ABSTRACT
Tonsils conduct immune surveillance of antigens entering the upper respiratory tract. Despite their immunological function, they are also sites of persistence and invasion of bacterial pathogens. Actinobacillus suis is a common resident of the tonsils of the soft palate in pigs, but under certain circumstances it can invade, causing septicemia and related sequelae. Twenty-four putative adhesins are predicted in the A. suis genome, but to date, little is known about how they might participate in colonization or invasion. To better understand these processes, swine tonsil lysates were characterized by mass spectrometry. Fifty-nine extracellular matrix (ECM) proteins were identified, including small leucine-rich proteoglycans, integrins, and other cell surface receptors. Additionally, attachment of the wild type and 3 adhesin mutants to 5 ECM components was evaluated. Exponential cultures of wild-type A. suis adhered significantly more than stationary cultures to all ECM components studied except collagen I. During exponential growth, the A. suis Δflp1 mutant attached less to collagen IV while the ΔompA mutant attached less to all ECMs. The ΔcomE1 strain attached less to collagen IV, fibronectin, and vitronectin during exponential growth and exhibited differential attachment to collagen I over short adherence time points. These results suggest that Flp1, OmpA, and ComE1 are important during early stages of attachment to ECM components found in tonsils, which supports the notion that other adhesins have compensatory effects during later stages of attachment.
INTRODUCTION
Tonsils are secondary lymphoid organs that are an important part of the host immune response against antigens entering the body through the mouth and nose. Pigs have 5 sets of tonsils located around the oropharyngeal cavity in the so-called Waldeyer's ring, the largest being the tonsils of the soft palate (1), which are thought to be functionally equivalent to the palatine tonsils in humans (2). The tonsils of the soft palate of swine have as many as 200 branching crypts that end blindly within the lymphoid tissue beneath (3). The lumina of the crypts are covered by a layer of nonkeratinized squamous epithelium that is continuous with the epithelium of the oral cavity, while lymphoepithelium lines the deeper areas of the crypts (4). Lymphoepithelium has a thinner epithelial lining interspersed with goblet cells and M cells, the latter of which can be used by some pathogens to invade the tonsil (3, 5). The lymphoepithelium also allows passage of lymphocytes, plasma cells, and macrophages (2). The epithelium of the crypts overlays a basement membrane (sometimes fragmented), which in turn is supported by an interstitial extracellular matrix (ECM) of various thickness, depending on the location in the tonsil (3). Despite studies of the ultrastructure and fine structure of the tonsils of swine (1, 3) and studies into the ECM components and host cell receptors present in tonsils of other species (6–8), little has been done to characterize these components in the tonsils of the soft palate of swine.
Components of the ECM have important functions in the host, providing both structure to tissues and an anchor to the overlying epithelium (9). Cell signaling also occurs between the epithelial cells and the ECM components via integrins and other receptors expressed on the basolateral side of the cell surface that can affect the morphology of the cytoskeleton (9). The ECM components, including glycoproteins, proteoglycans, and other molecules, are present in different regions of the connective tissue. The specific composition of the ECM can vary depending on the body site. For example, laminin and collagen IV form independent networks in the basement membrane and are interconnected by nidogens and perlecan (9). Fibronectin and collagen I are ubiquitous ECM components in the connective tissues of the body and are mainly found in the interstitial matrix (9), although fibronectin can also be found in a soluble form in plasma in humans (10). Another adhesive glycoprotein, vitronectin, is also abundant in blood plasma and in the ECM, where it participates in adhesion and tissue repair, assists in formation of the membrane attack complex, and promotes neutrophil infiltration during infection (11). There are also extracellular proteoglycans called small leucine-rich proteoglycans (SLRPs) that serve the role of “tissue organizers” during development, attachment, and wound repair by interacting with fibrillar collagens, growth factors, and host receptors (12). Other important components of the ECM include hyaluronan, numerous types of collagen, elastin and its associated proteins, proteases such as matrix metalloproteases and plasmin(ogen), growth factors, and many other molecules (13).
In addition to their role in host tissue integrity and structure, the components of the ECM are also used by many bacteria as a site of attachment, colonization, and sometimes invasion (2). It should be noted that in order for bacteria to gain access to the ECM, they must first breach the barrier of the epithelium. This can be accomplished when damage occurs to the epithelium that exposes the ECM, by breaking the tight junctions between epithelial cells, by invading the epithelial or M cells and emerging on the basolateral side, or by being taken across the epithelial barrier by leukocytes (2, 3, 14, 15).
The processes of attachment and colonization are mediated primarily by fimbrial and afimbrial bacterial adhesins and various host cell receptors, including ECM components. A limited number of studies has been done to identify host receptors in the oral cavity of swine, but among members of the family Pasteurellaceae that are human pathogens, the trimeric autotransporter adhesin DsrA of Haemophilus ducreyi has been shown to bind to fibronectin and vitronectin in vitro (16), and the EmaA autotransporter of Aggregatibacter actinomycetemcomitans binds to collagen V (17). In addition, the purified lipoprotein e (P4) of nontypeable Haemophilus influenzae binds with high affinity to laminin, fibronectin, and vitronectin (18), while the OmpA outer membrane protein (OMP) of the important veterinary pathogen Pasteurella multocida adheres to cell surface fibronectin (19), as does OmpA of Mannheimia haemolytica (20). The ComE1 protein, which is present in several members of the family Pasteurellaceae, has also been shown to attach to fibronectin (21).
Actinobacillus suis, a member of the family Pasteurellaceae, is a Gram-negative facultative anaerobe that is a common resident of the tonsils of the soft palate of swine (22). Typically, it resides harmlessly at this site, but unknown factors can cause this organism to invade the bloodstream, resulting in often fatal septicemia and sequelae such as meningitis, pleuritis, and arthritis (23). Little is known about the pathogenesis of A. suis, including how it attaches to and colonizes the tonsils, where it typically resides, or by what mode it enters the bloodstream. Previous studies have shown that an OmpA homologue may be important for A. suis infection (24) and crossing the blood-brain barrier (25), but nothing is known about the host receptor for OmpA or the receptors for the other 23 putative adhesins identified by characterizing the genome of A. suis (26). Therefore, the first objective of this study was to identify ECM components in the tonsils of the soft palate of swine, and the second objective was to investigate the possible interaction of 3 putative adhesins with some of these molecules. The adherence of exponentially growing and stationary-phase cultures of isogenic mutants of flp1, comE1, and ompA to selected ECM components was compared to that of the wild type. The flp1 gene encodes the pilin of the type IVb pilus of the tight adherence (tad) locus, which is involved in biofilm formation in many members of the family Pasteurellaceae and other Gram-negative organisms (27). The comE1 gene encodes a putative fibronectin-binding protein previously described by Mullen et al. (21) in other members of the Pasteurellaceae. The A. suis ompA gene encodes an OMP that attaches to swine tonsil explants and brain microvascular epithelial cells (25).
MATERIALS AND METHODS
Bacterial strains and growth conditions.Strains and plasmids used in this study are listed in Table 1. Actinobacillus suis H91-0380, a virulent O2:K2 clinical isolate (28, 29), and unmarked isogenic mutants (Δflp1, ΔompA, and ΔcomE1 strains) generated from this strain (A. R. Bujold, J. Labrie, M. Jacques, and J. I. MacInnes, submitted for publication) were grown in brain heart infusion (BHI; BD, Sparks, MD) as previously described (31).
Strains, plasmids, and genes used to characterize differential attachment to ECM components
Tonsil preparation for MS.Approximately 1 g of frozen tonsil tissue, collected from pigs exhibiting no signs of clinical disease, was minced and added to 3 ml ice-cold phosphate-buffered saline (PBS). The sample was fully homogenized using a Bio-Gen PRO200 homogenizer (Diamed, Mississauga, ON) in five 15-s intervals at a speed ramping from 5,000 to 35,000 rpm. The homogenate was freeze-thawed twice at −70°C and spun at 10,000 × g at 4°C for 20 min to remove cellular debris. Total protein concentration was quantified by bicinchoninic acid (BCA) assay (Pierce, Thermo Fisher Scientific, Waltham, MA) and diluted to a final concentration of 1 mg/ml prior to trypsinization and analysis by mass spectrometry (MS).
The procedure described above was conducted a second time using Tris-buffered saline (TBS) instead of PBS. TBS homogenate was further processed using the anionic cell lysis kit SP810 (Protea Biosciences, Inc., Morgantown, WV) according to the manufacturer's instructions, including the degradation of the anionic acid labile surfactant (AALS) II. This AALS tonsil lysate was trypsinized and fractionated by gradual elution by strong cation exchange chromatography from SCX SpinTips (Protea Biosciences, Inc., Morgantown, WV) using increasing concentrations of ammonium formate (20 mM to 500 mM) in 10% acetonitrile.
The crude lysates in PBS and TBS and the AALS processed lysates were submitted to the SPARC BioCentre at the Hospital for Sick Children (Toronto, ON); the fractionated AALS lysate samples were submitted to the University of Guelph Advanced Analysis Centre (Guelph, ON) for analysis by MS.
MS.Samples analyzed at the SPARC BioCentre were trypsinized, and the peptides were loaded onto a 150-μm-inner-diameter (i.d.) precolumn (Magic C18; Michrom Biosciences, Bruker Co., Billerica, MA) at 4 μl/min and separated over a 75-μm-i.d. analytical column packed into an emitter tip containing the same packing material. The peptides were eluted over 60 min at 300 nl/min using a 0 to 40% acetonitrile gradient in 0.1% formic acid with an EASY-nLC nanochromatography pump (Proxeon Biosystems, Odense, Denmark). The peptides were eluted into an LTQ-Orbitrap hybrid mass spectrometer (Thermo-Fisher, Bremen, Germany) operated in a data-dependent mode. MS was acquired at 60,000-FWHM (full width at half maximum) resolution in the Fourier transform mass spectrometer (FTMS), and tandem MS (MS/MS) was carried out in the linear ion trap. Six MS/MS scans were obtained per MS cycle. The raw data were searched using Mascot 2.3.02 (Matrix Sciences, London, United Kingdom), and search results were analyzed using Scaffold 3.4.3 (Proteome Software Inc., Portland, OR) with Swiss-Prot, UniProt (http://www.uniprot.org/), and Ensembl (http://www.ensembl.org/) databases queried. Protein and peptide thresholds were set to 95% with a minimum of two peptides detected for positive protein identification.
The fractionated samples analyzed at the Advanced Analysis Centre were trypsinized and loaded on a 1200 high-performance liquid chromatography (HPLC) device (Agilent Technologies Co. Ltd., Santa Clara, CA) interfaced with a UHD 6530 Q-TOF mass spectrometer (Agilent Technologies Co. Ltd., Santa Clara, CA). A C18 column (100 mm by 2.1 mm; volume, 2.7 μm; AdvanceBio peptide mapping column; Agilent) was used for chromatographic separation using the solvents water with 0.1% formic acid for solvent A and acetonitrile with 0.1% formic acid for solvent B. The following mobile-phase gradient procedure was used: initial conditions, 2% B increasing to 45% B in 40 min and then to 55% B in 10 min, followed by column wash at 95% B and 10 min of reequilibration. The first 2 and last 5 min of gradient were sent to waste rather than the spectrometer. The flow rate was maintained at 0.2 ml/min, the mass spectrometer electrospray capillary voltage at 4.0 kV, and the drying gas temperature at 350°C with a flow rate of 13 liters/min. The nebulizer pressure was 40 lb/in2, and the fragmentor was set to 150 lb/in2. Nitrogen was used as nebulizing and drying gas as well as collision-induced gas. The mass-to-charge ratio was scanned across the m/z range of 300 to 2,000 m/z in 4-GHz extended dynamic range positive-ion auto-MS/MS mode. Three precursor ions per cycle were selected for fragmentation. The instrument was externally calibrated with an ESI TuneMix (Agilent Technologies Co. Ltd., Santa Clara, CA). The sample injection volume was 100 μl. Raw data files were loaded directly into PEAKS 7 software (Bioinformatics Solutions Inc., Waterloo, ON), where the data were refined and subjected to de novo sequence identification and database searching. Methionine oxidation modifications were considered within the search parameters. The tolerance values used were 10 ppm for parent ions and 0.5 Da for fragment ions.
Mutant generation.Unmarked isogenic mutants were generated for the flp1, ompA, and comE1 genes of A. suis H91-0380 using a modification of the sacB transconjugation method (32), as previously described (Bujold et al. submitted). Briefly, primers were designed with adapters for restriction enzyme (RE)-cut sites to PCR amplify 5′ (flanked by SalI and XhoI cut sites) and 3′ (flanked by XhoI and NotI cut sites) fragments of each gene. The PCR products for each gene were digested with appropriate REs, purified, and ligated into the pEMOC2 plasmid that was linearized with SalI and NotI REs. The plasmid construct was electroporated into Escherichia coli β2155 and mobilized into A. suis H91-0380 by filter mating at 37°C and 5% CO2 for 16 to 24 h on BHI agar plates containing 1 mM diaminopimelic acid (DAP) and 10 mM MgSO4. Cells were harvested from the filter by centrifugation and resuspended in 500 μl sterile PBS, and serial dilutions were plated on BHI agar containing 5 μg/ml chloramphenicol and incubated as before for 24 to 48 h. Overnight cultures of transconjugants (confirmed by colony PCR) were subjected to counterselection by growth in Mueller-Hinton (MH) broth at 37°C and subsequent plating on MH agar containing 10% sucrose, followed by incubation at 30°C for up to 48 h. Mutants were confirmed by colony PCR and sequencing.
Comparison of human and pig ECM components used for ECM attachment assays.The sequence identity and similarity of the human and pig ECM components fibronectin, vitronectin, laminin, collagen I, and collagen IV, used for ECM attachment assays with A. suis, were analyzed with ClustalW (http://www.genome.jp/tools/clustalw/) for multiple-sequence alignment and with the “Ident and Sim” feature of the Sequence Manipulation Suite (www.bioinformatics.org/sms). A search for conserved protein motifs was also done for human and pig vitronectin and collagen I using MotifFinder (http://www.genome.jp/tools/motif/) and the PROSITE, Pfam, and NCBI-CDD libraries.
Attachment assays.Single colonies of A. suis H91-0380 wild-type, Δflp1, ΔompA, and ΔcomE1 strains grown on Columbia agar plates containing 5% sheep's blood (BAPs) were used to inoculate 3 ml of BHI and incubated overnight at 37°C with shaking at 200 rpm. A 500-μl inoculum was added to 25 ml prewarmed BHI and incubated as before. Culture was removed at 60 min postinoculation (mpi; exponential phase) and 180 mpi (stationary phase) for all strains, except for the ΔompA strain, where the stationary-phase culture was taken at 210 mpi to compensate for a lower growth rate, as determined in growth curve studies (see Fig. S1 in the supplemental material). The number of CFU per milliliter was determined by plating 10-fold dilutions on BAPs. For attachment assays, 100 μl exponential- and stationary-phase cultures of each strain was added to Stripwell plates precoated with one of the purified human extracellular matrix components: collagen I, collagen IV, fibronectin, laminin, and vitronectin (EMD Millipore, Billerica, MA). The sequence identity of the human and pig ECMs was evaluated to ensure homology (see Table 3). In each experiment, one ECM-coated well inoculated with cell-free BHI and one bovine serum albumin (BSA)-coated well inoculated with bacterial culture were used as controls. After addition of the culture, wells were sealed with adhesive film (Fisher Scientific, Waltham, MA), centrifuged at 1,500 rpm for 10 min, and then incubated at 37°C in the presence of 5% CO2. At 0, 15, 30, 45, 60, and 120 min postattachment (mpa), culture was removed from wells for each ECM component and washed with PBS three times. The negative-control wells were incubated for 120 mpa before washing.
When the assays were complete, cells were heat fixed in the wells at 55°C for 20 min. Ninety-five microliters of crystal violet (1%) was then added to each well, and the strips were incubated at room temperature for 45 min. Following incubation with crystal violet, the wells were washed 5 times with sterile deionized water, air dried for 30 min at room temperature, and then destained with 100 μl 95% ethanol for 15 min. A plate reader (Beckman Coulter Inc., Brea, CA) was used to measure absorbance at 595 nm, blanked with the absorbance from the culture in the BSA-coated well. Blanked absorbance measurements were standardized to A595/input CFU based on plate counts.
Attachment assays were conducted in biological triplicate on three different days for each strain.
Statistical analysis.Wild-type profile data were analyzed using PROC MIXED repeated-measures analysis with a Tukey's post hoc test in SAS (v. 9.4; SAS Institute Inc., Cary, NC). The statistical model was Y = μ + culturei + timej + culture × timeij + εijk, where time was the repeated measure. Variance/covariance structures were tested, and the variance/covariance structure with the lowest Akaike information criterion/Bayesian information criterion (AIC/BIC) value was used for analysis. Wild-type post hoc planned contrasts among the culture × timeij interactions were done to compare (i) different time points in the same growth phase and (ii) the same time points between the exponential and stationary growth phases.
Mutant data were analyzed using PROC MIXED repeated-measures analysis with a Tukey's post hoc test in SAS. The statistical model was Y = μ + straini + culturej + timek + strain × cultureij + strain × culture × timeijk + εijkl, where time was the repeated measure. Variance/covariance structures were tested, and the variance/covariance structure with the lowest AIC/BIC value was used for analysis. Post hoc planned contrasts among the strain × culture × timeijk interactions were done to compare (i) different time points for the same strain and same growth phase, (ii) the same time points for the same strain and between the 2 growth phases, and (iii) the same time points among different strains within the same growth phase.
RESULTS
Proteins in pig tonsils.A total of 1,677 proteins were found in the tonsil samples analyzed by MS (see Table S1 in the supplemental material). Among these, 59 proteins of interest were identified: 35 ECM components, 5 members of the small leucine-rich proteoglycan (SLRP) family, 7 receptors, and 12 additional proteins (Table 2). The ECM components included fibronectin, laminin, vitronectin, collagen I, and collagen IV.
Extracellular matrix proteins identified in tonsils of the soft palate of swine by MS and bioinformatic analysis
Comparison of common human and pig ECM components.ClustalW, Ident and Sim, and MotifFinder were used to determine if the human ECM components fibronectin, laminin, vitronectin, collagen I, and collagen IV were comparable to homologous pig proteins (Table 3). Human and pig fibronectin, laminin, and collagen IV share >90% identity and similarity. The human and pig vitronectin proteins are 69.9% identical and 75.1% similar, while collagen I human isoform COL1A1 and pig COL1A2 are 60.6% identical and 67.3% similar.
Protein sequence identity and similarity of human and pig ECM components used for ECM attachment assays
Human and pig vitronectin proteins contain the same motifs: a somatomedin B (SMB) domain at the N terminal followed by an RGD motif and 4 hemopexin motifs (11). Both human and pig collagen I isoforms contain multiple collagen triple helix repeats (various numbers in human and pig) and a fibrillar collagen C-terminal domain. Only the human sequence contains a von Willebrand factor type C domain at the N terminus, which appears to be a conserved motif of COL1A1 but not COL1A2 (33, 34).
Attachment of A. suis H91-0380 wild type to ECM components.To determine whether A. suis H91-0380 wild type adhered to the common ECM components collagen I, collagen IV, fibronectin, laminin, and vitronectin, attachment was measured for exponential- and stationary-growth-phase cultures incubated for 0, 15, 30, 45, 60, and 120 mpa (Fig. 1). A significant reduction in attachment was observed in stationary-phase compared to exponential-phase cultures with collagen IV, fibronectin, laminin, and vitronectin. No significant effects were observed for wild-type attachment at different times to any ECM component tested.
Attachment of A. suis H91-0380 to purified ECM components during exponential (exp) and stationary (sta) growth. Attachment results have been corrected for nonspecific binding by subtracting the average signal from BSA-coated wells from each experimental sample. Error bars represent standard errors of the means (SEM). Statistical significance is expressed with asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Each data point is an average from three independent biological replicates.
Attachment of A. suis mutants to ECM components.The attachment of three A. suis H91-0380 knockout mutants (Δflp1, ΔompA, and ΔcomE1 strains) to various purified human ECM components was also measured and compared to the wild type. The average attachment of the Δflp1 mutant to collagen IV during exponential phase was significantly less than that of the wild type (Fig. 2). There was also no measurable attachment of the Δflp1 mutant to collagen I during exponential phase compared to background (BSA-coated well) signal; however, this result was not significant due to variations between biological replicates.
Attachment of A. suis H91-0380 wild-type and Δflp1 strains to ECM components during exponential (exp) and stationary (sta) growth. Attachment results have been corrected for nonspecific binding by subtracting the average signal from BSA-coated wells from each experimental sample. Error bars represent standard errors of the means (SEM). Statistical significance is expressed with asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Each data point is an average from three independent biological replicates.
There was significantly less attachment of the ΔompA mutant during exponential growth than the wild type with all ECMs tested (Fig. 3). In the case of collagen I and collagen IV, there was no measurable attachment during exponential phase relative to background signal.
Attachment of A. suis H91-0380 wild-type and ΔompA strains to ECM components during exponential (exp) and stationary (sta) growth. Attachment results have been corrected for nonspecific binding by subtracting the average signal from BSA-coated wells from each experimental sample. Error bars represent standard errors of the means (SEM). Statistical significance is expressed with asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Each data point is an average from three independent biological replicates.
Compared to the wild type, there was significantly less attachment of the ΔcomE1 strain during exponential growth to collagen IV, fibronectin, and vitronectin (Fig. 4). Furthermore, differential attachment over time (strain-culture-time effects) was observed for collagen I (Fig. 5), where differences in attachment between wild-type and ΔcomE1 strains during exponential phase were observed at 0, 30, 45, and 120 mpa.
Attachment of A. suis H91-0380 wild-type and ΔcomE1 strains to ECM components during exponential (exp) and stationary (sta) growth. Attachment results have been corrected for nonspecific binding by subtracting the average signal from BSA-coated wells from each experimental sample. Error bars represent standard errors of the means (SEM). Statistical significance is expressed with asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Each data point is an average from three independent biological replicates.
Attachment of A. suis H91-0380 wild-type and ΔcomE1 strains during exponential (exp) and stationary (sta) phase to collagen I over different incubation times. Attachment results have been corrected for nonspecific binding by subtracting the average signal from BSA-coated wells from each experimental sample. Error bars represent standard errors of the means (SEM). Statistical significance is expressed with asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Each data point is an average from three independent biological replicates.
DISCUSSION
In this study, we sought to identify ECM components and host receptors present in the tonsils of the soft palate of swine. We further set out to characterize the attachment of a virulent clinical isolate of A. suis and three isogenic adhesin gene mutants to purified components of the ECM. Tonsils from healthy pigs were analyzed by mass spectrometry to identify ECM components and host cell receptors that might be used by bacteria for host-pathogen interactions. Among the larger and more common ECM components, collagen types I, III, IV, V, VI, VII, XII, XIV, XV, and XVIII were identified in this study, as were fibronectin, several laminin isoforms, and vitronectin. To our knowledge, this is the first report of the characterization of ECM components in swine tonsils. Collagen types I, II, III, IV, VII, and XVIII have been found in human tonsils, as have fibronectin, laminin, vitronectin, tenascin, and thrombospondin (7, 8). Maatta et al. (7) also detected collagen type XVII in the crypt epithelium of human palatine tonsils, but this type of collagen was not identified in the current study; thrombospondin also was not detected. This finding was unexpected given its reported presence in the basement membrane and vessels of the extrafollicular areas of human tonsils (8). Of the other ECM components identified by MS in this study, the SLRP lumican has been detected in low levels in human tonsil (35). Some of the proteins detected (Table 2), such as fibrinogen, are important in wound healing, and it therefore seems reasonable that it would be present in tonsils to aid in tissue repair.
Due to their abundance in host connective tissues and the high sequence similarity between human and pig ECMs (Table 3), the ability of A. suis and A. suis adhesin mutants to bind to purified human collagen I, collagen IV, fibronectin, laminin, and vitronectin was evaluated. As might have been predicted from earlier expression studies (Bujold et al., submitted), there was significantly greater attachment of A. suis H91-0380 wild type to collagen IV, fibronectin, laminin, and vitronectin during exponential phase relative to stationary phase (Fig. 1).
In the isogenic Δflp1 mutant, the pilin gene of the tight adherence (tad) locus was disrupted. Previous studies of members of the family Pasteurellaceae (Actinobacillus, Aggregatibacter, Haemophilus, and Pasteurella), as well as Yersinia and other genera, have demonstrated that the type IVb pili encoded by the tad locus are involved in biofilm formation and host colonization (27, 36). To date, no host receptor has been described for these pili, and it has been proposed that the type IVb pili engage in nonspecific attachment through autoaggregation (37). In this work, there was no difference in attachment of the Δflp1 mutant to ECM components compared to the wild type with most of the ECM components tested, with the exception of collagen IV attachment, which was significantly reduced in exponential cultures (Fig. 2). Furthermore, there appeared to be little attachment of the Δflp1 mutant to collagen I, but due to the degree of variation between biological replicates, no significant difference between the Δflp1 mutant and wild-type cultures at exponential phase was detected. These results suggest that the type IVb pili encoded by the tad locus could contribute to more than just autoaggregation and that these pili may attach to collagen IV, and perhaps collagen I, although further work remains to be done to confirm the latter assertion. Further, this finding is noteworthy because collagen IV is a component of the basement membrane, which suggests that the tad locus would aid organisms in attaching at this site after having breached the epithelial barrier.
We next examined the attachment of the ΔompA mutant of A. suis to ECM components. The outer membrane protein OmpA has been previously shown to confer cell viability and serum resistance and to participate in attachment of A. suis to tonsil explants (25) and of Mannheimia haemolytica to bronchial epithelial cells and fibronectin (20, 38). In the current study, we observed significantly less attachment of the ΔompA mutant (versus the wild type) during exponential phase to all ECM components tested, even when corrected for lower growth rate (Fig. 3). It might also be noted that in previous studies (Bujold et al., submitted), the ΔompA mutant produced significantly less biofilm than the wild type or other mutant strains. In our previous studies, we found the expression of ompA to be upregulated in aerobic growth, particularly during stationary phase (Bujold et al., submitted). Therefore, it is somewhat surprising that the attachment of the ΔompA mutant was not also reduced compared to that of the wild-type strain during stationary phase. This may have been the result of other adhesins participating in attachment to ECM components during stationary phase, thereby compensating for the effects of ΔompA mutant attachment at this time point, although further studies are needed to test this hypothesis.
Homologues of comE1 in other members of the family Pasteurellaceae have been shown to bind to fibronectin, and in Pasteurella multocida, this interaction has been shown to involve the FnIII9–10 region of fibronectin (21, 39). With the A. suis ΔcomE1 mutant, there was reduced attachment of exponential cultures to collagen IV, fibronectin, and vitronectin compared to the wild-type strain (Fig. 4). In addition to these findings, there was differential binding of exponential cultures of the ΔcomE1 mutant to collagen I over time (Fig. 5), whereby attachment at 0, 30, 45, and 120 mpa was attenuated compared to that of the corresponding wild-type cultures. These results suggest that binding of A. suis to ECM components changes over short periods of time, although the factors that contribute to this phenotype are not yet clear. It also appears that, in addition to previous findings by Mullen et al. (21), the comE1 homologue of A. suis participates in more than just fibronectin binding. It would also be worthwhile to examine whether the arginyl-glycyl-aspartic acid (RGD) motif is important for interactions of comE1 with these ECM components and perhaps other proteins not examined here. This hypothesis is supported by the fact that collagen IV, fibronectin, laminin, vitronectin, and over 100 other proteins contain RGD motifs (13). The importance of RGD motifs in cell adhesion and their recognition by many integrins (40) suggests that more specific examination of the interactions of these motifs with A. suis is warranted in the future.
Although it is widely accepted that initial attachment and colonization of the host is an important early step in the pathogenesis of many organisms, little is known about these mechanisms for A. suis in the tonsils of the soft palate of swine. Here, we have for the first time identified several ECM components, including glycoproteins, proteoglycans, and other proteins in these tonsils. We have further demonstrated that A. suis can attach to purified collagen IV, fibronectin, laminin, and vitronectin during exponential growth and to a lesser degree during the stationary phase. Using knockout mutants, we have demonstrated that the biofilm-associated Flp1 pilin encoded by the tad locus is required for optimal binding of A. suis to collagen IV. In contrast, previous work has failed to identify a receptor (27). In addition, we found that disrupting the gene encoding the outer membrane protein OmpA resulted in reduced attachment to all ECM components tested. Finally, we found that disrupting comE1, a homologue of a previously described fibronectin-binding protein, resulted in reduced attachment of A. suis to collagen IV, fibronectin, and vitronectin, as well as reduced attachment to collagen I in a time-dependent manner. Taken together, these findings suggest that ECM proteins present in the tonsils of the soft palate of swine, and presumably other species, play a role in the colonization of bacterial pathogens. The characterization of A. suis attachment to the selected ECM components provides a basis for further studies into the mechanisms of attachment, and perhaps even invasion, of A. suis and other members of the family Pasteurellaceae via the tonsils.
ACKNOWLEDGMENTS
We thank Anne Laarman for assistance with statistical analysis.
This work was funded by discovery grants from the Natural Sciences and Engineering Research Council of Canada to J.I.M. A.R.B. was supported by an Ontario Veterinary College Ph.D. Scholarship and an Ontario Graduate Scholarship.
FOOTNOTES
- Received 31 May 2016.
- Returned for modification 16 June 2016.
- Accepted 22 July 2016.
- Accepted manuscript posted online 1 August 2016.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00456-16.
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
