|Home | About | Journals | Submit | Contact Us | Français|
Transcription initiation is a critical step in bacterial gene regulation and is often controlled by transcription regulators. The alternate sigma factor (σ54) is one such regulator that facilitates activator-dependent transcription initiation and thus modulates the expression of a variety of genes involved in metabolism and pathogenesis in bacteria. This study describes the role of σ54 in the nosocomial pathogen Enterococcus faecalis. Biofilm formation is one of the important pathogenic mechanisms of E. faecalis, as it elevates the organism's potential to cause surgical site and urinary tract infections. Lysis of bacterial cells within the population contributes to biofilm formation by providing extracellular DNA (eDNA) as a key component of the biofilm matrix. Deletion of rpoN rendered E. faecalis resistant to autolysis, which in turn impaired eDNA release. Despite the significant reduction in eDNA levels compared to the parental strain, the rpoN mutant formed more robust biofilms as observed using laser scanning confocal microscopy and Comstat analysis, indicating and emphasizing the presence of other matrix components. Initial adherence to a polystyrene surface was also enhanced in the mutant. Proteinase K treatment at early stages of biofilm development significantly reduced the accumulation of biofilm by the rpoN mutant. In conclusion, our data indicate that other factors in addition to eDNA might contribute to the overall composition of the enterococcal biofilm and that the regulatory role of σ54 governs the nature and composition of the biofilm matrix.
As opportunistic pathogens, enterococci are the third leading cause of hospital-acquired or associated infections, as they are responsible for 11.2% of surgical site infections (SSI), 14.9% of urinary tract infections (UTI), and 16% of reported bloodstream infections (25). The ability to form a biofilm is an important aspect of the lifestyle of the organism, as biofilm formation is thought to be a property associated with the establishment of SSI and UTI (34), both of which can serve as foci to establish bloodstream infections. Biofilms are aggregates of bacteria that are covered in exoploymer matrix and are more resistant to antibiotics than their planktonic counterparts (15, 26). In several bacterial species, nucleic acids, polysaccharides, proteins, and lipids constitute the exopolymer matrix (19). The components of the biofilm matrix form a physical barrier that enhances the inaccessibility of the biofilm cells to antibiotics and the immune system, thereby making the infection difficult to eradicate (33). Extracellular DNA (eDNA) serves as an important biofilm matrix component in several microbial model systems, including but not limited to Neisseria meningitidis, Listeria monocytogenes, Pseudomonas aeruginosa, Enterococcus faecalis, Staphylococcus aureus, and Staphylococcus epidermidis (2, 23, 29, 32, 36, 47, 48, 54). The expression of the two secreted E. faecalis proteases, gelatinase and serine protease, is regulated in a quorum-dependent manner by the Fsr regulatory system (22, 45, 46), and these proteases direct biofilm development by modulating the eDNA matrix via regulation of the extent of autolysis (54) in a fratricidal manner (51). In an attempt to identify other factors that govern eDNA release in E. faecalis, we identified rpoN, which encodes σ54, in a preliminary transposon mutagenesis screen.
Transcription initiation is one of the important stages of gene regulation, and sigma factors play a crucial role in determining the controlled response of a subset of genes tied to a given environmental stimulus. Sigma factors reversibly bind to RNA polymerases and drive promoter-specific transcription initiation. In prokaryotes, two distinct families of sigma factors have been studied, sigma 70 (σ70) and σ54. The σ70 family also includes several related alternate sigma factors. Sigma 54 shares no structural homology with sigma 70, possesses a distinct consensus binding sequence (−24/−12; TTGGCACNNNNNTTGCT) and, unlike sigma 70, facilitates activator-dependent transcription initiation (24, 38).
Sigma 54 plays an important role in the virulence of several bacteria but does not share the same function in all pathogens (30). In Vibrio fischeri, σ54 influences biofilm formation, motility, and symbiotic colonization of squid and negatively regulates bioluminescence (58). Quorum-sensing regulation in Vibrio cholerae O1 strains is dependent on rpoN (28). Sigma 54 is required for biofilm formation by Burkholderia cenocepacia as well as its survival within macrophages (51). In the major food-borne pathogen Listeria monocytogenes, σ54 is essential for its osmotolerance potential (41) and is responsible for mesentericin sensitivity (14, 41), whereas in Pseudomonas aeruginosa, σ54 influences the activity of isocitrate lyase (21), alginate biosynthesis (6), and pilin and flagellin production, in addition to several other virulence determinants (44). Sigma 54 also regulates biofilm formation, enterocyte effacement, acid tolerance, flagellar biosynthesis, and several other processes in Escherichia coli (3, 49, 60).
In E. faecalis, σ54 is responsible for sensitivity to class IIa bacteriocins, such as mesentericin and divercin (9, 13). The basis for the class IIa bacteriocin sensitivity is due to the role of σ54 in regulating four distinct sugar phosphotransferase systems (PTSs) that are dependent on four known σ54 enhancer binding proteins (LpoR, MphR, MpoR, and MptR) (13). MptD, a component of the mannose PTS, is thought to serve as the cellular receptor for the class IIa bacteriocins (24). However, additional roles for σ54 in enterococcal biology remain to be elucidated.
In this study, we investigated the role of σ54 in eDNA release, autolysis, and biofilm formation, and we demonstrate a functional role for σ54 in regulating initial adherence of cells to substrate as well as the overall composition of the biofilm matrix.
The bacterial strains and plasmids used in this study are listed in Tables 1 and and2,2, respectively. E. coli Electro10 Blue (E10-Blue; Stratagene) was used for construction of plasmids and was cultured in Luria-Bertani (LB) broth supplemented with appropriate antibiotics. E. faecalis strains were cultured in either Trypticase soy broth containing 0.25% glucose (TSB) or Todd-Hewitt broth (THB; BD Biosciences) containing appropriate antibiotics whenever required. Chloramphenicol (Cm) and spectinomycin (Spec) were used for selection of E. coli at concentrations of 10 μg/ml and 150 μg/ml, respectively. For E. faecalis, Cm, Spec, and tetracycline (Tet) were used at 15 μg/ml, 500 μg/ml, and 15 μg/ml, respectively. When required, X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; Amresco) was used at a concentration of 80 μg/ml for both E. coli and E. faecalis.
An E. coli-Enterococcus temperature-sensitive cloning vector, pLT06 (56), was used to generate an isogenic in-frame deletion of rpoN in E. faecalis V583. Upstream and downstream regions flanking rpoN (ef0782) were amplified by PCR from a V583 genomic template by using the primer pair RpoNP1/RpoNP2 and RpoNP3/RpoNP4, respectively (Table 3). The primers RpoNP1/RpoNP2 and RpoNP3/RpoNP4 were designed with EcoRI/BamHI and BamHI/PstI restriction sites, respectively. The resultant PCR products were digested with BamHI, ligated, and reamplified with primers RpoNP1 and RpoNP4. For the construction of the deletion vector, the amplified product was digested with EcoRI and PstI followed by ligation to similarly digested pLT06. The ligation was electroporated into competent E10-Blue cells for propagation, and blue colonies were selected on LB agar containing chloramphenicol and X-Gal at 30°C. Clones were screened for the appropriate insert by using the primers OriF and SeqR. A positive plasmid designated pKS70 was confirmed by restriction digestion and electroporated into E. faecalis V583 cells (12), and VI01 was subsequently generated following the protocol previously described (56) and confirmed by PCR using the primers RpoNUp and RpoNDown. Using pKS70, ~98% of the rpoN gene was deleted, leaving seven codons at the 5′ end and two codons at the 3′ end. The next adjacent gene is ef0783, which encodes an O-acetyltransferase. This gene is located approximately 200 bp downstream of rpoN, and the strategy used to delete rpoN does not alter the expression of ef0783 (data not shown).
The temperature-sensitive cloning vector pLT06 (56) was used to generate markerless gene complementation of rpoN in VI01. The rpoN gene (ef0782) along with flanking regions was amplified by PCR from a V583 genomic template using primers RpoNP1 and RpoNP4 (Table 3). For the construction of rpoN markerless complementation vector pVI12, the amplified product was digested with EcoRI and PstI followed by ligation with similarly digested plasmid vector pLT06. The ligation was electroporated into competent E10-Blue cells for propagation, and blue colonies were selected on LB agar containing chloramphenicol and X-Gal at room temperature. Clones were screened for the appropriate insert by using the primers OriF and SeqR. A positive plasmid designated pVI12 was confirmed by restriction digestion and electroporated into E. faecalis VI01 cells (12), and VI40 (markerless complement) was generated following the protocol previously described (56) and was confirmed by PCR using primers RpoNUp and RpoNDown.
E. faecalis V583, VI01, and VI40 were grown on LB agar containing 0.2% fructose and 10 mM 2-deoxy-d-glucose (2DG) (24). 2DG is a toxic homologue of glucose and enters cells via the mannose PTS permease (5). In E. faecalis, mannose PTS expression is controlled by σ54. Strains resistant to 2DG do not express a functional mannose PTS permease (24). Hence, growth on medium containing 2DG was used as a marker to confirm deletion of rpoN.
Overnight cultures were centrifuged for 10 min at 13,000 rpm, and the resulting supernatant was filtered (0.2-μm pore size; Nalgene) to obtain cell-free supernatants. The supernatants were tested for the presence of eDNA by using 1 μM SYTOX Green (Invitrogen, Molecular Probes).
The eDNA was also precipitated from the culture filtrate with an equal volume of isopropanol. The precipitated eDNA was washed in 75% ethanol, air dried, and dissolved in TE buffer (10 mM Tris-Cl, 1 mM EDTA; pH 8.0) and visualized on 1% agarose gels after staining with ethidium bromide.
The autolysis assay was performed as previously described (15).
eDNA in biofilms was quantified using a previously described protocol with some modifications (36). Briefly, biofilms were grown in a 96-well polystyrene plate in TSB for 24 h at 37°C. After 24 h, the supernatant was discarded and the biofilm was suspended in resuspension buffer (50 mM Tris-Cl [pH 8], 10 mM EDTA, 500 mM NaCl). The resuspended biofilm was centrifuged, and eDNA was quantified in the supernatant with 1 μM SYTOX Green (Invitrogen, Molecular Probes).
Confocal laser scanning microscopy (CLSM) was performed on 1-day-old biofilms as described previously (54). E. faecalis strains VI01 and VI40 were transformed with pMV158GFP (39) to generate VI29 and VI41, respectively, both of which expressed green fluorescent protein (GFP) constitutively. VT09 [V583(pMV158GFP)] (54) along with VI29 and VI41 were used for confocal imaging. Briefly, biofilms were grown on sterile glass coverslips placed in six-well tissue culture plates. The coverslip was submerged in 5 ml of TSB containing tetracycline for plasmid maintenance. After 24 h of growth, the biofilm was gently washed with sterile phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4 [pH 7.4]) and stained with 1 μM SYTOX orange (Invitrogen) for 6 to 7 min. The coverslip were inverted on a clean glass slide and sealed using clear nail polish. The biofilm was visualized using a Zeiss LSM 5 Pa laser scanning confocal microscope.
To visualize the biofilms formed by VT09, VI29, and VI41 macroscopically, biofilms were grown as described for confocal analysis, with the exception that after 24 h of growth, the biofilms were gently washed with sterile phosphate buffer, then fresh TSB was added, and the biofilms were grown for an additional 24 h, at which time the biofilm was washed and imaged with an AlphaImager system (Alpha Innotech, San Leandro, CA).
Adherence of E. faecalis strains to flat-bottom polystyrene plates (BRAND, Germany) was tested using a previously described protocol (27) with some modifications. Cultures grown overnight were diluted 1:10 in fresh TSB, and 200 μl was transferred to a flat-bottom 96-well polystyrene microtiter plate. After 2 h of incubation at 37°C, the supernatant was discarded and the wells were gently washed with sterile PBS. The adherent cells were resuspended in 200 μl PBS by vigorous pipetting, diluted, and plated on THB agar for colony counting. Also, the initial load was calculated by plating the diluted culture on THB agar for colony counting. The adherence potential of the strains was calculated as the percentage of initial load that adhered.
Biofilms were grown on 96-well round-bottom tissue culture-treated polystyrene plates (Techno Plastic Products, Switzerland) as previously described (22). At 6, 12, and 24 h, the biofilms were treated with 1 μg/ml proteinase K (Amresco), and this treatment remained for the remainder of the experiment. The 24-h treatment was allowed to stand for 1 h prior to processing the biofilm. An untreated control was included to determine the effect of treatment. After 24 h of growth, the biofilm was quantified by the crystal violet staining method (22). Each assay was performed in triplicate and repeated four times.
Statistical analysis of quantitative detection of eDNA, adherence assay results, and Comstat analysis of biofilms were performed using GraphPad Prism 4 software (San Diego, CA). One-way analysis of variance followed by Dunn's multiple comparison tests was performed to determine statistical significance.
The rpoN deletion mutant VI01 (ΔrpoN) was constructed using the markerless deletion vector pKS70. Initial growth curves of the wild-type V583 strain, the rpoN deletion mutant (VI01), and its complement (VI40) were assessed in TSB. No alterations in the growth of the 3 strains were observed (see Fig. S1 in the supplemental material). The 2DG-resistant phenotype was confirmed by growth on medium containing 2DG. VI01 grew to the final dilution of 10−8, while the parental strain V583 and the rpoN complement VI40 were significantly inhibited and grew only at dilutions of 10−3 and 10−4. Complementation confirmed that there were no polar effects of the gene deletion and attributed the 2DG-resistant phenotype to the targeted deletion of rpoN (Fig. 1).
On the basis of phenotype characterization from a preliminary transposon mutagenesis screen, we tested for eDNA in the supernatants of planktonic cultures by using SYTOX Green. A smaller amount of eDNA was detected in VI01 culture supernatants than with the wild-type V583 strain, whereas the markerless complementation of the rpoN mutant restored the phenotype to wild type (Fig. 2A). eDNA in the supernatants of the various strains was also confirmed by visualization on an ethidium bromide-stained 1% agarose gel after precipitation of eDNA with isopropanol (Fig. 2B).
Given the fact that planktonic growth and biofilms are two different lifestyles of the bacterium, we tested to see the effect of rpoN deletion on eDNA during biofilm development. As observed in planktonic cultures, a lesser amount of eDNA was detected in the VI01 biofilm than in the wild type, and this was attributed to the deletion of rpoN, as the complementation restored the eDNA detected in the biofilm to wild-type levels (Fig. 2C).
Because eDNA release in E. faecalis is dependent upon cell death by autolysis (54) and the rpoN mutant is defective in eDNA release, we hypothesized that σ54 may differentially modify the rate of autolysis in E. faecalis. In the autolysis assay, we observed that VI01 showed a significant decrease in the rate of autolysis, a phenotype readily complemented by introducing the gene in single copy to its native locus (Fig. 3).
eDNA has been shown to be an important matrix component in E. faecalis biofilms (54). The decreased levels of eDNA in VI01 led us to the hypothesis that VI01 may form less-dense biofilms than the wild-type V583. However, CLSM analysis of 24-h-old biofilms grown on glass coverslips showed that VI29 (ΔrpoN, GFP+) formed a thicker biofilm (as measured based on z-stack thickness, using an LSM image examiner) than those formed by the wild-type strain VT09 or the complemented strain VI41 (Fig. 4b). The appearance of the VI29 biofilm suggested early initiation of microcolony development, which was confirmed by macroscopic examination of the biofilms after 2 days of growth on coverslips (Fig. 4a). Despite the increased thickness and overall biofilm biomass of the rpoN mutant strain (VI29) compared to the parental and complemented strains (Table 4), very few random dead cells and DNA (as detected by SYTOX orange staining) were observed within the biofilm. Consistent with earlier observations on the role of cell death and of eDNA as a matrix component (54), regions within the wild-type VT09 contained concentrated foci of DNA and dead cells, which was phenocopied by the complement VI41 strain (Fig. 4b).
In order to determine whether increased biofilm formation by the rpoN mutant was due to its initial adherence ability, we calculated the percentage of the initial inoculum that adhered to 96-well microtiter plates after 2 h. Adherence of VI01 to a polystyrene plate was significantly enhanced in comparison to the wild type. In addition, markerless complementation of VI01 (VI40) reduced the adherence potential to wild-type levels (Fig. 5).
On the basis of the macroscopic observations and CLSM of the VI29 biofilm and the relative lack of eDNA detection for this mutant, we hypothesized a role for a different polymer matrix that promotes biofilm formation in the rpoN mutant. To test the role of proteins in VI01 biofilm, we examined the effect of proteinase K treatment on biofilm development. The wild-type and complemented strains exhibited decreased biofilm when treated with proteinase K only after 24 h of biofilm growth. In contrast, reduction in VI01 biofilm was significant when treated with proteinase K after 6 h of biofilm growth and continued to respond to treatment after 12 h and 24 h of biofilm growth (Fig. 6).
The role of σ54 in regulating numerous biological properties, including those related to virulence, has been well documented for a variety of bacterial species (3, 10, 21, 42, 49, 51, 52, 55, 58, 59). However, its role in E. faecalis has been limited to observations made regarding its contribution to sensitivity to class IIa bacteriocins through the regulation of sugar PTSs (9, 13, 22). Identification of σ54 as a potential regulatory protein in the cascade of biofilm development was an interesting breakthrough, and we focused our efforts on elucidating its effect on E. faecalis V583 biofilm. The role of autolysis (54) and fratricide (53) has been well documented in enterococcal biofilm formation and has been shown to be important in providing eDNA as a key biofilm matrix component. However, the observation that biofilm formation was enhanced in the rpoN mutant despite the increased resistance to autolysis and the absence of eDNA was an unexpected finding.
One possible explanation for the increased resistance to autolysis observed in the rpoN mutant could be novel modifications of the cell wall or alteration of the modifications, such as O-acetylation (43) or d-alanylation (17) on the cell wall that protect against lysis. Deletion of rpoN did not alter the autolysin profile of E. faecalis when the micrococcal cell wall was used as a zymogram substrate (data not shown), ruling out the possibility of inactive autolysins. Also, the deletion of rpoN did not have a measurable effect on the secretion of the extracellular proteases GelE and SprE (data not shown), which have been previously shown to contribute to autolysis in E. faecalis (54, 57). A significant reduction in cell death due to impaired cell lysis occurred in E. faecalis V583 (ΔrpoN) planktonic and biofilm cultures, suggesting the requirement of a functional σ54 for regulation of susceptibility to cell lysis. In P. aeruginosa, deletion of rpoN abolishes cell death in the microcolonies during biofilm maturation and has been related to the expression of surface structures (type 4 pili and flagella) whose expression is regulated by σ54 (56). Additionally, σ54-dependent gene regulation promotes phage-induced lysis in P. aeruginosa (10). There are seven phages associated with E. faecalis V583, with one of them being a part of the core genome (37). It will be interesting to test the role of σ54-dependent transcription of phage particle proteins and host cell lysis and the contribution of σ54 to biofilm development.
Enhancement of biofilm formation in the absence of a well-characterized matrix component in the rpoN mutant indicates that a substantial knowledge gap still exists in unraveling factors associated with E. faecalis biofilm development. Cellular processes regulated by σ54 are attractive in this regard, to begin revealing the interplay between metabolism and biofilm development, as one of the few characterized roles for σ54 is the regulation of four sugar PTS pathways. It is noteworthy that deletion of the genes encoding the four known enhancer binding proteins (LpoR, MphR, MpoR, and MptR) did not reduce eDNA release, impair autolysis, or alter biofilm development (data not shown), suggesting that σ54 might act as a repressor of genes independent of enhancer protein function. The idea that σ54 levels in the cell or within the population might be regulated raises an interesting experimental question. Our observation that the wild-type and rpoN complement strains could grow on 2DG at a much lower frequency (10−4 and 10−5, respectively) relative to the rpoN mutant parallels a recent report by Flanagan et al. (18) that resistance to the E. faecalis plasmid-encoded bacteriocin MC4-1 (a class IIa bacteriocin) was dependent on point mutations within the rpoN gene that occurred at high frequency (10−3 to 10−4). This resistance was shown to be reversible to a susceptible phenotype by point mutations that also occurred within rpoN as second-site suppressors. These combined observations suggest that there are hot spots for mutation within rpoN and could be a mechanism for phase variation within the E. faecalis population.
In Vibrio vulnificus, σ54 positively regulates the gene encoding ADP-glycero-manno-heptose-6-epimerase (gmhD), which is responsible for production of lipopolysaccharide and exopolysaccharide, both of which are required for biofilm formation (31), while in B. cenocepacia σ54 controls motility, which in turn plays a role in biofilm formation (51). However, in E. coli K-12, rpoN deletion enhances biofilm formation (3). Such different effects of rpoN on the biofilm-forming potential of bacteria provide a clear example of how the knowledge regarding a gene whose function was first reported to be restricted to nitrogen assimilation has evolved to govern virulence-related functions in addition to bacterial metabolism. Our data provide additional support for the expanding role of σ54 in the world of low-GC Gram-positive bacteria.
Biofilm formation is a multistep process that begins with the attachment of bacteria to a substrate, followed by colonization via further recruitment of more bacteria or by cell division. Initial attachment of a bacterial cell to a surface is an important stage in biofilm development and determines the fate of this process. In P. aeruginosa (8, 35), Staphylococcus aureus (11), and Streptococcus spp. (40) it has been shown that a defect in initial adherence of a bacterium affects biofilm formation and subsequently influences the virulence of these pathogens. Our data indicate a similar influence of the attachment process in enterococcal biofilm development, wherein deletion of rpoN increases the adherence potential of the pathogen, which subsequently results in a more dense biofilm.
Other than DNA, other molecules, such as proteins and polysaccharides, have been suggested to be important constituents in the polymer matrix of several bacteria (16, 19). Robust biofilm formation by VI01 despite the significant reduction in eDNA led us to test for the presence of other matrix components by using compounds capable of dissolving the aforementioned components. The reduced ability of VI01 (ΔrpoN) to form a biofilm when treated with proteinase K suggests a role for protein in either adhesion or matrix composition to promote E. faecalis biofilm and is consistent with recent observations by Guiton et al. (20). Those authors observed that colonization of an implanted piece of urinary catheter as well as the bladder epithelium was dependent on a functioning sortase enzyme for the proper anchoring of proteins to the cell wall, which in turn promoted cellular adhesion. In S. aureus, a biofilm defect in mutants that overproduce extracellular protease was rectified by the addition of α2-macroglobulin, a general protease inhibitor, indicating a vital role for proteins in either cellular adhesion or the biofilm matrix (4). Similarly, in Bacillus subtilis, TasA is required for the structural integrity and development of biofilms (7). In E. faecalis biofilms (54), eDNA is known to be a crucial matrix component in the early stages of biofilm development, but by 24 h of growth in the biofilm DNase has a minimal effect on disrupting the biofilm. Here we show that in E. faecalis V583, proteins are likely to serve as important matrix components during the later stages of biofilm development, as a reduction in biomass was observed only at 24 h and not at earlier time points. This suggests the time-dependent involvement of different polymers in the overall development of the biofilm.
Complementation studies of the rpoN mutant by using a low-copy-number plasmid did not result in complete reversal of the phenotype to wild-type levels in experiments that involved stressing of cells (osmotic shock and 2DG toxicity) (data not shown). This was primarily due to plasmid loss in the absence of selection and suggested a survival advantage for E. faecalis in the absence of σ54 under certain stress conditions. The inability to fully complement an rpoN mutant has also been reported for L. monocytogenes (41). Similarly, in a V. fischeri squid colonization model (58), the level of colonization varied with the complemented strain, and only some animals exhibited wild-type levels of colonization. For this reason, we utilized a complementation strategy that restored the function of the gene by placing it at its native locus in single copy.
A literature survey for σ54 and its biological roles revealed a bias toward Gram-negative species, with P. aeruginosa, Vibrio spp., and E. coli being the most studied. In an attempt to identify the distribution of rpoN in low-GC Gram-positive organisms, we performed a BLAST search using σ54 of E. faecalis V583 as the query. Among the organisms queried, only L. monocytogenes, B. subtilis, C. difficle, and C. perfringens appeared to have homologues, whereas in S. aureus, S. pneumoniae, and S. pyogenes homologues to σ54 were absent. The basis for this distribution among enterically adapted organisms as well as the potential genes regulated by σ54 await further study.
In conclusion, the results from this study show that σ54 in E. faecalis V583 contributes to cell death and eDNA release and that in its absence, E. faecalis adapts an alternate matrix to establish biofilms. Understanding the mechanism underlying the phenotypes observed in this study is the main focus of ongoing studies in our laboratory.
We are grateful to Vinai Thomas for the preliminary screen of the E. faecalis transposon library and to Dan Boyle for assistance with confocal imaging.
This work was supported by Public Health Service grant AI77782 from the National Institutes of Health.
Published ahead of print 11 November 2011
Supplemental material for this article may be found at http://jb.asm.org/.