COMT Accommodates Ligands via an Induced-Fit Mechanism
Previous crystal structures of the rat COMT isoform show minor conformational differences between apo-COMT and holo-COMT 
. Therefore, we performed DMD simulations of COMT in the absence of all substrates to determine whether the conformation of the human isoform remains identical to the crystal structure. An alignment of the crystal structure and simulated structure reveals few differences in the tertiary structure between the two with an overall root mean square deviation (RMSD) of 1.5 Å (). Secondary structural elements are completely retained in the absence of ligands. The most notable difference lies within the catechol-binding loop, which migrates closer to the active site in the absence of ligands.
Structures of COMT open and closed conformations.
Although the overall structure of COMT remains identical in the absence of ligands, the solvent accessibility of the active site changes (). In the crystal structure, COMT forms two pockets that accommodate the adenosine and methionine side-chains of SAM. The adenosine pocket partially collapses, and the methionine-binding motif completely closes in the absence of SAM. Comparison of the two active sites shows two main conformations of COMT: an open and a closed state. Within the closed state of COMT, SAM cannot bind inside the active site and therefore methylation cannot occur. To accommodate SAM, the COMT must initially open so that SAM can access the active site. Since only the adenosine-binding motif is partially open in the closed COMT structure, this region of the SAM cofactor may be responsible for initial binding and perhaps induces the open state. Because the closed COMT state makes the active site inaccessible, all docking simulations are performed with the open state of COMT under an assumption of induced fit. Within the open state, we refer to the absence of ligands as apo-COMT and the presence of all ligands as holo-COMT.
Structural Characterization of SAM Conformations within Apo-COMT and Holo-COMT
We determine the binding poses of SAM in the absence and presence of a divalent cation metal by performing docking simulations within the open state of COMT using MedusaDock. In the absence of metal and substrate, portions of SAM bind to a different groove of COMT (). The adenosine moiety of SAM remains identical to the holo-COMT structure, with the Ile91 side-chain packing on top of the pyrimidine portion of adenine and the imidazole of His142 participating in a perpendicular edge-to-face aromatic interaction with the pyrimidine (). The imidazole of adenine participates in an additional edge-to-face interaction with the indole side-chain of Trp143. Additional polar contacts are made between residues 118-120 and the purine nitrogens. The two hydroxyl groups of the ribose participate in hydrogen bonding interactions with the side-chain of Glu90. A single water molecule from the active site satisfies an additional hydrogen bond requirement of the amine bound at C6 of the adenosine motif.
Binding configurations of SAM with and without metal.
Similarities between the structures of SAM docked inside apo-COMT and holo-COMT end at the sulfonium center. In the apo-COMT complex, the terminal amine group of SAM is involved in a hydrogen-bonding network with the side-chains of Asp141 and Asn170 and the backbone carbonyl of Met40 (). The terminal carboxyl group participates in hydrogen bonding with Lys144 and a single water molecule. In the holo-COMT complex, the primary role of this water molecule is to occupy a coordination site of the divalent metal cation. Its secondary role is to form a hydrogen bond with the carboxyl terminus of the methionine side-chain (). In absence of the metal, satisfying the hydrogen bond of the carboxyl terminus becomes its primary role. Yet, its role is non-essential for the overall conformation of SAM in apo-COMT since Lys144 can satisfy both hydrogen bond requirements of the carboxyl terminus alone. The interactions highlighted here with methionine underlie an important point for why catalysis cannot occur without a metal. Lys144 is responsible for deprotonating the catechol to create the oxyanion responsible for attacking the methyl group. In this particular pose, the Lys144 is preoccupied in a hydrogen-bonding network and is unavailable to deprotonate.
Prior to catechol binding, a divalent metal cation must first displace the positively charged amine group. Several divalent cations are capable of contributing to enzymatic activity, although the native metal is magnesium in vivo 
. Here we modeled the divalent metal cation using Zn2+
, which is 80% as effective as magnesium. We find that upon metal binding, steric occlusion prevents the methionine of SAM from binding to the negatively charged pocket and is forced into an interior groove. The amine group of the methionine maintains a hydrogen bond with Asp141, albeit at a different position, but additionally forms hydrogen bonds with the backbone carbonyl of Gly66 and the side-chain of Ser72. The amide backbones of Ser72, Val42, and an additional water molecule form hydrogen bonds with the carboxyl group of SAM. Additional hydrophobic interactions are formed between the methionine side-chain and residues 40, 42, 60, 68, and 89 inside this pocket ().
The results found here are initially surprising because it is expected that SAM binds to apo-COMT as found in the crystal structure of the holoenzyme complex. The amine and carboxyl tail of SAM form favorable van der Waals contacts and satisfy their hydrogen bonds in both conformations. However, the carboxylate side-chains within the active site preferentially bind to the amine side-chain of SAM, as the polar contacts are stronger than those shared with the backbone carbonyl groups of holo-COMT.
Most of the lowest energy docking poses generated for apo-COMT by MedusaDock are identical () and deviate from the crystal structure (). However, the SAM conformation found for apo-COMT could potentially be an artifact of the MedusaDock/MedusaScore suite. To test the validity of our docking results, we utilize the docking program Glide (see Methods
) to score the poses found for apo-COMT. In our initial docking test, we docked the crystal structure of COMT to SAM. Because Glide only allows flexibility to be modeled within the ligand, the lowest energy pose found was identical to the crystal structure (with an energy of −13.45 kcal/mol). We adjust for the lack of flexibility within the active site side-chains by redocking SAM to the MedusaDock-generated structure of COMT. Remarkably, we find that apo-COMT conformation of SAM is recapitulated with Glide (). Furthermore, this conformation is scored with a lower energy (−15.93 kcal/mol) compared to the crystal structure conformation.
Analysis of SAM docking poses generated by MedusaDock and Glide.
Role of Divalent Metal Cation in Methyltransferases
The role that magnesium plays in COMT activity has remained unclear to date. Without any metal bound to COMT, methylation is 6.2% as effective as compared to with all cofactors present 
. Reports from the crystal structure suggest that Mg2+
is necessary for correctly aligning the oxyanion of the catechol substrate with the carbocation of SAM. However, it is unknown what the exact coordination complex of Mg2+
is in vivo
due to the use of inhibitors in all holo-COMT crystal structures. These inhibitors usually contain electron-withdrawing groups on the catechol to lower the nucleophilicity of the reactive oxygen atom. Molecular dynamic simulations using a natural catechol substrate show catechol as a monodentate ligand of Mg2+ 
. The hydroxyl that coordinates with Mg2+
is a further topic of debate.
Here we suggest that a divalent metal cation may also be essential for structural rearrangements of the SAM cofactor. In our model, the orientation of the methionine without metal present positions the donating methyl group 6.1 Å away from the oxyanion, compared to a separation of 2.7 Å in the presence of metal (). Furthermore, the oxyanion and sulfonium no longer form a 180° angle in between the methyl group. This angle decreases to ~45° in the absence of metal. Therefore the probability of SAM methylating the catechol is lowered.
Our results described here are derived purely from computation. Thus, crystallization of COMT in the absence of catechol substrate would be required to validate our mechanism. Two structures would be needed to support our mechanism: 1) COMT crystallized with SAM (or S-adenosyl-homocysteine) to show the alternate binding pose; 2) COMT crystallized with SAM and metal to show SAM in its holo-COMT conformation. Agreement between our docking poses and the proposed crystal structures would demonstrate the existence of this alternative SAM conformation in apo-COMT, and furthermore, show that the metal is required for proper SAM binding.
The mechanism proposed here is of broad biological interest since SAM serves as a common methyl donor for many methylation reactions, including CpG methylation 
. The binding site of SAM on COMT is homologous with many SAM-dependent methyltransferase structures, with several unique residues that form part of the catechol-binding site. Thus, our mechanism could be applicable to a broad range of methyltransferases that require a divalent metal cation and SAM.