Bacterial strains and plasmids used in this work are listed in Table 1. Mutant strains of C. violaceum were derived from the wild-type strain ATCC 12472 (Vasconcelos et al., 2003). All bacterial strains of C. violaceum and E. coli were grown at 37°C in Luria-Bertani (LB) medium. Antibiotic susceptibility assays were performed on Mueller-Hinton (MH) medium (Sigma). For plasmid selection in general cloning procedures, LB medium was supplemented with the antibiotics tetracycline (10 μg/ml), kanamycin (50 μg/ml), and ampicillin (100 μg/ml). Oligonucleotide sequences are listed in Supplementary Table S1.

The in-frame emrR, emrCAB, emrRCAB, and cviI gene deletions were constructed by a two-step allelic exchange procedure with the vector pNPTS138 as previously described (da Silva Neto et al., 2012). The resulting mutant strains ΔemrR, ΔemrCAB, ΔemrRCAB, and ΔcviI were confirmed by PCR using specific primers (Supplementary Table S1). For complementation of emrR mutants, a DNA fragment containing the full emrR gene was cloned into the vector pMR20 and this construct was introduced into the mutant strains by conjugation (da Silva Neto et al., 2012).

Selection was performed by plating 500 μl of an LB-grown overnight culture of C. violaceum ATCC 12472 on LB plates with increased concentrations of nalidixic acid (0.5–64 μg/ml; 1–7 times the MIC). Several spontaneous resistant colonies that appeared on these plates after incubation at 37°C for 24 h were cultivated in LB without antibiotic. After testing these colonies for nalidixic acid susceptibility (see item MIC on LB plates), resistant colonies (4 μg/ml to 512 μg/ml) were selected for sequencing the full emrR and the QRDR region of gyrA, using specific primers (Supplementary Table S1). For Sanger DNA sequencing, the amplicons of emrR and gyrA were obtained by colony PCR and sequenced in both strands using BigDye Terminator V3.1 (Applied Biosystems).

Disk diffusion assays were performed as recommended by the (Clinical and Laboratory Standards Institute [CLSI], 2013). Briefly, C. violaceum strains, grown on MH plates for 20 h, were resuspended in sterile saline and adjusted to 0.5 MacFarland turbidity standard. These suspensions were seeded onto MH plates using a sterile swab. On the surface of seeded plates were placed disks impregnated with 24 antibiotics (BD BBL^TM Sensi-Disc^TM Antimicrobial Susceptibility Test Discs) (Supplementary Table S2). The growth inhibition halos were recorded after 24 h incubation at 37°C. Disk diffusion assays were performed at least in biological triplicate.

MICs of nalidixic acid, kanamycin, streptomycin, tetracycline, doxycycline, erythromycin, chloramphenicol, and cefotaxime for each C. violaceum strain were determined by a broth macrodilution method according to CLSI guidelines (Clinical and Laboratory Standards Institute [CLSI], 2013). The MIC values were determined as the lowest concentration of the antibiotic that inhibited visible bacterial growth after incubation of the cultures in MH broth for 24 h at 37°C under shaking.

The strains were grown on LB plates for 24 h, and streaked on LB plates with increasing concentrations of nalidixic acid (0.5, 1.0, and 2.0 μg/ml, for emrRCAB mutants). The susceptibility profiles of the spontaneous nalidixic acid-resistant mutants were determined using this method (1 μg/ml to 512 μg/ml of nalidixic acid). After 24 h incubation at 37°C, the bacterial growth was recorded.

The presence of N-acyl-L-homoserine lactones (AHLs) in the culture supernatant of C. violaceum strains was evaluated by the production of violacein in the biosensor strain ΔcviI, using an agar plate assay, as previously described (Morohoshi et al., 2008). Briefly, filtered supernatants (80 μl) of overnight cultures were added inside wells sunken on LB agar plates soaked with the ΔcviI strain. After 48 h incubation at 25°C, the appearance of a violacein halo was recorded on the plates. Biofilm formation was assayed on polypropylene tubes with the strains cultured in LB medium for 16 h under static conditions, using the crystal violet method (Azeredo et al., 2017).

Chromobacterium violaceum strains were grown at 37°C in LB until mid-log phase (OD₆₀₀ of 0.8–1.0). After cell harvesting by centrifugation, the total RNA was extracted with TRIzol reagent (Ambion), and purified with the illustra RNAspin Mini RNA isolation kit (GE Healthcare), which includes a DNase treatment step. To test emrCAB-inducing conditions by Northern blot, the wild-type strain ATCC 12472 was grown at 37°C in LB until OD₆₀₀ of 0.8. Then, the culture was split into four aliquots and either left untreated or treated for 10 min with salicylate (0.1, 1, and 10 mM), nalidixic acid (0.1, 0.2, and 0.5 mM), or ethidium bromide (0.1, 0.2, and 0.5 mM), and the cells were used for RNA extraction. RNA concentration was determined with a NanoDrop spectrophotometer (Thermo Scientific) and RNA integrity was checked by using formaldehyde-denaturing agarose gels.

A detailed description of the custom-designed oligonucleotide microarray slides (Agilent Technologies) was published previously (Previato-Mello et al., 2017). All procedures for cRNA labeling, hybridization, and washing of the slides as well as data acquisition, extraction, and normalization were performed exactly as described (Previato-Mello et al., 2017), and following manufacturer’s instructions (Agilent Technologies). Data sets included three independent biological experiments with RNA extracted from C. violaceum ATCC 12472 and ΔemrR strains grown in LB at 37°C until mid-log phase as stated above. Differentially expressed genes were those that had their expression levels altered at least two-fold (ΔemrR/WT).

Microarray raw data have been deposited in the Gene Expression Omnibus (GEO) database¹ with accession number GSE112521.

Samples of total RNA (7 μg) extracted as stated above were used for Northern blot assays as previously described (da Silva Neto et al., 2012; Previato-Mello et al., 2017). Specific probes for each indicated gene were amplified by PCR (primers listed in Supplementary Table S1), and labeled with [α-³²P]dCTP (PerkinElmer) by using an Exo-Klenow enzyme DECAprime II kit (Ambion). After membrane hybridization in ULTRAhyb buffer (Ambion) and washing, the signal was detected by autoradiography.

The coding region of the C. violaceum emrR gene (CV_0769) was amplified by PCR using specific primers (Supplementary Table S1), and cloned into the vector pET15b as a 501-pb NdeI/BamHI DNA fragment. The recombinant His-EmrR protein was produced in E. coli BL21(DE3) after induction with 1 mM IPTG, and purified by NTA-resin affinity chromatography (Qiagen) as previously described (da Silva Neto et al., 2012).

Proteins from total extracts of C. violaceum strains were separated on 15% SDS-PAGE gels, and transferred onto a nitrocellulose membrane (Amershan Protran). Membranes were blocked and incubated with a 1:1,000 dilution of anti-EmrR mouse polyclonal antiserum. After membrane incubation with a secondary anti-mouse IgG conjugated to peroxidase, detection was performed using the LumiGLO western blotting protein detector kit as recommended by the manufacturer (KPL). The polyclonal anti-EmrR antibody was developed in mice according to an experimental protocol approved by the Local Ethical Animal Committee (CEUA) of FMRP-USP (protocol number 147/2014).

DNA probes corresponding to promoter regions of indicated genes were amplified by PCR from the ATCC 12472 genome, using specific primers (Supplementary Table S1). These DNA fragments were end labeled with [γ-³²P]ATP (PerkinElmer) by using T4 polynucleotide kinase (Thermo Scientific) and purified with the Wizard^® SV gel and PCR clean-up system (Promega). The DNA binding reactions containing the DNA probes and different amounts of DTT-reduced His-EmrR were performed in an interaction buffer, as previously described (da Silva Neto et al., 2012; Previato-Mello et al., 2017).

In the genome of C. violaceum ATCC 12472 there are at least 15 genes encoding MarR family transcription factors (Vasconcelos et al., 2003), but only ohrR has been characterized (da Silva Neto et al., 2012; Previato-Mello et al., 2017). Analysis of sequence alignment and genomic position indicated that CV_0769 and the nearby genes CV_0768, CV_0767, and CV_0766 resemble the emrR, emrCAB cluster (Figure 1A) described in other bacteria, which encode the MarR family transcription factor EmrR and the MFS-type efflux pump EmrCAB (Lomovskaya and Lewis, 1992; Huang et al., 2013). To define the role of EmrR in antibiotic resistance, we constructed and characterized an emrR null mutant strain by antimicrobial susceptibility assays. Antibiogram tests with 24 antibiotics (Supplementary Figure S1) and determination of MIC with eight antibiotics (Supplementary Table S3) revealed that the ΔemrR mutant showed increased resistance specifically to nalidixic acid, a quinolone (a 13-mm decrease in the halo and a fourfold increase in the MIC). After complementation of the ΔemrR mutant (Figure 1B), this increased resistance to nalidixic acid was reverted, as determined by disk diffusion (Figure 2A), and MIC (Table 2). A phenotype of decreased violacein production, verified for the ΔemrR mutant in LB liquid cultures, was also complemented (Figure 2B). These results indicate that EmrR controls antibiotic resistance and pigment production in C. violaceum.

To evaluate whether the emergence of nalidixic acid-resistant mutants can occur by point mutation in emrR, we isolated spontaneous mutants from C. violaceum wild-type cultivated in LB with increasing concentrations of nalidixic acid (0.5–64 μg/ml), and determined the MIC of these mutants (Table 3). DNA sequencing revealed that three isolates with MIC of 4 μg/ml had two types of mutation in emrR (substitution and deletion), while for the 25 remaining isolates sequenced the mutations occurred in gyrA, a hotspot gene for mutations that confer quinolone resistance (Giles et al., 2004). We selected two spontaneous mutants, emrR_R92H and gyrA_T85I, for further characterization (Figure 3). In emrR_R92H, histidine replaced arginine at position 92 of EmrR. As this substitution arose in the second (position 91–106) of a four-element fingerprint signature for the MarR family predicted in EmrR (Figure 3A) and the levels of the protein EmrR were increased in the emrR_R92H strain probably by loss of self-repression of the regulator (Figure 1B), we suggest that this point mutation affects the DNA binding activity of EmrR. In fact, emrR_R92H displayed the same antibiotic resistance profile observed for the null mutant ΔemrR, namely increased resistance to nalidixic acid, but not to other quinolones (Figures 3A,B). Although all point mutations in gyrA mapped within the quinolone resistance determining region (QRDR) of GyrA (position 67–106), the substitution of threonine for isoleucine at position 85 of GyrA had particular impact on antibiotic resistance, as the gyrA_T85I mutant showed the highest level of resistance to nalidixic acid, and presented cross resistance to other quinolones, such as ciprofloxacin, norfloxacin, and levofloxacin (Table 3 and Figure 3C). These results indicate that point mutations in EmrR and GyrA are related to the emergence of quinolone resistance in C. violaceum.

To identify the full repertoire of genes regulated by EmrR, we compared, by DNA microarray analysis, the transcriptome of the wild-type and ΔemrR mutant strains grown in LB medium, at log phase. From this analysis, 22 genes were upregulated and 14 genes were downregulated in the ΔemrR mutant compared to wild-type strain (Supplementary Table S4), indicating that EmrR acts mostly as a transcription repressor. The EmrR-repressed genes discussed in the text are shown in Figure 4A, and most of them were validated by Northern blot (Figure 4B) and EMSA (see the next item). Among the genes upregulated in the ΔemrR mutant, CV_0766 (emrB), CV_0767 (emrA), and CV_0768 (emrC) showed more than ten-fold increase in expression in the mutant. These genes compose the operon emrCAB, which encodes the putative MFS-type efflux pump EmrCAB (Figure 1A). However, other genes encoding putative transport proteins were upregulated (pcaK, CV_1769, crcB, CV_3014), including three MFS-type transporters. Interestingly, some genes from a pathogenicity island required for C. violaceum virulence (cipA, cipB) (Miki et al., 2010), and genes related to the response of C. violaceum to oxidative stress (garA, gstA) (Previato-Mello et al., 2017) were also upregulated in the ΔemrR mutant (Supplementary Table S4 and Figure 4). These data suggest that EmrR acts as a repressor of several putative transporters, including the EmrCAB efflux pump, and also regulates genes involved in other processes (virulence and oxidative stress).

Most MarR family transcription factors act as repressors by direct binding in the promoter region of their target genes (Perera and Grove, 2010). Thus, we employed EMSA to determine whether EmrR acts as a repressor by direct interaction with promoters of the EmrR-repressed genes. After purification of EmrR as a Histag recombinant protein with high degree of purity and homogeneity (Supplementary Figure S2), the labeled DNA fragments for promoter region of nine selected genes were incubated with increasing concentrations of EmrR. For all probes, the shift was first observed at 100 nM EmrR, whereas for a non-specific probe, the shift occurred only at 500 nM EmrR (Figure 5). Competition assays using specific and non-specific unlabeled DNA confirmed the specificity of EmrR binding (Figure 5, right panels). These data confirm that EmrR directly repressed these genes, including the emrR gene itself, as well as the operon emrCAB.

DNA microarray analysis and Northern blot assay using the wild-type and ΔemrR strains indicated that EmrR repressed the operon emrCAB (Figure 4). To further reinforce the regulation of emrCAB by EmrR, the Northern blot assays were performed with other strains and conditions (Figure 6). When the strains were grown in LB medium, the operon emrCAB was highly expressed in both the ΔemrR and emrR_R92H mutant strains, whereas in the wild-type and complemented strains its expression was fully repressed, suggesting that ΔemrR and emrR_R92H produce the EmrCAB efflux pump at high levels (Figure 6A). To test if emrCAB expression can increase in wild-type C. violaceum in response to other compounds, salicylate, ethidium bromide, and nalidixic acid were tested with Northern blot assays. Only salicylate induced the operon emrCAB in a dose-dependent manner (Figure 6B). To confirm whether the phenotypes of increased nalidixic acid resistance and decreased violacein production in the ΔemrR mutant were because this strain overexpresses emrCAB, we constructed and characterized ΔemrCAB and ΔemrRCAB mutant strains (Figure 7). Deletion of emrCAB in the wild-type strain had no effect on susceptibility to several antibiotics (Supplementary Figure S1), including nalidixic acid (Table 2, Supplementary Figures S1, and Figure 7B), nor altered violacein production (Figure 7C) in relation to the wild-type strain, suggesting that the absence of the EmrCAB efflux pump can be replaced by other efflux pumps. On the other hand, deletion of emrCAB in the ΔemrR background (strain ΔemrRCAB) restored nalidixic acid susceptibility (Table 2 and Figures 7A,B) and violacein production (Figure 7C) altered in the ΔemrR mutant, confirming that EmrR controls these phenotypes via EmrCAB. As violacein production is activated by AHL-mediated signaling in C. violaceum (Morohoshi et al., 2008; Stauff and Bassler, 2011), we thought that the absence of violacein in ΔemrR could be related to the secretion of AHL mediated by EmrCAB. Indeed, using the biosensor strain ΔcviI, which does not produce AHLs nor violacein, but does produce violacein in response to exogenous AHL, we verified that ΔemrR accumulated extracellular AHLs (Figures 7D,E) and presented decreased biofilm formation (Figure 7F), suggesting that overexpression of EmrCAB decreases the intracellular accumulation of AHLs and of nalidixic acid.