Strains used in this study are shown in Table 1 (Kilian and Holmgren, 1981; Kilian et al., 1989; Bishop et al., 2009). The S. mutans strains UA159 and OMZ175 were used as reference in specific assays (Ajdić et al., 2002; Abranches et al., 2011). These strains were routinely grown from frozen stocks in brain heart infusion (BHI) agar (BD Difco, MI, United States) at 37°C in a 10% CO₂ atmosphere. For phenotypic analyses, overnight (18 h) cultures obtained in BHI with adjusted absorbances were diluted into fresh BHI supplemented or not with 10% of human saliva (BHIS) or in chemically defined medium (CDM; Biswas et al., 2007), and incubated at 37°C in 10% CO₂ (90% air) or in aerobiosis under rotational aeration (160 rpm).

Saliva samples were collected from a healthy reference volunteer, according to a protocol previously approved by the Ethics Committee of the Piracicaba Dental School, State University of Campinas (CEP/FOP-UNICAMP; protocol No. 3.365.892). The reference volunteer (30 years of age) was a non-smoker, with good general, and oral health; the absence of active lesions of dental caries or periodontal disease was verified in clinical intra-oral examination performed by a dental professional (VAF). Whole saliva samples were collected at morning, after at least 2 h of fasting. Briefly, salivation was stimulated by parafilm chewing, and then whole saliva was collected in glass tubes maintained on ice bath. Saliva samples were then clarified by centrifugation (40,000 × g; 15 min, 4°C), sterilized by filtration under vacuum through filter with pore sizes of 0.22 μm diameter (Millipore Express™ Plus, Merck KGaA, Germany), and stored at −70°C until use. Serum and blood samples were also collected from the reference healthy volunteer (Alves et al., 2020) by using a standard protocol, as previously approved by the Ethics Committee of the Piracicaba Dental School, State University of Campinas (UNICAMP) (CAAE: 83140418.0.0000.5418).

Growth curves of S. sanguinis strains were determined in BHI. Briefly, 18 h cultures with adjusted A_{550 nm} (0.08) were diluted (1:6) into fresh BHI (Schott Duran, Germany) and incubated at 37°C under the aerobiosis (in a shaker incubator, China) or in 10% CO₂ atmosphere. The absorbances of cultures were then determined at each 1 h over an 8-h period. Three assays were performed for each strain at each atmospheric condition.

Biofilms were formed in 96-well flat bottom polystyrene plates (CralPlast) and quantified as previously described (Camargo et al., 2018), with minor modifications. Briefly, 18 h BHI cultures with adjusted absorbances (A_{550 nm} of 0.3) were diluted (1:10) into 5 ml of fresh medium (BHI or BHIS) supplemented with 1% sucrose. Diluted cultures were transferred to 96-well plates (200 μl/well) and incubated (37°C) under aerobiosis (Thermo Shaker Incubator, China) for 18 h. Next, the culture fluids were removed from each well and the plates were gently washed three times with the distilled water to remove weakly adhered cells. The biofilms were then stained with 1% crystal violet for 30 min at room temperature. Afterward, the stain was eluted in ethanol from biofilms, and A_{575 nm} of the eluates was determined and expressed as indirect measures of biofilm biomass. Planktonic growth (A_{550 nm}) was also assessed in the same bath cultures to control the bacterial growth.

The initial phases of biofilm formation by S. sanguinis strains were investigated by scanning electron microscopy (SEM) as previously described (Moraes et al., 2014), with minor modifications. Briefly, glass slides (10 mm × 10 mm) were treated with 1 ml of sterile saliva (18 h at 4°C) in 24-well microplates (Corning, NY, United States). Slides were then incubated with strain cultures in BHI supplemented with 1% sucrose, during 4 h at 37°C under rotational aeration (at 160 rpm). Afterward, slides were washed three times with 0.9% saline, fixed with 2.5% glutaraldehyde (Sigma-Aldrich, United States) and dehydrated in ethanol. Specimens were then sputter coated with gold and analyzed in a scanning electron microscope (JSM5600 LV; JEOL, Japan). Representative images of the specimens were obtained under magnification of 5,000× and 10,000×. Slides incubated with uninoculated medium were used as negative controls.

Amounts of eDNA were measured by the quantitative PCR (qPCR) as described elsewhere (Camargo et al., 2018). Briefly, culture supernatants were obtained from 1 ml of cultures in BHI medium (aerobiosis at 37°C) at the late log phase of growth (A₅₅₀ of 0.7) by centrifugation (twice at 16,000 × g at 4°C for 10 min), followed by filtration through polyethersulfone membranes (0.22-μm pore diameter; Millipore, United States) for the removal of remaining cells. The qPCR was performed in a StepOne real-time PCR system (Life Technologies, United States) with volumes of 1 μl of culture supernatants mixed with 3.4 μl molecular-grade water, 5 μl of Power SYBR green PCR master mix (Thermo Fischer Scientific, United States), and 0.3 μl of a 10 mM stock solution of each primer for gene encoding 16S rRNA (Forward: 5′-CGTAAACGATGAGTGCTAGGTG-3′; Reverse: 5′-TAGAGCGGTCAGAGGGATGT-3′). The qPCR cycling conditions were 95°C for 10 min followed by 40 cycles of 95°C for 15 s, 58°C for 15 s, and 72°C for 30 s. The DNA concentration was calculated based on average threshold cycle values against a 10-fold dilution series of purified SK36 genomic DNA in the same medium. Sterile culture medium was used as a negative control in qPCRs. Three independent experiments were performed in duplicate.

Bacterial binding to complement proteins [C3b, C1q, serum amyloid protein (SAP), C4b-binding protein (C4BP), and factor H (FH)] were determined as previously described (Alves et al., 2016, 2020) with some modifications. Briefly, approximately 10⁷ CFU from mid-log phase cultures (A_{550 nm} 0.3) were harvested by centrifugation (10,000 × g, 4°C), washed twice with PBS (pH 7.4) and resuspended in 20 μl of 20% serum in PBS. After incubation (37°C, 30 min), cells were washed twice with PBS-Tween 0.05% (PBST) and incubated with specific antibodies diluted in PBST, which included fluorescein isothiocyanate (FITC)-conjugated polyclonal goat anti-human C3 IgG antibody (1:300; on ice for 40 min) (ICN, United States), FITC-conjugated polyclonal goat anti-human C1q (1:300; 37°C for 60 min) (LSBio, United States) (Domenech et al., 2013), FITC-conjugated polyclonal anti-human SAP (1:200; 37°C for 60 min) (LSBio, United States) (Yuste et al., 2007), or FITC-conjugated polyclonal rabbit anti-human C4BP (1:225 in PBST; 25°C for 60 min) (LSBio, United States) (Domenech et al., 2013). Factor H was detected by incubation with goat anti-human FH IgG (1:100, 37°C for 30 min) (Calbiochem) followed by incubation with FITC-conjugated anti-goat IgG (1:1,000; 4°C, for 40 min). After antibody probing, cells were washed twice with PBST and fixed in 3% paraformaldehyde in PBS for flow cytometry analyses using a FACSCalibur flow cytometer (BD Biosciences, United States). A total of 25,000 bacterial cells were gated using forward and side scatter parameters. Results were expressed as the geometric mean fluorescence intensity (MFI) of positive cells for each complement protein and as fluorescent index (FI) determined by multiplying MFI by the percentage of positive cells (Alves et al., 2016). Control samples included cells treated only with PBS instead of serum. In addition, heat-inactivated sera (56°C for 20 min) were applied as negative controls in preliminary experiments with each of the tested strains and antibodies, showing minimal effects on comparative analyses of the strains.

Human PMN were isolated from the fresh heparinized blood samples as previously described, with modifications (Alves et al., 2016), using centrifugation over a double gradient composed of 1,119 and 1,083 density Histopaque (Sigma-Aldrich, United States); red blood cells were removed by hypotonic lysis. Isolated PMN was suspended in RPMI 1640 [GIBCO (Thermo Fisher Scientific, United States), Life Technologies, NY, United States] medium supplemented with inactivated 10% fetal bovine serum. Cell viability (>98%) and purity (>95%) were monitored by the trypan blue exclusion and May-Grunwald Giemsa staining, respectively. Bacteria used in the phagocytosis assays were previously labeled with FITC as described elsewhere (Alves et al., 2016). Briefly, 500 μl of bacterial strains (A_{550 nm} 0.3) were washed twice in PBS, suspended in FITC (Sigma) solution [1.7 mg/ml in carbonate buffer (Na₂CO₃ 0.15 M, 0.9% NaCl; pH 9)], and the suspensions were incubated for 1 h (shaking at room temperature in the dark). Next, cells were harvested and washed three times with PBST, and aliquots were stored overnight in 10% glycerol at −70°C.

To analyze serum-mediated phagocytosis, aliquots containing 10⁷CFU of FITC-labeled bacteria were added to wells of 96-well plates containing 2 × 10⁵ PMNs in 50 μl of RPMI medium supplemented with 20% of human serum to a multiplicity of infection (MOI) of 200 bacteria per PMN. Plates were then incubated for 5 min (37°C, 10% CO₂, gentle shaking), and cells were fixed by addition of 100 μl/well of 3% paraformaldehyde. To assess serum-independent phagocytosis, parallel assays were performed in the absence of human serum. PMNs were then analyzed using FACSCalibur (BD Biosciences, United States), and the frequency of phagocytosis expressed as the number of PMN cells with intracellular bacteria, within a total of 10,000 PMN analyzed (Alves et al., 2016).

The binding of bacteria to plasma and/or ECM proteins was analyzed as described elsewhere (Agarwal et al., 2013) with modifications. Briefly, black polystyrene 96-well microtiter plates were treated (18 h at 4°C) with each human protein [plasma fibronectin, plasma fibrinogen, plasma plasminogen, or type I collagen from human fibroblast (Sigma-Aldrich, United States); 5 μg/ml in PBS pH 7.2]. Afterward, the plates were washed (three times with PBST) and incubated (2 h at room temperature) with blocking solution (50 mM Tris–HCl – pH 8, 150 mM NaCl, 0.1% Tween 20, 3% fish gelatin). Next, 100 μl of suspensions of FITC-labeled bacteria in carbonate buffer (0.15 M NaCl, 0.1 M Na₂CO₃; pH 9.6) (containing 1 × 10⁸ ufc) was added per well, and plates incubated during 1 h at 37°C. Plates were then washed three times with washing buffer (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 0.1% Tween 20) for removal of unbound bacteria, and fluorescence signal quantified with the help of a microplate fluorescence reader (Synergy H1, BioTek, CA, United States). The control samples for each strain included wells treated with each human protein, but not incubated with FITC-labeled bacteria, and wells treated with BSA followed by the incubation with the bacterial suspensions.

Primary HCAEC cells were obtained from Lonza and cultivated in basal medium (EBM-2, Lonza, United States) supplemented with EGM-2MV (Lonza, United States). Invasion assays were performed as described elsewhere (Abranches et al., 2011), with modifications. Briefly, HCAEC cells were seeded (1 × 10⁵ cells/well) in 24-well culture plates (Corning, NY, United States), medium removed, and cells were washed three times with pre-warmed HEPES buffer. Next, HCAEC were added of 1 ml of bacterial suspensions containing 1 × 10⁷ cfu in antibiotic-free EGM-2 medium supplemented with 2% of fetal bovine serum (to a MOI of 100:1), and incubated (5% CO₂, 37°C) during periods of 2, 4, and 6 h. After the incubation periods, culture supernatants were removed and the HCAEC cells were washed three times with pre-warmed HEPES buffer and incubated with EBM-2 medium supplemented with penicillin G (50 μg/ml) and gentamycin (300 μg/ml) during 1 h for killing of extracellular bacteria. Afterward, culture fluids were removed for plating onto BHI (to monitor the presence of extracellular bacteria) and the HCAEC was washed with pre-warmed HEPES buffer as described, and lysed in cold ultrapure type I H₂O at room temperature for 20 min. Serial dilutions of cell lysates were plated in BHI agar for bacterial counting after 48 h of incubation (10% CO₂, 37°C). The counts of cfu in the initial inoculum were used as reference for calculating the percentages of invasion. To analyze the effects of human serum in bacterial invasion, similar assays were performed by incubating the HCAEC cells with bacteria previously treated (37°C, 20 min) with 20% human serum or with PBS (negative control) during 2 h.

Strain comparisons with the reference strain SK36 were performed using a non-parametric Mann–Whitney U-test. The flow cytometry data (MFI, FI, and frequencies of phagocytosis by PMN) were compared using Kruskal–Wallis with post hoc Dunn’s test or Mann–Whitney U-test. The Pearson correlation was used to identify associations between binding to complement proteins and/or host plasma/ECM proteins and HCAEC invasion. Differences were considered significant when a p value of <0.05 was obtained.

Growth characteristics of strains were assessed by comparing growth curves in BHI medium at different oxygen tensions, including static incubation at 10% CO₂ (90% air) (Supplementary Figures 1A,B) and culture aeration (Supplementary Figures 1C,D). Strain SK72 showed the lowest growth in both atmospheric conditions among all strains analyzed (Supplementary Figures 1B,D), whereas SK678 showed increased growth under medium aeration when compared to SK36 and other strains (Supplementary Figures 1C,D).

Streptococcus sanguinis SK36 depends on the presence of saliva and production of eDNA to initiate biofilms (Kreth et al., 2017). Additionally, the biofilm maturation in vitro requires the presence of sucrose, the only substrate for glucan synthesis by the glucosyltransferase GtfP expressed by SK36 (Xu et al., 2007; Moraes et al., 2014). Here, we investigated the biofilm initiation on saliva-coated glass surfaces and biofilm maturation in the presence of sucrose using microtiter plate assays. Consistently with our previous studies (Moraes et al., 2014), SK36 was able to initiate biofilms on glass slides coated with human saliva, forming a homogeneous monolayer with spaced small micro colonies after 4 h of grow (Figure 1), and similar phenotypes were also observed in the majority of strains, except for SK115 and for the blood isolates SK678 and SK1056. We then assessed strain capacities to mature biofilms during 18 h of growth in BHI with 1% sucrose, observing biofilms with robust biomass (Figure 2A) formed by strains defective in biofilm initiation (SK115, SK678, and SK1056) (Figure 1), a trait also observed in SK160 and SK330 (Figure 2A). On the other hand, strains SK49 and SK72 showed reduced capacities to mature the biofilms in the presence of sucrose (Figure 2A). Parallel analysis of planktonic growth of cultures used for biofilm formation confirmed that reduced biofilm maturation was not due to growth defects (data not shown). We thus investigated the influence of saliva components in biofilm maturation by performing the similar assays using BHI w/1% sucrose supplemented with 10% of filter-sterilized human saliva. As shown in Figure 2B, saliva supplementation clearly reduced the biofilm maturation of SK115, SK678, SK1056, and SK330, but increased biofilm maturation by SK72. On the other hand, SK49 retained defects in biofilm formation in saliva supplemented BHI, whereas saliva supplementation did not significantly change the biofilm maturation in SK36, SK160, and SK353 (Figure 2B). Thus, the capacity of S. sanguinis to initiate biofilms on saliva-coated surfaces are not associated with their ability to mature biofilms in the presence of sucrose, a process that may be compromised or enhanced by salivary components in a strain-dependent fashion.

To further investigate if differences in biofilm maturation were associated with bacterial binding to sucrose-derived EPS, we analyzed profiles of bacterial aggregation in BHI and CDM media, with and without 1% sucrose. As shown in Figure 2C, strains with increased capacity to form mature biofilms (SK115, SK160, SK1056, and SK353) showed increased aggregation in BHI or CDM supplemented with sucrose when compared to SK36, SK49, and SK72. The appearance of the bacterial aggregates was, however, variable between strains. For example, strains SK115, SK678, and SK1056 formed silky flocculates in BHI-sucrose and/or CDM-sucrose, which did not resemble the clump aggregates mediated by the sucrose-derived glucan typical of cariogenic S. mutans strains (Alves et al., 2016; Harth-Chu et al., 2019; Figure 2C). In addition, the aggregates formed by SK160 in BHI/CDM supplemented with sucrose were numerous, but of reduced size, making then difficult to notice in the digital images of tubes shown in Figure 2C. None of the strains formed aggregates in media not supplemented with sucrose, either in CDM (Figure 2C) or in BHI (data not shown). Thus, apart from differences in aggregate structure, strains with increased capacity to mature biofilms in the presence of sucrose, also showed significant aggregation mediated by sucrose-derived EPS.

Streptococcus sanguinis SK36 produces H₂O₂ under aerobiosis, which in turn, promotes the production of eDNA by mechanisms not entirely understood (Kreth et al., 2009). Because eDNA is required for biofilm initiation in SK36 (Moraes et al., 2014), we compared kinetics of H₂O₂ and eDNA production in S. sanguinis strains grown in BHI. As shown in Figure 3A, all the strains produced H₂O₂, which increased in concentration until 6–8 h. Hydrogen peroxide achieved similar concentration at 8 h of growth in most strains, except in strain SK678, which showed reduction in the production of H₂O₂ (Figure 3B). Of note, SK678 showed reduced capacity to initiate biofilms (Figure 1). Kinetics of eDNA production during these same culture conditions were also assessed and showed increasing levels of eDNA mostly from 4 to 8 h of growth (increases from 9.91 to 2345.6 ng/ml) (Figure 3C). However, amounts of eDNA in 8 h cultures were more variable between strains, ranging from 28.66 to 2345.6 ng/ml (Figure 3D), and were compatible with profiles of H₂O₂ production, although could not be significantly correlated with amounts of H₂O₂ produced at this same time point (Pearson correlation; r: 0.417, p = 0.264). Strain SK160 produced atypical high amounts of eDNA (more than 2,000 ng/ml), although final concentration of H₂O₂ produced by this strain were similar to that produced by most strains tested (Figure 3B). Cultures of SK678 (a low H₂O₂ producer) yielded reduced mounts of eDNA. However, low eDNA yield was also observed in strains which produce significant amounts of H₂O₂, for example, SK115, which was also defective in biofilm initiation. Therefore, although the production of H₂O₂ is a conserved function of S. sanguinis, few strains differ in amounts of H₂O₂ produced in a fashion compatible with profiles of eDNA production and biofilm initiation phenotypes.

Streptococcus sanguinis strains show reduced binding to C3b (a central effector component of the complement system) when compared to other commensal species of oral streptococci, but a diversity in C3b binding was detected in this species (Alves et al., 2019). In S. mutans, C3b deposition is impaired by surface-bound glucan EPS and it seems to be also influenced by components of BHI medium (Alves et al., 2016). Thus, we firstly investigated the effects of BHI in S. sanguinis binding to C3b, by comparing levels of surface bound C3b after serum treatment of strains previously grown in BHI and in CDM. Significant inter-strain variation was detected in C3b deposition as expressed in MFI or FI (Figures 4A,B). Although inter-strain variation in C3b binding observed in BHI-grown strains was comparable to variations in CDM-grown strains, intensities of C3b deposition were significantly higher in BHI- than in CDM-grown bacteria for several strains (Figures 4A,B), suggesting that BHI components adsorbed to bacterial cells may favor C3b deposition. Increases in C3b deposition in BHI-grown compared to CDM-grown strains were more evident when surface-bound C3b was measured using FI (Figure 4B), because this index reflects the intensities of C3b deposition, as well as the proportion of C3b-positive cells, being more accurate than MFI for inter-strain comparisons. Because, C3b is a major opsonin recognized by PMN, we further investigated if differences in C3b deposition in BHI-grown bacteria treated with serum were associated with the frequencies of phagocytosis of strains by PMN isolated from human peripheral blood in serum-dependent way. As shown in Figure 4C, phagocytosis of all S. sanguinis strains, except for strains SK353 and SK160, did not occur in the absence of human serum, indicating that S. sanguinis phagocytosis is mostly dependent on serum. Importantly, the inter-strain variations in C3b-binding (measured either in MFI or FI) were significantly associated with frequencies of phagocytosis by PMN (Figures 4D,E).

To further access the effects of sucrose-derived EPS in S. sanguinis evasion of C3b deposition and phagocytosis, intensities of C3b binding were compared in a sub-set of strains grown either in CMD or CDM supplemented with 1% sucrose (CDMS). As shown in Supplementary Figures 2A,B, the growth in CDMS significantly reduced the C3b deposition (measured either in MFI and/or FI) in strains SK36, SK330, SK678, and SK1056, when compared to these respective strains grown in CDM. On the other hand, C3b deposition was not affected by the medium supplementation with sucrose in SK72 and SK160. Importantly, changes in the C3b deposition were consistent with changes in serum-mediated phagocytosis accessed in the same culture conditions (Supplementary Figure 2C). As expected, phagocytosis was limited in the absence of serum when strains were either grown in CDM or CDMS (Supplementary Figure 2C). These findings indicate that sucrose-derived EPS not only promotes the biofilm maturation, but also impairs the C3b-mediated opsonophagocytosis by human PMN in most strains, including blood isolates (such as SK678 and SK1056).

To further explore the potential mechanisms underlying inter-strain diversity in S. sanguinis binding to C3b, we investigated the binding of S. sanguinis strains to serum components involved in complement activation [C1q and the pentraxin serum amyloid P component (SAP)], as well as to the fluid-phase complement downregulators [Factor H (FH) and the C4b-binding protein (C4BP)]. As shown in Figure 5A, strains significantly differ in binding to these components. Importantly, strain binding to C3b was positively associated with the binding to the initial recognition proteins of the classical pathway of complement activation, C1q (Figure 5B) and SAP (Figure 5C). In addition, although no significant associations between C3b deposition and binding to C4BP or FH were detected in our correlation analyses (Figures 5D,E), strain SK49 which clearly evades C3b deposition and opsonophagocytosis by PMN, showed increased binding to C4BP compared to SK36 (Figure 5A). Therefore, variation in S. sanguinis strain binding to complement activators and/or fluid-phase regulators are associated with strain susceptibility to C3b deposition.

Bacterial invasion and persistence into endothelial cells are important mechanisms of pathogenicity in cardiovascular and other systemic infections (Kozarov, 2012; Konradt and Hunter, 2018). Thus, we investigated the capacities of S. sanguinis strains to invade primary HCAEC. Initially, we defined the kinetics of invasion of primary HCAEC by S. sanguinis strains in the absence of human serum, using the invasive and non-invasive S. mutans strains (OMZ175 and UA159, respectively) as reference (Abranches et al., 2011; Alves et al., 2020). As expected, OMZ175 showed increased invasion from 2 to 6 h of incubation with HCAEC, achieving the highest invasion percentages, whereas UA159 was not invasive (Figure 6A). Two S. sanguinis strains (SK678 and SK353) showed invasion kinetics similar to that of OMZ157. In addition, strains SK1056, SK330, and SK160 also showed frequencies of HCAEC invasion which were significantly higher than invasion rates of SK36 (Figure 6A). The strain SK115 was not able to invade HCAEC, similar to that observed for non-invasive UA159. We next investigated the influence of human serum in strain invasiveness, by exposing HCAEC cells to S. sanguinis strains previously treated with 20% human serum. As shown in Figure 6B, for most of the strains tested, the pre-treatment with human serum increased the invasiveness in approximately 100- to 600-fold at 2 h of bacterial challenge in relation to serum-independent rates of invasion. Serum treatment promoted, however, moderate effects on SK160 and SK115 invasiveness, implying defects of these strains in endothelial cell invasion or in intracellular persistence (Figure 6B). As verified in the absence serum, strains SK678 and SK353 showed the highest invasiveness in the presence of serum, reaching percentages of invasion similar to that found in OMZ175 (Figure 6B).

Endothelial cells dynamically interact with glycoproteins present in soluble form in plasma and as structural components of sub-endothelial ECM (Davis and Senger, 2005), which in turn, may function as receptors for bacterial binding and invasion. Thus, as a first step to investigate the potential mechanisms involved in S. sanguinis invasiveness into HCAEC, we screened for phenotypes of binding to plasma/ECM glycoproteins reported to modulate bacterial invasiveness, such as fibronectin, fibrinogen, plasminogen, and collagen type I (Lemichez et al., 2010; Abranches et al., 2011; Singh et al., 2012; Alves et al., 2020; Nomura et al., 2020). The S. mutans strains UA159 and OMZ175 were also used as reference for these phenotypes (Avilés-Reyes et al., 2014; Alves et al., 2020). As shown in Figure 7A, SK36 and SK72 showed the highest binding to fibronectin, fibrinogen, and plasminogen among all the S. sanguinis and S. mutans strains. Strains SK36 and SK72 also showed significant binding to type I collagen, but none of the S. sanguinis strains achieved intensities of collagen binding observed for OMZ175, which expresses the collagen-binding protein Cnm (Nomura et al., 2012; Avilés-Reyes et al., 2017). The other S. sanguinis strains showed similar profiles of binding to the tested glycoproteins. No significant associations were found between rates of serum-dependent invasion with each individual plasma/ECM glycoprotein tested (data not shown) or with scores of binding intensities to all the glycoproteins analyzed (sum of relative intensities of binding to the four glycoproteins) (Figure 7B). Because EC, such as HCAEC, expresses the receptors for C3b and C1q (Soares and Bach, 1999; Langeggen et al., 2002; Fischetti and Tedesco, 2006; Oroszlán et al., 2007; Yin et al., 2008), we further investigated the associations between invasiveness into HCAEC with intensities of binding to C3b and C1q. As shown in Figure 7C, serum-dependent invasiveness positively correlated with C3b binding. In addition, strain binding to C1q tended to correlate with strain invasiveness, although correlation did not achieve statistical significance (p = 0.08) (Figure 7D). Finally, scores of FI measures of C3b binding added of scores of bindings to the four glycoproteins (Figure 7E) showed increased correlation with bacterial invasiveness (r = 0.76; p < 0.05; Figure 7E), suggesting that strain-specific interactions with complement and ECM glycoproteins modulates the capacity to invade EC.