Inhibition of Fibrinolysis by Streptococcal Phage LysinSM1

ABSTRACT Expression of bacteriophage lysinSM1 by Streptococcus oralis strain SF100 is thought to be important for the pathogenesis of infective endocarditis, due to its ability to mediate bacterial binding to fibrinogen. To better define the lysinSM1 binding site on fibrinogen Aα, and to investigate the impact of binding on fibrinolysis, we examined the interaction of lysinSM1 with a series of recombinant fibrinogen Aα variants. These studies revealed that lysinSM1 binds the C-terminal region of fibrinogen Aα spanned by amino acid residues 534 to 610, with an affinity of equilibrium dissociation constant (KD) of 3.23 × 10−5 M. This binding site overlaps the known binding site for plasminogen, an inactive precursor of plasmin, which is a key protease responsible for degrading fibrin polymers. When tested in vitro, lysinSM1 competitively inhibited plasminogen binding to the αC region of fibrinogen Aα. It also inhibited plasminogen-mediated fibrinolysis, as measured by thromboelastography (TEG). These results indicate that lysinSM1 is a bi-functional virulence factor for streptococci, serving as both an adhesin and a plasminogen inhibitor. Thus, lysinSM1 may facilitate the attachment of bacteria to fibrinogen on the surface of damaged cardiac valves and may also inhibit plasminogen-mediated lysis of infected thrombi (vegetations) on valve surfaces.

genome-wide association studies (12), and for this reason, clinical isolates are sometimes identified as "Streptococcus mitis/oralis" (13)(14)(15)(16)(17). These organisms are a leading cause of IE, with mortality ranging from 6% to 30% (18,19). Despite the increasing importance of endocarditis due to S. mitis/oralis, especially in view of the high prevalence of multidrug resistance among these strains, relatively little is known about the virulence determinants of S. mitis/oralis. Our previous studies have identified several surface adhesins of S. oralis strain SF100 (formerly identified as S. mitis), such as PblA, PblB, and lysin SM1 , that mediate binding to human platelets and enhance virulence in animal models of IE (20)(21)(22)(23). Lysin SM1 is encoded by a lysogenic bacteriophage (SM1) and has been shown to have at least two pathogenetic functions. First, lysin is essential for the export of the phage-encoded adhesins, PblA and PblB. In addition, extracellular lysin can bind phosphocholine residues on the bacterial cell wall, where it can mediate bacterial binding to fibrinogen (21). Deletion of the lysin SM1 gene in SF100 resulted in significantly lower binding of the organism to fibrinogen and platelets in vitro and delayed the onset of platelet aggregation by this strain (20).
Fibrinogen is a 340-kDa glycoprotein comprising three pairs of distinct polypeptide chains (Aa, Bb, and g; Fig. 1) that are linked by 29 disulfide bridges (24,25). It can be polymerized by the hydrolytic catalysis of its terminal ends by thrombin, resulting in fibrin clots or thrombi. Fibrin polymers can be degraded by a proteolytic process known as fibrinolysis, which is tightly controlled by a series of cofactors, inhibitors, and receptors (26)(27)(28). The high-affinity binding of plasminogen, a serine protease, to the distal portion of each aC region of fibrinogen Aa chains is the first step of fibrinolysis. Bound plasminogen is then activated to plasmin by cleavage at AA561 by tissue-type plasminogen activator (t-PA), thereby triggering fibrinolysis (29)(30)(31).
With a view toward better understanding how lysin SM1 interacts with fibrinogen, we identified the specific binding site for lysin SM1 within the fibrinogen Aa chain and investigated the effect of this interaction on clotting and fibrinolysis. Our studies indicate that a specific interaction of the binding domain in fibrinogen Aa chain overlaps a region bound by plasminogen. Moreover, binding of this region by lysin SM1 inhibits plasmin-mediated fibrinolysis.

RESULTS
Lysin SM1 binding to the aC region of fibrinogen Aa. We previously showed that recombinant lysin SM1 encoded by bacteriophage SM1 binds to the Aa chain of human fibrinogen and that this interaction enhances the attachment of S. oralis SF100 to human platelets (21). The domain of lysin that bound to the Aa chain was contained within the region spanned by amino acid residues 102 to 198 (97 amino acids [AA]) (20). To identify the regions within fibrinogen Aa that bound lysin SM1 , we expressed and purified recombinant forms of the whole Aa (610 aa; variant 1), N-terminal region (AA1-182; variant 2) and C-terminal region (aC region; AA183-610; variant 3) and examined their binding by lysin 102-198 by far-Western blotting ( Fig. 2A and B). Lysin 102-198 (10 mg) bound to variant 1 and 3, but not variant 2, indicating that the lysin SM1 binding domain was located on the aC region of Aa chain. To identify the specific binding regions within this domain, several soluble truncated forms of the region fused to MalE were isolated and tested for lysin 102-198 binding (Fig. 2C). Recombinant lysin  bound to the variants containing the region spanned by amino acid residues 534 to 610 (variant 8), the C-terminal end of fibrinogen Aa chain.
To better define the plasminogen binding sites on fibrinogen, we examined the binding of recombinant human plasminogen with the above-described fibrinogen Aa chain subdomains, as measured by far-Western blotting. As expected, plasminogen (10 mg) bound the aC region (variants 1 and 3; Fig. 2B). Plasminogen also interacted with an ;60-kDa protein (Fig. 2B, lane 1), which was identified as a fragment of variant 1 by liquid chromatography-tandem mass spectroscopy (LC-MS/MS) (data not shown). In addition, plasminogen bound the variants containing the region spanned by amino acid residues 534 to 610 (variant 8) and 248 to 376 (variant 5; aC-RU) (Fig. 2C), indicating that it has two separate binding sites on the Aa chain (AA183-376 and AA534-610).
We next used overlapping 19-amino acid peptides fused with maltose binding protein (MBP) to localize the lysin binding segment between residues 534 and 610 of Fg  (Fig. S1). Lysin SM1 was shown to bind variants 8, 9, 10, and 12, but not variant 11 or 13, indicating that the binding peptide of the fibrinogen Aa chain is localized to AA572-590 (AGSEADHEGTHSTKRGHAK). To further demonstrate that lysin SM1 binding is specific, we made targeted point mutations within the lysin SM1 binding region (AA102-198) and assessed binding to the Fg Aa chain (Fig. S2). Of the four substitutions tested individually, both H111A and D188A were markedly reduced in binding to MalE:Aa 534-610 . These findings indicate that lysin SM1 binding to Fg Aa is specific and that these are key residues for this interaction.
To directly compare lysin SM1 and plasminogen binding to the aC region, equal amounts (0.1 mM) of variants 4 (Aa 183-248 ), 5 (Aa 248-376 ), and 8 (Aa 534-610 ) were immobilized in 96-well plates, and binding of recombinant lysin 102-198 and purified plasminogen was measured by enzyme-linked immunosorbent assay (ELISA). As expected, lysin SM1 bound to variant 8, with binding reaching a plateau at 75 mM and with an apparent K D of 4.8 Â 10 25 M (Fig. 3A). Lysin  showed no binding activity with variants 4 and 5. In contrast, plasminogen bound to variant 5 (K D = 8.5 Â 10 25 M), and variant 8 (K D = 1.8 Â 10 25 M; Fig. 3A). To validate these specific interactions, we also examined whether purified fibrinogen, variant 8, or variant 5 could inhibit lysin SM1 binding to immobilized fibrinogen. When lysin SM1 (1 mg) was coincubated with 0 to 100 mM of these proteins (Fig. 3B), subsequent binding to fibrinogen by lysin SM1 was effectively blocked by purified fibrinogen and variant 8, but not variant 5 (Fig. 3B). In addition, we found that plasminogen (0 to 50 mM) inhibited lysin SM1 (1 mg) binding to immobilized fibrinogen (Fig. S3). These data indicate that the fibrinogen binding sites for both lysin SM1 and plasminogen are colocalized within the same domain in the aC region (Aa 534-610 ) of the fibrinogen Aa chain. We next assessed the impact of lysin SM1 expression on the binding of streptococci to fibrinogen. Wild-type (WT) SF100 and its Dlysin isogenic mutant (PS1006) were compared for binding to immobilized human fibrinogen and recombinant Aa truncates. As shown in Fig. 3C, the WT bound to only fibrinogen and variant 8, but not variant 5. Compared with the WT strain, PS1006 had significantly reduced binding to both fibrinogen (P = 0.01) and variant 8 (P , 0.001). These findings strongly suggest that lysin SM1 on the surface of S. oralis mediates binding to the fibrinogen aC region and that the binding domain is located within residues 534 to 610.
Quantitative assessment of lysin SM1 binding to the aC region by surface plasmon resonance. We analyzed by surface plasmon resonance (SPR) the binding affinity of lysin SM1, lysin  , and plasminogen to purified human fibrinogen, by measuring the dissociation constant (K D ), a specific type of equilibrium constant that measures the propensity of dissociation between two components (Fig. 4A). Increasing concentrations of lysin SM1 (0 to 150 nM), lysin 102-198 (0 to 150 nM), and plasminogen (0 to 250 nM) were flowed over immobilized fibrinogen, and the K D was calculated for each protein. The K D values of lysin SM1 , lysin  , and plasminogen to immobilized fibrinogen were determined to be 3.15 Â 10 25 , 2.32 Â 10 25 , and 1.04 Â 10 25 M, respectively. We next analyzed the binding affinities of lysin SM1 , lysin  , and plasminogen to recombinant forms of variant 8 (Aa 534-610 ) (Fig. 4B). Lysin SM1 , lysin  , and plasminogen showed high levels of binding to this peptide, with affinities of 3.23 Â 10 25 , 1.73 Â 10 25 , and 1.72 Â 10 25 M, respectively. These values are within the range reported for other bacterial fibrinogen binding proteins, such as Srr1 of Streptococcus agalactiae and SdrG of Staphylococcus aureus (32,33).
Inhibition of plasminogen binding to fibrinogen by lysin SM1 . Since lysin SM1 and plasminogen bound the aC region of fibrinogen Aa with similar affinities, we next determined whether lysin SM1 , lysin 102-198 , or lysin 1-102 could competitively inhibit plasminogen binding to immobilized fibrinogen, as measured by ELISA. Immobilized fibrinogen was preincubated with 0 to 100 mM the lysin SM1 variants, followed by incubation with 100 nM plasminogen. As shown in Fig. 5A, minimal inhibition was detected for Inhibition of Fibrinolysis by Phage Lysin ® lysin 1-102 , but more than 75% inhibition of plasminogen binding to immobilized fibrinogen was seen with either lysin SM1 or lysin 102-198, which were significant compared with untreated fibrinogen (P , 0.05).
To confirm the above-described findings, we also examined by SPR the impact of lysin on plasminogen binding to fibrinogen (Fig. 5B). Lysin SM1 (1 mM) was streamed over immobilized fibrinogen followed by the addition of plasminogen (100 nM). Similar to what was seen by ELISA, the affinity (K D ) of plasminogen binding to fibrinogen was 1.92 Â 10 25 M. This was reduced to 4.84 Â 10 22 M in the presence of lysin SM1 . We then released the bound lysin from the immobilized fibrinogen by washing the sensor chip surface with a low-pH glycine buffer (pH 2.0). The K D value of plasminogen binding to immobilized fibrinogen on the chip surface was restored to 1.68 Â 10 25 M, indicating that lysin SM1 competitively inhibited plasminogen binding to fibrinogen.
We next examined whether native lysin SM1 produced by strain SF100 had similar effects on plasminogen binding. As expected (21), lysin SM1 was found in cell wall extracts and in the culture supernatants of SF100, but not for PS1006 (Fig. 5C). We detected about 0.45 6 0.036 mg/ml of lysin SM1 in the culture supernatant of SF100, as measured by ELISA. To assess the impact of lysin SM1 on plasminogen binding to immobilized fibrinogen, we pretreated fibrinogen-coated wells with 0 to 100 ml of supernatants collected from WT or PS1006 cultures, followed by incubation with plasminogen. As was seen with recombinant lysin SM1 , the supernatant from WT SF100 significantly inhibited plasminogen binding (P , 0.006 for volumes above 12.5 ml), but supernatants from PS1006 had no effect (Fig. 5D).
Inhibition of fibrinolysis by blocking plasminogen binding to the aC region by lysin SM1 . Fibrinolysis requires the binding of plasminogen to the C-terminal region of fibrinogen or fibrin, followed by its cleavage by tissue plasminogen activator (tPA), thereby generating the active protease plasmin (34,35). To assess the impact of lysin SM1 on fibrinolysis, we examined the impact of lysin on clot formation and dissolution in vitro, using thromboelastography (TEG). Fibrinogen was preincubated with 13.7 mM albumin (as a control), followed by adding thrombin (to activate fibrin formation and polymerization) and plasmin (to initiate fibrinolysis). Clotting was detectable within 2 min, as indicated by an increase in the elastic modulus, and peaked at 10 min. This was followed by a decline in the modulus, indicating ongoing fibrinolysis, which reached lower than zero shear modulus strength (kdyne/cm 2 ) after 24 min (Fig. 6A). Fibrinolysis was completely blocked by addition of epsilon-aminocaproic acid (EACA; 130 mg/ml), a standard lysine analogue used to competitively inhibit plasmin-induced fibrinolysis (36). When fibrinogen was preincubated with lysin SM1 , clotting reached significantly higher levels at 10 min and peaked at 15 min. These high levels of clotting and resistance to proteolysis were sustained even at 30 min postexposure to thrombin and plasmin (P , 0.001).
The above-described studies demonstrated that lysin SM1 could inhibit fibrinolysis. To determine whether this was due to the competitive inhibition of plasmin binding to the aC region, fibrinogen was preincubated with lysin 102-198 or lysin  . When tested by TEG, preincubation with lysin 1-101 had a minimal effect on plasmin-induced fibrinolysis. In contrast, lysin 102-198 reduced fibrinolysis (P , 0.001) to levels that were comparable to those seen with lysin SM1 , indicating that the inhibition of fibrinolysis by lysin SM1 is due to its blocking of plasmin binding.
In vivo, fibrinolysis requires the conversion of plasminogen to plasmin by tPA. Inhibition of Fibrinolysis by Phage Lysin ® However, tPA can only activate plasminogen that is bound to fibrinogen. We therefore examined whether lysin SM1 binding to fibrinogen could inhibit fibrinolysis induced by tPA. Fibrinogen was mixed with lysin 102-198 or 13.7 mM albumin (as a control) and incubated for 4 min with thrombin, followed by the addition of plasminogen and tPA. As expected, tPA induced extensive clot lysis when mixed with plasminogen alone (Fig. 6B). However, tPA failed to induce lysis in the presence of plasminogen and lysin  (P , 0.001). Since tPA can only activate plasminogen bound to fibrinogen, these data indicate that the blocking of tPA-mediated fibrinolysis by lysin SM1 is due to the inhibition of plasminogen binding to the fibrinogen aC region, such that tPA can no longer generate plasmin and clot lysis.

DISCUSSION
Lysin SM1 is a key adhesin of S. oralis SF100, mediating bacterial binding to platelets in vitro through its interaction with fibrinogen on the platelet surface. However, it was unknown which regions of fibrinogen were bound by lysin SM1 , in part because this adhesin has no structural homology to other known fibrinogen binding proteins, as measured by amino acid sequence alignment (T-Coffee) and protein structure homology-modeling (SWISS-MODEL) (37,38). Our studies indicate that lysin SM1 binds residues 534 to 610 of the fibrinogen Aa chain. This differs from other known fibrinogen binding proteins of other bacteria, such as staphylococcal ClfB and Srr1 and Srr2 of Streptococcus agalactiae (both bind AA283-410 of the Aa chain), staphylococcal SdrG (AA1-25 of the b chain), and staphylococcal ClfA, FnBPA, and FnBPB (AA6-20 of the g chain) (20,21,33,(39)(40)(41). Staphylococcal bone sialoprotein-binding protein (Bbp) binds the same region (AA561-575) of the Aa chain as lysin SM1 (42). However, lysin SM1 differs from at least some of these proteins in its impact on clotting. In particular, SdrG of Staphylococcus epidermidis inhibits coagulation by binding to the thrombin cleavage sites on fibrinogen (33), and Bbp of Staphylococcus aureus has anticoagulant action through an unknown mechanism via binding to AA561-575 (40). In contrast, lysin SM1 has no direct effect on clot formation, at least as measured by TEG, but does have strong anti-fibrinolysis effects, by inhibiting the binding of plasminogen to the aC region of fibrinogen Aa chain.
Pathogenic bacteria can produce and secrete activators or inhibitors of fibrinolysis that may impact their survival and dissemination. At least two distinct mechanisms involving plasminogen have been observed. First, plasminogen binding proteins of bacteria, such as streptokinase from group A, C, and G streptococci, staphylokinase of S. aureus, and Pla of Yersinia pestis (39)(40)(41)(42)(43)(44), can bind free circulating plasminogen and convert it to plasmin (43)(44)(45)(46)(47)(48). Second, proteins on the surface of bacteria, such as GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and enolase of streptococci, plasminogen-binding protein (PAM) of Streptococcus pyogenes, and OspA/C of Borrelia burgdorferi, bind plasminogen, which is then converted to plasmin by tPA (49)(50)(51)(52). Here, we report a novel mechanism for inhibiting fibrinolysis, in which lysin SM1 competitively inhibits plasminogen binding to the fibrinogen Aa 534-610 region, such that it can no longer activate fibrinolysis. We also found that plasminogen bound a second region (AA248-376) within the aC region of fibrinogen. However, the binding affinity for this region was about five times lower than at the primary binding site (AA534-610). This lower affinity would explain our finding that inhibition of fibrinolysis by lysin SM1 did not appear to be affected by plasminogen binding to the second binding site.
Our previous studies using an animal model of infective endocarditis demonstrated that loss of lysin SM1 expression by SF100 was associated with decreased virulence, as measured by reduced levels of bacteria (CFU/g of tissue) within vegetations on cardiac valves, as well as in kidneys and spleens (21). Part of this reduced virulence is likely due to the loss of fibrinogen-mediated binding to platelets, which is a key step for the initial attachment of bacteria to damaged valve surfaces, as well as for the subsequent formation of infected vegetations. These structures are composed of bacteria embedded in a biofilm containing platelets, fibrinogen, and fibrin (53). Vegetation formation is thought to protect bacteria from phagocytosis and render these organisms less susceptible to antimicrobials. Moreover, larger vegetations are associated with increased embolization to target organs, such as the kidneys, spleen, and brain (54)(55)(56). Fibrinolysis may serve to mitigate vegetation formation, as both in vitro and in vivo studies have shown that tPA may reduce vegetation size and facilitate antimicrobial therapy (57)(58)(59)(60). Thus, an additional mechanism by which lysin SM1 may enhance virulence is by blocking plasminogen binding to fibrinogen/fibrin within vegetations, thereby inhibiting tPA activation and vegetation lysis.
Although these studies examined a single strain of S. oralis, our findings are likely to be applicable to a broad range of organisms. Metagenomic studies by Willner et al. indicate that bacteriophage SM1 is highly prevalent in the oral microbiome and that it is the most common bacteriophage of Gram-positive organisms in the oral cavity (61). Moreover, the lysin SM1 gene was among the open reading frames (ORFs) of SM1 most frequently detected. This group also examined the published salivary metagenomes from nine individuals, and all were found to contain pblA and pblB, two phage morphogenesis genes adjacent to lysin on the SM1 genome. More recently, metagenomic studies of salivary specimens from children detected bacteriophage SM1 in 29 of 30 individuals (62). In addition, lysin is among the most commonly expressed genes of streptococcal bacteriophages within the oral microbiome. These findings strongly indicate that SM1 or similar bacteriophages encoding a lysin SM1 homolog are highly prevalent in the oral microbiome (63). Our own searches for homologs of lysin SM1 indicate that numerous strains of not only S. oralis, but also and S. mitis and Streptococcus pneumoniae (data not shown) encode such homologs, indicating that lysin SM1 may be widely prevalent in a variety of streptococcal species.
In summary, we propose that expression of streptococcal phage lysin SM1 impacts the pathogenesis of infective endocarditis in the following manner: (i) damage of the endocardium by inflammation or hemodynamic trauma induces the deposition of platelets, fibrinogen, and fibrin polymerization onto the valve surface (Fig. 7A); (ii) attachment of streptococci encoding lysin SM1 , such as S. oralis SF100, to the altered surface. This initiates endocardial infection and attachment of free or cell wall lysin SM1 to the aC region of fibrinogen, thereby blocking the binding of circulating plasminogen and tPA (Fig. 7B); (iii) the further deposition and polymerization of fibrinogen onto the infected endocardium along with the proliferation of bacteria on the valve surface, resulting in extensive vegetation formation (Fig. 7C).

MATERIALS AND METHODS
Strains and growth conditions. The bacteria and plasmids used in this study are listed in Table S1. Strain SF100 is an endocarditis-associated clinical isolate (64). Originally identified as S. mitis by conventional clinical laboratory methods, we have recently sequenced the complete genome of this strain. BLAST analysis (v2.7.1) was carried out to identify to which species it shows similarity. The average nucleotide identity (ANI) values used to compare the genome of SF100 with S. mitis and S. oralis were determined using the OrthoANIu algorithm (https://www.ezbiocloud.net/tools/ani) (65). In addition, the whole-genome sequences were aligned with 120 bacterial marker genes, using GTDB-Tk (v1.3.0) (66), and the best-fit model was selected using ModelTest-NG (v0.1.6) (67). A phylogenetic tree was constructed for the PROTGAMMALGF model using RAxML (v8.2.12), including bootstrap analysis based on 100 replicates (68). BLAST analysis of the complete SF100 genome found the highest similarity with S. oralis ATCC 35037 (GenBank accession no. LR134336), with ANI values of 95.5% and 86.2%, respectively, for S. oralis ATCC 35037 and S. mitis NCTC12261 (CP028414). Because an ANI cutoff of 95 to 96% is used for species definition (65), these results indicated that SF100 should be classified as a strain of S. oralis. In the phylogenetic tree analysis (Fig. S4), the S. mitis and S. oralis groups were clearly separated, and SF100 clustered with S. oralis strains, further indicating that SF100 is a member of this species.
SF100 was grown in Todd-Hewitt broth (Difco, Franklin Lakes, NJ) supplemented with 0.5% yeast extract (THY). Escherichia coli strains were grown at 37°C under aeration in Luria broth (LB, Difco). Appropriate concentrations of antibiotics were added to the medium, as required.
Cloning and expression of fibrinogen Aa and its variants. cDNA encoding the Aa chain of human fibrinogen was generously provided by Susan Lord (University of North Carolina at Chapel Hill) (69)(70)(71). Full-length and truncates of the Aa chain were cloned into pMAL-C2X (New England Biolabs, Ipswich, MA) with specific primer sets (Table S2) to express maltose binding protein MalE-tagged versions of variants. All recombinant proteins were purified by affinity chromatography with amylose resin according to the manufacturer's instructions (New England Biolabs).

Inhibition of Fibrinolysis by Phage Lysin
® number of bound bacteria was determined by staining with crystal violet (0.5% vol/vol) for 1 min, as described previously (32).
Surface plasmon resonance (SPR) spectroscopy. SPR spectroscopy was performed using a Reichert-4 SR7500DC system (Reichert Technologies, Munich, Germany). Purified human fibrinogen (0.1 mM) in sodium acetate buffer (pH 5.5) was covalently immobilized on a plain gold surface polyethylene glycol (PEG) sensor chip. Increasing concentrations (range, 0 to 250 mM) of lysin SM1 , lysin  , or plasminogen in PBS were flowed over fibrinogen at a rate of 30 ml/min with 3 min association and dissociation times. The sensorgram data were subtracted from the corresponding data from the reference flow cell and analyzed using Scrubber2 software (Reichert Technologies). A plot of the level of binding (response units) at equilibrium against a concentration of analyte was used to determine the K D .
Analysis of lysin SM1 expression by Western blotting. S. oralis SF100 and PS1006 were harvested by centrifugation of liquid cultures at an A 600 of 0.8, and the pellet was lysed with 6 M urea. The culture supernatants were concentrated by centrifugation with Amicon Ultra-50 units (Merck Millipore). The samples were separated by SDS-PAGE with 3 to 8% Tris-acetate gels (Invitrogen) under reducing conditions and then were transferred to nitrocellulose membranes. After blocking, the membranes were incubated with rabbit anti-lysin SM1 IgG (1:3,000), followed by incubation with HRP conjugated goat anti-rabbit IgG (1:5,000).
Statistical analysis. Data expressed as means 6 standard deviations (SD) were compared for statistical significance using the unpaired t test with Prism v7.0 (GraphPad Software, Inc., La Jolla, CA, USA). P , 0.05 was considered to be statistically significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.