The Calcium-induced Conformation and Glycosylation of Scavenger-rich Cysteine Repeat (SRCR) Domains of Glycoprotein 340 Influence the High Affinity Interaction with Antigen I/II Homologs*

Background: AgI/II homolog interaction with GP340 is crucial for bacterial attachment to tooth surface. Results: Tandem SRCR domains efficiently adhere/aggregate bacteria. Calcium-induced conformational switch and O-linked carbohydrates of SRCRs are necessary for the interaction with AgI/II homologs. Conclusion: High affinity interactions are dictated by calcium and carbohydrates. Significance: Oral streptococci adhere to specific calcium-induced conformation of immobilized SRCRs and to its carbohydrates. Oral streptococci adhere to tooth-immobilized glycoprotein 340 (GP340) via the surface protein antigen I/II (AgI/II) and its homologs as the first step in pathogenesis. Studying this interaction using recombinant proteins, we observed that calcium increases the conformational stability of the scavenger-rich cysteine repeat (SRCRs) domains of GP340. Our results also show that AgI/II adheres specifically with nanomolar affinity to the calcium-induced SRCR conformation in an immobilized state and not in solution. This interaction is significantly dependent on the O-linked carbohydrates present on the SRCRs. This study also establishes that a single SRCR domain of GP340 contains the two surfaces to which the apical and C-terminal regions of AgI/II noncompetitively adhere. Compared with the single SRCR domain, the three tandem SRCR domains displayed a collective/cooperative increase in their bacterial adherence and aggregation. The previously described SRCRP2 peptide that was shown to aggregate several oral streptococci displayed limited aggregation and also nonspecific adherence compared to SRCR domains. Finally, we show distinct species-specific adherence/aggregation between Streptococcus mutans AgI/II and Streptococcus gordonii SspB in their interaction with the SRCRs. This study concludes that identification of the metal ion and carbohydrate adherence motifs on both SRCRs and AgI/II homologs could lead to the development of anti-adhesive inhibitors that could deter the adherence of pathogenic oral streptococci and thereby prevent the onset of infections.

SAG is a large glycoprotein complex that contains glycoprotein 340 (GP340), sIgA, and an unknown 80-kDa protein (10,19). Among these, the major component GP340 is known to aggregate several species of bacteria, including mutans, viridans streptococci, and Helicobacter pylori (18,20), and is thereby considered an innate immune response factor. GP340 is a 340-kDa protein that contains 14 SRCR domains, two C1r/C1s Uegf Bmp1 (CUB) domains, and one zona pellucida (ZP) domain (Fig. 2). The 13 SRCR domains are present in tandem at the N terminus, followed by an intriguingly nested 14th SRCR domain between two CUB domains, with a ZP domain at the C terminus. The SRCR domains are interspersed with regions termed SID, an acronym for the SRCR interspersed domains. Except between the 4th and the 5th SRCR domain, all other tandem repeats contain the SID domain (11). These SRCR domains belong to an ancient class of proteins and are present in protozoan parasites like Cryptosporidium, Toxoplasma, Plasmodium, and in the algae Chlamydomonas (21,22). They also appear in the entire animal kingdom, beginning with sponges, and are highly conserved, where a single SRCR domain usually contains 100 -110 amino acids (23). GP340 SRCR domains were recently shown to aid in transcytosis of HIV into vaginal epithelial cells (24,25). This highlights the role of the GP340 SRCR domains in infection, where it serves as a portal of entry into the host for both bacteria and viruses that result in various human diseases (26 -29).
In a systematic study conducted with various oral streptococci, Loimaranta et al. (30) classified bacterial recognition properties of GP340 into three different groups as follows: group I strains were both aggregated by and adhered to GP340; group II preferentially adhered, and group III preferentially aggregated. Using a peptide-based approach, Bikker et al. (17,31) identified a consensus peptide SRCRP2 (QGRVEVLYRG-SWGTVC) derived from the 14 SRCR domains of GP340, which aggregated several species of bacteria, and also inhibited the adherence of AgI/II to SAG (32). In a subsequent study using alanine scanning, the most important residues involved in aggregation were deduced to reside within the "VEVLXXXXW" motif (31). In these studies, the SID domains that are predicted to host the glycosylation sites were classified into two different groups, namely SID20 and SID22 based on sequence homology, and neither one displayed aggregation nor adherence to bacteria (17,31).
In this study, we have reported the adherence characteristics and the multicomponent adherence mechanisms adopted by S. mutans AgI/II and S. gordonii SspB with the recombinantly expressed SRCR domains. We also have presented comparative results on the adherence and aggregation properties of the SRCRP2 peptide and recombinantly expressed SRCRs. The results emanating from this study would foster the development of inhibitors to AgI/II homologs on oral streptococci.
Surface Plasmon Resonance-Real time binding analyses of the SRCR domains with AgI/II fragments were carried out using the BIAcore 2000 system. The CM5 chip was labeled with ligands iSRCR 1 or iSRCR 123 or SAG as described previously (34,35), using the amine coupling kit (GE Healthcare). The control and experimental surfaces were blocked using 1 M ethanolamine. Various concentrations of analytes (0.125 to 2.5 M) of S. mutans AgI/II or S. gordonii SspB fragments (supplemental Table S2) were injected over the prepared chip surfaces, and dissociations were measured for 8 -10 min at a flow rate of 20 l/min of binding buffer (20 mM HEPES, pH 7.4, 150 mM  NaCl, 2.5 mM CaCl 2 ) at 25°C. Self-adhesion of iSRCR 1 or iSRCR 123 (2 M) was also determined in a similar manner as described above. Between experiments, the regeneration of the chip surface was accomplished using solutions as shown in supplemental Table S2. Finally, to determine the effect of calcium, SPR analysis was carried out by dialyzing the analytes and ligands in binding buffer devoid of CaCl 2 .
On-chip enzymatic deglycosylation of the iSRCR 1 and iSRCR 123 was carried out to remove N-and O-linked carbohydrates. Briefly, after immobilization of iSRCR 1 and iSRCR 123 on the CM5 sensor chip, the deglycosylation was carried out by incubating the chip surface with a mixture containing a total reaction volume of 40 l made up of 4 l of 10ϫ G7 reaction buffer, 4 l of 10% Nonidet P-40, 4 l of neuraminidase (Sigma), 18 l of water, and 10 l of O-glycosidase (New England Biolabs) and similarly following the manufacturer's protocols for endoglycosidase H (New England Biolabs). Later, the chip was sealed and incubated overnight at 37°C. Subsequently, the chip was thoroughly washed with binding buffer (20 mM HEPES, 150 mM NaCl, 2.5 mM CaCl 2 ) to remove the deglycosylating enzymes and other remnants. Similar to the experimental procedure described above, in a separate experiment an off-chip deglycosylation of SRCRs was carried out and later checked to verify deglycosylation using MALDI-TOF MS analysis (data not shown). Binding studies with FL AgI/II and FL SspB and their subfragments were then carried out as described above and regenerated as described in supplemental Table S2. All experiments were carried out in triplicate, and the kinetics of the association (K A ) and dissociation (K D ) rate constants were deduced using the 1:1 Langmuir kinetic model on the BIAevaluation software (36).
The utilization of a bivalent adherence model to elucidate the kinetics had inherent difficulties in clearly distinguishing affinities for each region, particularly for FL AgI/II and FL SspB . In addition, the SRCR holding two distinct surfaces compounded the elucidation of individual kinetics, and presently there are no modeling protocols available to determine the individual affinities for such a system; therefore, for simplicity we have utilized a single site 1:1 Langmuir model. The larger 2 values observed for the full-length AgI/II and SspB are directly attributable to the multiple binding sites.
The concentration (C in micromolars) of analyte (FL AgI/II or FL SSpB at 2 M) that adhered to the immobilized ligand (iSRCR 1 and iSRCR 123 ) within the flow cell was calculated using the formula C ϭ (RU/M r ) ϫ (1/V), where RU is resonance unit (1 RU ϭ 1 pg of bound protein); M r is molecular weight of analyte, and V is volume of flow cell (1.2 ϫ 10 Ϫ10 liters).
Competition Adherence Assay-To determine whether AgI/II domains bound to the same site on SRCR domains, competitive binding SPR experiments were conducted in triplicate as described previously (13), where each fragment, 2 M AgI/II (FL, A 3 VP 1 , or C 123 ) or SspB (FL, A 3 VP 1 , or C 123 ), was initially passed over the CM5 chip surface immobilized with either iSRCR 1 , iSRCR 123 , or SAG for 60 s to saturate available binding sites. The response curve of AgI/II or SspB fragment was first recorded, where the maximal RU (RU 1 ) was considered as the base line for the second injection, and thereafter, the competing fragment was injected, and its response was recorded as RU 2 .
The adherence of the second fragment was then calculated (RU 2 Ϫ RU 1 ) for all SPR competing assay as reported earlier (13).
ELISA-The binding between commercially synthesized (Think Peptides, Inc.) SRCRP2 peptide (QGRVEVLYRGS-WGTVCK-(FAM)) with fluorescein amidite (FAM) at the carboxyl end and AgI/II homologs was analyzed. Briefly, AgI/II homologs (10 g/well) in carbonate/bicarbonate buffer, pH 9.6, were coated on a black ELISA plate individually, washed with binding buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, and 2.5 mM CaCl 2 ), and blocked with 3% nonfat dry skim milk. Serial dilution of the SRCRP2 peptide (200 l) ranging from 0.03 to 1 mg/ml in binding buffer were then incubated with all fragments of AgI/II and SspB coated on wells for 3 h at room temperature. The coated wells without fluorescently labeled SRCRP2 peptides were used as controls. Later, the wells were washed with binding buffer, and the data were recorded at an excitation wavelength of 495 nm and emission at 519 nm using Synergy 2multimode microplate reader and the results were then analyzed (Synergy, Inc.).
Adherence/Inhibition Studies-SRCRP2 peptide at different concentrations (5, 10, 50, 100, and 200 M) was incubated with 2 M each of FL, A 3 VP 1 , and C 123 of AgI/II and SspB at room temperature for 3 h, and then their interaction with immobilized iSRCR 1 or iSRCR 123 on the CM5 chip was analyzed. Running buffer containing 20 mM HEPES, pH 7.4, 150 mM NaCl, and 2.5 mM CaCl 2 at 25°C with a flow rate of 20 l/min was used throughout the experiment. The CM5 chip was regenerated with buffer containing 1 M NaCl and 20 mM EDTA, pH 7.4, after each reaction cycle. Direct adherence of 200 M SRCRP2 peptide alone served as the control, and all calculations on the adherence inhibition were assessed using the BIAevaluation software.
Aggregation Assay-Aggregation assays were performed as described earlier (34) with slight modifications. Briefly, S. mutans UA159 and S. gordonii DL1 cells were grown in TSY broth media (30 g/liter of trypticase soy broth and 0.5 g/liter yeast extract, pH 7.2) overnight at 37°C in the presence of 5% CO 2 . The bacteria were centrifuged at 5000 ϫ g and washed with a buffer containing 20 mM HEPES, pH 7.4, 150 mM NaCl and resuspended to an approximate OD 700 of 1. The bacterial suspension (900 l) was mixed with 6 l of 0.1 M CaCl 2 and 100 l of SAG or iSRCR 1 or iSRCR 123 (10 M) or SRCRP2 peptide (400 g/ml). The aggregation of bacteria was then measured by recording OD 700 over 60 min at 5-min intervals, where the buffer alone was used as control. All experiments were carried out at least five times, and the results were analyzed with oneway analysis of variance. Post hoc testing at p Ͻ 0.05 was considered statistically significant, and results were presented as the percentage of cells aggregated.
Confocal Microscopy-S. mutans UA159 and S. gordonii DL1 were grown overnight in TSY media on an eight-well LabTek chamber slide system (Sigma). The cells were fixed with 3% paraformaldehyde, washed with binding buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, and 2.5 mM CaCl 2 ), and thereafter iSRCR 1 or iSRCR 123 (10 M) was added to the cells and incubated for 60 min. The unbound SRCRs were removed by repeated washing using the binding buffer. Subsequently, Alexa Fluor 488-conjugated anti-His tag antibody (EMD Millipore) (1:50 dilution) that can bind to the His tag on SRCRs was added. After 60 min of incubation, the unbound antibody was washed away thoroughly using binding buffer, and the chamber walls were gently removed. Coverslips were then mounted with 15 l of Fluoromount-G with DAPI (Southern Biotech Inc.,) to stain bacterial nuclei and were sealed until ready to be imaged. The experiment without SRCRs served as control. All slides were imaged using a Leica SP1 UV confocal laser scanning microscope and a Zeiss LSM 710 confocal laser scanning microscope at the High Resolution Imaging Facility (University of Alabama at Birmingham).
Glycoprotein Staining and GC-MS Analysis of SRCRs Carbohydrates-The iSRCR 1 and iSRCR 123 proteins were electrophoretically separated on a 12.5% SDS-polyacrylamide gel and stained by glycoprotein staining kit (Pierce), where horseradish peroxidase (HRP) and soybean trypsin inhibitor were used as positive and negative controls, respectively. The glycosyl composition analysis of purified iSRCR 1 and iSRCR 123 was done by the preparation and gas chromatography-mass spectrometry (GC-MS) of trimethylsilyl methyl glycosides as described previously (37).
Circular Dichroism-Spectroscopic studies were carried out on an Olis DSM 100 circular dichroism spectrophotometer with a 0.2-mm path length quartz cell. Control recombinant iSRCR 1 or iSRCR 123 at a concentration of 1 mg/ml in a buffer containing 20 mM HEPES, pH 7.4, 150 mM NaCl at 22°C was scanned between 200 and 260 nm, and the spectra was recorded (10 times). Similarly, the conformational changes of SRCRs on addition of different concentrations of calcium (1, 2.5, 10, and 100 mM) in binding buffer were analyzed. Using CONTIN/LL algorithm implemented in CDPRO (38), the protein secondary structures were analyzed.
Differential Scanning Calorimetry-The thermostability of SRCRs in the presence of calcium ions was analyzed using MicroCal MC-II differential scanning calorimeter (GE Healthcare) as described earlier (39). Briefly, iSRCR 1 or iSRCR 123 at a concentration of 1 mg/ml was mixed and incubated with different concentrations of CaCl 2 ranging from 0 to 100 mM to final volume of 400 l of buffer containing 20 mM HEPES and 150 mM NaCl. Buffer without SRCRs served as control. Data were recorded with calorimetric scanning rates that ranged from 30 to 90°C/h at 30 p.s.i. pressure. The data collected were analyzed for the unfolding temperature (T m ) and the calorimetric (⌬H cal ) and van't Hoff (⌬H v ) unfolding enthalpies using the Origin 7.0383 software package (MicroCal).
Analytical Ultracentrifugation-iSRCR 1 or iSRCR 123 (0.5 mg/ml) in a buffer containing 20 mM Tris, pH 8.0, 150 mM NaCl, and 1 mM EDTA were subjected to sedimentation velocity experiments on a Beckman Optima XL-A as described previously (12). Briefly, the samples were centrifuged to 45,000 rpm with the temperature maintained at 20°C and absorbance at 280 nm across the cell recorded every 5 min. Using Sednterp, buffer density values of 1.0052 g/ml, protein partial specific volumes of 0.720 and 0.714 g/ml, and hydration values of 0.365 and 0.370 g/g for iSRCR 1 and iSRCR 123 , respectively, were calculated (40,41).

RESULTS
Adherence Assays and Quantitation-The adherence affinities between immobilized iSRCRs and the analytes FL, A 3 VP 1 , and C 123 of AgI/II and SspB are summarized in Table 1 and supplemental Fig. S2, A-C. All fragments of AgI/II and SspB interact with nanomolar affinity to immobilized SRCRs. Although FL AgI/II and FL SspB displayed similar affinities with iSRCR 1 and iSRCR 123 , the quantity of protein that adhered to iSRCR 123 was higher (16% for FL AgI/II and 43% for FL SspB ) than that of iSRCR 1 (supplemental Fig. S3).
Competitive Binding Experiments-FL AgI/II was able to inhibit the binding of A 3 VP 1 AgI/II and C 123 AgI/II by 46 and 36% respectively, whereas FL SspB inhibited A 3 VP 1 SspB and C 123 SspB by 54 and 23% with immobilized iSRCR 1 (Fig. 3). In all other cases, A 3 VP 1 or C 123 of AgI/II and SspB did not significantly inhibit the adherence of each other. These results validate that a single SRCR domain contains the two distinct surfaces that specifically bind A 3 VP 1 as well as C 123 fragments of the AgI/II homologs. Similar inhibition was observed with immobilized iSRCR 123 domains where FL AgI/II inhibited the binding of A 3 VP 1 AgI/II and C 123 AgI/II by 44 and 25%. However, FL SspB had limited inhibitory effects with immobilized iSRCR 123 , where A 3 VP 1 and C 123 displayed 68 and 76% inhibition respectively. This shows that AgI/II and SspB are different in their adherence characteristics, although they are highly homologous (57% identity and 70% homology).
The results from the competitive adherence experiments conclusively provide evidence for multiple adherence sites within a single SRCR domain and have thus narrowed down the region of adherence to the SRCR domains of GP340. This concurs with our earlier observation of multiple sites on SAG (12).

Role of Calcium (Calcium-mediated Adherence/Stability)-
In the absence of calcium, there was no adherence between AgI/II homologs and SRCRs (Fig. 4, A-D). In the presence of  calcium, CD spectra had recorded a significant reduction in ␣-helices that was correspondingly compensated with an increase in ␤-sheet content (Table 2), whereas no such changes were observed with AgI/II or SspB (data not shown). Upon calcium addition, the stability (thermal unfolding) of iSRCR 1 increased in a dose-dependent manner (Table 3), and at 100 mM CaCl 2 , the SRCR domain unfolded at an uncharacteristically surprising 90°C. Although the thermal unfolding curves of iSRCR 1 were simple and easy to interpret, iSRCR 123 was more complex to interpret as it involved multiple domains (data not shown).
The homologous structures of SRCR domains from both group A (with six cysteines) and group B (with eight cysteines) have been determined (42,43); however, to date there are no crystal structures of the SRCR domains of GP340. Both iSRCR 1 and iSRCR 123 domains possess similar secondary structures compared with the solved crystal structures PDB2JA4, PDB1BY2, and PDB1P57 (23,42,44), thus indicating a possible adoption of similar SRCR folds.
Effect of Carbohydrates on the Adherence of AgI/II-The presence of glycosylation on iSRCR 1 and iSRCR 123 was initially confirmed using glycoprotein staining (supplemental Fig. S4). Further glycan profile analysis of both iSRCR 1 and iSRCR 123 indicated that they are predominantly O-glycosylated with Gal␤1-3-GalNac and mannose carbohydrates ( SRCRP2 (Bikker Peptide)-Initial ELISAs (Fig. 5A) demonstrated the adherence of the SRCRP2 peptide to AgI/II, SspB, and their subfragments. When incubated at low concentration (5 M) with FL AgI/II and A 3 VP 1 AgI/II , the SRCRP2 peptide improved the adherence by 8 and 13% respectively to iSRCR 1 (data not shown) and 8 and 16% respectively to iSRCR 123 , whereas C 123 AgI/II had no changes in adherence (Fig. 5B). Only FL SspB improved the adherence by 97% with iSRCR 123 at lower concentration (5 M of SRCRP2 peptide), whereas A 3 VP 1 SspB and C 123 SspB did not alter the adherence characteristics to either iSRCR 1 (data not shown) or iSRCR 123 (3 and 5%), respectively (Fig. 5B). Also, the SRCRP2 alone (control) did not show any binding with SRCRs. These results indicated that SRCRP2 peptide does not inhibit the adherence of AgI/II and SspB to iSRCR 1 and iSRCR 123 and that the adherence site was different from that of the aggregation sites present on AgI/II and SspB.
Aggregation Assays-In the presence of iSRCR 123 , 69% of S. mutans and 48% of S. gordonii aggregated, whereas iSRCR 1 aggregated 17% of S. mutans and 12% of S. gordonii (Fig. 7, A  and B). The positive control SAG aggregated S. mutans by 74% and S. gordonii by 72%. Earlier studies with the consensus peptide, SRCRP2 derived from SRCR domains, aggregated a variety of bacteria (17,31); however, in this study compared with iSRCR 123 , the SRCRP2 peptide displayed very limited aggregation with S. mutans (13%) and S. gordonii (11%) similar to that of a single SRCR domain. Even at higher concentrations, the    (Fig. 8, A and B). From the X-Y panel view, it is noticeable that iSRCR 1 adhered poorly compared with iSRCR 123 , underscoring that multiple SRCR domains have better adherence capability compared with that of a single SRCR domain.
Analytical Ultracentrifugation-We sought to answer the question of the spatial organization of the SRCR domains, particularly whether they might be elongated similarly to AgI/II through ultracentrifugation experiments. From their observed frictional ratios (iSRCR 1 ϭ 1.59 and iSRCR 123 ϭ 1.80), resultant prolate ellipsoid ratios (iSRCR 1 ϭ 7.18 and iSRCR 123 ϭ 10.36), and calculated dimensions (iSRCR 1 ϭ 12.60 ϫ 1.75 nm and iSRCR 123 ϭ 22.08 ϫ 2.13 nm), it is evident that both iSRCR 1 and iSRCR 123 will have extended structures (Table 5). However, these are not extended as linear rigid structures but exist in a flexible nonlinear conformation forming curvy tertiary structures.

DISCUSSION
Oral streptococci primarily attach to tooth-immobilized GP340 via AgI/II homologs and subsequently colonize and infect the host (19,45). For the past 3 decades, this interaction has been studied using GP340 extracted from the saliva of either single or multiple donors who have inherent allelic variability (30,46,47). For the first time in this study using recombinantly expressed SRCR domains of GP340 (Drosophila expression system), we established a benchmark and elucidated the intricate components involved in this bacterial adhesion. Nanomolar affinity interactions between the SRCRs and AgI/II homologs (Table 1 and supplemental Fig. S2, A-C) were deduced from SPR data. The adherence kinetics of the C 123 SspB (present near cell the wall) had distinctive sensorgrams, where they did not remain bound to the immobilized SAG or SRCR domains (supplemental Fig. S2C). Overall, these results imply that SAG-binding protein AgI/II of pathogenic S. mutans contains a locking mechanism to remain bound, whereas the C-terminal region of the commensal S. gordonii SspB does not. In isothermal titration calorimetry experiments (supplemental Fig. S5), the affinity between the AgI/II and SRCR in solution was in the micromolar range and indicated that the nature of the interaction is very different in solution compared with an immobilized state.
This study also demonstrates that the adherence surfaces for A 3 VP 1 and C 123 of AgI/II are contained within a single SRCR domain, and it is therefore the minimal adherence region (Fig.  3, A-E).
The concentration of calcium within the oral cavity has been estimated to be between 1.2 and 2.8 mmol/liter, and using sim- FIGURE 5. A, these curves illustrate the direct adherence of SRCRP2 peptide to FL AgI/II A 3 VP 1 AgI/II , and C 123 AgI/II of S. mutans (bold lines) as well as FL SspB , A 3 VP 1 SspB , and C 123 SspB of S. gordonii (dotted lines). Fluorescein-tagged SRCRP2 (FAM) was serially diluted (0.03-1 mg/ml), and its interaction with immobilized AgI/II and SspB and their subfragments was measured at OD 519 . The results illustrated that the SRCRP2 peptide adheres well with AgI/II and SspB. B, adherence/inhibition of FL AgI/II and FL SspB and subfragments (2 M) in the presence of SRCRP2 peptide at various concentrations (0.005 to 0.200 mM) with immobilized iSRCR 123 . In control experiments, SRCRP2 peptide alone does not display any measurable interaction with iSRCR 123 . More importantly SRCRP2 does not inhibit the adherence of AgI/II and SspB (and their subfragments) to the SRCRs. Surprisingly, SRCRP2 increased the adhesiveness of FL SspB confounding us, as it could be interpreted as nonspecific adherence/aggregation. ilar concentrations, we discovered that calcium induces secondary structural changes ( Table 2) and increases the thermal stability of the SRCRs (Table 3). Because the oral cavity is subject to environmental changes, including pH and temperature changes (hot and cold food and beverages), perhaps the observed thermal stability could be a direct consequence of evolution. It would be interesting to see whether the SRCR domains from sea urchin, which has 57% sequence identity with human GP340 SRCR domain, possess these thermal properties that would directly link it to evolution of the SRCR domains within the human oral cavity. In SPR experiments (supplemental Fig. S6), calcium induced a large change in RUs while interacting with the immobilized SRCRs. Such phenomena have been observed in other proteins that endure calcium-induced structural and conformational changes (48). This opens up the possibility that the SRCRs undergo a distinct conformational change in the immobilized state, to which the bacterial AgI/II homologs adhere with nanomolar affinity, whereas in a solution state they adhere with micromolar affinity (supplemental Fig.  S5). Physiologically, this could represent an evolution of streptococci in the oral cavity, where GP340 (being an innate immunity molecule) would aggregate the microbes and clear them into the gut, and to survive, bacteria have developed specific higher affinity to the tooth-immobilized conformation of GP340.
The positive glycostaining (supplemental Fig. S4) and the glycan profile analysis of recombinant SRCRs (Table 4) indicated that they are predominantly O-glycosylated. Deglycosylation of SRCRs with O-glycanase reduced the adherence affinity (Table 1), and therefore, for the first time we have now quantitatively determined that the high affinity interactions observed between AgI/II homologs and SRCRs/SAG is directly attributed to the carbohydrate adherence. The results are plotted as percentage of aggregation measured at OD 700 at 5-min intervals for 1 h. Bacterial cells in buffer alone were used as control. Differences in aggregation detected between groups were analyzed using one-way analysis of variance, where *, p Ͻ 0.05 was considered significant, and error bars represent the standard deviation. The images display S. mutans and S. gordonii cells stained with DAPI (blue) and anti-His tag Alexa Fluor 488 antibody (green). The observed green fluorescence depicts the binding of iSRCR 1 and iSRCR 123 to S. mutans and S. gordonii, whereas the control S. mutans and S. gordonii were counterstained by DAPI alone. Also, it is evident from the images that iSRCR 123 adhered more profoundly than iSRCR 1 . Compared with iSRCR 1 and iSRCR 123 , the SRCRP2 peptide (Fig. 5B) showed limited aggregation with AgI/II homologs. In addition, SRCRP2 did not offer significant inhibition. There are two possibilities, one is that the aggregation site is different from that of the adhesion in AgI/II homologs, or this could be interpreted as nonspecific adherence by the peptide. We lean toward nonspecific adherence as SRCRP2 increased the adhesiveness of only FL SspB . These results now demonstrate that the peptide does not aggregate well nor does it inhibit the SRCR/ GP340-binding motif/site on AgI/II homologs.
Confocal microscopic images (Fig. 8, A and B) and aggregation assays (Fig. 7, A and B) show that iSRCRs bind to S. mutans and S. gordonii cells, where iSRCR 123 attaches extensively compared with iSRCR 1 . Detailed SPR analysis based on protein adherence to chip surface indicated higher amounts of FL AgI/II and FL SspB adhered to immobilized iSRCR 123 than iSRCR 1 (supplemental Fig. S3). Based on these observations, we conclude that longer tandem SRCR domains would more efficiently agglutinate various bacteria. With GP340 being an innate immunity molecule, the number of tandem repeats it takes to efficiently agglutinate bacteria could have been evolutionarily determined, and it is interesting to note that in humans GP340 contains 14 SRCR domains, of which 13 are tandem repeats, whereas in other vertebrates the number of tandem repeats are comparatively lower (26,27).
One surprising result that came out of these studies is that the SRCR domains self-associate with nanomolar affinities, thus indicating that this association is highly specific, as nonspecific interactions traditionally appear to fall within the micromolar range (Fig. 6, A-D). GP340 is known to exist as a higher order complex, and these aggregates could be as large as 5000 kDa (19,49). Thus far, the aggregation property of GP340 has been attributed to the ZP domain; as in other mammalian proteins, the ZP domain was shown to be involved in self-aggregation (50). Furthermore, the tertiary architecture analysis of tandem SRCR domains indicate that they may not strictly form a linear elongated structure (Table 5) but could form a curvy centipede-like extended structure, similar to that observed in electron microscopy images of GP340 (28). Combined together, these results open up several possible models for bacterial aggregation/adherence, wherein one potential model could simulate the bacterial proteins to be sandwiched between two SRCR domains (GP340s) (Fig. 9).
Earlier studies have shown that the SRCR domains of GP340 play a pivotal role in mediating HIV adhesion/clearance through GP120 (16,51). Although GP340 acted as a clearance mechanism in the oral cavity, the case was very different on the vaginally derived GP340, which is immobilized on the cell surface, where this was shown to mediate transcytosis from apical to basolateral surface in both genital tract epithelial cells in culture and with endocervical tissue (52). Similarly, in SPR experiments, immobilized SRCRs adhere tightly to AgI/II homologs, and in the fluid phase SRCRs aggregate S. mutans and S. gordonii, a double-faceted property, where on the one hand it acts as a portal of entry for microbes while immobilized and on the other hand as a clearance mechanism within the oral cavity in fluid state. This property indicates that SRCRs could possibly adopt different secondary structural conformations in fluid and immobilized states, and this conformation could be induced by calcium ions.
Summarizing our findings, we report that the minimal adherence region is restricted to a single SRCR domain, which carries the two distinct surfaces that adhere to A 3 VP 1 as well as C 123 of both AgI/II and SspB. Better adherence and aggregation of bacteria are observed with increasing numbers of SRCR domains. These SRCR domains attain stability in the presence of calcium, and calcium mediates structural changes that are essential for the adherence of AgI/II homologs. Furthermore, glycosylations play a significant role in the adherence to AgI/II and SspB. Interestingly, the SRCRs self-associate and the tandem domains may adopt a curvy centipede-like structure. Although there are similarities in the binding of AgI/II and SspB, there are certainly distinct differences pointing toward species specificity in their adherence.
Overall, these results now point to the fact that focusing on the SRCRs and elucidating the molecular motifs involved in adherence would aid in the development of interventional therapeutics. Such studies could potentially result in the identification and development of small molecule inhibitors or passive immunization therapies that could impede oral streptococcal adherence to tooth surfaces and alleviate the global burden of dental caries.