Critical Evaluation of Polarizable and Nonpolarizable Force Fields for Proteins Using Experimentally Derived Nitrile Electric Fields

Jacob M. Kirsh, Jared Bryce Weaver, Steven G. Boxer, Jacek KozuchJournal of the American Chemical Society 2024 · DOI

TL;DR

This paper critically evaluates how well different molecular dynamics (MD) force fields (FFs) can simulate the local electric fields within proteins. By comparing experimental measurements of electric fields using a nitrile vibrational probe with simulations from both fixed-charge (AMBER) and polarizable (AMOEBA) FFs, they find that polarizable FFs are significantly better at capturing these crucial electrostatic interactions, especially in hydrogen-bonding environments. This work highlights the importance of using more sophisticated FFs for accurate modeling of protein electrostatics.

What Problem Does This Paper Address?

Molecular dynamics (MD) simulations are powerful tools for understanding protein function, and electrostatics play a critical role in processes like protein folding, binding, and catalysis. However, the accuracy of these simulations heavily relies on the chosen force field (FF), which describes the interactions between atoms. A major challenge has been the lack of experimental methods to directly benchmark the local electric fields predicted by FFs within proteins, especially in complex environments like those involving hydrogen bonds. Previous attempts to use vibrational frequencies of nitrile probes were complicated by hydrogen bonding, leading to ambiguous results. This paper addresses this gap by employing a novel approach using the transition dipole moment (TDM) of nitrile probes, which, as previously shown by the authors, provides a more robust measure of electric fields, even in the presence of hydrogen bonding. They aim to compare these experimental electric field values with those simulated by two common FFs: the fixed-charge AMBER and the polarizable AMOEBA.

Methods

The core of this study involves comparing experimental electric field measurements with MD simulations.

  • Experimental Measurement: The researchers utilized a previously established method where nitrile vibrational probes (specifically, o-cyanophenylalanine, or oCNF, incorporated into photoactive yellow protein, PYP) act as sensors for local electric fields. The key innovation is using the transition dipole moment (TDM) of the nitrile's vibrational mode, which correlates linearly with the electric field. This overcomes limitations of using vibrational frequencies, which are more sensitive to hydrogen bonding. They derived experimental electric field values (FC=N,TDM) for four different oCNF-substituted PYP variants.
  • Molecular Dynamics Simulations: They performed MD simulations of these PYP variants using two distinct FFs:
  • AMBER ff99SB-ILDN: A widely used fixed-charge (FC) FF that only considers atomic partial charges.
  • AMOEBA (AMBER-Polarizable): A polarizable (POL) FF that includes charges, dipoles, quadrupoles, and polarizabilities, allowing for more nuanced electrostatic descriptions. Simulations were run for 100-200 ns per variant.

  • Electric Field Extraction: From the MD trajectories, they extracted the local electric fields experienced by the nitrile group (FC=N,MD).

  • Analysis: The study then systematically compared the experimentally derived FC=N,TDM values with the computationally derived FC=N,MD values from both FFs. They analyzed the distributions of simulated electric fields, deconstructing them into hydrogen-bonding (HB) and non-hydrogen-bonding (non-HB) populations. Further analysis involved examining H-bond geometries (distance and angle) and electrostatic potential (ESP) maps of the nitrile group to understand the FF-dependent differences.

Datasets

The primary "dataset" used for evaluation consists of experimental electric field measurements derived from the vibrational Stark effect (VSE) of nitrile probes incorporated into four specific photoactive yellow protein (PYP) variants (F280CNF, F620CNF, F920CNF, F960CNF). These experimental values are then compared against the output of MD simulations performed on the same protein structures using the AMBER and AMOEBA force fields. No new benchmark datasets were explicitly introduced, but the experimental data itself serves as a crucial benchmark for evaluating the FFs.

Key Insights and Ideas

  • Polarizable FFs are Crucial for H-Bonding: The most significant insight is that polarizable force fields (like AMOEBA) are essential for accurately capturing local electric fields in proteins, particularly when hydrogen bonding is involved. Fixed-charge FFs (like AMBER) tend to underestimate these fields in such environments.
  • Nitrile TDM as a Robust Probe: The paper reinforces the utility of the nitrile TDM as a reliable experimental probe for electric fields, overcoming the ambiguities previously associated with using vibrational frequencies in hydrogen-bonding contexts.
  • FFs Dictate H-Bond Geometry: The choice of FF dramatically influences the simulated geometry of hydrogen bonds involving the nitrile. Polarizable FFs lead to more "head-on" and shorter H-bonds, which better aligns with the expected electron distribution of the nitrile.
  • Origin of FF Differences: The improved performance of AMOEBA is attributed to its inclusion of higher-order multipole moments and polarizability, which results in a more focused negative charge density along the nitrile's bond axis. This more accurately models the nitrile's lone pair and its interactions.
  • Implications for Drug Discovery: The findings suggest that for computational screening of nitrile-containing drugs (where H-bonding interactions are critical for ligand binding), polarizable FFs like AMOEBA should be preferred, despite their higher computational cost.

Key Results

  • AMOEBA Recapitulates Experiment Better: The polarizable AMOEBA FF showed significantly better agreement with the experimentally derived TDM electric fields (FC=N,TDM) compared to the fixed-charge AMBER FF.
  • AMBER Underestimates Fields: For nitriles experiencing moderate field strengths and especially those involved in hydrogen bonding, the AMBER FF consistently underestimated the electric fields.
  • Significant FF Dependence in H-Bonding: The electric field distributions for nitriles in H-bonding environments were substantially affected by the FF choice. For instance, the H-bonding fraction predicted by AMOEBA was up to 6.5 times higher than that predicted by AMBER for some variants.
  • Quantitative Correlation: Weighted linear regressions showed that the correlation between FC=N,TDM and FC=N,MD had slopes of 0.56 ± 0.04 for AMBER and 1.17 ± 0.09 for AMOEBA, indicating that AMOEBA's simulated fields were much closer to the experimental values (a slope near 1 signifies good agreement).
  • Geometric Differences: Average H-bond distances were shorter by up to 0.1 Å and angles were larger with AMOEBA compared to AMBER for the oCNF nitriles.

Limitations and Future Work

  • Computational Cost: The authors acknowledge that performing polarizable MD simulations (like with AMOEBA) is computationally more expensive than fixed-charge simulations. However, they note that advances in hardware (GPUs) and software are making these simulations more feasible.
  • Generalizability: While this study focused on nitriles in PYP, the authors suggest that further work is needed to explore the FF dependence of other functional groups and protein environments. They specifically mention backbone carbonyls as a promising target for future benchmarking.
  • Engineered Perturbations: The paper concludes by suggesting that future work could involve further engineered local perturbations to nitrile environments to gain deeper insights into FF capabilities. They also plan to report and analyze these within the framework developed in this study.

Summary updated

Abstract

Molecular dynamics (MD) simulations are frequently carried out for proteins to investigate the role of electrostatics in their biological function. The choice of force field (FF) can significantly alter the MD results, as the simulated local electrostatic interactions lack benchmarking in the absence of appropriate experimental methods. We recently reported that the transition dipole moment (TDM) of the popular nitrile vibrational probe varies linearly with the environmental electric field, overcoming well-known hydrogen bonding (H-bonding) issues for the nitrile frequency and, thus, enabling the unambiguous measurement of electric fields in proteins (<i>J. Am. Chem. Soc.</i> <b>2022</b>, <i>144</i> (17), 7562-7567). Herein, we utilize this new strategy to enable comparisons of experimental and simulated electric fields in protein environments. Specifically, previously determined TDM electric fields exerted onto nitrile-containing <i>o</i>-cyanophenylalanine residues in photoactive yellow protein are compared with MD electric fields from the fixed-charge AMBER FF and the polarizable AMOEBA FF. We observe that the electric field distributions for H-bonding nitriles are substantially affected by the choice of FF. As such, AMBER underestimates electric fields for nitriles experiencing moderate field strengths; in contrast, AMOEBA robustly recapitulates the TDM electric fields. The FF dependence of the electric fields can be partly explained by the presence of additional negative charge density along the nitrile bond axis in AMOEBA, which is due to the inclusion of higher-order multipole parameters; this, in turn, begets more head-on nitrile H-bonds. We conclude by discussing the implications of the FF dependence for the simulation of nitriles and proteins in general.
BibTeX
@article{kirsh2024critical,
  author = {Jacob M. Kirsh and Jared Bryce Weaver and Steven G. Boxer and Jacek Kozuch},
  title = {{Critical Evaluation of Polarizable and Nonpolarizable Force Fields for Proteins Using Experimentally Derived Nitrile Electric Fields}},
  journal = {Journal of the American Chemical Society},
  year = {2024},
  doi = {10.1021/jacs.3c14775},
}
Figure