Escherichia coli strain DH5α was used for plasmid construction and was maintained on Luria-Bertani (LB) agar or broth (Difco) at 37°C. Wild-type S. flexneri serotype 2a strains 301 and 2457T, and serotype 5a strain M90T, were grown on tryptic soy agar (Difco) containing 0.01% (w/v) Congo red or in LB broth at 30 and 37°C. When necessary, nalidixic acid (50 μg/mL), streptomycin (50 μg/mL), or chloramphenicol (30 μg/mL) were added to the growth media.

As shown in Supplementary Figure S1, gene fragment R0139, amplified from S. flexneri strain 2457T using primers R0139-1 and R0139-2 (Supplementary Table S1), was ligated into chloramphenicol resistance-conferring pir-dependent suicide vector pXL275, generating recombinant plasmid pXL275-R0139. pXL275-R0139 was then transformed into E. coli S17-λpir, and transferred into S. flexneri strain 2457T by conjugation according to the method of Klümper et al. (2014). Following homologous recombination into plasmid pSf-R27 at the R0139 site, the resulting pSf-R27 plasmid contained the chloramphenicol resistance marker. Using helper plasmid pRK2013, the recombinant pSf-R27 plasmid was transferred into S. flexneri strain 301, as described previously (Zhang et al., 2013). Resulting transconjugants were named 301/pSf-R27, and were purified and used for further analyses.

Recombinant plasmid pAK-sfh was prepared by ligating sfh, amplified from S. flexneri strain 2457T using primers sfh-1 and sfh-2 (Supplementary Table S1), into low-copy-number vector pAK (Zhao et al., 2010). The recombinant plasmid was then transformed into electrocompetent S. flexneri 301 cells using a Bio-Rad electroporation apparatus to generate strain 301/pAK-sfh.

Bacteria were grown to stationary phase at 37°C in LB medium adjusted to pH 5.0. Aliquots (1 mL) of culture were then centrifuged for 5 min at 2,000 × g, and the resulting pellets resuspended in 1-mL volumes of LB medium adjusted to pH 2.5 or pH 5.0 (Lin et al., 1995; Yang et al., 2015). The suspensions were incubated for 30 min at 37°C with shaking at 220 rpm. Bacterial cells were then collected by centrifugation as described above, and the resulting pellets washed three times with PBS. Cells were then stained using a Cell Viability Kit (BD Biosciences) for 15 min and measured using a FACScan flow cytometer (BD Biosciences). The data were analyzed using CellQuest software. Two or three repetitions were performed for each experiment.

Preparation of whole-cell protein extracts (complex and monomer samples) and 2-DE analysis (BN-PAGE and IEF/SDS-PAGE) was performed as previously described (Zhu et al., 2010a; Niu et al., 2013).

All of the protein spots generated by 2-DE were analyzed by MALDI-TOF/TOF MS. The protein spots were carefully excised from the gel, destained using destaining solution (50% acetonitrile, 25 mM acid ammonium carbonate), and then digested for 13 h using sequencing-grade modified trypsin (Roche). Peptides from the digested proteins were used for MALDI-TOF/TOF analysis. MALDI-TOF MS was performed using an Ultraflex^III MALDI-TOF mass spectrometer (Bruker Daltonics) operating in reflectron mode with 20 kV accelerating voltage and 23 kV reflecting voltage. A saturated solution of α-cyano-4-hydroxycinnamic acid in 50% acetonitrile and 0.1% trifluoroacetic acid was used as the matrix. A 1-μL volume of a mixture of the matrix and sample solutions at a 1:1 ratio was applied to the Score384 target well. The SNAP algorithm (S/N threshold: 5; Quality Factor Threshold: 30) in FlexAnalysis v.2.4 was used to select the 150 most prominent peaks in the mass range m/z 700–4,000. The subsequent MS/MS analysis was performed in a data-dependent manner, and the 10 most abundant ions were subjected to high energy collision-induced dissociation analysis. The collision energy was set to 1 keV, and nitrogen was used as the collision gas.

To deal with one PMF and multiple TOF/TOF spectra from one sample as a single combined dataset, the raw data were first merged into one MGF file using Biotools v3.0 software, and then searched using Mascot 2.1 (Matrix Science Ltd.) against the S. flexneri 2a 2457T genome database to eliminate redundancy resulting from multiple members of the same protein family. The database contains 4,540 entries, including all of the predicted open reading frames on the chromosome of S. flexneri 2a 2457T (GenBank GI:30043918), the virulence plasmid pCP301 (GenBank GI:18462515) from S. flexneri 2a 301, and the large IncHI plasmid R27 (GenBank GI:7800243) from Salmonella Typhi. Results were checked against the NCBInr database (version 20061021, 4,072,503 sequences) to eliminate known contaminants. The search parameters used were as follows: trypsin digestion with one missed cleavage; carbamidomethyl modification of cysteine as a fixed modification and oxidation of methionine as a variable modification; peptide tolerance maximum, ±100 ppm; MS/MS tolerance maximum, ±0.6 Da; peptide charge, +1; monoisotopic mass. Scores >21 are considered significant (P < 0.05) for a local MS/MS search. For unambiguous identification of proteins, more than five peptides must be matched.

For quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) analysis, bacteria were cultured at 37°C and harvested in stationary phase. Total RNA was isolated from the cultures using RNeasy Mini Spin Columns (Qiagen) and treated with RNase-free DNase I (New England Biolabs) according to the manufacturer's instructions. cDNA was generated from 3 μg of each RNA sample using a Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific).

qRT-PCR analysis was carried out using an iCycler iQ Real-Time PCR System (BioRad) in 50-μL reaction mixtures containing 1 μL of cDNA, 600 nM of each primer, and iQ SYBR Green Supermix (BioRad) according to the manufacturer's instructions. Reactions were carried out under the following conditions: 30 s at 95°C, followed by 40 cycles of 15 s at 95°C, 30 s at the specific annealing temperature, and 30 s at 72°C. The qRT-PCR reactions were performed in triplicate for each of the three biological replicates tested. Data were analyzed using CFX Manager 2.1 (BioRad). Gene-specific primers (Supplementary Table S1) were designed using Primer Premier 5.0 software (Premier Biosoft). Relative amounts of cDNA were normalized to the amounts of 16S rRNA cDNA in each sample. Results represent the mean from at least three independent experiments.

End-biotinylated primer sets gadAp1/gadAp2 and gadBp1/gadBp2 were designed to amplify the gadA and gadB DNA regulatory regions, respectively (Supplementary Table S1). Using DNA extracted from wild-type strain 301 as a template, the biotin-labeled gadB DNA regulatory region was amplified and then coupled to Dynabeads M-280 Streptavidin (Invitrogen) according to the manufacturer's instructions. An end-biotinylated DNA fragment, which does not bind potential target proteins, was also included as a negative control. Approximately 15 mL of an overnight bacterial culture were then harvested, resuspended in 1 mL of TGED buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 10% (v/v) glycerol, 1 mM DTT, 0.01% Triton X-100) containing 100 mM NaCl, and then lysed by sonication. The DNA-coupled magnetic beads were incubated with the cell lysate for 30 min at room temperature in TGED buffer containing 100 mM NaCl, and then washed three times in the same buffer. Proteins were eluted on ice in 50 μL of TGED buffer containing 1 M NaCl. Following purification using a 2-D Clean-Up Kit (GE Healthcare), proteins were pipetted into 10 kDa ultrafiltration tubes (Amicon Ultra-0.5; Millipore) and centrifuged at 13,000 × g for 5 min. The flow-through was discarded and 0.1% (m/v) DTT in 50 mmol/L NH₄HCO₃ was added to the ultrafiltration tubes. Samples were incubated for 30 min and then centrifuged as described above. The flow-through was again discarded and 0.25% (m/v) IAA in 50 mmol/L NH₄HCO₃ was added to the tubes and incubated for 30 min in the dark. The samples were centrifuged as described above and then washed at least three times with 50 mmol/L NH₄HCO₃ (the flow-through was discarded after each centrifugation step).

The resulting proteins were digested with trypsin (Roche, 1:50, w/w) for 16 h at 37°C, and flow-through peptides were collected in a new tube for liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis. Peptides were separated on a 25-cm C18 nano column using EASY-nLC (Thermo Fisher Scientific). Samples were loaded and then eluted for 30 min using a 4–90% ACN fraction-optimized non-linear gradient in 0.1% formic acid. Eluted peptides were detected using an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific), and then semi-quantified using a spectral counting approach as described previously (Gao et al., 2015). Each sample was independently analyzed three times, and all spectra were compared against the S. flexneri 2a 2457T database described above using PD v.2.1 software (Thermo Fisher Scientific).

His-tagged HNS and Sfh proteins were expressed in E. coli BL21 (DE3) cells using the pET28a vector. The corresponding gene regions were amplified from 2457T using primers hnsp1/hnsp2 and sfhp1/sfhp2, respectively. gadA (350 bp) and gadB (450 bp) DNA regulatory region fragments were amplified by PCR using 5′-end-biotinylated primers. All primer sequences are listed in Supplementary Table S1. The PCR fragment was gel-purified twice using a High Pure PCR Product Purification Kit (Roche). The EMSA assay was conducted using a LightShift Chemiluminescent EMSA Kit and a Chemiluminescent Nucleic Acid Detection Module (Thermo Fisher Scientific) as per the manufacturer's instructions.

The SDS polyacrylamide gels were transferred to PVDF membranes at 15 V for 1.5 h. The resulting PVDF membranes were blocked with 10% (w/v) skim milk powder in TBS (100 mmol/l Tris-HCl, pH 7.5, 0.9% (w/v) NaCl) containing 0.1% (v/v) Tween 20 (TBST) for 1 h. Membranes were incubated with anti-GadB antibody (Abmart Corp.) diluted in TBST for 1–2 h at room temperature or at 4°C overnight at the recommended concentration, followed by detection using ECL reagents (Thermo Fisher Scientific) and manual film development.

To examine the molecular mechanism of acid resistance in S. flexneri, we compared the acid resistance of three different strains. Strains 2457T and 301 were identified as serotype 2a and showed ~98% genome homology, while M90T is a serotype 5a strain. The stress tolerance of the strains was examined using survival assays, which showed that the survival rate of 2457T was obviously lower than those of the other two S. flexneri strains at pH 2.5 (Figure 1A).

Because the glutamic acid decarboxylase system is the most effective acid-tolerance pathway in Shigella, we examined the expression levels of soluble protein complexes among the three strains. Strains were grown aerobically in 100 mL of LB medium at 37°C, and then the protein complexes were separated and analyzed by BN-PAGE 2-DE. Compared with strains 301 and M90T, the abundance of the GadA/B complex in strain 2457T was significantly reduced (Figures 1B–D,H). IEF/SDS-PAGE was then performed to analyze the expression levels of protein monomers. Whole cell protein extracts were separated using IPG strips with a linear gradient of pH 4–7 in the first dimension, and results showed a corresponding decrease in the levels of GadA and GadB in strain 2457T (Figures 1E–G,I).

Plasmid pSf-R27 is present in strain 2457T but not in strain 301 (Supplementary Figure S2) (Wei et al., 2003). To determine whether pSf-R27 is responsible for the differences in GadA/B expression between the two strains, pSf-R27 was transformed into strain 301. As expected, IEF/SDS-PAGE showed that levels of GadA and GadB expression were significantly decreased in recombinant strain 301/pSf-R27 (Figures 2A–C), suggesting that pSf-R27 inhibits the expression of GadA and GadB. To verify the proteomic data, the effects of pSf-R27 on gadA and gadB mRNA levels were quantitated using qRT-PCR analysis. Consistent with the protein expression data, pSf-R27 reduced the levels of gadA and gadB mRNA in strain 301 (Figure 2D).

To clarify the mechanism of pSf-R27 regulation of GadA/B expression, the region upstream of gadA/B was amplified using biotinylated primers and then linked to streptavidin-coated magnetic beads to capture interacting proteins in S. flexneri strains 2457T and 301. Captured proteins were identified by LC-MS/MS. Information regarding the 10 most abundant proteins is provided in Table 1. According to the MS data, three H-NS family members (Sfh, H-NS, and StpA) were identified. Protein Sfh, the coding sequence for which is located on plasmid pSf-R27, was present at a higher abundance (semi-quantified by PSM numbers) in strain 2457T than the well-known regulator StpA, which belongs to the H-NS protein family (Zhang and Belfort, 1992). StpA also has a similar structure and function to H-NS (Sonnenfield et al., 2001; Deighan et al., 2003), which is a selective binding protein that represses gadA/B transcription and expression in E. coli (Heath and Rock, 1996). Identification of H-NS binding to the gadA/B promoter in S. flexneri confirmed the applicability of our experiments.

We further examined the effects of Sfh binding to the gadA/B regulatory region in vitro by electrophoretic mobility shift assays. The results indicated that the gadA/B sequence was shifted by H-NS and Sfh (Figure 3), which is in keeping with the proposed binding of Sfh upstream of the gadA/B sequence.

Because Sfh is located on plasmid pSf-R27 and was shown to bind to the regulatory region of gadA/B, we examined whether Sfh regulates the expression of GadA/B using IEF/SDS-PAGE analysis. As expected, the expression of GadA/B was reduced in strain 301/pAK-sfh (strain 301 expressing recombinant plasmid pAK-sfh) (Figures 4A–C). Moreover, qRT-PCR analysis indicated that Sfh inhibits the expression of gadA/B at the transcriptional level (Figure 4D).

Because GadA and GadB play important roles in acid resistance, we examined whether plasmid pSf-R27, and Sfh in particular, regulates acid tolerance in S. flexneri. Survival of strains 301 and 2457T was examined following growth at pH 2.5 and compared with that at pH 5.0 using flow cytometry. As expected, the acid tolerance of strain 301 was significantly higher than that of strain 2457T. Further, the presence of plasmid pSf-R27 and the expression of Sfh reduced the acid tolerance of strain 301 (Figures 5A,B).

We also examined the expression of GadB in each strain using western blot analysis, and the results were mainly consistent with those generated by IEF/SDS-PAGE analysis (Figure 5C). The expression of GadB in strain 2457T was significantly lower than that in strain 301, and the presence of plasmid pSf-R27 or expression of Sfh in strain 301 reduced the detectable levels of GadB. These results indicated that the acid tolerance of S. flexneri is closely related to the expression levels of GadA/B.

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.