ABSTRACT
Aggregatibacter actinomycetemcomitans is hypothesized to colonize through the interaction with collagen and establish a reservoir for further dissemination. The trimeric adhesin EmaA of A. actinomycetemcomitans binds to collagen and is modified with sugars mediated by an O-antigen polysaccharide ligase (WaaL) that is associated with lipopolysaccharide (LPS) biosynthesis (G. Tang and K. Mintz, J. Bacteriol. 192:1395–1404, 2010). This investigation characterized the function and cellular localization of EmaA glycosylation. The interruption of LPS biogenesis by using genetic and pharmacological methods changed the amount and biophysical properties of EmaA molecules in the outer membrane. In rmlC and waaL mutant strains, the membrane-associated EmaA was reduced by 50% compared with the wild-type strain, without changes in mRNA levels. The membrane-associated EmaA protein levels were recovered by complementation with the corresponding O-polysaccharide (O-PS) biosynthetic genes. In contrast, another trimeric autotransporter, epithelial adhesin ApiA, was not affected in the same mutant background. The inhibition of undecaprenyl pyrophosphate recycling by bacitracin resulted in a similar decrease in the membrane-associated EmaA protein. This effect was reversed by removal of the compound. A significant decrease in collagen binding activity was observed in strains expressing the nonglycosylated form of EmaA. Furthermore, the electrophoretic mobility shifts of the EmaA monomers found in the O-PS mutant strains were associated only with the membrane-associated protein and not with the cytoplasmic pre-EmaA protein, suggesting that this modification does not occur in the cytoplasm. The glycan modification of EmaA appears to be required for collagen binding activity and protection of the protein against degradation by proteolytic enzymes.
INTRODUCTION
The periodontal pathogen Aggregatibacter actinomycetemcomitans can be recovered from multiple physiologic sites within the oral cavity, as well as from extraoral infective lesions (20, 23). The colonization of disparate tissues by this microorganism depends on the expression of adhesins, including fimbrial and nonfimbrial adhesins (31, 37). Autotransporter proteins (type V secretion system) are well-represented among these nonfimbrial adhesins, which include the epithelial adhesin Aae (24), the multifunctional adhesin ApiA (Omp100) (17, 37), and the extracellular matrix protein adhesin A (EmaA) (16, 33). Generally, autotransporter proteins contain an N-terminal passenger domain, which typically mediates a virulence function, and a C-terminal β-barrel domain for outer membrane insertion (8). The adhesins of A. actinomycetemcomitans vary in molecular mass and the number of proteins required for forming a biologically functioning cell surface structure. A single Aae monomer (101 kDa) is sufficient for binding to epithelial cells (8, 24), whereas three identical monomers are required for the final assembly of the functional trimeric outer membrane structures for either ApiA (32 kDa) (13) or EmaA (202 kDa) (25).
The EmaA monomers are synthesized as a preprotein by the ribosome in the cytoplasm, and the pre-EmaA protein contains a 56-amino-acid signal peptide that is removed after translocation across the inner membrane (11). Following inner membrane translocation, the C termini of three EmaA monomer proteins are proposed to integrate into the outer membrane to form a pore for the translocation of the remaining N-terminal portions of the proteins across the outer membrane, as suggested for other trimeric autotransporters (8). EmaA molecules assemble into antenna-like appendages on the surface of the bacterium and specifically bind to collagen (16, 25, 32). It has also been proposed that collagen interacts with the head domain of EmaA, which correspond to the N termini of the proteins, or the globular ends of the antenna-like cell surface appendages (1, 36). The EmaA structures, which extend at least 150 nm from the surface, have been implicated in the initiation of infective endocarditis (31).
Recent studies suggested that EmaA undergoes posttranslational modification with sugars that are associated with the O-antigen polysaccharides (O-PS) moiety of A. actinomycetemcomitans lipopolysaccharides (LPS) (32). Protein glycosylation in bacteria is suggested to occur by either direct attachment of the sugar to the protein by cytoplasmic oligosaccharyltransferases or by utilization of enzymes that are shared or evolutionarily related to LPS biosynthetic enzymes (9). The machinery of LPS biogenesis involves the attachment of O-PS sugars onto the lipid carrier undecaprenyl monophosphate (UndP), through phosphorylation and the formation of UndPP (undecaprenyl pyrophosphate)-linked O-PS sugars (19, 22). The O-PS or glycans are translocated across the inner membrane by translocases and conjugated to either the lipid A core oligosaccharide (for the biosynthesis of LPS) or protein substrates (protein glycosylation) by either oligosaccharyltransferases or O-antigen ligases (9). The O-antigen ligase (Waal) of A. actinomycetemcomitans, which is associated with conjugation of the O-PS to the lipid A core oligosaccharide, has been implicated in the ligation of the O-PS sugars to the EmaA monomers (32).
Protein glycosylation has been identified in both Gram-positive and -negative bacteria (26). Flagellar proteins (27), fimbriae (pili) (21), nonfimbrial adhesins (32), and secreted proteins (18) have all been observed to be modified with glycans. Glycosylation has been proposed to contribute to proper protein folding to prevent degradation, the acceleration of protein secondary structure formation, and the enhancement of the thermodynamic stability of the protein (3). Disruption of the glycosylation pathways in bacterial pathogens results in a negative impact on bacterial colonization and invasion of host cells (28). In addition to their roles in pathogenicity, glycans are also proposed to play an important role in protein-protein interactions (14).
In this study, the role of glycan modification in EmaA was studied in strains defective in O-PS biosynthesis. The amounts of membrane-associated EmaA proteins were substantially reduced in strains with interrupted O-PS biogenesis. These strains were demonstrated to have little collagen binding activity, which was similar to the emaA mutant strain. Furthermore, the evidence presented also suggests that the glycosylation of EmaA takes place on the periplasmic side of the inner membrane.
MATERIALS AND METHODS
Bacterial strains.The wild-type (WT) strain of A. actinomycetemcomitans, VT1169, is a spontaneously rifampin- and nalidixic acid-resistant serotype b strain that has been described previously (16, 30, 32). The mutant, complemented, and EmaA overexpression strains were all derived from strain VT1169 (Table 1). The rmlC and waal mutants were generated and characterized previously (32). A. actinomycetemcomitans strains were stored at −80°C and grown in 3% Trypticase soy broth and 0.6% yeast extract (TSBYE; Becton Dickinson) in a 37°C incubator with 10% humidified carbon dioxide. All mutant strains were grown in TSBYE medium supplemented with 50 μg/ml spectinomycin. Strains transformed with replicative plasmids (Table 1) were grown in TSBYE medium supplemented with 1 μg/ml chloramphenicol.
Strains and plasmids
Bacitracin treatment.The effect of bacitracin (BAC; Sigma-Aldrich, St. Louis, MO) on the growth of A. actinomycetemcomitans was determined spectrophotometrically based on the optical density at 495 nm (OD495). Concentrations ranging from 100 to 5,000 μg/ml, with 100-μg/ml increases, were added to the TSBYE broth (1 ×107 cells in a volume of 200 μl), and bacterial growth was monitored over a period of 24 hours. The MIC of bacitracin against A. actinomycetemcomitans was determined. The concentrations that did not impact bacterial growth were used to characterize the effects of bacitracin on LPS and EmaA biosynthesis.
Cellular fractionation of A. actinomycetemcomitans.Membrane and cytoplasmic proteins of A. actinomycetemcomitans were prepared as described previously (16). Briefly, 200 ml of late-logarithmic-phase cells (OD495, 0.4 to 0.5) were harvested and lysed using a French press minicell. Unbroken cells were removed by centrifugation at 7,650 × g, and membrane fragments were collected by centrifugation at 100,000 × g. Inner and outer membrane proteins were separated by incubation in 1% sodium N-lauryl sarcosine with agitation at room temperature for 30 min and centrifugation at 15,600 × g for 30 min at 4°C. The sarcosine-soluble inner membrane proteins remained in the supernatant, and the sarcosine-insoluble outer membrane proteins were associated with the pellet. Protein concentrations were determined spectrophotometrically, based on the absorbance at 280 nm (A280).
The bacitracin-treated cells were prepared as follows: bacteria were grown on fresh TSBYE plates without bacitracin for 2 days, and one colony was grown in 5 ml TSBYE with freshly prepared 600 μg/ml bacitracin for 24 h and then transferred into 200 ml fresh TSBYE with 600 μg/ml bacitracin and grown to late logarithmic phase. The bacitracin-treated cells were fractionated as described above. To remove the bacitracin, 5-ml aliquots of 48-hour bacitracin-treated cells were recovered by centrifugation at 4,000 × g for 10 min, washed once with fresh TSBYE broth, resuspended in 200 ml fresh TSBYE, and grown for 18 to 24 h before cellular fractionation.
Immunoblot analysis.Protein samples were diluted in 10 mM HEPES with 2% SDS for protein concentration measurement based on either absorbance at 280 nm (A280) or the results of a bicinchoninic acid (BCA) protein assay (Pierce Thermo Scientific, Rockford, IL). Equivalent amounts of protein from different strains were dissolved in electrophoresis loading buffer containing 10 mM HEPES, 2% SDS, 5% (vol/vol) β-mercaptoethanol, 2% (vol/vol) glycerol, and 0.05% (wt/vol) bromophenol blue at 100°C for 5 min and loaded onto 4-to-15% gradient polyacrylamide Tris-HCl Procast gels (Bio-Rad, Hercules, CA). The separated proteins were transferred to a Westran polyvinylidene difluoride (PVDF) membrane (Whatman Inc., Piscataway, NJ) and probed with anti-ApiA polyclonal antibody (13) or anti-EmaA monoclonal antibody (33). The immune complexes were detected with either horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG for EmaA detection or goat anti-rabbit IgG for ApiA detection (Jackson Laboratory, Bar Harbor, ME). Signal was detected using the SuperSignal West Pico or Femto maximum sensitivity chemiluminescent substrates (Pierce Biotechnology, Rockford, IL) and visualized by exposure to Kodak X-Omat LS film (Carestream Health, Rochester, NY).
The EmaA membrane protein tends to aggregate after solubilization of the membrane, and the monomers do not consistently resolve in the separating gel (36). Therefore, an immunodot blot format was used to quantify the amount of EmaA in the various strains. Total membrane protein of each sample was prepared in 100 μl of 10 mM HEPES with 2% SDS, boiled, and loaded on an Optitran nitrocellulose membrane (Whatman Inc., Piscataway, NJ) by using a dot blot apparatus (Bio-Rad, Hercules, CA). After 1 h of incubation, the membranes were probed with an anti-EmaA monoclonal antibody followed by HRP-conjugated goat anti-mouse antibodies as described previously (33).
Immunoblot assays or immunodot blot assays were performed 4 times using samples prepared from four independent experiments. Densities were determined using ImageJ 1.43u (http://rsb.info.nih.gov/ij/).
RNA isolation and quantitative, real-time RT-PCR analysis.The total RNA was purified from cells grown to early logarithmic phase (OD495, 0.25 to 0.30) as described previously (30). Quantitative real-time reverse transcription-PCR (RT-PCR) was performed as described previously (30). Briefly, the purified total RNA was pretreated with DNase I (Invitrogen, Carlsbad, CA) and reverse transcribed into cDNA, using random hexamers. The quantitative real-time PCR was performed using an ABI Prism 7900HT sequence detection system at the Vermont Cancer Center of the University of Vermont. The primers and probes used in this study are listed in Table 2. The relative expression level of each gene were calculated based on the threshold cycle (CT) and the following formula for the relative expression level (2−ΔΔCT): ΔΔCT = CT(target) − CT(16S rRNA) − CT(calibrator).
Primers and probes for quantitative real-time PCR
The expression level was based on the mRNA levels of emaA, which were detected from three independently isolated RNA samples, and the 16S rRNA gene was used for normalization of the RT-PCR data. The wild-type strain VT1169 was chosen as the calibrator.
Analysis of LPS by GC/MS.LPS samples were purified using the hot phenol-water method, as described earlier (32). The glycosyl composition of the extracted LPS was analyzed using combined gas chromatography/mass spectrometry (GC/MS), as described previously (32), at the Complex Carbohydrate Research Center, University of Georgia.
Analysis of protein gels by LC/MS.Equivalent amounts of cytoplasmic protein were dissolved in electrophoresis loading buffer, boiled for 5 min, and loaded onto a 5-to-15% gradient polyacrylamide–SDS gel, with a 3% stacking gel. Electrophoresis was performed at 60 V, 4°C, for 24 h. The gel was silver stained (Invitrogen, Carlsbad, CA), and the bands of interest were processed for ion trap liquid chromatography/MS (LC/MS) as described earlier (32). The LC/MS was performed at the Vermont Genetics Network proteomics facility located at the University of Vermont.
Transmission electron microscopy.The EmaA structures were visualized using transmission electron microscopy (TEM) of whole-cell mount preparations as described earlier (11, 25). Briefly, bacteria were grown to early logarithmic phase (OD495, 0.2 to 0.3), pelleted at 960 × g, resuspended in phosphate-buffered saline (PBS; 10 mM NaH2PO4, 150 mM NaCl; pH 7.4), applied to grids, and negatively stained using Nano-W (Nanoprobes, Yephank, NY). Data collection was carried out using a Tecnai 12 electron microscope (FEI, Hillsboro, OR) equipped with a LaB6 cathode, a 14-μm, 2,048-pixel charge-coupled-device (CCD) camera (TVIPS, Gauting, Germany) and a dual-axis tilt tomography holder (Fischione, Export, PA), operating at 100 kV. Micrographs were recorded with the CCD camera at a nominal magnification of ×52,000, which corresponded to a 0.25-nm pixel size on the specimen scale.
Three-dimensional Matrigel collagen binding assay.Collagen binding activity was analyzed using an assay described previously (33). Briefly, 50 μg of acid-soluble human type V collagen (Sigma-Aldrich, St. Louis, MO) was mixed with 25 μl of Matrigel (BD Biosciences, San Jose, CA) to a final volume of 50 μl with TSBYE in individual wells of sterile, 96-well tissue culture plates. Following incubation at 37°C for 150 min to solidify the mixture, 200 μl of TSBYE was added and incubated overnight. Approximately 5.0 × 107 early-logarithmic-phase cells in 100 μl were added to each well, incubated at 37°C for 2 h, and removed and washed three times with PBS with agitation. A 200-μl volume of TSBYE was added to each well and incubated overnight at 4°C to solubilize the gel. The bacterial suspension was removed from each well and diluted, and CFU were enumerated on TSBYE agar plates. Five individual experiments were performed in triplicate for each strain examined. Unpaired t tests were performed for comparisons between two strains, and one-way analysis of variance (ANOVA) was performed to compare multiple strains (GraphPad Prism, version 5.04; La Jolla, CA). P values of <0.05 were considered significant.
RESULTS
The biochemical properties of two trimeric autotransporter adhesins of A. actinomycetemcomitans were investigated in rmlC and waaL mutant backgrounds. The TDP-4-keto-6-deoxy-d-glucose 3,5-epimerase (RmlC) and the O-antigen ligase (WaaL) are involved in the biosynthesis of rhamnose and the ligation of O-PS (with its repeating trisaccharide unit composed of rhamnose, fucose, and N-acetylglactosamine) to the lipid A core oligosaccharide of serotype b A. actinomycetemcomitans, respectively (32, 35). These enzymes have also been implicated in the glycosylation of EmaA, as suggested by an increase in the electrophoretic mobility of EmaA in the LPS mutant strains and a positive fucose-specific Lens culinaris agglutinin (LCA) lectin reaction (32). In this study, the A. actinomycetemcomitans trimeric autotransporter multifunctional adhesin ApiA was also investigated. The separated outer membrane proteins were probed with a polyclonal antibody specific for ApiA to determine changes, if any, in the electrophoretic mobility of this protein in the rmlC and waaL mutants. The mobility and the amount of ApiA in the outer membrane of the rmlC mutant and wild-type strains were similar (Fig. 1A and B). In addition, no difference in the electrophoretic mobility of ApiA was observed in the waaL mutant (data not shown), suggesting an absence of posttranslational modification of ApiA by the LPS biosynthetic machinery.
Identification and quantification of ApiA in A. actinomycetemcomitans strains. (A) Outer membrane proteins were obtained by incubation of total membrane fragments with sodium N-lauryl sarcosine. Proteins were separated on SDS-polyacrylamide gels, transferred to a PVDF membrane, and probed with the anti-ApiA antibody. Proteins were detected from the WT, TDP-4-keto-6-deoxy-d-glucose 3,5-epimerase mutant (rmlC), and the rmlC-complemented strain (rmlC/rmlC+). (B) Quantification of ApiA. The integrated signal intensities were quantified using the ImageJ program and analyzed with one-way ANOVA. The level of the proteins in the wild-type strain was arbitrarily set as 1.0, and the quantification was based on four independent experiments.
In contrast, a decrease in the amount of membrane-associated EmaA in the rmlC and waaL mutant strains was observed (Fig. 2A), in addition to the increase in the electrophoretic mobility of the protein compared to the wild-type strain (Fig. 3), as reported previously (32). Quantification of immunodot blot assay results demonstrated that the inactivation of these LPS biosynthetic enzymes resulted in a 50% reduction in the amount of EmaA in the strains investigated (Fig. 2B). The decrease in the amount of EmaA was at the posttranscriptional level, since the quantitative RT-PCR suggested little change in the level of emaA mRNA in either the rmlC or the waaL mutant strains compared with the wild type (P > 0.05) (Fig. 2C). The amount of EmaA in the membrane was restored to the wild-type level in these two mutants following the complementation of the rmlC or the waaL gene in trans (Fig. 2A and B). The data suggest that the decreases in the amount of EmaA observed in the membrane fraction of the rmlC and waaL mutant strains occur posttranscriptionally.
EmaA expression in different A. actinomycetemcomitans strains. (A) Immunodot blot analysis. Equivalent amounts of total membrane protein were loaded on a nitrocellulose membrane using a dot blot apparatus and probed with the anti-EmaA monoclonal antibody. (B) Quantification and comparison of EmaA protein levels. The EmaA protein level in the wild-type strain (with the 67-μg loading amount) was set as 1.0, and the quantification was based on four independent experiments. (C) Quantitative RT-PCR analysis of emaA. The expression level was based on the mRNA levels of emaA, detected from three independently isolated RNA samples. rmlC, the rmlC mutant; waaL, O-antigen ligase mutant. ***, P < 0.001 compared with WT.
Bacitracin treatment and EmaA glycosylation. Equivalent amounts of total membrane protein from each strain were prepared and separated by electrophoresis, transferred to a PVDF membrane, and probed with a monoclonal antibody specific for EmaA. WT, wild-type VT1169; rmlC, the rmlC mutant; WT+Bac, VT1169 grown in TSBYE medium with 600 μg/ml bacitracin for 48 h; WT+/−Bac, the bacitracin was removed after 48 hours and the cells were transferred into fresh TSBYE medium for 24 h.
The effect of the change of the O-PS composition on EmaA, as demonstrated with LPS mutant strains (32), was further investigated by using a pharmacological approach that employed the peptide antibiotic bacitracin. Bacitracin binds to UndPP and inhibits the generation of UndP, a carrier lipid required for the transport of precursor polysaccharides across the bacterial inner membrane for assembly of lipopolysaccharides (22). The MIC of bacitracin for A. actinomycetemcomitans VT1169 was determined to be 3.3 mg/ml. The addition of 600 μg/ml of bacitracin into the TSBYE medium resulted in a change in the carbohydrate composition of the O-PS (Table 3), but it did not alter the growth characteristics of A. actinomycetemcomitans. Therefore, 600 μg/ml of bacitracin was used for all subsequent experiments. The minimum concentration of bacitracin required for a change in the O-PS composition was not determined. All three sugars, rhamnose (Rha), fucose (Fuc), and N-acetylglactosamine (GalNAc), which are associated with the serotype b O-PS of A. actinomycetemcomitans, were substantially reduced (70% to 80%) in the LPS purified from the bacitracin-treated cells compared with the wild-type strain, as determined by gas chromatography/mass spectrometry (Table 3).
Comparison of glycosyl composition and fatty acids of purified LPS samples
In addition to the change in the O-PS composition of the LPS, bacitracin treatment impacted the electrophoretic mobility of EmaA. The membrane-associated EmaA from the bacitracin-treated cells migrated faster than the cells grown in the absence of the antibiotic (Fig. 3). This change was reversed by removal of the compound and regrowth of the treated cells in fresh TSBYE medium without bacitracin (Fig. 3, WT+/−Bac). A difference in the electrophoretic mobility of the membrane-associated EmaA was observed in the bacitracin-treated strain versus the rmlC mutant strain (Fig. 3), indicating a difference in the amount of glycosylation. However, the EmaA molecular mass from either the rmlC mutant or the bacitracin-treated cells was observed to be different from that of the wild-type strain grown in the absence of bacitracin (Fig. 3).
The bacitracin-treated cells also displayed a 40% decrease in the amount of EmaA associated with the membrane (Fig. 4A and B), in addition to the mobility shift. Furthermore, the reduction in membrane-associated EmaA following growth in bacitracin correlated with an observed decrease in EmaA surface structures (Fig. 5). The regrowth of the bacitracin-treated cells in fresh TSBYE medium without the compound restored the level of EmaA in the membrane (Fig. 4, WT versus WT+/−Bac) and the number of surface structures to that observed in the wild-type strain (Fig. 5). The apparent decrease or degradation of EmaA in the bacitracin-treated cells was similar to that observed with the rmlC and waaL genetic mutants.
Quantification of membrane-associated EmaA from bacitracin-treated and untreated cells. (A) Immunodot blot analysis. Equivalent amounts of total membrane protein were loaded on the nitrocellulose membrane using a dot blot apparatus and probed with the anti-EmaA monoclonal antibody. (B) Quantification and comparison of EmaA protein. WT, wild-type VT1169; WT+Bac, VT1169 grown in TSBYE medium with 600 μg/ml bacitracin for 48 h; WT+/−Bac, the bacitracin was removed after 48 h and the cells were transferred into fresh TSBYE medium for an additional 24 h. The EmaA protein level in the wild-type strain (with the 67-μg loading amount) was set as 1.0, and the quantification was based on four independent experiments. ***, P < 0.001 compared with WT.
Transmission electron microscopy visualization of EmaA structures of whole mount preparations. Bacteria were grown to early logarithmic phase and negatively stained with Nano-W. WT, wild type VT1169; rmlC, the rmlC mutant; rmlC/emaA+, the rmlC mutant transformed with plasmid pKM2/emaA; WT+Bac, VT1169 grown in TSBYE medium with bacitracin; WT+/−Bac, the bacitracin was removed after 48 h and the cells were transferred into fresh TSBYE medium and grown for an additional 3 to 4 h; emaA, the emaA mutant. Bar, 100 nm. Arrows indicate EmaA structures.
The apparent decrease in the amount of EmaA on the surface may lead to the loss of collagen binding activity, as observed for the rmlC mutant strain (32). Alternatively, the loss of collagen binding activity in the rmlC mutant, as well as other O-PS mutants (data not shown), may be due to the absence of EmaA glycosylation. To elucidate the mechanism that resulted in the collagen binding defect, the O-PS mutant strains with lower levels of EmaA proteins on the surface were transformed with a replicating plasmid carrying the emaA gene with the endogenous promoter (Table 1, pKM2/emaA) to restore the number of EmaA appendages on the surface. Those strains, including the rmlC mutant transformed with pKM2/emaA and the EmaA overexpression strain treated with bacitracin, were characterized and analyzed for collagen binding activity.
Both the rmlC/emaA+ strain and WT/emaA+ Bac-treated strain were observed to have membrane-associated EmaA in amounts comparable to the wild-type strain in the total membrane fraction (Fig. 6A) and outer membrane fraction (Fig. 6B). However, the partitioning of EmaA in sodium N-lauryl sarcosine, the detergent used to differentiate inner and outer membrane proteins, was different between the glycosylation states. EmaA from both the rmlC mutant and bacitracin-treated strains, with or without the overexpression plasmid, was more soluble in the detergent than the EmaA from the wild type (Fig. 6C), suggesting a difference in the biophysical properties of EmaA proteins due to the presence or absence of glycan conjugants. Transmission electron microscopy demonstrated the presence of EmaA appendages on the cell surface of all strains synthesizing EmaA (Fig. 5). The number of EmaA structures appeared to be reduced in the rmlC mutant and WT bacitracin-treated cells, respectively, compared with the wild-type cells.
Quantification of total membrane-associated and sarcosine-soluble and -insoluble EmaA proteins. (I) Immunodot blot analysis. Equivalent amounts of total membrane protein (A) and the corresponding sarcosine-insoluble fractions (B) and sarcosine-soluble fractions (C) were loaded on the nitrocellulose membrane by using a dot blot apparatus and probed with the anti-EmaA monoclonal antibody. (II) Quantification of EmaA proteins from total membrane, sarcosine-insoluble, and -soluble fractions. The EmaA protein levels in the wild-type strain (with the 67-μg loading amount for the total membrane and 34 μg for the sarcosine-insoluble and -soluble fractions) were set as 1.0, and the quantification was based on four independent experiments. WT, wild-type VT1169; rmlC, the rmlC mutant; rmlC/emaA+, the rmlC mutant transformed with the plasmid pKM2/emaA to overexpress EmaA; emaA, emaA mutant; WT+Bac, VT1169 grown in the presence of bacitracin; WT/emaA++Bac, the emaA overexpression strain (VT1169/emaA+) treated with bacitracin. *, P < 0.05; **, P < 0.01; ***, P < 0.001 compared with WT.
The collagen binding activities of the emaA plasmid-transformed rmlC mutant (rmlC/emaA+) and the bacitracin-treated EmaA overexpression strain were significantly different (P < 0.001) from the wild-type strain (Fig. 7), even though these strains contained equivalent amounts of EmaA in the membrane and on the bacterial surface (Fig. 5 and 6). The collagen binding activities of the rmlC/emaA+ strain and WT/emaA+ Bac-treated strain were similar to the emaA mutant strain (P > 0.05) (Fig. 7).
Analysis of collagen binding activities of different A. actinomycetemcomitans strains. Human type V collagen was mixed with Matrigel to form a 3D matrix. Early-logarithmic-phase cells were added to the wells, and bound bacteria were plated following solubilization of the gel by incubation at 4°C. WT, wild-type VT1169; rmlC, the rmlC mutant; rmlC/emaA+, the rmlC mutant transformed with plasmid pKM2/emaA; WT+Bac, VT1169 grown in TSBYE medium with bacitracin; WT/emaA++Bac, the emaA overexpression strain (VT1169/emaA+) treated with bacitracin; emaA, the emaA mutant. ***, P < 0.001 compared with WT.
The carbohydrate modification of EmaA has been suggested to take place external to the cytoplasm. To address this hypothesis, equivalent concentrations of cytoplasmic proteins from the wild type and rmlC mutant were separated by SDS-PAGE and transferred, and the electrophoretic mobilities of the EmaA monomers from the individual strains were compared. The electrophoretic mobility of the monomeric EmaA from the cytoplasm appeared to be similar between the wild type and the rmlC mutant (Fig. 8A). Interestingly, protein aggregation was not observed with the cytoplasmic form of EmaA, in contrast with that associated with the membrane-associated form (Fig. 8A), suggesting varied sensitivities to the detergent of these two forms of EmaA. The lack of difference in the electrophoretic mobility of EmaA from the cytoplasm, which was confirmed by mass spectrometric analysis, of the rmlC mutant and the wild type was also observed in the silver-stained polyacrylamide gels (Fig. 8B). Taken together, the data suggest that the glycosylation of EmaA, required for both protein stability and biological function, does not occur in the cytoplasm.
Electrophoretic mobility of cytoplasmic (Cyto) and membrane (Mem) forms of EmaA. (A) Equivalent amounts of membrane and cytoplasmic proteins from each strain were prepared and separated by electrophoresis, transferred to a PVDF membrane, and probed with anti-EmaA monoclonal antibody. WT/emaA+ (Mem), membrane proteins of EmaA overexpression strain; rmlC/emaA+ (Mem), membrane proteins of the rmlC mutant strain overexpressing EmaA; WT/emaA+ (Cyto), cytoplasmic proteins of EmaA overexpression strain; rmlC/emaA+ (Cyto), cytoplasmic proteins from the rmlC mutant strain overexpressing EmaA. (B) Silver-stained SDS-polyacrylamide gel of separated cytoplasm samples. Equivalent amounts of cytoplasmic proteins of the WT, the rmlC mutant, and the emaA mutant were separated by SDS-polyacrylamide gels and silver stained. The cytoplasm EmaA monomers of the WT and the rmlC mutant (indicated by the black arrow) were confirmed by mass spectrometric analysis.
DISCUSSION
Two adhesins of A. actinomycetemcomitans have been suggested to be glycosylated: fimbriae, which are composed of the repeating fimbrial subunits Flp1 (∼6.5 kDa) (10, 34) and EmaA (32). The proteins required for the biogenesis of fimbriae are encoded by the 14-gene tad locus, which includes the assembly of a type IV secretion apparatus (12). The glycosylation of Flp1 is suggested to occur on serine and asparagine residues located near the carboxyl terminus of the protein (10) by an undefined mechanism. EmaA is encoded on a stand-alone gene (16). The glycosylation of EmaA utilizes the biosynthetic machinery of the O-polysaccharides of lipopolysaccharides (32). Furthermore, the covalent attachment of the carbohydrate moieties to the EmaA protein backbone requires WaaL (32), which is also the enzyme responsible for ligation of the UndPP-linked O-PS to the lipid A core oligosaccharide in the periplasm (19, 22).
Glycosylation is not associated with all autotransporter proteins in A. actinomycetemcomitans. We observed an increase in the electrophoretic mobility of the EmaA protein isolated from the O-PS mutants, whereas the electrophoretic mobilities of the trimeric autotransporter protein ApiA and the monomeric autotransporter protein Aae (30) were unchanged in the same genetic backgrounds. The data suggest that the ApiA and Aae autotransporter proteins are not modified by O-PS, and glycosylation is a specific characteristic of EmaA in this serotype b strain. The extent of the exclusivity for the modification of the autotransporter EmaA adhesin by the LPS biogenesis machinery in A. actinomycetemcomitans is unknown.
The requisite LPS biosynthetic pathway for EmaA glycosylation was further supported by the effects of bacitracin on O-PS composition and properties of EmaA. The phosphorylation of UndP and dephosphorylation of UndPP are crucial steps for the generation and recycling of the lipid carrier UndP, which is required for the transport of O-PS precursor sugars across the inner membrane in Gram-negative bacteria (22). Not only did the inhibition of the UndPP dephosphorylation by bacitracin alter the O-PS composition of A. actinomycetemcomitans, but also it resulted in an electrophoretic mobility shift, which lends additional support to our hypothesis that the posttranslational modification of EmaA is dependent on the LPS biosynthetic machinery. Furthermore, the alteration of LPS biosynthesis when we used either bacitracin treatment or the generation of LPS genetic mutants changed the biophysical properties of EmaA, as suggested by the change in the sarcosine solubility of the protein in the absence of glycan conjugants. The absence or decrease in the amount of glycosylation may change the hydrophobicity of the protein and alter the solubility properties of the protein.
The amount of surface EmaA molecules was dramatically reduced in both the O-PS mutant and the bacitracin-treated cells, in addition to the changed biophysical properties of EmaA. A posttranscriptional mechanism for the reduction in the amount of membrane-associated EmaA is proposed based on the equivalent levels of emaA mRNA observed in these strains. The decrease in the amount of EmaA protein may be due to either the efficiency of protein translation for this specific mRNA, proteolysis due to incorrect membrane insertion, or aberrant protein glycosylation and subsequent protein degradation. Presently, we do not have the experimental tools to investigate the kinetics of protein translation in A. actinomycetemcomitans. Also, we are not aware of published data which suggest any experimental approaches for targeting specific mRNAs in terms of protein translation. The amounts of ApiA in the bacterial strains were found to be equivalent, as were other membrane proteins, including Aae and the outer membrane protein Omp34 (30), which suggests that the membranes of these strains remain competent for protein translocation and membrane insertion.
Glycosylation, however, is suggested to stabilize the protein from enzymatic degradation (2). A decreased amount of an unglycosylated autotransporter, adhesin (AIDA-1) of Escherichia coli, has been reported compared with the glycosylated form (4). In this study, we have demonstrated a reduction in the relative amounts of membrane-associated EmaA in the mutant and bacitracin-treated cells compared with the wild-type strain. The absence of a change in the amount of ApiA and other membrane proteins (30) implies that alterations in the amount and composition of the O-PS are related specifically to a decrease in membrane-associated EmaA. The change in EmaA concentration does not appear to be a general proteolytic degradation of membrane proteins, due to stress responses after alteration of the membrane composition as observed in other bacteria (6). Therefore, the glycoconjugants associated with the EmaA adhesin appear to protect the protein from degradation in the periplasm and/or on the cell surface.
The number of EmaA structures present on the bacterial surface required for adhesion to collagen is unknown. Therefore, the loss of collagen binding activity associated with the O-PS mutants or bacitracin-treated cells may be attributed to suboptimal concentrations (numbers) of structures associated with the outer membrane. The overexpression of emaA in the O-PS mutant strains, the rmlC mutant, and the bacitracin-treated cells resulted in the expression of the EmaA protein and structures at an approximate wild-type level. However, the elevated amount of EmaA in the membrane in either the rmlC mutant or the bacitracin-treated cells did not significantly restore the collagen binding activity to the wild-type level. Therefore, the reduction of collagen binding in these strains is not attributed to the amount of EmaA protein, indicating a role for the carbohydrate moiety of EmaA in our Matrigel collagen binding assay. The glycans associated with the EmaA protein expressed by wild-type cells appear to contribute to the collagen binding activity. A linkage between protein glycosylation and function has been observed in Gram-negative bacteria, including flagellar glycosylation and motility (15, 27, 29), the glycosylation of adhesin (AIDA-1), and binding to epithelial cells (4). The presence of glycans associated with EmaA may be either necessary for the direct interaction with collagen or facilitate the folding of the proteins into an active conformation upon surface expression.
The interaction of the O-antigen polysaccharide and protein glycosylation biosynthetic pathways has been documented for several bacterial species, including A. actinomycetemcomitans EmaA (reviewed in references 9 and 32). The glycosylation of EmaA appears to be unique in utilizing the O-PS ligase (WaaL) for the covalent modification of EmaA with O-PS sugars (32). The WaaL ligase is an inner membrane enzyme, functioning on the periplasmic side of the inner membrane (19), which suggests that EmaA glycosylation should occur within the periplasmic space. This assumption is supported by the absence of a difference in the electrophoretic mobility of isolated cytoplasmic EmaA from the rmlC mutant or the bacitracin-treated cells (data not shown) compared with the wild-type strain. The absence of a difference in the electrophoretic mobility of cytoplasmic EmaA in these strains is in sharp contrast to the change in the mobility of the membrane-associated EmaA derived from cells of the same bacterial preparation (Fig. 8A). Therefore, the data further support our hypothesis that the glycosylation of EmaA occurs in the periplasm of the bacterium.
A. actinomycetemcomitans appears to have evolved a shared biosynthetic pathway for EmaA glycosylation and LPS biosynthesis. Our data suggest that modification of EmaA is important in maintaining a protein structure that is resistant to proteolysis and/or required for the interaction of the monomers with periplasmic proteins that are involved in secretion of outer membrane proteins (7). Furthermore, these glycan moieties are important for the interaction with collagen in our three-dimensional Matrigel collagen binding assay. EmaA is an important virulence determinant in the initiation of infective endocarditis (31) and may be also important for establishing a reservoir for both oral and extraoral reinfection. Sharing of these biosynthetic pathways between LPS and EmaA may improve the bacterial fitness for survival and dissemination in the host.
ACKNOWLEDGMENTS
We thank Toshihisa Kawai, The Forsyth Institute, Cambridge, MA, for providing the anti-ApiA (Omp100) antibody. We also acknowledge Bin Deng for the mass spectrometric analysis (Vermont Genetics Network).
This research was supported by NIH-NIDCR grant RO1-DE13824 awarded to K.P.M. The Vermont Genetics Network is supported by P20 RR16462 from the INBRE program of the National Center for Research Resources.
FOOTNOTES
- Received 6 April 2012.
- Returned for modification 6 May 2012.
- Accepted 30 May 2012.
- Accepted manuscript posted online 11 June 2012.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
