D3R Grand Challenge 3: blind prediction of protein-ligand poses and affinity rankings.

The Drug Design Data Resource aims to test and advance the state of the art in protein-ligand modeling by holding community-wide blinded, prediction challenges. Here, we report on our third major round, Grand Challenge 3 (GC3). Held 2017-2018, GC3 centered on the protein Cathepsin S and the kinases VEGFR2, JAK2, p38-α, TIE2, and ABL1, and included both pose-prediction and affinity-ranking components. GC3 was structured much like the prior challenges GC2015 and GC2. First, Stage 1 tested pose prediction and affinity ranking methods; then all available crystal structures were released, and Stage 2 tested only affinity rankings, now in the context of the available structures. Unique to GC3 was the addition of a Stage 1b self-docking subchallenge, in which the protein coordinates from all of the cocrystal structures used in the cross-docking challenge were released, and participants were asked to predict the pose of CatS ligands using these newly released structures. We provide an overview of the outcomes and discuss insights into trends and best-practices.

CSAR 2014: A Benchmark Exercise Using Unpublished Data from Pharma.

The 2014 CSAR Benchmark Exercise was the last community-wide exercise that was conducted by the group at the University of Michigan, Ann Arbor. For this event, GlaxoSmithKline (GSK) donated unpublished crystal structures and affinity data from in-house projects. Three targets were used: tRNA (m1G37) methyltransferase (TrmD), Spleen Tyrosine Kinase (SYK), and Factor Xa (FXa). A particularly strong feature of the GSK data is its large size, which lends greater statistical significance to comparisons between different methods. In Phase 1 of the CSAR 2014 Exercise, participants were given several protein-ligand complexes and asked to identify the one near-native pose from among 200 decoys provided by CSAR. Though decoys were requested by the community, we found that they complicated our analysis. We could not discern whether poor predictions were failures of the chosen method or an incompatibility between the participant's method and the setup protocol we used. This problem is inherent to decoys, and we strongly advise against their use. In Phase 2, participants had to dock and rank/score a set of small molecules given only the SMILES strings of the ligands and a protein structure with a different ligand bound. Overall, docking was a success for most participants, much better in Phase 2 than in Phase 1. However, scoring was a greater challenge. No particular approach to docking and scoring had an edge, and successful methods included empirical, knowledge-based, machine-learning, shape-fitting, and even those with solvation and entropy terms. Several groups were successful in ranking TrmD and/or SYK, but ranking FXa ligands was intractable for all participants. Methods that were able to dock well across all submitted systems include MDock,1 Glide-XP,2 PLANTS,3 Wilma,4 Gold,5 SMINA,6 Glide-XP2/PELE,7 FlexX,8 and MedusaDock.9 In fact, the submission based on Glide-XP2/PELE7 cross-docked all ligands to many crystal structures, and it was particularly impressive to see success across an ensemble of protein structures for multiple targets. For scoring/ranking, submissions that showed statistically significant achievement include MDock1 using ITScore1,10 with a flexible-ligand term,11 SMINA6 using Autodock-Vina,12,13 FlexX8 using HYDE,14 and Glide-XP2 using XP DockScore2 with and without ROCS15 shape similarity.16 Of course, these results are for only three protein targets, and many more systems need to be investigated to truly identify which approaches are more successful than others. Furthermore, our exercise is not a competition.

Discovery of GSK1997132B a novel centrally penetrant benzimidazole PPARγ partial agonist.

The peroxisome proliferator-activated receptor γ (PPARγ) is a ligand-activated nuclear receptor, thought to play a role in energy metabolism, glucose homeostasis and microglia-mediated neuroinflammation. A novel benzimidazole series of centrally penetrant PPARγ partial agonists has been identified. The optimization of PPARγ activity and in vivo pharmacokinetics leading to the identification of GSK1997132B a potent, metabolically stable and centrally penetrant PPARγ partial agonist, is described.

Phenoxyacetic acids as PPARδ partial agonists: synthesis, optimization, and in vivo efficacy.

A series of phenoxyacetic acids as subtype selective and potent hPPARδ partial agonists is described. Many analogues were readily accessible via a single solution-phase synthetic route which resulted in the rapid identification of key structure-activity relationships (SAR), and the discovery of two potent exemplars which were further evaluated in vivo. Details of the SAR, optimization, and in vivo efficacy of this series are presented herein.

Structure of Rev-erbalpha bound to N-CoR reveals a unique mechanism of nuclear receptor-co-repressor interaction.

Repression of gene transcription by the nuclear receptor Rev-erbalpha plays an integral role in the core molecular circadian clock. We report the crystal structure of a nuclear receptor-co-repressor (N-CoR) interaction domain 1 (ID1) peptide bound to truncated human Rev-erbalpha ligand-binding domain (LBD). The ID1 peptide forms an unprecedented antiparallel beta-sheet with Rev-erbalpha, as well as an alpha-helix similar to that seen in nuclear receptor ID2 crystal structures but out of register by four residues. Comparison with the structure of Rev-erbbeta bound to heme indicates that ID1 peptide and heme induce substantially different conformational changes in the LBD. Although heme is involved in Rev-erb repression, the structure suggests that Rev-erbalpha could also mediate repression via ID1 binding in the absence of heme. The previously uncharacterized secondary structure induced by ID1 peptide binding advances our understanding of nuclear receptor-co-repressor interactions.

Discovery of a novel class of PPARdelta partial agonists.

Anthranilic acid GW9371 was identified as a novel class of PPARdelta partial agonist through high-throughput screening. The design and synthesis of SAR analogues is described. GSK1115 and GSK7227 show potent partial agonism of the PPARdelta target genes CPT1a and PDK4 in skeletal muscle cells.

Co-crystal structure guided array synthesis of PPARgamma inverse agonists.

PPARgamma-activating thiazolidinediones and carboxylic acids such as farglitazar exert their anti-diabetic effects in part in PPARgamma rich adipose. Both pro- and anti-adipogenic PPARgamma ligands promote glucose and lipid lowering in animal models of diabetes. Herein, we disclose representatives of an array of 160 farglitazar analogues with atypical inverse agonism of PPARgamma in mature adipocytes.

Substituted 2-[(4-aminomethyl)phenoxy]-2-methylpropionic acid PPARalpha agonists. 1. Discovery of a novel series of potent HDLc raising agents.

The peroxisome proliferator activated receptors PPARalpha, PPARgamma, and PPARdelta are ligand-activated transcription factors that play a key role in lipid homeostasis. The fibrates raise circulating levels of high-density lipoprotein cholesterol and lower levels of triglycerides in part through their activity as PPARalpha agonists; however, the low potency and restricted selectivity of the fibrates may limit their efficacy, and it would be desirable to develop more potent and selective PPARalpha agonists. Modification of the selective PPARdelta agonist 1 (GW501516) so as to incorporate the 2-aryl-2-methylpropionic acid group of the fibrates led to a marked shift in potency and selectivity toward PPARalpha agonism. Optimization of the series gave 25a, which shows EC50 = 4 nM on PPARalpha and at least 500-fold selectivity versus PPARdelta and PPARgamma. Compound 25a (GW590735) has been progressed to clinical trials for the treatment of diseases of lipid imbalance.

X-ray crystal structures of the estrogen-related receptor-gamma ligand binding domain in three functional states reveal the molecular basis of small molecule regulation.

X-ray crystal structures of the ligand binding domain (LBD) of the estrogen-related receptor-gamma (ERRgamma) were determined that describe this receptor in three distinct states: unliganded, inverse agonist bound, and agonist bound. Two structures were solved for the unliganded state, the ERRgamma LBD alone, and in complex with a coregulator peptide representing a portion of receptor interacting protein 140 (RIP140). No significant differences were seen between these structures that both exhibited the conformation of ERRgamma seen in studies with other coactivators. Two structures were obtained describing the inverse agonist-bound state, the ERRgamma LBD with 4-hydroxytamoxifen (4-OHT), and the ERRgamma LBD with 4-OHT and a peptide representing a portion of the silencing mediator of retinoid and thyroid hormone action protein (SMRT). The 4-OHT structure was similar to other reported inverse agonist bound structures, showing reorientation of phenylalanine 435 and a displacement of the AF-2 helix relative to the unliganded structures with little other rearrangement occurring. No significant changes to the LBD appear to be induced by peptide binding with the addition of the SMRT peptide to the ERRgamma plus 4-OHT complex. The observed agonist-bound state contains the ERRgamma LBD, a ligand (GSK4716), and the RIP140 peptide and reveals an unexpected rearrangement of the phenol-binding residues. Thermal stability studies show that agonist binding leads to global stabilization of the ligand binding domain. In contrast to the conventional mechanism of nuclear receptor ligand activation, activation of ERRgamma by GSK4716 does not appear to involve a major rearrangement or significant stabilization of the C-terminal helix.

A critical assessment of docking programs and scoring functions.

Docking is a computational technique that samples conformations of small molecules in protein binding sites; scoring functions are used to assess which of these conformations best complements the protein binding site. An evaluation of 10 docking programs and 37 scoring functions was conducted against eight proteins of seven protein types for three tasks: binding mode prediction, virtual screening for lead identification, and rank-ordering by affinity for lead optimization. All of the docking programs were able to generate ligand conformations similar to crystallographically determined protein/ligand complex structures for at least one of the targets. However, scoring functions were less successful at distinguishing the crystallographic conformation from the set of docked poses. Docking programs identified active compounds from a pharmaceutically relevant pool of decoy compounds; however, no single program performed well for all of the targets. For prediction of compound affinity, none of the docking programs or scoring functions made a useful prediction of ligand binding affinity.

Substrate specificity and novel selective inhibitors of TNF-alpha converting enzyme (TACE) from two-dimensional substrate mapping.

We report a systematic analysis of the P1' and P2' substrate specificity of TNF-alpha converting enzyme (TACE) using a peptide library and a novel analytical method, and we use the substrate specificity information to design novel reverse hydroxamate inhibitors. Initial truncation studies, using the amino acid sequence around the cleavage site in precursor-TNF-alpha, showed that good turnover was obtained with the peptide DNP-LAQAVRSS-NH2. Based on this result, 1000 different peptide substrates of the form Biotin-LAQA-P1'-P2'-SSK(DNP)-NH2 were prepared, with 50 different natural and unnatural amino acids at P1' in combination with 20 different amino acids at P2'. The peptides were pooled, treated with purified microsomal TACE, and the reaction mixtures were passed over a streptavidin affinity column to remove unreacted substrate and the N-terminal biotinylated product. C-terminal cleavage products not binding to streptavidin were subjected to liquid chromatography/mass spectrometry analysis where individual products were identified and semiquantitated. 25 of the substrates were resynthesized as discrete peptides and assayed with recombinant TACE. The experiments show that recombinant TACE prefers lipophilic amino acids at the P1' position, such as phenylglycine, homophenylalanine, leucine and valine. At the P2' position, TACE can accommodate basic amino acids, such as arginine and lysine, as well as certain non-basic amino acids such as citrulline, methionine sulfoxide and threonine. These substrate preferences were used in the design of novel reverse hydroxamate TACE inhibitors with phenethyl and 5-methyl-thiophene-methyl side-chains at P1', and threonine and nitro-arginine at P2'.

A ligand-mediated hydrogen bond network required for the activation of the mineralocorticoid receptor.

Ligand binding is the first step in hormone regulation of mineralocorticoid receptor (MR) activity. Here, we report multiple crystal structures of MR (NR3C2) bound to both agonist and antagonists. These structures combined with mutagenesis studies reveal that maximal receptor activation involves an intricate ligand-mediated hydrogen bond network with Asn770 which serves dual roles: stabilization of the loop preceding the C-terminal activation function-2 helix and direct contact with the hormone ligand. In addition, most activating ligands hydrogen bond to Thr945 on helix 10. Structural characterization of the naturally occurring S810L mutant explains how stabilization of a helix 3/helix 5 interaction can circumvent the requirement for this hydrogen bond network. Taken together, these results explain the potency of MR activation by aldosterone, the weak activation induced by progesterone and the antihypertensive agent spironolactone, and the binding selectivity of cortisol over cortisone.

Metalloproteinase inhibitors for the disintegrin-like metalloproteinases ADAM10 and ADAM17 that differentially block constitutive and phorbol ester-inducible shedding of cell surface molecules.

The transmembrane metzinkin-proteases of the ADAM (a disintegrin and a metalloproteinase)-family ADAM10 and ADAM 17 are both implicated in the ectodomain shedding of various cell surface molecules including the IL6-receptor and the transmembrane chemokines CX3CL1 and CXCL16. These molecules are constitutively released from cultured cells, a process that can be rapidly enhanced by cell stimulation with phorbol esters such as PMA. Recent research supports the view that the constitutive cleavage predominantly involves ADAM10 while the inducible one is mediated to a large extent by ADAM17. We here describe the discovery of hydroxamate compounds with different potency against ADAM10 and ADAM17 and different ability to block constitutive and inducible cleavage of IL6R, CX3CL1 and CXCL16 by the two proteases. By screening a number of hydroxamate inhibitors for the inhibition of recombinant metalloproteinases, a compound was found inhibiting ADAM10 with more than 100-fold higher potency than ADAM17, which may be explained by an improved fit of the compound to the S1' specificity pocket of ADAM10 as compared to that of ADAM17. In cell-based cleavage experiments this compound (GI254023X) potently blocked the constitutive release of IL6R, CX3CL1 and CXCL16, which was in line with the reported involvement of ADAM10 but not ADAM17 in this process. By contrast, the compound did not affect the PMA-induced shedding, which was only blocked by GW280264X, a potent inhibitor of ADAM17. As expected, GI254023X did not further decrease the residual release of CX3CL1 and CXCL16 in ADAM10-deficient cells verifying that the compound's effect on the constitutive shedding of these molecules was exclusively due to the inhibition of ADAM10. Thus, GI254023X may by of use as a preferential inhibitor of constitutive shedding events without effecting the inducible shedding in response to agonists acting similar to PMA.

Structural disorder in the complex of human pregnane X receptor and the macrolide antibiotic rifampicin.

The human nuclear xenobiotic receptor, pregnane X receptor (PXR), detects a variety of structurally distinct endogenous and xenobiotic compounds and controls expression of genes central to drug and cholesterol metabolism. The macrolide antibiotic rifampicin, a front-line treatment for tuberculosis, is an established PXR agonist and, at 823 Da, is one of the largest known ligands for the receptor. We present the 2.8 A crystal structure of the ligand-binding domain of human PXR in complex with rifampicin. We also use structural and mutagenesis data to examine the origins of the directed promiscuity exhibited by the PXRs across species. Three structurally flexible loops adjacent to the ligand-binding pocket of PXR are disordered in this crystal structure, including the 200-210 region that is part of a sequence insert novel to the promiscuous PXRs relative to other members of the nuclear receptor superfamily. The 4-methyl-1-piperazinyl ring of rifampicin, which would lie adjacent to the disordered protein regions, is also disordered and not observed in the structure. Taken together, our results indicate that one wall of the PXR ligand-binding cavity can remain flexible even when the receptor is in complex with an activating ligand. These observations highlight the key role that structural flexibility plays in PXR's promiscuous response to xenobiotics.

A structural basis for constitutive activity in the human CAR/RXRalpha heterodimer.

The X-ray crystal structure of the human constitutive androstane receptor (CAR, NR1I3)/retinoid X receptor alpha (RXRalpha, NR2B1) heterodimer sheds light on the mechanism of ligand-independent activation of transcription by nuclear receptors. CAR contains a single-turn Helix X that restricts the conformational freedom of the C-terminal AF2 helix, favoring the active state of the receptor. Helix X and AF2 sit atop four amino acids that shield the CAR ligand binding pocket. A fatty acid ligand was identified in the RXRalpha binding pocket. The endogenous RXRalpha ligand, combined with stabilizing interactions from the heterodimer interface, served to hold RXRalpha in an active conformation. The structure suggests that upon translocation, CAR/RXRalpha heterodimers are preorganized in an active conformation in cells such that they can regulate transcription of target genes. Insights into the molecular basis of CAR constitutive activity can be exploited in the design of inverse agonists as drugs for treatment of obesity.

Progesterone receptor ligand binding pocket flexibility: crystal structures of the norethindrone and mometasone furoate complexes.

Although progesterone, the natural ligand of the progesterone receptor (PR), has a hydrogen atom at the 17alpha position, other potent steroid agonists such as norethindrone and mometasone furoate have larger substituents at this position that are accommodated by the PR ligand binding pocket. Crystallographic analysis of PR ligand binding domain complexes clearly demonstrated that these moieties were accommodated by local shifts of the protein main chain and by adoption of alternative side chain rotamer conformations of ligand-proximal amino acids. These conformational changes imparted a ligand-specific volume to the binding pocket, from 490 A3 in the metribolone complex to 520 A3 in the norethindrone complex, 565 A3 in the progesterone complex, and 730 A3 in the mometasone furoate complex. Despite these marked alterations in binding pocket volume, critical interactions essential for establishment of an active AF2 conformation were maintained.

Crystal structures of the catalytic domain of phosphodiesterase 4B complexed with AMP, 8-Br-AMP, and rolipram.

Phosphodiesterase catalyzes the hydrolysis of the intracellular second messenger 3',5'-cyclic AMP (cAMP) into the corresponding 5'-nucleotide. Phosphodiesterase 4 (PDE4), the major cAMP-specific PDE in inflammatory and immune cells, is an attractive target for the treatment of asthma and COPD. We have determined crystal structures of the catalytic domain of PDE4B complexed with AMP (2.0 A), 8-Br-AMP (2.13 A) and the potent inhibitor rolipram (2.0 A). All the ligands bind in the same hydrophobic pocket and can interact directly with the active site metal ions. The identity of these metal ions was examined using X-ray anomalous difference data. The structure of the AMP complex confirms the location of the catalytic site and allowed us to speculate about the detailed mechanism of catalysis. The high-resolution structures provided the experimental insight into the nucleotide selectivity of phosphodiesterase. 8-Br-AMP binds in the syn conformation to the enzyme and demonstrates an alternative nucleotide-binding mode. Rolipram occupies much of the AMP-binding site and forms two hydrogen bonds with Gln443 similar to the nucleotides.

Subtype specific effects of peroxisome proliferator-activated receptor ligands on corepressor affinity.

Natural ligands for nuclear receptors are believed to activate gene transcription by causing dissociation of corepressors and promoting the association of coactivator proteins. Using multiple biophysical techniques, we find that peptides derived from one of the nuclear receptor interacting motifs of the corepressors nuclear receptor corepressor (NCoR) and silencing mediator of retinoid and thyroid receptors (SMRT) are able to bind the ligand binding domains (LBD) of all three PPAR (peroxisome proliferator activated receptor) subtypes. Using these peptides as tools, we find that ligands designed as selective agonists for PPAR gamma promote the association of coactivator peptides and dissociation of corepressor peptides as expected on PPAR gamma but surprisingly have varied effects on the binding of corepressor peptides to the other PPAR subtypes. In particular, some members of a class of L-tyrosine-based compounds designed as selective agonists for PPAR gamma reduce the affinity for corepressor peptides on PPAR gamma but increase the affinity for the same peptides on PPAR delta and in one case on PPAR alpha. We provide structural data that suggests that the molecular basis for these observations are variations in the ligand binding pockets of the three PPAR subtypes that are perturbed differentially by individual ligands and result in altered presentations of the overlapping coactivator/corepressor binding surfaces.

Activation of nuclear receptors: a perspective from structural genomics.

Crystal structures of more than two dozen different nuclear receptor ligand binding domains have defined a simple paradigm of receptor activation, in which agonist binding induces the activation function-2 (AF-2) helix to form a charge clamp for coactivator recruitment. Recent structural studies present a surprising contrast. Activation of the mouse LRH-1 receptor is independent of a bound agonist despite its large ligand binding pocket, whereas the activation of the Drosophila DHR38 receptor is dependent on ecdysteroids even though the receptor lacks a ligand binding pocket. These new findings shed light on the diverse structural mechanisms that nuclear receptors have evolved for activation, and have important implications in their respective signaling pathways.

The first completed genome sequence from a teleost fish (Fugu rubripes) adds significant diversity to the nuclear receptor superfamily.

Defining complete sets of gene family members from diverse species provides the foundation for comparative studies. Using a bioinformatic approach, we have defined the entire nuclear receptor complement within the first available complete sequence of a non-human vertebrate (the teleost fish Fugu rubripes). In contrast to the human set (48 total nuclear receptors), we found 68 nuclear receptors in the Fugu genome. All 68 Fugu receptors had a clear human homolog, thus defining no new nuclear receptor subgroups. A reciprocal analysis showed that each human receptor had one or more Fugu orthologs, excepting CAR (NR1I3) and LXRbeta (NR1H2). These 68 receptors add striking diversity to the known nuclear receptor superfamily and provide important comparators to human nuclear receptors. We have compared several pharmacologically relevant human nuclear receptors (FXR, LXRalpha/beta, CAR, PXR, VDR and PPARalpha/gamma/delta) to their Fugu orthologs. This comparison included expression analysis across five Fugu tissue types. All of the Fugu receptors that were analyzed by PCR in this study were expressed, indicating that the majority of the additional Fugu receptors are likely to be functional.