CHARMM-GUI QM/MM Interfacer for a Quantum Mechanical and Molecular Mechanical (QM/MM) Simulation Setup: 1. Semiempirical Methods

Abstract

Quantum mechanical (QM) treatments, when combined with molecular mechanical (MM) force fields, can effectively handle enzyme-catalyzed reactions without significantly increasing the computational cost. In this context, we present CHARMM-GUI QM/MM Interfacer, a web-based cyberinfrastructure designed to streamline the preparation of various QM/MM simulation inputs with ligand modification. The development of QM/MM Interfacer has been achieved through integration with existing CHARMM-GUI modules, such as PDB Reader and Manipulator, Solution Builder, and Membrane Builder. In addition, new functionalities have been developed to facilitate the one-stop preparation of QM/MM systems and enable interactive and intuitive ligand modifications and QM atom selections. QM/MM Interfacer offers support for a range of semiempirical QM methods, including AM1(+/d), PM3(+/PDDG), MNDO(+/d, +/PDDG), PM6, RM1, and SCC-DFTB, tailored for both AMBER and CHARMM. A nontrivial setup related to ligand modification, link-atom insertion, and charge distribution is automatized through intuitive user interfaces. To illustrate the robustness of QM/MM Interfacer, we conducted QM/MM simulations of three enzyme-substrate systems: dihydrofolate reductase, insulin receptor kinase, and oligosaccharyltransferase. In addition, we have created three tutorial videos about building these systems, which can be found at https://www.charmm-gui.org/demo/qmi. QM/MM Interfacer is expected to be a valuable and accessible web-based tool that simplifies and accelerates the setup process for hybrid QM/MM simulations.

Figure 1

Schematic overview of the QM/MM system building using QM/MM Interfacer.

Figure 2

Ligand modification in QM/MM Interfacer. (A) Ligand selection, visualization, and modification through 2D-sketchpad. (B) Two different embedding schemes: (left) matching elements and bond orders and (right) matching any atom. The modified ligand is represented by a magenta-colored ball-and-stick model.

Figure 3

QM region selection and charge distribution scheme in QM/MM Interfacer. (A) Atom selection web-implementation. (B) n-Link charge distribution scheme, where Q_X refers to any QM atom and M_X refers to any MM atom. The QM-MM boundary is defined along the Q₁ and M₁ bond, with an introduced H-link atom indicated by L. The charge of the M₁ atom is set to zero, and the residual charge of the QM region and the M₁ atom is evenly distributed among the M₂ atoms directly bonded to the M1 atom. (C) Automatically generated charge distribution table.

Figure 4

(A) The overall structure of the E. coli DHFR enzyme in complex with H₂folate and NADPH (PDB ID: 1rx2). The protein is shown as a color-coded cartoon, the substrate H₂folate as a yellow stick, and the NADPH cofactor as a purple stick. For the enzyme, the M20 (residues 9–23), C–D (residues 64–71), and βF-βG (residues 116–132) loops are shown in blue, orange, and purple, respectively. (B) Schematic of the hydride transfer reaction catalyzed by DHFR, where the arrow indicates the transfer of a hydride from NADPH to H₃folate⁺.

Figure 5

(A) Comparison of calculated potentials of mean force (PMF) profiles between the AM1-QM/MM and SRP-AM1-QM/MM models, where the reaction coordinate ξ is defined as the difference of the distances of the transferred hydride (H_R) from the donor C₄ of NADPH to the acceptor C₆ of H₃folate⁺ in Figure 4B. (B) Average distances and errors of the distances of the transferring hydride (H_R) from the donor (C₄) and the acceptor (C₆) carbon atoms, determined from each umbrella sampling window of the original AM1-QM/MM (black) and SRP-AM1-QM/MM (red) simulations. The transition state (TS) region is indicated by a bar. (C–D) Snapshots taken from the TS region: (C) the original AM1-QM/MM simulation and (D) the SRP-AM1-QM/MM simulation. The color scheme for (C) is the same as in Figure 4, and in (D) slightly lighter colors are used for each loop and ligand. The transferred hydride (H^–) is shown as a green sphere and the hydrogen-link atoms are shown as a white stick (Figure S1).

Figure 6

Comparison of IRK structures: (A) the initial structure and (B–C) snapshots after 500 ps MD simulation of (B) AM1-QM/MM and (C) AM1/d-PhoT-QM/MM systems. The protein is shown in a gray cartoon, the A-loop in a blue tube, and the αC-helix in a green cartoon. The A-loop and αC-helix for the AM1-QM/MM and AM1/d-PhoT-QM/MM simulations are shown in their respective lighter colors. The inset of each figure shows an enlarged view of ATP, Mg²⁺ ions, substrate Tyr, and Mg²⁺-coordinating IRK residues. For clarity, the Mg²⁺-coordinated waters are not shown. In each figure, the H-link atoms are indicated by a purple sphere.

Figure 7

Time series of IRK RMSD, key distances of the reacting groups, and angle formed by the nucleophilic Tyr-OH, P_γ, and O_3β atom of ATP: (left panel) AM1-QM/MM and (right panel) AM1/d-PhoT-QM/MM simulations. The key distances are for the Tyr-HH atom to the oxygen atom of Asp1132, which is hydrogen bonded, and the Tyr-OH atom to the P_γ atom of ATP. (B) Schematic of the IRK-catalyzed reaction in which Asp1132 acts as a general base to deprotonate the substrate Tyr residue. The deprotonated Tyr residue then nucleophilically attacks the P_γ atom of ATP, producing ADP and phosphorylated Tyr as the product.

Figure 8

QM/MM equilibrated structures of C. lari OST-membrane systems from PDB ID: (A) 5ogl and (B) 6gxc. The membrane system is shown in a ball-and-stick model, the protein OST in a green cartoon, and the substrate peptide in a blue tube, respectively. The protein and substrate residues, LLO, Mn²⁺ ion, and its coordinating waters are shown in a ball-and-stick model.

Figure 9

Two proposed mechanisms of N-linked glycosylation within OST. (A) The reactive Asn becomes a better nucleophile through hydrogen bonding with Asp56 and Glu319. (B) Deprotonation at the Asn precedes the nucleophilic attack, forming an imidate tautomer. The activated N atom of Asn then undergoes a nucleophilic attack at the C1 atom of LLO, resulting in a cleavage of the C1-OX bond, producing the reaction product.