The wild-type SS2 strain ZY05719 was isolated from a diseased pig during an outbreak in the Sichuan province of China. The bacterial strains and plasmids used in this study are listed in Table 1; the sequences of all primers are listed in Supplementary Table 1. S. suis strains were grown at 37°C in Todd–Hewitt broth (THB, Oxoid Cheshire, United Kingdom) or THB agar (THA). Escherichia coli strains were grown on Luria–Bertani (LB) agar plates or in LB broth at 37°C. For mutant selection, 100 μg/mL spectinomycin was added to S. suis medium. Ampicillin (100 μg/mL) was used to maintain plasmids pBADHisA and pBADHisA-pelB, and 50 μg/mL kanamycin or 150 μg/mL erythromycin was used to maintain the pTCV-lac plasmid.

An overnight Top10 bacterial solution containing pBADHisA and pBADHisA-pelB plasmids was diluted 1:100 in fresh LB broth, supplemented with 100 μg/mL ampicillin (LB-ampicillin), and grown to OD₆₀₀ of 0.2–0.3. Each culture was divided in two aliquots: one was supplemented with 0.2% L-arabinose and the other one not. A growth curve was constructed based on hourly OD₆₀₀ measurements from at least three independent experiments. At the same time, after adding 0.2% L-arabinose, samples were diluted every 3 h and colony-forming units (CFU) were counted by spreading the serially diluted PBS in a 10-fold suspension of bacteria on LB agar plates for a 24 h of incubation period at 37°C.

Nine putative type II TA systems in SS2 ZY05719 were predicted by TAfinder. DNAStar Lasergene 7¹ and BLAST from NCBI² were used to analyze DNA and amino acid sequences. The antitoxin and toxin three-dimensional (3D) structure was predicted using the SWISS-MODEL server, whereas the secondary structure was predicted using PHYRE2². The promoter of Xress-MNTss was predicted using the SoftBerry website.

Mutants were constructed via natural DNA transformation, with some modifications (Peterson et al., 2004; Zaccaria et al., 2014). The up and downstream sequences of the target gene were amplified by PCR with primer pairs, from the genomic DNA of strain ZY05719. The up and downstream sequences were fused with the sacB-spc cassette by overlap PCR. The linear fusion DNA fragment used for the mutants and synthetic peptide were added to the 100-ul bacteria [optical density at 600 nm (OD6₀₀), 0.042]. To generate ΔXress-MNTss, ΔMNTss, and ΔXress, the composite samples were incubated at 37°C for 2 h under static conditions and then plated in THB containing spectinomycin. The sacB gene, which is sensitive to sucrose, was used as a negative control. Next, the fusion homologous fragment without any cassette was transferred to the primary positive mutant for the second transformation, after which the transformed bacteria were maintained on a THB plate containing 10% (w/v) sucrose. Construction of D-MNTss and D-Xress point mutant strains followed a similar protocol to that of deletion strains, except that the translation initiation codon ATG was mutated to CTG.

The Xress-MNTss promoter (300 bp) was amplified from ZY05719 genomic DNA. The PCR product was ligated in the pTCV-lac reporter plasmid to obtain pTCV-lac-300, which was then transferred to wild-type, deletion, and point mutant strains. The β-galactosidase activity assay was performed according to Miller’s method (Aviv and Gal-Mor, 2018) with some modifications. The overnight culture broth was diluted 1:100 with fresh THB and placed in a CO₂ incubator at 37°C for cultivation. Upon reaching log phase (OD₆₀₀ ∼ 0.6), 2 ml of bacterial cells culture was collected by centrifugation, washed twice with sterile phosphate-buffered saline (PBS), and resuspended in 200 μL pre-cooled Z-buffer containing 50 mM β-mercaptoethanol. Subsequently, 0.1% SDS and chloroform were added, the suspension was thoroughly mixed and placed in a 30°C water bath for 5 min; after which 100 μL O-nitrophenyl-β-D-galactopyranoside (4 mg/mL) was added, mixed well, and the reaction was allowed to proceed at 30°C until the solution was no longer yellow. At that point, 250 μL sodium carbonate was added to stop the reaction, the solution was centrifuged, and 250 μL of supernatant was aliquoted to a 96-well plate. Absorbance at 420 nm (A₄₂₀) and 550 nm (A₅₅₀) was recorded with a microplate reader, and β-galactosidase activity was calculated using the following formula:

where MU = Miller units; t = reaction time; and v = volume of culture assayed in milliliters. At least three independent cultures of each strain were assayed in each experiment.

Whole-cell RNA from bacteria in log phase was extracted using TRIzol (TaKaRa) according to the manufacturer’s instructions. After removing any contaminating DNA with gDNA wiper, the extracted RNA served as template to synthesize cDNA using a HiScriptII first-strand cDNA synthesis kit (Vazyme). The QuantStudio 6 Flex RT-PCR system and ChamQ Universal SYBR qPCR master mix (Vazyme) were used to determine the concentration of selected transcripts. The housekeeping gene parC was used as an internal reference (Wu et al., 2014), and the 2^{–Δ Δ CT} method was employed to calculate the relative fold change. At least three replicates were performed for each sample (Livak and Schmittgen, 2001).

For the cotranscription test, total RNA was extracted using a bacterial RNA extraction kit and divided in two aliquots: one was reverse-transcribed into cDNA, and the other was not (negative control). Primers were designed to span zy05719_RS04595-zy05719_RS04600, zy05719_RS04600- zy05719_RS04605, and zy05719_RS04605- zy05719_RS04610 (Supplexmentary Table 1). Negative controls and cDNA were used as templates for co-transcription analysis.

The Xress coding sequence was amplified, cut with restriction enzymes, and ligated to pCold^TM II to generate the prokaryotic expression vector pCold^TM II-Xress. Separately, the antitoxin coding sequence was amplified, the stop codon was removed, the linker GGGGSGGGGSGGGGS was added to connect the antitoxin to the toxin coding sequence, and the construct was ligated in pCold^TM II to generate pCold^TM II-Xress-MNTss. The His-tag was placed at the N-termini of the protein. The pCold^TM II-Xress and pCold^TM II-Xress-MNTss plasmids were transformed into BL21(DE3) competent E. coli, which were cultured to OD₆₀₀ ∼ 0.4–0.6. At that point, 0.5 mM isopropyl-D-thiogalactopyranoside was added and cells were cultured at 16°C for another 16 h. Next, the cells were harvested and sonicated in lysis buffer (20 mM Na₃PO₄⋅12H₂O, 0.5 mM NaCl, 30 mM imidazole, pH 7.4), after which the antitoxin Xress and the Xress-MNTss protein complex were purified on a His-tag Ni-NTA affinity chromatography column. The protein was eluted with a step-wise gradient using imidazole concentrations ranging from 50 to 500 mM. The protein concentration was determined by performing a Bradford assay with bovine serum albumin as a standard.

For EMSA, native polyacrylamide gel electrophoresis was performed by incubating the purified protein with Xress-MNTss promoter fragments probe. The latter (300 bp) were amplified by PCR and purified using a kit (TaKaRa). The negative-control probe was amplified from 16S rRNA. The purified protein and DNA probe were added to the binding buffer (10 mM Tris-base, 50 mM KCl, 5 mM MgCl₂, 1 mM dithiothreitol, 0.05% Nonidet P-40, 2.5% glycerol, pH 7.5), incubated at 37°C for 30 min, and then subjected to 6% native polyacrylamide gel electrophoresis in 0.5 × TBE buffer (44.5 mM Tris-base, 44.5 mM boric acid, 1 mM EDTA, pH 7.5) at 200 V for 45 min. The gel was stained in 0.5 × TBE containing ethidium bromide for 20 min and images were taken.

The MICs of antibiotics against S. suis were determined according to the Clinical and Laboratory Standards Institute Guidelines (Humphries et al., 2018). The strains were diluted 1,000-fold in THB, and 180 μL of the inoculated culture was added to the first well of a 96-well microtiter plate and 100 μL to the other wells. Next, 20 μL antibiotics was added to the first well, mixed, and 100 μL of the mixture was added to the following well. The procedure was repeated until the 10th well. The 11th well served as the positive control and the 12th well as the negative control. The 96-well plate was then incubated at 37°C for 20 h, and the results were recorded. Each experiment was repeated independently three times.

To assess the virulence of the deletion strain in vivo, we randomly divided BALB/c mice into five groups of 10 mice each, and challenged them with 5 × 10⁸ colony-forming units (CFU)/mouse. The control group was challenged with PBS. Clinical symptoms and survival of the mice were monitored for 7 days. Additionally, a bacterial load assay was performed to evaluate the proliferation capacity in vivo. Each group consisted of six mice, and the intraperitoneal injection dose used was 3 × 10⁸ CFU/mouse of ZY05719 and ΔXress. At 6 h postinfection, the mice were anesthetized with isoflurane and euthanized by CO₂. Blood, brains, livers, and spleens were harvested, weighed, and homogenized in PBS. Bacteria were isolated from these homogenates and blood by plating serial 10-fold dilutions on a THB-agar medium to enumerate CFU. Animal experiments were conducted at the Animal Center Laboratory of Nanjing Agricultural University and were approved by the Jiangsu Provincial Laboratory Animal Monitoring Committee.

Biofilm formation was analyzed by staining with 0.1% crystal violet in a 96-well plate. Briefly, bacteria were grown to OD₆₀₀ = 0.6, diluted 1:100, inoculated into a 96-well plate, and incubated at 37°C for 3 days. THB served as a control. The medium was discarded and the cells were gently washed twice with PBS to remove any unattached bacteria. After fixing with methanol for 30 min, the fixation solution was discarded, the samples were dried at room temperature, and the biofilm was stained with 0.1% crystal violet for 30 min. The samples were washed with tap water, dried, 33% acetic acid was added, the mixture was placed on a shaker at 80 rpm/min for 30 min to release crystal violet, and absorbance was measured at 600 nm. Ten independent cultures were used for each strain.

Adhesion experiments were performed in vitro according to established methods. Briefly, Hep-2 human laryngeal epithelial cells were cultured in Dulbecco’s modified Eagle medium (DMEM) containing 10% fetal bovine serum until they reached a monolayer in 24-well cell plates (5 × 10⁵ cells/well). The bacteria were cultured to log phase, washed with DMEM, and diluted to a density of 5 × 10⁷ CFU/mL. Each cell plate well was infected with 1 mL of bacterial solution, centrifuged at 800 × g for 15 min, and incubated at 37°C for 2 h. The cells were washed five times with sterile PBS to remove floating bacteria, treated with 100 μL of 0.25% trypsin-EDTA at 37°C for 10 min, and rinsed with 900 μL sterile deionized water to release any bacteria adhering to the cells. Finally, the bacteria were diluted and counted on a THB plate. The results were expressed in terms of relative adhesion frequency, compared with the adhesion frequency of the wild-type (set to 100%). Data were obtained from at least three independent experiments.

To quantify the level of capsule production, we performed a dot blot assay. Briefly, 5 μL of bacteria, twofold serially diluted in PBS, were spotted onto a nitrocellulose membrane and fixed with 70% ethanol for 5 min. After air-drying, the membranes were blocked with blocking solution (5% w/v skim milk in PBS containing 0.05% Tween 20) for 2 h. Anti-SS2 polyclonal antibody (1:500 dilution) was applied to probe the nitrocellulose membrane spotted with bacteria. After washing with PBS-Tween 20 buffer, the membrane was incubated with horseradish peroxidase-conjugated anti-rabbit secondary antibody (1:2,000 dilution, Boster). The signal was detected using the Tanon High-sig ECL western blotting kit. The average gray value was quantified using ImageJ software³. The experiment was repeated three times, and the ZY05719 capsule-deletion strain (ΔCPS) was used as a control.

Streptococcus suis morphology was observed using a transmission electron microscope. Briefly, bacteria were grown to mid-exponential phase, centrifuged at 5,000 × g for 10 min, and then fixed with 2.5% glutaraldehyde for at least 2 h. The samples were dehydrated in propylene oxide for 10 min and visualized using a Hitachi H-7650 apparatus according to the manufacturer’s instructions. This experiment was performed independently three times.

All experiments were repeated at least three times. GraphPad Prism version 8 was used for analysis and plotting. Statistical significance was set to P < 0.05, and an unpaired two-tailed Student’s t-test or log-rank (Mantel-Cox) test was applied to analyze the data.

Nine putative type II TA systems in SS2 strain ZY05719 were predicted using TAfinder (Table 2), an online tool in TADB (Shao et al., 2011). The localization of TA systems in the complete genome of ZY05719 and related sequence information are shown in Supplementary Figure 1. Based on a comparative analysis, TA1, TA4, TA7, TA8, and TA9 were found to be homologous to type II TA systems RelBE1 (SC84), SezAT (05ZYH33), RelBE2 (SC84), yefM-yoeB (SC84), and ParDE (SC84), respectively (Yao X. et al., 2015; Zheng et al., 2015; Xu et al., 2018). Instead, TA2, TA3, Xress-MNTss, and TA6 have not been characterized yet.

Next, we identified the putative TA systems. Using the pBADHisA plasmid, we found that the toxins T2/T3, MNTss, and T6 elicited no toxic effect in the cytoplasm, with addition of the inducer L-arabinose (Figure 1A). Thus, we chose plasmid pBADHisA-pelB to induce toxin secretion in the periplasmic space. The periplasmic localization (pBADHisA-pelB) was achieved by fusion to the PelB leader sequence (Lei et al., 1987). In order to rule out the possibility of protein accumulation in the periplasm and lead to growth arrest, we added a negative control (pBADHisA-pelB-RHSse). Growth of the resulting E. coli strain slowed significantly when L-arabinose was added, indicating a bactericidal effect (Figure 1B). The accumulation of RHSse (QRR36965.1) protein in the periplasm did not lead to growth inhibition, indicating that the growth inhibition caused by T2/T3, MNTss, and T6 was due to toxic effects. And T2/T3 showed bacteriostatic effect, while MNTss and T6 showed significant bactericidal effect (Figure 1C). These results demonstrated that the toxins T2/T3, MNTss, and T6 exerted a toxic effect in the periplasmic space rather than the cytoplasm. To identify the antitoxin of MNTss, we ligated the antitoxin and toxin coding sequences in pBADHisA-pelB, and assessed bacterial growth to determine if the antitoxin neutralized the toxin. As shown in Figure 1D, E. coli containing toxin-only pBADHisA-pelB-MNTss was significantly inhibited, whereas that harboring pBADHisA-pelB-Xress-MNTss could largely neutralize the toxin, indicating that MNTss and Xress formed an active TA system.

Further, the 3D structure of the antitoxin Xress (Supplementary Figure 2A) and toxin MNTss (Supplementary Figure 3A) was predicted using the SWISS-MODEL protein homology-modeling server⁴. Xress was predicted to belong to the Xre family (Supplementary Figure 2B) of transcriptional regulators harboring a helix-turn-helix domain. MNTss was predicted to belong to the MNT family (Supplementary Figure 3B) and contain a nucleotidyltransferase domain similar to the kanamycin nucleotidyltransferase of Staphylococcus aureus. So we named this TA system Xress-MNTss.

Since studies have reported that antitoxin has a regulatory effect (Guo et al., 2019), and the previous analysis found that Xress has the potential to regulate genes. So we analyze the gene sequence structure of the Xress-MNTss system, the SoftBerry website⁵ was used to predict the Xress-MNTss promoter (Figure 2A). Several pairs of palindromic sequences were found in the Xress-MNTss promoter using an online prediction website⁶. The antitoxin or TA complex of a typical type II TA system can bind to its own promoter to ensure auto-regulation. To determine whether this was the case of Xress-MNTss, we constructed theΔXress-MNTss and CΔXress-MNTss strains, as well as inserted the Xress-MNTss promoter in the pTCV-lac plasmid. An in vivo promoter activity assay revealed significantly more β-galactosidase activity in the ΔXress-MNTss strain compared to the wild-type and CΔXress-MNTss, suggesting that the Xress-MNTss system was capable of inhibiting its own promoter (Figure 2C).

To directly determine whether the antitoxin regulated the Xress-MNTss promoter in vivo, ΔXress and ΔMNTss were constructed (Supplementary Figure 4), toxin and antitoxin expression was measured in the ΔMNTss, ΔXress and ΔXress-MNTss mutants. Toxin expression was 25-fold higher in the ΔXress mutant than in the wild-type, whereas the antitoxin was significantly downregulated (Figure 2B). Because these results suggested that the antitoxin Xress was likely to negatively regulate the Xress-MNTss promoter, toxin and antitoxin point mutants were constructed (Figure 2D). Transcription of toxin in the antitoxin point mutant was significantly upregulated, and transcription of antitoxin was significantly downregulated in toxin point mutation (Supplementary Figure 5). The pTCV-lac plasmid was transformed into the point mutant strains and ΔXress to analyze promoter activity in vivo. As expected, promoter activity was fourfold higher in the antitoxin point mutant (Figure 2E) and sixfold higher in the ΔXress (Supplementary Figure 6). This finding indicated that the antitoxin exerted a negative regulatory effect on the Xress-MNTss system.

And in order to assess direct binding of Xress-MNTss to its own promoter, an electrophoretic mobility shift assay (EMSA) was conducted with purified antitoxin protein (Figure 3A) and TA complex (Figure 3B). The antitoxin Xress and TA Xress-MNTss complex could retard the mobility of the Xress-MNTss promoter in a dose-dependent manner (Figures 3C,D), revealing that the antitoxin and TA complex bound directly to the Xress-MNTss promoter to regulate the Xress-MNTss system.

Further use the island viewer website⁷, the Xress-MNTss system was found to be located on a putative genomic island (Figure 4A). Gene structure analysis revealed a streptomycin resistance gene (zy05719_RS04610) downstream of the Xress-MNTss system, and prediction by the BPROM suite in SoftBerry showed that zy05719_RS04595–zy05719_RS04610 formed a single operon and shared the same promoter (Figure 4B). ZY05719_RS04610 is homologous to streptomycin adenylyltransferase (SSUSC84_0863) in S. suis SC84. Given that the antitoxin Xress and Xress-MNTss complex bound to the Xress-MNTss promoter and had a negative auto-regulatory effect, the Xress-MNTss system likely regulated the expression of zy05179_RS04610, too. Indeed, zy05719_RS04610 was significantly downregulated in ΔXress-MNTss and ΔXress, but significantly upregulated in ΔMNTss (Figure 4C), reflecting the higher sensitivity to streptomycin in the former two and resistance in the latter (Table 3). Hence, zy05719_RS04610 might mediate resistance to streptomycin in ZY05719. However, the regulation mechanism of streptomycin resistance needs to be further explored.

Because Xress has the ability to regulate target genes, we wondered whether Xress also regulates other genes to cause corresponding phenotypic changes. And given that the TA system has been reported to participate in biofilm formation, this characteristic was investigated. Biofilm formation was significantly increased in the ΔXress strain compared to that in the wild-type (Figures 5C,D), confirming the involvement of the Xress-MNTss system.

Based on reports of the involvement of the TA system in pathogenicity, BALB/c mice were used to explore whether the Xress-MNTss system contributed to bacterial virulence. Mice infected with ZY05719, ΔXress-MNTss, and ΔMNTss strains showed clinical symptoms and mortality rates of 100%, 90%, and 100%, respectively. In contrast, survival was significantly increased in mice infected with ΔXress (Figure 5A). And the bacterial abundances of ΔXress in the blood, brain, liver, and spleen were significantly lower than that of ZY05719 (Figure 5B). No significant growth difference was observed between wild-type and mutant strains when cultured in THB (Supplementary Figure 7). These results showed that the Xress-MNTss system contributed to pathogenicity of S. suis ZY05719. Furthermore, ΔXress adhered significantly better to Hep-2 cells than the wild-type (Figure 6A). Previously, loss of the S. suis capsule was shown to increase adhesion to epithelial cells. Here, dot blot analysis was used to verify loss of the capsule in ΔXress. Compared to the wild-type strain, the dot-blot signal was significantly higher in ΔXress, and was comparable to the ΔCPS control (Figures 6B,C). Transmission electron microscopy analysis of wild-type and ΔXress cells showed significantly decreased capsule thickness in the latter (Figure 6D). Taken together, these results suggested that the Xress-MNTss system participated in biofilm formation and capsule synthesis.