Group A Streptococcus strain A20 (wild type strain) is an emm1-type strain and has been previously described (Chiang-Ni et al., 2009). Strain AP3 was provided by Prof. Jiunn Jong Wu (Department of Medical Laboratory Science and Biotechnology, College of Medicine, National Cheng Kung University, Tainan, Taiwan). This strain was isolated from the spleen of A20-infected BALB/c mouse (subcutaneous infection) after 3 days of infection. In this study, the genome sequence of AP3 was generated using the Miseq sequencer (Illumina) according to the manufacturer’s protocols. Sequencing reads were mapped; and SNPs and Indels were called relative to the A20 genome sequence (NCBI accession number: NC_018936). The 143T deletion in the covS gene was found and confirmed by traditional Sanger sequencing method. In addition, another six SNPs and an Indel were found in the repeat sequence regions of transposases or rRNA genes (data not shown). GAS strains were cultured on the trypticase soy agar with 5% sheep blood or in the tryptic soy broth (Becton, Dickinson and Company, Sparks, MD) supplemented with 0.5% yeast extract (TSBY) and described in Table 1. E. coli DH5α was purchased from Yeastern (Yeastern Biotech Co., LTD, Taipei, Taiwan) and was cultured in Luria-Bertani (LB) broth at 37°C with vigorous aeration. When appropriate, antibiotics chloramphenicol (25 and 3 μg/ml for E. coli and GAS, respectively), erythromycin (125 and 5 μg/ml for E. coli and GAS, respectively) and spectinomycin (100 μg/ml) were used for selection.

To treat GAS strains with neutral and acidic broths, bacterial pellets (collected from cultures in 40 ml of TSBY for 12-16 h at 37°C) were washed and resuspended in 1 ml of fresh TSBY broth. Seventy-five microliter of bacterial suspension was transferred to TSBY broth that was adjusted to pH 7.5 (supplemented with 0.1 M of HEPES) or pH 6.0 (supplemented with 0.1 M of MES) by 1 N HCl or NaOH (Dalton and Scott, 2004). Although culture broths were buffered by salts, pH of bacterial suspensions was decreased significantly after 4-5 h of incubation. Therefore, GAS strains were cultured at different pH conditions for 3 h at 37°C. The O.D.₆₀₀ of the bacterial suspensions (both pH 7.5 and pH 6.0) after 3 h incubation were around 0.4-0.6.

Group A Streptococcus genomic DNA extraction, RNA extraction, and reverse transcription were performed as described previously (Chiang-Ni et al., 2009). Real-time PCR reactions were performed in a 20 μl mixture containing 1 μl of cDNA, 0.8 μl of primers (10 μM), and 10 μl of SensiFAST^TM SYBR Lo-ROX pre-mixture (Bioline Ltd, London, UK) according to the manual. The expression level of the target gene was normalized to the gyrA and analyzed using the ΔΔCt method (7500 software v2.0.5, Applied Biosystem^®, Thermo Fisher Scientific Inc.). In addition, all values of control and experiment groups were divided by the mean of control samples before statistical analysis (Valcu and Valcu, 2011). Primers used for real-time PCR analysis were designed by Primer3¹ (v.0.4.0) according to the MGAS5005 sequence (NCBI reference sequence: NC_007297.1) and are described in Table 2. Southern blot and Northern blot analyses were performed as described previously (Chiang-Ni et al., 2012). Ten microgram of genomic DNAs extracted from GAS strains were digested by restriction enzymes for Southern blot analysis. Three microgram of RNAs extracted from GAS strains under different culture conditions were subjected to Northern blot analysis. DNA probes used for detecting the lacD.1 gene and chloramphenicol cassette were amplified by primers described in Table 2 and labeled with alkaline phosphatase (AlkPhos Direct Labeling and Detection System, GE Healthcare, Chicago, IL, USA) according to the manual. After hybridization, the membrane was washed and the signal was detected by the Gel Doc XR+ system (BioRad, Hercules, CA, USA).

Plasmid pMW506, the covR gene interrupted by a chloramphenicol cassette via the MslI site on E. coli vector pSF152, was provided by Prof. Jiunn Jong Wu. The plasmid pMW506 was electroporated into A20 and AP3 by the method described previously (Chiang-Ni et al., 2008) to interrupt the covR gene by a double-cross homologous recombination (designed SW656 and SW934, respectively).

To construct covR/covS-complemented strain, the covR/S allele (2505 bp, including promoter region) was amplified from wild type A20 strain using primers CovR/S-F-3 and CovR/S-R-2 (Table 2) and ligated into the BamHI site of E. coli-GAS shuttle vector pTRKL2 to create pCN111. The plasmid pTRKL2 was kindly provided by T. R. Klaenhammer (Department of Food Science, Southeast Dairy Foods Research Center, North Carolina State University, Raleigh, USA). The backbone of pTRKL2 is a pAMß1-derived Gram-positive vector pIL252. The copy number of pTRKL2 is corresponded to pIL252, which is 6-9 copies in a streptococcal and lactococcal host and 30-40 copies in E. coli (O’Sullivan and Klaenhammer, 1993). The plasmid pCN111 and vector pTRKL2 were electroporated into AP3 to generate the covR/S-complemented strain (SCN121) and vector control strain (SCN127).

Phosphorylation site mutation (D53A) in the covR gene was generated by overlap PCR with primers covR-F-2, covR-D53A-R, covR-R-2, and covR-D53A-F listed in Table 2. The PCR product (1.1 kb) with D53A amino acid substitution was confirmed by sequencing and ligated into temperature-sensitive E. coli-Lactococcus shuttle vector pG⁺HOST9 via the EcoRI site. The constructed plasmid was transformed into wild type A20 strain and cultured at 30°C. Transformants were transferred to 37°C to force plasmid integration via a single homologous recombination. Finally, transformants in which the plasmid excised from chromosome via a second recombination were selected in the antibiotic-free plate at 30°C. The phosphorylation site mutation was further verified by sequencing and the strain was designed as SCN128.

To construct the lacD.1 isogenic mutant, pG⁺HOST9 (erythromycin resistant) and E. coli vector pSF152 (spectinomycin resistant) were ligated into the EcoRI site. The erythromycin cassette of the constructed vector was further removed by BamHI digestion to generate plasmid pCN143. The lacD.1 gene with upstream (833 bp) and downstream (921 bp) regions was amplified by primers lacD.1-F-4 and lacD.1-R-2 and ligated into pCN143 with the BamHI site. The lacD.1 gene was removed by inverted PCR with primers lacD.1-F-3 and lacD.1-R-3 (Table 2) and replaced by a chloramphenicol cassette from Vector 78 (Tsou et al., 2010) to generate pCN145. The plasmid pCN145 was transformed into A20, AP3, and SCN128 and transformants were selected by chloramphenicol at 30°C. Transformants were transferred to 37°C with chloramphenicol selection to force plasmid integration via a single or double homologous recombination. Transformants with double homologous recombination were screened by antibiotics (spectinomycin sensitive and chloramphenicol resistant) and the deletion of lacD.1 gene was further confirmed by sequencing. Isogenic lacD.1 mutants of A20, AP3, and SCN128 were designed as SCN139, SCN140, and SCN141, respectively.

High-copy-number E. coli-Streptococcus shuttle vector pDL278 (Chiang-Ni et al., 2012) was employed to construct the rgg over-expression strain. The rgg gene with its native promoter was amplified by primers rgg-F-4 and rgg-R-3 (1958 bp) and ligated into pDL278 with the BamHI site (designed as pCN138). pCN138 was transformed into covS mutant AP3 to generate rgg over-expression strain SCN134.

Skim-milk agar was prepared by Columbia agar base (Becton, Dickinson and Company, Sparks, MD, USA) supplemented with 3% skim milk. GAS strains were sub-cultured on the Skim-milk agar and incubated at 5% CO₂, 37°C incubator for 12-16 h. Casein hydrolysis by GAS strains results in the appearance of a clear zone around bacterial colonies. For Western blot analysis, 30 μl of GAS culture supernatant was subjected to 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by transfer to the polyvinylidene difluoride membrane (Millipore, Billerica, MA, USA). The membrane was blocked with 5% skim milk in PBST buffer (PBS containing 0.2% of tween-20) at 37°C for 1 h, and SpeB protein was detected by the anti-SpeB antibody (Toxin Technology, Inc., Sarasota, FL, USA) according to the manual. After hybridization, the membrane was washed with PBST and hybridized with the secondary antibody, peroxidase conjugated goat anti-rabbit IgG (1:10,000 dilution; Cell Signaling Technology, Inc., Danvers, MA, USA) at room temperature for 1 h. The blot was developed using Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific Inc., Rockford, IL, USA) and the signal was detected by the Gel Doc XR+ system (BioRad).

Group A Streptococcus strains were cultured in TSBY broth supplemented with or without 20 mM Mg⁺⁺. Bacteria were grown to the stationary phase of growth, and bacterial pellets were washed and resuspended in the Buffer A. Bacterial cells were disrupted by the bead beater (Mini-Beadbeater, BioSpec Products Inc., Bartlesville, OK, USA) and supernatants were collected for further analysis. Phos-tag SDS-PAGE was prepared according to the manual. Briefly, 10 μg of bacterial total protein was mixed with the 4× protein dye and subjected to 10% SDS-PAGE containing 10 μmol/L of Phos-tag (Wako Pure Chemical Industries Ltd, Richmond, VA, USA) and 0.5 μM of MnCl₂ directly without boiling. Phosphorylated proteins were separated on Phos-tag SDS-PAGE for 120-140 min at 100 V at 4°C. The gel was washed with transfer buffer (39 mM of glycine, 48 mM of Tris, 0.037% of SDS, and 20% of methanol) supplemented with 1 mM of EDTA for 40 min. The gel was washed again by EDTA-free transfer buffer for another 20 min and proteins were transferred onto the PVDF membrane. The membrane was blocking by 5% skim milk in PBST buffer at 37°C for 1 h. Mouse anti-CovR serum (kindly provided by Prof. Jiunn Jong Wu) in 1:5,000 dilutions was used as the primary antibody for hybridization. After hybridization, the primary antibody was washed from the membrane with PBST buffer and hybridized with the secondary antibody, peroxidase conjugated goat anti-mouse IgG (1:10,000 dilution; Millipore, Billerica, MA, USA) at room temperature for 1 h. The blot was developed using Pierce ECL Western Blotting Substrate and the signal was detected by the Gel Doc XR+ system (BioRad).

Statistical analysis was performed by using the Prism software, version 4 (GraphPad, San Diego, CA, USA). Significant differences in multiple groups were determined using ANOVA. Post-test for AVOVA was analyzed by Tukey’s Honestly Significant Difference Test. A p-value <0.05 was taken as significant.

Although CovR protein can be autophosphorylated in the absence of CovS, inactivation of CovS results in a dramatically decreased level of phosphorylated CovR (Miller et al., 2001; Dalton and Scott, 2004; Horstmann et al., 2015). However, expression of speB is downregulated in covS mutants compared with wild type strains (Sumby et al., 2006; Walker et al., 2007; Trevino et al., 2009; Tran-Winkler et al., 2011). These results suggest that CovR acts as a repressor of speB expression, even in the absence of sensor kinase CovS. To elucidate the role of CovR in the regulation of speB expression in the presence or absence of CovS, isogenic mutants lacking functional covR were constructed from wild type A20 (SW656) and covS mutant AP3 (SW934) strains. The interruption of the covR gene by a chloramphenicol (Cm) cassette in SW656 and SW934 was further confirmed by Southern blot analysis. Results showed that the Cm cassette was only detected in SW656 and SW934 with a size of 3.3 kb, but not in their parental strains (Figure 1A). Expression of speB in the stationary phase of growth was upregulated in SW656 compared with A20 (Figure 1B). In addition, expression of speB was also upregulated in SW934 compared with AP3 (Figure 1B). Furthermore, SW934, but not AP3, hydrolyzed casein in the skim-milk agar plate (Figure 1B, lower panel). These results indicate that CovR acts as a negative regulator of speB expression in both wild type and covS mutant strains.

Phosphorylated CovR protein has a greater affinity for the speB promoter DNA than non-phosphorylated CovR (Miller et al., 2001). Therefore, we proposed that CovR does not directly repress the expression of speB in the covS mutant. Sumby et al. (2006) showed that the expression of SpeB-positive regulator Rgg in the covS mutant is significantly decreased compared with that in the wild type strain. In line with the previous observation, we also found that expression of rgg was downregulated in the covS mutant AP3 compared with that in the wild type A20 strain (Figure 2A). In addition, covR/covS trans-complemented strain SCN121, but not vector control strain SCN127, restored rgg expression in the stationary phase of growth of AP3 (Figure 2A). To elucidate the role of rgg repression in speB expression in the covS mutant, rgg was overexpressed by the high-copy number vector pDL278 in AP3 (SCN134). Expression of speB is repressed in the covS mutant; therefore, the value of the fold change in expression cannot be precisely evaluated by a relative quantification PCR method. Therefore, the expression of speB in AP3 and SCN134 was analyzed by Northern blotting. Results showed that speB transcripts were detected in SCN134 but not in AP3 (Figure 2B). Compared to the wild type A20 strain, the expression of speB in the stationary phase of growth in SCN134 was only partially restored (7 h of incubation, Figure 2B). Expression of speB is restricted to the stationary phase of growth (Unnikrishnan et al., 1999; Neely et al., 2003); however, we found that speB transcripts can be detected in SCN134 in the early- and late-exponential phases of growth (3 and 5 h of incubation, respectively, Figure 2B). SpeB protein in the bacterial culture supernatant was further detected by Western blot. SpeB is secreted as an inactive precursor (42 kDa), which is autocatalytically cleaved into an active protease (28 kDa), and produces at least six intermediates during this process (Doran et al., 1999; Collin and Olsen, 2000; Zimmerlein et al., 2005). Our results show that the mature SpeB protein and its intermediates were detected in the culture supernatant of SCN134, but not in AP3 during the stationary phase of growth (Figure 2C). These results indicate that rgg repression contributes directly to speB inhibition in the covS mutant.

Results from Figure 2A show that the expression of rgg was repressed in AP3 compared with that in the wild type A20 strain. To elucidate the role of CovR in regulation of rgg expression in the absence of CovS, the expression of rgg in AP3 and its covR isogenic mutant SW934 was analyzed. The results showed that expression of rgg during the stationary phase of growth increases significantly in SW934 compared with AP3 (Figure 3A), suggesting that CovR negatively regulates rgg expression. AP3 is not the covS isogenic mutant of A20. Although we considered that only a T nucleotide deletion in the covS in AP3 genome is important, the possibility that other unidentified mutations in AP3 contribute to the rgg repression cannot be completely excluded. Regulation by CovS is thought to be solely through CovR (Trevino et al., 2009). Therefore, to study the role of CovS inactivation in rgg expression, a wild type A20 strain was constructed with the mutation in phospho-aspartic acid residue (D53A) of CovR (SCN128) (Dalton and Scott, 2004). Expression of rgg was significantly decreased in SCN128 compared with the wild type A20 strain (Figure 3A). In addition, speB transcripts and protease activity were not detected in SCN128 (Figure 3B and data not shown). These results suggest that the nonphosphorylated CovR protein in the covS mutant acts as a transcriptional repressor of rgg expression.

Results showed in Figure 3 suggests that the decrease in the level of phosphorylated CovR results in downregulation of rgg expression. The CovR/CovS system responds to changes in Mg⁺⁺ concentrations by changing the phosphorylation status of CovR (Tran-Winkler et al., 2011; Horstmann et al., 2015). Therefore, the expression of rgg and the CovR-regulated gene ska in A20 and AP3 under Mg⁺⁺ stimulation was further analyzed. As expected, the expression of ska was decreased significantly with Mg⁺⁺ stimulation in A20 but not in AP3 (Figure 4A and data not shown). However, there was no statistical difference in the levels of rgg expression between Mg⁺⁺ treated and nontreated A20 or AP3 (Figure 4A and data not shown). Horstmann et al. (2015) showed that serotype M1 GAS strains have high levels of phosphorylated CovR in the absence of Mg⁺⁺ stimulation; therefore, differences in the expression levels of rgg in A20 with or without Mg⁺⁺ treatment could not be clearly observed. Therefore, CovR phosphorylation levels in A20 and AP3 in the presence with or without Mg⁺⁺ were further analyzed by Phos-tag western blot (Barbieri and Stock, 2008; Horstmann et al., 2015). Results showed that phosphorylated CovR protein was only clearly observed in A20 but not in AP3 (Figure 4B). In addition, a slight increase in phosphorylated CovR protein in A20 was found under Mg⁺⁺ stimulation (Figure 4B), indicating that A20 has high levels of phosphorylated CovR in the absence of Mg⁺⁺ stimulation. Horstmann et al. (2015) showed that levels of phosphorylated CovR in M1 GAS strain decreases significantly in the presence of antimicrobial peptide LL37. To further elucidate the role of phosphorylation level of CovR in the regulation of rgg expression, the wild type A20 strain was treated with LL37 and the expression of ska and rgg was analyzed. A20 strain was grown to the late-exponential phase of growth and then treated by 150 and 300 nM LL37 for another 3 h. After LL37 treatment, the expression of ska increased significantly compared with that in A20 incubated in TSBY broth without supplemented with LL37 (Figure 4C). In addition, the expression of rgg was decreased in the presence of 150 and 300 nM of LL37 (Figure 4C). Furthermore, the role of phosphorylated and non-phosphorylated CovR protein in the regulation of rgg expression was further evaluated by covR isogenic mutants of A20 and AP3. Expression of rgg was significantly increased in SW656 and SW934 compared with the parental strains (Figure 4D), indicating that CovR acts as a transcriptional repressor in both wild type and covS mutant strains. In addition, these results also suggest that non-phosphorylated CovR can more strongly repress rgg expression than phosphorylated CovR.

Expression of speB in SCN134 was only partially restored (Figure 2B), suggesting that the regulatory activity of Rgg in SCN134 could still be inhibited. Therefore, the expression of lacD.1, a negative regulator of Rgg, was analyzed. The results showed that expression of lacD.1 was significantly increased in the covS mutant AP3 compared with that in the wild type A20 strain (Figure 5). In addition, the AP3 covR/S trans-complemented strain SCN121, but not the vector control strain SCN127, showed similar lacD.1 expression levels to A20 in the stationary phase of growth (Figure 5). The role of CovS inactivation in lacD.1 upregulation was further elucidated by SCN128. Compared to the wild type A20 strain, SCN128 showed a significantly increase in lacD.1 expression. These results indicate that expression of lacD.1 is upregulated in the covS mutant.

The increase in lacD.1 expression may lead to Rgg inactivation in the covS mutant AP3 (Loughman and Caparon, 2006). Therefore, the role of LacD.1 in speB inhibition in the covS mutant was further analyzed. Isogenic lacD.1 mutant of A20, AP3, and SCN128 were constructed. The deletion of lacD.1 gene in lacD.1 mutants was confirmed by sequencing analysis (data not shown). In addition, a 2.7 kb region of the lac.1 operon of wild type and lacD.1 mutant strains was amplified by PCR with primers lacD.1-F-2 and lacD.1-R-2 (Table 2), and PCR products were further digested with the restriction enzyme BamHI. The BamHI site is located within the lacD.1 gene; therefore, as expected, only PCR products amplified from the wild type strains were digested by BamHI (Figure 6A). The lacD.1 and lacD.2 genes have 70% sequence identity; therefore, the lacD.1 mutants were further verified by Southern blot analysis. Results showed that no lacD.1 gene was detected in DNA extracted from the lacD.1 isogenic mutants of A20 and AP3 (Figure 6B). In addition, the Cm cassette was only detected in the lacD.1 mutants (Figure 6B). However, deletion of the lacD.1 gene in AP3 (SCN140) and SCN128 (SCN141) did not restore speB expression (Figure 6C). Expression of speB is induced by acidic pH but inhibited by neutral pH (Loughman and Caparon, 2006; Chiang-Ni et al., 2012). LacD.1 has been shown to play critical roles in the inhibition of speB expression under neutral pH culture conditions (Loughman and Caparon, 2006). However, this study found that speB expression was inhibited in both the wild type A20 strain and its lacD.1 isogenic mutant (SCN139) under neutral pH culture conditions (Figure 6D). In addition, the expression level of speB in SCN140 under acidic condition was still lower than that of A20 and SCN139 under pH 7.5 condition (Figure 6D). These results indicate that LacD.1 is not involved in the repression of speB expression in the covS mutant.