The bacterial strains and plasmids used in this study are listed in Table 1. The virulent S. suis strain SC-19 was isolated from a diseased pig during the 2005 outbreak in Sichuan, China (Li et al., 2009). Since the genome of SC-19 has not been sequenced, the genome sequence of the strain S. suis 05ZYH33 (GenBank accession number CP000407) was used as reference for gene clone, transcriptomic, and phosphoproteomic analysis. S. suis 05ZYH33 was isolated from an infected human during the same outbreak in Sichuan (Lun et al., 2007). Both of these two isolates are serotype 2. Bacteria were grown in TODD-Hewitt broth (THB; OXOID, England) medium or plated on THB Agar (THA; OXOID) with 5% (v/v) sheep blood at 37°C. Erythromycin (90 μg/ml) was added to screen the mutant strain and erythromycin (90 μg/ml) and spectinomycin (100 μg/ml) were added to select for a complementary strain.

Escherichia coli (E. coli) strain DH5α (Trans, China) was used as host strain for cloning and E. coli strain BL21 (DE3) (Trans) was used to express His-tag fusion proteins. E. coli strains were cultured in Luria-Bertani (LB) Broth (Difco, France) or plated on LB Agar at 37°C. If necessary, kanamycin (25 μg/ml) was added.

The stk deletion strain was obtained using an existing method (Takamatsu et al., 2001b; Zhu et al., 2014). Primers used in this study were designed according to the genome sequence of S. suis 05ZYH33 and are listed in Table S1. Primers SU-F/SU-R and SD-F/SD-R were used to amplify the upstream and downstream regions of stk. Moreover, the fragments were cloned into pSET4s, respectively. Finally, the erythromycin-resistance expression cassette (erm^R) was amplified from pAT18 with primers Erm-F/Erm-R and cloned to achieve the stk-knockout vector pSET4s-S. The pSET4s-S was then electroporated into SC-19. Through homologous recombination, the stk gene was replaced the erm^R via double-crossover incident.

The promoter sequence of IMPDH was selected to drive the expression of stk gene in SC-19 (Takamatsu et al., 2001a; Zhu et al., 2014). The promoter and the coding sequence of stk were cloned into pSET2 to obtain the recombinant plasmid pSET2-S. The pSET2-S was transformed into the mutant strain Δstk to acquire the complementary strain CΔstk.

To further confirm the mutant strain Δstk and the complementary strain CΔstk, we performed Western blot analysis. The overnight cultures of SC-19, Δstk and CΔstk, cultured in THB were collected by centrifugation and resuspended in bacterial lysis buffer (50 mM Tris/HCl, pH 8.5, 100 mM NaCl, 2 mM ethylene diamine tetraacetic acid, 100 μg/ml lysozyme, 1 mM phenylmethanesulfonyl fluoride, 0.5% Triton X-100). Bacterial cells were lysed with a French pressure cell press. Quantified bacterial lysate was separated on SDS-PAGE and then electrotransferred to PVDF membrane (Invitrogen, USA) (Tan et al., 2015). Mouse anti-STK serum was produced as described previously (Li et al., 2011) by using recombinant STK protein. The PVDF membranes (Invitrogen, USA) were probed with primary antibodies against STK (1:1000) or GidA as control (1:1000; stored at −80°C in our laboratory). After wash, the membranes were incubated with goat anti-mouse IgG (H+L)-HPR (1:5000; Southern Biotech, USA). Antibody-tagged protein bands were detected by using Western ECL Substrate Kit (Bio-Rad, USA).

To confirm the mutant strain Δstk and the complementary strain CΔstk, we also performed RT-PCR (Tan et al., 2015). Briefly, RNA was isolated using SV Total RNA Isolation System (Promega, USA) according to the manufacturer's instructions. Then, cDNA was synthesized using HiScript Q Select RT SuperMix (Vazyme, China) according to the manufacturer's instructions.

To confirm whether the upstream and downstream genes of stk in the mutant were affected or not, the genes were amplified using the primers SSU05_0427-F/SSU05_0427-R (for upstream gene) and SSU05_0429-F/SSU05_0429-R (for downstream gene; Table S1) using the SC-19 cDNA as template.

The coding sequence of stk was amplified using primers stk-F/stk-R from the SC-19 genome. The PCR product was restricted with EcoRI and XhoI and then inserted into the digested pET-28a vector to generate the recombinant plasmid pET-stk, which was then transformed into E. coli BL21 (DE3) cells. Expression was induced by 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) (Sigma, USA) at 18°C for 12 h. His-tagged STK was purified using Ni-NTA columns (GE Healthcare, Sweden) according to the manufacturer's recommendation. The purified protein was identified by Western blot analysis using the His-tag rabbit-polyclonal antibody (Sigma, USA) and STK-mouse polyclonal antibody (made in our laboratory), respectively.

The kinase assay was performed as described previously (Boitel et al., 2003; Fernandez et al., 2006). The reaction was conducted in 50 μl of kinase buffer (50 mM Hepes, 1 mM DTT, 0.01% Brij35, pH 7.0) containing 2 mM MnCl₂, 100 μM ATP and 1 μCi of [γ-³²P]-ATP (PerkinElmer, UK). The reaction was initiated by adding the kinase at 30°C for 10 min and stopped by adding SDS-PAGE sample buffer plus EDTA (25 mM final). A total of 20 μl of the reaction was subjected to electrophoresis. In each case, the reaction products were separated on a 12% SDS-polyacrylamide gel and the radiolabelled proteins were visualized using auto-radiography. To obtain relative quantification of the radiolabelled ATP incorporation, the radioactive samples were analyzed using a PhosphorImager apparatus (Fujifilm, Japan). For the substrate phosphorylation, myelin basic protein (MBP; Sigma, USA) was used as a positive control to test the kinase activity. The enzyme/substrate ratio was 1:10 with 0.4 μM kinase.

The strains SC-19, Δstk and CΔstk at the mid-log phase were collected and then resuspended twice using ddH₂O. Each sample (20 μl) was fixed on glass slides (Shitai, China) through flaming. A Gram staining kit (Jiancheng, China) was utilized according to the manufacturer's instruction. The stained samples were observed under light microscope.

Scanning electron microscope (SEM) observation was performed following previous methods (Wang et al., 2011; Fleurie et al., 2012). The bacteria were grown in THB broth at 37°C and harvested at OD₆₀₀ of 0.6. Bacterial suspensions were spotted onto glass coverslips (0.17 mm in thickness and 20 mm in diameter; WHB, China) and washed with phosphate-buffered saline. The cells were fixed with 2.5% glutaraldehyde overnight at 4°C. All samples were dehydrated with a series of gradient ethanol and air dried. The dried samples were covered with a 10 nm gold/platinum layer (JSM-6390LV, JEOL, Japan) and observed via SEM (JFC-1600, JEOL, Japan).

All SEM images were analyzed with Image J. The sizes (in pixels) of bacterial chains were defined as their two-dimensional area based on the “Auto Threshold” function, which defined the cell chain outline. Shapes that were not chains and chains on the border of images were manually excluded from subsequent analyses (Dalia and Weiser, 2011). For chain length analysis, data are the result of at least 100 chains analyzed per sample from at least 100 fields.

Growth rates of the SC-19, Δstk and CΔstk were detected through the measurement of the density changes represented by OD₆₀₀ nm values and CFU counts of the cultures. Different strains were grown overnight in THB medium and then the initial OD₆₀₀ nm of all subcultures were adjusted to 0.014. The diluted cells were incubated at 180 rpm/m at 37°C in the same medium and OD₆₀₀ nm values were read every hour until the growth process entered stationary phase. Meanwhile, 100 μl of the bacterial culture was diluted and then vortexed to break the chains. Bacteria numbers were recorded every hour by viable count. All the growth data were analyzed with Origin 8.0. The growth rate at each time point was calculated based on the function of “Non-linear Curve Fit” and “Mathematics.” The average growth rate (AGR) was calculated from data in mid-log phase based on the function of “Linear Curve Fit” (OriginLab, Northampton, MA, USA).

All mice used in our study were purchased from the Wuhan Institute of Biological Products (Wuhan, China). The study was performed strictly in accordance with the Ethics Committee of Huazhong Agricultural University according to Hubei Province Laboratory Animal Management Regulations-2005. All efforts were made to minimize suffering.

To probe the possible role of the stk gene in S. suis virulence, 24 female specific-pathogen-free (SPF) Kunming mice (4–6 weeks old) were divided into three groups. Groups 1 and Group 2 were intraperitoneally injected with 1 × 10⁹ colony-forming unit (CFU) mid-log-phase cells of either SC-19 or Δstk. Saline was applied in Group 3 as negative control. Clinical signs and survival time were recorded. The mice were observed for 7 days to obtain steady survival curves.

To better evaluate the pathogenicity of Δstk, we performed a determination of viable bacteria in organs assay as described previously (Gao et al., 2016). Six-week-old female SPF Kunming mice (6 mice per group) were intraperitoneally inoculated with 5 × 10⁸ CFU of a 1:1 mixture of mid-log phase SC-19 and Δstk. Saline was applied as negative control in six mice. At 6, 18, and 52 h post-infection (hpi), blood samples, brains, and lungs were obtained from each group. The samples were homogenized after weighing, and serial dilutions were plated onto THA. To count the colonies, we used streptomycin (20 μg/ml) for SC-19, whereas 20 μg/ml streptomycin and 90 μg/ml erythromycin were used for Δstk.

To investigate gene expression profiles between SC-19 and Δstk, RNA-Seq was performed in the BGI, Shenzhen. Three biological replicates were mixed as one sample before RNA-Seq. RNA isolation and purification were the same as aforementioned. RNA-Seq based transcriptomic profiles were conducted as previous cited (Wilhelm and Landry, 2009). After the quality inspection using an Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System, the library was sequenced using Illumina HiSeq 2000. Quality control of raw reads was performed and clean reads were mapped onto the complete reference S. suis 05ZYH33 genome.

The gene expression level was calculated by using RPKM (reads per kb per million reads) method. The number of reads that uniquely aligned to a unique gene was normalized to RPKM (Mortazavi et al., 2008). The RPKM method eliminates the effect of different gene lengths and sequencing levels on the calculation of gene expression. Therefore, the gene expression can be directly calculated by comparing the different gene expression level among samples (Chen et al., 2014). To identify DEGs between two samples, a statistical analysis of the frequency of each unique-match read in each library was performed by referring to “the significance of digital gene expression profiles” (Audic and Claverie, 1997; Chen et al., 2014). The probability of gene A expressed equally between two samples can be calculated with the following formula, wherein the total clean tag number of the sample 1 is N₁, and total clean tag number of sample 2 is N₂; gene A holds x tags in sample 1 and y tags in sample 2; p-value corresponds to differential gene expression test (Audic and Claverie, 1997; Chen et al., 2014).

or

False Discovery Rate (FDR) was used in the multiple hypothesis testing to correct for p-value (Chen et al., 2014). Following the formula below, assuming W DEGs had been selected, M genes of those were really differential expressed, whereas H genes indicated no difference which were false positive. If we decide that the error ratio “Q = H/W” must stay below a cutoff (e.g., 5%), we should preset the FDR to a number no larger than 0.05. FDR-values were calculated according to the previous algorithm (Li J. et al., 2015)

To detecte DEGs, we used FDR ≤ 0.001 and log transformed fold change (the ratio of RPKM values) > 2.0. Moreover, enrichment analysis based on KEGG pathway database (http://www.genome.jp/kegg/) was done using the R package GAGE v2.22 (Luo et al., 2009). Pathway-level differential expression was visually checked using Pathview (Luo and Brouwer, 2013). Functional classification of gene products was predicted using BLASTP v2.5.0+ (Altschul et al., 1997) by the clusters within the orthologous group (COG) database v2014 (Galperin et al., 2015). The q-value cutoff for BLAST searching was set to 0.1. Putative virulence-associated genes (VAGs) of S. suis were screened by searching all protein sequences against Virulence Factor Database (VFDB; Chen et al., 2005) using BLASTP. The E-value cutoff for BLAST searching was set to 1e-20. The RNA-Seq data have been submitted to the NCBI Gene Expression Omnibus (GEO) database under the accession number GSE87759.

A subgroup (n = 10) of DEGs was selected to cross-validate the RNA-Seq data with SYBR green detection. The primers (Table S1) were designed according to the genomic sequence of S. suis 05ZYH33. RNA extraction was carried out as described above. qRT-PCR was performed on an ABI 7300 HT Sequence Detection System using the ABI Power SYBE Green PCR Master Mix. The 16S rRNA served as an internal reference gene. The relative expression level was measured using the 2^−ΔΔct method. Data were reported as mean relative expression levels (± standard deviation) between SC-19 and Δstk. Five depressed genes in Δstk (SSU05_1776, SSU05_0272, SSU05_1815, SSU05_0309, and SSU05_2154) and five stimulative genes in Δstk (SSU05_0792, SSU05_0358, SSU05_1011, SSU05_0906, and SSU05_1532) were chosen to do qRT-PCR for RNA-Seq confirmation. The assays were replicated thrice.

Subsequently, we performed qRT-PCR to compare the transtriptional levels of the above genes between SC-19 and CΔstk related to Δstk. The relative expression level was measured using the 2^−ΔΔct method. Data were reported as mean relative expression levels (± standard deviation) between SC-19 and Δstk, as well as between CΔstk and Δstk.

SC-19 and Δstk cells at mid-log phase were cultured in THB as described above. Three independent biological replicates of bacterial pellets were then treated with SDT buffer (4% SDS, 100 mM Tris-HCl, 1 mM DTT, pH 7.6) and heated for 15 min at 100°C. The cell suspensions were sonicated for 5 min (10 s sonication with 15 s interval, 80 W) on ice and protein concentrations in the supernatants were determined through Bradford protein assay (Gao et al., 2016). Each sample (20 μg) was then prefractionated using 12% SDS-PAGE. Subsequently, the gel was processed using 200 μl UA buffer (8 M Urea, 150 mM Tris-HCl, pH 8.0) and subjected to in-gel tryspin digestion at 37°C for 16 h (Hu et al., 2014). The resulting tryptic peptides were labeled according to the protocol of iTRAQ Reagent-8 plex Multiplex Kit (Applied Biosystems, USA). The labeled peptides were mixed, concentrated using a vacuum concentrator and resuspended in 500 μl 1 × DHB buffer (3% DHB, 80% CAN, 0.1% TFA). Afterward, TiO₂ beads were added and agitated for 40 min. The centrifugation was performed, resulting in the first beads. The supernatant from the first centrifugation was mixed with additional TiO₂ beads, resulting in the second beads that were collected as before. Both bead groups were combined and washed thrice with 50 μl of washing buffer I (30% ACN, 3% TFA) and then washed thrice with 50 μl washing buffer II (80% ACN, 3% TFA) to remove the non-absorbed material. Finally, the phosphopeptides were eluted with 50 μl of elution buffer (40% ACN, 15% NH₄OH), followed by lyophilization and MS analysis.

Phosphopeptide solution (6 μl) was mixed with 0.1% (v/v) fluoroacetic acid (20 μl). Equal amounts of digested protein were loaded into a Thermo Scientific EASY column (2 cm^*100 μm 5 μm-C18) and then washed with solvent A (99% H₂O, and 0.1% formic acid). By applying solvent B (84% acetonitrile, 16% H₂O, and 0.1% formic acid), the peptides were eluted from the trapping column over a Thermo scientific EASY column (75 μm* 100 mm 3 μm-C18) with a gradient (0–55% B for 220 min at 250 nL/min, 55–100% B for 8 min, and 100% B for 12 min) using Thermo scientific Easy nLC system. For MS analysis, peptides were analyzed in positive ion mode. MS/MS was carried out with a Q-Exactive mass spectrometer (Thermo Finnigan, USA) setting in a positive ion mode and data-dependent manner choosing the most abundant precursor ions using a full MS scan from m/z 350–1800, with a resolution of 70,000 at m/z 200. Target value determination was based on automatic gain control and dynamic exclusion duration was 30 s. MS/MS scan was acquired at a resolution of 17,500 at m/z 200. Normalized collision energy was 29 eV and the underfill ratio was set to 0.1%. Quantitation was achieved by comparing the peak areas and resultant peak ratios among the four MS/MS reporter ions from 114 to 117 Da.

The data files produced by LC-MS/MS were processed by Proteome Discoverer 1.4 and searched by Mascot 2.2 (Matrix Science, MA) against 26,496 S. suis protein-coding sequences deposited in the Uniprot database (downloaded on August 4, 2015). For peptides after phosphopeptide enrichment, the following options were used. Peptide mass tolerance: ± 20 ppm; MS/MS tolerance: 0.1 Da; enzyme: trypsin; max missed cleavages: 2; fixed modifications: carbamidomethyl (C), iTRAQ8plex (K) and iTRAQ8plex (N-term); variable modifications: oxidation (M), phosphorylation (Ser/Thr/Tyr). For peptides, proteins, and phosphosites identification, FDR was estimated and the threshold was set to 1%. The peptides were determined as true phosphorylation based on the following criteria: Phospho RS score > 50 and Phospho RS site probabilities > 75%. Proteome Discoverer 1.4 software was used to extract the peak intensity within 20 ppm of each expected iTRAQ reporter ion from each analyzed fragmentation spectrum. Only spectra in which all the expected iTRAQ reporter ions detected were used for quantification.

The intensity of the reporter ions was used for phosphopeptide quantification. We normalized the phosphopeptide ratios by dividing by the median ratio of all peptides identified. As for the quantitative analysis, the log₂ fold-change values (SC-19/Δstk) were calculated for each phosphopeptide. Only phosphopeptides detected in at least two out of the three biological replicates were used for assessment of significant change. The t-test was employed to identify significant changes between the wild type strain and the mutant strain among the three biological replicates. The phosphopeptides that passed t-test with p < 0.05 were considered to be significantly regulated (Fan et al., 2014). We also included the cutoff for the log₂ fold change values, in which the phosphorylation changes were considered highly significant if the log₂ value ≥ 0.26 or ≤ −0.26. The data were then normalized and logarithmically transformed as FC. FC (Δstk/SC-19) ≥ 1.2 and ≤ 0.83 were used to represent up- or down-regulations. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaino et al., 2016) partner repository with the dataset identifier PXD005663.

The means of two groups were compared using Student's t-test (unpaired, two-tailed) in GraphPad Prism 5 (San Diego, USA), with p < 0.05 considered to be statistically significant. Unless otherwise specified, all the experiments were performed in triplicate at least thrice.

As previously described, the STK in S. suis consists of a cytoplasmic kinase domain (residues 11–267 aa) and an extracellular C-terminal region composed of four penicillin-binding proteins (residues 373–434 aa; 441–500 aa; 507–568 aa; 575–638 aa) and a Ser/Thr kinase-associated domain (PASTA; residues 348–629 aa; Figure 1A; Beilharz et al., 2012). Protein sequence alignments between STK and its homolog PknB (Rv0014c in M. tuberculosis H37Rv; NC_000962.3; Boitel et al., 2003) showed that Ser/Thr kinase domains share 43% similarity (Figure 1B). The STK expression was confirmed by Western blot analysis (Figure 1C). The ex vivo kinase assay demonstrated that STK possesses autophosphorylation and substrate phosphorylation activities (Figure 1D).

We constructed a deletion mutant of stk, as confirmed by RT-PCR (Figure S1A) and Western blot (Figure S1B). The RT-PCR showed that the stk upstream and downstream genes transcriptions were not affected by stk deletion. Western blot showed the STK protein did not expressed in Δstk, but presented in SC-19 and CΔstk.

The effect of stk knockout on morphology of S. suis was investigated using Gram staining (Figure 2A) and SEM (Figure 2B). The results revealed that stk inactivation triggered significant chain elongation (Figure 2C).

To characterize SC-19, Δstk and CΔstk, we measured growth by detecting OD₆₀₀ and CFU counts. The AGRs of mid-log phase (2–4 h) displayed no significant difference among the three strains detected by the OD₆₀₀ value (AGR of SC-19 was 0.51% per min; AGR of Δstk and CΔstk was 0.49% per min; Figure 2D). Whereas, the AGRs of mid-log phase (2–4 h) detected by CFU counts showed that Δstk grew much slower than SC-19 (AGR of SC-19 was 1.47 × 10⁵ CFU per min and Δstk was 6.58 × 10⁴ CFU per min; Figure 2E). Although the growth ability of CΔstk was not fully restored by introducing a complementary plasmid in the mutant, CΔstk grew much faster than Δstk during 2–4 h (AGR of CΔstk was 1.18 × 10⁵ CFU per min).

We tried to use the endogenous promoter of stk gene to construct the complementary strain, but it failed (no expression; the endogenous promoter needs further characterization). Therefore, we used a known promoter in S. suis genome, the promoter of IMPDH (Zhu et al., 2014). Maybe this promoter is not a strong promoter to restore all the phenotypes. Similar facts have also been reported in other studies (Ju et al., 2012).

Experimental infection of mice was performed to estimate the difference of viability in vivo between Δstk and SC-19. All SC-19-infected mice displayed severe clinical symptoms, such as septicemia, and most (7/8) died during the first day of infection. By contrast, all Δstk-infected mice displayed slight clinical symptoms and had none mortality during the 7-day observation period (Figure 3A). Therefore, the pathogenicity of Δstk was markedly attenuated.

To further evaluate the pathogenicity of Δstk, we determined the differences of viable bacteria in organs using intraperitoneal inoculation. Bacteria were recovered from blood (Figure 3B), brains (Figure 3C), and lungs (Figure 3D) at different time points post-infection. The bacterial loads in blood, brains, and lungs were lower in Δstk than SC-19, from 6 to 52 hpi. The mutant strain was almost cleared at the 52 hpi (Figures 3B–D). These results indicated that the survival ability of Δstk in the host was weakened compared with the parental strain SC-19.

To get a glimpse of transcriptomic dysregulation of Δstk, gene expression profiles were determined using RNA-Seq for Δstk and SC-19. A total of 2,116 genes were expressed with read counts greater than five in at least one sample. Among these, 539 were differentially expressed (25% of expressed genes) in Δstk compared with SC-19, with 320 up-regulated and 219 down-regulated (p < 0.05 and log transformed fold change > 2.0; Table S2). These genes were involved in diverse physiological activities, including carbohydrate metabolism, amino acid metabolism, nucleotide metabolism, and translation. As an internal control, stk transcript was not detected for Δstk but presented in SC-19.

The effectiveness of the RNA-Seq data was cross-validated using qRT-PCR. Ten genes with various expression levels, as determined by RNA-Seq analysis were chosen for qRT-PCR analysis (Table 2). The correlation between the two methods was high (R² = 0.903; Figure S2A).

The transcription levels of 10 selected genes were compared between SC-19 and CΔstk related to Δstk (Figure S2B). The differential expression of these target genes of SC-19 and CΔstk showed identical trend (five up-regulated and five down-regulated), although the fold changes were displayed significant different between SC-19 and CΔstk (p < 0.05). This indicates that the STK complementation works but not completely restored.

To get an overview of major perturbed functions in Δstk compared with the wild type SC-19, a KEGG pathway enrichment analysis was conducted. Seven metabolic pathways were significantly repressed by stk-deletion (q < 0.1; Table 3). Among these, five pathways, namely, carbohydrate metabolism (ssu00010 and ssu00500), amino acid metabolism (ssu00290), nucleotide metabolism (ssu00230), PTS (ssu02060), as well as the other two pathways (ssu03010 and ssu00970) are crucial for the central metabolism of bacterial cells and are vital in translation. Most genes in the glycolysis were significantly down-regulated, including ptsG (SSU05_0397), crr (SSU05_0398), pgk (SSU05_0157), adhA (SSU05_0279), and adhE (SSU05_0280) (Figure 4). Notably, expression levels of rpsN (SSU05_1535), rpsR (SSU05_1832), rpsP (SSU05_0796), rpsU (SSU05_1433), and glyS (SSU05_1764) translation-associated genes in the pathways of ribosome and aminoacyl-tRNA biosynthesis were decreased in Δstk. A total of 78% (28 out of 36) involved in various PTSs were down-regulated, including manX/Y/Z (SSU05_1778–1780)—encoding mannose transporter, ptsG/crr/bglF (SSU05_0397, SSU05_0398, and SSU05_1490) encoding maltose/glucose transporter, celC (SSU05_0709) encoding cellobiose transporter, and ulaA/ulaB (SSU05_2062, and SSU05_2063) encoding ascorbate transporter. Down-regulation of these pathways and related genes were potentially consistent with the decreased bacterial cell growth in the mutant strain.

The stk deletion significantly reduced virulence of S. suis. Among 180 VAGs predicted by VFDB and known VAGs of S. suis (Fittipaldi et al., 2012), 52 genes were DEGs in Δstk, including 32 down-regulated genes and 20 up-regulated genes (Table 4). These DEGs were mainly assigned into three classes: (i) 10 genes involved in adherence and immune evasion, including 9 down-regulated genes, such as adsA (SSU05_1000), clpC (SSU05_0389), and clpE (SSU05_0390); (ii) 17 genes involved in metal ion uptake, including 11 down-regulated genes, such as hitC (SSU05_1029), fbpC (SSU05_0551), and mgtB (ssu05_1418); (iii) 6 genes involved in ABC-type multidrug transport systems, including 5 up-regulated genes.

To investigate the Ser/Thr protein kinase activity and the substrates of STK in Δstk and SC-19, we conducted phosphoproteomic analysis. In total, 32 phosphoproteins were detected. Of these 12 were differentially expressed phosphoproteins (DEPPs), including 9 down-regulated phosphoproteins (FC < 0.83) and 3 up-regulated phosphoproteins (FC > 1.2; Table 5; Table S4). As an internal control, auto-phosphorylation of STK was not detected in Δstk. Among 12 DEPPs detected, all of the cell division associated proteins, including FtsA, GpsB, DivIVA, and MapZ were down-regulated. Another protein DnaK, which is a classical molecular chaperone, was down-regulated, as well as the predicted RNA-binding protein Jag. Moreover, down-regulated phosphorylation was detected in the predicted periplasmic solute-binding protein (SSU05_1717), an uncharacterized protein (SSU05_0066), and a hypothetical protein (SSU05_0636). Phosphorylation levels of three proteins, namely translation elongation factor (EF-Tu) involved in protein synthesis, fructose-bisphosphate aldolase (FBA) and GAPDH involved in glycolysis were up-regulated. Moreover, all of these three proteins were also involved in bacteria evasion of host defense, adhesion and invasion.

Inactivating S. suis STK affects the bacterial phenotypes, including cell chain length and growth kinetics (Figure 2). Consistently, the results have been reported in a previous study (Zhu et al., 2014). Several reports have also stated that stk-deficient could affect cell shapes and cell sizes in other Gram-positive bacteria, such as S. pneumoniae (Echenique et al., 2004; Fleurie et al., 2012), S. aureus (Donat et al., 2009), S. mutans (Banu et al., 2010), and M. tuberculosis (Kang et al., 2005).

In our study, we have used both OD₆₀₀ value and CFU counts to measure the growth rate of SC-19, Δstk, and CΔstk. Through these two methods, we found during the mid-log phase, the growth rate of the three strains detected by OD₆₀₀ value displayed no significant difference (Figure 2D), whereas the CFU counts showed the growth rate of Δstk was much slower than SC-19 and CΔstk during the mid-log phase (Figure 2E). Similar phenomenon has been also appeared in our previous study (Gao et al., 2016). The longer chains of the mutant strain may change the light-scattering properties of S. suis (Zheng et al., 2011) and other bacteria (Sha et al., 2004). Hence, growth detection through OD value was not adequate, and the CFU determination can accurately reflect the growth ability.

Phosphorylation of factors involved in cell division could function as an internal clock that regulates the sequence and timing of the cell cycle events (Grangeasse and Lesterlin, 2015). According to the phosphoproteomic analysis in this study, phosphoproteins associated with cell growth and division, such as DivIVA and MapZ, were down-regulated in Δstk (Table 5). This provides an explanation for the weakened growth ability and elongated chain length of Δstk. DivIVA has been identified as a STK-substrate in S. pneumoniae (Nováková et al., 2010). Moreover, S. pneumoniae with the non-phosphoablative form of DivIVA possess an elongated shape with a polar bulge and aberrant spatial organization of nascent peptidoglycan (Fleurie et al., 2012). These findings suggest that phosphoablative form of DivIVA is crucial for maintaining cell shape and division. MapZ is a newly discovered cell division protein essential in proper septum placement and likely functions as a marker of the cell division site (Holecková et al., 2014). S. pneumoniae with a non-phosphoablative form of MapZ exhibited cell shape and viability defects (Fleurie et al., 2014). Similar functions of DivIVA and MapZ may exist in S. suis. Hence, STK of S. suis may influence the cell growth and division of S. suis by regulating phosphorylation of DivIVA and MapZ. The other two down-regulated phosphoproteins in Δstk, FtsA, and GpsB involved in cell division, may also contribute to the STK regulated cell division process. In B. subtilis, GpsB is phosphorylated by STK and its phosphorylation regulates STK activity through a negative feedback loop (Pompeo et al., 2015). In E. coli, FtsA is crucial in recruiting of FtsZ filaments to the membrane and negative regulated FtsZ organization (Loose and Mitchison, 2014).

The deletion of stk in S. suis also resulted in the alteration of bacterial pathogenecity. This attenuation may result from the impaired growth of Δstk, and also can be due to direct effect on the expression of virulence genes. As previously reported, stk-deletion strain displayed reduced ability to adhere to and invade in epithelial cells and increased sensitivity to phagocytosis (Zhu et al., 2014). In this study, analysis of gene expression profiles indicated that several genes associated with adherence and immune evasion were repressed in Δstk (Table 4), such as adsA (−2.85-fold), clpC (−5.68-fold) and clpE (−11.12-fold). The gene adsA encodes an adenosine synthase AdsA. The adsA mutant of B. anthracis was easier to be cleared compared with the wild-type strain, implicating its role in immune evasion in the host (Thammavongsa et al., 2009). The gene clpC encodes a heat shock protein ClpC with ATPase activity. In L. monocytogenes, ClpC is a general stress response protein required for in vivo survival by promoting early bacterial escape from the phagosome of macrophages (Rouquette et al., 1998). Moreover, ClpC is required for adhesion and invasion. A clpC-deficient mutant of S. pneumoniae displayed decreased expression of CbpA (a structural adhesion) and pneumolysin (Nair et al., 2000). clpE acts synergistically with clpC in virulence and the clpE mutant of L. monocytogenes exhibited a significant reduction in virulence (Nair et al., 1999). Therefore, our results indicated that these down-regulated VAGs potentially contributed to decreased adherence to epithelial cells, increased immune evasion and increased sensitivity to phagocytosis of Δstk.

The Δstk mutants displayed defects in their ability to adapt to various environmental conditions, such as high temperature, high osmotic, oxidative, and low acidic pH stress (Zhu et al., 2014). These may due to the down-regulation of 11 genes encoding iron ABC transporters (HitC/FbpC, MgtB) that mediate the uptake of nutrients and several metabolism-related enzymes (i.e., ManN, PurA, and IMPDH; Table 4). S. suis requires nutrients, including trace metals, whose availability within the infected host is relatively low. The hitC/fbpC operon encodes an iron transport system that is responsible for utilizing iron bound to transferrin or iron chelates (Sanders et al., 1994). Mg²⁺ transporters encoded by mgtB are unique transport systems with unusual mechanisms for mediating Mg²⁺ movement through the membrane (Moncrief and Maguire, 1999). manN (−3.89-fold), whose product is the component of the mannose-specific PTS, can regulate hemolysin gene expression (Wilson et al., 2007). Down-regulation of these VAGs may hinder the acquisition of nutrients of S. suis, hence decrease the adaptation of S. suis to various conditions.

Other VAGs of S. suis, such as fhb (−3.45-fold) and ssnA (−3.53-fold), have been identified previously. The gene fhb encodes a factor H binding protein, which is a major antiphagocytic factor in preventing C3b deposition and blocking activation of the alternative pathway of the complement system (Pian et al., 2012). The cell wall-anchored DNase encoded by ssnA plays a role in disruption of neutrophil extracellular traps (De Buhr et al., 2014). Down-regulation of fhb and ssnA may be associated with the attenuated virulence.

According to the phosphoproteomic analysis, the phosphorylation levels of three proteins associated with virulence were up-regulated in the Δstk (Table 4). They are EF-Tu, GAPDH, and FBA. Interestingly, all of them are the bacterial outer surface proteins that contribute to evasion of host defense, adhesion and invasion (Tan et al., 2008; Li Q. et al., 2015). Ser/Thr phosphorylation of EF-Tu has been reported in E. coli, Thermus thermophilus, Streptomyces coelicolor, and B. subtilis (Pereira et al., 2011). EF-Tu phosphorylation could accelerate its release from the translation site (Absalon et al., 2009). In B. subtilis, phosphorylation impairs the essential GTPase activity of EF-Tu by preventing its release from the ribosome, leading to the overall protein synthesis inhibition (Pereira et al., 2015). However, the role of EF-Tu phosphorylation in pathogen virulence has not been reported. GAPDH is the surface expression protein with the ability to bind albumin (Brassard et al., 2004), which serves as a receptor for the plasmin receptor and contributes to host cytoskeletal protein binding and signal transduction between bacteria and host tissues (Gase et al., 1996; Jobin et al., 2004). FBA catalyzes the reversible reaction of fructose 1,6-bisphosphate transferring into glyceraldehydes-3-phosphate and dihydroxyacetone (Jado et al., 1999). The effects of GAPDH and FBA phosphorylation on virulence remain unknown. The biological significance of phosphorylation modulation by STK on these virulence associated proteins deserves further investigations.

In our study, evaluating of the transcriptional status of gene pathways revealed that seven pathways are significantly repressed in Δstk (Table 3). These repressions could be an important reason of the impaired growth of Δstk.

Glycolysis is the most critical phase in glucose metabolism to produce energy in the form of ATP during bacterial cellular respiration (Kresge et al., 2005; Mulukutla et al., 2014). Our study found more than half of the genes (17 out of 27) involved in glycolysis were repressed in Δstk, such as crr (−6.67-fold), bglA (−4.61-fold), and pstG (−20.63-fold). The gene crr, which encodes a glucose-specific phosphotransferase enzyme component for glucose transport into the cytoplasm, is involved in glycolysis, starch and sucrose metabolism (Han et al., 2007). bglA encodes a 6-phospho-β-glucosidase to convert host derived-sugars into usable monosaccharaides that contribute to energy generation for bacterial survival (Terra et al., 2015). pstG codes for a glucose transporter that enables catalysis of glucose to transform into glucose-6p (Buhr et al., 1994). Down-regulating the enzymes mentioned above in Δstk may hinder glucose utilization, nutrient storage, and ATP synthesis.

More than two thirds of the genes (34 out of 54) involved in the purine metabolism pathway were repressed in Δstk (Table S3). Previous studies have emphasized the importance of nucleotide biosynthesis in bacteria. Auxotroph mutants of nucleotide biosynthesis in Salmonella (McFarland and Stocker, 1987), S. aureus (Mei et al., 1997), and S. pneumoniae (Polissi et al., 1998) have shown impaired growth.

Our results showed that translation-associated pathways (ssu00970 and ssu03010) are the most affected pathways in Δstk, which has not been reported in other bacteria (Table 3). In Δstk, 88% of the genes involved in ribosomes (ssu03010) were decreased (46 out of 52), and in aminoacyl-tRNA biosynthesis (ssu00970), 84% of the genes were decreased (21 out of 25). The ribosome plays a major role in determining the overall gene expression profile of the cell (Starosta et al., 2014). Meanwhile, glyS (−2.51-fold) encoding a glycyl-tRNA synthetase has been suggested to be involved in DNA stabilization under extreme environmental conditions (Kumar et al., 2011). It has been previously shown that alterations of genes involved in translation can provoke slower growth of E. coli (Ruusala et al., 1984; Starosta et al., 2014).

Phosphoproteome analysis in this study has revealed that the phosphorylations of two glycolytic enzymes (FBA, GAPDH) and a translation asscociated protein EF-Tu were up-regulated (Table 4). These proteins are also involved in S. suis virulence, which have been discussed above. The phosphorylation of a chaperone DnaK was down-regulated in Δstk. DnaK is induced in E. coli upon heat shock and promotes ATP-dependent refolding or degradation of damaged proteins (Sherman and Goldberg, 1993). Previously, in vivo phosphorylation has been shown to enhance DnaK binding to polypeptide substrates (Panagiotidis et al., 1994). Modulations of phosphorylation levels on these proteins may also lead to impaired metabolism of Δstk.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.