*Present address: Sir William Dunn School of Pathology, University of Oxford, Oxford OX1 3RE, UK
The replication checkpoint control in Bacillus subtilis: identification of a novel RTP-binding sequence essential for the replication fork arrest after induction of the stringent response
Article first published online: 1 MAR 2002
Blackwell Science Ltd, Oxford
Volume 31, Issue 6, pages 1665–1679, April 1999
How to Cite
Autret, S., Levine, A., Vannier, F., Fujita, Y. and Séror, S. J. (1999), The replication checkpoint control in Bacillus subtilis: identification of a novel RTP-binding sequence essential for the replication fork arrest after induction of the stringent response. Molecular Microbiology, 31: 1665–1679. doi: 10.1046/j.1365-2958.1999.01299.x
- Issue published online: 1 MAR 2002
- Article first published online: 1 MAR 2002
We have shown previously that induction of the stringent response in Bacillus subtilis resulted in the arrest of chromosomal replication between 100 and 200 kb either side of oriC at distinct stop sites, designated LSTer and RSTer, left and right stringent terminators respectively. This replication checkpoint was also shown to involve the RTP protein, normally active at the chromosomal terminus. In this study, we show that the replication block is absolutely dependent upon RelA, correlated with high levels of ppGpp, but that efficient arrest at STer sites also requires RTP. DNA–DNA hybridization data indicated that one or more such LSTer sites mapped to gene yxcC (−128 kb from oriC ). A 7.75 kb fragment containing this gene was cloned into a theta replicating plasmid, and plasmid replication arrest, requiring both RelA and RTP, was demonstrated. This effect was polar, with plasmid arrest only detected when the fragment was orientated in the same direction with respect to replication, as in the chromosome. This LSTer2 site was further mapped to a 3.65 kb fragment overlapping the next40 probe. Remarkably, this fragment contains a 17 bp sequence (B′-1) showing 76% identity with an RTP binding site (B sequence) present at the chromosomal terminus. This B′-1 sequence, located in the gene yxcC, efficiently binds RTP in vitro, as shown by DNA gel retardation studies and DNase I footprinting. Importantly, precise deletion of this sequence abolished the replication arrest. We propose that this modified B site is an essential constituent of the LSTer2 site. The differences between arrest at the normal chromosomal terminus and arrest at LSTer site are discussed.
Environmental stress induces a readjustment of the synthesis of many macromolecules in bacterial cells. Thus, nutritional deprivation, energy starvation, temperature shift or osmotic shock lead to the accumulation of the alarmone ppGpp, guanosine 3′,5′-bis(diphosphate). The intracellular level of ppGpp is regulated by two enzymes in Escherichia coli, RelA (ppGpp synthetase I), which is associated with the ribosomes, and the bifunctional SpoT protein, involved in both the synthesis and the degradation of the alarmone (Gentry and Cashel, 1996). In Bacillus subtilis, a single enzyme associated with ribosomes appears to carry both activities (Wendrich and Marahiel, 1997). In both E. coli and B. subtilis, amino acid starvation results in the accumulation of uncharged tRNAs, which activate RelA, leading to the accumulation of ppGpp and many metabolic changes, including the inhibition of stable RNA synthesis. This pleiotropic response to nutritional downshift is termed the stringent response (for a review, see Cashel et al., 1996). Increased levels of ppGpp elicited by the overexpression of relA, without amino acid deprivation, also inhibit the synthesis of stable RNA and certain proteins, while stimulating the synthesis of other proteins (Schreiber et al., 1991). This suggests that ppGpp levels normally regulate many synthetic processes in the cells directly or indirectly. For example, in E. coli, ppGpp appears to stimulate the synthesis of σS, a positive regulator of the expression of many stationary-phase genes (Gentry et al., 1993). ppGpp has also been implicated in growth rate control under steady-state conditions. Thus, it has been shown that the intracellular concentration of ppGpp is inversely correlated with the synthesis of stable RNA and also with the growth rate. In a strain ΔrelAΔspoT, lacking ppGpp, the proportion of mRNA and stable RNA synthesis becomes independent of the growth rate (Hernandez and Bremer, 1991).
The onset of DNA replication is subject to stringent control in both E. coli and B. subtilis (Levine et al., 1991). In addition, at least in E. coli, studies have indicated that elements involved in the regulation of the initiation of DNA replication, including the synthesis of the initiation protein DnaA, Fis and IHF (Chiaramello and Zyskind, 1990; Ninneman et al., 1992; Aviv et al., 1994) and transcription in the oriC region (Rokeach et al., 1987; Ogawa and Okazaki, 1991), are subject to stringent control. However, even in cells in which DnaA has accumulated to a high level, induction of the stringent response still inhibits the initiation of DNA replication from oriC (Levine et al., 1991). Other studies have shown that a high level of ppGpp caused by the overexpression of relA, in the absence of amino-acid starvation, inhibited the initiation of DNA replication in E. coli (Schreiber et al., 1995). This indicates that augmentation of the intracellular level of ppGpp itself may in some way control DNA replication at the level of initiation in E. coli. However, the precise mechanism of inhibition of DNA synthesis by ppGpp under these conditions is still unknown.
The effect of the induction of the stringent response on DNA replication in B. subtilis was first shown by Séror et al. (1986) and has been studied extensively using a temperature-sensitive initiation mutant, dnaB37, to provide cells synchronized with respect to DNA replication. In contrast to E. coli, the induction of the stringent response inhibits DNA replication, not at the level of initiation at oriC but approximately 200 kb downstream on either side of oriC, that is after replication of approximately 10% of the B. subtilis chromosome. Interestingly, this situation has been exploited to demonstrate that this amount of DNA replication was sufficient to allow separation of newly replicated origins, as shown by immunofluorescence studies (Sharpe and Errington, 1998). Importantly, when the stringent response is lifted, replication resumes close to or at the blocked sites, independently of any protein or RNA synthesis (Levine et al., 1991; Séror et al., 1991). The analysis of sequences duplicated during the stringent response suggested the existence of several stop sites, called STer sites (stringent terminus) on the left and the right arms of the chromosome (Levine et al., 1995). Interestingly, previous studies of the dnaB37 mutant have shown that DNA replication, resulting from premature initiation induced by transfer from non-permissive temperature to permissive temperature, in this case without any inhibition of protein synthesis, is also blocked downstream of the origin. On the basis of the determination of the copy number of different chromosomal markers and the analysis of the size of replication loops by electron microscopy, it appears that, after premature initiation, DNA replication is also blocked transiently at similar positions to STer sites downstream of the origin (Henckes et al., 1989). Thus, in B. subtilis, a second level of post-initiation control is involved in the regulation of DNA replication, acting as a checkpoint to impede the movement of the replisome after the induction of the stringent response or after premature initiation of DNA replication.
We have shown that the replication terminator protein (RTP) is required for efficient inhibition of DNA synthesis after the induction of the stringent response (Levine et al., 1995). RTP in B. subtilis, like Tus in E. coli, is involved in the termination of normal cycles of chromosome replication (for a review, see Hill, 1992; Baker, 1995). This protein interacts with specific DNA terminators, Ter sequences, located in a region diametrically opposite to oriC (Carrigan et al., 1987). Each terminator element, whose effects are strictly polar, comprises two binding sites of different affinity for an RTP dimer (Lewis et al., 1990). The B site, also called the core sequence, is first filled by RTP, then a second dimer contacts the A site by co-operative interaction with the first dimer (Langley et al., 1993; Sahoo et al., 1995a). The interaction of two dimers with the terminator is absolutely required for the function of the nucleoprotein complex (Smith et al., 1994; Sahoo et al., 1995a; Manna et al., 1996a). RTP has been crystallized (Bussiere et al., 1995), and the different regions responsible for binding to DNA and the interaction between the two dimers on each DNA terminator have been identified (Manna et al., 1996a; Pai et al., 1996). Although RTP and Tus share no significant homology, both are able to block DNA unwinding catalysed by the replicative helicases DnaB and PriA of E. coli. (Lee et al., 1989; Khatri et al., 1989; Kaul et al., 1994; Sahoo et al., 1995b). Moreover, RTP was shown to interact with DnaB in vitro (Manna et al., 1996b)
In this study, we present more detailed evidence for the existence of specific DNA replication arrest sites on the chromosome, dependent upon RTP, after the induction of the stringent response in B. subtilis. We have mapped an apparent arrest site within a 3.65 kb region encompassing genes yxcC to iolS, which is capable of arresting replication of a plasmid during the stringent response. This arrest is specifically dependent upon RTP and RelA and displays polarity. Importantly, the yxcC gene contains a previously identified motif with 76% identity to the RTP B-site, which indeed binds RTP efficiently in vitro, as shown by DNA gel retardation and DNA footprinting. Deletion of this sequence abolished the arrest.
Role of RTP and RelA in DNA replication arrest obtained in the presence of Arg-HX
In order to analyse the contribution of RTP and RelA to the inhibition of DNA replication downstream of the origin after the induction of the stringent response, residual incorporation of radioactive precursors was determined in different derivatives of the dnaB37 mutant listed in Table 1: rtp+relA+ (OMM242); Δrtp relA+ (OMG250); rtp+relA− (OMM243), Δrtp relA− (OMG251). Cultures containing uniformly labelled DNA ([3H]-thymidine) were grown at 30°C and transferred to 45°C for 30 min in order to synchronize DNA replication as described previously (Levine et al., 1991). Upon return to the permissive temperature, 30°C, separate cultures were treated with arginine hydroxamate (Arg-HX), the drug used to induce the stringent response. Control cultures were also treated with chloramphenicol or rifampicin in order to measure the completion of one round of DNA replication or to prevent further initiation of replication respectively.
As expected, the addition of rifampicin completely abolished new rounds of replication, whereas, an approximate 100% increment in newly replicated DNA (measured by [3H]-thymidine incorporation), representing one round of DNA synthesis, was obtained in the presence of chloramphenicol (see Fig. 1A). In contrast, in the untreated controls, an additional round of DNA replication can occur before the cells move into stationary phase under the conditions used in these experiments. The results in Fig. 1 also show that induction of the stringent response, which leads to a 20-fold increase in the intracellular level of ppGpp under the conditions used in this study (Levine et al., 1991), limited residual DNA synthesis to 15–20% in different experiments, consistent with a replication arrest downstream of the origin, as described earlier (Fig. 1A; see also Levine et al., 1995). However, after the addition of Arg-HX in the absence of RTP (Fig. 1B), replication continued until at least 60% of the chromosome was replicated (Fig. 1B). In the relA− strain, the alleviation of the effect of Arg-HX treatment was even more dramatic, with the addition of the drug having virtually no effect on residual DNA replication (Fig. 1A). A similar result was obtained with the relA−Δrtp double mutant (Fig. 1B). The results demonstrated that RelA and, therefore, presumably the production of ppGpp may be sufficient to retard replication fork movement, with the presence of RTP ensuring that the final arrest takes place in a relatively well-defined region, at the STer sites. Alternatively, ppGpp may have two effects: in collaboration with RTP to effect an efficient arrest; plus a less specific effect to reduce replisome movement even in the absence of RTP.
Residual DNA synthesis after the induction of the stringent response was also measured in a dnaB37, spac-1 rtp strain (OMG270), which constitutively overproduces RTP by a factor of 50–100, relative to the wild-type level, and in the mutant dnaB37, spac-1 rtp, relA− (Table 1). We have shown previously with OMG270 using DNA–DNA hybridization methods that DNA synthesis is limited to a small region of a maximum of 100 kb around oriC, with even the most proximal markers not being fully replicated when RTP is overproduced (Levine et al., 1995). In agreement with our previous experiments, we show in this study (Fig. 1C) that the level of residual incorporation of [3H]-thymidine in OMG270 overproducing RTP and treated with Arg-HX was extremely low, as found with rifampicin treatment. Importantly, high-level expression of RTP in OMG270 in the absence of Arg-HX had no detectable effect on growth (data not shown). However, 1Fig. 1C shows that the replication arrest close to the origin induced by the addition of Arg-HX in the presence of high levels of RTP still requires RelA, as this effect was also abolished in the relA− mutant. This result provides evidence that the activity of RelA and RTP are indeed somehow co-ordinated to effect replication arrest.
It is important to note that, in previous studies using radiolabelled thymidine, we demonstrated that, in the absence of RTP, replication continued well past LSTer2 (Levine et al., 1995). Moreover, earlier studies (Séror et al., 1986) have indicated that induction of the stringent response did not inhibit the incorporation of the radiolabelled precursors into DNA. Such data indicate that replication does not appear to terminate at LSTer2, simply because the uptake of radioactive thymidine is being inhibited by the induction of the stringent response, as has been reported for E. coli (Wegrzyn and Taylor, 1992). Nevertheless, in this study, we also measured the increase in DNA in the different strains used in Fig. 1, in the presence or in the absence of the stringent response, using a chemical method based on DNA fluorescence measurements (see Experimental procedures). The results obtained were essentially identical to those shown in Fig. 1. This confirmed that, in B. subtilis, under these conditions, radioactive thymidine incorporation does reflect the amount of DNA synthesis.
Analysis of the precise sequences replicated during the stringent response on the left arm of the chromosome
We have presented previously evidence (Levine et al., 1995) that replication forks may halt in several regions or sites on the right and the left arms of the chromosome after the induction of the stringent response. Thus, such data indicate two candidate arrest regions on the right arm, which appear to be located between spoOH (+ 115 kb) and next5 (+ 166 kb) and between soft10 (+ 199 kb) and lipA (+ 251 kb). In this study, we focused our attention on the arrest of replication located to the left of oriC. Using DNA–DNA hybridization, we analysed the replication of numerous markers after the induction of the stringent response in OMM242 (dnaB37 rtp+relA+). Our results indicated the presence of two possible LSTer regions, LSTer1 between omg80 (−94 kb) and omg84 (−104 kb) and LSTer2 covering a region of approximately 8 kb identified by the probes omg100 (−127 kb) and omg104 (−135 kb) (Fig. 2). Subsequent studies were devoted to attempts to identify precisely the arrest site in the LSTer2 region.
Cloning the next40–omg103 region into a theta replicating plasmid
In attempting to localize the position of the arrest site in the LSTer2 region more precisely, we decided to clone this region in a B. subtilis, E. coli shuttle plasmid, pWS64-2 (Table 1). This plasmid is derived from pAMβ1, which replicates in B. subtilis by a theta unidirectional mechanism (Bruand et al., 1991).
Replication of plasmid pWS64-2 during the stringent response.
In designing experiments to detect RelA, RTP-dependent replication arrest sites, it was essential to establish conditions in which the production of ppGpp did not inhibit replication by a mechanism distinct from that correlated with LSTer sites. The replication of many replicons, such as colE1-type plasmids, pSC101 and λ, is inhibited when the intracellular concentration of ppGpp is increased either by amino acid starvation or by overexpression of relA (Szalewska-Palasz et al., 1994; Herman and Wegrzyn, 1995; Wröbel and Wegrzyn, 1998). In contrast, replication of a plasmid carrying the ori region from RK2 is inhibited after amino-acid starvation, but not when the intracellular level of ppGpp is increased without nutritional downshift. Thus, replication of most, but not all, replicons is dependent on the ppGpp concentration (Herman and Wegrzyn, 1995). In order to test whether pWS64-2 still replicates in the presence of a high level of ppGpp in the cell, incorporation of [3H]-thymidine into plasmid DNA was measured in the control and in the Arg-HX-treated cells. In this experiment, we also measured the incorporation of radioactive precursors into plasmid DNA after the addition of rifampicin and chloramphenicol. Two successive pulses of [3H]-thymidine of 15 min each were given to the different cultures, the first one immediately after the addition of the drugs and the second 15 min later. Plasmids were extracted, purified by CsCl gradient centrifugation, and the radioactivity incorporated into plasmid DNA was determined. As shown in Table 2, the level of [3H]-thymidine incorporated into plasmid DNA was severely affected by the addition of rifampicin and chloramphenicol. In contrast, the incorporation of radioactive precursors in the plasmid extracted from the control and the Arg-HX-treated cells was very similar after a 0–15 min pulse, and some reduction was observed after a longer pulse. In such conditions, we did not detect any accumulation of replication intermediates in the presence of Arg-HX (data not shown). Therefore, induction of the stringent response per se does not inhibit replication of the parental vector pWS64-2 under the conditions of the experiment, and this plasmid was thus considered to be suitable for identifying putative LSTer, terminator sequences.
Construction of plasmids pOMG4-a and pOMG4-b.
The 7.75 kb fragment containing next40, omg101 and omg103 was cloned into pWS64-2 at the SmaI site in both orientations (Fig. 3). In pOMG4-a, the insert containing the putative LSTer2 site is in the same orientation as in the chromosome relative to the direction of the replication fork. In pOMG4-b, the insert is in the opposite orientation.
ppGpp and RTP are required for the detection of an arrest in plasmid pOMG4-a
The copy number of pWS64-2 is relatively high in B. subtilis, 50–100 copies per cell. However, RTP is normally maintained at a very low level probably because of autorepression at the transcriptional level (Ahn et al., 1993). Therefore, to facilitate the analysis of the influence of RTP on plasmid DNA synthesis during the stringent response, it was necessary to use OMG219, a construct expressing rtp from the inducible promoter spac-1 (see Experimental procedures). Western blotting indicated that, in this strain, RTP was in fact overexpressed approximately 50-fold after the addition of IPTG (data not shown).
As a control for the production of functional RTP protein, OMG219 was transformed by pWS65-2 (Table 1), a derivative of pWS64-2 carrying 250 bp of the terminus region of the chromosome with the terminator sequences IRI and IRII. In this construct, the replication fork is impeded by RTP bound to IRII (Smith and Wake, 1992). OMG219 transformed by this plasmid was grown in minimal medium at 37°C to mid-exponential phase, diluted in the same medium plus or minus IPTG and grown for 2 h. Plasmids were extracted using the neutral lysis procedure and restricted by Bgl II. The fragments were submitted to agarose gel electrophoresis and analysed by Southern blotting with a probe corresponding to the restriction fragment Bgl II–SmaI located between the origin of replication and the multiple cloning site (Fig. 3). As expected, a single 3.4 kb band was observed when pWS65-2 was extracted from OMG219 grown without IPTG (Fig. 4A, lane 1). After overproduction of RTP, an additional slow-moving band was detected in pWS65-2 (Fig. 4A, lane 2). Such species correspond to replication intermediates, blocked by RTP, in this case at the normal chromosomal terminus, generating branched molecules after restriction enzyme digestion.
OMG219 transformed by pOMG4-a was grown in minimal medium at 37°C to mid-exponential phase and diluted in the same medium plus or minus IPTG. The cells were grown for 2 h, then half of the culture was treated with Arg-HX for 20 min in each case. Cultures were harvested, and plasmids were extracted by the neutral lysis procedure. The plasmid DNA was restricted by BamHI and EcoRI, generating four fragments (Fig. 3), the A fragment of 3.65 kb (containing next40 and a small part of omg101), the 1.65 kb B fragment (omg101) and the 2.44 kb C fragment (essentially omg103), as well as the linearized vector of 8.05 kb. The fragments were submitted to agarose gel electrophoresis and analysed by Southern blotting with probes specific for fragments A, B and C. If a functional STer site is present in the insert, it should lead to the accumulation of blocked (slowly migrating) replication intermediates in the presence of a high level of ppGpp and RTP. The results of this experiment are presented in Fig. 4. In addition to the 3.65 kb band expected for the A fragment, small but clearly detectable amounts of a more slowly migrating band were obtained reproducibly when plasmid DNA was analysed from cells overexpressing rtp, in which the stringent response had also been induced (Fig. 4B, lane 4). This species was not detected with plasmids extracted from control cells (lane 1), or from Arg-HX-treated cells without RTP (lane 2), or from cells overproducing RTP untreated with Arg-HX (lane 3). The novel band detected in 4Fig. 4B most probably reflects the accumulation of Y-shaped molecules resulting from a replication arrest in the A fragment, which is strictly dependent upon the overproduction of RTP and the induction of the stringent response. 4Figure 4D and E also presents the results of the analysis of the B fragment (lanes 1–4) and the C fragment. In each case, whatever the conditions of culture, no retarded species were detected. Therefore, our results showed that, after IPTG treatment and the induction of the stringent response, the replisome is apparently halted only in the A fragment. The relatively low intensity of the slowly migrating species indicated that replisome arrest was only being detected in a minority of plasmid molecules. This is not unexpected, and a number of hypotheses may be advanced to explain this (see Discussion). Moreover, even in the control experiment (Fig. 4A) using a plasmid containing an authentic terminator (A + B sites), only a minority of molecules were arrested when RTP was overproduced.
Enrichment for slowly migrating molecules by BND cellulose chromatography
In order to increase the sensitivity of the detection of Y-shaped molecules, we decided to use a benzoylated naphthoylated DEAE (BND) cellulose column, which specifically retains single-strand DNA. Thus, the experiment described in 4Fig. 4C was repeated, but this time the plasmid DNA was applied to a BND cellulose column and, after digestion by restriction enzymes, replication intermediates were analysed as before. As shown in 4Fig. 4C (lane 1), a larger amount of the slowly migrating band of an apparent size of 5.5 kb, corresponding to that detected previously, was found reproducibly. In fact, one additional slowly migrating band of 7.3 kb was also detected in this experiment, but not characterized further. Finally, the samples were treated with mung bean endonuclease, a single-strand specific endonuclease. As expected, the slowly migrating bands were sensitive to this enzyme and disappeared (Fig. 4C, lane 2), confirming that they corresponded to Y-structures. Similar results were obtained in the presence of S1 endonuclease (data not shown).
When the polarity of the 7.75 kb BamHI--EcoRI fragment (pOMG4-b) is reversed, an RTP, ppGpp-dependent arrest is not detectable
The same experiment described earlier for pOMG4-a was carried out with pOMG4-b, which contains the same insert but in the opposite orientation (Fig. 4F–H). Plasmid DNA was extracted, treated with the same restriction enzymes and analysed as before. In contrast to the results with pOMG4-a, no retarded species were detected in the A fragment in pOMG4-b even when the cells were treated by both IPTG to induce RTP and Arg-HX (Fig. 4F). This suggests that the arrest of replication detected in pOMG4-a is dependent upon the orientation of the insert relative to the direction of the replication fork. The analysis of the B and C fragments are shown in Fig. 4G and H respectively. As in pOMG4-a, no slowly retarded species was detected with the B′ and C′ probes.
Organization of the LSTer2 region
We have shown that, after the induction of the stringent response, an arrest of DNA replication can be detected at at least one site in a plasmid construct pOMG4-a within the 3.65 kb A fragment, which encompasses the genes yxcC to iolS. As this arrest requires RTP, in addition to the induction of the stringent response, we have searched for sequences homologous to the consensus RTP binding sites in the LSTer2 region of the B. subtilis chromosome. The chromosomal terminator B site, which binds RTP tightly, is relatively well conserved in all the chromosomal terminators identified in strains W168 and W23 by Wake and colleagues (Franks et al., 1995). The A site alone interacts very poorly with RTP and displays a degenerated sequence with only the first three bases and the fifth base being conserved in the different terminators (Franks et al., 1995). As expected, we did not find complete terminators including both an A site and a B site in the 3.65 kb A fragment. However, four sequences diverging from the consensus B site by four mismatches (76% homology) could be detected (Fig. 5). It was necessary, therefore, to test these different sequences in vitro for their ability to bind RTP.
RTP protein binds specifically to the B′-1 sequence, as determined by gel retardation analysis and DNase I footprinting
The binding of RTP to each of the four B′ sequences in vitro was investigated by gel retardation experiments, using polymerase chain reaction (PCR) products containing the putative RTP binding site. As a control, we used a 250 bp DNA fragment from the chromosomal terminus region carrying the two DNA terminators, IRI and IRII. This fragment corresponds to the insert that impeded the replication fork in the presence of a high level of RTP in pWS65-2 (Fig. 4A). As expected (see Lewis et al., 1990), a four-step mobility shift was observed for this fragment (Fig. 6A), reflecting the binding of two dimers of RTP on each IR terminator (B + A sites). Figure 6 presents the results of RTP-binding assays performed with the PCR products containing B′-1, B′-2, B′-3 and B′-4 sequences. The DNA fragment harbouring the B′-1 site clearly showed a reduction in mobility when increasing amounts of RTP were added (Fig. 6B). As the length of the DNA fragments containing IRI-IRII and B′-1 is equivalent and as the two experiments were done in parallel, it appears that retardation of the B′-1 fragment is accompanied by the binding of only one RTP dimer. A control experiment also showed that the gel mobility shift was not affected by mixing samples with an unrelated 180 bp fragment (equivalent c.p.m.). In contrast, RTP did not interact detectably in vitro with B′-2, B′-3 or B′-4 sequences (Fig. 6C), as indicated by the absence of any gel retardation effects (data not shown). Thus, among the four B′ sequences identified in the 3.65 kb fragment A, only the B′-1 sequence interacted detectably with RTP in vitro.
In order to confirm that RTP protein binds specifically to the B′-1 sequence, a DNase I footprinting experiment was then performed with the same 250 bp DNA fragment. Importantly, the experiment revealed that RTP protected from DNase I cleavage a region of 17 bp, corresponding to the B′-1 sequence (Fig. 7), thus confirming the specificity of the binding of RTP to B′-1, with no co-operative binding to a sequence adjacent to B′-1. Finally, a gel retardation experiment was repeated using many concentrations of RTP with the B′-1 sequence used in 6Fig. 6A and a DNA fragment of 239 bp containing a perfect B site but no A site. In this case, the results were quantified by phosphorimaging, as described in Experimental procedures. The apparent dissociation constant (KD) for the RTP/B′-1 interaction was found to be 1.45 × 10−11 M (dimer of RTP), differing only by a factor of three from the KD for the interaction between RTP and a perfect B site (0.45 × 10−11 M). This has to be compared with the KD values of 1.2 × 10−11 M and 2 × 10−11 M reported previously for authentic terminators at the terminus (Lewis et al., 1990; Smith et al., 1996).
The B′-1 sequence, within the 3.65 kb A fragment, is essential for the stringent arrest
In order to test if the B′-1 sequence, which interacts with RTP in vitro, is involved in the plasmid replication arrest in vivo, the 7.75 kb LSTer2 region was deleted for 18 bp encompassing B′-1 and cloned into pWS64-2 (see Experimental procedures). The replication of the recombinant plasmid, pOMG12, was analysed as before in strain OMG219 cultivated in various conditions. In this case, no retarded species were detected with the A fragment, even when the plasmid DNA was extracted from cells overexpressing rtp, in which the stringent response had also been induced (data not shown). This indicates clearly that the RTP binding site B′-1 in the A fragment is required to halt replisome movement after the induction of the stringent response.
Analysis of an RTP mutant indicates that the interaction between two RTP dimers may not be required for an arrest at an LSTer site
The replication arrest at the Ter sites in the terminus region is absolutely dependent upon the binding of two RTP dimers to the overlapping B and A sites of the DNA terminators. In contrast, the B′-1 RTP binding sites involved in the arrest of plasmid replication in LSTer2, after the induction of the stringent response, are not associated with an adjacent A site. This is deduced from the sequence, from gel retardation analysis and from DNase I footprinting experiments. Therefore, in order to test the possibility that the arrest at the STer site may have different requirements from the arrest at the terminus, we used a mutation in RTP, Y88F, identified by Manna et al. (1996a), which prevents the interaction between two RTP dimers binding to the core B site and the adjacent auxiliary (A) site. Such a mutation does not allow termination at the terminus, either in vitro (Manna et al., 1996a) or in vivo (G. Wake, personal communication). Two strains derived from dnaB37 were constructed, OMG257, which contains the mutation mentioned above, and OMG275, the control strain carrying the wild-type rtp gene behind the same promoter (spac-1 ). A DNA–DNA hybridization experiment showed that, as expected, the induction of the stringent response provoked an arrest of DNA replication downstream of the origin in OMG275 (Fig. 8A). Note that the amount of RTP produced in the absence of IPTG is sufficient to halt replication fork movement. As shown before, in a mutant deleted for rtp, there is no arrest after induction of the stringent response (Fig. 8B). In fact, in OMG257, containing the RTP (Y88F) mutation preventing the interaction between two RTP dimers, the arrest of DNA replication still occurs downstream of the origin on the left and the right arms (Fig. 8B). Therefore, the arrest of DNA replication at an STer site does not seem to require co-operative binding of two dimers of RTP on adjacent sites, thus indicating that the mechanism of arrest involving RTP might be different from that operating at the normal terminus. These results support both the mobility shift and the DNase I footprinting experiments.
We have shown previously that the induction of the stringent response blocks DNA replication in B. subtilis approximately 200 kb downstream of the origin on both the left and the right arms (Levine et al., 1991). Subsequently, we showed that, after induction of the stringent response, the replication forks in B. subtilis appeared to halt in several different regions or sites (STer) on both arms of the chromosome and that this process involved the replication terminator protein (RTP) (Levine et al., 1995). Nevertheless, whereas in the absence of RTP, the replication forks are able to pass through the STer sites during induction of the stringent response, our previous study, based on DNA–DNA hybridization data, indicated that full replication of the chromosome was not completed. This suggested that replication may be partially impeded even in the absence of RTP. In this study, we have assessed the relative roles of RTP and RelA (controlling the stringent response) in blocking replication by direct measurement of residual DNA replication (using radiolabel or fluorescence of DNA) after the addition of Arg-HX to induce the stress response. Noticeably, in the absence of RelA and therefore in the absence of ppGpp or, at least, at greatly reduced levels of ppGpp, the addition of Arg-HX had no detectable effect on residual DNA synthesis, demonstrating that the arrest sites were abolished completely. In contrast, in the absence of RTP, we again found that, although replication arrest was alleviated, full replication of the chromosome was not apparently completed. One simple interpretation of these results is that high levels of ppGpp may eventually retard replication fork movement, with RTP ensuring that the final arrest takes place in specific sites, at the STer sites. Alternatively, even in the absence of RTP, an additional factor may participate with high levels of ppGpp to inhibit replication.
We have now analysed more precisely the region of the replisome arrest involving RTP and ppGpp, and the results demonstrate that the replisome was impeded in two relatively well-defined regions on the left arm of the chromosome. Thus, approximately half of the replication forks appeared to be arrested in a 10 kb region between omg80 (−94 kb) and omg84 (−104 kb), designated LSTer1, with a second arrest in an 8 kb region, LSTer2, delimited by the chromosomal markers omg100 (−127 kb) and omg104 (−135 kb). In attempting to define the precise nature of an arrest site further, it was necessary to clone the corresponding chromosomal region into a plasmid. Then, it was essential to optimize the conditions to give maximal levels of ppGpp synthesis, by arginine hydroxamate treatment, commensurate with minimum inhibition of plasmid replication. This was achieved with plasmid pWS64-2, which has the advantage of a theta unidirectional replicating mechanism, allowing easy identification of arrested replication forks, detectable as Y-shaped molecules.
The region LSTer2 (7.75 kb) was inserted into the vector in both orientations relative to the direction of replication fork movement, and DNA synthesis of the recombinant plasmids was studied in a strain in which the rtp gene was under the control of an IPTG-inducible promoter. We showed that the inhibition of replication of pOMG4-a could be detected when both RTP and ppGpp accumulated in the cells. The replication arrest, which was completely dependent on RelA and RTP, occurred mainly at one site located in the 3.65 kb part of the insert, which contains the genes yxcC to iolS. The arrest mechanism also exhibited polarity, as the replication of the plasmid carrying the insert in the opposite orientation was not impeded.
The arrest detected with pOMG4-a occurred at low frequency in the population of plasmid DNA molecules. This may indicate that the arrest with LSTer2 in a plasmid may be less efficient than when present in the chromosome. This might result from several factors, for example the level of supercoiling, the pattern of transcription in the plasmid or the precise context of the LSTer site with respect to upstream or downstream sequences. It is also important to note that we estimate that less than 20% of plasmid DNA molecules in our experiments actually initiated their replication between the moment of the addition of Arg-HX in the culture and the moment when the cells were harvested. In addition, when an authentic chromosomal DNA terminator was present in pWS64-2, termination arrest by RTP was relatively inefficient (Smith and Wake, 1992; see also Fig. 4A).
RTP normally functions as a contrahelicase, which exerts its effect by specific binding to overlapping DNA terminator sequences A and B, constituting a so-called IR element. Importantly, we detected four putative binding sites for RTP sharing homology with the 17 bp terminator B site in a 7.75 kb sequence encompassing the LSTer2 region,. Moreover, three of these, B′-1, B′-2 and B′-3, are present in the 3.65 kb fragment A, in which the arrest was detected. When analysed in vitro, a 250 bp fragment containing B′-1, located in the gene yxcC, was clearly shown to bind RTP (in the absence of ppGpp), with a KD similar to that reported previously for RTP binding to IR elements. In contrast, the adjacent B-like sequences, B′-2 and B′-3, failed to give detectable binding. In addition, a DNase I footprinting experiment has shown that RTP protein binds specifically to a sequence exactly covering the B′-1 sequence. Importantly, precise deletion of the B′-1 sequence abolished the plasmid arrest. This indicates the strong possibility that B′-1 is involved in plasmid and chromosomal replication arrest observed in vivo and, therefore, very probably constitutes an essential part of LSTer2.
Understanding the mechanism blocking the replication fork downstream of the origin after induction of the stringent response is of great interest, particularly in relation to the normal chromosomal termination mechanism. Although our results strongly suggest the involvement of a 17 bp B-like site (B′-1) fixing RTP, the adjacent sequence does not include any discernible A site. Thus, a 250 bp fragment containing B′-1 was apparently only able to bind one RTP dimer in vitro, and a DNase I footprinting experiment using this fragment showed protection of B′-1 with no adjacent A sequence. In contrast to the binding of two dimers to an A, B site (28 bp) required for replication arrest at the terminator, our results support the idea that binding of a second, immediately adjacent dimer of RTP is not necessary to block replication forks at STer sites. However, previous studies (Smith and Wake, 1992; Sahoo et al., 1995a; Manna et al., 1996a) have shown that the binding of RTP to a single B site at the chromosomal terminus is not sufficient to impede replisome movement either in vitro or in vivo. In fact, it has been proposed that the binding of a second dimer to the A site provokes a conformational change in the RTP–B site complex, stabilizing the double-stranded structure of the two overlapping DNA sites, and thus inhibiting the replicative helicase (Manna et al., 1996b; Kralicek et al., 1997). Our results therefore raise the possibility that replication arrest at STer sites proceeds via a mechanism differing in some important aspects from chromosomal termination. In this case, helicase inhibition may be achieved by a single dimer bound to a B-like site, facilitated in some way by the action of ppGpp. This small molecule could bind to RTP (or to a protein bound to RTP) and induce a conformational change, resulting in stabilization of the RTP dimer at the STer site, thus enabling inhibition of helicase activity. ppGpp, which has been shown to inhibit the activity of several enzymes (Cashel et al., 1996), might also itself inhibit the activity of a particular component of the replisome in the presence of RTP. In agreement with the single dimer hypothesis, a mutation described by Manna et al. (1996a), which prevents the interaction between two RTP dimers (and prevents RTP arresting the replication fork at the normal terminus), still allowed the arrest of DNA replication downstream of the origin, after induction of the stringent response (Fig. 8). Importantly, our data do not rule out other factors that may be required to effect an efficient arrest in addition to a B-like sequence. Indeed, other results indicate that the cloning of B′-1 alone into a plasmid is not sufficient to promote RelA, RTP-dependent replication arrest (data not shown). Additional factors necessary for fork arrest could include other as yet unidentified DNA sequences in the vicinity of STer sites, for example, necessary for binding accessory proteins. Moreover, although we have identified and characterized one high-affinity RTP binding site apparently implicated in the arrest at LSTer 2, we have not excluded the possibility of other efficient binding sites in the vicinity, also involved in the arrest mechanism. One possibility might be that a small number of RTP dimers, closely spaced but not necessary interacting, could progressively slow or even halt a replication fork under conditions pertaining to a stringent arrest. As indicated above, the level of local supercoiling or even the relative activity of transcription through STer sites could also influence the efficiency of the arrest mechanism. It is interesting to note that chromosomal transcription in the LSTer2 region (see Fig. 5) is predicted to be exclusively in the direction opposite to that of replication. The absence of such opposing transcription, as also pointed out by Mohanty et al. (1996), may be an important factor in maintaining stable RTP–DNA interactions.
In all probability, it is important that there are major differences in the requirements for RTP to block the replisome, presumably via the helicase, at the normal terminus or at the stringent termini, if the danger of a replication arrest during a normal replication round is to be avoided. Specific differences in the nature of the RTP binding site and the mechanism of RTP–DNA stabilization at terminator and STer sites would therefore seem to be appropriate. It is tempting to conclude that ppGpp plays an important role in imparting specificity for fork arrest at STer sites. In this way, the alarmone concentration in the cell could provide a sensing mechanism designed as a checkpoint control to allow continued replication or fork arrest according to the prevailing nutritional status of the medium. Interestingly, recent studies on Mixococcus xanthus (Harris et al., 1998) indicate that high levels of ppGpp induce the formation of fruiting bodies, therefore supporting the idea of ppGpp as a developmental trigger. Another example of a phosphorylated nucleotide acting as a cell cycle regulator is AP4A, recently reported to be implicated in the timing of cell division in E. coli (Nishimara et al. (1997).
Bacterial strains, plasmids and growth conditions
Strains used in this study are shown in Table 1. OMG219 was derived from SU230, trpC2, neo spac-1 rtp (Table 1), a B. subtilis strain overexpressing rtp constitutively, previously constructed by Smith and Wake (1992). However, efficient use of such a construct requires the presence of lacI. Therefore, the lacI gene under the control of penicillinase promoter from B. licheniformis was inserted at the amyE locus. The strain was constructed as follows: the HindIII–BamHI fragment of pDG148 (Table 1) containing ppen lacI was inserted into the polylinker of pDG364 (Table 1). This plasmid contains the amyE gene disrupted by the cat gene. The BamHI and HindIII sites used to insert lacI were located between the cat gene and the front part of amyE. The recombinant plasmid, pOMG3, was linearized and used to transform the MO1099 strain (Table 1) carrying an erm cassette at the amyE locus. Selection was made for CamR clones, followed by screening for EmS. The chromosomal DNA of strain OMG210, amyE:: cat lacI, was used to transform SU230. The transformants were selected for neomycin and chloramphenicol resistance. The phenotype Amy− was confirmed. This strain was designated OMG211. OMG219 was derived from OMG211 by transformation with chromosomal DNA from HVS567 (Table 1). OMG219 was checked for its hypersensitivity to UV radiation and mitomycin C to confirm its RecA− status. Western blotting experiments demonstrated that the construct OMG219 produced the expected levels of RTP. The production of RTP increased proportionally to the concentration of IPTG added to the culture, with 0.5 mM IPTG giving maximal overexpression of rtp, about 50-fold higher than in the wild type.
Δrtp was introduced into OMM243 by transformation with chromosomal DNA from SU171, selecting for chloramphenicol resistance, to produce OMG251. Similarly, rtp from SU230, under the control of a spac-1 promoter, was introduced into OMM243, selecting for neomycin resistance, thus resulting in OMG271.
Strain OMG257 was derived from SU358 (provided by G. Wake) bearing the Y88F mutation in the rtp gene under the control of the spac-1 promoter at the amyE locus, disrupted by the cat gene. The cat gene was substituted for erm by homologous recombination with the plasmid pCm::erm (ECE 72 from Bacillus Genetic Stock Centre). The chromosomal DNA of this new strain was used to transform OMG250. The transformants were selected for resistance to erythromycin and chloramphenicol. The phenotype Amy− was confirmed. This strain was designated OMG257.
Bacteria were grown at 37°C in a shaking water bath in minimal Spizizen medium (Spizizen, 1958) supplemented with 0.5% w/v glucose, 0.02% w/v casein hydrolysate, 40 μg ml−1 tryptophan. In the case of the strains derived from OMM242 carrying the mutations dnaB37, thyA and thyB, cultures were grown at 30°C in the same medium supplemented with 5 μg ml−1 thymine. Transformation of B. subtilis was carried out according to the procedure of Kunst et al. (1994). The mixture was plated on nutrient broth with appropriate antibiotics (5 μg ml−1 chloramphenicol, 2 μg ml−1 neomycin, 2 μg ml−1 erythromycin and 10 μg ml−1 tetracycline).
pOMG12-a is derived from pWS64-2 and contains the 7.75 kb LSTer 2 region deleted for 18 bp encompassing the B′-1 sequence. The construction was performed as follows: a 1 kb fragment extending from the beginning of the A fragment (proximal to oriB ) to the sequence 5′-ACACCGTTTGAAGCGTGCGC-3′, just upstream of B′-1, was amplified by polymerase chain reaction (PCR). The oligonucleotides were designed to create EcoRI and BamHI restriction sites at the extremity of the PCR product, which were used to insert the fragment in pWS64-2. The resulting plasmid, pOMG11, was digested with BamHI and XbaI and ligated with a PCR product of 6.73 kb, starting immediately downstream of B′-1 (5′-ATAGGCGAGCAAATCCCGGTGAC-3′) and terminating at the end of the C fragment. The recombinant plasmid, pOMG12, was transformed into B. subtilis strain OMG219. pOMG12 carries the same insert as pOMG4-a, except that the B′-1 sequence (18 bp) was deleted and was substituted by a BamHI restriction site.
Estimation of the amount of DNA synthesized
DNA was uniformly labelled by cultivation at 30°C for at least 20 generations in the presence of [methyl-3H]-thymidine at a concentration of 7.4 kBq ml−1 (0.2 μCi ml−1). TCA-precipitable radioactivity was measured as described previously (Laurent and Vannier, 1973). DNA was also measured by fluorescence, essentially according to Boulanger et al. (1997). DNA was obtained from 1 ml of culture after lysozyme and SDS treatment of cells in 0.4 ml of TENsac solution (10 mM Tris-HCl, 10% sucrose, 25 mM NaCl). DNA solution (10 μl) was diluted in water to 100 μl and added to 900 μl of a solution of the fluorescent probe YO-PRO-1 (Molecular Probes; 2 μg ml−1 in H2O) in a quartz cuvette. Excitation and emission wavelengths were set at 490 nm and 509 nm respectively. The signal was measured in a spectrofluorimeter (Kontron Instruments).
Radiolabelling of DNA
DNA replication of B. subtilis was synchronized as described previously (Levine et al., 1987). In brief, OMM242 was grown at 30°C to a density of 3 × 107 cells ml−1 and shifted to 45°C for 30 min. The culture was then returned to 30°C for subsequent analysis. Under these conditions, excellent synchrony was observed. as shown by measuring the rate of DNA synthesis (Levine et al., 1991). For the purpose of measuring chromosomal replication, cells were labelled for the times indicated in the text with [methyl-3H]-thymidine at a concentration of 0.74 MBq ml−1 (20 μCi ml−1).
Hybridization of radioactive chromosomal DNA to DNA probes immobilized on filters
The experiments were carried out according to Levine et al. (1995). Probe DNA was loaded onto nitrocellulose filters (Schleicher & Schuell). Each filter contained about 200 ng of DNA per kilobase of probe. Blank filters contained 100 μg of denatured calf thymus DNA. The labelled chromosomal DNA was extracted and sonicated to reduce DNA fragments to a length of 400 bp. A fixed amount of labelled chromosomal DNA was denatured and allowed to reassociate with the excess of probe DNA fixed on the different filters for 4 days at 42°C. After washing, the radioactivity of each filter was determined. Background values given by calf thymus DNA filters corresponded to 2–5 × 10−5 of the inputs.
Isolation of the intermediates of plasmid DNA replication
Strain OMG219, transformed by the different plasmids, was grown in minimal medium at 37°C to an A570 of 0.3, diluted in the same medium and divided into four parts. The expression of rtp was induced in two of them by the addition of IPTG (1 mM). The different cultures were grown for 2 h and then Arg-HX (100 μg ml−1) was added to one of the two IPTG-treated cultures and to a non-induced culture. The concentration of Arg-HX used to induce the stringent response in OMG219 was reduced because this strain is hypersensitive to the drug, probably because of its RecA− status. After 20 min, cells were harvested, centrifuged for 10 min at 10 000 g, washed twice in TSE buffer (10 mM Tris-HCl, pH 8.0, 300 mM NaCl, 10 mM EDTA) and stored at −80°C. Extraction of plasmid DNA was performed by the neutral lysis method (Horiuchi et al., 1987). In order to enrich for replication intermediates in the preparation, 10 μg of plasmid DNA was digested by the appropriate restriction enzymes and then applied to a benzoylated naphthoylated DEAE (BND) cellulose column as described by Dijkwel et al. (1991). The principle behind the use of this resin is based on its capacity to retain the molecules containing single-strand DNA. Degradation of branched DNA molecules by the mung bean nuclease was carried out as follows: a sample enriched in replication intermediates containing 2 μg of DNA was precipitated with ethanol. The pellet was washed with ethanol (70%), dried and resuspended in 200 μl of 30 mM sodium acetate, pH 4.6, 50 mM NaCl, 1 mM zinc acetate, 0.001% Triton X-100 and divided into two parts. Mung bean nuclease (0.05 units) was added to one part, and both samples were incubated at 37°C for 10 min. The reaction was stopped by the addition of 1 μl of 10% SDS.
The Western blotting method was carried out as reported previously (Levine et al., 1995). Total protein extract (30 μg) was separated by electrophoresis on SDS–PAGE (17% N,N′-methylenebisacrylamide ratio 37:1) for 3 h at 25 mA and transferred to nitrocellulose membrane, according to the procedure of Otter et al. (1987). After blocking the unreacted sites, the membrane was incubated with rabbit antiserum against RTP provided by Professor R. G. Wake (1:1000 dilution in PBS) for 60 min at room temperature. The membrane was then washed for 4 × 10 min in PBS containing 0.1% Tween 20. The protein A–alkaline phosphatase conjugate (1:1000 dilution in PBS) was added. After 60 min incubation at room temperature, the membrane was washed for 4 × 10 min in PBS–Tween 0.1% and for 10 min in the detection reaction buffer (100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 100 mM MgCl2). Revelation of the signal was performed in the same buffer with nitroblue tetrazolium (50 mg ml−1 in 50% dimethyl formamide) and 5-bromo-4-chloro-3-indolyl phosphate (25 mg ml−1 in 25% dimethyl formamide). Once the desired band intensity was obtained, the reaction was stopped by placing the membrane in distilled water.
Source of RTP
RTP was purified in the laboratory of Professor R. G. Wake after the removal of nucleic acids, essentially by CM Sephadex chromatography followed by FPLC MonoS chromatography (Lewis et al., 1990).
Gel retardation assays and DNase I footprinting
The PCR products containing selected sequences were 5′ end-labelled with T4 polynucleotide kinase (Promega) and [γ-32P]-ATP. The binding reactions were performed according to Smith et al. (1996) with slight modifications. DNA fragments (0.2 pmol) were incubated with increasing amounts of RTP (0, 0.4, 0.8, 1.6, 3.2 and 6.4 pmol of RTP monomers) in a total volume of 20 μl containing 50 mM Tris-HCl (pH 7.8), 10% glycerol, 10 mM MgCl2, 50 mM KCl, 1 μg of herring sperm DNA and 200 μg of BSA. The mixtures were incubated for 30 min at 23°C, and 10 μl was loaded on a 6% polyacrylamide gel in 36 mM Tris-HCl (pH 7.5), 30 mM NaH2PO4. The gel was run at 7 V−1 cm and visualized by autoradiography. In order to compare the affinity of RTP to the B′-1 sequence and to the B site of TerI, gel retardation assays were performed using 0.1 pmol of DNA and a wide range of RTP protein amounts (from 0 to 11.2 pmol). The 239 bp DNA fragment containing the B site of TerI was obtained by PCR using pWS69-1 (Smith and Wake, 1992) as substrate. Dried gels were read with the PhosphorImager (Storm 860) and processed with imagequant 4.2 software (Molecular Dynamics). DNase I footprinting experiments were performed essentially as described by Lewis et al. (1990), except that 10% glycerol was used.
We would like to thank Professor Barry Holland for critical reading of the manuscript, and Dr Laurent Jannière for useful discussions. We are particularly grateful to Professor R. Gerry Wake for providing plasmids pWS64-2, pWS65-2, the purified RTP protein and strain SU358. These studies were supported by grants from Centre National de la Recherche Scientifique, EEC (Science Stimulation no. SCI-CT91-0713), Human Frontier Science Program (no. RG-386/95), Association pour la Recherche sur le Cancer and Ligue Nationale Française contre le Cancer. S.A. wishes to acknowledge the receipt of a fellowship from Ministère de l'Enseignement Supérieur et de la Recherche and from Association pour la Recherche sur le Cancer.
- 91996) The stringent response. In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology, 2nd edn. Neidhardt, F.C., Ingraham, J.L., Low, K.B., Magasanik, B., Schaechter, M., and Umbarger, H.E. (eds). Washington, DC: American Society for Microbiology Press, pp. 1410–1438., , , (