2.1. Expression and purification of Mtr_Re and Mtr_Ms

The genes for R. erythropolis PR4 Mtr (Mtr_Re; locus tag RER_26020) or for M. smegmatis MC² 155 Mtr (Mtr_Ms; locus tag LJ00_12995) were cloned into a pET-22b(+) plasmid (constructed using NdeI and XhoI sites; GenScript) and transformed into competent E. coli One Shot BL21 (DE3) cells (Invitrogen, Thermo Fisher Scientific). The cells were grown in LB medium containing 100 µg ml⁻¹ ampicillin. Protein expression was induced by adding isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM on reaching an OD_600 nm of 0.4–0.6, and the cultures were incubated at 15°C for 24 h with shaking before the cells were harvested and stored at −80°C.

The cells were thawed, dissolved in 50 mM Tris–HCl pH 7.5, 1 mM DTT, 5 µg ml⁻¹ DNase with a cOmplete Protease Inhibitor Cocktail tablet (Roche) at a 1:10 cell wet weight:buffer ratio and lysed by sonication. Mtr_Re and Mtr_Ms were precipitated with 0.6 and 0.4 g ml⁻¹ ammonium sulfate, respectively, dissolved in 50 mM Tris–HCl pH 7.5, 1 mM DTT and desalted by dialysis (SnakeSkin dialysis tubing, 10K molecular-weight cutoff, Thermo Fisher Scientific). The desalted proteins were filtered through a 0.45 µm filter (Merck Millipore), applied onto a HiTrap Q FF anion-exchange column (Cytiva) and eluted with a linear gradient to 50 mM Tris–HCl pH 7.5, 1 mM DTT, 1 M NaCl. Finally, the proteins were purified on a Superdex 200 Increase 10/300 GL column (Cytiva) in 50 mM HEPES pH 7.5, 150 mM NaCl. The eluted Mtr_Re and Mtr_Ms both showed approximate molecular weights of 100 kDa, corresponding to the dimeric form of Mtr, as validated by calibration of the size-exclusion column (Gel Filtration LMW Calibration Kit, Cytiva). Protein fractions were pooled and concentrated to 30 mg ml⁻¹ using Amicon Ultra-15 filter units (50 kDa molecular-weight cutoff, Merck Millipore), flash-frozen in liquid nitrogen and stored at −80°C. All chromatographic steps were performed on an ÄKTApurifier FPLC system (GE Healthcare) and all expression and purification steps were monitored by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). UV–visible (UV–Vis) spectra were recorded on an Agilent Cary 60 spectrophotometer and the protein concentrations were estimated using an extinction coefficient at λ_max of ɛ_465 nm = 11.5 mM⁻¹ cm⁻¹.

2.2. Protein crystallization and structure determination

Initial crystallization screening of both Mtr proteins was performed using the sitting-drop vapor-diffusion crystallization method with a Mosquito robot (SPT Labtech). Initial hits giving rod-shaped crystals of Mtr_Re (30 mg ml⁻¹) of approximately 100 × 50 × 50 µm in size were obtained with the MIDASplus screen from Molecular Dimensions and were further optimized in 45%(w/v) polypropylene glycol 400, 4%(v/v) ethanol. Rod-shaped crystals of Mtr_Ms (12.5 mg ml⁻¹) of approximately 200 × 30 × 30 µm in size were obtained with Index from Hampton Research [0.02 M magnesium chloride hexahydrate, 0.1 M HEPES pH 7.5, 22%(w/v) poly(acrylic acid sodium salt) 5100]. The crystals were grown at 20°C, cryoprotected in 50%(w/v) polypropylene glycol 400 [and 4%(v/v) ethanol for Mtr_Re] and flash-cooled in liquid nitrogen.

X-ray diffraction data were collected at 100 K on beamlines ID23-2 and ID30B for Mtr_Re and Mtr_Ms, respectively, at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. Diffraction images were processed with XDS (Kabsch, 2010) and AIMLESS (Evans, 2011; Agirre et al., 2023) to resolutions of 2.9 Å for Mtr_Re and 4.7 Å for Mtr_Ms. The structures contained two molecules per asymmetric unit. The structure of Mtr_Re was solved by molecular replacement using Phaser (McCoy et al., 2007) with an AlphaFold (Jumper et al., 2021) ab initio model of R. erythropolis DN1 Mtr as a template (pLDDT > 90), and the solved structure was further used as a template to solve the Mtr_Ms structure, in which the chains were rebuilt with CHAINSAW (Stein, 2008). Several rounds of rebuilding and various refinement strategies were performed using Coot (Emsley et al., 2010) and Phenix (Liebschner et al., 2019). The Mtr_Re data were twinned, and the twin law h, −h − k, −l with a twin fraction of 0.47 was used in all refinements in Phenix. The best Mtr_Re structure was obtained by refining XYZ in both reciprocal and real space, translation–libration–screw (TLS) rotation factors and individual B factors, and applying noncrystallographic symmetry (NCS) restraints. A similar refinement was performed for Mtr_Ms but using isotropic B factors and no TLS rotation factors. For the Mtr_Ms structure, a round of refinement with PDB-REDO was performed (Joosten et al., 2014). The presence of disulfide bonds between the redox-active Cys pairs was confirmed in both chains in Mtr_Re by omit maps, although some minor negative electron density was present indicating slightly incomplete occupancy of the disulfide bridges. PDBePISA was used to analyze the buried surface area of the Mtr dimer. The absorbed X-ray dose was calculated using RADDOSE-3D (Zeldin et al., 2013). All structural figures were prepared with PyMOL version 2.5 (Schrödinger).

2.3. Docking analysis and protein–ligand interactions

Docking studies between Mtr_Re and MSH or MSSM were performed with CB-Dock2 (Cavity-detection guided Blind Docking; Liu et al., 2022). Structure-based blind-docking calculations were performed using the dimeric structure of Mtr_Re and the experimental MSH structure from M. tuberculosis mycothiol S-transferase (PDB entry 8f5v; Jayasinghe et al., 2023) or an MSSM model, with human GR with GSSG bound (PDB entry 1gra; Karplus & Schulz, 1989) as a reference model for template-based blind docking. To generate schematic diagrams of protein–ligand interactions, LigPlot+ (Wallace et al., 1995; Laskowski & Swindells, 2011) was used, applying a 4 Å cutoff radius, to show hydrogen bonds and hydrophobic contacts, which are represented by dashed lines and by arcs with spokes radiating towards the ligand atoms that they contact, respectively.

2.4. Bioinformatics analysis: sequence-similarity networks, multiple sequence alignment and phylogenetic analysis

Sequence-similarity networks (SSNs) were generated with the EFI Enzyme Similarity Tool EFI-EST (https://efi.igb.illinois.edu/efi-est/; Oberg et al., 2023), using M. tuberculosis Mtr as a search sequence, in order to analyze the relation to other similar oxidoreductases. The search included bacterial, archaeal and eukaryotic taxonomic groups using the UniRef50 and UniRef90 sequence databases, retrieving a total of 16 589 homologous sequences. Sequences were grouped with an alignment score of 120 and nodes representing sequences sharing >60% identity, dividing the main homologous oxidoreductases into separate clusters. Figures illustrating SSN analyses were created in Cytoscape (version 3.9) with the organic layout (Shannon et al., 2003). The ten largest clusters of sequences were selected and assembled into seven groups with respect to their annotated function. Multiple sequence alignments were performed in JalView (Waterhouse et al., 2009) with Clustal Omega (Sievers et al., 2011) and phylo­genetic tree analysis with average distances using the BLOSUM62 matrix on two or three selected sequences from each of the ten clusters, including the sequences for the top DALI hits.

2.5. Structure comparison: structural alignment search with DALI

A search for similar structures to Mtr in the Protein Data Bank (PDB) was performed with the DALI (Distance-matrix ALIgnment) protein structure-comparison server (Holm, 2022) using the Mtr_Re structure as a search template. For the top selected hits, a multiple structure sequence alignment was generated with DALI including secondary-structure assignments by DSSP (Kabsch & Sander, 1983; Touw et al., 2015) through DALI and was presented with JalView.

2.6. Analysis of conserved residues

The degree of conservation of residues in the Mtr_Re and GR (PDB entry 1gra; Karplus & Schulz, 1989) structures was evaluated with ConSurf (Ashkenazy et al., 2010, 2016; Celniker et al., 2013). The analysis was based on identification of homologous sequences from the UniRef90 database using the HMMER algorithm (Finn et al., 2015) and multiple sequence alignment with MAFFT (Katoh & Standley, 2013). Of the homologous sequences passing the standard threshold, ConSurf selected a sample of 150 representative sequences. The sequences were inspected in JalView and all 150 sequences from the GR search were annotated as GRs, while only 97 of the 150 sequences selected from the Mtr search were annotated as Mtrs. Therefore, ConSurf was re-run using the 97 Mtr sequences to obtain a conservation degree based only on annotated Mtrs. A nine-bin colored scale was used to show the conservation of each residue, from most variable (turquoise) to most conserved (maroon), when generating three-dimensional figures with PyMOL or coloring the residues in LigPlot+.

3.1. Overall structures of Mtr_Re and Mtr_Ms

Two structures of Mtrs from homologous Actinobacteria are presented in this work; Mtr_Ms from the M. tuberculosis model organism M. smegmatis and Mtr_Re from R. erythropolis, a biocatalyst used for the bioremediation of toxic compounds. The crystal structures of Mtr_Re and Mtr_Ms confirm that Mtr is a homodimer, as also supported by molecular-weight estimations during protein purification. Each monomer, composed of 458 and 461 residues in Mtr_Re and Mtr_Ms, respectively, belongs to the pyridine nucleotide-disulfide oxidoreductase (PNDO) superfamily (Lu et al., 2020) and consists of three domains, namely a NAD(P)H-binding domain, a FAD-binding domain and a dimerization/interface domain (Fig. 1). In FAD-containing NAD(P)H-dependent oxidoreductases, two globular dinucleotide-binding three-layer ββα-sandwich Rossmann-like fold domains are fused into a single chain responsible for the binding of FAD and NAD(P)H (Ojha et al., 2007) through conserved amino-acid sequence motifs (Susanti et al., 2017; Dym & Eisenberg, 2001; Hammerstad & Hersleth, 2021), as also seen in the Mtr structures. Insight into the dimeric state of Mtr from M. tuberculosis has previously been given by Grüber and coworkers, demonstrated by a low-resolution structure derived from SAXS data, as well as dynamic light-scattering (DLS) studies (Kumar et al., 2017). The latter study, however, proposed an extended conformation of the NAD(P)H-binding domain of one of the monomers, resulting in an asymmetric dimer assembly which indicates domain flexibility in solution. This feature is not seen in crystal structures of other homologous oxidoreductases, or in the Mtr_Re or Mtr_Ms crystal structures, which are both composed of a symmetrical dimer assembly.

Figure 1 Crystal structure of MtrRe, showing the biological dimer in (a) and a single monomer (chain A) in (b), displaying the FAD-binding domain in orange, the NAD(P)H-binding domain in green and the interface domain in blue. In (b), additional structural features are indicated. The FAD cofactor and catalytic residues are represented as sticks and colored by atom type.

The crystal contacts between the monomers were analyzed with PISA using the Mtr_Re structure, showing a total buried surface area of 3405 Ų, which is in strong agreement with the total interface area calculated for the dimeric GR (Karplus & Schulz, 1987). Moreover, the total solvent-accessible surface area of Mtr is 34 240 Ų, the complex-formation significance score (CSS) is 0.668 and the solvation free-energy gain upon the formation of the interface (ΔⁱG) is −38.0 kcal mol⁻¹. Its formation entails a solvation free energy (ΔG^int) of −51.6 kcal mol⁻¹, while the free energy of dissociation (ΔG^diss) is 42.2 kcal mol⁻¹. Therefore, the dimeric Mtr crystal structure shown in this work strongly corresponds to the in vivo biological assembly.

Little deviation is observed between Mtr_Re (2.9 Å resolution) and Mtr_Ms (4.7 Å resolution), which show highly similar overall folds with an r.m.s.d. of 1.0 Å (Table 1 and Fig. 2). Despite the low resolution obtained for the latter structure, it is clear that it adopts the structurally conserved topology that is seen for members of the PNDO superfamily and that major features are invariant between the two structures in this study. In addition, a strongly comparable A_465 nm/A_280 nm ratio for both proteins, as well as the yellow crystals obtained of Mtr_Re and Mtr_Ms, support the presence of FAD in both proteins, although FAD could not be built into the Mtr_Ms structure due to its low resolution and lack of significant density. Together, these results demonstrate that the detailed structural features observed and described for the structure of Mtr_Re are representative of both Mtr_Re and Mtr_Ms, as well as other Mtrs across species (Fig. 2). In Mtr_Re, both monomers show strong density for the FAD cofactor, which is tightly stabilized in its binding pocket through several polar interactions. The conserved and redox-active Cys pair responsible for substrate reduction is located on the si face of the isoalloxazine ring of FAD, with the cysteines (Cys39 and Cys44) forming a disulfide bond. The conserved His–Glu ion pair (His442 and Glu447) essential for proton transfer is located at the C-terminal end, positioning it in close proximity to the FAD cofactor of the opposite monomer (Fig. 1). No electron density was observed for NADPH in the Mtr structures reported in this study, as is often the case for oxidoreductase structures where no NADPH has been included in the crystallization setup. However, the putative NAD(P)H-binding site in Mtr is lined by the conserved amino-acid sequence motif [GXGXXA/G for the pyrophosphate group of NAD(P)H] commonly found in oxidoreductases utilizing this cofactor (Dym & Eisenberg, 2001; Hammerstad & Hersleth, 2021). Part of the HRRXXXR binding motif for the 2′ phosphate group of NADPH, found, for example, in E. coli low M_r TrxR (Laurent et al., 1964), is lacking in Mtr. However, amino-acid substitutions in this motif are commonly seen in NADPH-consuming reductases, and the basic amino acids present in Mtr_Re (for example Arg199 and Arg205) are likely to serve a homologous role in the stabilization of the 2′ phosphate group of the pyridine nucleotide. Moreover, Mtr has previously been shown to be selective for NADPH over NADH, as well as indicating a strict preference for the 2′-phosphate regioisomer when assayed with 3′-NADPH (Patel & Blanchard, 1999).

Figure 2 Structural alignment of MtrRe (multicolored) and MtrMs (gray), showing the similar overall fold of the biological dimers of the two crystal structures. The FAD cofactor corresponds to MtrRe and is shown as sticks and colored by atom type.

3.2. Sequence and structure comparison of Mtr with homologous oxidoreductases

The sequence and structure of Mtr were compared with those of other enzymes using different approaches. Structural comparison of Mtr (Mtr_Re) with deposited PDB structures using the DALI protein structure comparison server shows that Mtr is highly similar to other disulfide oxidoreductases. Hits with the highest Z-score for each of the eight most similar protein types are shown in Table 2, which correlates to the results from the SSN analysis. High structural similarity is observed to MerA (PDB entry 5x1y; Bafana et al., 2017), DLD (PDB entry 1ebd; Mande et al., 1996), GR (PDB entry 6b4o; Center for Structural Genomics of Infectious Diseases, unpublished work), TryR (PDB entry 2tpr; Kuriyan, Kong et al., 1991), GAR (PDB entry 2rab; Van Petegem et al., 2007) and TrxR (PDB entry 3dgz; B. E. Eckenroth, R. J. Hondal & S. J. Everse, unpublished work) (r.m.s.d.s of 1.8–2.2 Å), with somewhat lower similarity to quinone reductase (LpdA; PDB entry 1xdi; Argyrou et al., 2004) and CoADR (PDB entry 5l1n; Sea et al., 2018) (r.m.s.d.s of 2.7–2.8 Å).

The overall fold of Mtr highly resembles that of previously characterized oxidoreductases, as seen from the structural alignment of Mtr with selected homologous structures (Fig. 3a). All structures share the overall conformation and arrangement of their three domains and the same orientation of the FAD in the FAD-binding domain, as well as the location of the NAD(P)H-binding site. The probable and conserved positioning of NADPH in Mtr can be demonstrated using the coordinates of GR (PDB entry 1grb; Karplus & Schulz, 1989; Fig. 3b). A conformational change of an aromatic residue, Phe178, would be required for NADPH binding in Mtr, a rearrangement that has previously been described for GR. In GR, the equivalent Tyr197 on the re face of FAD shields the flavin cofactor, blocking the nicotinamide-binding pocket, but rotates away from the isoalloxazine ring of FAD upon NADPH binding (Fig. 3b), further allowing hydride transfer from NADPH to FAD and ultimately transferring an electron pair to the proximal cysteine of the redox-active Cys pair (Karplus & Schulz, 1989; Berry et al., 1989). Hence, by direct comparison, the NADPH nicotinamide ring of Mtr_Re would be stabilized between the re face of the isoalloxazine ring of FAD and the phenyl group of Phe178 through stacking interactions. In Mtr, Arg199 and Arg205, which are also conserved in GR (Arg218 and Arg224), are likely to be involved in electrostatic interactions with the 2′ phosphate group of NADPH, which can also be noted as an RXXXXXR motif (Figs. 3b and 4). The Mtr crystal structures presented in this work confirm that Mtr is composed of the low M_r TrxR-like fold with an additional C-terminal interface domain, as described for related oxidoreductases such as GR (Hammerstad & Hersleth, 2021).

Figure 3 (a) Structural alignment of MtrRe and the homologous oxidoreductases GR (PDB entries 1gra and 1grb; Karplus & Schulz, 1989; NADPH from the latter), DLD (PDB entry 2eq9; T. Nakai & N. Kamiya, unpublished work) and MerA (PDB entries 1zk7 and 4k7z; Ledwidge et al., 2005; A. Dong, M. Falkowski, M. Malone, S. M. Miller & E. F. Pai, unpublished work; NADPH from the latter). Cofactors and the redox-active Cys pairs are represented as sticks and colored by atom type. (b) Overlay of MtrRe and GR (PDB entries 1gra and 1grb) showing the arginines stabilizing the 2′ phosphate group of NADPH, and Phe178 in Mtr corresponding to Tyr197 in GR which undergoes conformational change when NADPH binds.

Figure 4 Multiple structural sequence alignment generated by DALI for the structures listed in Table 2. The figure was generated with JalView and the sequences are colored according to percentage identity. The consensus secondary-structure assignments by DSSP (H/h, helix; E/e, strand; L/l, coil) are shown below the alignment. The characteristic motifs are shown above the sequences and the green lines indicate the residues that line the binding pocket in Mtr and GR.

To more generally characterize Mtr with respect to other homologous enzymes, a sequence-similarity network (SSN) was generated. The SSN investigation showed that the top ten clusters consisted of other NAD(P)H-dependent disulfide/thiol oxidoreductases containing the canonical low M_r TrxR fold with an additional interface domain. The clusters were divided into seven groups according to their annotated function (Fig. 5). Actinobacterial Mtrs cluster together with a group of archaeal dihydrolipoamide dehydrogenases (DLDs; colored orange); the latter is part of a larger group of three DLD clusters (colored red). The close relationship between Mtrs and archaeal DLDs is also seen from the phylogenetic analysis (Supplementary Fig. S1). GRs cluster into a separate cluster, however, containing a few sequences of the related glutathione amide reductases (GARs) and trypanothione reductases (TryRs). Mercuric reductases (MerAs), eukaryotic high M_r TrxRs and the most distant group of CoADRs each make up distinct clusters, as seen in the SSN and phylo­genetic analysis. CoADRs differ functionally from the remaining enzymatic clusters in view of their single active-site Cys residue used for catalysis and cluster into an independent clade in the phylogenetic tree. Three clusters containing sequences annotated as DLDs, PNDOs and FAD-dependent oxidoreductases or MerAs, with the latter lacking the metal-binding NmerA domain, make up a final miscellaneous group. The SSN shows that the largest groups of homologous enzymes to Mtr are the GRs, DLDs, MerAs and high M_r TrxRs; however, the closest group that cluster together with Mtrs are the archeal group annotated as DLDs. This close relationship is interesting and will need further investigation to reveal whether there are some additional functional relationships between these groups.

Figure 5 Comparison of Mtr with homologous oxidoreductases through sequence-similarity networks (SSNs). The SSN displays the ten largest clusters. The clusters are individually colored and assembled into seven groups with respect to their annotated function. (GR, glutathione reductases; TryR, trypanothione reductases; GAR, glutathione amide reductases; TrxR, thioredoxin reductases; Mtr, mycothiol disulfide reductases; MerA, mercuric reductases; DLD, dihydrolipoamide dehydrogenases; CoADR, coenzyme A disulfide reductases).

3.3. The putative MSSM binding site

Through docking studies and structural comparison with the crystal structure of human GR (with GSSG bound; PDB entry 1gra; Karplus & Schulz, 1989), the binding of MSSM to the active site of Mtr was examined. Mtr shows high structural similarity to GR, and a similar reaction mechanism to that of GR has been proposed for Mtr in the reduction of MSSM (Patel & Blanchard, 1999). Therefore, we expected MSSM to bind to Mtr in a similar way as GSSG binds to GR, placing the substrate disulfide in the vicinity of the redox-active Cys pair on the si face of FAD, allowing reduction of MSSM. Moreover, as GR is only functional as a homodimer, with each substrate-binding site being formed by both subunits (Schulz et al., 1978), this is also expected for Mtr.

Initial structure-based blind-docking calculations were performed between homodimeric Mtr_Re and the reduced product MSH (PDB entry 8f5v; Jayasinghe et al., 2023), returning two significant solutions with Vina scores of −6.8 and −5.8. The MSH molecules are docked into the cavity corresponding to the GSSG binding cleft in GR, placing the MSH thiol near the catalytic Cys pair in Mtr, stabilized by a significant number of putative polar contacts (data not shown).

Further docking studies revealed that MSSM can also fit into the expected substrate-binding site with its disulfide positioned in close proximity to the catalytic cysteines of Mtr (Vina score of −7.0; Fig. 6, Supplementary Fig. S2 and Fig. 4). This positioning of the substrate would facilitate the initial nucleophilic attack performed by the N-terminal cysteine of Mtr (Cys39) on the MSSM disulfide bond, forming a mixed disulfide, commonly described as the first catalytic step performed by GR and similar FDRs (Fagan & Palfey, 2010; Deponte, 2013; Berry et al., 1989; Pai & Schulz, 1983; Kallis & Holmgren, 1980). Analogous to GR, deprotonation of the Cys39 thiol group in Mtr, leading to the formation of this intermolecular disulfide bond, could be accelerated owing to the histidine residue (His442) of the conserved His–Glu ion pair located in the opposite subunit of the Mtr homodimer, which is positioned close to the redox-active Cys pair and FAD cofactor (Figs. 1b and 6a). In GR, the interaction between the analogous histidine and glutamate residues has been proposed to facilitate deprotonation in a similar way as in serine proteases, and the histidine has furthermore been suggested to protonate the thiolate leaving group of the first GSH molecule leaving the active site (Pai & Schulz, 1983; Veine et al., 1998; Wong & Blanchard, 1989; Wong et al., 1988; Arscott et al., 2000; Fig. 6c). A nearby tyrosine residue (Tyr114, human GR numbering) was proposed to be involved in assisting the acid catalyst histidine (Krauth-Siegel et al., 1998); however, this has subsequently been disproved by others (Deonarain et al., 1989). This tyrosine is, however, only conserved in 9% of GRs in the ConSurf search and is not conserved in Mtrs, where it is replaced by a glycine residue (Gly95) that is unlikely to play a catalytic role in the oxidative half-reaction of MSSM reduction.

Figure 6 Comparison of the putative MSSM binding site in MtrRe (a, b) with the GSSG binding site from the GR–GSSG complex (PDB entry 1gra; Karplus & Schulz, 1989) (c, d). Coordinates for NADPH were taken from PDB entry 1grb (Karplus & Schulz, 1989). (a) and (c) show the respective disulfide substrates bound in proximity to the redox-active Cys pairs. The His–Glu ion pairs are located on opposite monomers of the dimers. In (b) and (d) the active sites are shown from a different angle in surface representation, revealing differences in the space available for substrate binding in Mtr and GR, respectively, and showing the degree of conservation of the residues lining the substrate-binding clefts as evaluated with ConSurf. Variable residues are colored turquoise and highly conserved residues are colored maroon. Cofactors, catalytic residues and substrates are represented as sticks and colored by atom type. In (e) and (f) the binding sites of Mtr and GR are overlaid, showing the increase in the size of the binding pocket that is largely due to a shift of the α3 helix in Mtr compared with GR.

The α-helix (numbered α3) enclosing one side of the substrate-binding pocket (Trp77–Asp102 in Mtr_Re) is slightly shifted in the Mtr structure (Figs. 6e and 6f) compared with GR (PDB entry 1gra; Karplus & Schulz, 1989), creating a larger binding cavity lined with highly conserved residues (Figs. 6b and 6d and Supplementary Fig. S2). Consequently, this allows more space for the larger MSSM substrate to bind, in agreement with the larger size of MSSM compared with GSSG. It is notable that the residues in the α3 helix are more conserved among Mtrs than among GRs (Figs. 6b and 6d). That the α3 helix is more straight in GR, while it is bent in Mtrs, could be due to the three glycine residues (known to disorder helices) found in both Mtr_Re and Mtr_Ms but not in GR (Figs. 6e and 4).

Overall, our docking calculations provide insight into a likely MSSM binding site in Mtr that is compatible with the expected mechanism for MSSM reduction, supported by functional studies as well as by structural similarity to GR (Patel & Blanchard, 1999; Holsclaw et al., 2011; Kumar et al., 2017; Karplus & Schulz, 1989).