The expression of plasmid‐borne virF of Shigella encoding a transcriptional regulator of the AraC family, is required to initiate a cascade of events resulting in activation of several operons encoding invasion functions. H‐NS, one of the main nucleoid‐associated proteins, controls the temperature‐dependent expression of the virulence genes by repressing the in vivo transcription of virF only below a critical temperature (∼32°C). This temperature‐dependent transcriptional regulation has been reproduced in vitro and the targets of H‐NS on the virF promoter were identified as two sites centred around −250 and −1 separated by an intrinsic DNA curvature. H‐NS bound cooperatively to these two sites below 32°C, but not at 37°C. DNA supercoiling within the virF promoter region did not influence H‐NS binding but was necessary for the H‐NS‐mediated transcriptional repression. Electrophoretic analysis between 4 and 60°C showed that the virF promoter fragment, comprising the two H‐NS sites, undergoes a specific and temperature‐dependent conformational transition at ∼32°C. Our results suggest that this modification of the DNA target may modulate a cooperative interaction between H‐NS molecules bound at two distant sites in the virF promoter region and thus represents the physical basis for the H‐NS‐dependent thermoregulation of virulence gene expression.
Bacteria entering the human host encounter an increase of their growth temperature to 37°C; this environmental change represents a cue that can trigger virulence gene expression in human pathogens such as Shigella species and enteroinvasive Escherichia coli (EIEC) (Maurelli and Sansonetti, 1988). These microorganisms cause disease by a similar, complex mechanism of pathogenicity which depends on the expression of chromosomal or plasmid‐borne virulence genes (Sansonetti et al., 1982; Finlay and Falkow, 1997). The virulence genes carried by pINV are organized in regulons coordinately regulated by a central modulator, VirR (Maurelli and Sansonetti, 1988), corresponding to the major nucleoid protein H‐NS (Lammi et al., 1984; Spassky et al., 1984; Pon et al., 1988), which represses their expression at 30°C or at low osmolarity (Tobe et al., 1991, 1993; Dagberg and Uhlin, 1992; Porter and Dorman, 1994). In addition to chromosome‐encoded H‐NS (VirR), two pINV‐encoded activators, the proteins VirF and VirB, are involved in the transcriptional control of the invasion genes (Adler et al., 1989). In a cascade model, VirF activates transcription of the gene coding for the secondary regulator VirB which, in turn, activates several operons encoding the invasion genes. The activation of virB transcription by VirF is highly sensitive to changes in DNA topology and is antagonized at 30°C by H‐NS (Tobe et al., 1993, 1995). Thus, the cellular level of VirF is crucial for the expression of the invasive phenotype of Shigella and E.coli EIEC and, while the expression of virF is not constitutive at 30°C, it is induced at 37°C (Tobe et al., 1991; Colonna et al., 1995) and may also be regulated at the post‐transcriptional level through tRNA modifications (Durand et al., 1994, 1997). Furthermore, virF inactivation or hyperexpression results in a complete loss of the invasive phenotype or in the induction of invasiveness at the non‐permissive temperature, respectively. More recently, it has been shown that a constitutively higher level of virF mRNA and β‐galactosidase expressed from a virF–lacZ fusion can be found in cells with an hns‐defective background (Prosseda et al., 1998). These data suggested that, as in the case of other virulence‐regulating genes such as cfaD (Jordi et al., 1992), virB (Tobe et al., 1993) and papI (Forsman et al., 1992), virF expression might also be negatively controlled by H‐NS in a temperature‐dependent manner. This premise was supported further by data suggesting that this protein may interact with the promoter region of virF (Prosseda et al., 1998).
H‐NS is known to affect, primarily at the transcriptional level, the expression of a fairly large number of genes (reviewed in Atlung and Ingmer, 1997), and although the molecular basis of its regulatory activity probably rests on its preferential interaction with intrinsically curved DNA (Yamada et al., 1990, 1991; Tanaka et al., 1991), its ability to induce bending of non‐curved DNA (Spurio et al., 1997) and its ability to induce negative supercoiling (Tupper et al., 1994), different mechanisms of repression have been proposed (Higgins et al., 1988; Goransson et al., 1990; Hulton et al., 1990; Owen‐Hughes et al., 1992; Falconi et al., 1993; Ueguchi and Mizuno, 1993; Tupper et al., 1994; Barth et al., 1995). To explain how H‐NS may regulate gene expression in a temperature‐dependent manner is even more intriguing. In fact, while induction of gene expression by a temperature increase is not restricted to the pathogenicity genes, but also concerns the heat shock regulon, the participation of a DNA‐binding protein such as H‐NS in thermoregulation is a unique characteristic of the virulence genes. Furthermore, unlike the heat response, which entails a temperature increase above the normal growth temperature and which produces a transient expression of the genes belonging to the regulon, temperature‐dependent induction of the vir regulon in Shigella involves a temperature shift within the normal range of growth temperatures and the stable activation of the regulon. Thus, it is clear that temperature control over virulence expression operated by H‐NS entails a unique mechanism, clearly different from that responsible for the activation of the heat shock regulon.
To explain the thermoregulation in Shigella pathogenicity expression, it is commonly assumed that the temperature control acts via the plasmid‐encoded VirF and VirB proteins and involves an H‐NS and/or environmentally induced alterations in DNA supercoiling. In fact, it has been reported that H‐NS can alter DNA topology constraining negative supercoiling (Tupper et al., 1994). Furthermore, it is known that changes in DNA supercoiling can occur in response to the same environmental factors (e.g. temperature, osmolarity and phase of growth) which influence the expression of both Salmonella and Shigella virulence genes (Galan and Sansonetti, 1996). More specifically, since VirF can bind to virB in the absence of a thermal signal, it has been postulated that H‐NS antagonizes this transcriptional activator only at low temperature via changes in topology (Tobe et al., 1993, 1995). A similar interpretation has been offered for thermoregulation of virulence mediated by histone‐like proteins in Yersinia (Mikulskis and Cornelis, 1994). However, since the effect of hns mutations on the supercoiling of reporter plasmids is not straightforward, being sometimes negligible (Kawula and Orndorff, 1991; Yasuzawa et al., 1992) and, in other cases, of opposite sign in different types of bacteria (Dorman et al., 1990), alternative explanations have also been formulated. It has been proposed, for instance, that thermoregulation by H‐NS could be mediated via the bacterial translational machinery which would produce higher levels of this protein at 30°C compared with 37°C. In turn, this increased concentration of H‐NS could cause an increased competition with VirF for binding to virB promoter (Tobe et al., 1993) or a change in the H‐NS quaternary structure favouring its specific (inhibitory) interaction with the upstream region of virB (Hromockyj et al., 1992). The recent data suggesting that H‐NS may also directly control the expression of virF in a temperature‐dependent manner opens up new perspectives for the elucidation of the mechanism responsible for thermoregulation of pathogenicity in Enterobacteriaceae.
Thus, in the present study we have undertaken the task of establishing whether H‐NS indeed plays a role in the temperature‐dependent control of virF expression and of defining the molecular basis of this thermoregulation. By in vitro and in vivo footprinting, we have demonstrated that H‐NS binds, at 30°C but not at 37°C, to two sites in the upstream region of virF, one overlapping the −35 and −10 promoter elements. Furthermore, we have demonstrated that interaction at this site is sufficient for transcriptional repression of virF by H‐NS but that both sites are necessary for efficient thermoregulation. Analysis at different temperatures of the intrinsic DNA curvature of fragments containing different segments of the virF promoter showed that this parameter may cause a structural modification of DNA within the virF regulatory region, which controls the accessibility of the DNA target to H‐NS.
Temperature‐ and H‐NS‐dependent activity of the virF promoter in vivo
As mentioned above, previous data had shown that the in vivo expression of virF present on a recombinant plasmid is influenced by temperature and pH in wild‐type but not in hns‐defective E.coli strains. Furthermore, purified H‐NS was found to interact with the virF promoter, at least as judged by electrophoretic gel‐shift experiments (Prosseda et al., 1998). The affinity of H‐NS for the virF promoter was estimated to be similar to that displayed for virB, whose interaction with H‐NS is well documented (Tobe et al., 1993) and was attributed to an intrinsic curvature predicted in this DNA. These results suggested that virF expression is under the control of H‐NS which represses transcription by binding to its promoter at low pH and low temperature (Prosseda et al., 1998).
To confirm this hypothesis and, more importantly, to elucidate the molecular basis of a possible H‐NS‐dependent thermoregulation of virF expression, we sought to characterize, both in vivo and in vitro, the nature, conditions and functional consequences of the interaction between H‐NS and virF. To this end, the constructs and the DNA fragments schematically presented in Figure 1 were prepared and transformed into E.coli cells (either wild‐type or carrying two different hns alleles). As seen in Table I, the level of β‐galactosidase activity expressed in vivo from the virF–lacZ fusion carried by pFABlac112 (Figure 1) was found to be 4–5 times lower at 30°C compared with 37°C in the two wild‐type strains (MC4100 and HN4122), while it was essentially the same at 30 and 37°C in the two hns null alleles.
When pFABlac112 was replaced by pFBlac2, which carries a shorter upstream fragment of virF (Figure 1), repression of lacZ at 30°C (compared with its expression at 37°C) was <2‐fold and, again, seen only in the wild‐type hns background (Table I). These data confirmed and extended the previous observation that in vivo H‐NS may control virF expression in a temperature‐dependent manner (Prosseda et al., 1998).
Binding of H‐NS to the virF promoter region is temperature dependent
The above result prompted us to map the binding sites of H‐NS on the virF promoter by performing in vitro [DNase I and dimethyl sulfate (DMS)] and in vivo (DMS) footprinting experiments at different temperatures in the presence and absence of H‐NS. DNase I footprinting carried out on supercoiled plasmid (pMYSH6504) DNA containing the whole virF gene (Figure 1) revealed the existence of two fairly extended sites (sites I and II) which were protected preferentially by H‐NS when the DNA target and the purified protein were incubated at 28°C. Although the precise borders of these sites are not very sharp, it is relevant that there is a long (160 bp) non‐protected fragment between site I and site II. Site I spans approximately from −46 to +46 on one filament and from −54 to +33 on the other (Figure 2A and B) and site II from −278 to −234 on one filament and −274 to −218 on the other (Figure 2C and D). Within both sites, a few DNase I‐hypersensitive diester bonds (indicated by the arrows in Figure 2B and C) are visible. In contrast to the results obtained at 28°C (Figure 2A–D), very little or no protection by H‐NS was observed when identical footprinting experiments were carried out at 37°C; in fact, within the limits of accuracy of this type of analysis, we cannot detect any significant difference in the DNase I digestion patterns obtained at 37°C in the absence and presence of up to 500 nM H‐NS in the regions of H‐NS site I (not shown) and site II (Figure 2E). These results indicate that the interaction of H‐NS with these two DNA targets is strongly affected by a moderate (10°C) temperature variation.
The interaction of H‐NS with the virF promoter region was also investigated in vivo analysing the DMS modification on plasmid pMYSH6504 (Figure 1) in E.coli YK4122 (wt) and YK4124 (hns2) at 37 and 28°C. As seen in Figure 3A and B, there is a substantial difference between the in vivo footprinting patterns obtained at 37°C and at 30°C in the strain containing wild‐type H‐NS; two stretches of DNA with several bases, mostly G residues whose positions are indicated as C on the complementary strand, are substantially less accessible to DMS at the lower temperature. On the contrary, in the hns− background, there is no difference in the DMS reaction patterns obtained at 37 and 30°C and there is essentially no difference in the reactivity patterns of these DNA samples and the DNA sample extracted from YK4122 (wt) cells grown at 37°C (cf. the appropriate lanes of Figure 3A and B), taking into account that slightly less total DNA was loaded in the case of wild‐type lanes. Since the two strains are isogenic except for the presence or absence of a functional hns gene and since the regions of reduced reactivity with DMS at 30°C closely correspond to the two H‐NS‐binding sites identified by DNase I footprinting in vitro, these results suggest that in vivo there is a temperature‐dependent contact between these DNA segments and H‐NS.
To confirm this premise, DMS footprinting experiments were also performed in vitro in the presence and absence of H‐NS (Figure 3C and D). As seen in the figure, most of the bands of site I (i.e. −16, +4, +15, +23, +39, +44, +52 and +55) and of site II (i.e. −249, −250, −254, −270, −274, −275, −277 and −278) which displayed a notable intensity decrease in the DMS probing in vivo are also strongly protected from chemical attack in vitro at 28°C (cf. Figure 3A and D for site II, and Figure 3B and C for site I). In vitro H‐NS protection of both sites was again observed only at low (28°C) but not at high (37°C) temperature. Since DMS is known to methylate guanine residues at the N‐7 position in the major groove of DNA (Borowiec and Gralla, 1986), the shielding effect of H‐NS is compatible with the premise that this protein interacts with the major groove of DNA, as suggested by Tippner et al. (1994).
Taken together, these experiments demonstrate that the temperature‐dependent protection of sites I and II of virF detected in vivo in a wild‐type hns background is due to the interaction of these sites with H‐NS without the involvement of any other cellular protein. The localization of H‐NS‐binding sites, identified by the different footprinting techniques used in this study, is summarized in Figures 4 and 5. As seen in Figure 4, there is excellent correspondence between the sites protected by H‐NS from DNase I digestion on both filaments and DMS reaction in vitro at 28°C and, in turn, between these sites and those protected in vivo by H‐NS at the same temperature.
The dimensions of the two sites occupied by H‐NS at low temperature and the finding that H‐NS has similar affinities for them (as seen in Figure 2A–D, they are protected similarly by comparable amounts of protein) suggest that both sites might be occupied simultaneously and somewhat cooperatively by several H‐NS molecules. Moreover, the two sites are not entirely independent of each other, since DMS footprinting carried out on plasmid pFB41, which unlike pMYSH6504 lacks the promoter‐distal H‐NS site (Figure 1), revealed only a very weak protection in the region of H‐NS‐binding site I (cf. Figure 3C and E). These results suggest the existence of long‐range protein–protein interactions among H‐NS molecules occupying sites I and II of virF. These interactions could be favoured by the intrinsic DNA curvature predicted between the two H‐NS‐binding sites (Figure 5) which may contribute to bringing into close proximity these sites which are ∼160 bp apart (Wang and Giaever, 1988; Travers, 1989, 1995). Finally, for the interpretation of functional data (see below), it seems relevant to notice that the H‐NS promoter‐proximal site (site I) includes both −35 and −10 elements of the virF promoter. This H‐NS‐binding model is reminiscent of that proposed for the H‐NS–hns interaction (Falconi et al., 1993) and consistent with the known preference of H‐NS for binding curved DNA (Yamada et al., 1990, 1991; Owen‐Hughes et al., 1992; Zuber et al., 1994) as well as with the essential role attributed to protein–protein interactions in the correct functioning of this DNA‐binding protein (Spurio et al., 1997).
In vitro transcription of virF is influenced by H‐NS and temperature
If H‐NS alone is directly responsible for the thermoregulation of virF expression in vivo, as suggested by the results presented in Table I and by its temperature‐dependent interaction with the virF promoter, it might be possible to reproduce this effect in a purified in vitro system. That this is indeed the case is shown by the results of experiments in which the virF promoter activity was examined in the presence and absence of H‐NS as a function of temperature. Thus, the addition of purified H‐NS was found to inhibit severely (∼80% up to 15 min incubation) transcription of virF at 30°C when supercoiled plasmid pMYSH6504 (Figure 1) was used as template (Figure 6A), while transcription at 37°C was only slightly affected by H‐NS (Figure 6B). When the transcriptional activity of virF in the presence or absence of H‐NS was measured as a function of temperature between 26 and 40°C, it was found that the extent of transcriptional inhibition caused by a given amount of H‐NS was not constant and did not vary linearly as a function of the incubation temperature (Figure 6C). Instead, below 30°C, H‐NS caused a 5‐fold inhibition of virF transcription but had negligible effects on samples incubated above the critical temperature of 32°C (Figure 6C). The extent of the H‐NS repression seen in this experiment at 30°C closely corresponds to that found in vivo at the same temperature (Table I).
The relevance of DNA supercoiling for the inhibition of virF promoter function by H‐NS was examined in the following experiment in which the plasmid template pMYSH6504 was linearized by EcoRI digestion prior to the transcription test. In a preliminary experiment, the activity of the virF promoter on a supercoiled and linearized (L) template was analysed at 30 and 37°C (Figure 6D). As seen from this panel, the supercoiled (S) is transcribed better than the linearized (L) template, and temperature has little and no influence on the activity of the linearized and supercoiled promoter, respectively. Unlike that seen with the supercoiled DNA template, H‐NS did not inhibit transcription at either 30 or 37°C (Figure 6E and F), indicating that inhibition of virF transcription occurs only on supercoiled DNA at 30°C. The specific requirement for template supercoiling for H‐NS inhibition is clearly indicated by the fact that the level of transcription of the supercoiled template at 30°C in the presence of H‐NS is way below the level of transcription obtained at either 30 or 37°C on the linearized template in the presence of the same amount of H‐NS (cf. Figure 6A, E and F).
As seen above, the upstream region of virF contains two H‐NS‐binding sites designated I and II (Figure 5). Since site I overlaps the conserved elements of the virF promoter, it is reasonable to assume that this site is involved in transcriptional repression by H‐NS whose interaction with DNA might occlude the access of RNA polymerase to the −35 and −10 consensus sequences.
To determine whether the more upstream H‐NS site II is also required for repression, a shorter construct (pFB41), which lacks ∼150 bp of the distal portion of the virF promoter region (Figure 1), was used as a transcriptional template. As seen from Figure 7, while essentially no inhibition of transcription by H‐NS was seen at 37°C (Figure 7B), at 30°C this protein produced a significant decrease (85%) of virF transcription but only for very short incubation times (<4 min) while the inhibition was relieved with longer incubation, dropping to 50 and 25% after 6 and 15 min, respectively (Figure 7A). These data are in full agreement with the observation that in vivo there is only a minor difference in the transcriptional activity of virF between wild‐type and hns2 strains transformed with the same pFBlac2 construct (Table I).
Taken together, these transcription results and the aforementioned very weak H‐NS footprints on site I in the absence of site II indicate that without the latter site, H‐NS forms an unstable nucleoprotein complex which cannot withstand a prolonged competition with the RNA polymerase.
Influence of temperature on DNA topology of the virF promoter
Since superhelicity of pMYSH6504 was found to be essential for repression of the virF promoter activity by H‐NS (Figure 6) and since the temperature may influence the supercoiling level of plasmids (Goldstein and Drlica, 1984; Lopez‐Garcia and Forterre, 1997), the distribution of pMYSH6504 topoisomers extracted from wild‐type (MC4100) and from two hns mutant strains (HN4104 and YK4124) grown in LB medium at 30 and 37°C was analysed. No detectable differences in the level of superhelicity were detected when the topoisomers of the plasmid isolated from wild‐type or H‐NS‐deficient cells were resolved by chloroquine agarose gel electrophoresis (Figure 8), a result similar to that previously obtained by Yasuzawa et al. (1992) with different reporter plasmids. As to the effect of temperature, our results indicate that going from 30 to 37°C, there is a slight increase in the level of supercoiling. Since this phenomenon occurs in all the genetic backgrounds assayed, and since its extent is very low, we can conclude that the substantial temperature‐dependent change in the level of virF expression in the presence of H‐NS cannot be attributed to a variation of DNA supercoiling caused by this protein. Furthermore, when in vitro DMS footprinting by H‐NS was compared using supercoiled and EcoRI‐linearized pMYSH6504, identical protections were found at 28°C (like that seen in Figure 3C and D) while no protection in either case was found at 37°C (not shown). These results indicate that DNA supercoiling does not modulate the interaction of H‐NS with its target but is necessary for successful competition of H‐NS with the RNA polymerase.
H‐NS binds DNA with low sequence specificity (Rimsky and Spassky, 1990; Lucht et al., 1994), but with a definite preference for curved DNA (Yamada et al., 1990, 1991). Computer predictions, based on published estimates of dinucleotide wedge angles (Bolshoy et al., 1991), suggest the presence of an intrinsic DNA curvature located approximately between −80 and −40 in the virF promoter region (Figure 5). Since intrinsic curvature is another structural property of DNA known to be influenced by temperature (Diekmann and Wang, 1985), it could be suspected that changes of curvature within the promoter region might account for the temperature‐ and H‐NS‐dependent repression of virF expression.
To test this hypothesis, the electrophoretic mobility of three different DNA fragments derived from the virF promoter region was studied as a function of temperature between 4 and 60°C. The DNA fragments examined in this experiment are designated in Figures 1 and 5 as: A (222 bp), corresponding to the most upstream portion (−345/−124) of virF, which contains the H‐NS‐binding site II; B (239 bp), corresponding to the proximal portion of virF (−135/+104), which contains the predicted curvature, the H‐NS‐binding site I and the conserved elements of the promoter; and A + B (449 bp), corresponding to the entire promoter region (−345/+104). A 317 bp fragment derived from the hns promoter region, known to be curved (Falconi et al., 1993), was also tested. These fragments, obtained by Pwo PCR amplification, were subjected to polyacrylamide gel electrophoresis at different temperatures which were monitored during the run by a temperature probe polymerized within the gel. The electrophoresis carried out at 60°C yielded mobility values closely corresponding to the known sizes of the fragments, while the mobilties determined at the lowest temperature (4°C) yielded mobility values corresponding to apparent sizes deviating, to different extents, from the actual sizes of the fragments; the maximum and minimum deviations were observed for the A + B (>35%) and for the A (<3%) fragment, respectively (Figure 9A and B). With increasing temperature from 4 to 60°C, each fragment exhibited a different rate of change in its apparent size towards the real value. The A + B fragment of virF displays a steep decrease in apparent size between 4 and 32°C, followed by a shallower decline between 32 and 60°C, while the other three fragments display the steepest reduction of their apparent size in a single step between 20 and 28°C.
The abrupt change of the electrophoretic mobility displayed at 32°C by the fragment which corresponds to the entire virF promoter region suggests that this DNA segment undergoes a specific and temperature‐dependent structural transition which is complete just above 32°C. This modification is probably responsible for modulating the access of H‐NS to the virF promoter and the consequent capacity of this protein to repress its activity. This premise is entirely consistent with the evidence presented above that this DNA fragment is endowed with unique, temperature‐dependent structural and functional characteristics. In fact, it has been shown that: (i) H‐NS can stably bind to sites I and II of the virF promoter and inhibit transcription in vivo and in vitro only below 32°C (Figure 6A–C); (ii) essentially no protection by H‐NS of either site II or site I was obtained at any temperature when virF promoter fragments A and B were used in footprinting experiments (Figure 3E and not shown); and (iii) when the virF promoter was deprived of its most upstream region corresponding to fragment A, transcriptional inhibition by H‐NS, though still somewhat temperature dependent, became very inefficient, both in vitro (Figure 7A and B) and in vivo (Table I). The functional relevance of the temperature‐dependent conformational transition responsible for the abrupt change in electrophoretic mobility of the A + B fragment seen in Figure 9 is supported further by the finding that a similar transition could also occur under conditions similar to those in which temperature‐dependent repression of the virF promoter is observed. In fact, as seen in Figure 9A (open symbols), Mg2+, at the same concentration (10 mM) present in the transcription reactions (Figures 6 and 7), did not significantly influence the electrophoretic mobility at different temperatures of any of the DNA fragments examined.
The ability of bacteria to infect hosts depends on prompt adaptation to changing environmental conditions such as temperature, pH, osmolarity, and availability of oxygen and nutrients (Dorman and Ni Bhrian, 1993). Much of the adaptation is known to rely on changes in the transcriptional activity of the bacterial genome, but how these environmental parameters can trigger modulations in gene expression remains a central, yet still unanswered question. The notion that H‐NS, together with temperature, pH and osmolarity, plays a crucial role in the regulation of several virulence genes (e.g. cfa/I, sfa, pap, virB and virG) is supported by the evidence that there is a higher and temperature‐independent expression of these operons in strains carrying hns‐defective alleles (e.g. drdX and virR) (Maurelli and Sansonetti, 1988; Goransson et al., 1990; Dagberg and Uhlin, 1992; Jordi et al., 1992; Morschhauser et al., 1993). In these processes, the expression of the target operons can be prevented by H‐NS either directly or through the repression of their transcriptional activators (Goransson et al., 1990; Tobe et al., 1991; Jordi et al., 1992; Lambert et al., 1992; Morschhauser et al., 1993); how a DNA‐binding protein which plays a role in the structural organization of the nucleoid and controls a fairly large number of genes in a temperature‐independent manner can also sense changes in environmental temperature and translate them into the regulation of virulence genes remains a fundamental yet still intriguing issue.
The primary event following the upshift of Shigella to the host temperature (37°C) is the synthesis of VirF, which in turn triggers a regulatory cascade involving the activation of virB and virG (Adler et al., 1989; Tobe et al., 1991). Unlike the cases of virB and virG, for which the thermoregulation by H‐NS is well documented, only recently has it been suggested that virF also might be thermoregulated by H‐NS and that its activation requires only the temperature stimulus (Colonna et al., 1995; Porter and Dorman, 1997; Prosseda et al., 1998).
In this study, after confirming that below 32°C and throughout the cellular growth cycle, the activity of Shigella virF promoter is indeed repressed by H‐NS, we reproduced a similar temperature‐dependent regulation in in vitro transcriptional tests. With these premises, we considered the activation of virF an ideal model to study the molecular mechanism of H‐NS‐dependent thermoregulation of virulence expression. Indeed, this is the first report showing that in a thermoregulated virulence system, H‐NS is able to bind to and repress the promoter of a primary regulator as a function of temperature. In fact, through the reductionist approach used in this study, we have been able to show that the low temperature and the presence of H‐NS are necessary and sufficient conditions for repression and that transition from repression to derepression occurs at ∼32°C, within a very narrow temperature range. This finding, the excellent quantitative correspondence between the H‐NS transcriptional repression observed in vivo and in vitro and the clear‐cut evidence that in vivo and in vitro DMS footprinting of the virF promoter occurs exclusively at low temperature in the presence of H‐NS indicate that thermoregulation of virF expression is solely under H‐NS control and does not depend on other factors, such as σS, which is required for the expression of other virulence genes (Loewen and Hengge‐Aronis, 1994) and is expressed preferentially at low temperature (Sledjeski et al., 1996). As mentioned in the Introduction, it has been suggested that variations of DNA supercoiling may represent the key to the control of bacterial virulence. Our results have indeed shown that a supercoiled DNA template is a necessary condition for H‐NS repression. We have shown, however, that within the critical temperature range there is no H‐NS‐dependent change in DNA topology in vivo of the very same plasmid carrying virF. Furthermore, it was found that DNA supercoiling does not influence the interaction of H‐NS with the virF promoter region since identical H‐NS footprints were obtained on supercoiled and relaxed DNA. Thus, our results (Figure 6D) suggest that RNA polymerase interacts with the promoter in different ways depending on whether the template is supercoiled or relaxed, to give complexes which may or may not be inhibited successfully by H‐NS. Another, not mutually exclusive, explanation for the requirement for a supercoiled template to elicit H‐NS and temperature‐dependent repression is that the temperature‐dependent conformational transition of DNA which we suggest to control H‐NS repression (see below) could be facilitated by supercoiling.
Since the quaternary structure of H‐NS was found to be essential for its biological activity (Spurio et al., 1997), we considered the possibility, suggested by Hromockyj et al. (1992), that temperature‐dependent changes of H‐NS quaternary structure might result in temperature‐dependent gene repression. Our results have shown, however, that the aggregation state of H‐NS is not influenced, either in vivo or in vitro, by temperature changes within the range (30–34°C) critical for virF derepression (R.Spurio, M.Falconi, C.O.Gualerzi and C.L.Pon, unpublished observation). On the other hand, the present data indicate that the temperature sensor within the virF system is the DNA structure which, within a very narrow temperature range, allows a more or less efficient interaction with the H‐NS repressor. In fact, the DNase I and DMS footprinting data have shown that H‐NS binds, only below 32°C, to two sites separated by a fairly long (160 bp) stretch of DNA in which our computer simulation predicts an intrinsic curvature. Our data have allowed us to correlate the simultaneous occupation of these two sites with the temperature‐dependent virF repression by H‐NS. Indeed, the binding of H‐NS to its promoter‐proximal site (site I) is very unstable in the absence of the upstream site (site II) so as to give only a very weak protection from DNase I and DMS and to cause only a transient and very weak inhibition of transcription both in vitro and in vivo.
Unlike the case of the protein repressor, the electrophoretic analysis carried out at different temperatures has shown that the structure of the virF promoter region is highly sensitive to temperature changes within the relevant temperature range. In fact, four different DNA fragments, each containing one or two H‐NS‐binding sites, were found to display, at low temperature (4°C), abnormally low mobilities which increased towards the theoretically expected values with increasing temperature. This change in the electrophoretic mobility occurs in different ways for the fragment (A + B) containing both H‐NS site I and II of virF, for the fragment containing only the binding sites I (fragment B) or II (fragment A) of virF and for the fragment containing the two H‐NS‐binding sites present in the upstream region of the hns promoter. In fact, the mobility increase is gradual, with a sharper step at ∼30°C for the latter three fragments, while for the first fragment it is fairly sharp between 4 and 32°C and more gradual between 32 and 60°C. The different behaviour indicates that this fragment is endowed with a unique conformation compared with the other three, and its biphasic mobility change suggests that it exists in two different conformations, one, amenable to H‐NS repression below 32°C and the other, insensitive to H‐NS above this temperature. The nature of the conformational transition occurring in this fragment around 32°C is not known at present. The correlation between the in vivo and in vitro transcriptional activity of the virF promoter, the demonstrated requirement for binding of H‐NS to both sites for temperaturedependent repression and the results of the footprinting experiments carried out below and above 32°C suggest, however, that a temperature‐sensitive ‘hinge’ is present on the DNA spacer between H‐NS site I and site II. Rotation of the helical axis about this hinge would place in or out of phase the H‐NS molecules weakly bound to the two distant sites. The low‐temperature‐repressible DNA conformation would be that allowing a cooperative stabilizing interaction between the H‐NS molecules bound to two sites. The DNA conformational transition monitored in Figure 9 is observed on a linearized DNA fragment on which H‐NS is unable to repress transcription (Figure 6E and F); however, it is possible that negative supercoiling might be necessary to stabilize or favour this transition.
Materials and methods
Bacterial strains and general procedures
Escherichia coli K12 strains used in this study were MC4100 [F– araD139 Δ (argF‐lac) U169 rpsL150 relA f. bB5301 deoC1 pstF25 rbsR] (Casadaban et al., 1980), HN4104 (MC4100 hns118) (Colonna et al., 1995), HN4122 [F– ara Δ(lac pro) mal thi], YK4122 (W3110 trp+) and YK4124 (YK4122 hns2) (Yasuzawa et al., 1992). HN4124 (HN4122 hns2) was constructed by transduction with P1vir grown on YK4124. The presence of hns‐defective alleles was monitored by the appearance of red colonies on MacConkey base agar containing 0.2% salicine. When appropriate, the following antibiotics/chemicals were added to the media at the following concentrations: ampicillin 50 μg/ml; kanamycin, 30 μg/ml; tetracycline, 5 μg/ml; hygromycin 75 μg/ml; X‐gal 20 μg/ml. Isolation of plasmids, restriction digestions, cloning, electrophoresis and purification of DNA fragments, Southern and Northern blotting were carried out according to Sambrook et al. (1989). DNA probes were labelled with [α‐32P]dATP by the random priming method (Amersham Kit). Purification of H‐NS was performed as described (Falconi et al., 1988). β‐galactosidase assays were performed as described by Miller (1992) on SDS–chloroform‐permeabilized cells grown in LB broth (pH 7) supplemented with ampicillin. PCR was performed as previously described (Colonna et al., 1995). The sequence of the PCR‐generated fragments was checked by the dideoxy chain‐terminating method (Sambrook et al., 1989).
pMYSH6504 is a pBR322‐derived vector containing the virF gene of S.flexneri 2a virulence plasmid pMYSH6000 (Sakai et al., 1986). Plasmids containing the virF–lacZ fusions were constructed by cloning PCR‐generated fragments of the virF promoter region carried by the pMYSH6504 into the multicloning site of the 'lac. YA translational fusion vector pMC1403 (Casadaban et al., 1980) (Figure 1). The forward primers used for amplification were designed with an EcoRI site at the 5′ end and correspond, for pFABlac112, to a sequence (−356 to −339) internal to the vector (QH8 5′‐CTA CGA ATT CCG CTT CCT TTA GCA GC‐3′) and, for pFBlac2, to nucleotides −143 to‐121 (BX71 5′‐TTG AAT TCA AAT ACT TAG CTT G‐3′); the reverse primer designed with a BamHI site corresponds to nucleotides +90 to +115 (QH7 5′‐TAG GGA TCC AAG CGA ACC TTT ATA TC‐3′) in both constructs. The two fragments (QH8–QH7 = 458 bp and BX71–QH7 = 246 bp) containing the virF transcriptional and translational signals and the first 16 codons of virF were fused to the lacZ gene of pMC1403 digested with EcoRI and BamHI enzymes. To construct pFB41, a 997 bp fragment containing the virF gene was obtained by amplification of pMYSH6504 (Sakai et al., 1986) using BX71 and CF4 (5′‐CGGGATCCAAATTTTTTATGATA‐3′ corresponding to +831 to +852) as primers and cloned into the EcoRI–BamHI sites of pBR322.
DNase I footprinting
Supercoiled plasmid DNA (200 ng per sample) was pre‐incubated for 20 min at the indicated temperature with increasing amounts of purified H‐NS in 30 μl of BB buffer containing 40 mM HEPES‐HCl pH 8, 100 mM KCl, 10 mM Mg acetate and 0.5 mM dithiothreitol (DTT). After addition of 30 ng of DNase I, incubation was continued for 40 s, the reactions were stopped by placing the samples on ice and by addition of 1.5 μl of 0.5 M EDTA, 10 μl of Na acetate 3 M. Each DNA sample (100 μl), precipitated with 2.5 vols of ethanol in the presence of 1 μg of tRNA carrier, was resuspended in 10 μl of Polymed PCR buffer supplemented with 3 mM MgCl2, 100 μM of each NTP, 4 pmol of 5′‐32P‐end‐labelled primer and 0.5 U of TaqI polymerase (Polymed) and subjected to 25 cycles of linear PCR (denaturation 1 min at 95°C, primer annealing 1 min at 46°C, except for F322 whose Tm was 48°C, and primer extension 1 min at 72°C). The extension products were separated on a 7% sequencing gel. The following pairs of convergent oligonucleotides BX8 (5′‐GCG AAC CTT TAT ATC T‐3′, coordinates +104 to +89) and F322 (5′‐AGA AGC TGC ATA AGC TC‐3′, coordinates −106 to −90), and FO1 (5′‐CGC TTC CTT TAG CAG C‐3′, coordinates −345 to − 330) and F321 (5′‐ACT TTT CTT AGC AAT ATC TG‐3′, coordinates −169 to −188) were used to detect H‐NS protections, on both DNA strands, at sites I and II, respectively.
In vivo DMS footprinting
The E.coli strains YK4122 (wt) and YK4124 (hns2), harbouring pMYSH6504, were grown to A620 = 0.5 at 30 and 37°C and then incubated for 3 min with 10 mM DMS followed by ‘quenching’ of DMS by addition of 2 vols of cold phosphate‐buffered saline (PBS) buffer (Sambrook et al., 1989) containing 3 mM β‐mercaptoethanol. The chilled cells were immediately collected, washed once with 2 ml of 0.9% NaCl solution, and plasmid DNA was extracted according to the boiling method (Sambrook et al., 1989). Aliquots of DNA, partially cleaved at G and A residues, were subjected to primer extension by TaqI polymerase and a 32P‐end‐labelled oligonucleotide essentially as described above for DNase I footprinting.
In vitro DMS footprinting
Supercoiled (or EcoRI‐linearized) plasmid DNA (200 ng per sample) was pre‐incubated at 28 or 37°C for 15 min with increasing amounts of purified H‐NS in 50 μl of BB buffer (see above). After addition of 1 μl of DMS to a final concentration of 200 mM, the incubation was continued for 1 min before adding 200 μl of an ice‐cold solution containing 500 mM Na acetate (pH 7), 250 mM β‐mercaptoethanol, 5 mM EDTA and 25 μg/ml tRNA. Plasmid DNA, precipitated twice with 3 vols of ethanol, subsequently was processed as described above for DNase I footprinting.
Electrophoretic mobility assay of DNA fragments at different temperatures
The virF fragments used for these assays were obtained by PCR amplification of pMYSH6504 using as forward primer FO1 (see above) or BX7 (5′‐CAA ATA CTT AGC TTG T‐3′, coordinates −135 to −120) and as reverse primers BX8 (see above) or PR8 (5′‐GCA CTC AAA GGG ACT A‐3′, coordinates −135 to −124). The fragment encompassing the intrinsic curvature of the hns promoter region was obtained by PCR amplification of E.coli chromosomal DNA using QC2 (5′‐GAA GAC TGA AAG GTC G‐3′) and HC9 (5′‐CGC ACG AAG AGT ACG G‐3′) corresponding respectively to positions 593–608 and 894–909 of the hns sequence (Pon et al., 1988). The electrophoretic mobility of these DNA fragments (FO1‐BX8, 449 bp; BX7‐BX8, 239 bp; FO1‐PR8, 222 bp; and QC2‐HC9, 317 bp) and that of non‐bent marker fragments (Pharmacia's 100 bp ladder) was analysed by electrophoresis on 0.75 mm thick 5% polyacrylamide gels [29.2:0.8 acryl:bis in TBE (Tris–HCl 90 mM, H3BO3 90 mM, 2.5 mM Na2EDTA, pH 8.6) or in TBM (Tris–HCl 90 mM, H3BO3 90 mM, 10 mM MgCl2, pH 8.6)] run at different temperatures (4, 20, 28, 32, 40 and 60°C for TBE gels; 20, 28, 32 and 40°C for TBM gels) at 5 V/cm, with (TBM gels) or without (TBE gels) buffer recirculation.
Computer‐generated predictions of intrinsic curvature were obtained with WEDGE420, a program developed by one of the authors (G.M.) for the analysis of structural parameters of DNA molecules on ×86 PC platforms running DOS (version 3.2 and up) or Windows (version 3.1 and up) operating systems. Input DNA sequences can be up to 20 kb in size. The bending profile, based on the assumptions of Trifonov's ’wedge' model for bent DNA (Bolshoy et al., 1991) and generated using published (and user‐editable) estimates of dinucleotide wedge angles, is not meant to be a projection of the double helix path on a plane, but rather a representation of the DNA axis with its curves laid on a single plane. Further details and a copy of the executable code and of the accessory files are freely available from the author ( ).
Analysis of DNA supercoiling
The E.coli MC4100, HN4104 and YK4124 strains harbouring pMYSH6504 were grown to A600 = 0.5 in LB medium at 30 and 37°C. The plasmid was extracted by alkaline lysis (Birnboim and Doly, 1979) with the following modification: to the DNA pellet, dissolved in TE, 0.5 vol. of NH4 acetate (7.5 M) was added and, after centrifugation, the DNA present in the supernatant was ethanol precipitated. Topoisomers of pMYSH6504 were resolved on 1% agarose gels in the presence 30 μg/ml chloroquine run for 30 h at 2 V/cm at room temperature in TAE buffer (Tris–HCl 40 mM, Na acetate 25 mM, Na2EDTA 1 mM, pH 8.3). After electrophoresis, the chloroquine was removed by soaking the gel for several hours in distilled water. Plasmid DNA was transferred onto a nitrocellulose filter by Southern blotting and hybridized with a probe consisting of 32P‐labelled, EcoRI‐linearized pMYSH6504. The radioactive bands were detected and quantified by Molecular Imager (Bio‐Rad).
We thank Drs C.Sasakawa for plasmid pMYSH6504, M.J.Casadaban for plasmid pMC1403, Y.Kano for strains YK4122 and YK4124, and P.Lejeune for strain TP504. We are also grateful to Professor Cynthia L.Pon and Dr Roberto Spurio for invaluable help and for critical discussion. This work was supported by grants from the Italian Consiglio Nazionale delle Ricerche, MURST (PRIN ‘Protein–nucleic acid interactions’), Università Roma La Sapienza (GPA 1996–1998) and in part from the ‘Istituto Pasteur–Fondazione Cenci Bolognetti’ Foundation.
- Copyright © 1998 European Molecular Biology Organization