Escherichia coli TOP10 and DH5α were used for cloning. M. smegmatis mc²155 was used throughout this work. The E. coli strains were grown in Luria-Bertani (LB) broth, while M. smegmatis was cultured in Middlebrook 7H9 broth (Difco) containing 10% oleic acid-albumin-dextrose-catalase supplement (Becton Dickinson) and 0.05% Tween 80. All strains were grown at 37°C with shaking at 200 rpm. Hygromycin (200 μg ml⁻¹ for E. coli and 100 μg ml⁻¹ for M. smegmatis), kanamycin (50 μg ml⁻¹ for E. coli and 25 μg ml⁻¹ for M. smegmatis), 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (Xgal 50 μg ml⁻¹) and sucrose (2% w/v) were used for selection or screening as appropriate. Ciprofloxacin, Norfloxacin, and Ofloxacin stock solutions (10 mg/ml⁻¹) were prepared in NaOH 1 M and Rifampicin stock solution (10 mg/ml⁻¹) was prepared in methanol. Mycobacterial biofilms were grown in Sauton medium (0.5 g KH₂PO₄, 0.5 g MgSO₄ 7H₂O, 2.0 g citric acid, 0.05 g ferric ammonium citrate, 60 ml glycerol, 4.0 g asparagine, 0.1 ml ZnSO₄ 1% solution, distilled water up to 1,000 ml, and pH 7.4).

Primer sequences and plasmids are listed in Supplementary Tables S1, S2, respectively. The M. smegmatis (ΔMSMEG_3763) mutant strain, carrying an in frame deletion in MSMEG_3763, was isolated using a two-step homologous recombination strategy (Parish et al., 1999). A 806 bp fragment (up), containing the upstream flanking regions of MSMEG_3763 was PCR-amplified from M. smegmatis mc²155 genomic DNA using the forward upMS3763f and reverse upMS3763r primers. The amplified fragment (up) was cloned into HindIII and KpnI restriction sites of p2NIL, yielding the pBD01 plasmid. A 825 bp fragment (dw), containing the downstream flanking regions of MSMEG_3763, was PCR-amplified from M. smegmatis mc²155 genomic DNA using the forward dwMS3763f and reverse dwMS3763r primers. The amplified fragment (dw) was cloned into KpnI and PacI restriction sites of pBD01, yielding the pBD02 plasmid. To obtain the suicide delivery vector pBD03, the pGOAL19 cassette was cloned into the PacI restriction site of the pBD02 plasmid. pBD03 was electroporated into M. smegmatis mc²155 and single crossover events (SCOs), positive in the Xgal screening and resistant to kanamycin and hygromycin, were verified by colony PCR. Among the positive SCOs, a single colony was streaked onto fresh media without selection, and incubated at 37°C for 5 days to allow the second recombination event to occur, before selection on plates containing sucrose and Xgal. The white sucrose-resistant colonies were screened for kanamycin and hygromycin sensitivity, and then analyzed by colony PCR to confirm that a second recombination event had occurred, yielding a deletion of 771 bp in the coding region of MSMEG_3763. One positive clone was chosen for further experiments and herein named M. smegmatis (ΔMSMEG_3763). The in-frame deletion event in M. smegmatis (ΔMSMEG_3763) was verified by sequencing.

To complement the gene-inactivating mutation in M. smegmatis (ΔMSMEG_3763), a DNA fragment containing the 770 bp coding sequence of MSMEG_3763 (including start and stop codons) was PCR-amplified from M. smegmatis mc²155 with forward cMS3763f and reverse cMS3763Salr primers (Supplementary Table S2). The forward primer included an optimized Shine–Dalgarno sequence. The resulting 805 bp fragment was cloned into the EcoRI-SalI sites of pMV306hsp (Addgene#26155) to obtain pBD04, carrying the MSMEG_3763 gene under the control of the Mycobacterium bovis BCG hsp60 promoter. The recombinant plasmid pBD04 was verified by PCR and sequencing (Microgem Laboratory Research). pBD04 was electroporated into M. smegmatis (ΔMSMEG_3763) to generate the complemented M. smegmatis (ΔMSMEG_3763 pBD04) strain. Recombinant clones were selected using kanamycin, and the presence of pBD04 was verified by plasmid extraction followed by PCR analysis.

Efflux pump activity was determined using ethidium bromide (EtBr) as previously described (Rodrigues et al., 2013). Mycobacterium smegmatis strains were grown in 7H9/OADC medium at 37°C to an OD₆₀₀ of 0.6 (corresponding to 10⁸ CFU ml⁻¹), the pellet washed and resuspended in PBS supplemented with 0.05% Tween80 and EtBr solution to final concentration of 2 μg ml⁻¹. Aliquots of 195.5 μl of the bacterial suspension were added to 96 well microtitre plate containing 4.5 μl of water or 4.5 μl of verapamil (VPL) to a final concentration of 250 μg ml⁻¹. The assay was conducted at 37°C in a Promega GloMax® Discover Microplate Reader using the excitation and emission wavelengths of 520 nm and 560–640 nm, respectively. Fluorescence data was acquired every 60 s for 90 min at 37°C. Fluorescence values were expressed as percentage increase with respect to the value of each sample at 1 min. EtBr and VPL were used at ¼ the M. smegmatis minimum inhibitory concentration (MIC) in order not to compromise the cellular viability, as confirmed by c.f.u. counting. Each experiment was repeated in triplicate.

The MIC determination protocol was adapted from Agrawal et al. (2015). In brief, wells of a 96 well microtitre plate were filled with 50 μl 7H9 media except for the first column. Double of the required drug concentration was prepared and 100 μl volumes were added to the first column. This was diluted to reach the desired concentration of rifampicin, or serially diluted to halve the concentration by mixing with media only in the subsequent wells to the penultimate column for CIP. No drug was added to the last column as a no antibiotic control. The M. smegmatis strains were grown in replicates in 7H9 medium to an OD₆₀₀ of 0.6, diluted 1,000 times (corresponding to 10⁵ CFU ml⁻¹; Supplementary Figure S1), and 50 μl of the diluted culture was added to each well. The plate was sealed with parafilm to avoid drying of the cultures, and incubated with mild shaking (100 rpm) at 37°C for 40 h, followed by addition of 30 μl Sigma Resazurin sodium salt dye (filter sterilized, 0.2 mg ml⁻¹ final concentration) to each well and incubation for a further 6 h before imaging. For EtBr and VPL MIC determination, 20 μl of Promega CellTiter-Blue® was used as indicator. MICs were determined as the values of the first well showing no growth as indicated by color change.

Biofilm formation assay was performed as previously described but with some modifications (O’Toole 2011). Bacterial cultures were grown in 7H9/ADC 0.05% Tween80 medium. The well-grown primary cultures (~OD₆₀₀ 1.5) of wild-type, M. smegmatis (ΔMSMEG_3763) and M. smegmatis (ΔMSMEG_3763 pBD04) strains were washed twice with Sauton medium to remove Tween80. The cultures were diluted 1:500 to inoculate 7 ml Sauton medium in 60 mm diameter polystyrene Petri dishes. Petri dishes were incubated for 3–6 days at 37°C in a humidified incubator, and the formation of surface pellicles was monitored. Quantification of biofilm was performed as follows: well-grown primary cultures (~OD₆₀₀ 1.5) were washed with Sauton medium and diluted to a final OD₆₀₀ of 0.05 in the same medium, then distributed into 96-well polystyrene microtitre plates and incubated at 37°C in a humidified incubator for 4–7 days. Each well received an inoculum of 200 μl. After incubation, contents of the wells were aspirated out by syringe and the wells were washed with sterile water. The wells were then stained for 15 min, adding 300 μl 1% Crystal Violet (w/v) solution to each well. The stain was removed, washing the plate with sterile water, and plates were left to air dry. 300 μl 80% ethanol was added to each stained well and the plates were incubated for 15 min at room temperature. The content of each well was mixed by pipetting, and 200 μl transferred to an optically clear flat-bottom 96-well plate. Optical density (OD) was measured at 550 nm using a Biotek Synergy HT plate reader.

Amino acid sequences were retrieved and analyzed with ExPASy tools using the MSMEG_3763 gene sequence as input. The MSMEG-3763 aggregation profile was evaluated using the AGGRESCAN server and the presence of putative transmembrane helices predicted by the TMHMM server on the basis of the amino acid sequence (Krogh et al., 2001). The three-dimensional (3D) structure of MSMEG-3763 was predicted using the I-TASSER server (Interactive Threading ASSEmbly Refinement; Roy et al., 2010). I-TASSER is a computational structure modeling method that uses a combinatorial approach of comparative modeling, threading, and ab initio modeling (Roy et al., 2010). In particular, the structure predictions by I-TASSER rely on template proteins with known structures obtained from databases and the prediction procedures are based on matching the query sequence against a non-redundant sequence database. According to I-TASSER criteria, the selection of the representative 3D model was performed by evaluating the C-score. This latter parameter is a scoring function based on the relative clustering structural density and the consensus significance score of the multiple threading templates. The C-score is usually in the range from −5 to 2, where a score of higher value indicates a predicted model with a high confidence and vice versa. Additionally, the quality of the selected model was further evaluated using PROCHECK and MolProbity (Laskowski et al., 1996; Davis et al., 2007). The selected 3D model was visualized using PyMol (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC) and Chimera (Pettersen et al., 2004). The secondary structure elements were identified by DSSP (Hess et al., 2008; Joosten et al., 2011; Huang and MacKerell, 2013). The membrane protein modeling of MSMEG-3763 was performed using a two-step procedure. First, the protein was dissolved in dipalmitoyl phosphatidylcholine (DPPC) phospholipid bilayer by GROMACS (Hess et al., 2008) using the CHARMM36 force field (Huang and MacKerell, 2013). The protein was inserted into the membrane using the structural information obtained as reported above by TMHMM server. Finally, in order to define the correct orientation and position of MSMEG-3763 inside the membrane a minimization protocol was applied.

Molecular docking studies of rifampicin and ciprofloxacin with MSMEG-3763 were conducted using EAdock algorithm implemented in the Swissdock server (Grosdidier et al., 2011). The molecular docking protocol involved the preparation of MSMEG-3763 and ligand structures. In particular, for the protein the predicted 3D model was used as the reference structure, whereas the rifampicin and ciprofloxacin structures were downloaded from Protein Data Bank (PDB) database. In accord with the docking protocol, all residues of the protein were held fixed and a binding pocket was not defined. The docking procedure was validated by comparing the binding affinity values estimated for the ciprofloxacin/MSMEG-3763 (-5.9 kcal/mol) and rifampicin/MSMEG-3763 (-6.1 kcal/mol) complexes with that obtained from the ofloxacin/MSMEG-3763 (-4.2 kcal/mol) model. The binding affinity values were obtained for each docking model by using PRODIGY webserver.¹ All docked poses were visualized and analyzed using Chimera (Pettersen et al., 2004).

The M. smegmatis (ΔMSMEG_3763) mutant strain, carrying a 771 bp deletion in the coding region of MSMEG_3763, and its isogenic M. smegmatis (ΔMSMEG_3763 pBD04) complemented strain were constructed to characterize the function of the putative MSMEG-3762/63 mycobacterial efflux pump. The growth rate of the mutant and complemented strains were comparable to that of wild type in standard laboratory conditions (data not shown).

Ethidium Bromide accumulation is a well-established method to monitor efflux activities in bacteria. The assay was used to assess the role of MSMEG-3763 as a component of the putative efflux pump by measuring differences in EtBr accumulation between wild type M. smegmatis (wt), deletion mutant (ΔMSMEG_3763), and complemented (ΔMSMEG_3763 pBD04) strains in the absence and presence of VPL, an efflux pump inhibitor (EPI). First, MICs of EtBr and VPL were determined as 8 μg ml⁻¹ and 900 μg ml⁻¹ for EtBr and VPL, respectively (data not shown). The efflux inhibition assays were conducted using concentrations that did not affect mycobacterial viability (¼ MICs). As expected, the M. smegmatis (ΔMSMEG_3763) mutant strain showed higher EtBr accumulation compared to the wild type and the complemented strains (Figure 1A). When the bacterial cultures were exposed to both EtBr and VPL, increased EtBr accumulation was observed for the three strains over the same incubation time. Once again, EtBr accumulation was higher for the mutant compared to the wild type and the complemented strains (Figure 1B), demonstrating that loss of MSMEG-3763 reduced the ability of mycobacteria to remove EtBr, and suggesting that this complex possessed efflux pump activity.

In order to verify the involvement of the MSMEG-3762/63 efflux pump in the removal of antimicrobial compounds used in anti-TB therapy, the MICs for first- and second-line anti-TB drugs were determined using the Resazurin Reduction Microtiter Assay (REMA). The MIC of rifampicin was lower (2 μg ml⁻¹) for M. smegmatis (ΔMSMEG_3763) compared to M. smegmatis (wt; 3 μg ml⁻¹; Figure 2A). As expected, the MIC of rifampicin for complemented M. smegmatis (ΔMSMEG_3763 pBD04) was the same as that observed for the wild type strain. The loss of MSMEG-3763 did not change the efficacy of other first-line antibiotics, isoniazid, and pyrazinamide (data not shown). The MICs of second-line TB antibiotics were also determined. An increase in sensitivity of M. smegmatis (ΔMSMEG_3763) to ciprofloxacin was observed. The MICs of ciprofloxacin for M. smegmatis (wt) and complemented M. smegmatis (ΔMSMEG_3763 pBD04) were (0.25 μg ml⁻¹), while the MIC for M. smegmatis (ΔMSMEG_3763) was 0.12 μg/ml⁻¹(Figure 2B). No difference in MIC was observed for other second-line anti-TB drugs, NOR or ofloxacin (data not shown). Thus, abrogation of MSMEG-3762/63 efflux pump function increased mycobacterial drug sensitivity to both rifampicin and ciprofloxacin.

A bioinformatic analysis of the MSMEG-3763 protein based on its primary aa sequence was performed. The aggregation profile of the protein was determined using AGGRESCAN and TMHMM servers (Figure 3A), showing that the regions spanning residues 29–35, 37–48, 60–71, 73–83, 121–135, 141–165, 182–190, and 230–248 are clearly characterized by high aggregation prone sequences. Further analysis using the server TMHMM suggested that, in the regions showing high aggregation propensity, the protein may form transmembrane helices (Figure 3B).

To better understand the functional proprieties of MSMEG-3763 a detailed description of the structural features is required. Therefore, to elucidate the molecular details of the protein, we performed 3D modeling of the structure using I-TASSER. The computational modeling of MSMEG-3763 generated five models with C-scores ranging from −4.36 to 0.29. The 3D model with the highest c-score was used as a reference for the structural analysis. Additionally, the good quality of the selected 3D model was confirmed by evaluating the structural statistics (Supplementary Figure S2A) and the Ramachandran plots (Supplementary Figure S2B) respectively, obtained by analyzing the predicted model using the software MolProbity (Davis et al., 2007), with 86 and 11% of aa-residues situated in most favorable (MF) and allowed regions (AR), and PROCHECK (Laskowski et al., 1996), with 84 and 13% aa-residues situated in MF and AR, respectively. The 3D structural model of MSMEG-3763 (Figure 3C) indicates that the protein presents as a globular fold composed of 10 α-helices (A = 10–27, B = 29–47, C = 60–91, D = 94–101, E = 106–136, F = 143–169, G = 172–188, H = 201–207, I = 213–223, and L = 232–257) joined by short non-helical regions that provide a rigid structural framework. In particular, the hydrophobic portion of the protein is a typical helix bundle composed of six α-helices (Figure 3C, in red) that likely forms a channel inside the membrane. As proposed by the model reported in Figure 3D, the hydrophilic portions, according to TMHMM predictions, probably correspond to the outer and the cytosolic regions of the protein. Therefore, structural modeling of the putative transmembrane component of the MSMEG-3762/63 efflux pump predicts a structure suited to purpose.

To gain structural insights into the interaction between rifampicin and ciprofloxacin compounds and the membrane protein MSMEG-3763 (Figure 4A), we performed a series of molecular docking studies that represent a powerful technique for studying protein-ligand complexes. In particular, molecular docking aims to predict at atomic level the binding mode (i.e., position and orientation) and affinity of a complex formed by two or more constituent molecules with known structures. The three-dimensional model of protein-ligand complexes of MSMEG-3763 was generated using the EAdock algorithm (Figure 4B). The optimized structure of the ciprofloxacin/MSMEG-3763 complex indicated that the recognition mechanism of ciprofloxacin by MSMEG-3763 is principally mediated by the residues Phe¹⁰⁷, Leu¹⁵⁷, Leu¹⁶¹, Val¹⁸², Pro¹⁸⁵, Leu¹⁸⁸, Leu¹⁸⁹, Trp²⁰⁵, Val²⁰⁶, Ala²⁰⁹, Leu²¹⁰, Pro²¹¹, Phe²⁴², Leu²⁴⁵, Ala²⁴⁶, Leu²⁴⁹ forming a shallow hydrophobic pocket on the surface of MSMEG-3763 (Figure 4C). Instead, for the rifampicin/MSMEG-3763 complex (Figure 4D) the molecular docking structure predicted that rifampicin binds into a hydrophobic cleft on the surface of MSMEG-3763 formed by Phe¹⁰⁷, Leu¹¹⁰, Ala¹¹¹, Gly¹¹⁴, Val¹⁵⁶, Val¹⁵⁹, Gly¹⁶⁰, Val²⁴⁴, Leu²⁴⁷, and Ala²⁵¹. These docking results therefore suggest that this efflux pump has the capacity to bind, and potentially export, the antimicrobial drugs rifampicin and ciprofloxacin. To validate the docking protocol applied to obtain the structural model of ciprofloxacin/MSMEG-3763 and rifampicin/MSMEG-3763 complexes, we estimated for both models the binding affinity and then compared these values with that estimated for the ofloxacin/MSMEG-3763 model (Supplementary Figure S3) obtained using the same docking parameters. Notably, no difference in MIC of ofloxacin was observed for M. smegmatis wt and ∆3763 mutant strain. The comparison of the binding affinity values (ciprofloxacin/MSMEG-3763 = -5.9 kcal/mol; rifampicin/MSMEG-3763 = -6.1 kcal/mol; and ofloxacin/MSMEG-3763 = -4.2 kcal/mol) clearly indicates that the formation of ciprofloxacin/MSMEG-3763 and rifampicin/MSMEG-3763 complexes is favored with respect to the ofloxacin/MSMEG-3763 complex, confirming the MIC data.

These findings demonstrate that the applied docking protocol allows a structural model for ciprofloxacin and rifampicin binding to be obtained, providing a reasonable description of the molecular determinants involved in the recognition mechanism of MSMEG-3763 for both compounds.

Bacterial efflux pumps have roles in many physiological functions that influence pathogenicity besides drug export, including biofilm development. To test if the absence of MSMEG-3763 affected the ability to form a mycobacterial biofilm, a comparative analysis of the three strains was performed. Pellicles formed by M. smegmatis (wt) were observed on the surface after 3 days of incubation and continued to develop up to the last day of observation (day 6; Figure 5A). The M. smegmatis (ΔMSMEG_3763) deletion mutant also formed a biofilm, but the reticulated appearance associated with formation of the extracellular matrix was not observed on the surface of the liquid growth medium. This observation was further confirmed by a crystal violet quantification of the biofilms (Figure 5B). The amount of biofilm formed by the efflux pump mutant M. smegmatis (ΔMSMEG_3763) was reduced from 33–61%, depending on the day of sampling, compared with the wild type. As expected, the ability to form a biofilm was restored nearly to wild type levels in the M. smegmatis (ΔMSMEG_3763 pBD04) complemented strain (Figure 5B). Loss of efflux function through MSMEG-3762/63 therefore interferes with biofilm development, influencing mycobacterial interactions with the environment that may affect pathogenicity.

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.