Small-molecule Crystal Structures Research Articles

Broadening the scope ofActa Crystallographica Section C

Acta Crystallographica Section C: Crystal Structure Communications celebrates its 30th anniversary in 2013. In 1983, Section C was created out of a merger of the Short Structural Papers, which had been included in Section B since 1972, with the successful journal Crystal Structure Communications, which had been published by the University of Parma, Italy, in the period 1972–1982. The new Section C adopted Crystal Structure Communications as its subtitle and published papers focusing on crystal structures of small molecules and continuous solids. The recognition of a continued demand to be able to publish very short reports of crystal structure determinations resulted in Section C including a Short Format Papers subsection in the period 1985–1993, CIF-access papers in the period 1997–1999 and Electronic Papers in 2000, and then the ultimate creation of Acta Crystallographica Section E in 2001. Since then, Acta Crystallographica Section C has specialized in the rapid publication of articles that emphasize interesting scientific results and analyses based on the determination of high-quality or challenging crystal and molecular structures of interest in the fields of chemistry, biochemistry, mineralogy, pharmacology, physics and materials science. The journal has a reputation for publishing high-quality crystal structures, as well as articles describing difficult or challenging structures and the strategies used to address the difficulties. The depth of discussion has been an important aspect of articles presented in Section C.

Energetically unfavorable amide conformations for N6‐acetyllysine side chains in refined protein structures

The reversible acetylation of lysine to form N6-acetyllysine in the regulation of protein function is a hallmark of epigenetics. Acetylation of the positively charged amino group of the lysine side chain generates a neutral N-alkylacetamide moiety that serves as a molecular "switch" for the modulation of protein function and protein-protein interactions. We now report the analysis of 381 N6-acetyllysine side chain amide conformations as found in 79 protein crystal structures and 11 protein NMR structures deposited in the Protein Data Bank (PDB) of the Research Collaboratory for Structural Bioinformatics. We find that only 74.3% of N6-acetyllysine residues in protein crystal structures and 46.5% in protein NMR structures contain amide groups with energetically preferred trans or generously trans conformations. Surprisingly, 17.6% of N6-acetyllysine residues in protein crystal structures and 5.3% in protein NMR structures contain amide groups with energetically unfavorable cis or generously cis conformations. Even more surprisingly, 8.1% of N6-acetyllysine residues in protein crystal structures and 48.2% in NMR structures contain amide groups with energetically prohibitive twisted conformations that approach the transition state structure for cis-trans isomerization. In contrast, 109 unique N-alkylacetamide groups contained in 84 highly accurate small molecule crystal structures retrieved from the Cambridge Structural Database exclusively adopt energetically preferred trans conformations. Therefore, we conclude that cis and twisted N6-acetyllysine amides in protein structures deposited in the PDB are erroneously modeled due to their energetically unfavorable or prohibitive conformations.

Conformational analysis of retinoic acids: Effects of steric interactions on nonplanar conjugated polyenes

Retinoic acids and other vitamin A analogs contain a trimethylcyclohexenyl ring in conjugation with a polyene chain joined at carbon-6 (C6) and carbon-7 (C7). A MP2-SCS/cc-pVDZ//B3LYP/6-31G(d) 2-D potential energy surface was computed for all-trans retinoic acid, which had 6 minima (3 enantiomeric pairs). The global minima were distorted s-gauche enantiomers (τ6–7=±53°) with half-chair conformations of the ring. Distorted s-gauche enantiomers (τ6–7=±55°) with inverted half-chair ring conformations were 1.7kJ/mol above the global minima. The s-trans enantiomers (τ6–7=±164°) were 11.3kJ/mol above the global minima. Steric energies were computed by the method of Guo and Karplus to identify key structural elements in retinoic acids which determines their conformation. Small molecule crystal structures in the CCDC database with trimethylcyclohexenyl ring and exocyclic double bonds have ring-chain geometries near to one of the six energy minima of retinoic acids, except for retinaldehyde iminium cations.

Low-barrier hydrogen bonds in proteins

Hydrogen bonding interactions are one of the most important chemical interactions among materials, especially biological materials, which help confer specificity, which is crucial for their efficient functioning. Recently, low-barrier hydrogen bonds (LBHBs) have been proposed to play a critical role in enzyme catalysis. In this review, tools to identify LBHBs are described, along with analyses of neutron crystal structures of small molecules to identify geometric parameters characteristic of LBHBs, which are assumed to be characterized by dynamic disorder along the hydrogen bond (H-bond) of the bonding hydrogen atom. The analysis of protein structures determined by neutron diffraction indicates that LBHBs are found to occur in both active site and non-active site regions of a protein. Moreover, very short H-bonds are observed in the vicinity of folding cores identified through nuclear magnetic resonance studies on two proteins, SUMO-1 and HIV-1 protease. This observation suggests that LBHBs may also be important in the context of folding of proteins.

Reading PDB: Perception of Molecules from 3D Atomic Coordinates

The analysis of small molecule crystal structures is a common way to gather valuable information for drug development. The necessary structural data is usually provided in specific file formats containing only element identities and three-dimensional atomic coordinates as reliable chemical information. Consequently, the automated perception of molecular structures from atomic coordinates has become a standard task in cheminformatics. The molecules generated by such methods must be both chemically valid and reasonable to provide a reliable basis for subsequent calculations. This can be a difficult task since the provided coordinates may deviate from ideal molecular geometries due to experimental uncertainties or low resolution. Additionally, the quality of the input data often differs significantly thus making it difficult to distinguish between actual structural features and mere geometric distortions. We present a method for the generation of molecular structures from atomic coordinates based on the recently published NAOMI model. By making use of this consistent chemical description, our method is able to generate reliable results even with input data of low quality. Molecules from 363 Protein Data Bank (PDB) entries could be perceived with a success rate of 98%, a result which could not be achieved with previously described methods. The robustness of our approach has been assessed by processing all small molecules from the PDB and comparing them to reference structures. The complete data set can be processed in less than 3 min, thus showing that our approach is suitable for large scale applications.

Conformational bias in small-molecule crystal structures is rare (and explicable)

Conformational bias in small-molecule crystal structures is rare (and explicable)

Use of Small-Molecule Crystal Structures To Address Solubility in a Novel Series of G Protein Coupled Receptor 119 Agonists: Optimization of a Lead and in Vivo Evaluation

G protein coupled receptor 119 (GPR119) is viewed as an attractive target for the treatment of type 2 diabetes and other elements of the metabolic syndrome. During a program toward discovering agonists of GPR119, we herein describe optimization of an initial lead compound, 2, into a development candidate, 42. A key challenge in this program of work was the insolubility of the lead compound. Small-molecule crystallography was utilized to understand the intermolecular interactions in the solid state and resulted in a switch from an aryl sulphone to a 3-cyanopyridyl motif. The compound was shown to be effective in wild-type but not knockout animals, confirming that the biological effects were due to GPR119 agonism.

The good, the bad and the twisted: a survey of ligand geometry in protein crystal structures.

The protein databank now contains the structures of over 11,000 ligands bound to proteins. These structures are invaluable in applied areas such as structure-based drug design, but are also the substrate for understanding the energetics of intermolecular interactions with proteins. Despite their obvious importance, the careful analysis of ligands bound to protein structures lags behind the analysis of the protein structures themselves. We present an analysis of the geometry of ligands bound to proteins and highlight the role of small molecule crystal structures in enabling molecular modellers to critically evaluate a ligand model’s quality and investigate protein-induced strain.Electronic supplementary materialThe online version of this article (doi:10.1007/s10822-011-9538-6) contains supplementary material, which is available to authorized users.

Systematic conformational bias in small-molecule crystal structures is rare and explicable

Analysis of the Cambridge Structural Database, together with DFT and crystal structure prediction calculations, show that the observation of higher-energy planar conformers of biphenyl (BP) and cyclobutane (CB) is possible because of improved intermolecular interactions in their crystal structures. Such intermolecular/intramolecular energy compensation almost always occurs when crystallographic and molecular symmetry elements coincide. For BP and CB, almost exclusively, a crystallographic inversion centre coincides with a centre of symmetry in a Ci-symmetric molecule. We conclude that the observation of higher energy conformers (with the compensation of conformational energies up to ≈8–10 kJ.mol.−1 above the global minimum) together with this symmetry coincidence is rare. The work shows that concerns, expressed by some drug discovery chemists and other scientists, that conformations observed in crystal structures are systematically biased due to ‘crystal packing effects’ is overstated: only 16% of BP and CB fragments are exactly planar in small-molecule crystal structures, while the remaining conformations are close to their gas-phase energy minima. Thus, crystal structure conformations are good guides to conformational preferences in other phases and in other applications, e.g. in conformer generation or in the study of protein–ligand binding.

Hydrogen-bond stabilization in oxyanion holes: grand jeté to three dimensions

We recently reported crystallographic evidence that the hydrogen bonds which can stabilize oxygen-centered negative charge within enzyme oxyanion holes are rarely found in the place they should be expected on the basis of the analysis of small-molecule crystal structures. We investigated this phenomenon using calculations on simplified active site models. A recent paper suggested that several aspects of the analysis required further exploration. In this paper we: (i) review the results of our crystallographic study; (ii) report molecular dynamics studies which investigate the effect of protein movement; (iii) report ONIOM calculations which trace the reaction coordinate for an oxyanion hole reaction in the presence of a complete enzyme active site. These results show that the limitations of gas phase calculations on simplified models do not invalidate our comparison of competing active site geometries. These new results reaffirm the conclusion that oxyanion holes are not usually stabilized by planar arrangements of H-bonds, and that this sub-optimal transition state stabilization leads to better overall catalysis.

A very short history of structure-based design: how did we get here and where do we need to go?

Beddell et al. [1] published the first structure-based design paper in 1976 on hemoglobin ligands using Kendrew wireframe models.The first protein–ligand docking paper was in 1982 [2]. The PDB contained about 200 protein X-ray crystal structures in 1982, but very few were drug discovery targets. Many medicinal chemists believed that relevant macromolecule crystal structures were not likely to ever become available early enough during the lifetime of a drug discovery project to matter. Many also thought that designing an optimal ligand would be trivially obvious if they were lucky enough to have the high-resolution structure of their target. The initial flurry of enthusiasm in the 80s around docking and de novo design yielded to the realization in the 90s that scoring is the hard problem, not design. It’s relatively easy to design a molecule to optimize any given scoring function; the hard part is recognizing which apparently complementary molecules actually will bind to their target. This is the crux of the still-unsolved scoring problem. Predicting free energies of binding in aqueous solution has proven to be far more difficult than most of us realized. Despite 30 years of effort with many different approaches, many different investigators, and massive amounts of computer time, we still cannot reliably predict relative binding free energies with sufficient accuracy to drive organic synthesis during hit-to-lead optimization. CASP (Critical Assessment of Protein Structure, http:// predictioncenter.org) showed the protein folders in 1994 that true, blind prediction is much harder than retrospective ‘‘prediction’’ [3]. Their field has advanced steadily since then. Our own field just began to learn this recently, thanks to the CCDC (Cambridge Crystallographic Data Centre) and SAMPL (http://sampl.eyesopen.com) ‘‘contests’’, plus other critical analyses (e.g., [4]). Blind predictions for small molecule crystal structures began with the CCDC in 1999 [5]. Anthony Nicholls, Vijay Pande, and others started SAMPL0 in 2007 to predict small molecule solvation free energies. SAMPL3 was held this year and included blind prediction of host–guest complexes and trypsin-fragment binding. The initial CCDC contest results were disappointing, but helped stimulate dramatic progress. The most recent CCDC contests proved that high-resolution, accurate, blind prediction of small molecule crystal structures is possible for some small organic molecules, albeit with heroic amounts of computing [6, 7]. This is really exciting work: it proves that current theory and its implementation are sufficient to solve this problem in several non-trivial cases. SAMPL0–3 showed that true blind predictions of solvation free energies, host–guest complexes, protein-fragment structures and relative binding affinities are still remarkably difficult and typically have surprisingly large errors [8–10]. Literature in our field is still cluttered with work claiming to predict what’s already known. It is not even clear where the major errors are (partial charges, dielectric model, fixed versus polarizable force fields, torsion terms, entropy changes due to torsional degrees of freedom, water model, etc.). We still don’t understand why protein–ligand binding free energy doesn’t continue to increase with increasing size of the binding interface [11]. Brute force computing has not helped. Major improvements in our methods are required. The PDB now has over 75,000 structures, of which about 50,000 are protein–ligand complexes (http://www.pdb.org). High-resolution X-ray crystal structures are now routinely J. Blaney (&) Computational Chemistry and Cheminformatics, Small Molecule Drug Discovery, Genentech, 1 DNA Way, South San Francisco, CA 94080, USA e-mail: blaney.jeff@gene.com

Stereoelectronic Requirements for Optimal Hydrogen-Bond-Catalyzed Enolization

Protein crystallographic analysis of the active sites of enolizing enzymes and structural analysis of hydrogen-bonded carbonyl compounds in small molecule crystal structures, complemented by quantum chemical calculations on related model enolization reactions, suggest a new stereoelectronic model that accounts for the observed out-of-plane orientation of hydrogen-bond donors (HBDs) in the oxyanion holes of enolizing enzymes. The computational results reveal that the lone-pair directionality of HBDs characteristic for hydrogen-bonded carbonyls is reduced upon enolization, and the enolate displays almost no directional preference for hydrogen bonding. Positioning the HBDs perpendicular to the carbonyl plane induces strain in the catalyst-substrate complex, which is released upon enolization, resulting in more favorable kinetics and thermodynamics than the in-plane arrangement of HBDs.

QM methods in structure based design: Utility in probing protein–ligand interactions

Small changes in ligand structure can lead to large unexpected changes in activity yet it is often not possible to rationalize these effects using empirical modeling techniques, suggesting more effective methods are required. In this study we investigate the use of high level QM methods to study the interactions found within protein–ligand complexes as improved understanding of these could help in the design of new, more active molecules. We study aspects of ligand binding in a set of protein ligand complexes containing ligand efficient, fragment-like inhibitors as these structures are often challenging to determine experimentally. To assess the reliability of our theoretical models we compare the MP2/6-31+G** QM results to the original X-ray coordinates and to QM/MM B3LYP/6-31G*//UFF results which we have previously reported. We also contrast these results with data obtained from an analysis of the distribution of comparable interactions found in (a) high resolution kinase complexes (≤1.8 Å) from the PDB and (b) more generic, small molecule crystal structures from the CSD.

A simple procedure for the derivation of electron density based surfaces of drug-receptor complexes from a combination of X-ray data and theoretical calculations.

To contribute to an understanding of biological recognition and interaction, an easy-to-use procedure was developed to generate and display molecular surfaces and selected electron density based surface properties. To overcome the present limitations to derive electron densities of macromolecules, the considered systems were reduced to appropriate substructures around the active centers. The combination of experimental X-ray structural information and aspherical atomic electron density data from theoretical calculations resulted in properties like the electrostatic potential and the Hirshfeld surface which allowed a study of electronic complementarity and the identification of sites and strengths of drug-receptor interactions. Applications were examined for three examples. The anilinoquinazoline gefitinib (Iressa(R)) belongs to a new class of anticancer drugs that inhibit the tyrosine kinase activity of the epidermal growth factor receptor (EGFR). In the second example, the interaction of epoxide inhibitors with the main protease of the SARS coronavirus was investigated. Furthermore, the progesterone receptor complex was examined. The quantitative analysis of hydrogen bonding in the chosen substructure systems follows a progression elaborated earlier on the basis of accurate small molecule crystal structures. This finding and results from modified substructures suggest that also the surface properties seem robust enough to provide stable information about the recognition of interacting biomolecular species although they are obtained from medium molecular weighted subfragments of macromolecular complexes, which consist of no more than approximately 40 residues.

Cellulose and the twofold screw axis: modeling and experimental arguments

Crystallography indicates that molecules in crystalline cellulose either have twofold screw-axis (21) symmetry or closely approximate it, leading to short distances between H4 and H1′ across the glycosidic linkage. Therefore, modeling studies of cellobiose often show elevated energies for 21 structures, and experimental observations are often interpreted in terms of intramolecular strain. Also, some computer models of cellulose crystallites have an overall twist as well as twisted individual chains, again violating 21 symmetry. To gain insight on the question of inherent strain in 21 structures, modeling was employed and crystal structures of small molecules were surveyed. (Residues in a disaccharide cannot be related by 21 symmetry because they are not identical but if their linkage geometry would lead to 21 symmetry for an infinite cellulose chain, the disaccharide would have 21 pseudo symmetry.) Several initial structures in quantum mechanics (QM) studies of cellobiose minimized to structures having 21 pseudo symmetry. Similarly, a number of relevant small molecules in experimental crystal structures have pseudo symmetry. While the QM models of cellobiose with 21 pseudo symmetry had inter-residue hydrogen bonding, the experimental studies included cellotriose undecaacetate, a molecule that cannot form conventional hydrogen bonds. Limitations in characterizing symmetry based on the linkage torsion angles ϕ and ψ were also explored. It is concluded that 21 structures have little intrinsic strain, despite indications from empirical models.

The X-ray Structure of RU486 Bound to the Progesterone Receptor in a Destabilized Agonistic Conformation

Here we describe the 1.95 A structure of the clinically used antiprogestin RU486 (mifepristone) in complex with the progesterone receptor (PR). The structure was obtained by taking a crystal of the PR ligand binding domain containing the agonist norethindrone and soaking it in a solution containing the antagonist RU486 for extended times. Clear ligand exchange could be observed in one copy of the PR ligand binding domain dimer in the crystal. RU486 binds while PR is in an agonistic conformation without displacing helix 12. Although this is probably because of the constraints of the crystal lattice, it demonstrates that helix 12 displacement is not a prerequisite for RU486 binding. Interestingly, B-factor analysis clearly shows that helix 12 becomes more flexible after RU486 binding, suggesting that RU486, being a model antagonist, does not induce one fixed conformation of helix 12 but changes its positional equilibrium. This conclusion is confirmed by comparing the structures of RU486 bound to PR and RU486 bound to the glucocorticoid receptor.

Orthogonal dipolar interactions between amide carbonyl groups

Orthogonal dipolar interactions between amide C=O bond dipoles are commonly found in crystal structures of small molecules, proteins, and protein-ligand complexes. We herein present the experimental quantification of such interactions by employing a model system based on a molecular torsion balance. Application of a thermodynamic double-mutant cycle allows for the determination of the incremental energetic contributions attributed to the dipolar contact between 2 amide C=O groups. The stabilizing free interaction enthalpies in various apolar and polar solvents amount to -2.73 kJ mol(-1) and lie in the same range as aromatic-aromatic C-H...pi and pi-pi interactions. High-level intermolecular perturbation theory (IMPT) calculations on an orthogonal acetamide/N-acetylpyrrole complex in the gas phase at optimized contact distance predict a favorable interaction energy of -9.71 kJ mol(-1). The attractive dipolar contacts reported herein provide a promising tool for small-molecule crystal design and the enhancement of ligand-protein interactions during lead optimization in medicinal chemistry.

Close Contacts between Carbonyl Oxygen Atoms and Aromatic Centers in Protein Structures: π···π or Lone-Pair···π Interactions?

Lone-pair...pi and, more recently, pi...pi interactions have been studied in small molecule crystal structures, and they are the focus of attention in some biomolecules. In this study, we have systematically analyzed 500 high-resolution protein structures (resolution < or =1.8 A) and identified 286 examples in which carbonyl oxygen atoms approach the aromatic centers within a distance of 3.5 A. Contacts involving backbone carbonyl oxygens are frequently observed in helices and, to some extent, in strands. Geometrical characterization indicates that these contacts have geometry in between that of an ideal pi...pi and a lone-pair...pi interaction. Quantum mechanical calculations using 6-311++G** basis sets reveal that these contacts give rise to energetically favorable interactions and, along with MD simulations, indicate that such interactions could stabilize secondary structures.

Introduction and validation of an invariom database for amino-acid, peptide and protein molecules

A database of invarioms for structural refinement of amino-acid, oligopeptide and protein molecules is presented. The spherical scattering factors of the independent atom or promolecule model are replaced by ;individual' aspherical scattering factors that take into account the chemical environment of a bonded atom. All amino acids were analysed in terms of their invariom fragments. In order to generate 73 database entries that cover this class of compounds, 37 model compounds were geometry-optimized and theoretical structure factors were calculated. Multipole refinements were then performed on these theoretical structure factors to yield the invariom database. Validation of this database on an extensive number of experimental small-molecule crystal structures of varying quality and resolution shows that invariom modelling improves various figures of merit. Differences in figures of merit between invariom and promolecule models give insight into the importance of disorder for future protein-invariom refinements. The suitability of structural data for application of invarioms can be predicted by Cruickshank's diffraction-component precision index [Cruickshank (1999), Acta Cryst. D55, 583-601].

An Experiment in Crystal Structure Prediction by Popular Vote

The ability to identify the crystal structures of small molecules by visual inspection, given a list of computer-generated low-energy possibilities, has been tested in an experiment conducted at an international crystallographic conference. The surprising result of the test was that the experimentally observed crystal structures were the least popular of five choices for the two molecules included in the test, casting doubt on the reliability of crystallographic intuition as a complement to computational methods in crystal structure prediction.