The amino acid sequences of enolase proteins of S. suis (SSU1320), S. pneumoniae (SPD_1012), S. pyogenes (Spy0731), B. lactis (AAZ22544.1), L. plantarum (JDM1_0656), and S. mutans (SMU.1247) were aligned using ClustalX2 (Larkin et al., 2007). Protein structural information is more reliable than the corresponding amino acid sequence for inferring the function of protein as it changes less than the protein sequence during evolution (Rost, 1999). Therefore, we predicted S. suis P1/7 enolase secondary structure with the sequence alignment outputs using ESPript 3.0 (Robert and Gouet, 2014) to investigate homologous motifs among different enolase proteins. The enolase gene sequence identity of various streptococcal strains was analyzed using the Nucleotide Basic Local Alignment Search Tool (BLASTn).

The basic purpose of protein superimposition is to develop an optimal conformational alignment of protein structures with minimized root-mean-square-deviation (RMSD) of alpha-carbon (Cα) atom pairs and to analyze the conservation of protein folds across the aligned proteins. The tertiary structure of S. suis P1/7 enolase (SSU1320) was predicted based on the enolase gene coding sequence and using the resolved three-dimensional structure of S. suis strain 05ZYH33 enolase (PDB entry 4ewj) as the template through the Robetta server (Lu et al., 2012). S. pneumoniae enolase and L. plantarum enolase, of which Plg-binding motifs had been determined (Bergmann et al., 2003; Vastano et al., 2013), were selected as structural templates to respectively perform pairwise structural rigid-body superimposition with predicted S. suis enolase three-dimensional structure in PyMoL software (The PyMOL Molecular Graphics System, Version 2.5.2 Schrödinger, LLC). In each superimposition, S. suis enolase was computationally rotated and transformed to align to the template enolase with minimized RMSD of C_α atom pairs. A typical RMSD value for homologous proteins was proposed as 3 Å (Chothia and Lesk, 1986). Lower RMSD values predict higher similarity between two proteins. Protein superimposition outputs containing specific homologous amino acid residues or motifs were generated by PyMoL software and exported as images. Plg-binding motifs of S. pneumoniae enolase and L. plantarum enolase were used to predict Plg-binding motifs of S. suis enolase in silico because of the structural similarity. All PyMoL commands needed during the superimposition could be found in the PyMoL Command Reference web (https://pymol.org/pymol-command-ref.html).

Four eno mutant strains were generated by the CRISPR/Cas9-mediated genome editing. This involved four steps: (i) the construction of gRNA-Cas9 co-expression plasmid, (ii) the construction of the DNA repair template, (iii) the transformation of S. suis, and (iv) validation of the gene editing. Guide RNAs (gRNAs) which form an alternative to the crRNA:tracrRNA were designed in Benchling (Biology Software. Retrieved from https://benchling.com) and potential off-target scores were checked using CRISPR RGEN Tools (http://www.rgenome.net/cas-designer/) (Bae et al., 2014; Park et al., 2015). The best gRNA sequence was cloned into plasmid pSS by annealing of two oligonucleotides containing overhangs that anneal to the BsaI digested vector. Plasmid pSS encodes Cas9 and a chloramphenicol-resistance gene (Gussak et al., 2023). Details of thermocycler conditions for generating gRNA and inserting gRNA into pSS are included in the Supplementary Material . Four different repair templates were generated with the mutated Plg-binding sites with approximately 1000 bp of homologous sequence upstream and downstream the eno gene and cloned into plasmid pUC57 which encodes a kanamycin-resistance gene (Gussak et al., 2023). Detailed sequences of all DNA fragments included in the repair templates can be found in Supplementary Table S1 and corresponding thermocycler conditions are included in the Supplementary Material . At predicted Plg-binding sites Lys-261, Lys-252/Glu-253/Lys-255, Lys252/Lys-261, and Lys-434/435, the lysine residues were substituted by alanine and glutamic acid by glycine. For remaining residues in the gRNA complementary region, we replaced codons with amino acids representing same proteins so that the repair template could escape the Cas9-mediated double stranded DNA breaks.

Escherichia coli DH5α cells (NEB, C2987H) were transformed with the shuttle vectors harboring gRNA-Cas9 or repair templates and were grown in the Luria Bertani Broth (LB) with the required antibiotics: 25 µg/ml of chloramphenicol or 50 µg/mL of Kanamycin, respectively. Finally, we transformed co-expression vectors containing the respective gRNAs and corresponding repair templates into wild-type S. suis, as previously described (Zaccaria et al., 2016) with slight modification. Transformed S. suis strains were grown overnight at 37 ˚C with 5% CO₂. The overnight culture was diluted 1:40 with fresh Todd Hewitt Broth (THB; Difco Laboratories, Detroit, MI, The United States) with 0.2% yeast extract (THY) medium and incubated at 37 ˚C until the optical density at 600 nm (OD_600nm) was between 0.035 and 0.058. Aliquots of 100 µL of S. suis liquid culture were transferred into 1.5 mL Eppendorf Safe Lock Tubes™ and combined with 5 µL synthetic competence-inducing peptide solution (Zaccaria et al., 2016) (5 mM, GenScript), 1 µg of pSS/gRNA-Cas9, and 2 µg of DNA repair template. After 2 h of incubation at 37 ˚C with 5% CO₂, the mixture was spread onto THY agar medium containing chloramphenicol at a final concentration of 5 µg/mL.

S. suis serotype 2 strain P1/7 and the four isogenic strains carrying the mutations in the predicted Plg-binding sites eno²⁶¹ , eno^252/253/255 , eno^252/261 , and eno^434/435 were grown overnight in 10 mL THY at 37 ˚C with 5% CO₂. After overnight culture, the bacterial suspension was centrifuged at 4000 rpm for 10 min, resuspended in 10 mL fresh THY medium, measured for OD_600nm, and diluted to OD_600nm = 0.08. Subsequently, wild-type and S. suis enolase mutants were cultured an additional one hour to reach concentrations of about 3×10⁸ CFU/mL and used for growth and Plg binding experiments. S. suis strains were cryopreserved in 50% glycerol-THY at -80 ˚C.

The human cerebral microvascular endothelial cell line (hCMEC/D3) used as a model of the BBB was purchased from Merck (Cat.SCC066) and cultured in commercial EndoGRO™-MV Complete Media Kit (Milipore, Cat. # SCME004) supplemented with 1 ng/mL of FGF-2 (Cat.GF003) according to the manufacturer’s recommendations.

Bacteria were cultured as described above and growth measured by OD_600nm using a spectrophotometer (Molecular Devices, CA). Measurements were recorded for three independent bacterial cultures at each timepoint. Additionally, colony forming units (CFU) of bacteria in batch culture were counted. At each timepoint, a 1:10 serial dilution of each batch culture in PBS was performed and 10 µL from each dilution was spread on THY agar media using triplicate plates for each timepoint. Plates were incubated overnight at 37 °C with 5% CO₂ and colonies counted the next day. The average CFU/mL of the S. suis culture was calculated from three independent experiments for each timepoint.

Protein expression plasmid pNIC28-Bsa4 was used for expression of wild-type and mutant enolase as previously described with minor modification (Charalampos Aslanidis, 1990). The eno genes were amplified using the genomic DNA as template from wild-type S. suis or plasmids carrying the eno²⁵¹ homologous repair templates as described in the Supplementary Material . The PCR products were recovered after agarose gel electrophoresis, treated by T4 DNA polymerase and dCTP, and then combined with T4 DNA polymerase and dGTP-treated pNIC28-Bsa4. After 2 h of incubation at 22 ˚C, 5 µL of each pNIC-eno construct was transformed to chemically competent E. coli DH5α cells (NEB, C2987H) which were subsequently plated onto LB agar media containing kanamycin. The colonies containing the correct pNIC-eno constructs were identified by colony PCR and then cultured in fresh LB medium with 50 µg/mL of kanamycin at 37 ˚C overnight with shaking at 220 rpm. The pNIC-eno plasmids were isolated using the Qiagen QIAprep^® Spin Miniprep Kit (Qiagen) according to the manufacturer’s instructions. The concentration of plasmid DNA was measured by DS-11 spectrophotometer (Denovix, Wilmington, DE, The United States). Constructs was verified by sequencing the purified DNA plasmid (Eurofins Genomics).

For protein expression, Rosetta (DE3) cells were transformed with the plasmids of wild-type enolase and eno²⁵¹ mutant by heat shock. After overnight culture, transformed Rosetta cells were grown in LB with 30 µg/mL of kanamycin and 40 µg/mL of chloramphenicol at 2 L scale. Bacterial cultures with an OD_600nm of 0.6 were induced by isopropyl-β-D-thiogalactopyranoside (IPTG) at a final concentration of 0.4 mM during overnight growth at 18 ˚C. Bacterial cells were harvested by centrifugation for 15 min at 4000 rpm and resuspended in 20 mL MiliQ water and then stored at -20 ˚C until further use. Resuspended bacterial cells were thawed and transferred in lysis buffer (25 mM Tris, 200 mM NaCl, 1 mM TCEP, pH 8.0). Cells were lysed by sonication: 125 s-on and 10 s-off on ice with an output power setting of 60%. Cell debris and insoluble proteins were removed by centrifugation for 30 min at 21,000 rpm at 4 ˚C, while the supernatant was loaded onto a column containing 3 mL Nickel beads and equilibrated with wash buffer (25 mM Tris, 200 mM NaCl, 1 mM TCEP, 20 mM imidazole, pH 8.0). The Nickel beads were washed with 20 mL wash buffer to elute unbound proteins and the his-tagged protein was eluted with 6 mL elution buffer (25 mM Tris, 200 mM NaCl, 1 mM TCEP, 200 mM imidazole, pH 8.0). All fractions of the eluate were analyzed by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). For both wild-type and eno²⁵¹ enolase proteins, the first elution fraction was used for further purification using size-exclusion chromatography. The recovered proteins were incubated with 50 µL of 6 mg/mL his-TEV protease, transferred to a dialysis membrane and dialyzed overnight at 4 ˚C to remove the imidazole. Nickle beads were then used to remove the histone affinity tag essentially using the same procedure described above. Proteins were concentrated by Amicon Ultra-15 concentrator with 30,000 Da cut-off filters to a final volume of 1 mL and injected into a S200 16/60 gel filtration column, equilibrated in PBS. Fractions were collected and chromatograms were recorded. Aliquots of the fractionated proteins were prepared in PBS and snap-frozen in liquid nitrogen before storage at -80 ˚C. The final concentrations of purified soluble wild-type and eno²⁵¹ proteins in PBS were 2.5 mg/mL and 2.6 mg/mL, respectively.

To measure the activity of wild-type and eno²⁵¹ enolase, the conversion of 2-phospho-D-glycerate (2-PG) to phosphoenolpyruvate (PEP) by purified enolase protein was measured, as previously described (Pancholi and Fischetti, 1998). Briefly, different concentrations of 2-PG (0.5, 1, 2.5, 3, and 6 mM) were incubated with 5 µg of purified enolase protein at 37 ˚C for 3 min in HEPES/NaOH buffer (100 mM HEPES, 10 mM MgCl₂, 7.7 mM KCl; pH 7.0) with a final volume of 750 µL. The enolase activity was recorded as the rate of change in the absorbance at 240 nm with a 5-s interval for a period of 3 min. To calculate the released rate of PEP, the Lambert-Beer formula (A = α_m cl) was used (α_m : the extinction coefficient for PEP, 1.164×10^-3 M^-1). The relation between the concentration of 2-PG (S) and rate of PEP formation (v) was presented in Michaelis-Menten plot according to the equation v = d[P]/dt = V_max × [S] × (K_M+[S])^-1. V_max represented the maximum rate reached by the reaction at saturating substrate concentration and a given enzyme concentration. K_M was equal to the substrate concentration when the rate of product formation was half of V_max. The reciprocal of both sides of the Michaelis-Menten equation was used to Lineweaver-Burk plot. The kinetic coefficients, K_M and V_max, were calculated from the intercepts and slopes of the Lineweaver-Burk plot (Lineweaver and Burk, 1934).

Wild-type and four S. suis enolase mutants were cultured in 10 mL THY medium overnight at 37 ˚C and centrifuged at 4000 rpm for 10 min. Bacterial pellets were resuspended in 10 mL fresh THY medium and the OD_600nm was measured. All bacterial suspensions were diluted to OD_600nm of 0.08 in 10 mL THY and subsequently subcultured for 1 h at 37 ˚C with 5% CO₂ to obtain expected final concentrations for all strains based on the growth curve information. After subculturing, all cultures were pelleted and resuspended with PBS to the concentration of 1×10⁹ CFU/mL. Using these concentrated bacterial solutions, Pln activity assays were performed as previously described (Bergmann et al., 2005b, Vastano et al., 2013; Buscetta et al., 2016) with minor modifications. Briefly, 500 µL of bacterial suspension (5×10⁸ CFU S. suis) was incubated with 10 µL of human Plg (1 mg/mL, Sigma-Aldrich) in the presence or absence of 5.6 µL of recombinant human tPA (42.8 KIU/mL, abcam) for 2 h at 37 ˚C. Bacteria incubated with 20 µL of human Pln (0.3 units/mL, Sigma-Aldrich) for 2 h was used as positive control. After incubation, bacteria were washed with 600 µL of PBS to remove unbound Plg and then resuspended in 400 µL of chromogenic substrate D-Val-Leu-Lys-p-nitroanilide (VALY) (400 µM, Sigma-Aldrich). Absorbance at 405 nm was measured soon after addition of VALY but before resuspension (A_t0). After 18 h of incubation at 37 ˚C with 5% CO₂, bacteria were pelleted and the absorbance of 100 µL of the supernatant was measured at 405 nm (A_t18) in a 96-microtiter plate. The activity of S. suis-bound Pln was calculated as ΔAbsorbance= A_t18-A_t0. As a negative control, bacteria were not incubated with Plg and only with tPA. The blank control was bacteria alone which were not incubated with Plg or tPA. All data were normalized relative to the blank control.

To determine the relation between Plg-binding ability and Pln activity of all S. suis strains in this study, a Pln activity standard curve was generated using the Plasmin Activity Assay Kit (ab204728) by plotting the rate of Relative Fluorescence Units (RFU) increase (ΔRFU/h) as a function of the amount of Pln (µg). As the initial amount of Plg added was the same for all strains, the Plg binding % reflects the Plg binding ability of wild-type and mutant strains of S. suis.

The hCMEC/D3 cell line was grown to confluency on collagen type I pre-coated 24-well plate Transwell inserts (Corning, Cat.353096). For each insert, 500 µL of cell suspension with the concentration of 3×10⁴ cells/mL in EndoGRO™-MV Complete medium was seeded, which yielded a density of 5×10⁴ cell/cm². The wild-type and enolase mutant S. suis were diluted accordingly in EndoGRO™-MV Complete medium to a Multiplicity of Infection (MOI) of 10 bacteria per hCMEC/D3 cell. 500 µL of bacterial suspension with or without Plg plus tPA was inoculated in the upper chamber of inserts and 1.5 mL of pre-warmed EndoGRO™-MV Complete medium was added to the lower chamber of each well. 10 µL of 10 mM EDTA was added to the insert as a positive control for tight junction disruption, while the empty insert (without cells) was used as the blank for trans-endothelial electrical resistance (TEER) measurements. After 3 hours, all inserts were moved to a new 24-well plate containing 1.5 mL pre-warmed medium and TEER values were measured again. The lower chamber medium from the original plates were collected and serially diluted with PBS to enumerate translocated bacteria in each well. The translocation ratio was calculated after 3 hours as the bacterial CFU in the lower chamber divided by the bacterial CFU in the inoculum added to the upper chamber.

At least three independent experiments were performed in each study and triplicate samples used for the measurements. Graphs were created using GraphPad Prism software version 9 (GraphPad Software). Results were presented as means ± standard deviations (SD). Student’s t test, One-way ANOVA, and Two-way ANOVA were implemented where appropriate to analyze the differences between groups in different experiments. A P value of < 0.01 was considered statistically significant.

BLASTn search results showed that the enolase gene (eno) is present in all S. suis strains, with 98.62% to 100% nucleotide sequence identity ( Supplementary Table S2 ), corroborating previous reports that eno is highly conserved in S. suis (Zhang et al., 2000; Morgulis et al., 2008). Sequences of enolase proteins from S. pneumoniae, S. pyogenes, B. lactis, L. plantarum, and S. mutans that have been reported to bind to Plg via specific Plg-binding motifs were aligned with the enolase amino acid sequence from the virulent S. suis strain P1/7 to reveal the conservation of proposed Plg-binding sites across all species ( Figure 1 ).

To identify potential Plg-binding sites in S. suis enolase, the predicted tertiary structure was superimposed on the corresponding structures of S. pneumoniae enolase and L. plantarum enolase, for which Plg-binding motifs have been reported (Bergmann et al., 2003; Ehinger et al., 2004; Bergmann et al., 2005b, Vastano et al., 2013) ( Figure 2 ). The predicted tertiary structure of S. suis enolase was similar to the crystal structure of S. pneumoniae enolase ( Figure 2A ). The RMSD (see Methods) of Cα atoms in the two superimposed proteins was 0.269 Å, which is lower than 3 Å, the value considered as threshold to delimit homologous proteins (Chothia and Lesk, 1986), suggesting the structure of both proteins is highly conserved. The pneumococcal Plg-binding site FYDKERKVY (in bold the amino acid residues crucial for Plg binding) is identical in S. suis and located with an RMSD of only 0.699 Å in the superimposed proteins. Four amino acids of S. pneumoniae enolase previously proposed to interact with Plg, Asp-250, Lys-251, Glu-252, and Lys-254, showed orthologous residues Asp-251, Lys-252, Glu-253, and Lys-255 in S. suis enolase (Bergmann et al., 2003; Ehinger et al., 2004) ( Figure 2A -a). Although the C-terminal double-lysine residues of S. suis and S. pneumoniae enolases aligned with an RMSD value of 0.005 Å, site-directed mutagenesis of these lysine residues in S. pneumoniae eno gene did not impair Plg-binding, suggesting that it might not be a universal Plg-binding site (Bergmann et al., 2005b, Bergmann et al., 2003; Ehinger et al., 2004). Therefore, we hypothesized that Lys-434 and Lys-435 were also less likely to be a Plg-binding site in S. suis ( Figure 2A -b). Our results disproved this hypothesis as described in the sections below.

Superimposition of the predicted tertiary structure of S. suis enolase with the crystal structure of L. plantarum enolase yielded a RMSD value of 0.177 Å, showing that the tertiary structure of L. plantarum enolase was more like S. suis than S. pneumoniae enolase ( Figure 2B ). The region from residue 249 to residue 261 with sequence FYDKERKVYDYTK in S. suis enolase and the region from residue 248 to residue 259 on L. plantarum enolase were closely aligned in the superimposed protein structures, with a RMSD value of 0.212 Å. Because Lys-259 has been proposed in protein as an effective Plg-binding site of L. plantarum enolase (Vastano et al., 2013), and this protein appears highly similar to S. suis enolase, we hypothesized that S. suis enolase Lys-261 might also contribute to Plg-binding ( Figure 2B ).

The five putative Plg-binding sites encoded by the S. suis eno gene were modified by CRISPR/Cas9 gene editing to introduce amino acid substitutions predicted to affect binding. The substitution of Asp-251 (eno²⁵¹) was not possible. Four obtained enolase mutant S. suis strains (eno²⁶¹ , eno^252/253/255 , eno^252/261 , and eno^434/435 ) were verified by sequencing and their growth were compared in THY ( Figures 3 , 4A ). Plate counting of bacteria showed that the wild-type and all four enolase mutant strains remained in lag phase for about 2 hours, entered exponential phase after about 2 hours and stationary phase after around 6 hours ( Figure 4B ). Strains eno²⁶¹ , eno^252/261 , and eno^434/435 reached approximately 1 × 10⁹ CFU/mL after 6 h growth which was similar to wild-type at 6 h. Although eno^252/253/255 appeared to grow faster than other strains reaching a density of 1.7 × 10⁹ CFU/mL after 6 h (P = 0.003; three independent experiments), nonparametric multiple unpaired t-test analysis did not find significant differences between growth of eno^252/253/255 and the wild-type strain across all timepoints, revealing that the site-directed substitutions on Lys-252/Glu-253/Lys-255 of S. suis enolase have no effect on bacterial growth.

As we failed to construct a viable eno²⁵¹ mutant strain, the recombinant wild-type and eno²⁵¹ mutant proteins were expressed and purified in vitro to detect whether the enzyme activity of eno²⁵¹ mutant protein was different from wild-type enolase. 2-PG-to-PEP conversion rates in the presence of different concentrations of substrate (2-PG: 0, 0.5, 1, 2.5, 3, 6 mM) were displayed in in a Michaelis-Menten plot Figure 5A , and the corresponding Lineweaver-Burk plot in Figure 5B . The values of K_M and V_max for wild-type enolase and eno²⁵¹ were determined to be 3.087 mM and 527.9 μM/min (wild-type), and 1.260 mM and 279.8 μM/min (eno²⁵¹ ), respectively. The maximum rate of the enolase reaction (V_max) was lower for eno²⁵¹ than for wild-type enolase in the presence of excess substrate while the substrate affinity (Km) of eno²⁵¹ was increased compared to wild-type. These results suggested that eno²⁵¹ has a higher affinity to the substrate 2-PG but a lower catalytic reaction rate. We speculate that this might be due to structural changes that affect the enzyme's active site or its ability to undergo conformational changes necessary for catalysis, explaining why we could not recover a viable eno²⁵¹ mutant.

We determined Plg-binding ability of S. suis by conversion of bound Plg to Pln with tPA and measuring Pln activity with a chromogenic substrate. A standard curve was generated with purified Pln to calculate the amount of Pln in the assay ( Figure 6A ). All four enolase mutants bound significantly less Plg than the wild-type S. suis strain (P < 0.01), showing that the amino acid substitutions indeed reduced Plg-binding ( Figure 6B ). The similar effect of different amino acid substitutions on bacterial Plg-binding ability suggested that all potential binding sites contribute to binding affinity. All identified Plg-binding sites likely contribute to Pln-binding because all four enolase mutants were observed to bind less Pln than wild-type S. suis.

To investigate if Plg binding to S. suis enolase contributes to S. suis translocation across the BBB, we compared transmigration of the enolase mutants (eno²⁶¹ and eno^252/253/255 ) with that wild-type S. suis across hCMEC/D3 cell monolayers. Before starting the experiment, the TEER of the hCMEC/D3 cells monolayer was measured using an epithelial Volt-Ohmmeter (EVOM3) device (World Precision Instruments, Europe) to ensure that all inserts contained intact monolayers with proper barrier function. Only inserts with TEER values higher than 100 Ω × cm² were used for S. suis translocation assay (Santiago-Tirado et al., 2019) ( Supplementary Figure S1 ). Log-phase wild-type and enolase mutants with human Plg plus tPA were applied to the upper chamber at a MOI of 10 bacteria per endothelial cell, while S. suis incubated with human Pln was regarded as the positive control. After 3 h of infection, the medium in the lower chamber was removed and dilutions were plated on THY agar to enumerate CFU of S. suis that had crossed the hCMEC/D3 monolayer. The translocation ratio was calculated as the ratio of translocated S. suis to the inoculum added to the upper chamber. The addition of Plg to wild-type S. suis significantly increased bacterial translocation across the BBB but had no effect on translocation of the eno^252/253/255 and eno²⁶¹ mutants ( Figure 7 ). These results indicated that Plg-binding sites in enolase contribute to wt S. suis translocation across the BBB ( Figure 7 ).