速報 (Selected Papers)
Nonlocal Modulation of Antigen–Antibody Interactions Underlying A107R-Mediated Affinity Gain in VHH D2-L29: MD Simulations with MM-PBSA–Based Thermodynamic Analysis
2025 年 24 巻 3 号 p. 77-79
詳細
2025 年 24 巻 3 号 p. 77-79
Single-domain VHH antibodies offer attractive developability, but rational affinity improvement requires atomistic insight into how sequence changes reshape binding. We investigated the A107R substitution in VHH D2-L29 bound to hen egg-white lysozyme using long-timescale molecular dynamics simulations with MM-PBSA–based thermodynamic analysis. Computed binding free energies recapitulated the experimentally observed increase in affinity. Per-residue decomposition of the change in binding free energy indicated that residue 107 (A107R) and Tyr100 had the largest favorable changes, whereas Asn102 had the largest unfavorable change. Together, the results indicate that molecularly detailed, interface-wide analysis of nonlocal interaction changes is essential for rationally engineering VHH variants with enhanced affinity.
Single-domain VHH antibodies (nanobodies) are compact, highly specific binders that often exhibit remarkable stability and solubility, making them attractive for next-generation diagnostics and therapeutics [1]. Computationally guided design efforts have successfully identified affinity-enhancing substitutions in single-domain antibodies [2]. Here, rather than relying on extensive screening, we focus on a mechanism-based strategy for rational mutant design to increase VHH affinity.
However, predicting the thermodynamic impact of a single substitution remains challenging. Even a single substitution can alter the structure and dynamics of the antigen–antibody interface; thus, the effect may extend beyond the mutated site and propagate across the interface [3]. In this study, we aim to clarify the influence of the A107R substitution in VHH D2-L29 bound to hen egg-white lysozyme (HEL). In a previous study [2], A107R was reported to increase affinity. To investigate the mutation’s effect on the antigen–antibody interface at the molecular level, we combine long-timescale molecular dynamics (MD) simulations with binding free-energy calculations based on the molecular mechanics Poisson–Boltzmann surface area (MM-PBSA) approach.
We investigated two complexes—the wild-type (WT) VHH D2-L29 bound to hen egg-white lysozyme (HEL) and its A107R variant—and performed three independent replicas for each system. Starting coordinates were obtained from the Protein Data Bank (PDB ID 1ZV5) [4], and the A107R substitution (Ala→Arg at position 107) was introduced on the WT complex using Discovery Studio Visualizer (BIOVIA). (See Fig. 1.) Proteins were described with the AMBER99SB-ILDN force field and water with TIP3P; each complex was solvated and Cl− ions were added to neutralize protein charges in dodecahedral boxes. As a result, the WT system comprised the D2-L29(WT)-HEL complex, 19,253 water molecules, and 11 Cl− ions, whereas the A107R system comprised the D2-L29(A107R)-HEL complex, 19,258 water molecules, and 10 Cl− ions.
Crystal structure of the hen egg-white lysozyme (HEL)–VHH D2-L29 (WT) complex (PDB ID: 1ZV5 [4]). HEL is shown in purple and the VHH in green. The black arrow indicates Ala107 (shown in ball-and-stick representation).
After steepest-descent minimization, the systems were equilibrated for 300 ps under heavy-atom restraints and then simulated for 600 ns per replica (three replicas per system). Temperature (298 K) and pressure (1 bar) were maintained with the Nose–Hoover thermostat and the c-rescale barostat. Long-range electrostatics were treated with particle-mesh Ewald (PME), and all covalent bonds were constrained with LINCS. The time step was set to 3 fs.
As suggested by the backbone RMSDs, the final 150 ns of each trajectory were used for the following binding free energy analysis. Binding free energies were estimated with the MM-PBSA method [5]. We used the single-trajectory protocol, evaluating complex, receptor, and ligand energies from the same coordinate frames to suppress noise from uncorrelated sampling. The binding free energy was evaluated as the sum of molecular-mechanics terms (van der Waals, electrostatics) and solvation terms (polar Poisson–Boltzmann and non-polar SASA-based); entropic contributions were excluded. Per-residue energy decomposition was performed under identical settings for WT and A107R to quantify local contributions. In this study, we used the implementation defaults for the Poisson–Boltzmann and non-polar parameters.
All MD simulations were performed with GROMACS 2024.1 [6], and MM-PBSA calculations were carried out using gmx_MMPBSA [7].
MM-PBSA predicted ∆Gb values of −70.0 ± 3.9 kcal mol−1 (WT) and −83.6 ± 2.8 kcal mol−1 (A107R), indicating stronger binding for the variant; this trend is consistent with experimental measurements [2]. Intuitively, one might expect that the mutation-induced change in binding affinity would be dominated solely by interfacial interactions in the vicinity of residue 107 (A107R); however, per-residue decomposition of the change in binding free energy (∆∆Gb) indicated that residue 107, Tyr100, Tyr110, Asp44, and Trp111 all became more favorable upon the A107R substitution, whereas Asn102, Tyr103, Gln105, Arg45, and Phe37 became less favorable (Table 1). In particular, the change in the per-residue contribution to binding for Tyr100 was −1.9 kcal mol−1, comparable to that for residue 107 (−1.6 kcal mol−1). For Tyr100, the favorable change in electrostatics (−3.0 kcal mol−1) was partially offset by the change in polar solvation (+2.0 kcal mol−1). For residue 107 (A107R), the unfavorable change in electrostatics (+17.7 kcal mol−1) was overcompensated by favorable changes in polar solvation (−15.8 kcal mol−1) and in van der Waals interactions (−2.8 kcal mol−1). Taken together, affinity enhancement arises from a nonlocal reorganization of interactions across the interface rather than a single contact, underscoring the importance of atomistic, interface-wide energetic analysis.
| Favorable changes | Unfavorable changes | ||
| Residue a | ∆∆Gb b | Residue a | ∆∆Gb b |
| Tyr100 | −1.9 ± 2.3 | Asn102 | +1.4 ± 0.5 |
| Ala107 c | −1.6 ± 0.3 | Tyr103 | +0.8 ± 0.5 |
| Tyr110 | −0.9 ± 0.2 | Gln105 | +0.6 ± 0.5 |
| Asp44 | −0.7 ± 0.8 | Arg45 | +0.5 ± 0.5 |
| Trp111 | −0.7 ± 0.4 | Phe37 | +0.4 ± 0.1 |
a Residues are labeled by their WT identities. b Changes upon the A107R substitution. Values represent the mean ± standard error of the mean (SEM) from three independent MD replicas and are reported in kcal mol−1. c Mutation site in WT.
In this study, MD simulations with MM-PBSA calculations reproduced the experimental trend and revealed that A107R increases affinity via nonlocal interfacial rebalancing. These findings show that detailed MD-based analysis is essential for the rational optimization of VHH antibodies.
This research used computational resources of TSUBAME (25IAO) and Research Center for Computational Science, Okazaki, Japan.
