|Home | About | Journals | Submit | Contact Us | Français|
Clinical isolates of the porcine pathogen Actinobacillus pleuropneumoniae often form adherent colonies on agar plates due to expression of an operon, pgaABCD, encoding a poly-β-1,6-N-acetyl-d-glucosamine (PGA) extracellular matrix. The adherent colony phenotype, which correlates with the ability to form biofilms on the surfaces of polystyrene plates, is lost following serial passage in broth culture, and repeated passage of the nonadherent variants on solid media does not result in reversion to the adherent colony phenotype. In order to investigate the regulation of PGA expression and biofilm formation in A. pleuropneumoniae, we screened a bank of transposon mutants of the nonadherent serovar 1 strain S4074T and identified mutations in two genes, rseA and hns, which resulted in the formation of the adherent colony phenotype. In other bacteria, including the Enterobacteriaceae, H-NS acts as a global gene regulator, and RseA is a negative regulator of the extracytoplasmic stress response sigma factor σE. Transcription profiling of A. pleuropneumoniae rseA and hns mutants revealed that both σE and H-NS independently regulate expression of the pga operon. Transcription of the pga operon is initiated from a σE promoter site in the absence of H-NS, and upregulation of σE is sufficient to displace H-NS, allowing transcription to proceed. In A. pleuropneumoniae, H-NS does not act as a global gene regulator but rather specifically regulates biofilm formation via repression of the pga operon. Positive regulation of the pga operon by σE indicates that biofilm formation is part of the extracytoplasmic stress response in A. pleuropneumoniae.
Actinobacillus pleuropneumoniae is the causative agent of porcine pleuropneumonia, a highly contagious economically important endemic disease of pigs worldwide, for which there is no fully efficacious vaccine (9). This bacterium is a strict parasite of the porcine respiratory tract and is most often recovered from sequestered lung lesions and/or from tonsils of chronically infected pigs and is recovered less frequently from the nasal cavities (14, 15, 60). Many of the virulence factors that contribute to the ability of A. pleuropneumoniae to cause acute disease have been identified (23, 49). However, less is known about the factors that contribute to this bacterium's ability to persist and cause chronic infection.
A common finding among clinical isolates of A. pleuropneumoniae is that, when cultured on agar, many form sticky, adherent colonies that are difficult to remove from the agar surface. This phenotype has previously been referred to as “rough” (28), but in order to avoid confusion with “rough” lipopolysaccharide (LPS) mutants, we have chosen to use the term “adherent” to describe the colony morphology. More than half of the 77 field isolates tested by Kaplan and Mulks (28), and 70% of our own strain collection (currently, 625 field strains from the United Kingdom, Denmark, and China), were found to form adherent colonies on agar. This phenotype has been shown to correlate with the ability to form biofilms in the wells of polystyrene plates, as well as at the air-liquid interface in broth culture, and to be dependent on the production of an extracellular matrix composed of poly-β-1,6-N-acetyl-d-glucosamine (poly-β-1,6-GlcNAc or PGA) (26, 28, 29).
Biofilm formation contributes to chronic infection by numerous bacterial pathogens (16, 43). However, it is not yet known if the ability to form biofilms contributes to persistence of A. pleuropneumoniae within the porcine respiratory tract. All 15 reference strains of A. pleuropneumoniae contain the PGA biosynthetic gene pgaC, but only 2 of these strains exhibit a biofilm phenotype (26, 28). Kaplan and Mulks (28) found that, following serial passage in broth culture, clinical isolates of A. pleuropneumoniae lost the ability to form adherent colonies. Even acute experimental infection of pigs did not result in reversion of the nonadherent variants to the biofilm-positive phenotype (28), and they suggested that the change of phenotype was irreversible. However, it was recently shown that growth of A. pleuropneumoniae in certain broth cultures resulted in increased biofilm formation (31), suggesting that the genes required for biofilm formation are subject to regulation by medium components. The mechanism of regulation was not determined, though reduced biofilm production was associated with an increased concentration of zinc in the medium.
During preparation of a bank of signature-tagged transposon mutants of the nonadherent serovar 1 strain S4074T (49), we noted that 62/2,064 mutants were adherent to the agar surface. Interestingly, only 1 of these mutants (previously identified as mclA::Km) was attenuated for acute infection in pigs (49), whereas the other 61 mutants were recovered from infected lungs, indicating no loss of virulence during acute infection. The mclA (also called rseA [regulator of sigma E]) gene encodes an anti-sigma factor that negatively regulates activity of the extracytoplasmic stress response sigma factor RpoE/σE (1). Studies of different bacteria have shown that σE regulates expression of, among others, numerous genes encoding proteins involved in stress response and in repair and maintenance of the bacterial envelope (18, 27, 52). Using promoter analysis to predict possible members of σE regulons in other bacteria, Rhodius et al. (47) suggested that, in addition to a common core set of genes involved in these processes, σE may also regulate organism-specific genes important for pathogenesis. AlgU, the σE homologue in Pseudomonas aeruginosa, Burkholderia pseudomallei, and Xylella fastidiosa, regulates, among others, factors that contribute to biofilm formation and/or structure (30, 37, 50, 54). Mutations in MucA, the P. aeruginosa homologue of RseA, lead to overexpression of alginate, an exopolysaccharide that contributes to biofilm structure, but is not essential for initial biofilm formation, in this bacterium (37, 54). Recently, a second regulator of alginate synthesis in P. aeruginosa, MucR, has been described (25), indicating complex regulation of the mucoid phenotype in this bacterium.
In order to identify an other factor(s) regulating biofilm development in A. pleuropneumoniae, we have now mapped the insertions in the remaining 61 adherent STM mutants and found that they were all located in one of two Tn10 hot spots within the hns gene encoding a predicted H-NS-like regulatory protein. Homologues of H-NS have been identified in numerous Gram-negative bacterial species, but the role of most H-NS-like proteins has not yet been determined (6, 58). The function of H-NS has been most intensively studied in Salmonella and Escherichia coli, where its ability to bind to intrinsically curved AT-rich DNA has been associated with its role as a global gene regulator (19, 21) and more recently as a specific repressor of horizontally acquired genes with atypically low GC content (36, 39, 42). Most H-NS-like proteins share a conserved C-terminal DNA-binding domain (6), and there is evidence that H-NS-like proteins in other bacteria have a role in global transcriptional control (3, 57). In Salmonella and E. coli, H-NS is known to regulate hundreds of genes (36, 39, 42), whereas in other species, the number of regulated genes may be more restricted (58). Recently, Dalai et al. (17) reported that H-NS regulates biofilm formation in A. pleuropneumoniae, but they did not determine the mechanism of regulation. In this paper, we investigate the role of H-NS and σE in regulation of biofilm development in A. pleuropneumoniae.
A. pleuropneumoniae S4074T (here designated wild type [WT]) and its derivatives 5B3 and 19B10 (49), which contain Tn10 transposon insertions in hns and rseA (mclA), respectively, were used in this study. An additional 60 adherent mutants from the Sheehan et al. (49) STM screen and a serotype 1 clinical isolate (M2000) were also investigated. Bacteria were grown at 37°C in 5% CO2 in brain heart infusion (BHI) agar (Difco) supplemented with 1 μg/ml β-NAD (BHI-NAD agar) or in BHI-NAD broth. Tn10 insertion mutants were grown in medium containing 50 μg/ml kanamycin. For protein and RNA extraction, bacterial suspensions were standardized to an optical density at 600 nm (OD600) of 0.3, and 50 μl was spotted onto sterile 0.4-μm-pore-size membrane filter discs (Millipore) placed on BHI-NAD agar plates and grown overnight. E. coli strains XL1-Blue, MG1665, and MG1665Δhns were propagated on Luria-Bertani (LB; Difco) agar supplemented, when necessary, with 100 μg/ml amplicillin or 20 μg/ml chloramphenicol.
Unless otherwise stated, recombinant DNA techniques were carried out as described elsewhere (48). Sequences of all primers used for PCRs are listed in Table Table1,1, and those for quantitative reverse transcription-PCR (qRT-PCR) are listed in Table Table2.2. Locations of Tn10 insertions were determined by inverse PCR and sequencing as previously described (49), except that chromosomal DNA was obtained using the FastDNA Spin kit, according to the manufacturer's instructions (Qbiogene, United Kingdom). Mutations in hns were confirmed by direct PCR using one of the two hns-specific primers (hns_F or hns_R) in combination with the Tn10-specific primer BS401 (49).
The A. pleuropneumoniae hns gene was amplified by PCR using primers hnsExp_F and hnsExp_R, cloned into pGEM-T (Promega), and transformed into E. coli XL1-Blue. An ApaI-NotI fragment containing the hns gene was then subcloned into ApaI-NotI-digested pMC-Express (8). The resulting plasmid, pMCAphns, was transformed into E. coli MG1665Δhns (kindly provided by M. Goldberg, Birmingham, United Kingdom) and tested for the ability to complement motility in soft agar and mucoidy as described by Gordon et al. (24).
In order to determine if the phenotype associated with the rseA mutation was due to deregulation of RpoE, the rpoE gene was amplified by PCR (using primers rpoEExp_F and rpoEExp_R) and cloned into the vector pMC-Express (8) under the control of the sodC promoter of A. pleuropneumoniae, which confers constitutive expression. The resulting plasmid, pMCrpoE, was conjugated into S4074T, as previously described (49).
For RNA extractions, bacteria from 3 to 5 filter discs were vortexed in 4.5 ml Trizol LS (Invitrogen) and 0.5 ml water in the presence of glass beads. The samples were then transferred to a matrix tube and processed in the FastPrep instrument (Thermo Electron) for 40 s at a setting of 6.0. Cell debris and matrix were removed by centrifugation at 15,000 × g for 10 min at 4°C, and RNA in the supernatant was extracted with chloroform, precipitated with ethanol, and resuspended in 87 μl RNase-free water. Samples were treated with 3 μl DNase I (Invitrogen) at room temperature for 15 min and processed using the RNeasy minikit (Qiagen, United Kingdom), according to the manufacturer's protocol. RNA integrity was checked on 1% agarose gels, and concentrations were determined on a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies).
RNA samples were analyzed for the presence of contaminating DNA by PCR amplification of a 200-bp fragment of the A. pleuropneumoniae sodC promoter region using the primers sodCP_F and sodCP_R (Table (Table1).1). If required, a second DNase I digestion was performed to completely remove DNA from the sample. After the second DNase I treatment, the RNA was again purified, according to the RNeasy minikit cleanup protocol, and finally stored at −80°C. cDNA synthesis and labeling were done as previously described (12, 13). Labeled samples were combined and hybridized to the A. pleuropneumoniae 5b strain L20 microarray AppChip2 (31). The hybridization and washing conditions followed were those described previously (53).
A total of six DNA microarray experiments were performed for each comparative hybridization using RNA isolated from five independent pairs of cultures and incorporated a dye swap to account for dye bias. Slides were scanned using a GenePix 4000B scanner (Axon Instruments) and analyzed with GenePix Pro6.0 to generate the data format compatible for further statistical analysis using GeneSpring GX7.3 (Agilent). Microarray analysis was carried out as described previously (12, 13, 34, 45).
RNA samples from the WT, 19B10, and 5B3 (the strains used for microarray analysis), as well as from M2000 (an adherent serotype 1 clinical isolate), were prepared as described above, including three biological replicates of each. In addition, RNA was prepared from two further hns::Km mutant strains (3A3 with a 3′ Tn10 insertion and 24D3 with the same 5′ Tn10 insertion as that in 5B3) (Fig. (Fig.1)1) to confirm that the regulation of pgaA and rpoE was similar in the various hns mutant strains. Samples were reverse transcribed to first-strand cDNA using SuperScript III and random primers (Invitrogen), and real-time PCR was performed as previously described (40) using the primers for pgaA, hns, rpoE, and rpoD listed in Table Table2.2. An initial input of 100 ng per RNA sample was used. The threshold value of each gene was first normalized to the value of the constitutively expressed control gene rpoD (expression was unchanged in microarray analysis of either 5B3 or 19B10 compared to that of WT). Gene induction or reduction values were calculated by comparing the normalized values of the wild-type and mutant samples using the statistical formulation for the threshold cycle (ΔΔCT) method (comparative CT method) as described in the User Bulletin 2 ABI7700 Sequence Detection System (Applied Biosystems).
In order to confirm that overexpression of the pga operon in the rseA::Km (19B10) and hns::Km (5B3) strains is solely responsible for the adherent phenotype, we deleted the pgaC gene in these strains. For creation of the deletion construct, an 814-bp sequence upstream of pgaC was amplified using primers pgaCleft_F and pgaCleft_RSacI (incorporating a SacI restriction site by changing two bases in the native sequence), and an 871-bp sequence downstream of pgaC was amplified using primers pgaCright_FSphI (incorporating a SphI restriction site by changing two bases in the native sequence) and pgaCright_R. A chloramphenicol acetyltransferase gene (cat), originally from a Staphylococcus aureus plasmid, pC194 (35), was amplified from a previously constructed plasmid, pUSSCat (unpublished), using the primers CmApUSS_FBamSacI and CmApUSS_RBamSphI. These primers incorporate the A. pleuropneumoniae uptake signal sequence (USS) required for natural transformation in this bacterium (10, 46) and restriction sites for BamHI, SacI, and SphI, as indicated. The three different PCR products were digested with SacI and/or SphI, as required, cleaned using a QIAquick spin column (Qiagen), and ligated together using T4 DNA ligase (New England Biolabs). The ligation mix was diluted 1:100 and used as the template in a PCR using the pgaCleft_F and pgaCright_R primers. The resulting 2.7-kb fragment, with the pgaC gene entirely replaced by the cat gene, was used as donor DNA for natural transformation (on agar plates) as previously described (10). The pgaC deletion was transformed into the Shope (WT), 19B10, and 5B3 strains, using both PCR product and genomic DNA, resulting in the construction of ΔpgaC, 19B10/ΔpgaC, and 5B3/ΔpgaC, respectively. Deletion of pgaC was confirmed by PCR using the pgaCleft_F and pgaCright_R primers. In the double mutants, the presence of the rseA::Km or hns::Km mutation was confirmed by PCR using either the rseA_F or hns_F primer, respectively, in combination with the Tn10-specific primer BS401.
Plate-grown bacteria (WT, WT plus pMCrpoE, 5B3, 5B3/ΔpgaC, 19B10, and 19B10/ΔpgaC) were resuspended to an OD600 of 0.05 in BHI-NAD broth either containing 30 μg/ml dispersin B (kindly provided by J. Kaplan, New Jersey Dental School) in resuspension buffer (50 mM sodium phosphate, pH 5.8; 100 mM NaCl; 50% glycerol) or containing resuspension buffer alone. Aliquots of 100 μl were transferred in quadruplicate into the wells of a flat-bottom polystyrene microtiter plate (Greiner). Broth blanks were used as controls. Adherent bacteria were detected following 6 h of incubation at 37°C by crystal violet staining as previously described (40).
To map the transcriptional start site of the pga operon in both the hns::Km (5B3) and rseA::Km (19B10) strains, rapid amplification of 5′ cDNA ends (5′RACE) was performed on representative RNA samples that were used in the microarray experiments. The Invitrogen 5′RACE system was used according to the manufacturer's instructions. Briefly, cDNA was generated using the gene-specific primer pgaA_GSP1, reverse transcriptase, and an RNA template (2 μg of 19B10 and 4 μg of 5B3). cDNA was purified using a S.N.A.P. purification kit (Invitrogen). The presence of gene-specific cDNA was confirmed by PCR using primers pgaA_GSP2 and pgaA_GSP3. A homopolymeric tail was then added to the 3′ end of the cDNA using terminal deoxynucleotidyltransferase (TdT) and dCTP. Initially, this tailing reaction was performed at 37°C for 10 min, according to the manufacturer's instructions. However, attempts to PCR amplify gene-specific product from these samples using the abridged anchor primer (AAP) and the nested gene-specific primer pgaA_GSP2 failed. Subsequently, the tailing reaction was performed on ice for 60 min, as recommended for samples with strong secondary structure in the 3′ end of the cDNA. PCR amplification of these samples yielded specific product of approximately 800 bp for both 5B3 and 19B10. These PCR products were sequenced on both strands using pgaA_GSP4 and the nested anchor primer AUAP, and the transcriptional start site was identified.
A fragment of DNA containing 1,150 bases upstream, as well as the first 323 bases of pgaA, was PCR amplified using pgaEMSA_F and pgaEMSA_R. The PCR product was digested with AanI (Fermentas Life Sciences) and purified using a QIAquick PCR purification kit (Qiagen). The DNA fragments (0.5 μg) were combined with different concentrations of purified E. coli H-NS (kindly provided by M. Goldberg, Birmingham, United Kingdom) for 20 min at 37°C, as described by Ono et al. (41). The samples were subjected to electrophoresis through 1.5% agarose at 75 V for 5.5 h at 37°C. The gel was then stained with ethidium bromide to reveal the DNA-protein complexes.
The fully annotated microarray data have been deposited in MIAMExpress (accession no. E-MEXP-2461).
Nucleotide sequences of the DNA flanking the transposon insertion sites were obtained for each of the 61 adherent STM mutants and were used to search the A. pleuropneumoniae serovar 5b genome (22) for corresponding sequences. Four of the mutants (including 5B3 and 24D3) had Tn10 insertions at nucleotide 121 (5′ insertion), and the remaining 57 mutants (including 3A3) had insertions at nucleotide 392 (3′-end insertion) of the hns gene (Fig. (Fig.1).1). At both sites, hot spots for Tn10 insertion (5) were identified (Fig. (Fig.1).1). The insertion site at the 3′ end of hns (GtTTAGC) differs only at 1 bp (lowercase) from the known Tn10 hot spot (GCTNAGC), whereas the 5′ insertion site differs at 2 bp (GaaAAGC), which may explain the higher number of insertions at the 3′ site. The adherent mutants had been isolated from 29 of 48 separately tagged transposon mutant pools (49); thus, it is possible that some of the clones are siblings. There were at least 2 separate clones with 5′ insertions, and 29 separate clones with 3′ insertions, based on the distribution of mutants in the different tagged pools. Each mutant had a single transposon insertion (our unpublished data). The probability of secondary mutations being responsible for the adherent phenotype in all 31 independent hns mutants is highly unlikely.
The A. pleuropneumoniae H-NS protein sequence is 135 amino acids in length and shows 52% identity with H-NS from E. coli (Fig. (Fig.1).1). Genes encoding H-NS-related proteins are widespread in Gram-negative bacteria, and in some species, there are multiple genes encoding H-NS-like proteins (6, 58). No other hns-like gene was detected in the whole-genome sequence of A. pleuropneumoniae L20 (22). Expression of cloned A. pleuropneumoniae hns in E. coli MG1655Δhns resulted in restoration of motility and repression of mucoidy (Fig. (Fig.2),2), confirming that the A. pleuropneumoniae protein is a functional homologue of E. coli H-NS.
Transposon insertions at both sites in A. pleuropneumoniae hns, a monocistronic gene, resulted in the same adherent phenotype and similar results in quantitative reverse transcription-PCR (qRT-PCR) (see below), indicating loss of function of H-NS even when only the last 5 amino acids were affected. This is likely due to disruption of the DNA-binding domain, which is well conserved in H-NS proteins from various bacterial species and is located in the C terminus of the protein (7). The G+C content of the A. pleuropneumoniae hns gene is 39.95%, which does not differ greatly from the average amount of G+C content of this bacterium (42%).
In order to determine which A. pleuropneumoniae genes are regulated by H-NS, we analyzed the transcriptional profile of an hns::Km mutant (5B3) compared to that of the WT strain following 24 h of growth on BHI-NAD agar. H-NS is known to be a global regulator, affecting expression of hundreds of genes in E. coli and Salmonella spp. (36, 39, 42). Surprisingly, only three A. pleuropneumoniae genes, pgaABC, were upregulated by more than 4-fold in the hns::Km mutant (Table (Table3).3). This upregulation was confirmed by qRT-PCR (Fig. (Fig.3).3). No genes showed ≥4-fold reduction in expression in the hns::Km mutant. Expression of rpoE was <2-fold upregulated, as confirmed by qRT-PCR (Fig. (Fig.3).3). The qRT-PCR results for the additional hns mutants (24D3 and 3A3) showed almost identical levels of upregulation of pgaA (11.4 ± 0.2- and 12.6 ± 0.5-fold upregulated, respectively) and of rpoE (1.5 ± 0.03 for both), compared to 5B3 (11.5 ± 0.2-fold upregulated for pgaA and 1.2 ± 0.1-fold upregulated for rpoE).
In order to determine which A. pleuropneumoniae genes are regulated by σE, we analyzed the transcriptional profile of an rseA::Km mutant (19B10) compared to that of the WT strain following 24 h of growth on BHI-NAD agar. In the absence of negative regulation by RseA, σE is free to initiate transcription of genes under the control of this sigma factor. Twenty-four genes showed significant upregulation of ≥4-fold in the absence of RseA (Table (Table4).4). As with the hns::Km mutant, pgaABC were the most highly upregulated genes (Table (Table44 and Fig. Fig.3).3). Genes encoding other previously reported A. pleuropneumoniae adhesins, such as LPS or type IV pili, were not significantly upregulated in the rseA::Km mutant, although oapA, encoding a possible homologue of Haemophilus influenzae opacity protein A, was upregulated. No genes showed ≥4-fold reduction in expression in the rseA::Km mutant. Expression of hns was <2-fold upregulated in the rseA::Km mutant, as confirmed by qRT-PCR (Fig. (Fig.33).
That increased expression of the PGA matrix polysaccharide was solely responsible for the adherent colony phenotype of both the rseA::Km and hns::Km mutants was confirmed by the complete abrogation of biofilm formation of these strains in the presence of dispersin B and by deletion of the pgaC gene in the mutants (Fig. (Fig.4).4). Biofilm plate assays showed strong crystal violet staining was detectable only for the rseA::Km and hns::Km mutants, and the WT plus pMCrpoE, after 6 h of growth in the absence of dispersin B. The addition of dispersin B to the broth cultures completely inhibited formation of biofilm by these strains on the surface of polystyrene plates. The rseA::Km ΔpgaC and hns::Km ΔpgaC double mutants did not form biofilms (even in the absence of dispersin B) and formed smooth colonies on agar plates, conclusively demonstrating that the increased expression of PGA by the rseA::Km and hns::Km mutants was solely responsible for the adherent phenotype. The rseA::Km mutant and the WT plus pMCrpoE strain showed similar phenotypes both in the biofilm assay and on agar plates, providing evidence that overexpression of σE in both strains is responsible for the adherent phenotype.
The microarray results indicated that both H-NS and σE regulate expression of the pga operon; however, there is no increased expression of rpoE in the hns::Km mutant and no decrease in hns expression in the rseA::Km mutant. This raised the possibility that initiation of transcription of pgaA may be occurring from different promoter sites in the two mutants. Therefore, we mapped the transcriptional start site for pgaA using 5′RACE in both the hns::Km and rseA::Km mutants. The results indicated that, in both cases, transcription of the pga operon was initiated from a σE promoter (GAACTT-n16-TCAAA) located 468 bp upstream from the translational start site of pgaA (Fig. (Fig.5A).5A). Further examination of the sequence upstream of pgaA revealed that this region is extremely AT rich (70% A+T over 700 bases) and contains three predicted high-affinity H-NS binding sites (one copy of TCGATAATTT, and 2 copies of TCGATTATAT) (Fig. (Fig.5A).5A). A competitive gel shift assay confirmed that these sites specifically bind H-NS with high affinity (Fig. (Fig.5B),5B), indicating that H-NS can directly inhibit the transcription of pgaA.
The A. pleuropneumoniae serovar 1 type strain S4074T normally produces nonadherent colonies on agar plates and is not able to form a biofilm on the surface of polystyrene (28). During our previous STM analysis (49), designed to identify attenuated strains, we isolated an rseA::Km mutant (19B10) that was adherent on agar plates. RseA, the negative regulator of the extracytoplasmic stress response sigma factor σE, is a cytoplasmic membrane protein that acts to sequester σE, thus reducing or preventing transcription of genes under the control of this sigma factor until conditions arise that induce degradation of RseA (1). In the absence of functional RseA, expression of σE, as well as genes belonging to the σE regulon, is greatly enhanced. The attenuation of the A. pleuropneumoniae rseA::Km mutant suggests that overexpression of certain gene(s) regulated by σE may be detrimental to the ability of this organism to cause acute infection in pigs.
It is not likely that the cause of attenuation is related to the adherent colony phenotype of the rseA::Km mutant, since a further 61 adherent mutants that were among the 2,064 tested in the STM screen were not attenuated. In this study, we have mapped the transposon insertion sites in these 61 mutants and found that all are located in one of two Tn10 hot spots within the gene encoding a homologue of H-NS, a DNA-binding protein involved in regulating the structure and function of the bacterial chromosome (19-21). A recent report indicated that mutation of hns in A. pleuropneumoniae leads to increased biofilm formation and that providing the cloned hns gene in trans restores the nonadherent phenotype (17). However, Dalai et al. (17) did not investigate the genes involved in biofilm formation but rather went on to show that mutation of hns resulted in decreased expression of apxIIA and attenuation of virulence. We also found increased biofilm formation by our hns mutants. However, none of the hns mutants that we investigated were attenuated nor did we see any difference in expression of apxIIA. It is not clear why there is a discrepancy in expression levels of apxIIA, but it may be due to differences in medium components and growth conditions. In the Dalai et al. (17) study, virulence was assessed as a 50% lethal dose (LD50) following intraperitoneal injection in mice. The mouse is not the natural host for this bacterium nor is the peritoneum the natural site of infection. In pigs, as mentioned above, none of the 61 H-NS mutants that were part of our original STM screen were attenuated for virulence (49). In addition, strain 5B3 (with a single Tn10 insertion near the 5′ end of the hns gene) when given at 1 × 105 CFU by the intratracheal route (49) was as virulent as the WT strain, as assessed by lung lesions and clinical scores (our unpublished data). Thus, in contrast to the results obtained in mice, both STM and single-infection studies demonstrate that A. pleuropneumoniae lacking H-NS is fully virulent, being able to cause acute lung infection in pigs.
The A. pleuropneumoniae genome (22) encodes a number of possible factors that could contribute to adherence, or a rough colony morphology, including the extracellular matrix (encoded by pgaABCD), the type IV pilus (encoded by apfABCD), the lipopolysaccharide core (encoded by genes including galU, lbgB, rfaF, hldE, and kdkA), and a possible flp-tad-encoded bundle-forming pilus (29, 38, 55, 59). Recently, it was shown that the components of BHI broth from different suppliers may affect biofilm formation by A. pleuropneumoniae strains (31). Genes, including ftpA (fine tangled pilus major subunit), pgaABC, tadCD, and APL_0443 and APL_0104 (encoding autotransporter adhesins), were upregulated after growth of S4074T in BHI broth. The mechanism(s) of regulation was not determined, although addition of zinc to BHI broth was found to suppress biofilm formation (31). Li et al. (34) reported that a luxS mutant of S4074T exhibited enhanced biofilm formation compared to that of the parental strain after 24 h of culture in broth. However, as in the H-NS study by Dalai et al. (17), no attempt was made to detect the regulation of genes involved in adhesion in the luxS mutant.
The similar adherent colony phenotypes of the rseA::Km and hns::Km mutants of A. pleuropneumoniae S4074T led us to investigate the regulons governed by σE and H-NS in order to determine which genes under the control of these factors may be contributing to biofilm formation in this bacterium. H-NS has been shown to repress expression of E. coli RpoS/σS (4, 62), the stationary-phase sigma factor for which there is no homologue in A. pleuropneumoniae; therefore, we wanted to determine if H-NS was acting as a repressor of σE in A. pleuropneumoniae. However, the transcriptional profile of the A. pleuropneumoniae hns::Km mutant showed no significant change in expression of rpoE or rseA. Furthermore, if H-NS was acting to repress σE, it would have been expected that many genes would be common to both regulons. However, the only genes common to the two regulons were those encoding the PGA matrix, and surprisingly, these were the only genes significantly regulated by H-NS in A. pleuropneumoniae. This is in sharp contrast to the H-NS regulons of E. coli and Salmonella, which both contain hundreds of genes (36, 39, 42). It is possible that under other growth conditions, perhaps in concert with other regulators, H-NS may regulate more genes in A. pleuropneumoniae than we have currently detected. Regardless, it does not appear that the H-NS of A. pleuropneumoniae is a global regulator on the same scale as that seen in the Enterobacteriaceae.
In the σE regulon, the pga genes not only were the most highly upregulated but also were the only genes with >4-fold upregulation that have previously been reported to encode an adhesin of A. pleuropneumoniae. It is interesting to note that oapA, encoding a possible homologue of the H. influenzae opacity protein A, was significantly upregulated in the rseA::Km mutant. In H. influenzae, this protein has been shown to contribute to adhesion to epithelial cells (44, 61), though the function in A. pleuropneumoniae remains to be determined.
That increased production of the PGA matrix is solely responsible for the adherent colony phenotypes of both the rseA::Km and hns::Km mutants was confirmed by complete abrogation of adherence of these strains to wells of polystyrene plates in the presence of dispersin B, a PGA-degrading enzyme (29), and by the lack of biofilm formation by the rseA::Km ΔpgaC and hns::Km ΔpgaC double mutants in the absence of dispersin B. In contrast, biofilms produced by Aggregatibacter actinomycetemcomitans, where the Flp pilus is the major contributing factor, were previously shown to be partially resistant to dispersal by dispersin B (29).
In order to determine how σE and H-NS contribute to regulation of PGA formation in A. pleuropneumoniae, we investigated the promoter region of the pga operon. The sequence upstream of pgaA is extremely AT rich (70% A+T over 700 bases) and contains three predicted high-affinity H-NS binding sites (11, 32). There are 1 copy of TCGATAATTT and 2 copies of TCGATTATAT located 155, 248, and 325 bases upstream (respectively) of the translational start of the pgaA gene. A competitive gel shift assay showed that purified H-NS bound to sequences upstream of pgaA, with the fragment containing the three predicted high-affinity binding sites showing the greatest mobility shift. Specific binding of H-NS to these sites indicates that H-NS can directly inhibit PGA expression.
In addition to rpoE, other regulator genes, including APL_1409 (encoding a CueR-like regulator) and cdaR (encoding a putative sugar diacid utilization regulator), were upregulated in the rseA::Km mutant, making it difficult to determine if the pga operon is directly regulated by σE. It has been shown that derepression of H-NS-regulated genes can occur in the presence of other regulatory elements (such as members of the AraC, MarR, or LysR families of regulators), or alternative sigma factors (such as σS), in a process termed “counter-silencing” or “antisilencing” (21, 56). In particular, it has been shown in E. coli that H-NS more strongly inhibits transcription from σD promoters than from promoters recognized by σS (51). The fact that rpoE is not upregulated in the hns::Km mutant suggested that, in the absence of H-NS, the pga operon may be transcribed from a σD promoter. However, 5′RACE revealed that, in both mutants, transcription of the pga operon proceeded from a site located 468 bp upstream from the pgaA translational start. Examination of the sequence immediately preceding the transcriptional start site revealed a σE promoter sequence (GAACTT-n16-TCAAA). Our results indicate that, under the growth conditions tested, there is enough free σE available in the hns::Km mutant such that, in the absence of H-NS, expression of the pga operon can occur, although at a lower level than that seen in the rseA::Km mutant, where rpoE is upregulated and σE is not sequestered by RseA. In the rseA::Km mutant, PGA production is upregulated without altering the expression level of hns. This suggests that increased cellular levels of σE in the rseA::Km mutant are sufficient to displace bound H-NS, resulting in antisilencing of the pga operon.
We have shown here that transcription of the pga operon, the main factor associated with biofilm formation in A. pleuropneumoniae, is initiated from a σE promoter site and is specifically repressed by H-NS under conditions of low σE production. Repression by H-NS may indicate recent acquisition of the pga operon by lateral transfer. Similar genes encoding PGA are distributed among a wide variety of bacteria (33). The fact that A. pleuropneumoniae H-NS acts as a specific regulator of the pga operon, and not as a global regulator, may be due to the reduced genome size and G+C content and the restricted environmental niche of this bacterium compared to those of members of the Enterobacteriaceae.
The contribution of biofilm development to the virulence of A. pleuropneumoniae is still not clear, as both nonadherent and adherent isolates are equally capable of causing experimental disease, and the production of the PGA matrix is not sufficient to protect the rseA::Km mutant from host clearance mechanisms during acute infection (49). However, the fact that the majority of fresh clinical isolates of A. pleuropneumoniae are adherent on agar plates suggests that this phenotype may be important in vivo. The qRT-PCR results for the adherent serotype 1 clinical isolate (M2000) (Fig. (Fig.3)3) showed that rpoE expression was 4-fold higher and hns expression was 1.5-fold lower compared to those expression levels of the nonadherent serovar 1 strain S4074T. As expected, pgaA expression in M2000 was similar to that seen in the rseA::Km and hns::Km mutants of S4074T. Auger et al. (2) recently showed that the pgaBC genes were upregulated in A. pleuropneumoniae following contact with porcine respiratory tract epithelial cells, indicating that biofilm formation may contribute to colonization and/or persistence of this bacterium in vivo.
The fact that the pga operon genes are those most strongly upregulated in the A. pleuropneumoniae σE regulon indicates that biofilm formation is a major component of the extracytoplasmic stress response in this bacterium. In our previous STM study, it was noteworthy that, compared to many other organisms, a high percentage of attenuating mutations were identified in stress response genes (49). It may be that biofilm formation is part of the response to envelope damage caused by the host immune system, contributing to the persistence of A. pleuropneumoniae within tonsils or sequestered lung lesions. The downregulation of pga expression, or the degradation of the PGA matrix by dispersin B, may be important for the release of bacteria from the biofilm matrix in order to initiate acute infection. It is not clear which environmental signals and underlying mechanisms lead to the regulation of rpoE and hns and, as a consequence, PGA expression in A. pleuropneumoniae. These are likely to be complex and may result in populations of adherent and nonadherent bacteria within the different microenvironments (e.g., lungs versus tonsils) and at different stages of infection in the pig. Further studies are required to determine whether σE and H-NS are part of a larger regulatory network governing PGA expression and host-interactive biology.
This work was supported by the United Kingdom Biotechnology and Biological Sciences Research Council (P.R.L., J.T.B., A.N.R., and J.S.K.) and, in part, by the George John and Sheilah Livanos Charitable Trust (J.S.K., S.S., M.-S.L., and C.A.O.).
We are grateful to Jeff Kaplan for kindly providing dispersin B and to Martin Goldberg for the generous gifts of the E. coli MG1665Δhns strain and purified E. coli H-NS, as well as for helpful discussions. We are grateful to Charles Gervais for photographic expertise.
Published ahead of print on 5 March 2010.