Streptococcus mutans UA159 cells were cultured in BHI broth (Difco, Sparks, MD, United States) at 37°C under anaerobic conditions (90% N₂, 5% CO₂, and 5% H₂) and 1% sucrose was included in the medium for biofilm formation. E. coli strains were grown in LB medium (Difco, Sparks, MD, United States). Antibiotics and other supplements were used in the following final concentrations: Spe 800 μg/ml; kanamycin 50–100 μg/ml. The plasmid and primers used were provided in Supplementary Table S1. The CSP was synthesized by Life Invitrogen (Shanghai, China).

The S. mutans lmrB gene was amplified from genomic DNA using the polymerase chain reaction (PCR) system (TaKaRa) with appropriate primers (Supplementary Table S1). The amplicon was cloned in the pET28a expression vector to create the construct pET 28a-lmrB with His-tag. The resulting recombinant pET28a-lmrB with His-tag was sequenced for confirmation and transformed into the E. coli BL21 (DE3) cells to assess the in vivo role of LmrB in E. coli.

Localization of LmrB in E. coli was determined as previously described with modifications (Yeh et al., 2013). In brief, E. coli BL21 (DE3) cells harboring recombinant plasmid were cultured at 16°C with the addition of 0.3 mM isopropyl-β-D-thiogalactopyranoside to induce protein expression. After a further incubation for 4 h, the cells was collected, and then broken by sonication and ultracentrifuged at 100,000 × g for 60 min. The resulting pellet was re-suspended in buffer [10 mM Tris-HCl (pH 7.3), 5 mM MgCl₂, and 2% N-Lauroyl Sarcosine Sodium] and ultracentrifuged again. The total protein (the sonicated cell suspension) and the membrane protein (pellet from the second ultracentrifugation) were separated by SDS–PAGE and probed with an anti-His antibody. The proteins of E. coli cells transformed with pET28a were used as control.

Standard EtBr efflux assay was used to study the efflux activity of E. coli cells with pET28a-lmrB (Rodrigues et al., 2008). Cells were grown to an OD_{600 nm} of 0.6 and 200 μL of cells was loaded into 96-well black plates and mixed with EtBr for 30 min for EtBr accumulation. The EtBr-loaded cells were washed and re-suspended in PBS with or without the efflux pump blocker, carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (100 μM), added at 10 min (Bansal et al., 2016). The fluorescence was determined by a microplate reader (Promega GloMax, United States) at excitation and emission wavelengths of 525 and 605 nm. The obtained raw data were normalized to the EtBr-loaded cells, which was regarded as relative fluorescence equivalent to 1. In all the cases, E. coli BL21 (DE3) with pET28a vector was used as blank control.

Susceptibilities of E. coli BL21 (DE3) containing pET28a-lmrB against different groups of antibiotics including ampicillin, chlorhexidine (CHX), ofloxacin, and chemical agent like EtBr were carried out in 96-well polystyrene plates using a microdilution method as described previously (Xu et al., 2010). Bacterial growth was monitored after 24 h of incubation at 37°C and the MIC was defined as the lowest concentration of the antimicrobials that inhibited visible growth.

The amino acid sequence of LmrB was obtained from National Center for Biotechnology Information (NCBI) (NP_721163.1). The number of transmembrane domains was predicted by TMHMM server¹, and the secondary structure of the protein was determined using PREDICT PROTEIN (Rost et al., 2004). In addition, a 3D model of LmrB was built (E. coli YajR transporter was used as template) by SWIISS-MODEL using standard procedure and was analyzed.

The genomic nucleotide sequence encoding lmrB was obtained from NCBI GenBank (lmrB locus tag, SMU_745). Two 480 bp fragments containing regions of DNA upstream and downstream of lmrB were amplified from a genomic DNA template using the primers lmrB LF and lmrB RF listed in Table 1. The left flank region was digested with Xho I and Hind III, then ligated to MCS I of pFW5, which was digested similarly, generating plasmid pFW5:LF. The right flank region was then inserted into the MCS II of pFW5:LF using Pst I and Spe I, creating plasmid pFW5-LF:RF. The pFW5-LF:RF gene replacement vector was transformed into S. mutans UA159 in the presence of CSP. Transformants were selected on BHI agar plate containing 800 μg/ml Spe and confirmed by PCR and sequencing. Complementation construct containing the lmrB ORFs with the putative promoters were subcloned into the MCSI sites of pFW 5 to create pFW 5-lmrBcom. The construct was naturally transformed into lmrB mutant. Transformants without desired antibiotic resistance profiles were screened for complementation by PCR. The confirmed complemented strains were named pFW 5-lmrBcom.

To assess the effect of lmrB deletion on competence development, we examined the transformation of S. mutans upon lmrB deletion using the plasmid PVA838 (Moye et al., 2016). In brief, S. mutans UA159 and its derivatives grown to mid-logarithmic phase (OD₆₀₀ = 0.5) were diluted 1:100 in fresh BHI medium. Cultures were grown to an OD₆₀₀ of 0.1, and 1 μM of CSP and 100 ng of the plasmid PVA838 were added to cultures. After 3 h of growth, cultures were diluted and plated on BHI agar and on BHI agar containing erythromycin (10 μg/ml). After 24–48 h of incubation at 37°C in a 5% CO₂ aerobic atmosphere, CFUs were enumerated. Transformation efficiency is expressed as the number of erythromycin resistant colonies divided by the total number of viable colonies multiplied by 100.

The IPS content of bacterial cultures was determined as described (Busuioc et al., 2009). In brief, bacterial cultures were heated for 5 min at 100°C and bacteria were collected by centrifugation at 4,000 × g for 10 min, followed with washing twice with ice-cold water. Bacterial suspension was used for dry weight and IPS determination. The IPS was extracted with 5.3 M KOH and quantified using iodine solution [0.2% (w/v) iodine in 2.0% (w/v) potassium iodide solution] and glycogen was used as standard. The amount of IPS was normalized to the dry weight of bacteria.

The ability of S. mutans strains to undergo glycolysis in an increasingly acidic environment was evaluated by pH drop experiments with slight modifications to the previously published method (Cheng et al., 2014). S. mutans strains grown to mid-logarithmic phase (OD₆₀₀ = 0.5) were harvested by centrifugation (4,000 × g, 4°C, 10 min), washed with 0.5 mM potassium phosphate buffer containing 37.5 mM KCl and 1.25 mM MgCl₂ (pH 7.0) and re-suspended in 10% culture volume of the same solution. Glycolysis was initiated by the addition of glucose to give a final concentration of 1% (wt/vol). The resulting decrease in pH, as a result of S. mutans glycolytic activity, was monitored (Corning pH meter 240; Corning, NY, United States). To study differences in sugar storage, changes in pH were also monitored when cells were re-suspended in salt solution without addition of glucose.

Streptococcus mutans strains grown to mid-logarithmic phase (OD₆₀₀ = 0.5) were harvested by centrifugation and re-suspended in 5 ml 0.1 M glycine buffer (pH 7.0). Hydrogen peroxide (H₂O₂, Fisher Scientific, Fair Lawn, NJ, United States) was then added to give a final concentration of 0.02% (vol/vol) for oxidative stress tolerance assays. Catalase (10–25 kUnits/ml; Sigma) was added to inactivate the H₂O₂ after incubation for 15, 30, or 45 min. The culture was serially diluted in 0.1 M glycine buffer (pH 7.0), and the survival rate was determined by plating in triplicate on BHI plates (Cheng et al., 2016).

To further confirm the role of LmrB in oxidative stress tolerance, S. mutans was collected for RNA extraction after 2 h exposure to 0.02% H₂O₂ and cDNA synthesis was performed with a SuperScript cDNA Synthesis kit (Invitrogen). The lmrB expression under oxidative conditions was determined using qRT-PCR. 16S rRNA was used as the internal control with the following reaction program: 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 30 s. Threshold cycle (CT) values were determined and data were analyzed with the 2^-ΔΔCT method.

Streptococcus mutans UA159 strains grown to mid-logarithmic phase (OD₆₀₀ = 0.5) were diluted 1:100 in BHI containing 1% sucrose. For biofilm formation, aliquots (3 ml) of S. mutans strains in BHI-1% sucrose were placed into a 24-well plate containing polystyrene blocks and cultured at 37°C anaerobically for 24 h allowing biofilm formation. To determine the biomass and analyze structure alteration, SYTO 9 (Molecular Probes, Invitrogen, Carlsbad, CA, United States) was employed to label bacterial cells. As the main component of biofilm, exopolysaccharides were labeled with Alexa Fluor 647-labeled dextran conjugate (Molecular Probes) as previously described (Koo et al., 2010). Images were obtained with CLSM (LSM 710, Carl Zeiss, Germany) with a 20× objective and image series were generated by optical sectioning at each position. The three dimensional re-construction of biofilms was performed and the biomass of EPS and bacteria was analyzed using COMSTAT software.

The antimicrobial susceptibility of S. mutans strains was determined by the microdilution method as described previously (Xu et al., 2010). Briefly, 0.1 ml of S. mutans cells (OD₆₀₀ = 0.6) was added into 96-well plate that contained 0.1 ml of various antibiotics (CHX, penicillin, and erythromycin) at concentrations prepared from twofold serial dilutions. After 12 h of anaerobic growth at 37°C, bacteria were plated onto BHI agar plates and observed after overnight growth.

For biofilm formation abilities of S. mutans strains under antimicrobial pressure, S. mutans cells (OD₆₀₀ = 0.6) were diluted 1:100 in fresh BHI-1% sucrose with sub-inhibitory concentrations (sub-MIC) (0.25xMIC) of chlorhexidine for biofilm formation in flat-bottom 96-well microtiter plates (Corning Inc.) After 12 and 24 h of incubation, the biofilms were quantified using crystal violet staining.

RNA samples from the wild type and the lmrB mutant were extracted as follows: cells were grown to mid-logarithmic phase (OD₆₀₀ = 0.5) and collected by centrifugation. The obtained S. mutans cells were suspended in lysis buffer (30 mM Tris-HCl, 1 mM EDTA, 20 mg/mL lysozyme, [pH8.0]) and incubated at 37°C with gentle agitation for 30 min. Total RNA was purified using RNeasy Mini kits (Qiagen) according to the manufacturer’s instructions. Extracted RNA with purity (A260/A280) more than 2.0 was subjected to microarray preparation based on the manufacturer’s standard protocols. Briefly, 1 μg of total RNA from each sample was amplified and transcribed into fluorescent cRNA. The labeled cRNAs were hybridized onto the Whole Genome Oligo Array (4x44K, Agilent Technologies) and the arrays were scanned by the Agilent Scanner G2505B. Subsequent data processing was performed using the GeneSpring GX v11.5.1 software package (Agilent Technologies). Microarray data have been deposited at NCBI-GEO (accession no. GSE98374). DEGs with statistical significance were analyzed for protein–protein interactions (PPIs) through Cytoscape software (Institute of Systems Biology, Seattle, WA, United States) and functionally annotated through GO and Database for Annotation, Visualization and Integrated Discovery (DAVID) version 6.7. The expression of selected genes was validated by qRT-PCR with gene-specific primers shown in Supplementary Table S1.

Data analyses were performed using SPSS 13.0 and an exploratory data analysis was performed to check the assumptions of the equality of the variances and the normal distribution of errors. The data were analyzed using ANOVA, followed by Tukey’s multiple comparison tests and considered statistically significant at P < 0.05.

lmrB encoded a putative ATP-type multidrug resistance transporter (454 aa). The putative gene product was membrane associated, with 13 predicted transmembrane domains and the N-terminal region contained a MFS domain and a sugar transport domain (Figure 1A). To make sure that this protein could be expressed in correct form and could successfully locate on the membrane of E. coli cells, LmrB with His-tag was expressed in E. coli using pET28a. The total protein and the membrane protein of E. coli with and without isopropyl-β-D-thiogalactopyran -oside induction were isolated and probed with anti-His antibody. Using Western-blot analysis, bands corresponding to the expected size were identified in the total protein (Figure 1B, lanes 3 and 5) and membrane protein (Figure 1B, lane 2) of E. coli transformed with pET28a–lmrB-His-tag, while no band with the expected size was observed in the total protein (Figure 1B, lane 1) and membrane protein (Figure 1B, lane 4) of control group (E. coli cells transformed with pET28a). Notably, the band in the total protein was darker than that in the membrane protein, which suggested that only part of the recombinant protein was integrated into the E. coli cell membrane. The Western-blot analysis further confirmed that the recombinant protein was integrated into the membrane of E. coli (Figure 1C)

The structure topology of LmrB resembled E. coli YajR transporter, which belonged to MFS transporter (Figure 1D). Secondary structure analysis revealed 75.33% alpha-helix, 1.76% beta-sheet, and 22.91% loop in LmrB.

To assess whether LmrB possessed efflux activity, we examined the EtBr efflux activity of E. coli upon in trans expression of LmrB. As demonstrated in Figure 1E, the efflux activity of E. coli was higher for the bacterial culture harboring cloned pET28a-lmrB characterized by the low fluorescence intensity. And the EtBr efflux was suppressed with the addition of the efflux pump blocker CCCP. The results suggested that LmrB possessed efflux activity.

To confirm the function of LmrB as a potential multidrug efflux pump, we functionally expressed the candidate multidrug efflux pump in E. coli and examined the changes in the antimicrobials sensitivity. As listed in Table 1, a rise in the MIC values of CHX and ofloxacin was observed upon ectopic expression of LmrB in E. coli as compared with the control, with fourfold and eightfold increase in MIC, respectively. An increase in the MIC of EtBr was also noted. These results indicated that upon expression of LmrB, a multidrug resistance nature was conferred to E. coli cells. To further prove that the resistance displayed was due to efflux pump activity, we assessed the sensitivity of E. coli against ampicillin, an antibiotic targeted extracellularly. As expect, no change in MIC against ampicillin was observed in E. coli upon expression of LmrB.

In order to investigate the physiological function of LmrB in S. mutans, we constructed lmrB mutant and pFW5-lmrB complemented strains. The mutant and complementary strains were verified by PCR (Supplementary Figure S1). In view of the roles IPS played in supporting S. mutans persistence in batch cultures, we assessed the effect of lmrB inactivation on S. mutans capacity to store IPS. As demonstrated in Figure 2A, lmrB mutant have significantly higher amounts of IPS compared to the wild type. IPS utilization has been demonstrated to prolong acid production (Busuioc et al., 2009). We monitored the pH change with and without the addition of exogenous glucose to determine the ability of the lmrB mutant to reduce the extracellular pH through glycolysis. With the addition of glucose to cells re-suspended in buffer, no difference in kinetics and final pH was observed in the lmrB mutant and the wild type (data not shown). When no exogenous glucose was added, any drop in the extracellular pH could be attributed to IPS utilization. The lmrB mutant was found to be able to reduce the pH to lower values compared with the wild type in the absence of exogenous glucose (Figure 2B).

In order to identify whether LmrB was involved in the environment stress response, we measured the expression levels of lmrB under stress condition. It was found that the expression level of lmrB was enhanced by 4.23-fold when exposed to oxidative stress (p < 0.05) (Figure 3A). The involvement of LmrB in oxidative stress response was further supported by the H₂O₂ sensitivity assay, which showed significantly increased survival rate in lmrB mutant compared with the parental and complemented strains (p < 0.05) (Figure 3B). The result indicated that lmrB inactivation resulted in enhanced resistance to H₂O₂ in S. mutans.

The transformation efficiency of lmrB mutant was assessed without the addition of exogenous CSP. In the absence of CSP, we were able to obtain almost the same transformant colonies in the lmrB mutant and S. mutans (p > 0.05). Supplementary Figure S2 showed a typical example of the number of transformants obtained for each strain at the same serial dilution.

Since efflux pump was reported to closely associate with biofilm formation in many bacterial species, we explored the role of lmrB in biofilm formation of S. mutans using CLSM. It was found that the biofilm formed by lmrB mutant was significantly thicker (58.0 μm) than those formed by its wild type strain (34 μm) and the complemented strain (32 μm) (p < 0.05) (Figure 4A). Moreover, the EPS/bacteria ratio on the surface (thickness = 1–5 μm) and at the bottom (height = 47–50 μm) was significantly higher than that of the biofilms formed by the wild type and complemented strains (p < 0.05) (Figure 4B).

As previously reported, the inactivation of efflux pump might lead to multidrug hypersusceptibility. In order to explore how lmrB affected S. mutans antimicrobial sensitivity, we examined the susceptibility of the mutant to different species of antimicrobials. Unexpectedly, the lmrB mutant strain showed an overall decreased sensitivity to the treatment of different antimicrobials, especially the antimicrobial targeting bacterial cell wall (Figure 5A).

The abilities of S. mutans UA159 and the lmrB mutant to form biofilm under antimicrobial pressure were also examined. It was found that addition of 0.25xMIC of CHX to the medium differentially affected the ability of the lmrB mutant and the wild type to form biofilm in a time-dependent manner. As determined by the crystal violet assay, 0.25xMIC of CHX caused 26.3% reduction in 12 h biofilm of lmrB mutant (p > 0.05). For the wild type, the 12 h biofilm formation decreased 53.2% compared with that without CHX exposure (Figure 5B) (p < 0.05). However, after 24 h culture, the biomass of both the mutant and S. mutans UA159 biofilms returned to those of without CHX exposure.

To gain better insight into the global role of LmrB and how it relates to physiology and virulence in S. mutans, global gene expression was performed by comparative DNA microarray analysis. As shown in Supplementary Tables S2, S3, 107 genes were detected as differentially expressed, among which 66 genes were up-regulated and 41 genes were down-regulated. Based on the NCBI S. mutans genome annotation, 58 genes had assigned function whereas 49 genes were hypothetical proteins.

In order to gain insights into the biological significance of LmrB in S. mutans, we predicted and analyzed interactions among the identified DEGs. The DEGs were clustered in four interaction networks, including metabolic process, organelle part, transportation of carbohydrate and specific substrate and stress response (Figure 6). Among them, the three identified networks, cellular protein metabolic process, transportation of carbohydrate and specific substrate and stress response, were closely intertwined. Gene annotation enrichment and functional annotation clustering analysis were then conducted based on the identified protein networks. As shown in Figure 7, 20% and 14.23% of the genes related to response to stress and transmembrane transporter activity were affected by the lmrB inactivation. Moreover, starch and sucrose metabolism KEGG pathway was identified. Notably, lmrB deletion induced the expression of a set of functionally associated genes, like groES-groEL molecular chaperones, glgB-glgC-glgD-glgA-glgP, two-component system (TCS) scnRK and msmREFGK.

To validate the reliability of microarray data, we examined the expression levels of randomly selected genes using qRT-PCR, including up-regulated, down-regulated, and unchanged genes. As shown in Supplementary Figure S3, all genes selected for validation showed a similar expression trend in the qRT-PCR and microarray analysis. These results suggested a good concordance between microarray and qRT-PCR data.