How the stabilization of INK4 tumor suppressor 3D structure evaluated by quantum chemical and molecular mechanics calculations corresponds well with experimental results: interplay of association enthalpy, entropy, and solvation effects

The folding free energy of the INK4c tumor suppressor core, consisting of 10 helices, was determined as the sum of gas-phase interaction enthalpy, gas-phase interaction entropy, and dehydration and hydration free energy. The interaction energy and the hydration free energy were determined using the nonempirical density functional theory (DFT) method, augmented by a dispersion-energy correction term, the semiempirical density-functional tight-binding method covering the dispersion energy, and the density functional theory/conductor-like screening model (DFT/COSMO) procedure, whereas the interaction entropy was calculated with the empirical Cornell et al. force field. Alternatively, all contributions were evaluated consistently using empirical methods. All the values of the interaction energy of helix pairs are stabilizing, and the dominant stabilizing terms stem from the London dispersion energy and, in the case of charged systems, the electrostatic energy. The stabilization energy of the core, determined as the difference of the energy of the core and 10 separate helices, amounts to approximately 450 kcal/mol. Systematically, the difference in the hydration free energy of a helix pair and its separate components is smaller in magnitude than the interaction energy, and it is negative for some pairs while positive for others. The average total free energy of a core formation amounts to -29.6 kcal/mol (yielded by scaled quantum-chemical methods) and +13.9 kcal/mol (resulting from empirical methods). These values are considerably smaller than their single components, which are dominated by the interaction energy. The computationally predicted interval encloses the experimental value of the folding free energy (-2.8 kcal/mol).     

The design, synthesis and evaluation of high affinity macrocyclic carbohydrate inhibitors

Carbohydrate-protein interactions have been investigated for a model system of a monoclonal antibody, SYA/J6, which binds a trisaccharide epitope of the O-polysaccharide of the Shigella flexneri variant Y lipopolysaccharide. The thermodynamics of binding for the methyl glycoside of the native trisaccharide epitope, Rha-Rha-GlcNAc () to SYA/J6 over a range of temperatures exhibits strong, linear enthalpy-entropy compensation and a negative heat capacity change (DeltaC(p)=-152 cal mol(-1) degree(-1)). At 293 K the free energy of association is the sum of favourable enthalpy and entropy contributions (DeltaH=-3.9 kcal mol(-1) and -TDeltaS=-2.9 kcal mol(-1)). Crystal structures for SYA/J6 Fab detailed the position of the native trisaccharide epitope, Rha-Rha-GlcNAc, and facilitated a strategy to design a tighter binding, low molecular weight ligand. This involved pre-organization of the native trisaccharide in its bound conformation by addition of intramolecular constraints (a beta-alanyl or glycinyl tether). ELISA measurements indicated that the glycinyl tethered trisaccharide was not an optimal candidate for further analysis, while microcalorimetry provided data showing that the beta-alanyl tethered trisaccharide displayed a 15-fold increase in affinity for SYA/J6. Tethering resulted in a favourable entropic contribution to binding, relative to the native trisaccharide (-TDeltaDeltaS=-1.2 kcal mol(-1)). Potential energy and dynamics calculations using the AMBER Plus force fields indicated that trisaccharide adopted a rigid conformation similar to that of the bound conformation of the native trisaccharide epitope. While this strategy resulted in modest free energy gains by minimizing losses due to conformational entropy, thermodynamic data are consistent with significant contributions from solvent reorganization.     

An Artificial Receptor for Dimethyl Aspartate

[trans-5,15-Bis(2,7-dihydroxy-1-naphthyl)-2,3,7,8,12,13,17,18-octaethylporphyrinato]zinc(II) (1), a trifunctionalized porphyrin host, was prepared as a receptor for amino acid derivatives, particularly those having a hydrogen-bonding site in the side chain. The free energy changes for the binding of Leu-OMe, Asp-OMe, and Glu-OMe to 1 were -5.8 kcal/mol, -6.6 kcal/mol, and -5.9 kcal/mol, showing a selectivity for Asp-OMe. (1)H NMR titration experiments indicated that three simultaneous attractive interactions, one coordination interaction, and two hydrogen-bonding interactions, are operating in the host-guest complex. The preference for Asp-OMe over Glu-OMe was found to originate from the favorable enthalpy term for Asp-OMe. The free energy change, the enthalpy change, and the entropy change were determined and split into contributions arising from coordination interaction and from hydrogen-bonding interactions by use of reference hosts. Comparison of enthalpy and entropy changes suggests that the host-guest complex becomes more ordered as the number of recognition pairs increases.     

Enthalpies of hydration of N-methylacetamide by one, two, and three waters and the effect upon the C=O stretching frequency. An Ab initio DFT study

DFT calculations at the B3LYP/D95++(d,p) level of clusters of N-methylacetamide (NMA) with one, two, and three waters that were geometrically optimized on the counterpoise-corrected potential energy surfaces show that the gas-phase enthalpy of the interactions of NMA with three waters is -14.11 kcal/mol. Since the interactions between the three waters is 0.99 kcal/mol, the interaction enthalpy would become -15.10 if these interactions were subtracted. The internal geometry of the NMA molecule is sensitive to the degree of hydration, as are the H-bond lengths. Changes in the internal bond lengths with degree of hydration are approximately additive. The calculated C=O stretching frequencies correlate extremely well with the calculated C=O bond lengths, which suggests that the solvent effect upon this stretch could not be a purely electrostatic interaction. The calculated C=O stretch for NMA solvated by three waters in the gas phase agrees very well with that experimentally observed in aqueous solution.     

Comparison of 6-31G*-based MST/SCRF and FEP evaluations of the free energies of hydration for small neutral molecules

A comparison between Miertus–Scrocco–Tomasi (MST) SCRF and free energy perturbation (FEP) estimates of the free energy of hydration of eight small neutral molecules is presented. In both cases, the 6-31G* molecular electrostatic potential is used to describe the electrostatic properties of the molecules. The results demonstrate the ability of both methodologies to provide useful theoretical estimates of the total free energy of hydration; the average errors are only 1.5 kcal/mol (FEP) and 0.8 kcal/mol (MST/SCRF). The largest errors in the FEP and MST/SCRF results are less than 1.5 kcal/mol for all molecules except acetic acid, where the FEP method overestimates the free energy of hydration by 3.3 kcal/mol. © John Wiley & Sons, Inc.     

Expanding functionality of RNA: synthesis and properties of RNA containing imidazole modified tandem G-U wobble base pairs

Imidazole modification at C-5 of uridine that is part of tandem G-U wobble base pairs causes slight reduction of thermal stability (DeltaDeltaG(0)(310) < 0.4 kcal mol(-1)) and relatively small change in hydration of short RNA helices.     

Charge optimization increases the potency and selectivity of a chorismate mutase inhibitor

The highest affinity inhibitor for chorismate mutases, a conformationally constrained oxabicyclic dicarboxylate transition state analogue, was modified as suggested by computational charge optimization methods. As predicted, replacement of the C10 carboxylate in this molecule with a nitro group yields an even more potent inhibitor of a chorismate mutase from Bacillus subtilis (BsCM), but the magnitude of the improvement (roughly 3-fold, corresponding to a DeltaDeltaG of -0.7 kcal/mol) is substantially lower than the gain of 2-3 kcal/mol binding free energy anticipated for the reduced desolvation penalty upon binding. Experiments with a truncated version of the enzyme show that the flexible C terminus, which was only partially resolved in the crystal structure and hence omitted from the calculations, provides favorable interactions with the C10 group that partially compensate for its desolvation. Although truncation diminishes the affinity of the enzyme for both inhibitors, the nitro derivative binds 1.7 kcal/mol more tightly than the dicarboxylate, in reasonable agreement with the calculations. Significantly, substitution of the C10 carboxylate with a nitro group also enhances the selectivity of inhibition of BsCM relative to a chorismate mutase from Escherichia coli (EcCM), which has a completely different fold and binding pocket, by 10-fold. These results experimentally verify the utility of charge optimization methods for improving interactions between proteins and low-molecular weight ligands.     

Polar group enhanced gas-phase acidities of carboxylic acids: an investigation of intramolecular electrostatic interaction

We studied the effects of polar groups on the gas-phase acidities of carboxylic acids experimentally and computationally. In this connection, the gas-phase acidities (DeltaH(acid), the enthalpy of deprotonation, and DeltaG(acid), the deprotonation free energy) of borane-complexed methylaminoacetic acid ((CH(3))2N(BH(3))CH(2)CO(2)H) and methylthioacetic acid (CH(3)S(BH(3))CH(2)CO(2)H) were measured using the kinetic method in a flowing afterglow-triple quadrupole mass spectrometer. The values of DeltaH(acid) and DeltaG(acid) of (CH(3))2N(BH(3))CH(2)CO(2)H were determined to be 328.8 +/- 1.9 and 322.1 +/- 1.9 kcal/mol, and those of CH(3)S(BH(3))CH(2)CO(2)H were determined to be 325.8 +/- 1.9 and 319.2 +/- 1.9 kcal/mol, respectively. The theoretical enthalpies of deprotonation of (CH(3))2N(BH(3))CH(2)CO(2)H (329.2 kcal/mol) and CH(3)S(BH(3))CH(2)CO(2)H (325.5 kcal/mol) were calculated at the B3LYP/6-31+G(d) level of theory. The calculated enthalpies of deprotonation of N-oxide-acetic acid (CH(3)NOCH(2)CO(2)H, 329.4 kcal/mol) and S-oxide-acetic acid (CH(3)SOCH(2)CO(2)H, 328.6 kcal/mol) are comparable to the experimental results for borane-complexed methylamino- and methylthioacetic acids. The enthalpy of deprotonation of sulfone-acetic acid (CH(3)SO2CH(2)CO(2)H, 326.1 kcal/mol) is about 2 kcal/mol lower than the S-oxide-acetic acid. The calculated enthalpy of deprotonation of sulfoniumacetic acid, (CH(3))2S+CH(2)CO(2)H, is 243.0 kcal/mol. Compared to the corresponding reference molecules, CH(3)NHCH(2)CO(2)H and CH(3)SCH(2)CO(2)H, the dipolar group and the monopolar group substituted carboxylic acids are stronger acids by 11-14 and 97 kcal/mol, respectively. We correlated the changes of the acidity upon a polar group substitution to the electrostatic free energy within the carboxylate anion. The acidity enhancements in polar group substituted carboxylic acids are the results of the favorable electrostatic interactions between the polar group and the developing charge at the carboxyl group.     

Detailed considerations for a balanced and broadly applicable force field: a study of substituted benzenes modeled with OPLS-AA

Modern classical force fields have been traditionally parameterized by attempting to maximize agreement to any number of experimental and/or quantum mechanical target properties. As these force fields are pushed towards obtaining quantitative estimates of often subtle energetic differences, stringent and consistent parameterization criteria, particularly in regard to charge distributions, are required to ensure that systematic errors cancel, that parameters are transferable between molecules, and that performance does not significantly deteriorate when using more approximate methods, such as with continuum solvent models. Relative free energies of hydration are presented here for 40 mono- and disubstituted benzenes modeled with the OPLS-AA force field; heats of vaporization and pure liquid densities at standard conditions are presented when experimental data is available. Overall agreement between OPLS-AA and experiment is remarkable (average error = 0.5 kcal/mol for DeltaDeltaG(hydration), 1.0 kcal/mol for DeltaH(vap) (0), 0.02 g/mL for densities), yet several functional groups are identified as having consistent and correctable errors (alkyl-, nitro-, and thiobenzenes). Relative free energies of hydration obtained with rigorous free energy perturbations using explicit solvent are also compared with energies from minimizations using a generalized Born model (GB). There is high correlation between these estimates (R = 0.99), and as demonstrated here, reparameterization of the aforementioned groups can be guided with rapid GB calculations.     

Reaction mechanism and tautomeric equilibrium of 2-mercaptopyrimidine in the gas phase and in aqueous solution: a combined Monte Carlo and quantum mechanics study

A combined Monte Carlo and quantum mechanical study was carried out to analyze the tautomeric equilibrium of 2-mercaptopyrimidine in the gas phase and in aqueous solution. Second- and fourth-order M?ller-Plesset perturbation theory calculations indicate that in the gas phase thiol (Pym-SH) is more stable than the thione (Pym-NH) by ca. 8 kcal/mol. In aqueous solution, thermodynamic perturbation theory implemented on a Monte Carlo NpT simulation indicates that both the differential enthalpy and Gibbs free energy favor the thione form. The calculated differential enthalpy is DeltaH(SH)(-->)(NH)(solv) = -1.7 kcal/mol and the differential Gibbs free energy is DeltaG(SH)(-->)(NH)(solv) = -1.9 kcal/mol. Analysis is made of the contribution of the solute-solvent hydrogen bonds and it is noted that the SH group in the thiol and NH group in the thione tautomers act exclusively as a hydrogen bond donor in aqueous solution. The proton transfer reaction between the tautomeric forms was also investigated in the gas phase and in aqueous solution. Two distinct mechanisms were considered: a direct intramolecular transfer and a water-assisted mechanism. In the gas phase, the intramolecular transfer leads to a large energy barrier of 34.4 kcal/mol, passing through a three-center transition state. The proton transfer with the assistance of one water molecule decreases the energy barrier to 17.2 kcal/mol. In solution, these calculated activation barriers are, respectively, 32.0 and 14.8 kcal/mol. The solvent effect is found to be sizable but it is considerably more important as a participant in the water-assisted mechanism than the solvent field of the solute-solvent interaction. Finally, the calculated total Gibbs free energy is used to estimate the equilibrium constant.     

Thermochemistry of aqueous hydroxyl radical from advances in photoacoustic calorimetry and ab initio continuum solvation theory

Photoacoustic signals from dilute ( approximately 30 mM) solutions of H2O2 were measured over the temperature range from 10 to 45 degrees C to obtain the reaction enthalpy and volume change for H2O2(aq) --> 2 OH(aq) from which we ultimately determined DeltafG degrees , DeltafH degrees and partial molal volume, v degrees , of OH (aq). We find DeltarH = 46.8 +/- 1.4 kcal/mol, which is 4 kcal/mol smaller than the gas-phase bond energy, and DeltaVr = 6.5 +/- 0.4 mL/mol. The v degrees for OH in water is 14.4 +/- 0.4 mL/ml: smaller than the v degrees of water. Using ab intio continuum theory, the hydration free energy is calculated to be -3.9 +/- 0.3 kcal/mol (for standard states in number density concentration units) by a novel approach devised to capture in the definition of the solute cavity the strength and specific interactions of the solute with a water solvent molecule. The shape of the cavity is defined by "rolling" a three-dimensional electron density isocontour of water on the ab initio water-OH minimum interaction surface. The value of the contour is selected to reproduce the volume of OH in water. We obtain for OH(aq): DeltafH degrees = -0.2 +/- 1.4 and DeltafG degrees = 5.8 +/- 0.4 kcal/mol that are in agreement with literature values. The results provide confidence in the pulsed PAC technique for measuring aqueous thermochemistry of radicals and open the way to obtaining thermochemistry for most radicals that can be formed by reaction of OH with aqueous substrates while advancing the field of continuum solvation theory toward ab initio-defined solute cavities.     

Enthalpy of hydrophobic hydration separately estimated for each -CH2-, -CF2-, and -OCH2CH2- unit component of amphiphiles and of related compounds

The enthalpy of hydrophobic hydration, DeltaH(HH), was separately estimated for each apolar segment of amphiphiles and of related compounds. The specimens analyzed in this study were the homologs and analogs of sodium alkyl sulfate, alkyltrimethylammonium bromide, ammonium perfluoroalkylcarboxylate, and polyethylene glycols. The DeltaH(HH) was calculated from the adiabatic compressibility beta of aqueous solution systematically measured for each homologs at varying concentrations and temperatures. Those data reported in the previous paper were utilized. The DeltaH(HH) estimated for each -CH(2)-, -CF(2)-, and -OCH(2)CH(2)- segments in both hydrophobic and hydrophilic parts of the amphiphiles was found to change depending on its locations in the molecule. The DeltaH(HH) for the first -CH(2)- unit which is adjacent to the head sulfate group of H(CH(2))(n)SO(4)Na was found to be negligibly small; however, it increases gradually as the position of the unit moves apart from the anionic head group and attains as large as -26.5 kcal/mol for the eighth -CH(2)-. The change of DeltaH(HH) of -CF(2)- unit in F(CF(2))(n)COONH(4) shows the similar tendency to H(CH(2))(n)SO(4)Na. The value of DeltaH(HH) is as small as -8 kcal/mol for the first -CF(2)-, however, it increases to -16 kcal/mol for the fifth -CF(2)-, which is 1.5 times larger than that of the fifth -CH(2)- of alkyl sulfate. Contrary to this, all the DeltaH(HH) of the methylenes of cationic surfactant ranging from the second to the eighth -CH(2)- units of H(CH(2))(n)N(CH(3))(3)Br take a similar value around -8 kcal/mol of -CH(2)- irrespective of their positions. The first -CH(2)- unit which is equivalent to that of quaternary ammonium [HCH(2)](4)NBr shows slightly higher value of -9.0 kcal/mol of the first -CH(2)- than those for the second to eighth -CH(2)-. This seems to suggest that the cationic core nitrogen has an enhancing effect on hydrophobic hydration. The hydration enthalpy DeltaH(OEH) of each oxyethylene (OE) unit of polyethylene glycol was also estimated for H(OCH(2)CH(2))(p)OH (p=1-4) homologues to see the roles of ether oxygen and hydroxyl groups in hydrophobic hydration. The value of DeltaH(OEH) showed its maximum -13.1 kcal/mol for the OE unit located at the middle of three. The other two units at both ends were found to take the value of -9.9 kcal/mol, smaller than that at the center. This could suggest that the hydrophobic hydration of the ethylene segment is largely enhanced by the neighboring ether oxygens.     

Role of solvation in the reduction of the uranyl dication by water: a density functional study

We have studied the solvation of uranyl, UO(2)(2+), and the reduced species UO(OH)(2+) and U(OH)(2)(2+) systematically using three levels of approximation: direct application of a continuum model (M1); explicit quantum-chemical treatment of the first hydration sphere (M2); a combined quantum-chemical/continuum model approach (M3). We have optimized complexes with varying numbers of aquo ligands (n = 4-6) and compared their free energies of solvation. Models M1 and M2 have been found to recover the solvation energy only partially, underestimating it by approximately 100 kcal/mol or more. With our best model M3, the calculated hydration free energy Delta(h)G degrees of UO(2)(2+) is about -420 kcal/mol, which shifts to about -370 kcal/mol when corrected for the expected error of the model. This value agrees well with the experimentally determined interval, -437 kcal/mol < Delta(h)G degrees < -318 kcal/mol. Complexes with 5 and 6 aquo ligands have been found to be about equally favored with models M2 and M3. The same solvation models have been applied to a two-step reduction of UO(2)(2+) by water, previously theoretically studied in the gas phase. Our results show that the solvation contribution to the reaction free energy, about 60 kcal/mol, dominates the endoergicity of the reduction.     

Calculation of sequence‐dependent free energies of hydration of dipeptides formed by alanine and glycine

The relative free energies of hydration of the dipeptides glycylalanine and alanyl‐glycine in their naturally occurring form have been calculated both for the zwitterionic and protonated species. Emphasis was laid on comparisons between the conventional cutoff method and the Particle Mesh Ewald method to account for possible differences in electrostatic contributions to the free energy. Furthermore, the convergence behavior of the total free energy and its individual contributions were examined. The results, obtained by means of the thermodynamic integration technique as implemented in the free energy module of the AMBER program suite, suggest that in aqueous solution glycylalanine is more stable than alanylglycine by 2.7 kcal/mol in the zwitterionic form and by 3.5 kcal/mol in the protonated form. © 2001 John Wiley & Sons, Inc. J Comput Chem 22: 846–860, 2001     

Improved interaction potentials for charged residues in proteins

Electrostatic interactions dominate the structure and free energy of biomolecules. To obtain accurate free energies involving charged groups from molecular simulations, OPLS-AA parameters have been reoptimized using Monte Carlo free energy perturbation. New parameters fit a self-consistent, experimental set of hydration free energies for acetate (Asp), propionate (Glu), 4-methylimidazolium (Hip), n-butylammonium (Lys), and n-propylguanidinium (Arg), all resembling charged residue side chains, including beta-carbons. It is shown that OPLS-AA free energies depend critically on the type of water model, TIP4P or TIP3P; i.e., each water model requires specific water-charged molecule interaction potentials. New models (models 1 and 3) are thus described for both water models. Uncertainties in relative free energies of charged residues are approximately 2 kcal/mol with the new parameters, due to variations in system setup (MAEs of ca. 1 kcal/mol) and noise from simulations (ca. 1 kcal/mol). The latter error of approximately 1 kcal/mol contrasts MAEs from standard OPLS-AA of up to 13 kcal/mol for the entire series of charged residues or up to 5 kcal/mol for the cationic series Lys, Arg, and Hip. The new parameters can be used directly in molecular simulations with no modification of neutral residues needed and are envisioned to be particular important in simulations where charged residues change environment.     

QM/MM metadynamics study of the direct decarboxylation mechanism for orotidine-5'-monophosphate decarboxylase using two different QM regions: acceleration too small to explain rate of enzyme catalysis

Despite decades of study, the mechanism by which orotidine-5'-monophosphate decarboxylase (ODCase) catalyzes the decarboxylation of orotidine monophosphate remains unresolved. A computational investigation of the direct decarboxylation mechanism has been performed using mixed quantum mechanical/molecular mechanical (QM/MM) dynamics simulations. The study was performed with the program CP2K that integrates classical dynamics and ab initio dynamics based on the Born-Oppenheimer approach. Two different QM regions were explored. The free energy barriers for direct decarboxylation of orotidine-5'-monophosphate (OMP) in solution and in the enzyme (using the larger QM region) were determined with the metadynamics method to be 40 and 33 kcal/mol, respectively. The calculated change in activation free energy (DeltaDeltaG++) on going from solution to the enzyme is therefore -7 kcal/mol, far less than the experimental change of -23 kcal/ mol (for k(cat.)/k(uncat.): Radzicka, A.; Wolfenden, R., Science 1995, 267, 90-92). These results do not support the direct decarboxylation mechanism that has been proposed for the enzyme. However, in the context of QM/MM calculations, it was found that the size of the QM region has a dramatic effect on the calculated reaction barrier.     

Predicting hydration free energies with a hybrid QM/MM approach: an evaluation of implicit and explicit solvation models in SAMPL4

The correct representation of solute-water interactions is essential for the accurate simulation of most biological phenomena. Several highly accurate quantum methods are available to deal with solvation by using both implicit and explicit solvents. So far, however, most evaluations of those methods were based on a single conformation, which neglects solute entropy. Here, we present the first test of a novel approach to determine hydration free energies that uses molecular mechanics (MM) to sample phase space and quantum mechanics (QM) to evaluate the potential energies. Free energies are determined by using re-weighting with the Non-Boltzmann Bennett (NBB) method. In this context, the method is referred to as QM-NBB. Based on snapshots from MM sampling and accounting for their correct Boltzmann weight, it is possible to obtain hydration free energies that incorporate the effect of solute entropy. We evaluate the performance of several QM implicit solvent models, as well as explicit solvent QM/MM for the blind subset of the SAMPL4 hydration free energy challenge. While classical free energy simulations with molecular dynamics give root mean square deviations (RMSD) of 2.8 and 2.3 kcal/mol, the hybrid approach yields an improved RMSD of 1.6 kcal/mol. By selecting an appropriate functional and basis set, the RMSD can be reduced to 1 kcal/mol for calculations based on a single conformation. Results for a selected set of challenging molecules imply that this RMSD can be further reduced by using NBB to reweight MM trajectories with the SMD implicit solvent model.     

Prediction of hydration free energies for aliphatic and aromatic chloro derivatives using molecular dynamics simulations with the OPLS-AA force field

All-atom molecular dynamics computer simulations were used to blindly predict the hydration free energies of a range of chloro-organic compounds as part of the SAMPL3 challenge. All compounds were parameterized within the framework of the OPLS-AA force field, using an established protocol to compute the absolute hydration free energy via a windowed free energy perturbation approach and thermodynamic integration. Three different approaches to deriving partial charge parameters were pursued: (1) using existing OPLS-AA atom types and charges with minor adjustments of partial charges on equivalent connecting atoms; (2) calculation of quantum mechanical charges via geometry optimization, followed by electrostatic potential (ESP) fitting, using Jaguar at the LMP2/cc-pVTZ(-F) level; and (3) via geometry optimization and CHelpG charges (Gaussian03 at the HF/6-31G* level), followed by two-stage RESP fitting. Protocol 3 generated the most accurate predictions with a root mean square (RMS) error of 1.2 kcal mol(-1) for the entire data set. It was found that the deficiency of the standard OPLS-AA parameters, protocol 1 (RMS error 2.4 kcal mol(-1) overall), was mostly due to compounds with more than three chlorine substituents on an aromatic ring. For this latter subset, the RMS errors were 1.4 kcal mol(-1) (protocol 3) and 4.3 kcal mol(-1) (protocol 1), respectively. We propose new OPLS-AA atom types for aromatic carbon and chlorine atoms in rings with ≥4 Cl-substituents that perform better than the best QM-based approach, resulting in an RMS error of 1.2 kcal mol(-1) for these difficult compounds.