Correspondence: Gerhard Heusipp, Rhine-Waal University of Applied Sciences, Center for Research, Innovation and Transfer, Marie-Curie-Str. 1, 47533 Kleve, Germany. Tel.: +49 2821 80673 116; fax: +49 2821 80673 160; e-mail: email@example.com
Horizontal gene transfer plays an important role in bacterial evolution. DNA acquired by horizontal gene transfer has to be incorporated into existing regulatory networks. The histone-like nucleoid structuring protein H-NS acts as a silencer of horizontally acquired genes to avoid potential damage. However, specific regulators can overcome H-NS repression, resulting in the integration of newly acquired genes into existing regulatory networks. Here, we analyzed the influence of H-NS on the transcription of the Yersinia enterocolitica hreP gene and its regulators pypA, pypB, and pypC by establishing a dominant-negative H-NS version. Using transcriptional fusions and electrophoretic mobility shift assays, we show that H-NS silences hreP, pypA, pypB, and pypC by direct interactions. While the H-NS antagonist RovA activates pypC, it has no effect on pypA and pypB. Furthermore, H-NS affects biofilm formation in Y. enterocolitica.
Adaptation to changing environments is a prerequisite for a successful survival of bacteria in diverse habitats. Long-term adaptation occurs during evolution, where random mutations may result in changed phenotypes. Moreover, bacteria can acquire large DNA fragments by horizontal gene transfer, which plays a significant role in bacterial evolution, pathogenesis, and may lead to increased fitness and the ability to colonize new habitats (Dobrindt et al., 2004; Brzuszkiewicz et al., 2009; Lesic et al., 2012). Short-term adaptation is mediated by the controlled expression of certain genes in response to changes in environmental conditions that are mostly sensed and transduced by specific transcriptional regulators. This adaptation is especially important for pathogenic bacteria, as they are facing the attack of the immune responses of the host in addition to environmental changes in and outside the host. Considering this, it is obvious that also newly acquired genes have to be integrated into existing regulatory networks in the bacterial cell. H-NS, the histone-like nucleoid structuring protein, plays an important role in these regulatory processes in many Gram-negative bacteria by acting as a silencer of horizontally acquired genes (Dorman, 2004; Lucchini et al., 2006; Navarre et al., 2006). The H-NS protein binds preferentially to intrinsically curved AT-rich DNA often found in promoter regions of genes. Therefore, H-NS acts as a global repressor of gene expression in many bacteria.
The H-NS protein consists of an N-terminal dimerization domain and a C-terminal DNA-binding domain joined through a flexible linker (Arold et al., 2010). To silence gene expression, H-NS inserts into the minor groove of AT-rich DNA via an ‘AT-hook’ motif and oligomerizes to form a nucleoprotein complex that prevents the binding or activity of the RNA polymerase (Aravind & Landsman, 1998; Lucchini et al., 2006; Oshima et al., 2006; Gordon et al., 2011). Mutated H-NS versions lacking the oligomerization domain are no longer able to repress gene expression (Rimsky et al., 2001). H-NS-mediated repression of gene expression can be relieved by several mechanisms (Stoebel et al., 2008). For example, it has been described that transcriptional regulators like RovA of Yersinia can displace H-NS from promoters and activate gene expression (Heroven et al., 2004; Ellison & Miller, 2006). Other factors interfere with H-NS repression by forming heteromeric protein–protein complexes via the oligomerization domain, rendering H-NS nonfunctional (Madrid et al., 2007).
The genus Yersinia consists of three species pathogenic for humans. While Y. pestis is transmitted via a fleabite or aerosol and causes the devastating disease plague, Y. pseudotuberculosis and Y. enterocolitica are enteropathogenic. Comparative genomic analyses show that each of the three species has a set of genes common to all Yersiniae as well as a specific set allowing for different life styles outside the human host as well as during an infection (Wren, 2003). On a genomic level as well as on virulence level, even the species Y. enterocolitica is highly diverse. Several pathogenic and nonpathogenic isolates exist, indicating genomic exchange and adaptation (Howard et al., 2006). This is also mirrored in the genome sequence of a human pathogenic isolate showing various regions of genome flexibility and horizontal gene transfer (Thomson et al., 2006). The hreP gene, for example, is specific for Y. enterocolitica and has not been identified in any of the other Yersinia pathogens. Its relatively low GC% content and its location in the genome indicate horizontal acquisition (Young & Miller, 1997; Heusipp et al., 2001). It was shown that hreP is exclusively expressed during an experimental mouse infection by Y. enterocolitica. The hreP gene encodes a bacterial subtilisin/kexin-like protease that is necessary for full mouse virulence of Y. enterocolitica (Young & Miller, 1997; Heusipp et al., 2001). In a recent study, we identified the proteins PypA, PypB, and PypC as positive regulators and H-NS as a silencer of hreP transcription. PypA was identified as an inner membrane protein with no significant similarity to any known proteins, while PypB is a ToxR-like transmembrane transcriptional regulator and PypC is a cytoplasmic transcriptional regulator with an OmpR-like winged helix-turn-helix DNA-binding motif (Wagner et al., 2009). Interestingly, also pypA, pypB, and pypC have a low GC% content. Furthermore, pypB and pypC are associated with pathogenicity islands and flexible genome regions. While pypC is associated with the Yts2 type-II secretion system, pypB is the first gene of the tad operon encoding type-IVb pili (Schilling et al., 2010; Shutinoski et al., 2010). H-NS has been identified as a repressor of several horizontally acquired genes in Y. enterocolitica, and the transcriptional regulator RovA alleviates its repression in most cases (Cathelyn et al., 2007). The aim of this study was therefore to analyze the influence of H-NS on transcription of hreP and the pyp genes in more detail to better understand how horizontally acquired transcriptional regulatory networks evolve.
Materials and methods
Bacterial strains, plasmids, and oligonucleotides
All strains, plasmids, and oligonucleotides used are listed in Table 1. For routine growth, all strains were grown in Luria-Bertani (LB) broth or on agar plates at 26 °C for Y. enterocolitica or 37 °C for E. coli or as otherwise indicated. Antibiotics were used in the following final concentration: for Y. enterocolitica, nalidixic acid (Nal; 20 μg mL−1), kanamycin (Kan; 100 μg mL−1), streptomycin (Strep; 50 μg mL−1), and chloramphenicol (Cam; 12.5 μg mL−1); for E. coli, kanamycin (50 μg mL−1) and streptomycin (50 μg mL−1). To induce gene expression from the Para promoter, bacteria were grown in the presence of 0.2% [w/v] arabinose.
Table 1. Bacterial strains, plasmids, and oligonucleotides
Kanr, hns of Y. enterocolitica lacking its 3′ end coding for the DNA-binding domain under control of Para in pBAD18-Kan
Kanr, hns of Y. enterocolitica in pET24b(+), encodes H-NS with C-terminal 6xHis-tag
Sequence (5′→3′), restriction site (Restriction site is underlined)
Construction of plasmids and purification of recombinant H-NS
To construct plasmid pBAD-hnsΔ, an approximately 300-bp PCR fragment was generated excluding the sequence encoding the DNA-binding domain of H-NS, using the primers HNS-f2 and HNS-r2 and ligated via EcoRI/XbaI into pBAD18-Kan. The plasmid pET-hns was generated for expression of Y. enterocolitica H-NS with a carboxyterminal 6xHis-tag in E. coli. To this end, an approximately 400-bp PCR fragment was generated with primers HNS-f1 and HNS-r1 and ligated via NheI/HindIII into pET24b(+). To analyze transcriptional activation of pypA, pypB and pypC in an E. coli background, approximately 1-kb fragments upstream of the ATG start codon were amplified with primer pairs KW-pypA1000for/KW-pypA1000rev, JS–pypB5/JS-pypB6, and JS-pypC5/JS-pypC6, respectively, and ligated via BamHI/EcoRI (pypA) or XbaI/EcoRI (pypB, pypC) into the lacZYA reporter plasmid pKW1, resulting in pKW-A1000, pKW-B1000, and pKW-C1000, respectively.
Purification of recombinant H-NS via nickel-NTA affinity purification from E. coli BL21(DE3) bacteria carrying pET-hns was performed using standard conditions and buffers as described previously (Heusipp et al., 2001; Wagner et al., 2009).
β-Galactosidase activity assay
The β-galactosidase activity of reporter strains was determined as previously described and reported as arbitrary Miller units (Wagner et al., 2009). To compare β-galactosidase activities of Y. enterocolitica reporter strains with E. coli MC4100 and PD145 at lower temperature (26 °C), E. coli strains were grown at 30 °C, as there was only insufficient growth at 26 °C. Data are the means and standard deviations of at least three experiments, each performed in triplicate.
Biofilm assays and crystal violet staining
To determine the formation of biofilms, bacteria were inoculated in 96-well polyvinylchloride (PVC) microplates in a total volume of 200 μL in LB medium supplemented with 0.2% [w/v] arabinose to induce gene expression from Para. After growth at 26 °C for 24 h, supernatants were aspirated, and the wells were washed three times with ddH2O. For staining, wells were then incubated with 200 μL 1% [w/v] crystal violet for 15 min, washed again three times with ddH2O, and incubated with 200 μL 33% [v/v] glacial acetic acid to solubilize the crystal violet for an additional 20 min. Samples were then used to determine the OD at 590 nm.
Electrophoretic mobility shift assay (EMSA)
To analyze the interaction of recombinant H-NS with promoter regions of hreP, pypA, pypB, and pypC, we employed EMSA assays as described recently (Bossé et al., 2010). Briefly, PCR products representing approximately 500 nt upstream of the ATG start codon of hreP, pypA, pypB, and pypC were amplified using primer pairs JS–hreP2.rev/JS-flhB2X.rev, JS-pypA1/JS-pypA2, JS-pypB1/JS-pypB6, and JS–pypC1/JS-pypC6, respectively. The PCR fragments were purified, 0.5 μg of DNA was incubated with increasing amounts of recombinant H-NS in elution buffer [50 mM Tris–HCl (pH 8.0), 250 mM NaCl, 250 mM imidazole, 10% (v/v) glycerol, 0.1% (v/v) Triton X-100] for 20 min at 26 °C. As a nonspecific competitor DNA fragment, we included 0.5 μg of an approximately 200-bp PCR fragment generated with primers KR-cpxA1 and GH-cpx9, representing an internal fragment of the Y. enterocolitica cpxA gene. The DNA fragments were separated on 1.5% agarose gels and run at 75 V. DNA was detected after staining with ethidium bromide.
Results and Discussion
Establishing a dominant-negative H-NS variant to analyze the effect of H-NS on gene expression in Y. enterocolitica
In a previous study, we have shown that the hreP gene of Y. enterocolitica is positively regulated by PypA, PypB, and PypC, and that hreP transcription is increased in an E. coli hns mutant strain compared with its isogenic wild type, indicating that H-NS silences hreP (Wagner et al., 2009). As the pyp genes have characteristics of horizontally acquired genes, we anticipated that they also might be silenced by H–NS. Indeed, the β-galactosidase activities of pypA-lacZ, pypB-lacZ, and pypC-lacZ transcriptional fusions were repressed in the E. coli MC4100 strain compared with the hns mutant PD145 (Fig. 1). So far, effects of H-NS on transcription of Y. enterocolitica genes has been performed in E. coli strains, as hns mutants of Y. enterocolitica are not viable (Ellison & Miller, 2006). To confirm the effect of H-NS on hreP as well as transcription of the pyp genes in a Y. enterocolitica background, we constructed a truncated H-NS version that is able to multimerize, but not to bind to DNA, thereby acting as a dominant-negative H-NS protein (H-NSΔ). A similar approach has been successfully applied before using a naturally occurring H-NST protein of enteropathogenic E. coli (Banos et al., 2008). H-NS lacking its C-terminal nucleic acid-binding domain, but containing a functional dimerization domain, was cloned under the control of the Para promoter in pBAD18-Kan (pBAD-hnsΔ) and transferred to Y. enterocolitica transcriptional reporter strains. Similar to the effect of H-NSTEPEC, but in contrast to a mutation of the hns gene, the bacteria are viable. After 3 h of growth at 26 °C in the presence of 0.2% arabinose to induce transcription of hnsΔ from Para, the β-galactosidase activities of hreP-lacZ, pypA-lacZ, pypB-lacZ, and pypC-lacZ transcriptional fusions were found to be elevated in comparison with control cells, while there was – in contrast to data obtained in E. coli – no significant effect of H-NSΔ on pypA transcription (Fig. 2). This might be explained by differences in the genetic background; it is also possible that H-NSΔ is not able to completely block H-NS activity. Differences in β-galactosidase activities of the reporter fusions between E. coli and Y. enterocolitica might also be related to differences in the constructs used (pKW1-based plasmids in E. coli, chromosomally integrated reporter constructs in Y. enterocolitica). An effect of H-NSΔ on hreP and pypC gene expression was also detected after growth at 37 °C; however, the observed effects were slightly less pronounced, indicating that the effect of H-NS on gene expression in Y. enterocolitica is at least in part influenced by temperature (Fig. 2). This is not surprising, as H-NS functions by bridging DNA stretches and relies on DNA curvature, which is sensitive to physical parameters like osmolarity and temperature (Liu et al., 2010). This has been well demonstrated for the virF gene promoter of Shigella flexneri (Prosseda et al., 2004; Stoebel et al., 2008). Our data establish H-NSΔ as a tool to analyze the effects of H-NS on gene expression in Y. enterocolitica and show that hreP, pypB, and pypC are indeed silenced by H-NS. Data for pypA show some differences in E. coli and Y. entercolitica; however, we provide evidence that also pypA expression is influenced by H-NS. Nevertheless, further analysis is needed in future studies.
Recently, we identified a pypC homolog in the Y. enterocolitica genome that we termed pclR. Both genes are the first genes of operons encoding type-II secretion systems. While the pypC-associated Yts2 operon has a low GC% content, only the first two genes of the pclR-associated Yts1 operon, pclR and yts1C, are AT-rich (Shutinoski et al., 2010). Therefore, we tested whether the introduction of H-NSΔ results in anti-silencing of the pclR gene in a Y. enterocolitica pclR-lacZYA reporter strain. However, as there was no effect, H-NS does not seem to repress pclR transcription in Y. enterocolitica (data not shown).
H-NS directly interacts with the promoter regions of hreP, pypA, pypB, and pypC
As PypA, PypB, and PypC regulate hreP, and as the pyp genes themselves are part of a complex regulatory network (Wagner et al., 2009), effects of H-NS on one of these genes could be the result of either direct or indirect regulation. We used electrophoretic mobility shift assays (EMSA) to study the interaction of recombinant H-NS protein with the promoter regions of hreP, pypA, pypB, and pypC. PCR-amplified DNA fragments of approximately 500 nt upstream of the ATG start codon of each gene were incubated with increasing concentrations of H-NS. As a control, unspecific competitor DNA (an approximately 200-nt PCR fragment of the cpxA gene) was included in the reaction. As shown in Fig. 3, H-NS resulted in a specific mobility shift of hreP, pypA, pypB, and pypC promoter fragments, while the cpxA DNA fragment was not affected. These data support the hypothesis that the observed effects of H-NSΔ on transcription of hreP, pypA, pypB, and pypC are the result of a direct interaction of H-NS with the respective promoter regions. The observation that H-NS seems to directly interact with the pypA promoter region but does not significantly alter pypA transcription indicates that other factors in addition to H-NS influence pypA transcription in vivo.
Only pypC, but not pypA and pypB are activated by RovA overproduction
In previous analyses, RovA has been shown to activate gene expression of Yersinia genes repressed by H-NS by competing for binding sites in the respective promoter regions (Heroven et al., 2004; Ellison & Miller, 2006; Cathelyn et al., 2007). We further demonstrated that hreP is not activated by RovA overproduction (Wagner et al., 2009), but as many H-NS-repressed genes of Y. enterocolitica seem to be activated by RovA, we tested whether RovA activates pypA, pypB or pypC. The Y. enterocolitica pypA-lacZ, pypB-lacZ, and pypC-lacZ transcriptional reporter strains carrying pBAD-rovA or the empty plasmid pBAD18-Kan, as a control were grown for 3 h at 26 °C in the presence of 0.2% arabinose to induce RovA expression from Para-rovA before β-galactosidase activity was determined. As shown in Fig. 4, there is no effect of RovA overproduction on pypA activation. While transcription of pypB is induced only about twofold, pypC is strongly (11-fold) activated by RovA. We repeated the assay after 17-h growth, and under these conditions, there was no detectable effect of RovA on pypB, while pypC transcription increased 58-fold (data not shown). From this, we conclude that the influence of RovA on pypB transcription is probably negligible or indirect and of minor biological significance. However, RovA activates pypC transcription, and might also be involved in anti-repression of the pypC promoter by H-NS. This furthermore shows that the pyp regulatory network is quite complex and might integrate several inputs to tune the expression of virulence genes in Y. enterocolitica. For example, RovA overproduction activates pypC, but has no effect on hreP transcription, although PypC activates hreP. Future experiments will be aimed at an additional dissection of the regulatory cross-talk between hreP, the individual pyp genes, H-NS, and putative additional regulators. It will also be interesting to analyze whether PypB and PypC are able to compete for binding at individual promoters with H-NS and thereby contribute to anti-silencing by H-NS.
H-NS silences genes involved in biofilm formation of Y. enterocolitica
During the experiments described above to determine the effect of H-NSΔ on gene transcription, we observed that the bacteria formed biofilms at the liquid–air interface in the glass test tubes used for culturing after over-night growth at 26 °C. This prevented us from performing β-galactosidase assays as the bacteria could not easily be detached. To analyze this observation in more detail, we tested whether mutant strains of hreP and the pyp genes might be involved in biofilm formation. Therefore, we grew the different strains in 96-well PVC microtiter plates and determined biofilm formation using a crystal violet assay. Figure 5 shows that overproduction of H-NSΔ strongly increases biofilm formation in our mutant strains similar to the wild type after growth for 24 h at 26 °C. This shows that hreP, pypA, pypB, and pypC are not involved in biofilm formation after H-NS overproduction. The observed biofilm formation is therefore related to another H-NS regulated gene that needs to be determined.
This goes in line with other data from our group showing that the PypB-regulated tad locus encoding type-IVb pili is not involved in biofilm formation, although similar pili mediate the ability to grow in biofilms in other bacteria (Schilling et al., 2010). Although the pypC gene is associated with the Yts2 type-II secretion system, we could not identify any associated phenotype. Similarly, hreP is necessary for full virulence in the mouse infection model, but its mode of action could so far not been described further by in vitro analysis (Young & Miller, 1997; Heusipp et al., 2001).
In conclusion, we could show that hreP and the pyp genes are part of a complex regulatory network that is repressed by H-NS, further underlining the hypothesis that the genes have been acquired horizontally. Therefore, the pyp regulatory network might serve as a model to study the evolution of (horizontally acquired) regulatory networks. The data also underscore that it is important to study H-NS effects not only in E. coli as surrogate model organisms, and that dominant-negative versions of H-NS as the H-NSΔ protein are an important and valuable tool for further studies that can result in the identification of additional phenotypes as shown here for biofilm formation.
We thank the members of the Yersinia group at the ZMBE, Institute of Infectiology, for discussion. This work was supported by grants of the Deutsche Forschungsgemeinschaft (HE3079/9-1; Graduiertenkolleg GRK 1409/2). Petra Dersch is acknowledged for the gift of E. coli strains MC4100 and PD145.