Crystal structure of FAD-independent methylene-tetrahydrofolate reductase from Mycobacterium hassiacum

FAD-independent methylene-tetrahydrofolate (methylene-H4F) reductase (Mfr), recently identified in mycobacteria, catalyzes the reduction of methylene-H4F to methyl-H4F with NADH as hydride donor by a ternary complex mechanism. This biochemical reaction corresponds to that of the ubiquitous FAD-dependent methylene-H4F reductase (MTHFR), although the latter uses a ping-pong mechanism with FAD as prosthetic group. Comparative genomics and genetic analyses indicated that Mfr is indispensable for the growth of Mycobacterium tuberculosis, which lacks the MTHFR-encoding gene. Thus, Mfr is an excellent target enzyme for the design of antimycobacterial drugs. Here, we report the heterologous production, enzymological characterization and the crystal structure of Mfr from the thermophilic mycobacterium M. hassiacum (hMfr), which shows 78% sequence identity to Mfr from M. tuberculosis. Although hMfr and MTHFR show very low sequence identity and different catalytic mechanisms, their tertiary structures are highly similar, which suggests a divergent evolution of Mfr and MTHFR from a common ancestor. Most of the important active-site residues of MTHFR are conserved and equivalently positioned in the tertiary structure of hMfr. The Glu9Gln variant of hMfr exhibits a drastic reduction of the catalytic activity, which supports the predicted function of the glutamate residue as proton donor in both Mfr and MTHFR. The predicted nicotinamide binding site of hMfr is substantially narrower than the isoalloxazine binding site of MTHFR, which may reflect an evolutional adaptation to the different sizes of the coenzymes.

analyses indicated that Mfr is indispensable for the growth of Mycobacterium tuberculosis, which lacks the MTHFR encoding gene.Therefore, Mfr appears to be an excellent target for the design of antimycobacterial drugs.Here, we report the heterologous production, enzymological characterization, and the crystal structure of Mfr from the thermophilic mycobacterium Mycobacterium hassiacum (hMfr), which shows 78% sequence identity to Mfr from M. tuberculosis.Although hMfr and MTHFR have minor sequence identity and different catalytic mechanisms, their structures are highly similar, thus suggesting a divergent evolution of Mfr and MTHFR from a common ancestor.Most of the important active site residues of MTHFR are conserved and equivalently positioned in the tertiary structure of hMfr.
The Glu9Gln variant of hMfr exhibits a drastic reduction of the catalytic activity, which supports the predicted function of the glutamate residue as proton donor in both hMfr and MTHFR.Thus, highly similar binding modes for the C 1 -carriers and the reducing agents in hMfr and MTHFR are assumed.

| INTRODUCTION
Estimated ten million people are infected with tuberculosis each year and more than one million succumb to the disease. 1 Enhanced by increasing drug resistance, tuberculosis poses an outstanding public health threat. 2Therefore, the development of novel antimycobacterial agents is urgently needed. 3One of the key challenges in antimicrobial drug development is the discovery of new enzyme targets that are unique to the pathogen. 4,5e methylene-tetrahydrofolate (methylene-H 4 F) reductase (MTHFR) is a ubiquitous enzyme involved in the central carbon metabolism of eukaryotes, bacteria and most archaea.This enzyme catalyzes the reduction of methylene-H 4 F to methyl-H 4 F using NAD(P)H as reducing agent. 6MTHFR of Escherichia coli (eMTHFR) is a homotetramer (33 kDa per monomer) and contains FAD as prosthetic group.Catalysis proceeds via a ping-pong mechanism consisting of two half reactions.FAD is reduced by NAD(P)H in the reductive half reaction, and the generated FADH 2 subsequently reduces methylene-H 4 F in the oxidative half reaction.[13][14][15] The canonical MTHFR gene was, however, not found in the majority of mycobacterial genomes.Young et al. predicted by comparative genomic analysis and structure modeling that a gene of Mycobacterium tuberculosis may encode a noncanonical MTHFR. 16 1).Notably, the amino acid sequence identity of the canonical and noncanonical MTHFRs was very limited.The sequence identity is only 16%-17% and the coverage is 13%-26%.Deletion of MSMEG_6596 from M. smegmatis impaired growth of the strain in the absence of methionine. 17cently, Yu et al. reported that the protein encoded by Rv2172c of M. tuberculosis exhibits the same properties as the noncanonical MTHFR found in M. smegmatis.Gene deletion experiments revealed Rv2172c as essential for the growth of M. tuberculosis. 18Apparently, the noncanonical MTHFR is the only enzyme catalyzing the reduction of methylene-H 4 F in M. tuberculosis.Since the noncanonical MTHFR found in mycobacteria is a new type of methylene-H 4 F reductase, 17 this enzyme is referred to as methylene-tetrahydrofolate reductase (Mfr) in this article to distinguish it from the canonical MTHFR.
Since the genes encoding Mfr are found only in mycobacteria and some related bacteria, Mfr is an excellent target enzyme for the design of antimycobacterial agents.For the design of Mfr inhibitors, a high-resolution structure of Mfr is required.Here, we present the crystal structure of Mfr from Mycobacterium hassiacum (hMfr).
M. hassiacum is a thermophilic Mycobacterium, which is able to grow at 40-65 C. 19 The thermophilic properties of the enzymes from M. hassiacum are advantageous for crystallization. 20In addition, hMfr has 78% sequence identity (query coverage 87%) with its homolog from M. tuberculosis.Pronounced structural similarities of the active site architecture of hMfr and eMTHFR suggest an analogous catalytic setting of the hydride transfer reaction between NADH and methylene-H 4 F in hMfr and between FADH 2 and methylene-H 4 F in eMTHFR, respectively.To test this hypothesis, we performed a mutation analysis of the functional glutamate residue of hMfr.

| Purification and characterization of hMfr
The hMfr encoding gene was codon-optimized for the expression in E. coli, synthesized and cloned into pET-24b(+).The heterologously over-produced protein was purified to homogeneity (Table 1 and Figure 2A).The UV-visible spectrum of hMfr revealed only one peak at 280 nm, confirming the absence of FAD (Figure 2B), as reported for Mfr from M. smegmatis and M. tuberculosis. 17,18For crystallization, the purified enzyme fraction was further purified by size-exclusion chromatography.hMfr was eluted as a single symmetric peak at an elution volume corresponding to a molecular mass of approximately 30 kDa (Figure 2C).As the genome-deduced molecular mass of hMfr is 32 kDa, hMfr is present as a monomer.Kinetic measurements indicated that the hMfr activity is not dependent on FAD (Figure S1) in line with the findings for Mfr from M. smegmatis and M. tuberculosis. 17,18The apparent V max and K m values of hMfr are 18 μmole min À1 mg À1 and 160 μM for methylene-H 4 F and 17 μmole min À1 mg À1 and 16 μM for NADH, respectively (Table 2 and Figure S2).The biophysical and enzymatic properties of hMfr are in accordance with the previously published Mfr properties from M. smegmatis (see Table 2). 17hMfr shows an optimum temperature at 70 C (Figure 3A).The thermostability of hMfr was determined by

Methylene-H4F
F I G U R E 1 The reaction catalyzed by hMfr.A hydride is transferred from NADH to atom C11 of methylene-H 4 F that is part of the imidazolidine ring (blue).It has been postulated that a putative 5-iminium cation intermediate receives the hydride. 6easuring the enzyme activity after incubation of the crude extract of E. coli for 20 min at different temperatures.We observed a slight inactivation at 60 C in the assay condition and catalytic activity was undetectable after incubation at 70 C (Figure 3B).

| Crystal structure of hMfr
Purified hMfr was crystallized by the sitting-drop vapor-diffusion method.The crystal of hMfr contained one monomer in the asymmetric unit and diffracted to 1.8 Å resolution (Table 3).The phase problem was solved by the molecular replacement method using a model structure generated by Alphafold2 as template (Figure S3).The core structure of hMfr represents an β 8 α 8 or triose-phosphate isomerase (TIM) barrel fold.As typically found in TIM barrel family enzymes, several prolonged segments, inserted between the secondary structure elements, play essential functional roles.In hMfr, striking insertion segments include an α-helix bundle (212-243) after β7 and two elongated loops (loop 1: 51-69 after β2 and loop 2: 118-134 after β4) that form the outer wall of a putative active site cleft (Figure 4A).

| Comparison between the structures of hMfr and eMTHFR
Both hMfr and MTHFR are TIM barrel structures (Figure 4B) with a root-mean-square deviation (RMSD) of 4.7 Å involving 240 of 296 amino acids.Despite the low sequence similarities, the insertion Note: The cell extract was prepared from 3 g of cells (wet mass) and the enzyme was purified aerobically as described in the materials and methods section.
One unit (U) activity refers to the oxidation of 1 μmole NADH per min to form methyl-H 4 F from methylene-H 4 F. For kinetic analysis of the wild type and mutated hMfr, purification was performed by heat treatment and anion exchange using the Resource Q column (see Section 4).SDS-PAGE of the fraction is shown in Figure S5.The specific activity of the hMfr preparation purified by the shorter procedure was 14 ± 0.7 U/mg, which is similar to that of the preparation obtained by the longer procedure shown in this table.
F I G U R E 2 Biochemical characterization of hMfr purified from recombinant E. coli.(A) SDS-PAGE of 2 μg purified hMfr (see also Table 1).As a result, methyl-H 4 F could be modeled directly into the potential binding site of hMfr based on the structural alignment with the crystal structures of hMfr and eMTHFR from E. coli containing FAD and methyl-H 4 F (Figure 6). 10 The pterin ring of methyl-H 4 F confidently accommodates into the predicted pocket of hMfr.The glutamate tail after the para-aminobenzoate (PABA) ring of H 4 F interferes with the elongated segment after strand β8 (Figure 6C), which most likely changes its conformation in the hMfr-methyl-H 4 F complex.The binding of the smaller nicotinamide ring of NADH in comparison with the larger isoalloxazine of FAD appears to be optimized by narrowing the cleft via a conformational change of loop 1 (51-69) (Figure 6C).
Methylene-H 4 F and NADH were placed into the hMfr structure by in silico molecular docking (Figure S4).Their positions and conformations approximately correspond to those taken from the aligned eMTHFR- Note: The data of Mfr from M. smegmatis was obtained from the previous publication. 1713]15 Glu28 is considered as a proton donor for the formation of an iminium cation intermediate, which subsequently accepts the hydride from FADH 2 .The exchange of Glu28Gln renders the enzyme inactive and allowed the crystal structure determination of the ternary eMTHFR complex containing FAD and methyl-H 4 F. Gln183 and Asp120 form hydrogen bonds with the pterin ring atoms N1 and N3 of methyl-H 4 F, which are 3.1 and 2.9 Å long, respectively.Phe223 forms a π-π interaction with the PABA ring of methyl-H 4 F in eMTHFR.
In the hMfr structure, Glu9 and Gln177 are found in the equivalent positions of Glu28 and Gln183 of eMTHFR, respectively (Figure 7).These residues are strictly conserved among Mfr and MTHFR family members (Figure 5).In the molecular docking model (Figure S4), N10 of methylene-H 4 F is located nearby, which would allow it to interact with Glu9 via a water molecule.In the aligned sequences, the position of Asp120 of eMTHFR is replaced by Glu97 of hMfr (Figure 5).However, Glu97 is located at the beginning of α3 and points toward the solvent in the crystal structure of hMfr.Glu55 of the extended loop 1 occupies the place filled by Asp120 in eMTHFR and thereby probably stabilizes the pterin ring of methylene-H 4 F in an analogous manner.Leu221 of hMfr is found in the equivalent position of Phe223 of eMTHFR and may stabilize the PABA ring of methylene-H 4 F by hydrophobic interaction (Figure 7).
Based on this remarkably high structural relationship of the active site architecture of MTHFR and hMfr, we also assume a highly similar methylene/methyl-H 4 F binding mode in MTHFR and hMfr as shown in Figures 6 and S4.To substantiate our hypothesis, we constructed an hMfr mutant, in which Glu9 was exchanged to glutamine.In eMTHFR, the corresponding Glu28Gln mutation renders the enzyme inactive.Likewise in hMfr, the catalytic activity of the Glu9Gln variant of hMfr is reduced around 400-fold compared to that of the native enzyme, while the K m values of the substrates were only slightly changed (Table 2).This finding supported our hypothesis of a highly similar geometry between these glutamates and methyl-H 4 F in hMfr and eMTHFR, as well as a key role for these glutamates as proton donors to form an iminium ion intermediate as described by Trimmer et al. 13 It also implicates that the hydride transfer between C4 of NADH and C11 of methylene-H 4 F occurs at virtually the same place as the hydride transfer from the central pyrazine-ring moiety of FAD to methylene-H 4 F in MTHFR.Thus, the active site architecture for the NADH-to-methylene-H 4 F hydride transfer in hMfr corresponds to the reaction of the oxidative half reaction of MTHFR.
In previous studies, Mfr of M. tuberculosis was investigated by in vivo mutation experiments. 18In these experiments, the Arg159Asn and Leu214Ala mutations were introduced into a M. tuberculosis strain. 18Both strains showed impaired methionine biosynthesis and higher sensitivity to antifolates.Leu214 of Mfr from M. tuberculosis corresponds to Leu217 of hMfr, which is located next to the aforementioned Leu221 of hMfr and may play a role in interacting with the glutamate tail region of methyl H 4 F. Arg159 from M. tuberculosis corresponds to Arg162 in hMfr that is located close to the ADP part of the superposed FAD on hMfr and might play a role in binding of equivalent segments of NADH in hMfr.

| Possible inhibitors of hMfr
To further investigate, the potential of Mfr as a target enzyme for the development of antimycobacterial agents, isonicotinic acid, pyrazinoic acid, and prothionamide were tested as possible inhibitors.Isonicotinic acid and pyrazinoic acid are the active forms of the first-line drugs isoniazide and pyrazinamide 23,24 while prothionamide 25 belongs to the second-line drugs against M. tuberculosis.These three compounds resemble NADH in their structures and their mode of action is not fully elucidated yet.Therefore, they might be potential inhibitors of Mfr.However, no inhibitory effect could be detected for any of the tested compounds (Figure 8), suggesting that this set of currently used antimycobacterial agents does not affect the Mfr reaction.

| CONCLUSION
The noncanonical methylene-H 4 F reductase Mfr is a vital enzyme in mycobacteria, which includes the pathogen M. tuberculosis.Eukaryotic organisms including humans use instead the different FAD-containing MTHFR for the same reaction.Thus, Mfr has great potential for the design of antimycobacterial agents.We heterologously overproduced Mfr from the thermophilic mycobacterium M. hassiacum and characterized it as a monomeric enzyme that does not contain FAD as prosthetic group.The crystal structure of hMfr revealed a TIM barrel fold, which is also reported for the FAD-binding counterpart MTHFR.
Despite the low primary structure similarity and the different enzymatic mechanism, the tertiary structures of hMfr and MTHFR did not only show the same overall fold but also highly related binding sites for methyl-H 4 F. We mutated Glu9 of hMfr to Gln9 to mimic the Glu28Gln mutant of eMTHFR.This mutation resulted in a drastic reduction of the catalytic activity but not in a substantial loss of substrate affinity.In the assumed almost identical catalytic mechanism of MTHFR and hMfr, NADH appears to occupy the same site as FAD in eMTHFR in such a way that the hydride transfer geometry is strongly related to that of the oxidative half reaction of MTHFR.Moreover, the unexpectedly high similarities between the active sites of hMfr and MTHFR suggest a common ancestor for both methylene-H 4 F reductases.Taken together, the hMfr crystal structure and the deduced substrate binding mode can be applied for the development of inhibitors against Mfr.

| Construction of the vectors for overproduction of hMfr
The gene sequence of hMfr (Uniprot Accession: K5BDY6) was optimized for the codon usage of E. coli and synthesized by GeneScript as F I G U R E 4 Comparison of the tertiary structures of hMfr (PDB code: 8BGR) and eMTHFR (PDB code: 1ZP4). 10  For the preparation of hMfr for enzyme assays, a shorter method was applied.After the heat treatment, the supernatant was diluted 1:1 with 50 mM Mops/KOH pH 7.0 containing 2 mM DTT and the diluted protein solution was directly applied to the Resource Q column and eluted as described above.The specific activity of the wild type hMfr purified by the shorter method was similar to that shown in Table 1 (14 ± 0.7 U/mg) and SDS-PAGE showed only minor background protein bands.The corresponding fractions were collected and used for the determination of the optimum temperature, the kinetic constants, and the inhibition assays.

| Enzymological assay
The activity assay of hMfr was carried out in 1 mL quartz cuvettes filled with 0. absorbance at 340 nm was measured (ε 340 = 6.22 mM À1 cm À1 ).For the determination of the temperature dependency, the standard conditions were used and the cuvettes were incubated at different temperatures as indicated in Figure 3.For the inhibition assays, isonicotinic acid (Sigma-Aldrich), pyrazinoic acid (Sigma-Aldrich), and prothionamide (Sigma-Aldrich) were prepared as 10 mM stock solutions in anaerobic water and were added to the enzyme assay described above to a final concentration of 1 mM.As a negative control, we measured the activity of endogenous MTHFR from E. coli BL21(DE3) carrying an empty pET-24b(+) vector.
The endogenous MTHFR activity was 0.5 U/mg in cell extract and 0.2 U/mg after heat treatment under the standard assay conditions described above for measuring Mfr activity.Endogenous MTHFR was only found in flow-through fractions of the Resource Q column.
These experiments indicate that the purified hMfr fractions used for kinetic characterization do not contain the endogenous MTHFR activity from E. coli.

| Crystallization
Initially, crystallization experiments were set up at 8 C under anaerobic conditions using a N 2 /H 2 (95%/5%) gas mixture to avoid oxidation of methyl-/methylene-H 4 F. For sitting drop vapor diffusion method, During optimization, small rod-shaped crystals and thin plates emerged in several PEG solutions (PEG 1000-PEG 8000) in a concentration range between 10% and 30% (v/v) with no additives and no buffers.The best diffracting crystal was a thin plate, which grew in 15% (v/v) PEG 6000 with a final concentration of 30 mg/mL hMfr in the drop over the course of 4 weeks.The best crystal was transferred into the reservoir solution supplemented with 30% (v/v) glycerol and frozen.

| X-ray crystallographic analysis
The diffraction experiments were performed at 100 K on the beamline SLS X10SA (Villigen, Switzerland) equipped with a Dectris Eiger2 16 M detector.The dataset was processed with XDS and scaled with XSCALE. 26The phase problem was solved by the molecular replacement method using PHASER 27 with an Alphafold2 28 model of hMfr.
The Alphafold2 model was generated by the Colab Notebook 29 integrated into the CCP4 suite. 30The model was built and improved in COOT 31 and refined using Phenix.refine 32and Refmac. 33The final model was validated using the MolProbity 34 implement of Phenix.
Data collection, refinement statistics, and PDB code for the deposited structure are listed in Table 3.The figures of the structures and the ternary complex model were generated using the PyMOL version 2.4.1 (Schrödinger) and were edited using the Adobe Illustrator.

| Kinetic data processing and figure generation
The graphs and the kinetic constants analyses were performed with Python 3.7 using Jupyter Notebook (version 6.1.4) 35as development environment and the following packages: os, pandas, 36 seaborn, 37 matplotlib, 38 numpy, 39 and scipy. 40The determination of the kinetic constants is based on the experimentally determined average values of the specific activity and the corresponding standard deviation from three replicates per measurement point.The Michaelis-Menten equation (vi = vmax * substrate/(km + substrate)) was used as the corresponding function for scipy.curve_fit.The resulting standard deviations for the curve fitting process were calculated from the output of the pcov parameter from scipy.cuve_fit by calculating the diagonal of the resulting array and subsequently taking the square root of the diagonal.The chemical structures from Figure 1 were generated using ChemSketch and were edited using Adobe Illustrator.

| Molecular docking of substrates
Two different docking methods were used to dock the two substrates, methylene-H 4 F and NADH into hMfr.NADH was docked by CB-Dock2 41 into the Mfr-methyl-H 4 F complex structure where the methyl-H 4 F was transferred from the aligned eMTHFR_E28Q methyl-H 4 F complex.Methylene-H 4 F was docked into the Mfr structure using Autodock Vina implemented in AMDock. 42,43 (B) UV/Vis spectrum of purified hMfr (2.5 mg/mL) in 100 mM MOPS/NaOH pH 7.0 using a 1 cm light-path quartz cuvette.(C) Size-exclusion chromatogram of purified hMfr using a HiPrep 26/60 Sephacryl S-100 HR column.The peak at around 60 mL corresponds to hMfr.regions of hMfr and eMTHFR, forming the active site cleft, are architecturally rather related (see Figure4B).While the prolonged loop segments after β4 and β5 and those after β7 and β8 are involved in forming the putative NADH and methylene-H 4 F binding sites, respectively, the loop segment after β2 shields the binding site of both substrates.In both eMTHFR and hMfr, the loop following β4 is partially disordered.A multiple sequence alignment of Mfr and MTHFR family members also indicates a high conservation of the insertion segments at the predicted substrate binding sites (Figure5).Therefore, the binding modes of FAD and methyl-H 4 F, characterized in the substratecontaining crystal structure of eMTHFR, might largely reflect the binding mode of the substrates NADH/NAD + and methylene/methyl-H 4 F in the equivalent cleft of hMfr.
Apparent K m and V max were measured in the presence of 300 μM methylene-H 4 F or 100 μM NADH with the variable counter part of the substrates at 40 C. The assay was started by the addition of 10 μL of the hMfr solutions, which contained 1.9 mg/mL hMfr wild type enzyme or 150-200 mg/mL hMfr_E9Q.The standard deviations of the fitting analysis are shown.The original data of the experiments of hMfr are shown in Figure S1.F I G U R E 3 Thermophilic properties of hMfr.(A) Temperature dependency of the activity.(B) Residual activity in the cell extract after heat treatment for 20 min.Every data point consists of three independent measurements.The error bars represent the corresponding standard deviation.

4 . 2 |
The synthesized gene was cloned into pET-24b(+) between the NdeI and SalI restriction sites.The resulting plasmid pET-24b(+)_hMfr was transformed into chemically competent E. coli BL21(DE3) STAR cells (Thermo Fisher Scientific).The recombinant E. coli strain was cultivated at 37 C overnight in 100 mL of Lysogeny Broth (LB) medium as preculture.Forty milliliter of the preculture were used to inoculate two liters of prewarmed Terrific Broth (TB) medium supplemented with 50 μg/mL kanamycin sulfate in a 2 L flask (Thermo Fisher Scientific).The culture was stirred with 750 rpm and cultivated at 37 C until it reached an OD 600 of 0.6-0.8.Then the gene expression was induced by the addition of 1 mM (final concentration) of isopropylβ-D-1-thiogalactopyranoside (IPTG, Sigma-Aldrich).After 3 h of induction at 37 C, the cells were cooled down in an ice bath and centrifuged for 5 min at 13 000g.The cell pellets were stored at À20 C. Purification of hMfr All purification steps took place under a normal aerobic environment except for the preparation of crystals.The additional size-exclusion chromatography was performed in an anaerobic chamber (Coy) filled with a N 2 /H 2 (95%/5%) atmosphere.Around 3 g of cells were suspended 1:5 in 50 mM Mops/KOH pH 7.0 supplemented with 2 mM dithiothreitol (DTT, Sigma-Aldrich) and disrupted by sonication in an ice bath using a SONOPULS GM200 (Bandelin) with a KE73 tip with a 50% cycle and 160 W (2 min sonication time with 2 min pauses; in total 5 times).The cell debris were separated by centrifugation at 30000g for 30 min.To separate the thermostable hMfr from the E. coli proteins, the cell extract was incubated for 20 min at 50 C in a water bath and subsequently centrifuged for 20 min at 13 000g at F I G U R E 5 Primary structure comparison between Mfr and MTHFR.The secondary structure elements above and below the multiple sequence alignment refer to the crystal structure of eMTHFR (PDB: 1ZP4) and hMfr (PDB: 8BGR), respectively.Strictly conserved residues are depicted as white letters with red background.Residues with similar physicochemical properties are depicted as red letters in blue-lined boxes.The strictly conserved and catalytically essential glutamate residue is marked with a red triangle below the corresponding position.Residues that interact directly with methylene-H 4 F via hydrogen bonds in the ternary complex structure of MTHFR are indicated by blue triangles below the corresponding position.For the sake of readability, only the catalytic domains of MTHFR from Homo sapiens (residues 1-343) and Saccharomyces cerevisiae (residues 1-302) were included.The alignment was generated using MUSCLE,21 and the figure was created using ESPript 3.0.22F I G U R E 6 Predicted substrate binding sites of hMfr.(A) Cartoon and surface models of the complex structure of eMTHFR_E28Q (orange) in complex with methyl-H 4 F and FAD (PDB code: 1ZP4).The transparency of the surface is 10%.(B) Cartoon and surface models of the structure of hMfr.Methyl-H 4 F and FAD from the superimposed eMTHFR_E28Q complex are transferred and represented by orange colored carbon atoms.(C) Zoom-up views of methyl-H 4 F (left panel) and FAD (right panel) models in their protein environments.saturation 40% and the mixture was incubated under constant stirring in an ice bath for 20 min and then centrifuged for 20 min at 13 000g.The resulting supernatant was applied to two tandemly connected 5 mL HiTrap Phenyl Sepharose HP columns (Cytiva) equilibrated with 50 mM Mops/KOH pH 7.0 containing 2 mM DTT and 1.6 M ammonium sulfate.The column was washed with 1.1 M ammonium sulfate in the equilibration buffer.Elution proceeded in a single step with 0.7 M ammonium sulfate in 50 mM Mops/KOH pH 7.0 containing 2 mM DTT.The eluted protein fraction was pooled and desalted with 50 mM Mops/KOH pH 7.0 containing 2 mM DTT using a HiPrep 26/10 desalting column containing Sephadex G-25 resin.The desalted sample was applied to a 6 mL Resource Q column equilibrated with the same buffer.Elution took place with a linear gradient from 0 to 250 mM NaCl over 15 column volumes.hMfr eluted at a conductivity between 12 and 14 mS/cm.The corresponding fractions were pooled, concentrated with an Amicon filter (10 kDa cutoff, Merck) and applied to a HiPrep 26/60 Sephacryl S-100 HR column equilibrated with 20 mM MOPS/KOH pH 7.0 containing 2 mM DTT, 150 mM NaCl and 5% (v/v) glycerol.hMfr eluted at around 60 mL retention volume.The corresponding fractions were collected and used for the preparation of anaerobic crystal plates and the measurement of the UV/vis spectrum of hMfr.

7
7 mL of 100 mM Mops/NaOH pH 7.0 under an N 2 atmosphere and incubated at 40 C.In the standard assay, NADH (Roth) and methylene-H 4 F (Schircks Laboratories) were added to final concentrations of 100 and 300 μM, respectively.For the kinetic analysis, the concentration of NADH and methylene-H 4 F was set as the standard condition and the concentration of the counter substrate was varied, as shown in Figure S1.The assay was started by the addition of 10 μL of the hMfr solutions, which contained 1.9 mg/mL wild type hMfr or 150-200 mg/mL hMfr_E9Q variant.The decrease of NADH Conservation of the predicted active site residues of hMfr and of eMTHFR_E28Q.The residues and methyl-H 4 F derived from the crystal structure of the ternary eMTHFR_E28Q complex are shown with orange carbons and orange letters, while the residues from the hMfr apoenzyme are shown with blue carbons and blue letters.F I G U R E 8 The effect of different antimycobacterial agents on hMfr.Every bar represents the average of three independent measurements.The error bars are the corresponding standard deviation.A final concentration of 1 mM of the respective compound was used in the assays.
96-well screening plates (MRC 2 Well UVP, Hampton Research) were filled with 90 μL of reservoir solution and the 48-well optimization plates (MRC Maxi 48-Well Crystallization Plate (Swissci), Hampton Research) with 200 μL of reservoir solution.The crystallization plates filled with reservoir solutions were set up in the chamber 1 week prior to crystallization to make the solution anaerobic.First, crystals were found in several solutions of JB Screen Classic 1 (Jena Bioscience); for example, 25% polyethylene glycol (PEG) monomethyl ether 2000.
Later, Sah et al. found that the enzymes encoded by MSMEG_6596 and MSMEG_6649 in Mycobacterium smegmatis are noncanonical MTHFRs that do not contain FAD or any other prosthetic group but exhibit methylene-H 4 F reductase activity (Figure

1
Purification of hMfr expressed in E. coli.
Kinetic properties of the hMfr wild type, hMfr_E9Q mutant, and Mfr proteins from M. smegmatis (MSMEG_6596 and MSMEG_6649).
FAD-methyl-H 4 F-FAD structure.The distance between the hydridedonating C4 atom of NADH and the hydride-accepting C11 atom of methylene-H 4 F in the docking model is 3.2 Å. T A B L E 2 ray analysis statistics.
aValues relative to the highest resolution shell are within parentheses.b R free was calculated as the R work for 5% of the reflections that were not included in the refinement.