|Home | About | Journals | Submit | Contact Us | Français|
Gyrase-mediated hypernegative supercoiling is one manifestation of R-loop formation, a phenomenon that is normally suppressed by topoisomerase I (topA) in Escherichia coli. Overproduction of RNase HI (rnhA), an enzyme that removes the RNA moiety of R-loops, prevents hypernegative supercoiling and allows growth of topA null mutants. We previously showed that topA and rnhA null mutations are incompatible. We now report that such mutants were viable when RNase HI or topoisomerase III was expressed from a plasmid-borne gene. Surprisingly, DNA of topA null mutants became relaxed rather than hypernegatively supercoiled following depletion of RNase HI activity. This result failed to correlate with the cellular concentration of gyrase or topoisomerase IV (the other relaxing enzyme in the cell) or with transcription-induced supercoiling. Rather, intracellular DNA relaxation in the absence of RNase HI was related to inhibition of gyrase activity both in vivo and in extracts. Cells lacking topA and rnhA also exhibited properties consistent with segregation defects. Overproduction of topoisomerase III, an enzyme that can carry out DNA decatenation, corrected the segregation defects without restoring supercoiling activity. Collectively these data reveal 1) the existence of a cellular response to loss of RNase HI that counters the supercoiling activity of gyrase and 2) supercoiling-independent segregation defects due to loss of RNase HI from topA null mutants. Thus RNase HI plays a more central role in DNA topology than previously thought.
Negative DNA supercoiling is a regulated feature of bacterial chromosomes (Menzel and Gellert, 1983; DiNardo et al., 1982; Pruss et al., 1982; Drlica, 1992) that is important for most processes involving DNA strand separation. In E. coli, supercoiling regulation is achieved largely by DNA topoisomerases with opposing enzymatic activities. DNA gyrase, encoded by gyrA and gyrB, introduces negative supercoils, while two other enzymes, DNA topoisomerase I (topA) and DNA topoisomerase IV (parC and parE), remove excess negative supercoils (Pruss et al., 1982; Zechiedrich et al., 2000). Although the ability to create topA null mutants is a clear indication that topoisomerase I is not required for survival (Sternglanz et al., 1981; Stupina and Wang, 2005), growth of such mutants often requires compensatory mutations that reduce negative supercoiling (DiNardo et al., 1982; Pruss et al., 1982). Indeed, very high levels of supercoiling (hypernegative supercoiling, detected as plasmid topoisomers that cannot be resolved by electrophoresis in agarose gels (Pruss, 1985)) can accumulate in topA null mutants (Pruss, 1985) and correlate with growth inhibition (Massé and Drolet, 1999a,b).
One major pathway for generating hypernegative supercoiling in topA mutants involves R-loop formation coupled to gyrase activity (Massé and Drolet, 1999c). Hypernegative supercoiling is initiated by the accumulation of negative supercoils behind moving RNA polymerase when topoisomerase I is absent (Massé and Drolet, 1999c). These negative supercoils promote re-hybridization of nascent transcripts to the template DNA strand, which leaves the non-template strand unpaired and the DNA relaxed. Relaxed DNA is a substrate for gyrase, which then introduces more supercoils that lead to increased R-loop formation, more DNA relaxation, and increased gyrase action (Drolet et al., 1994; Phoenix et al., 1997; Drolet, 2006). Removal of RNA-DNA hybrids by overproduction of RNase HI suppresses the accumulation of hypernegative supercoils and the associated growth defect (Drolet et al., 1995; Massé and Drolet, 1999c). Growth inhibition and hypernegative supercoiling also correlate with major gene expression defects, as illustrated by the accumulation of truncated RNAs (Baaklini et al., 2004; Baaklini et al., manuscript submitted). Within this context, our inability to construct double topA rnhA null mutants did not come as a surprise (Massé and Drolet, 1999b; Drolet et al., 1995). Indeed, it was predicted that such double mutants would die from toxic levels of hypernegative supercoiling.
In the present work, we constructed topA rnhA double mutants that grew when rnhA was expressed from a plasmid-borne gene. Unexpectedly, we found that under non-permissive conditions the DNA of such mutants was relaxed, not hypernegatively supercoiled. Depletion of RNase HI activity in topA null mutants triggered a cellular response that inhibited supercoiling by gyrase. Moreover, the growth inhibition of topA rnhA null double mutants correlated with phenotypes reminiscent of segregation defects: extensive cell filamentation with abnormal nucleoid structures and accumulation of anucleate cells. Overproduction of topoisomerase III (topB), a type IA enzyme like topoisomerase I, corrected the segregation defects and allowed topA rnhA double mutants to grow without restoring supercoiling. Such activity was consistent with topoisomerase III being able to decatenate daughter DNA molecules (Hiasa et al., 1994) and use R-loops as a substrate (Broccoli et al., 2000). Thus high-level expression of a plasmid-borne topB gene provided a second way to allow conditional growth of topA rnhA double mutants. Below we describe measurements of bacterial DNA supercoiling and cell morphology that reveal an important role for RNase HI in preventing transcription from perturbing DNA topology beyond viable bounds.
We previously described the effects of topoisomerase I defects in living cells using a set of E. coli strains in which a temperature-sensitive gyrase mutation allows the growth of topA null cells at 37°C but not at 28°C owing to reactivation of gyrase (Drolet et al., 1995). In such strains hypernegatively supercoiled DNA accumulates following the temperature downshift unless RNase HI is overproduced (Massé and Drolet, 1999c). Overproduction of RNase HI at 28°C also restores growth. To better understand relationships among hypernegative supercoiling, R-loops, and growth inhibition, we constructed a conditional topA rnhA double mutant (transduction was used to introduce an rnhAcam allele into a topA20Tn10 gyrB(Ts) strain carrying a plasmid expressing rnhA under control of the arabinose-inducible PBAD promoter). Transductants were obtained by plating at 37°C in the presence but not in the absence of arabinose (not shown). When these transductants were transferred to liquid medium lacking arabinose, cells grew slowly, cell filamentation occurred, and culture growth stopped prematurely, irrespective of temperature (not shown). Thus, when RNase HI is absent, the growth defects of topA null gyrB(Ts) mutants were seen even at 37°C. Since colonies arose after plating on solid medium containing arabinose, we conclude that deleting rnhA in topA mutants is bacteriostatic, not bactericidal (not shown).
Supercoiling was measured with plasmid pGB2nusBkan, a low copy-number derivative of pSC101 that is prone to R-loop-dependent hypernegative supercoiling in topA null mutants (Broccoli et al., 2004). Plasmid topoisomers were resolved using agarose gel electrophoresis in the presence of 7.5 µg/ml of chloroquine, as previously described (Massé and Drolet, 1999c). Under this condition the more relaxed topoisomers migrated faster; hypernegatively supercoiled DNA also migrated rapidly. With a gyrB(Ts) control strain lacking only topA (strain PS152), hypernegatively supercoiled DNA accumulated after transfer of cells from 37 to 28°C (Fig. 1a, lane 2, indicated by [--]). When an rnhA null mutation was added (strain PS151, which also carried an inducible rnhA gene), and mutant cells from arabinose-containing agar were incubated overnight at 37°C in the absence of arabinose, followed by 1:1000 dilution into medium lacking arabinose, cell growth was sufficient for examination of plasmid supercoiling. A portion of the DNA was more relaxed at 37°C than in topA single mutant cells (Fig. 1a, compare lanes 1 and 3), and hypernegatively supercoiled DNA failed to accumulate following a transfer to 28°C (Fig. 1a, lane 4). The relaxed portion of topoisomers remained extensively relaxed following the temperature downshift (Fig. 1a, lanes 3 and 4, indicated by [rel]). These extensively relaxed topoisomers were not observed in wild-type (lanes 5 and 6) or in gyrB(Ts) (lanes 7 and 8) cells. Thus, removal of rnhA from a topA-deficient strain caused DNA relaxation and loss of hypernegative supercoiling.
When arabinose was added to the diluted (1:1000) overnight culture of the topA rnhA double mutant (strain PS151) to produce RNase HI from a plasmid-borne gene, supercoiling was restored at 37°C (Fig. 1a, compare lane 3, no arabinose, with lane 9, 0.05% arabinose) and accumulation of hypernegatively supercoiled DNA occurred following a temperature downshift (Fig. 1a, lane 10). Since RNase HI was overproduced in this experiment, topA null cells accumulated lower amounts of hypernegatively supercoiled DNA than observed with topA null cells having only wild-type levels of RNase HI (Fig. 1a, compare lanes 2, topA null cells, and 10, topA rnhA double mutant with 0.05% arabinose).
Two-dimensional gel electrophoresis confirmed the presence of hypernegatively supercoiled and extensively relaxed DNA, respectively, in the presence and absence of RNase HI (Fig. 1b; hypernegatively supercoiled DNA is seen at bottom part of the curves at high chloroquine concentration ((right panels, indicated by [--]); extensively relaxed DNA is at the far right part of the curves at lower chloroquine concentration (left and middle panels; lines were traced over the more relaxed topoisomers)). Additional two-dimensional gel analyses showed that the more relaxed topoisomers were not positively supercoiled (data not shown). Thus, removal of RNase HI relaxed DNA of topA null cells and prevented the accumulation of hypernegatively supercoiled DNA rather than increasing it.
Excess negative supercoiling in topA mutants originates largely from transcription. To determine whether the reduced supercoiling in the topA rnhA double mutant is specific to transcription-induced supercoiling, mutant cells (strain PS151) were treated with rifampicin, an inhibitor of RNA polymerase. When rifampicin was added to a culture of topA null cells (strain PS152, containing wild-type RNase HI) before the temperature downshift, almost all of the hypernegatively supercoiled topoisomers were lost (Fig. 2, top panels). When RNase HI was absent and most of the DNA was seen as extensively relaxed topoisomers, rifampicin had little effect (Fig. 2, bottom panels; a line was traced over the more relaxed topoisomers). Thus the extensive relaxation of supercoiling seen in topA rnhA double mutants was related to global supercoiling rather than being specifically linked to transcription. However, our results also show that the effect of the rnhA deletion on supercoiling following a temperature downshift was much stronger when transcription was allowed (Fig. 2, compare −rif with +rif, + and − RNase HI).
Changes in topoisomerase IV and gyrase concentrations were measured to address the possibility that levels of these proteins account for the extensive DNA relaxation associated with depletion of RNase HI activity in topA null cells. ParC and ParE were not overproduced in the topA20Tn10 gyrB (Ts) mutant when RNase HI was depleted, as assessed by western blot experiments (Fig. 3, compare lanes 5 and 6 with lanes 7 and 8) and northern blot experiments (data not shown). GyrA and GyrB levels also exhibited no significant difference between the presence and absence of RNase HI (Fig. 3, compare lanes 5 and 6 with lanes 7 and 8).
Lack of significant change in ParC, ParE, GyrA and GyrB level was also observed when the rnhA gene was inactivated in the ΔtopA gyrB (Ts) strain carrying pPH1243 (Fig. 3, lanes 1 to 4). This plasmid, which expressed topB (topoisomerase III) from an IPTG-inducible promoter, allowed the recovery of rnhAcam transductants in a ΔtopA recipient strain, but only when IPTG was present (Fig. 4a and data not shown). Western blot experiments with a transductant (strain SB383) confirmed that overproduction of topoisomerase III occurred (Fig. 4b). Plasmid pPH1243 also allowed topA rnhA double mutants to grow unless cultures were diluted into IPTG-deficient medium, in which case growth eventually stopped during log phase. Thus, introduction of two restrictive conditions, depletion of RNase HI or removal of topoisomerase III overexpression failed to reveal changes in gyrase and topoisomerase IV concentrations that could explain the extensive DNA relaxation observed with topA rnhA double mutants.
We also considered the possibility that supercoiling by gyrase was reduced. To test this hypothesis, we used the gyrAL83 allele to render gyrase resistant to norfloxacin, thereby allowing selective quinolone-mediated inhibition of topoisomerase IV, the only remaining relaxing activity in topA mutants (Zechiedrich et al., 2000). The gyrAL83 allele was introduced into isogenic ΔtopA gyrB (Ts) and ΔtopA gyrB (Ts) rnhAcam strains carrying pPH1243 to obtain, respectively, strains VU21 and PS160. Dilution into medium lacking IPTG gradually removed suppression of growth defect by topoisomerase III, but allowed enough growth for examination of plasmid supercoiling (as pointed out in a subsequent section, topoisomerase III overexpression has no effect on supercoiling).
Negative supercoiling increased at 37°C after inhibition of topoisomerase IV by norfloxacin in both the topA rnhA and the topA mutants following dilution (Fig. 5, compare topA rnhA, strain PS160, lanes 1 and 2 with topA, strain VU21, lanes 5 and 6). However, a significant proportion of the topoisomers remained relaxed in the topA rnhA mutant (Fig. 5, lane 2, indicated by [rel]). Following a shift to 28°C, the addition of norfloxacin to inhibit topoisomerase IV strongly stimulated the accumulation of hypernegatively supercoiled DNA in the topA mutant (Fig. 5, compare lanes 7 and 8, indicated by [--]), whereas it did not appreciably change the supercoiling level in the topA rnhA mutant (Fig. 5, compare lanes 3 and 4). Collectively these results show that topoisomerase IV is not responsible for the strong supercoiling deficit in topA rnhA double mutants. It is likely that supercoiling by gyrase is impaired when both RNase HI and topoisomerase I are absent.
GyrI is encoded by sbmC, a well-characterized inhibitor of gyrase (Nakanishi et al., 1998; Chatterji and Nagaraja, 2002). Since GyrI is part of the SOS regulon (Baquero et al., 1995), which is chronically expressed in rnhA mutants (Kogoma et al., 1993; McCool et al., 2004), it is expected to be at elevated levels in topA rnhA double mutants. To examine involvement of GyrI in the supercoiling inhibition in topA rnhA double mutants, an sbmCkan allele was introduced into isogenic ΔtopA gyrB(Ts) and ΔtopA gyrB(Ts) rnhAcam strains carrying pPH1243, to obtain, respectively, strains VU64 and VU70. Both extensive DNA relaxation and lack of hypernegative supercoiling, which are characteristic of the topA rnhA double mutant, were seen whether or not sbmC was present (Fig. 6, compare lanes 1 and 4 with lanes 2 and 5, respectively). The lack of a significant supercoiling effect from the absence of GyrI was also illustrated by comparing supercoiling in strains VU64 (Fig. 6, rnhA+, lanes 3 and 6) and VU70 (Fig. 6, rnhA−, lanes 2 and 5). Thus, GyrI is not responsible for the DNA relaxation seen in topA rnhA double mutants. We also examined a lexA3 mutant in which the SOS regulon is non-inducible. DNA was relaxed, and hypernegative supercoiling was absent in a lexA3 topA rnhA mutant (not shown).
The restoration of negative supercoiling is slow and requires a significant increase in the population cell density of strain PS151 (topA rnhA double mutant carrying pBADrnhA) following the addition of arabinose to produce RNase HI (not shown). This result suggested involvement of one or more stable factors in inhibiting supercoiling in topA rnhA double mutants. To test this idea, we prepared extracts of wild-type (strain AQ634), gyrB(Ts) (strain MA249), gyrB(Ts) topA (strain PS152) and gyrB(Ts) topA rnhA (strain PS151) cells. Extracts from both wild-type and gyrB(Ts) strains exhibited supercoiling activity (Fig. 7a), although with the gyrB(Ts) strain an endonuclease activity was also detected (lane 6). Maximum supercoiling activity was seen at 3.5 µg total protein in the reaction mixture for the gyrB(Ts) strain (lane 9); the same amount of protein from wild-type cells exhibited more supercoiling activity (compare wild-type, lane 4 with gyrB(Ts), lane 9), and a higher supercoiling level was achieved (not shown). Although supercoiling activity was also detected in crude extracts of topA null gyrB(Ts) cells (Fig. 7b, lanes 7 to 12, top panel), it was significantly lower than in extracts of isogenic topA+ gyrB(Ts) cells (Fig. 7a, lanes 6 to 10) and was better seen when the gel was probed with a [32P]-labelled DNA fragment (Fig 7b, lanes 7 to 12, bottom panel). Even with radioactive probing, supercoiling activity was undetectable in extracts from the topA rnhA double mutant grown in the absence of arabinose to deplete RNase HI activity (Fig. 7b, lanes 1 to 6, bottom panel). These results are consistent with crude extracts of topA rnhA null cells containing factor(s) that significantly reduce supercoiling activity.
As a further test for factor(s) inhibiting supercoiling, aliquots of extracts from both topA and topA rnhA double mutant cells were mixed with extracts from wild-type cells. As shown in Fig. 7c, extracts from the topA rnhA double mutant significantly inhibited supercoiling activity of wild-type extracts (compare lanes 1 and 5, which used 2.5 µg protein from the wild-type cell extract with, respectively, 5 and 0 µg protein from the topA rnhA double mutant extract). In contrast, inhibition of supercoiling activity by the extract from topA null cells was weak (Fig. 7c, compare lanes 1 and 6, using 2.5 µg protein from the wild-type cell extract with 5 µg protein from the topA rnhA or the topA null extracts). Thus, a strong inhibitory activity present in extracts of topA rnhA double mutant explains the failure to detect supercoiling activity and may explain, at least in part, the impairment of supercoiling by gyrase in topA rnhA double mutants.
The data described above failed to relate the growth defect of topA rnhA double mutants to excess negative supercoiling. Since in E. coli topoisomerase III is not involved in the regulation of supercoiling (Zechiedrich et al., 2000; Lopez et al., 2005), we expected DNA of topA rnhA double mutants to remain relaxed during suppression of growth defect by overexpression of topoisomerase III. To test this hypothesis we made use of the finding that in a topA null mutant hypernegative supercoiling of pPH1243 is stimulated both by transcription from the strong Ptrc promoter, which is activated by the addition of IPTG, and by translation inhibitors (Broccoli et al., 2004; Broccoli and Drolet, unpublished results; spectinomycin was added 15 min before shifting from 37 to 28°C). In one experimental set, IPTG was added throughout growth at 37°C to overexpress topoisomerase III; in the second set, IPTG was added after spectinomycin so that the transcription effect of IPTG on pPH1243 supercoiling could be observed in the absence of excess topoisomerase III. Fig. 8 shows that pPH1243 exhibited hypernegative supercoiling upon adding IPTG, either before (lane 2) or after (lane 6) the addition of spectinomycin to the topA null mutant. In contrast, relaxation of pPH1243 was observed whether or not topoisomerase III was overproduced and irrespective of the presence of spectinomycin in the topA rnhA double mutant (Fig. 8, lanes 3, 4, 7 and, 8). Two-dimensional gel electrophoresis confirmed relaxation of pPH1243 (data not shown). As expected, supercoiling was restored in this topA rnhA double mutant following the introduction of a plasmid carrying the rnhA gene (data not shown), indicating that relaxation is reversible and due to the absence of RNase HI. Collectively, these data support the assertion that overexpression of topoisomerase III suppresses the growth defect of topA rnhA double mutants without restoring supercoiling.
Despite the presence of a mechanism to prevent hypernegative supercoiling, topA rnhA double mutants do not grow without suppression. This result suggests the presence of supercoiling-independent mechanism(s) by which the absence of RNase HI dramatically perturbs cell physiology. As shown above, the growth defect of a topA rnhA double mutant can be rescued by overproducing topoisomerase III (Fig. 4), which can allow chromosome segregation by removing precatenanes during replication (Nurse et al., 2003). We therefore examined the possibility that the growth inhibition of topA rnhA double mutants might be related to segregation defects.
Cells were stained with DAPI and prepared for fluorescence microscopy such that both cell morphology and DNA content could be examined. In the absence of topoisomerase III overproduction, cells of the topA rnhA double mutant (strain SB383) formed long filaments full of DNA, and nucleoid structures were altered (Fig. 9a, panel 2). Anucleate cells also accumulated (Fig. 9b, yellow arrows), as did cells with very low amounts of chromosomal DNA (Fig. 9b, green arrows). Formation of such cells could result from septum closure on nucleoids, a typical manifestation of partition defects (guillotine effect; Niki et al., 1991). Clearly, the topA rnhA double mutant fails to produce a significant number of normal cells. However, more than half of the cells containing aberrant nucleoid structures likely remain viable, as cells of strain SB383 incubated in liquid medium lacking IPTG form colonies when plated on solid media containing IPTG to overproduce topoisomerase III (data not shown). Overproduction of topoisomerase III reduced both cell filamentation and DNA content, and it caused DNA to be more regularly distributed within the cells (Fig. 9a, panel 1). Topoisomerase III overproduction also reduced the number of anucleate cells by more than half. Segregation defects were also apparent in the topA20Tn10 rnhAcam gyrB (Ts) mutant carrying pBADrnhA (strain PS151; Fig. 9a, panel 4, -ara); they were corrected by the addition of arabinose to produce RNase HI (panel 3, +ara). Moreover, cells of single rnhA and topA null mutants failed to show major morphological or nucleoid defects (Fig. 9c, panels 2 and 3 respectively). Thus, combinations of topA and rnhA null mutations lead to phenotypes reminiscent of segregation defects, thereby explaining the growth inhibition phenotype of topA rnhA double mutants.
The work described above, which focused on the properties of topA rnhA double mutants, revealed new features of how DNA topology is regulated in bacterial cells. One level concerns DNA supercoiling. Basal supercoiling, which is defined as supercoiling in the absence of transcription (Drlica et al., 1988), is set largely by the supercoiling activity of gyrase opposed by the relaxing activities of topoisomerase I and topoisomerase IV. Transcription, through the generation of R-loops coupled with gyrase activity, produces hypernegative supercoiling that is countered by topoisomerase I and RNase HI. The growth defect associated with a deficiency of topoisomerase I is suppressed by partially defective gyrase, overexpression of topoisomerase IV, or overexpression of RNase HI. Surprisingly, topA rnhA null double mutants not only lacked hypernegative supercoiling but also contained DNA that was relaxed (Fig. 1 and Fig. 8). Evidence was found for the interesting possibility that RNase HI normally regulates factor(s) impairing supercoiling by gyrase. A second level of topology control concerns decatenation of replicated chromosomes. Decatenation is largely a function of topoisomerase IV (Zechiedrich and Cozzarelli, 1995). In our work, a topA rnhA double mutant exhibited a defect in chromosome segregation that was corrected by overexpression of topoisomerase III (Fig. 9), an enzyme that has no effect on supercoiling and is capable of decatenation both in vitro and in vivo (Hiasa et al., 1994; Nurse et al., 2003). Thus RNase HI, either directly or indirectly, influences both basal levels of supercoiling and chromosome segregation as discussed in the following sections.
Previous work showed that gyrase-mediated hypernegative supercoiling occurs during transcription in vitro in the absence of RNase HI (Drolet et al., 1994; Phoenix et al., 1997) and in topA null mutants when RNase HI is not overproduced (Massé and Drolet, 1999c). We expected removal of rnhA from topA mutants to lead to extensive hypernegative supercoiling, since cells lacking both topoisomerase I and RNase HI cannot be constructed (Massé and Drolet, 1999c; Drolet et al., 1995). Unexpectedly, hypernegative supercoiling failed to accumulate in topA rnhA double mutants that had been constructed by conditional expression of rnhA or topB. However, DNA of such mutants was more relaxed than usual. Relaxation was not an indirect consequence of cell death, because deleting rnhA in topA null mutants was bacteriostatic, not bactericidal. When arabinose was added to non-growing topA rnhA null cells carrying pBADrnhA to induce the synthesis of RNase HI, both growth and supercoiling were restored. More importantly, DNA remained extensively relaxed in a topA rnhA double mutant when cell growth was restored by overproducing topoisomerase III. Only when a plasmid carrying rnhA was introduced into this mutant that supercoiling was also restored. Therefore, DNA relaxation correlated with depletion of RNase HI activity in topA mutants.
Since supercoiling level is set primarily by opposing topoisomerase activities, we considered the possibility that the supercoiling deficit in topA rnhA null double mutants was due either to an excess of DNA relaxation activity by topoisomerases or to a loss of gyrase supercoiling activity. One well established mechanism by which excess negative supercoiling is prevented in topA mutants is the overproduction of topoisomerase IV (Kato et al., 1990; Free and Dorman, 1994). However, western blot experiments showed that topoisomerase IV is not overproduced in topA null cells lacking RNase HI activity. Moreover, inhibition of topoisomerase IV by norfloxacin failed to significantly raise supercoiling in a topA rnhA double mutant carrying the gyrAL83 quinolone-resistance allele. Thus topoisomerase IV was not responsible for the supercoiling deficit in topA rnhA double mutants. In agreement with the results of previous experiments showing that topoisomerase III is not involved in supercoiling regulation in E. coli (Zechiedrich et al., 2000; Lopez et al., 2005), overproducing this enzyme had no effect on supercoiling in either topA single or topA rnhA double mutants. Therefore, relaxation activity of topoisomerases is unlikely to explain the supercoiling deficit in topA null cells depleted of RNase HI activity.
Western blot experiments also demonstrated that levels of gyrase protein were not altered following the depletion of RNase HI activity in topA mutants. Two experiments indicated that the supercoiling deficit is related to impairment of supercoiling by gyrase. First, while inhibiting topoisomerase IV by norfloxacin in a single topA null mutant strongly promoted hypernegative supercoiling, it did not significantly stimulate supercoiling in a double topA rnhA null mutant. Second, supercoiling activity could not be detected in cell extracts of a topA rnhA null double mutant (strain PS151), and these extracts significantly inhibited the supercoiling activity when mixed with extracts from wild-type cells. Thus, the supercoiling deficit in topA null cells depleted of RNase HI activity is likely related to a cellular response that leads to the inhibition of gyrase.
The cellular response leading to the impairment of gyrase activity in double topA rnhA null mutants is currently unknown. The response is related neither to the SOS regulon, which is constitutively induced in the absence of RNase HI (Kogoma et al., 1993; McCool et al., 2004), nor to the presence of the gyrase inhibitor, GyrI, a member of the SOS regulon (Baquero et al., 1995). Previous results also failed to demonstrate an effect of GyrI on supercoiling in vivo (Chatterji et al., 2003). While supercoiling by gyrase can be prevented by a direct interaction between the enzyme and a specific inhibitor acting like GyrI, it may also be inhibited by low [ATP]/[ADP] (Westerhoff et al., 1998; Drlica, 1992) and indirectly by proteins that interact with DNA, such as the abundant nucleoid proteins Fis, H-NS and HU (Travers and Muskhelishvili, 2005a,b). Additional experiments are required to work through the many possible ways in which gyrase activity can be lowered.
The growth inhibition associated with topA rnhA double mutants correlated with phenotypes seen previously with chromosomal segregation defects: extensive cell filamentation, abnormal nucleoid structures, and accumulation of anucleate cells. Thus, the simultaneous absence of topoisomerase I and RNase HI leads to segregation defects not observed with cells lacking only one of the enzymes. R-loops may be involved, since a relationship between topoisomerase I and RNase HI is well established. If R-loops persist in topA rnhA double mutants, particularly at oriK sites where constitutive, stable DNA replication is initiated (Kogoma, 1997), excess replication could occur. Over-replication may saturate the segregation capacity of the cell, thus requiring more decatenation activity than can be provided by excess topoisomerase III.
Unregulated over-replication is also known to lead to collapse of replication forks, DNA double-strand breaks (Kouziminova et al., 2004; Simmons et al., 2004; Michel et al., 2007), and ultimately hyper-recombination, which can cause segregation defects (Lopez et al., 2005; Zhu et al., 2001; Zahradka et al., 1999; Magner et al., 2007). Accumulation of unresolved recombination intermediates interferes with chromosome segregation; resolution of these structures can be performed by topoisomerase III (Lopez et al., 2005; Zhu et al., 2001), even when it is present at normal or very low levels (less than 10 copies per cell; Digate and Marians, 1989).
The extensive DNA relaxation in topA rnhA double mutants could also contribute to segregation defects, since the inhibitory effect of temperature-sensitive gyrase mutations on segregation (Steck and Drlica, 1984) is thought to be due to DNA relaxation that then suppresses decatenation by topoisomerase IV (Zechiedrich et al., 1997; Holmes and Cozzarelli, 2000). Overproduction of topoisomerase III allows chromosome segregation when topoisomerase IV is inactive (Nurse et al., 2003). However, in the context of extensive DNA relaxation, the ability of topoisomerase III to perform decatenation might be reduced because single-stranded DNA regions, the substrate for topoisomerase III, are expected to be infrequent. Such regions can also be provided by R-loops (Broccoli et al., 2000).
The first function attributed to RNase HI in E. coli was a role in the removal of RNA primers of Okazaki fragments (Funnell et al., 1986; Ogawa and Okazaki, 1984). However, this enzyme cannot remove the last ribonucleotides at the RNA-DNA junctions. In fact, 5’-3’ exonuclease activity (e.g. polymerase I) plays the major role in the removal of RNA primers. Interestingly, RNase H activity in various bacterial species was recently shown to be dispensable for complete RNA primer removal (Fukushima et al., 2007). Only 5’-3’ exonuclease activity was shown to be indispensable for this process. Based on the results presented in the present work we propose that a major function of RNase HI, and possibly other bacterial RNase H molecules, involves the control of DNA topology via R-loops.
Escherichia coli strains used in this work are described in Table 1. Strains were constructed by transduction with phage P1vir as previously described (Miller, 1992). When needed, tetracycline (10 µg/ml), chloramphenicol (15 µg/ml) or kanamycin (50 µg/ml) was added to the medium. PCR was used to confirm the presence of only the rnhAcam allele and the sbmCkan allele, respectively, on the chromosome of the rnhA null and the sbmC null transductants.
pBAD18rnhA was constructed by placing an EcoRI-HindIII fragment carrying the rnhA gene under the control of the PBAD promoter of pBAD18 (Guzman et al., 1995). The rnhA gene was obtained by PCR from pSK760 (Drolet et al., 1995) by using d(GTCAGAATTCCAGGAAGTCTACCAGA) and d(GTCAAAGCTTGGCAATGTCGTAAACC) oligonucleotides. pGB2nusBkan is a pSC101 derivative that was constructed by inserting the pUC4K EcoRI fragment carrying a kanamycin-resistance cassette into the ScaI site of pGB2nusB (Friedman et al., 1990). pPH1243 is a pTrc99a derivative carrying the topB gene under the control of the IPTG-inducible Ptrc promoter (Broccoli et al., 2000).
Cells were grown overnight at 37°C in LB medium supplemented with cysteine (50 µg/ml, for the RFM475 derivatives), thymine (10 µg/ml, for the MA251 derivatives) or tryptophan (50 µg/ml, for VU35 and VU95 strains). When required, ampicillin (50 µg/ml), spectinomycin (30 µg/ml), arabinose (0.05%) or IPTG (1 mM) were added. Overnight cultures diluted in pre-warmed medium (37°C), were grown to an OD600 of ~0.5 at which time an aliquot of cells was recovered for plasmid extraction while the remaining culture was transferred to 28°C. Aliquots of cells were recovered for plasmid extraction at the indicated times. Growth was arrested by transferring cells into a tube filled with ice, thus immediately lowering the temperature of the culture to 0°C. Plasmid DNAs were extracted by alkaline lysis as previously described (24). For experiments using derivatives of MA251, overnight cultures were diluted 1:1000; they were diluted to an OD600 of 0.03 for the experiments using SB383, VU21, PS160, VU64, and VU70.
One-dimensional and two-dimensional agarose gel electrophoresis in the presence of chloroquine was performed in 0.5 × TBE as described (Massé et al., 1997). After electrophoresis, the gels were dried and prepared for in situ hybridization with random prime-labelled probes as described (Massé et al., 1997). Images were obtained by using a Phosphorimager Typhoon 9400 (Amersham Biosciences).
The equivalent of 200 µl of cell culture at an OD600 of 0.7 was used for Western blot analysis. The cell pellets were lysed by boiling in sodium dodecyl sulphate (SDS). The proteins were separated by SDS-polyacrylamide (7.5 %) gel electrophoresis. Western blots were performed as described previously (Sambrook et al.,1989) by using nitrocellulose membranes (Hybond-ECL, GE Healthcare). After transfer, the membranes were stained with Ponceau S (Fisher Scientific) to confirm that similar amounts of protein were loaded in each lane. ParC and ParE antibodies were obtained from Dr Kenneth J. Marians (Memorial Sloan-Kettering Cancer Center, New York, New York). GyrA and GyrB antibodies were purchased from John Innes Enterprises Ltd (John Innes Centre, Norwich Research Park Colney, Norwich, UK). TopB antibodies were obtained from Dr Russell DiGate (Department of Pharmaceutical Sciences, Philadelphia College of Pharmacy, Philadelphia, PA). The ECL Plus detection kit (GE Healthcare) was used to reveal the specific proteins.
Cells were grown in LB medium at 37°C to an OD600 of 0.7 and transferred to 28°C for 30 minutes. Cells were recovered and prepared for gyrase assays in crude extracts as previously described (DiNardo et al., 1982). 0.2 µg of relaxed pBR322 DNA (prepared by using wheat germ topoisomerase I from Sigma-Aldrich) was used in the assays.
RFM443, RFM475, PH379 and SB383 cells were grown overnight on LB plates supplemented, when required, with cysteine (50 µg/ml), ampicillin (50 µg/ml), chloramphenicol (10 µg/ml) and/or tetracycline (10 µg/ml). When needed, IPTG (1 mM) was added to the plates of SB383 cells. PS151 cells were grown overnight on LB plates supplemented with thymine (50 µg/ml) and, when needed, arabinose (0.05%). The plates were incubated at 37°C. After overnight growth, cells were resuspended in pre-warmed (37°C) liquid LB medium (supplemented as requested) to obtain a starting OD600 of about 0.01. Cells were grown at 37°C to an OD600 of 0.8. 150 µl of cells were harvested, centrifuged, and resuspended in 77% ethanol (fixing solution). The cells were washed with 500 µl of 0.9% NaCl, centrifuged and resuspended in 100 µl of 0.9% NaCl. 3 µl of the fixed samples were spread on slides pre-treated with a Poly-l-Lysine solution (Sigma) and allowed to air dry at room temperature. 5 µl of slow fade gold antifade reagent with DAPI (4’,6-diamidino-2-phenylindole; from Invitrogen) was deposited on the slides and sealed with a cover glass. Fluorescence pictures were obtained with a Nikon E600 equipped with a 100-W mercury lamp and standard DAPI filters using the X100 oil immersion objective. The images were captured on the computer using the Nikon ACT-1 software. Exposure time was 1/50s at maximum sensitivity. For phase contrast microscopy, the microscope was adjusted to the phase contrast optical system and pictures were taken at X100 oil immersion objective with the Ph3 annulus. Exposure time was 1/120s at normal sensitivity. The images were processed with Adobe Photoshop.
We thank Dr Russell DiGate and Kenneth Marians for antibodies and Dr Nicholas Cozzarelli for LZ1 strain. We also thank Patrick Hallenbeck for editing of the manuscript. This work was supported by grant FNR 12667 from the CIHR to M.D and NIH grant AI35257 to K.D. M.D. was a Chercheur-Boursier Senior from the FRSQ.