Oral streptococci utilize a Siglec-like domain of serine-rich repeat adhesins to preferentially target platelet sialoglycans in human blood

Abstract

Damaged cardiac valves attract blood-borne bacteria, and infective endocarditis is often caused by viridans group streptococci. While such bacteria use multiple adhesins to maintain their normal oral commensal state, recognition of platelet sialoglycans provides an intermediary for binding to damaged valvular endocardium. We use a customized sialoglycan microarray to explore the varied binding properties of phylogenetically related serine-rich repeat adhesins, the GspB, Hsa, and SrpA homologs from Streptococcus gordonii and Streptococcus sanguinis species, which belong to a highly conserved family of glycoproteins that contribute to virulence for a broad range of Gram-positive pathogens. Binding profiles of recombinant soluble homologs containing novel sialic acid-recognizing Siglec-like domains correlate well with binding of corresponding whole bacteria to arrays. These bacteria show multiple modes of glycan, protein, or divalent cation-dependent binding to synthetic glycoconjugates and isolated glycoproteins in vitro. However, endogenous asialoglycan-recognizing clearance receptors are known to ensure that only fully sialylated glycans dominate in the endovascular system, wherein we find these particular streptococci become primarily dependent on their Siglec-like adhesins for glycan-mediated recognition events. Remarkably, despite an excess of alternate sialoglycan ligands in cellular and soluble blood components, these adhesins selectively target intact bacteria to sialylated ligands on platelets, within human whole blood. These preferred interactions are inhibited by corresponding recombinant soluble adhesins, which also preferentially recognize platelets. Our data indicate that circulating platelets may act as inadvertent Trojan horse carriers of oral streptococci to the site of damaged endocardium, and provide an explanation why it is that among innumerable microbes that gain occasional access to the bloodstream, certain viridans group streptococci have a selective advantage in colonizing damaged cardiac valves and cause infective endocarditis.

Figure 1. Sialoglycan microarray analysis of binding specificities of adhesin-BRs and corresponding whole bacteria.

(A–C) Binding of the adhesin-BR proteins at 50 nM is presented. (n = 4, SD). (D) Adhesin-BR proteins and Cy3-labeled whole bacteria were assayed using the same sialoglycan microarrays. Heat map was generated using the method as previously described . R1 and R2 represent two different spacers.

Figure 2. Phylogeny of SRR glycoprotein BRs from a representative number of S. sanguinis and S. gordonii strains, and glycan binding to immobilized adhesin-BRs.

(A) ClustalW2 was used for both the sequence alignment and the neighbor-joining tree reconstruction. Corresponding SRR adhesin-BR sequence accession numbers are provided in Table S3. (B) PAA-supported sTa binds to both the Hsa-BR and BRs of the four GspB-orthologues. (C) PAA-SLn only binds to Hsa-BR. (n = 6, SEM).

Figure 3. Bacterial binding to saliva and glycoconjugates, and bacterium-mediated hemagglutination.

(A and B) Binding of DL1 vs. DL1Δhsa against immobilized whole human saliva, salivary ductal secretions, isolated and purified glycoproteins and related glycoconjugates, was tested by a dot blot assay. The binding was studied in the presence of EDTA (A) or divalent cations (B). (C–F) Bacterium-mediated hemagglutination using human RBCs and two-fold serial diluted bacterial suspensions. (C and D) DL1- and DL1Δhsa-mediated hemagglutination using non-sialidase treated (Ø) RBCs. (E and F) DL1- and DL1Δhsa-mediated hemagglutination using sialidase treated (S) RBCs.

Figure 4. Bacterial interactions with separated erythrocytes, platelets, and whole human plasma.

(A) Wild type bacteria only hemagglutinate non-sialidase treated RBCs. Mutant strains do not show any hemagglutination. Bars represent median values. (n = 3). (B) Bacterial adhesion to fixed, immobilized platelets. (n = 6, SEM). (C) Whole plasma samples were immobilized on dot blot in three-fold serial dilutions starting from the original concentration in whole blood. Each strain was tested to interact with either non-sialidase treated (Ø) or sialidase treated whole plasma (S). *P<0.05, ***P<0.001.

Figure 5. Representative flow cytometry profiles showing preferential platelet-bacterial adherence in intact whole human blood.

(A–C) Whole blood (WB) in PBS. (D–F) Two antibodies, CD235a-FITC and CD41a-APC, were added to WB. (G–I) DL1-Cy3 incubated with CD235a-FITC and CD41a-APC: (G) DL1-Cy3 also appears in Q4 in the FSC/SSC plot, the same as platelets. (J–L) DL1-Cy3, CD235a-FITC and CD41a-APC were added to WB: (J) Platelets/bacteria are gated as R1 and RBCs/WBCs as R2. (K) Events from gate R1. (L) Events from gate R2. Bacteria preferentially recognize platelets compared to RBCs. (M–O) DL1Δhsa-Cy3, CD235a-FITC and CD41a-APC were added to WB: (N) Events from gate R3, and DL1Δhsa-Cy3 bound platelets are drastically reduced compared to DL1-Cy3 bound platelets. (O) Events from gate R4. Blood samples from multiple donors were tested and consistent results were obtained.

Figure 6. Divalent cation effect on bacterial binding and preferential recognition of platelets over RBCs by Hsa-BR in whole blood.

(A) DL1 and DL1Δhsa binding to either platelets or RBCs were comparable in EDTA- or heparin-anticoagulated whole blood. Bar values = 100×[number of bacterium-bound platelets (or RBCs)/total number of platelets (or RBCs)]. (n = 3, SEM). (B) Quantification and comparison of Hsa-BR binding to platelets and RBCs in whole blood. Bar values = 100×[number of HsaBR-bound platelets (or RBCs)/total number of platelets (or RBCs)]. (n = 6, SEM). (C) Q1 shows mainly RBCs, and Q4 platelets. Only 2° antibody, anti-GST-APC, alone was added in whole blood. (D–F) 6 to 96 pmol of Hsa-BR was added to whole blood, followed by anti-GST-APC. ****P<0.0001.

Figure 7. Blocking of DL1-platelet binding by Hsa-BR in whole human blood.

(A) Representative FSC/SSC plot showing whole blood with added DL1-Cy3. Platelets and bacteria are gated as R1. (B–E) are gated from R1: (B) Whole blood with DL1-Cy3 and CD41a-APC. Non-DL1-bound platelets appear in Q1 and DL1-bound in Q2, while free DL1-Cy3 in Q3. (C–E) Whole blood with different concentrations of Hsa-BR added first, followed by DL1-Cy3 and CD41a-APC. Increasing amount of Hsa-BR results in decreasing count of DL1-bound platelets (Q2 percentages) and increasing numbers of free bacteria (Q3 percentages). (F) Quantification of Hsa-BR blocking of DL1 binding to platelets in whole blood. The percentage values of DL1-bound platelets were normalized, with the highest single value being 100%. (n = 6, SEM). *P<0.05, ***P<0.001.