Bispecific Estrogen Receptor α Degraders Incorporating Novel Binders Identified Using DNA-Encoded Chemical Library Screening.

Bispecific degraders (PROTACs) of ERα are expected to be advantageous over current inhibitors of ERα signaling (aromatase inhibitors/SERMs/SERDs) used to treat ER+ breast cancer. Information from DNA-encoded chemical library (DECL) screening provides a method to identify novel PROTAC binding features as the linker positioning, and binding elements are determined directly from the screen. After screening ∼120 billion DNA-encoded molecules with ERα WT and 3 gain-of-function (GOF) mutants, with and without estradiol to identify features that enrich ERα competitively, the off-DNA synthesized small molecule exemplar 7 exhibited nanomolar ERα binding, antagonism, and degradation. Click chemistry synthesis on an alkyne E3 ligase engagers panel and an azide variant of 7 rapidly generated bispecific nanomolar degraders of ERα, with PROTACs 18 and 21 inhibiting ER+ MCF7 tumor growth in a mouse xenograft model of breast cancer. This study validates this approach toward identifying novel bispecific degrader leads from DECL screening with minimal optimization.

Protein-RNA Docking Using ICM.

Protein-RNA interactions play an important role in many biological processes. Computational methods such as docking have been developed to complement existing biophysical and structural biology techniques. Computational prediction of protein-RNA complex structures includes two steps: generating candidate structures from the individual protein and RNA parts and scoring the generated poses to pick out the correct one. In this work, we considered three recently developed data sets of protein-RNA complexes to evaluate and improve the performance of the FFT-based rigid-body docking algorithm implemented in the ICM package. An electrostatic term describing interactions between negatively charged phosphate groups and positively charged protein residues was added to the energy function used during the docking step to take into account the greater role that electrostatic interactions play in protein-RNA complexes. Next, the docking results were used to optimize a scoring function including van der Waals, electrostatic, and solvation terms. This optimization yielded a much smaller weight for the solvation term indicating that solvation energy may be less important for the scoring of protein-RNA structures. Rescoring of the generated poses with the new scoring function led to much higher success rates, while pose clustering by contact fingerprints produced further improvements, achieving a success rate of 0.66 for the top 100 structures.

All-Atom Internal Coordinate Mechanics (ICM) Force Field for Hexopyranoses and Glycoproteins.

We present an extension of the all-atom internal-coordinate force field, ICMFF, that allows for simulation of heterogeneous systems including hexopyranose saccharides and glycan chains in addition to proteins. A library of standard glycan geometries containing α- and ß-anomers of the most common hexapyranoses, i.e., d-galactose, d-glucose, d-mannose, d-xylose, l-fucose, N-acetylglucosamine, N-acetylgalactosamine, sialic, and glucuronic acids, is created based on the analysis of the saccharide structures reported in the Cambridge Structural Database. The new force field parameters include molecular electrostatic potential-derived partial atomic charges and the torsional parameters derived from quantum mechanical data for a collection of minimal molecular fragments and related molecules. The ϕ/ψ torsional parameters for different types of glycosidic linkages are developed using model compounds containing the key atoms in the full carbohydrates, i.e., glycosidic-linked tetrahydropyran-cyclohexane dimers. Target data for parameter optimization include two-dimensional energy surfaces corresponding to the ϕ/ψ glycosidic dihedral angles in the disaccharide analogues, as determined by quantum mechanical MP2/6-31G** single-point energies on HF/6-31G** optimized structures. To achieve better agreement with the observed geometries of glycosidic linkages, the bond angles at the O-linkage atoms are added to the internal variable set and the corresponding bond bending energy term is parametrized using quantum mechanical data. The resulting force field is validated on glycan chains of 1-12 residues from a set of high-resolution X-ray glycoprotein structures based on heavy atom root-mean-square deviations of the lowest-energy glycan conformations generated by the biased probability Monte Carlo (BPMC) molecular mechanics simulations from the native structures. The appropriate BPMC distributions for monosaccharide-monosaccharide and protein-glycan linkages are derived from the extensive analysis of conformational properties of glycoprotein structures reported in the Protein Data Bank. Use of the BPMC search leads to significant improvements in sampling efficiency for glycan simulations. Moreover, good agreement with the X-ray glycoprotein structures is achieved for all glycan chain lengths. Thus, average/median RMSDs are 0.81/0.68 Å for one-residue glycans and 1.32/1.47 Å for three-residue glycans. RMSD from the native structure for the lowest-energy conformation of the 12-residue glycan chain (PDB ID 3og2) is 1.53 Å. Additionally, results obtained for free short oligosaccharides using the new force field are in line with the available experimental data, i.e., the most populated conformations in solution are predicted to be the lowest energy ones. The newly developed parameters allow for the accurate modeling of linear and branched hexopyranose glycosides in heterogeneous systems.

Are accurate computations of the 13C' shielding feasible at the DFT level of theory?

The goal of this study is twofold. First, to investigate the relative influence of the main structural factors affecting the computation of the (13)C' shielding, namely, the conformation of the residue itself and the next nearest-neighbor effects. Second, to determine whether calculation of the (13)C' shielding at the density functional level of theory (DFT), with an accuracy similar to that of the (13)C(α) shielding, is feasible with the existing computational resources. The DFT calculations, carried out for a large number of possible conformations of the tripeptide Ac-GXY-NMe, with different combinations of X and Y residues, enable us to conclude that the accurate computation of the (13)C' shielding for a given residue X depends on the: (i) (ϕ,ψ) backbone torsional angles of X; (ii) side-chain conformation of X; (iii) (ϕ,ψ) torsional angles of Y; and (iv) identity of residue Y. Consequently, DFT-based quantum mechanical calculations of the (13)C' shielding, with all these factors taken into account, are two orders of magnitude more CPU demanding than the computation, with similar accuracy, of the (13)C(α) shielding. Despite not considering the effect of the possible hydrogen bond interaction of the carbonyl oxygen, this work contributes to our general understanding of the main structural factors affecting the accurate computation of the (13)C' shielding in proteins and may spur significant progress in effort to develop new validation methods for protein structures.

Physics-based method to validate and repair flaws in protein structures.

A method that makes use of information provided by the combination of (13)C(α) and (13)C(ß) chemical shifts, computed at the density functional level of theory, enables one to (i) validate, at the residue level, conformations of proteins and detect backbone or side-chain flaws by taking into account an ensemble average of chemical shifts over all of the conformations used to represent a protein, with a sensitivity of ∼90%; and (ii) provide a set of (χ1/χ2) torsional angles that leads to optimal agreement between the observed and computed (13)C(α) and (13)C(ß) chemical shifts. The method has been incorporated into the CheShift-2 protein validation Web server. To test the reliability of the provided set of (χ1/χ2) torsional angles, the side chains of all reported conformations of five NMR-determined protein models were refined by a simple routine, without using NOE-based distance restraints. The refinement of each of these five proteins leads to optimal agreement between the observed and computed (13)C(α) and (13)C(ß) chemical shifts for ∼94% of the flaws, on average, without introducing a significantly large number of violations of the NOE-based distance restraints for a distance range ≤ 0.5 , in which the largest number of distance violations occurs. The results of this work suggest that use of the provided set of (χ1/χ2) torsional angles together with other observables, such as NOEs, should lead to a fast and accurate refinement of the side-chain conformations of protein models.

Towards crystal structure prediction of complex organic compounds--a report on the fifth blind test.

Following on from the success of the previous crystal structure prediction blind tests (CSP1999, CSP2001, CSP2004 and CSP2007), a fifth such collaborative project (CSP2010) was organized at the Cambridge Crystallographic Data Centre. A range of methodologies was used by the participating groups in order to evaluate the ability of the current computational methods to predict the crystal structures of the six organic molecules chosen as targets for this blind test. The first four targets, two rigid molecules, one semi-flexible molecule and a 1:1 salt, matched the criteria for the targets from CSP2007, while the last two targets belonged to two new challenging categories - a larger, much more flexible molecule and a hydrate with more than one polymorph. Each group submitted three predictions for each target it attempted. There was at least one successful prediction for each target, and two groups were able to successfully predict the structure of the large flexible molecule as their first place submission. The results show that while not as many groups successfully predicted the structures of the three smallest molecules as in CSP2007, there is now evidence that methodologies such as dispersion-corrected density functional theory (DFT-D) are able to reliably do so. The results also highlight the many challenges posed by more complex systems and show that there are still issues to be overcome.

Assessing the fractions of tautomeric forms of the imidazole ring of histidine in proteins as a function of pH.

A method is proposed to determine the fraction of the tautomeric forms of the imidazole ring of histidine in proteins as a function of pH, provided that the observed and chemical shifts and the protein structure, or the fraction of H(+) form, are known. This method is based on the use of quantum chemical methods to compute the (13)C NMR shieldings of all the imidazole ring carbons ((13)C(γ), , and ) for each of the two tautomers, N(δ1)-H and N(ε2)-H, and the protonated form, H(+), of histidine. This methodology enabled us (i) to determine the fraction of all the tautomeric forms of histidine for eight proteins for which the and chemical shifts had been determined in solution in the pH range of 3.2 to 7.5 and (ii) to estimate the fraction of tautomeric forms of eight histidine-containing dipeptide crystals for which the chemical shifts had been determined by solid-state (13)C NMR. Our results for proteins indicate that the protonated form is the most populated one, whereas the distribution of the tautomeric forms for the imidazole ring varies significantly among different histidines in the same protein, reflecting the importance of the environment of the histidines in determining the tautomeric forms. In addition, for ∼70% of the neutral histidine-containing dipeptides, the method leads to fairly good agreement between the calculated and the experimental tautomeric form. Coexistence of different tautomeric forms in the same crystal structure may explain the remaining 30% of disagreement.

Development of a new physics-based internal coordinate mechanics force field and its application to protein loop modeling.

We report the development of internal coordinate mechanics force field (ICMFF), new force field parameterized using a combination of experimental data for crystals of small molecules and quantum mechanics calculations. The main features of ICMFF include: (a) parameterization for the dielectric constant relevant to the condensed state (ε = 2) instead of vacuum, (b) an improved description of hydrogen-bond interactions using duplicate sets of van der Waals parameters for heavy atom-hydrogen interactions, and (c) improved backbone covalent geometry and energetics achieved using novel backbone torsional potentials and inclusion of the bond angles at the C(α) atoms into the internal variable set. The performance of ICMFF was evaluated through loop modeling simulations for 4-13 residue loops. ICMFF was combined with a solvent-accessible surface area solvation model optimized using a large set of loop decoys. Conformational sampling was carried out using the biased probability Monte Carlo method. Average/median backbone root-mean-square deviations of the lowest energy conformations from the native structures were 0.25/0.21 Å for four residues loops, 0.84/0.46 Å for eight residue loops, and 1.16/0.73 Å for 12 residue loops. To our knowledge, these results are significantly better than or comparable with those reported to date for any loop modeling method that does not take crystal packing into account. Moreover, the accuracy of our method is on par with the best previously reported results obtained considering the crystal environment. We attribute this success to the high accuracy of the new ICM force field achieved by meticulous parameterization, to the optimized solvent model, and the efficiency of the search method.

Quantum-mechanics-derived 13Calpha chemical shift server (CheShift) for protein structure validation.

A server (CheShift) has been developed to predict (13)C(alpha) chemical shifts of protein structures. It is based on the generation of 696,916 conformations as a function of the phi, psi, omega, chi1 and chi2 torsional angles for all 20 naturally occurring amino acids. Their (13)C(alpha) chemical shifts were computed at the DFT level of theory with a small basis set and extrapolated, with an empirically-determined linear regression formula, to reproduce the values obtained with a larger basis set. Analysis of the accuracy and sensitivity of the CheShift predictions, in terms of both the correlation coefficient R and the conformational-averaged rmsd between the observed and predicted (13)C(alpha) chemical shifts, was carried out for 3 sets of conformations: (i) 36 x-ray-derived protein structures solved at 2.3 A or better resolution, for which sets of (13)C(alpha) chemical shifts were available; (ii) 15 pairs of x-ray and NMR-derived sets of protein conformations; and (iii) a set of decoys for 3 proteins showing an rmsd with respect to the x-ray structure from which they were derived of up to 3 A. Comparative analysis carried out with 4 popular servers, namely SHIFTS, SHIFTX, SPARTA, and PROSHIFT, for these 3 sets of conformations demonstrated that CheShift is the most sensitive server with which to detect subtle differences between protein models and, hence, to validate protein structures determined by either x-ray or NMR methods, if the observed (13)C(alpha) chemical shifts are available. CheShift is available as a web server.

What can we learn by computing 13Calpha chemical shifts for X-ray protein models?

The room-temperature X-ray structures of ubiquitin (PDB code 1ubq) and of the RNA-binding domain of nonstructural protein 1 of influenza A virus (PDB code 1ail) solved at 1.8 and 1.9 A resolution, respectively, were used to investigate whether a set of conformations rather than a single X-ray structure provides better agreement with both the X-ray data and the observed 13Calpha chemical shifts in solution. For this purpose, a set of new conformations for each of these proteins was generated by fitting them to the experimental X-ray data deposited in the PDB. For each of the generated structures, which show R and Rfree factors similar to those of the deposited X-ray structure, the 13Calpha chemical shifts of all residues in the sequence were computed at the DFT level of theory. The sets of conformations were then evaluated by their ability to reproduce the observed 13Calpha chemical shifts by using the conformational average root-mean-square-deviation (ca-r.m.s.d.). For ubiquitin, the computed set of conformations is a better representation of the observed 13Calpha chemical shifts in terms of the ca-r.m.s.d. than a single X-ray-derived structure. However, for the RNA-binding domain of nonstructural protein 1 of influenza A virus, consideration of an ensemble of conformations does not improve the agreement with the observed 13Calpha chemical shifts. Whether an ensemble of conformations rather than any single structure is a more accurate representation of a protein structure in the crystal as well as of the observed 13Calpha chemical shifts is determined by the dispersion of coordinates, in terms of the all-atom r.m.s.d. among the generated models; these generated models satisfy the experimental X-ray data with accuracy as good as the PDB structure. Therefore, generation of an ensemble is a necessary step to determine whether or not a single structure is sufficient for an accurate representation of both experimental X-ray data and observed 13Calpha chemical shifts in solution.

Identifying native-like protein structures with scoring functions based on all-atom ECEPP force fields, implicit solvent models and structure relaxation.

Availability of energy functions which can discriminate native-like from non-native protein conformations is crucial for theoretical protein structure prediction and refinement of low-resolution protein models. This article reports the results of benchmark tests for scoring functions based on two all-atom ECEPP force fields, that is, ECEPP/3 and ECEPP05, and two implicit solvent models for a large set of protein decoys. The following three scoring functions are considered: (i) ECEPP05 plus a solvent-accessible surface area model with the parameters optimized with a set of protein decoys (ECEPP05/SA); (ii) ECEPP/3 plus the solvent-accessible surface area model of Ooi et al. (Proc Natl Acad Sci USA 1987;84:3086-3090) (ECEPP3/OONS); and (iii) ECEPP05 plus an implicit solvent model based on a solution of the Poisson equation with an optimized Fast Adaptive Multigrid Boundary Element (FAMBEpH) method (ECEPP05/FAMBEpH). Short Monte Carlo-with-Minimization (MCM) simulations, following local energy minimization, are used as a scoring method with ECEPP05/SA and ECEPP3/OONS potentials, whereas energy calculation is used with ECEPP05/FAMBEpH. The performance of each scoring function is evaluated by examining its ability to distinguish between native-like and non-native protein structures. The results of the tests show that the new ECEPP05/SA scoring function represents a significant improvement over the earlier ECEPP3/OONS version of the force field. Thus, it is able to rank native-like structures with C(alpha) root-mean-square-deviations below 3.5 A as lowest-energy conformations for 76% and within the top 10 for 87% of the proteins tested, compared with 69 and 80%, respectively, for ECEPP3/OONS. The use of the FAMBEpH solvation model, which provides a more accurate description of the protein-solvent interactions, improves the discriminative ability of the scoring function to 89%. All failed tests in which the native-like structures cannot be discriminated as those with low energy, are due to omission of protein-protein interactions. The results of this study represent a benchmark in force-field development, and may be useful for evaluation of the performance of different force fields.

Significant progress in predicting the crystal structures of small organic molecules--a report on the fourth blind test.

We report on the organization and outcome of the fourth blind test of crystal structure prediction, an international collaborative project organized to evaluate the present state in computational methods of predicting the crystal structures of small organic molecules. There were 14 research groups which took part, using a variety of methods to generate and rank the most likely crystal structures for four target systems: three single-component crystal structures and a 1:1 cocrystal. Participants were challenged to predict the crystal structures of the four systems, given only their molecular diagrams, while the recently determined but as-yet unpublished crystal structures were withheld by an independent referee. Three predictions were allowed for each system. The results demonstrate a dramatic improvement in rates of success over previous blind tests; in total, there were 13 successful predictions and, for each of the four targets, at least two groups correctly predicted the observed crystal structure. The successes include one participating group who correctly predicted all four crystal structures as their first ranked choice, albeit at a considerable computational expense. The results reflect important improvements in modelling methods and suggest that, at least for the small and fairly rigid types of molecules included in this blind test, such calculations can be constructively applied to help understand crystallization and polymorphism of organic molecules.

Use of decoys to optimize an all-atom force field including hydration.

A novel method of parameter optimization is proposed. It makes use of large sets of decoys generated for six nonhomologous proteins with different architecture. Parameter optimization is achieved by creating a free energy gap between sets of nativelike and nonnative conformations. The method is applied to optimize the parameters of a physics-based scoring function consisting of the all-atom ECEPP05 force field coupled with an implicit solvent model (a solvent-accessible surface area model). The optimized force field is able to discriminate near-native from nonnative conformations of the six training proteins when used either for local energy minimization or for short Monte Carlo simulated annealing runs after local energy minimization. The resulting force field is validated with an independent set of six nonhomologous proteins, and appears to be transferable to proteins not included in the optimization; i.e., for five out of the six test proteins, decoys with 1.7- to 4.0-A all-heavy-atom root mean-square deviations emerge as those with the lowest energy. In addition, we examined the set of misfolded structures created by Park and Levitt using a four-state reduced model. The results from these additional calculations confirm the good discriminative ability of the optimized force field obtained with our decoy sets.

Use of 13C(alpha) chemical shifts for accurate determination of beta-sheet structures in solution.

A physics-based method, aimed at determining protein structures by using NOE-derived distance constraints together with observed and computed 13C(alpha) chemical shifts, is applied to determine the structure of a 20-residue all-beta peptide (BS2). The approach makes use of 13C(alpha) chemical shifts, computed at the density functional level of theory, to derive backbone and side-chain torsional constraints for all of the amino acid residues, without making use of information about residue occupancy in any region of the Ramachandran map. In addition, the torsional constraints are derived dynamically--i.e., they are redefined at each step of the algorithm. It is shown that, starting from randomly generated conformations, the final protein models are more accurate than existing NMR-derived models of the peptide, in terms of the agreement between predicted and observed 13C(beta) chemical shifts, and some stereochemical quality indicators. The accumulated evidence indicates that, for a highly flexible BS2 peptide in solution, it may not be possible to determine a single structure (or a small set of structures) that would satisfy all of the constraints exactly and simultaneously because the observed NOEs and 13C(alpha) chemical shifts correspond to a dynamic ensemble of conformations. Analysis of the structural flexibility, carried out by molecular dynamics simulations in explicit water, revealed that the whole peptide can be characterized as having liquid-like behavior, according to the Lindemann criterion. In summary, a beta-sheet structure of a highly flexible peptide in solution can be determined by a quantum-chemical-based procedure.

A new force field (ECEPP-05) for peptides, proteins, and organic molecules.

Parametrization and testing of a new all-atom force field for organic molecules and peptides with fixed bond lengths and bond angles are described. The van der Waals parameters for both the organic molecules and the peptides were taken from J. Phys. Chem. B 2003, 107, 7143 and J. Phys. Chem. B 2004, 108, 12181. First, the values of the 1-4 nonbonded and electrostatic scale factors appropriate to the new force field were determined by computing the conformational energies of six model molecules, namely, ethanol, ethylamine, propanol, propylamine, 1,2-ethanediol, and 1,3-propanediol with different values of these factors. The partial atomic charges of these molecules were obtained by fitting to the electrostatic potentials calculated with the HF/6-31G quantum-mechanical method. Two different charge models (single- and multiple-conformation-derived) were also considered. We demonstrated that the charge model has a stronger effect on the conformational energies than the 1-4 scaling. The choice of a charge model affected the conformational energies of even the smallest molecules considered, whereas the effect of the 1-4 electrostatic or nonbonded scaling was apparent only for 1,3-propanediol. The best agreement with high-level ab initio data was obtained with the multiple-conformation-derived charges and with no scaling of the 1-4 nonbonded or electrostatic interactions (scale factors of 1.0). Next, the torsional parameters of a large number of neutral and charged organic molecules, assumed to be models of the side chains of the 20 naturally occurring amino acids, were computed by fitting to rotational energy profiles obtained from ab initio MP2/6-31G calculations. The quality of the fits was high with average errors for torsional profiles of less than 0.2 kcal/mol. To derive the torsional parameters for the peptide backbone, the partial atomic charges of the 20 neutral and charged amino acids were obtained by fitting to the electrostatic potentials of terminally blocked amino acids using the HF/6-31G quantum-mechanical method. Then, the phi-psi energy maps of Ac-Ala-NMe and Ac-Gly-NMe were computed using MP2/6-31G//HF/6-31G quantum-mechanical methods. The phi-psi energy map of Ac-Ala-NMe was used for refinement of the nonbonded parameters for the backbone nitrogen and hydrogen bonded to it. Subsequently, the main-chain torsional parameters were obtained by fitting the molecular mechanics energies to the phi-psi energy maps of Ac-Ala-NMe and Ac-Gly-NMe. The transferability of the entire force field was demonstrated by reproducing the main energy minima of terminally blocked Ala3 from the literature. The performance of the force field was also evaluated by simulating crystal structures of small peptides. By comparison of simulated and experimental data, examination of the torsional-angle and atom-positional root-mean-square deviations of the energy-minimized crystal structures from the corresponding X-ray model structures demonstrated high accuracy of the force field.

Coupling between conformation and proton binding in proteins.

Interest centers here on whether the use of a fixed charge distribution of a protein solute, or a treatment that considers proton-binding equilibria by solving the Poisson equation, is a better approach to discriminate native from non-native conformations of proteins. In this analysis of the charge distribution of 7 proteins, we estimate the solvation free energy contribution to the total free energy by exploring the 2(zeta) possible ionization states of the whole molecule, with zeta being the number of ionizable groups in the amino acid sequence, for every conformation in the ensembles of 7 proteins. As an additional consideration of the role of electrostatic interactions in determining the charge distribution of native folds, we carried out a comparison of alternative charge assignment models for the ionizable residues in a set of 21 native-like proteins. The results of this work indicate that (1) for 6 out of 7 proteins, estimation of solvent polarization based on the Generalized Born model with a fixed charge distribution provides the optimal trade-off between accuracy, with respect to the Poisson equation, and speed when compared to the accessible surface area model; for the seventh protein, consideration of all possible ionization states of the whole molecule appears to be crucial to discriminate the native from non-native conformations; (2) significant differences in the degree of ionization and hence the charge distribution for native folds are found between the different charge models examined; (3) the stability of the native state is determined by a delicate balance of all the energy components, and (4) conformational entropy, and hence the dynamics of folding, may play a crucial role for a successful ab initio protein folding prediction.

Evolution of physics-based methodology for exploring the conformational energy landscape of proteins.

The evolution of our physics-based computational methods for determining protein conformation without the introduction of secondary-structure predictions, homology modeling, threading, or fragment coupling is described. Initial use of a hard-sphere potential captured much of the structural properties of polypeptide chains, and subsequent more refined force fields, together with efficient methods of global optimization provide indications that progress is being made toward an understanding of the interresidue interactions that underlie protein folding.