to rescue InvA dinG rep cells because its defects are compensated for by the presence of the FIS sites and it forms more stable TEC than the RpoCΔ215–220 RNA Pol during rrn transcription. This observation supports the idea that there is no necessary correlation between the instability of TEC and the instability of the open complexes on rrn promoters. Furthermore, we isolated here a mutant (rpoBD444G) that affects transcription initiation at rrn promoters without affecting the expression of amino acids biosynthesis genes in a relA spoT context. To our knowledge, such a mutant was not reported before. Finally, rescue of a greA greB mutant at high temperature was previously described for a RNA Pol mutation isolated as increasing amino acids biosynthetic genes expression (Trautinger and Lloyd, 2002) whereas here it is observed for a RNA Pol mutant that does not affect these genes, nor rrn expression (rpoCH113R).
We used strains carrying a chromosome inversion to measure the consequences of a weaker stability of transcription complexes when the rate of replication arrest is increased either by the formation of R-loops, or by encountering oppositely oriented highly active RNA Pol in inverted rrn. With the exception of rpoCP451L, all RNA Pol mutations isolated here prevent R-loop formation. This means that RNA Pols that form unstable transcription complexes are less prone than the wild-type enzyme to R-loop formation. Conversely, only one (rpoBD444G) could prevent the lethality associated with the encounter of a series of RNA Pol transcribing a ribosomal operon in the direction opposite to replication. This means that even RNA Pols weakly bound to DNA will arrest replication forks when the latter collide with an oppositely oriented highly transcribed operon.
The general scheme that emerges from the present study and from previous studies of rep, dinG and uvrD single and multiple mutants is that Rep is the first factor that facilitates replication through transcribed sequences in vivo (Lane and Denhardt, 1975; Seigneur et al., 1998; Petit and Ehrlich, 2002; Guy et al., 2009; Boubakri et al., 2010). In the absence of Rep, UvrD becomes essential for E. coli viability and mutations in RNA Pol suppress this co-lethality, which points to UvrD as the main back-up to Rep for RNA Pol removal (Guy et al., 2009; Boubakri et al., 2010; this work). Two lines of evidence point to DinG as the other helicase that acts as a back-up. The first is our previous report that even in the absence of RecF, the combination of the rep, uvrD and dinG mutations is lethal in E. coli expressing the wild-type RNA Pol (Boubakri et al., 2010). This suggests that DinG removes wild-type RNA Pols from replication forks when Rep, UvrD and RecF are absent. The second, is our present observation that DinG is required for suppression of the rep uvrD co-lethality by most of the rposup mutations studied here, indicating that because they are unstable on DNA, the mutant RNA Pol can be removed from replication forks by DinG regardless of the presence of RecF. However, these mutated RNA Pols do not suppress the need for RecBC in a rep mutant, indicating that DinG and UvrD act in the context of restarting reversed replication forks. These helicases may recognize some features of reversed or of the restarting replication forks that are not shared by the original replication fork, whereas Rep may act within the context of the original replication fork that encounters the obstacle. It should be noted that the rposup mutations do not prevent spontaneous replication arrest in the E. coli chromosome because they did not relieve the growth defects of a priA mutant, which lacks the main replication restart protein PriA (strains JJC5540 and JJC5541 in Table S1, data not shown).
At low temperature, even in the presence of DinG, rep uvrD recF cells grow poorly on LB, indicating that DinG has a limited ability to remove the wild-type RNA Pol from replication forks under these growth conditions. Either transcription is a stronger obstacle to replication at low temperature, for example because transcriptional complexes are more stable, or for some reason DinG is less active. Provided that RecF is absent, several RNA Pol mutations isolated here bypass not only the need for Rep and UvrD but also the need for DinG (rep uvrD dinG recF rposup are viable, Fig. 3B). Removal of the mutant RNA Pol may then be catalysed either by some yet unknown function or by the replisome itself. Actually, in a purified system in vitro the wild-type RNA Pol initiation complex can be dislodged by the replisome (Pomerantz and O'Donnell, 2010). However, as RecF remains lethal in a rep uvrD dinG rposup mutant, replication forks are still arrested in these rposup cells that lack the three helicases, which is another indication that the dislodging of RNA Pol takes place during replication restart.
Our study of RNA Pol mutants that form unstable complexes on DNA provided conditions where viability is dependent on DinG only when RecF is present (the rep uvrD rposup combinations of mutations become LB-sensitive upon DinG inactivation only in a RecF+ context). This observation suggests that DinG counteracts a deleterious action of RecF; either DinG could remove RecFOR/RecA from DNA, or it could prevent RecFOR binding. It should be noted that UvrD can act both ways, it removes RecA from DNA or prevents RecFOR binding, depending on the cause of replication arrest (Lestini and Michel, 2007). The hypothetical removal of RecFOR and/or RecA from DNA by DinG needs to be tested in vitro; nevertheless, we do not favour this hypothesis because it is difficult to explain why DinG would remove RecF only in certain RNA Pol mutants. We favour the hypothesis that DinG and RecFOR compete for blocked replication forks in rep uvrD cells. Because DinG is unable to remove the wild-type RNA Pol, rep uvrD cells are killed by RecFOR. Similarly, the rep uvrD rpoN518D mutant is non-viable on LB at low temperature because under these conditions DinG cannot dislodge this mutated RNA Pol from DNA, letting RecFOR bind to arrested replication forks. Conversely, DinG is capable of removing all other RNA Pol mutants isolated here before RecFOR binds, and consequently the viability of the rep uvrD mutant becomes independent of the recF context in these RNA Pol mutants.
The strain backgrounds are MG1655, or JJC40, which is an hsdR Thr+ Pro+ derivative of AB1157 (leu-6 thi-1, his-4, argE3, lacY1, galK2, ara-14, xyl-5, mtl-1, tsx-33, rpsL31, supE44, hsdR, hsdM). JJC147 is HfrPK3-PO131 (thr1, leuB6, azi-15, tonA1, lacY1, supE44, argE86::Tn10). Strains genotypes and plasmids are described in Table S1. InvA is described in Boubakri et al. (2010), it is an MG1655 derivative, deleted for lacZ and attλ, which carries a 18 kb inversion (NT 4 025 300 to 4 043 723) encompassing the rrnA operon. Minimum medium is M9 supplied with thiamine 0.05%, CaCl2 100 µM, MgSO4 2 mM, glucose 0.4% (Miller, 1992). Casamino acids 0.2% were added for strain background other than MG1655. Most of the strains were constructed by P1 transduction (Miller, 1992). For Hfr conjugations, a mix of 10% donor cells/90% recipient cells was incubated for 15 min with low agitation at 37°C before plating on selective media. Null mutants were checked by polymerase chain reaction with external oligonucleotides that amplify a DNA fragment of different length for the wild-type and the interrupted alleles. Oligonucleotides used for checking mutations are shown in Table S4. uvrD mutants were checked for their UV-sensitive phenotypes and for their mutator phenotype (about 100-fold increase in the proportion of RifR clones in overnight cultures). The plasmids pGBts-rep (Petit and Ehrlich, 2002), pAM-rep (Lestini and Michel, 2008) and pAM-priA (Grompone et al., 2004) were described; they carry the rep or the priA gene under the control of their own promoter; they were segregated as published.
rpo mutations are rpoB A1331G (rpoBD444G), A1340G (rpoBH447R), A1552G (rpoBN518D), rpoCA338G (rpoCH113R), C1352T (rpoCP451L). Following P1 transduction or conjugation, screening for A1552G (rpoBN518D) was based the RifR phenotype it confers. Screening for A1331G (rpoBD444G) was done sequencing the rpoB region. Three mutations were screened after polymerase chain reaction amplification of the region either by sequencing or by looking for the presence of a restriction site that they create: A1340G (rpoBH447R) creates a BsiEI site, rpoCA338G (rpoCH113R) creates a BssHII site, C1352T (rpoCP451L) creates a BspMI site. Upon P1 co-transduction of rposup with thiC::Tn10 in a context where the mutation confers a phenotype, the phenotype and the presence of the mutation were tested in four to eight different TetR transductants and there was 100% correlation between the presence of the mutation and the appearance of the new phenotype. Twenty regularly spaced oligonucleotides in rpoB (10 on each strand) and 20 regularly spaced oligonucleotides in rpoC (10 on each strand) were used for sequencing of the entire genes.
Overnight cultures grown in MM (OD650 1 to 2) were diluted and plated on MM or LB plates, incubated at 37°C, 30°C or 25°C. LB plates at 37°C were counted after 24 and 48 h of incubation, MM plates at 37°C and LB plates at 30°C after 48 h and 3 days of incubation, and MM plates at 30°C and LB plates at 25°C after 3 and 4 days incubation.
We thank O. Espeli, R.L.G. Gourse and C.J. Herbert for very helpful reading of the manuscript. We are very grateful to Annie Kolb for discussions on RNA polymerase structure, and to Audrey Lagès and Marie Le Masson for excellent technical assistance. The work was supported by ANR-05-BLAN-0204-01, ANR-08-BLAN-0230-01 and Prix ‘coup d'élan’ from the foundation Bettencourt-Schueller.
Additional supporting information may be found in the online version of this article.
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.