The Complement proteins purified from human plasma C5, C6, C7, C8, C9 (C6–C9) and S-protein bound to C5b-9 (SC5b-9 complex), C5b6, and specific goat anti-human polyclonal antibodies were purchased from Complement Technology. Secondary anti-goat IgG antibody conjugated with peroxidase was purchased from KPL. Normal human serum (NHS) was obtained from healthy volunteers, after informed and signed consent (CEPSH #1206).

We employed two saprophytic (non-pathogenic) Leptospira strains: L. biflexa serovar Andamana strain CH11 and L. biflexa serovar Patoc strain Patoc I, and five pathogenic strains: L. interrogans serovar Pomona strain Pomona, L. interrogans serovar Kennewiki strain Fromm, L. interrogans serovar Copenhageni strain 10A, L. interrogans serovar Cynopteri strain 3522C and L. interrogans serovar Panama strain CZ 214. All Leptospira strains were cultivated for 7 days in modified Ellinghausen-McCullough-Johnson-Harris (EMJH) medium at 29°C under aerobic conditions (Barbosa et al., 2006). The supernatants containing the secreted proteases were obtained as described before by Fraga et al. (2014). Briefly, fresh bacterial suspensions containing 1.0 × 10⁹ leptospires were incubated at 37°C for 4 h. The supernatants were collected after centrifugation, passed through a 0.22-mm filter, aliquoted and immediately frozen at -80°C until use.

The recombinant PepSY-M4 fragment from thermolysin (IUBMB EC 3.4.24.27) contains the prodomain PepSY and the catalytic domains Peptidase_M4 and Peptidase_M4_C (amino acids 144–794) as described by Fraga et al. (2014). Escherichia coli DH5a and E. coli BL21 (SI) were used for cloning and expression, respectively.

Total SPL or SNPL protein concentrations were determined using Pierce BCA Kit. Before proteins were separated by 8–15% SDS-PAGE (Laemmli, 1970) they were solubilized in reducing loading buffer, heated for 3 min at 96°C. Both SPL and SNPL were considered free of lipopolysaccharide (LPS) according to the Limulus Amebocyte Lysate test (<0.125 endotoxins units/ml).

Supernatants of pathogenic Leptospira, SNPL (1.5 μg of total secreted proteins) or recombinant thermolysin (0.5–6 μg) were incubated with purified human C6, C7, C8, or C9 proteins (0.25 μg each) in a final volume of 25 μl (ratio: 5 supernatant proteins/1 Complement protein), up to 4 h at 37°C. We also used 10% NHS as a source of Complement proteins (in 22 μl of final volume). To evaluate the proteolytic specificity of SPL, proportionally lower concentrations of SPL proteins (0.07, 0.45, or 0.9 μg of total protein) were added to 5 μg of each C6–C9 proteins (respectively, ratios 0.01:1, 0.04:1, and 0.018/1 of supernatant protein/complement protein) diluted in PBS pH 7.4 in a final volume of 17 μl. When we assayed the proteolytic specificity of SPL to cleave purified C5, 0.06, 0.3, and, 0.6 μg of SPL were incubated with 3 μg of purified C5 in a final volume of 17 μl for 1 h or 4 h at 37°C.

We also evaluated the SPL cleavages of C6–C9 present in the SC5b-9 complex incubating 5 μg of this complex with 0.9 μg of SPL, for 4 h at 37°C (17 μl of final volume). In some experiments, SPL was pre-incubated with inhibitors of serine-, metallo-, cysteine-, or aspartyl-proteases [5 mM phenylmethylsulfonyl fluoride (PMSF); 5 mM 1,10-phenanthroline; 28 μM E-64 or 5 μM pepstatin, respectively] or 3 μM human plasma protease inhibitor α₂-macroglobulin (α₂M) (Roche) for 30 min before adding purified C6–C9 proteins. The cleavage products were subjected to 15% SDS-PAGE and analyzed by Western blot, as described by Fraga et al. (2014). After transferring proteins to nitrocellulose membranes, non-specific protein interactions were blocked with 10% skimmed milk diluted in PBS containing 0.05% Tween (PBS-T), followed by incubation with polyclonal goat anti-human C5–C9 antibodies, washed with PBS-T and incubated with rabbit peroxidase-conjugated anti-goat IgG antibodies. Positive signals were detected by enhanced chemiluminescence (West Pico, Pierce). When indicated, cleavage products were visualized after electrophoresis and staining with Coomassie Blue (Sigma–Aldrich). Molecular weight standards (Page Ruler) were purchased from Fermentas.

The interaction between thermolysin and C6 was analyzed by ligand affinity blotting (overlay) and ELISA. In the first case, recombinant thermolysin and a negative control (BSA) were subjected to 12% SDS–PAGE under non-reducing conditions and transferred to nitrocellulose membranes (Castiblanco-Valencia et al., 2012). After blocking with 10% skimmed milk in PBS-T, C6 (10 μg/ml) was added to the membrane for 90 min at 37°C. After washing, bound proteins were detected with goat anti-human C6 polyclonal antibody (1:5000), followed by incubation with peroxidase-conjugated secondary anti-goat antibody (1:5000). Positive signals were detected by enhanced chemiluminescence (West Pico, Pierce). The interactions between recombinant thermolysin and C7, C8, and C9 were also investigated by ELISA, following the same protocol. The interaction between Complement proteins (C6, C7, C8, and C9) and thermolysin was quantified by ELISA. Microtiter plates were coated with thermolysin or BSA (negative control) (100 μl; 10 μg/ml) for 16 h at 4°C. The wells were washed with PBS-T and blocked with PBS containing 3% BSA for 2 h at 37°C. After this period, Complement proteins (C6, C7, C8, or C9) were added (100 μl; 0–5 μg/ml) to the plate. The reaction was incubated for 1 h at 37°C and then washed with PBS-T. Bound proteins were detected with goat anti-human C6, anti-C7, anti-C8, or anti-C9 polyclonal antibodies (1:5000), followed by peroxidase-conjugated secondary anti-goat antibody (1:5000). As substrate, we used o-phenylenediamine dihydrochloride (Pierce) and the absorbance was measured at 492 nm. The dissociation constant (K_d) of the interactions was calculated by non-linear regression by fitting the data to the equation Y = B_max^∗X/(K_d + X) using GraphPad Prism 5.0 (GraphPad Software, Inc.). To test if the interaction of thermolysin with C6 is dependent on ionic strength or is affected by heparin, similar ELISA protocol was performed immobilizing thermolysin on microtiter wells (100 μl; 10 μg/ml) and then adding a fixed amount of C6 (1 μg) in the presence of increasing NaCl concentrations (0 to 600 mM, diluted in 10 mM Na₂HPO₄ and 1.8 mM KH₂PO₄) or heparin (0.05 to 1 μg, diluted in PBS), respectively.

To investigate if recombinant thermolysin would interfere in MAC-mediated erythrocyte lysis, we used a hemolytic assay previously described by Hallström et al. (2009) and da Silva et al. (2015). 2 × 10⁸ sheep erythrocytes/ml in Veronal-Buffered Saline were pre-incubated with purified C5b6 (2 μg/ml) in a final volume of 100 μl for 1 h at room temperature. LIC10301 is an outer membrane lipoprotein identified in pathogenic L. interrogans serovar Copenhageni, as predicted by the PSORT program¹ (Nakai and Kanehisa, 1991). LIC10301 does not bind Factor H or C4BP (Castiblanco-Valencia et al., 2012; da Silva et al., 2015). In different fresh tubes, leptospiral recombinant proteins, thermolysin (6.25 to 50 μg/ml) or LIC10301 (50 μg/ml, negative control) were pre-incubated with 2 μg/ml C7 for 30 min at 37°C. Subsequently, C8 (2 μg/ml) and C9 (0.2 μg/ml) proteins were added to the mixture for 15 min at 37°C. Next, C5b6-coated erythrocytes were added to the mixture of recombinant thermolysin-C7-C8-C9 or recombinant LIC10301-C7-C8-C9. After incubation for 30 min at 37°C, the erythrocyte suspensions were centrifuged and lysis was evaluated by measuring absorbance of the supernatants at 540 nm in the supernatants. The percentage of lysis was expressed considering 100% lysis when no leptospiral protein was added to the reaction. Statistical analyses were performed using ANOVA. Significant differences (^∗) were considered when p ≤ 0.05.

Our first approach was to investigate if SPL or SNPL could cleave purified component C5. As shown in the Figures 1A–C, C5-cleavage products were observed exclusively when we used SPL from L. interrogans serovar Kennewicki strain Fromm. Proteases present in the SPL cleaved C5 α chain, generating 35–40 kDa fragments even when using reduced amounts (0.6–0.06 μg) of SPL proteases to 3 μg C5 in 17 μl (Figure 1A). In addition, C5b as part of SC5b-9 complex was cleaved in the presence of SPL (Figure 1C). However, C5 cleavage was not observed when whole NHS was used as a source of this protein (Figure 1B).

Since C6, C7, C8, and C9 share common protein domains (Hadders et al., 2012; Bubeck, 2014), we decided to explore the proteolytic activity of leptospiral culture supernatants on these proteins. For this, seven different strains of Leptospira were selected: five pathogenic and two saprophytes (non-pathogenic). SPL or SNPL were independently incubated with purified C6, C7, C8, or C9 proteins (ratio w/w: 5 of supernatant: 1 of Complement protein). The cleavages were analyzed by Western blot employing specific antibodies against each protein.

The pattern of cleavage of C6–C9 proteins by SPL was similar for all five different strains of pathogenic leptospires employed in this study. On the other hand, no significant proteolytic activity on these purified Complement proteins was observed using SNPL from the two L. biflexa strains (Figure 2). The C6 polypeptide chain (∼100 kDa) was degraded by SPL producing mainly 95–90 kDa and 60 kDa fragments. The component C7 (95 kDa single chain) was also cleaved by SPL generating ∼80 and 55 kDa fragments. Intact C8 protein is composed of three polypeptide chains: α and β chains (both ∼64 kDa) and γ chain (∼22 kDa). SPL cleavage of C8 generated 50 and 40 kDa bands. The component C9 is composed of a single 71 kDa polypeptide chain and after incubation with SPL 50–40 kDa and ∼36 kDa fragments were observed by Western blotting analysis. In these assays, C6–C9 protein degradation by SPL (1.5 μg) was time-dependent and cleavage products could be observed before 30 min of incubation (Figure 3). The proteolytic activity specificity of SPL was observed even when relatively small amounts were used in relation to the concentrations of purified C6–C9 proteins as indicated in Figure 4, the cleavage products were still observed even when we used a SPL: Complement proteins mass ratio of 0.015. However, no C6–C9 cleavage products were observed when we treated NHS with SPL (Figure 5).

To identify which classes of proteases present in SPL are involved in the cleavage of purified C6–C9 proteins, we employed the following inhibitors: PMSF (inhibits serine proteases), E-64 (inhibits cysteine proteases), 1,10-phenanthroline (inhibits metalloproteases) and pepstatin (inhibits aspartyl proteases). Only 1,10-phenanthroline was able to inhibit the cleavage of the above Complement proteins by proteases present in supernatant from pathogenic L. interrogans serovar Kennewicki strain Fromm (Figure 6). This result strongly suggests that the metalloproteases represent one of the major classes of proteases present in the SPL involved in the cleavage of C6–C9 purified proteins. As shown in Figure 7 pathogenic L. interrogans serovar Kennewicki strain Fromm is resistant to NHS. However, bacterial survival diminishes by at least 50% in the presence of 1,10-phenanthroline-treated NHS which confirms the importance of secreted leptospiral metalloproteases for evasion from Complement activation.

Since no cleavage of C6-C9 proteins was observed when we incubated SPL with NHS, we investigated if α₂M, an abundant plasma protease inhibitor, would have the same effect than 1,10-phenanthroline. Figure 8 shows that when we pre-incubated SPL with α₂M before adding purified C6–C9 proteins, cleavage was also inhibited.

Next, we decided to investigate the proteolytic activity of the SPL proteases on the C6–C9 assembled in the MAC. Since most of circulating MAC complexes rapidly binds to the S-protein (also known as Vitronectin), SPL or SNPL were incubated with SC5b-9 complex for 2 h at 37°C under reducing and non-reducing conditions (Figure 9A). Again, only SPL presented proteolytic activity on the SC5b-9 complex. Under reducing conditions, we observed degradation generating products with 43–34 kDa proteins, and the concomitant generation of cleavage fragments sized between 26 and 17 kDa. Under non-reducing conditions, a marked degradation of proteins between 95 and 72 kDa was observed followed by several fragments with lower molecular mass. Different than what was described with individual C6–C9 proteins (Figure 2), SPL proteases were able to cleave C6 but not C7, C8, or C9 when they are organized in the MAC (data not shown).

Next, investigated if a recombinant Leptospira thermolysin previously showed be present in the SPL (Fraga et al., 2014) would cleave C6–C9 individual proteins or complexed in SC5b-9. Recombinant thermolysin (LIC13322, fragment PepSY M4) was incubated separately with each of purified C6, C7, C8, and C9 proteins or with the soluble SC5b9 complex and the cleavage products were analyzed by Western blot employing specific antibodies. In this assay, we observed that only C6 was cleaved by recombinant thermolysin LIC13322 (Figure 9B). No cleavage fragments were observed when C7, C8, or C9 were employed (data not shown). The recombinant thermolysin cleaved both purified C6 and as an integral part of the soluble SC5b9 complex. The cleavage of C6 by thermolysin has a similar pattern to that seen with it purified when we used SPL with a common 90 kDa degradation product. Still, the C6 cleavage was more pronounced in the presence of SPL when compared in the presence of recombinant thermolysin (60 kDa additional band) (Figure 9B).

The interaction between thermolysin and C6 was further explored by Far-western blot (overlay) in which recombinant thermolysin or BSA (negative control) were subjected to 12% SDS-PAGE and transferred to a nitrocellulose membrane. This membrane containing the immobilized proteins was incubated with purified C6 and the interaction was analyzed using anti-C6 specific antibody. Figure 10A shows that C6 interacts with immobilized recombinant thermolysin. We confirmed the interaction between thermolysin and C6 by ELISA, and the dissociation constant of the complex (K_d) was estimated to be of 31.8 ± 2.2 nM (Figure 10B).

We then investigated whether the thermolysin – C6 interaction is dependent on ionic strength. Since heparin affects MAC assembly (Sahu and Pangburn, 1993), we decided to investigate the effect of the addition of heparin on the thermolysin–C6 interaction. Thermolysin was immobilized and C6 protein was added in the presence of increasing concentrations of NaCl (0–600 mM; Figure 10C) or different amounts of heparin (0.05 to 1 μg; Figure 10D). We observed that the interaction between C6 and thermolysin was reduced only at high concentrations of NaCl (600 mM). Furthermore, we observed a dose-dependent inhibition by heparin (0.5 to 1.0 μg), suggesting that thermolysin may interact at least partially with C6 through heparin binding sites.

Several protein domains are common among C6–C9 proteins. We therefore asked if recombinant thermolysin, although it does not exert proteolytic activity on C7, C8, and C9, could interact directly with these molecules resulting in an additional inhibition effect by direct interaction, facilitating Leptospira Complement evasion. The interaction between thermolysin and the C7, C8, and C9 components of the MAC was analyzed by ELISA (Figure 11A). Recombinant thermolysin was immobilized and different amounts of the C6–C9 proteins were added separately to enable the calculation of the dissociation constant (K_d). We could indeed observe interactions between the Leptospira protease and the Complement C7, C8, and C9 components, with K_d values of 17.8 ± 1.3 nM; 62.0 ± 7.1 nM; and 56.8 ± 5.2 nM, respectively (Figure 11A). The interactions between thermolysin and C6–C9 components inhibits MAC-mediated hemolysis (Figure 11B) in a dose-dependent manner. These results demonstrated that the proteases secreted by the pathogenic Leptospira may inhibit the formation of MAC by cleaving C6, and/or by direct binding to C6-C9.