Binding studies of a putative C. pseudotuberculosis target protein from Vitamin B12 Metabolism

Vitamin B12 acts as a cofactor for various metabolic reactions important in living organisms. The Vitamin B12 biosynthesis is restricted to prokaryotes, which means, all eukaryotic organisms must acquire this molecule through diet. This study presents the investigation of Vitamin B12 metabolism and the characterization of precorrin-4 C(11)-methyltransferase (CobM), an enzyme involved in the biosynthesis of Vitamin B12 in Corynebacterium pseudotuberculosis. The analysis of the C. pseudotuberculosis genome identified two Vitamin B12-dependent pathways, which can be strongly affected by a disrupted vitamin metabolism. Molecular dynamics, circular dichroism, and NMR-STD experiments identified regions in CobM that undergo conformational changes after s-adenosyl-L-methionine binding to promote the interaction of precorrin-4, a Vitamin B12 precursor. The binding of s-adenosyl-L-methionine was examined along with the competitive binding of adenine, dATP, and suramin. Based on fluorescence spectroscopy experiments the dissociation constant for the four ligands and the target protein could be determined; SAM (1.4 ± 0.7 µM), adenine (17.8 ± 1.5 µM), dATP (15.8 ± 2.0 µM), and Suramin (6.3 ± 1.1 µM). The results provide rich information for future investigations of potential drug targets within the C. pseudotuberculosis’s Vitamin B12 metabolism and related pathways to reduce the pathogen’s virulence in its hosts.

. Molecular structure of Vitamin B 12 and reaction mechanism of CH 3 -group transfer by the precorrin-4 C(11)-methyltransferase (CobM). (A) Structure of Vitamin B 12 and B 12 -derived cofactors. In red, the methyl group transferred by CobM. (B) Reaction mechanism of CobM, the cofactor is SAM which donate the methyl group (in red) for the transfer reaction.
www.nature.com/scientificreports www.nature.com/scientificreports/ Methylmalonyl CoA mutase requires Ado-Cbl as a cofactor and catalyzes the final step in the methylmalonyl pathway, the reversible isomerization of R-methyl-malonyl-CoA to succinyl CoA, by doing so; it fulfills a critical function in modulating intracellular pools of propionyl-CoA, a toxic product of the catabolism of odd-and branched-chain fatty acids and cholesterol 20 . Methionine synthase uses the cofactor Met-Cbl to catalyze the conversion of homocysteine to the essential amino acid methionine 21 . Many bacteria like M. tuberculosis or E. coli possess a Vitamin B 12 -independent methionine synthase, MetE (5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase), which they employ for the same methyltransferase reaction and that represents a backup system for the Vitamin B 12 -dependent system 22 . The MetE gene can be located in the genome of several Corynebacterium species, e.g. C. glutamicum, C. efficiens, but not in C. pseudotuberculosis genomes deposited in the GenBank-NCBI database.
The Vitamin B 12 -dependent proteins such as the Vitamin B 12 -dependent ribonucleoside-diphosphate reductase, ethanolamine ammonia-lyase, and Vitamin B 12 import system permease protein BtuC are not conserved in the C. pseudotuberculosis genome, but are conserved in the genome of other Corynebacterium species.
Enzyme production and purification. Cp-CobM has a molecular weight of 28.974 kDa (272 amino acids).
After two step purification, the purity of the protein was confirmed on a denaturating SDS-PAGE ( Supplementary  Fig. S4A). Dimer conformation of the protein was confirmed by size exclusion chromatography ( Supplementary  Figs S4B, S5).

Investigation of the protein secondary structure under the influence of ligands. The far-UV
CD spectrum of Cp-CobM is characterized by the presence of two relative minima at 208 and 220 nm ( Fig. 2A) indicating a higher percentage of helical content than β-strand in the protein secondary structure.
The effect of SAM, adenine, dATP, and suramin on the protein secondary structure was tested, indicating a change in the protein secondary structure composition (Fig. 2B).
Binding studies between Cp-CobM and sAM, dAtp, adenine, and suramin using Nuclear Magnetic Resonance (NMR). STD-NMR experiments revealed that SAM, dATP, adenine, and the polyanion suramin interact specifically with Cp-CobM. Off-resonance and difference spectra were observed for each ligand independently. SAM has a distinct signal observed in the STD difference spectra, at δ8.14 ppm (Fig. 3A). Calculating the relative ratio between off-resonance and difference spectrum signal area is possible to assess the  www.nature.com/scientificreports www.nature.com/scientificreports/ binding epitopes and, thus, the degree of proximity between ligand atoms and the protein binding site. The closer the ligand atom is to the protein, the more saturation it will receive and, as a result, the higher its signal intensity in the difference spectrum. Binding epitopes for SAM are located at the adenine section of the SAM molecule. Accordingly, dATP and adenine were tested by STD NMR and the binding epitopes were located in the same area, when compared with SAM ( Fig. 3B,C). For SAM and dATP, hydrogen bound to carbon 2 is seen in the difference spectrum, while other atom signals are not present. Signals from the ribose hydrogens are not observed. On the other hand, for adenine, the highest STD-AF was seen for hydrogen bound to carbon 8 (atom #1 in Fig. 3C) and a hydrogen bound to carbon 2 signal is present. Binding epitopes for suramin were found in the naphthalene groups and the aromatic rings of the molecule (Fig. 3D).
However, the binding epitopes indicate differences; the dATP signal observed in the STD difference spectra at δ8.33 ppm (Fig. 3B), has no significant magnetization transfer. These differences can explain the role of the interactions of the methionine (SAM) or phosphate groups (dATP). The methionine molecule of SAM interacts with CobM proteins as demonstrated in Fig. 3A. Schubert et al. 1998, described the binding of a phosphate ion in B. megaterium-Cbif SAM binding site 10 , this phosphate ion interacts with His100 (His79 in Cp-CobM) and the backbone of Thr101 (Ser80 in Cp-CobM). The similarity of B. megaterium-Cbif in this region ( Supplementary Fig. S6) makes the binding of the phosphate group of dATP possible.
NMR competition experiments were performed to determine if adenine, dATP, and suramin compete for the SAM binding site or influence the SAM binding with Cp-CobM. As a result, ligands competing by the same binding site, which influences saturation transfer from the protein, and differences in STD-AF are observed. In the competition experiments between SAM and dATP, or adenine and suramin, the signals from SAM are not observed in the difference spectrum. The adenine STD effect is reduced by approximately 19% (δ8.09 ppm: −18%; δ8.03 ppm: −20%) and the STD effect for dATP is reduced by approximately 11% (δ8.11 ppm: −11%). These results indicate significant competition between SAM, adenine, and dATP for the same binding site of Cp-CobM.
In contrast, competition with SAM reduced the suramin STD effect by only −0.06% (Supplementary  Table S2), which is statistically insignificant when compared to the noise in the spectrum. This is probably because suramin does not interact with the SAM binding site of Cp-CobM. Suramin is a symmetric divalent polyanion with two naphthalene-trisulfonic acid head groups carrying the strong negatively charge that interacts with positively charged regions on protein surfaces and may cause changes in the protein secondary structure 23-25 . Determination of the dissociation constants between Cp-CobM and sAM, dAtp, adenine, and suramin by fluorescence spectroscopy. Interaction between Cp-CobM with SAM, adenine, dATP and suramin was investigated using intrinsic tryptophan (Trp) fluorescence approach. The maximum emission of the intrinsic tryptophan fluorescence was centered at 303 nm. The Results of the Cp-CobM fluorescence quenching after ligand titration were used to determine the dissociation constant (K d ), a nonlinear saturation curve approach and a modified Hill equation (Supplementary Fig. S7) was combined 26 .
The determined K d values for all tested ligands showed the following order SAM > suramin > dATP > adenine demonstrated in Table 2.
Interestingly, the quenching behavior of the Trp residues in Cp-CobM shows differences between the tested ligands. Therefore, SAM, dATP, and adenine showed a similar quenching pattern. In contrast, suramin had a completely different effect on the Trp quenching; the difference in the quenching can be explained by the difference www.nature.com/scientificreports www.nature.com/scientificreports/ of the suramine size and the localization of the binding area. The STD NMR competition experiments showed that SAM, adenine, and dATP compete for the same binding area, the already known SAM binding site. These molecules are small and will not exert a strong influence on the Trp quenching. Suramin is a big molecule and can induce structural changes 33 in the protein or, can serve as a shield to the Trp residues; both effects can induce the strong Trp quenching.

sequence analysis and homology modeling of Cp-CobM. A BLAST search for the Cp-CobM
sequence against the atomic coordinates held in the PDB demonstrated the highest sequence identity of 51% with the R. capsulatus-CobM whose three-dimensional structure has been determined (PDB: 3NEI) and 42% identity with B. megaterium-CbiF (PDB: 1CBF), the RMSD between both structures is 1.079 (1157 atoms).
Sequence alignments of C. pseudotuberculosis-CobM, R. capsulatus-CobM and B. megaterium-CbiF indicate the high conservation of regions important for function, such as the site of SAM/SAH interaction and the precorrin binding site, as well as residues in the dimer interface region (Fig. 4A).
A Cp-CobM homology model was generated based on the structure of the R. capsulatus-CobM, which was chosen based on the sequence identity of 51%. To obtain the protein models with the best quality for Cp-CobM, Cp-CobM-SAM and Cp-CobM SAH a cluster analysis was performed. For this analysis, the k-means method was used ranging from two to six. The correct cluster with the representative conformation was chosen by a combined analysis of the lowest Davies-Bouldin index (DBI) value and the highest pattern sequence based forecasting (pSF) (Supplementary Table S3). Additionally, the silhouette analysis (SI) was used to verify the best-formed cluster. To choose the representative conformation of the model, the distribution inside the cluster and the stability of the molecule over the MD simulation was used (supplementary Table S4). Calculation of the root mean square deviation (RMSD) and Radius of gyration (Rg) of Cα was determined to demonstrate the equilibrium and confluence of the systems ( Supplementary Fig. S8).
The comparison of the Cp-CobM homology model with the R. capsulatus-CobM structure is presented ( Supplementary Fig. S9). The Cp-CobM dimer is formed by two protein monomers that are twisted by 45° (Fig. 4B). The CobM monomer is composed of two α/β domains linked by a single coil forming a kidney shaped molecule 10 . The cavities formed between both domains are the SAM/SAH and the precorrin binding site. Both domains contain a five stranded β-sheet flanked by four α-helices 10 (Fig. 4C).

MD simulation of Cp-CobM complexed with SAM and SAH indicate dynamical flexibility dur-
ing cofactor binding. The SAM/SAH binding pocket of CobM proteins is located between the N-and C-terminal domains of the monomer. The residues involved in SAM/SAH binding are conserved between the species and include Pro10, Asp83, Ser112, Leu165, and Ala193 demonstrated for the R. capsulatus-CobM structure in complex with SAH. The protein ligand interaction is shown in a Ligplot figure (Fig. 5A) and demonstrates that the SAM interaction is stabilized by ten hydrogen bonds and eleven hydrophobic contacts. The adenine, ribose, and methionine parts of SAM are directly involved in the interaction (Fig. 5A). The NMR results demonstrated that adenine and dATP compete with SAM for the same binding site in the protein and there is a high possibility that they interfere mainly with the amino acids that interact with the SAM adenine part.
Docking experiments of SAH and SAM ligands in the Cp-CobM homology model and subsequently MD simulations indicated the interactions of amino acids involved in the SAH and SAM binding (Fig. 5B,C).
Comparison of the decomposition energies of both ligands indicates significant differences between SAM and SAH binding with Cp-CobM (Fig. 5D); these variations depend on the orientation of the ligand in the binding pocket and the number of hydrogen bonds formed. The largest differences can be observed for Pro10, Ser112, Leu165, Arg219 and Ala221. Depending on the ligands (SAH or SAM) for residues Ile36, Ser81, Asp83, Ile86, Leu165, Arg219, Thr220, and Ala221, a change in the coordination can be observed (Fig. 5B,C).
In the analog CbiF protein from B. megaterium, the transfer of the methyl group from SAM to precorrin-4 results in the release of SAH and precorrin-5, requiring significant conformational rearrangements in the www.nature.com/scientificreports www.nature.com/scientificreports/ enzyme 10 . To observe these conformational rearrangements in Cp-CobM model with and without ligands the Root Mean Square Fluctuation (RMSF) of the Cα atoms during the MD simulations were followed (Fig. 6).
The secondary structure and RMSF fluctuation of Cp-CobM and Cp-CobM-SAH are similar, but when compared with Cp-CobM-SAM complex they show significant differences in three regions that contain amino acids involved in the cofactor binding ( Fig. 6B-D).
Since SAM binding initiates an opening movement of the active site pocket, the putative precorrin-4 binding region becomes more accessible. Changes in the secondary structure of CobM were also observed during the CD experiments. A comparison of the secondary structure content of Cp-CobM and Cp-CobM-SAM complex using CD spectroscopy and MD simulations showed a decreased amount of α-helices after binding of the SAM molecules and a slight increase in random coiled elements. The data for β-sheets are slightly different between CD spectroscopy and MD simulations (Supplementary Table S5).
The RMSF analysis identified two protein regions, which undergo significant fluctuations. These areas consist of loop regions and the movements are not considered to be induced by ligand binding (Fig. 6, peaks labeled by asterisks).
The SAM/SAH binding area and the amino acids involved in the interaction to the cofactor undergo conformational changes during binding process. Figure 7A shows an overlay of the Cp-CobM amino acids involved in the binding with and without ligands and demonstrates the molecular movements after ligand interaction.
The strongest effect can be observed in Ile36 (Fig. 7A,B) located in the loop (Region I) with strong fluctuations during cofactor binding, which was described in Fig. 5. Interestingly, there is a displacement of approx. 4.42 Å after protein/cofactor binding (Fig. 7B colored by blue) related to the initial position (Fig. 7B side chain colored by white). During the SAH interaction, the same residue flips back 2.9 Å to the direction of the initial position ( Fig. 7B colored by green). We suppose that Ile36 interacts with the CH 3 group of SAM, which is transferred to precorrin-4. The distance between the amino acid and SAM CH 3 group is 3.8 Å, which makes a hydrogen bond unlikely, but hydrophobic interaction possible. Figure 7C Fig. S10).
The supplementary video S1 shows the dynamics of the protein after binding to the cofactor.

Conclusion
The complex biosynthesis of Vitamin B 12 involves several enzymes and one of these enzymes, Precorrin-4 C(11)-methyltransferase (CobM), functions in the aerobic pathway. The sequence of C. pseudotuberculosis CobM shares 40-50% sequence identity with the structures of CobM from R. capsulatus and B. megaterium.
CobM exists physiologically as a dimer (Fig. 4B) and the two α/β domains of the monomer are linked by a short peptide (Cp-CobM Val107 to Ser112). The SAM/SAH binding site is located in a shallow cavity 10 ; our results demonstrated upon binding of SAM that a relative hinge movement between the two domains occurs and this "opens" a cavity conformation, which is a prerequisite for precorrin-4 binding. A methyl group is extracted from SAM producing SAH followed by a further opening of the SAM/SAH binding site that allows the exit of SAH and precorrin-5.
MD simulations of a Cp-CobM homology model, CD, and NMR STD experiments all in the presence and in the absence of the ligands, identified regions in the protein that go through conformational rearrangements to facilitate the entrance of precorrin-4. CobM structure undergoes a conformational change that principally affects the access to the catalytic site increasing the accessibility. SAM, dATP, adenine, and suramin interactions with Cp-CobM were also investigated using STD-NMR. Results of NMR competition experiments showed that adenine and dATP compete with SAM for the same binding site and, as revealed by CD spectroscopy, cause similar structural changes. Suramin affects SAM binding to CobM, however, since the competition effect with SAM is reduced, a different mechanism is likely involved.
The dissociation constants between Cp-CobM and the four ligands could be determined using fluorescence spectroscopy.
Understanding the conformational changes that this enzyme undergoes after cofactor and substrate binding sheds light on the molecular basis of the protein functionality giving perspective for drug development to combat caseous lymphadenitis.

Material and Methods
In silico analysis. The genome sequence of C. pseudotuberculosis strain 1002 (bv. ovis) and C. pseudotuberculosis strain CIP52.97 (bv. equi), both deposited in the NCBI GenBank database (http://www.ncbi.nlm.nih.gov/genbank/) were analyzed with the enzymes involved in the Vitamin B 12 syntheses and Vitamin B 12 -dependent enzymes in perspective. Multiple CobM/CbiF sequences were retrieved from NCBI and sequence alignments were performed using MUSCLE 34 and Box Shade (http://www.ch.embnet.org/software/BOX_form.htm) web servers. The protein sequence from Cp-CobM was retrieved from UniProt database (http://www.uniprot.org/) (D9QAC4).  www.nature.com/scientificreports www.nature.com/scientificreports/ The Cp-CobM containing cell-suspension was mixed with lysozyme and incubated on ice for 1 h. Afterwards, the cell-suspension was lysed by sonication in four sets of 30 s pulses of 30% amplitude with 10 s intervals. This method produced the crude cell extract, which was centrifuged under 8,000 rpm at 6 °C for 90 min. The supernatant containing Cp-CobM was loaded onto a Ni-NTA gravity flow column pre-equilibrated with 20 mM Tris, pH 7.5, 500 mM NaCl, 10% (v/v) glycerol, 1 mM DTT. The Ni-NTA column with bounded Cp-CobM was washed extensively with the same buffer containing 20, 40, and 80 mM imidazole. Cp-CobM was eluted with imidazole concentrations of 250 and 500 mM. The elution fractions containing Cp-CobM were pooled and applied onto a size exclusion column (Superdex 75 10/300 GL -GE Healthcare, USA), pre-equilibrated with buffer containing 20 mM K 2 HPO 4 /KH 2 PO 4, pH 7.5, 150 mM NaCl, 1 mM DTT. Protein purity was evaluated by 15% SDS-PAGE. The amount of recombinant Cp-CobM produced was around 20 mg/L. Superdex 75 10/300 GL size exclusion column (GE Healthcare, USA) was calibrated using Albumin (MW: 65 kDa), Proteinase K (MW: 29.7 kDa) and Lysozyme (MW: 14.3 kDa). The column pre-equilibration was performed using a buffer solution containing: 20 mM K 2 HPO 4 /KH 2 PO 4, pH 7.5, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, and 0.5 mM PMSF.

Circular dichroism spectroscopy (CD).
For all CD measurements 15 repeated scans were performed with 5 of them used to establish the baseline. The wavelength range applied for far-UV spectra was from 200 nm to 260 nm in a time constant of 1 s and 100 nm/min continuous scanning mode, using a Jasco J-107 spectropolarimeter (Jasco, Japan). Cp-CobM was separately diluted in 5 mM K 2 HPO 4 /KH 2 PO 4 to a concentration of 3.1 µM to investigate the influence of SAM, adenine, dATP, and suramin. The protein was incubated with a double molar excess (6.2 µM) for 2 h prior to the measurements. The results are presented in molar ellipticity [θ], according to: where θ is the ellipticity measured at a given wavelength λ (deg), c is the protein concentration (mol L −1 ), l is the cell path length (cm) and, n is the number of amino acids. The results were analyzed and secondary structure amount determined using the CDpro software package 35 .

Saturation transfer difference by nuclear magnetic resonance (STD-NMR). Saturation Transfer
Difference (STD) identifies binding events that occur in the rapid exchange of the saturation transfer 36 , i.e., interactions that possess a dissociation constant (K d ) of µM to mM. Ligands in this range are of special relevance in www.nature.com/scientificreports www.nature.com/scientificreports/ the analysis of molecules that modulate protein function, rather than completely inhibiting it 37 or, in preliminary screening of ligands for fragment-based drug design 38,39 .
NMR data was collected in a Bruker AVANCE III HD (Bruker, Germany), operating at 600 MHz for one hour and equipped with triple resonance cryoprobe with a pulsed field gradient. Bruker pulse sequence STDDIFFESGP.3 was used to perform the STD experiments with protein suppression by a spin lock filter. Each experiment was set with 64 scans, saturation time of two seconds (with four seconds of recycle delay) applied by a Gaussian pulse at 35 dBW with an acquisition time of 1.9 seconds in a slide window of 14 ppm. Cp-CobM saturation was set to 0.0 ppm, in order to keep a safety distance of 1000 Hz from any ligand signal, while off-resonance was set to 20 ppm. Protein concentration was 20 µM while the concentrations of the ligands SAM, dATP, adenine, and suramin were set to 400 µM. Samples were prepared in 20 mM K 2 HPO 4 /KH 2 PO 4 pH 7.5, 150 mM NaCl with 10% D 2 O and transferred to 5 mm NMR tubes (Norell ® Sample Vault Series ™ ). STD effect is, calculated as: Where I off and I on are the integral of each ligand signal in the off-and on-resonance spectra, respectively. Data was collected in the STD experiment with each ligand individually and then for the competition between SAM and the other tested ligands.
Fluorescence spectroscopy. Fluorescence spectroscopy were used to determine the K d values for Cp-CobM with suramin, SAM, adenine, and dATP. A combination of a nonlinear saturation curve approach and a modified Hill equation were used to determine the K d values, previously described in Coronado et al. 26 .
To measure the fluorescence of the intrinsic tryptophan we used a quartz cuvette with 3 mm path lengths (105.253-QS, Hellma, Mühlheim, Germany) and the experiments were performed on a QuantaMaster40 spectrofluorometer (PTI, Birmingham, USA). The background intensities were corrected for all spectra. The excitation and emission wavelength was chosen in 295 nm and in the range of 285-400 nm, respectively. The emission spectrum was recorded with increments of 1 nm and, each point in the emission spectrum represents the average of 10 accumulations. 10 μM final concentration of the protein was used in a volume of 1.5 ml in a buffer containing 20 mM K 2 HPO 4 /KH 2 PO 4, pH 7.5, 150 mM NaCl, 1 mM DTT. The interaction between suramin, SAM, adenine, and dATP with Cp-CobM was investigated. The titration was performed stepwise with a ligand stock solution (1.0 mM + 10 μM protein) and the measurement was performed after each titration. To fit a saturation binding curve 40 , the quenching of the Cp-CobM fluorescence, ΔF (F max − F), at 303 nm for each titration point was used, based on equation (1) min d F min is the minimal fluorescence intensity in the presence of ligand; K d is the equilibrium constant for the protein-ligand complex. The "binding constant" K is defined as the reciprocal of K d .

Molecular dynamics and computational analysis. Initial structures. The atomic coordinates of R.
capsulatus-CobM (PDB: 3NEI; sequence identity 51%) were used as a template for comparative modeling by the satisfaction of spatial restraints as implemented in the program Modeller 9v13 44 .
To obtain a high quality model of Cp-CobM, the protein structure was solvated in water and the system was neutralized with ion molecules. Subsequently, Molecular Dynamics (MD) simulations was performed (100 ns) and cluster analysis were used to generate a stable and representative Cp-CobM model. AutoDock Vina 1.1.12 45 was used to perform docking studies for Cp-CobM-SAH and Cp-CobM-SAM. Polar hydrogens and partial charges were added to the protein using AutoDockTools program 46 , additionally, the rotational bonds in the ligands were defined.
The gridbox was used to define the search space near the ligand binding site. For each calculation, Autodock Vina scoring function ranked several structural poses. Cp-CobM-SAH and Cp-CobM-SAM complexes were used to initiate the MD simulations. Dimers were prepared by fitting the complexes models into the crystal structure, the biological form of the enzyme.
To prepare the ligands, Gaussian 09 47 was used at the level of theory HF/6-31 G* to optimize and calculate their electrostatic potentials (ESP). Subsequently, the restrained electrostatic potential (RESP) charges were determined using antechamber program 48 and the missing general amber force field (GAFF) parameters 49 were obtained using parmchk program.
Parameterization for molecular dynamics simulations. The general setup for the Cp-CobM homology simulations in native form and with ligands were performed with some modification as described in Coronado et al. 26 .
MD simulations were performed using the AMBER16 program 50 , all atoms protein interaction was described using FF14SB force field 51 , while GAFF and RESP charges describe SAH and SAM ligands. The protonation state was settled up using the H++ web-server 52 , at pH 7.4. The experimental systems were neutralized with Na + ions and settled in an octahedral or rectangular box of TIP3P water to at least 10 Å from any protein atom. Two steps www.nature.com/scientificreports www.nature.com/scientificreports/ of energy minimization for each system was performed to remove bad contacts from the initial structures. First, the energy minimization of the protein-complexes constrained (force constant of 50.0 kcal/mol.Å²) was achieved with 5,000 steepest descent followed by 5,000 conjugate gradient steps and by unconstrained energy minimization rounds (10,000 steps). After energy minimization, the system was gradually heated from 0 to 293 K for 300 ps under the constant number, volume, and temperature (NVT) ensemble, while the protein was restrained with a force constant of 25 kcal/mol.Å². Subsequently, an equilibration step was performed using the constant atom number, pressure, and temperature (NPT) ensemble for 2 ns. Finally, the production was run for 200 ns for each system and performed in NVT ensemble not having any restraints. The constant temperature (293 K) and pressure (1 atm) were controlled by Langevin coupling. The SHAKE constraints were applied to all bonds involving hydrogen atoms to allow a 2-fs dynamics time step. Long-range electrostatic interactions were calculated by the particle-mesh Ewald method (PME) 53 using 8 Å cutoff.
Structural and dynamical analyses. The results were analyzed by the CPPTRAJ program 54 of the AmberTools17 package. The system was visualized using VMD 55 and Pymol 56 programs. Root mean square deviation (RMSD) and Radius of gyration (Rg) of Cα were calculated to determine the system quality and stability ( Supplementary  Fig. S8).
Clustering analysis was performed with k-means method ranging from 2 to 6. To access the quality of clustering we used the DBI values and silhouette analyses. Protein flexibility was studied by Root Mean Square Fluctuation (RMSF) for the Cα atoms. The RMSF was calculated residue-by-residue over the equilibrated trajectories.
The generalized Born (GB)-Neck2 57 implicit solvent model (igb = 8) was used to determine the interaction energy between protein and ligands.
The calculated Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) energy showed the stability between the protein and the different ligands comprising the last 10 ns of the MD simulation, stripping all the solvent and ions.
The volume of the Cp-CobM SAM/SAM binding pocket was determined using the web tool POCASA1.1 58 .

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.