BUNDLE: a program for building the transmembrane domains of G-protein-coupled receptors

Modeling of mammalian olfactory receptors and docking of odorants

Olfactory receptors (ORs) belong to the superfamily of G protein-coupled receptors (GPCRs), the second largest class of genes after those related to immunity, and account for about 3 % of mammalian genomes. ORs are present in all multicellular organisms and represent more than half the GPCRs in mammalian species (e.g., the mouse OR repertoire contains >1,000 functional genes). ORs are mainly expressed in the olfactory epithelium where they detect odorant molecules, but they are also expressed in a number of other cells, such as sperm cells, although their functions in these cells remain mostly unknown. It has recently been reported that ORs are present in tumoral tissues where they are expressed at different levels than in healthy tissues. A specific OR is over-expressed in prostate cancer cells, and activation of this OR has been shown to inhibit the proliferation of these cells. Odorant stimulation of some of these receptors results in inhibition of cell proliferation. Even though their biological role has not yet been elucidated, these receptors might constitute new targets for diagnosis and therapeutics. It is important to understand the activation mechanism of these receptors at the molecular level, in particular to be able to predict which ligands are likely to activate a particular receptor ('deorphanization') or to design antagonists for a given receptor. In this review, we describe the in silico methodologies used to model the three-dimensional (3D) structure of ORs (in the more general framework of GPCR modeling) and to dock ligands into these 3D structures.

The use of G-protein coupled receptor models in lead optimization

With the emerging new crystal structures of G-protein coupled receptors (GPCRs), the number of reported in silico receptor models vastly increases every year. The use of these models in lead optimization (LO) is investigated here. Although there are many studies where GPCR models are used to identify new chemotypes by virtual screening, the classical application in LO is rarely reported. The reason for this may be that the quality of a model, which is appropriate for atomistic modeling, must be very high, and the biology of GPCR ligand-dependent signaling is still not fully understood. However, the few reported studies show that GPCR models can be used efficiently in LO for various problems, such as affinity optimization or tuning of physicochemical parameters.

Active machine learning for transmembrane helix prediction

Background

About 30% of genes code for membrane proteins, which are involved in a wide variety of crucial biological functions. Despite their importance, experimentally determined structures correspond to only about 1.7% of protein structures deposited in the Protein Data Bank due to the difficulty in crystallizing membrane proteins. Algorithms that can identify proteins whose high-resolution structure can aid in predicting the structure of many previously unresolved proteins are therefore of potentially high value. Active machine learning is a supervised machine learning approach which is suitable for this domain where there are a large number of sequences but only very few have known corresponding structures. In essence, active learning seeks to identify proteins whose structure, if revealed experimentally, is maximally predictive of others.

Results

An active learning approach is presented for selection of a minimal set of proteins whose structures can aid in the determination of transmembrane helices for the remaining proteins. TMpro, an algorithm for high accuracy TM helix prediction we previously developed, is coupled with active learning. We show that with a well-designed selection procedure, high accuracy can be achieved with only few proteins. TMpro, trained with a single protein achieved an F-score of 94% on benchmark evaluation and 91% on MPtopo dataset, which correspond to the state-of-the-art accuracies on TM helix prediction that are achieved usually by training with over 100 training proteins.

Conclusion

Active learning is suitable for bioinformatics applications, where manually characterized data are not a comprehensive representation of all possible data, and in fact can be a very sparse subset thereof. It aids in selection of data instances which when characterized experimentally can improve the accuracy of computational characterization of remaining raw data. The results presented here also demonstrate that the feature extraction method of TMpro is well designed, achieving a very good separation between TM and non TM segments.

Transmembrane helix prediction using amino acid property features and latent semantic analysis

Background

Prediction of transmembrane (TM) helices by statistical methods suffers from lack of sufficient training data. Current best methods use hundreds or even thousands of free parameters in their models which are tuned to fit the little data available for training. Further, they are often restricted to the generally accepted topology "cytoplasmic-transmembrane-extracellular" and cannot adapt to membrane proteins that do not conform to this topology. Recent crystal structures of channel proteins have revealed novel architectures showing that the above topology may not be as universal as previously believed. Thus, there is a need for methods that can better predict TM helices even in novel topologies and families.

Results

Here, we describe a new method "TMpro" to predict TM helices with high accuracy. To avoid overfitting to existing topologies, we have collapsed cytoplasmic and extracellular labels to a single state, non-TM. TMpro is a binary classifier which predicts TM or non-TM using multiple amino acid properties (charge, polarity, aromaticity, size and electronic properties) as features. The features are extracted from sequence information by applying the framework used for latent semantic analysis of text documents and are input to neural networks that learn the distinction between TM and non-TM segments. The model uses only 25 free parameters. In benchmark analysis TMpro achieves 95% segment F-score corresponding to 50% reduction in error rate compared to the best methods not requiring an evolutionary profile of a protein to be known. Performance is also improved when applied to more recent and larger high resolution datasets PDBTM and MPtopo. TMpro predictions in membrane proteins with unusual or disputed TM structure (K⁺ channel, aquaporin and HIV envelope glycoprotein) are discussed.

Conclusion

TMpro uses very few free parameters in modeling TM segments as opposed to the very large number of free parameters used in state-of-the-art membrane prediction methods, yet achieves very high segment accuracies. This is highly advantageous considering that high resolution transmembrane information is available only for very few proteins. The greatest impact of TMpro is therefore expected in the prediction of TM segments in proteins with novel topologies. Further, the paper introduces a novel method of extracting features from protein sequence, namely that of latent semantic analysis model. The success of this approach in the current context suggests that it can find potential applications in other sequence-based analysis problems.

Availability

and

Modelling the structures of G protein-coupled receptors aided by three-dimensional validation

Background

G protein-coupled receptors (GPCRs) are abundant, activate complex signalling and represent the targets for up to ~60% of pharmaceuticals but there is a paucity of structural data. Bovine rhodopsin is the first GPCR for which high-resolution structures have been completed but significant variations in structure are likely to exist among the GPCRs. Because of this, considerable effort has been expended on developing in silico tools for refining structures of individual GPCRs. We have developed REPIMPS, a modification of the inverse-folding software Profiles-3D, to assess and predict the rotational orientation and vertical position of helices within the helix bundle of individual GPCRs. We highlight the value of the method by applying it to the Baldwin GPCR template but the method can, in principle, be applied to any low- or high-resolution membrane protein template or structure.

Results

3D models were built for transmembrane helical segments of 493 GPCRs based on the Baldwin template, and the models were then scored using REPIMPS and Profiles-3D. The compatibility scores increased significantly using REPIMPS because it takes into account the physicochemical properties of the (lipid) environment surrounding the helix bundle. The arrangement of helices in the helix bundle of the 493 models was then altered systematically by rotating the individual helices. For most GPCRs in the set, changes in the rotational position of one or more helices resulted in significant improvement in the compatibility scores. In particular, for most GPCRs, a rotation of helix VII by 240–300° resulted in improved scores. Bovine rhodopsin modelled using this method showed 3.31 Å RMSD to its crystal structure for 198 C^α atom pairs, suggesting the utility of the method even when starting with idealised structures such as the Baldwin template.

Conclusion

We have developed an in silico tool which can be used to test the validity of, and refine, models of GPCRs with respect to helix rotation and vertical position based on the physicochemical properties of amino acids and the surrounding environment. The method can be applied to any multi-pass membrane protein and potentially can be used in combination with other high-throughput methodologies to generate and refine models of membrane proteins.

Advances in the Development and Application of Computational Methodologies for Structural Modeling of G-Protein Coupled Receptors

BACKGROUND: Despite the large amount of experimental data accumulated in the past decade on G-protein coupled receptor (GPCR) structure and function, understanding of the molecular mechanisms underlying GPCR signaling is still far from being complete, thus impairing the design of effective and selective pharmaceuticals. OBJECTIVE: Understanding of GPCR function has been challenged even further by more recent experimental evidence that several of these receptors are organized in the cell membrane as homo- or hetero-oligomers, and that they may exhibit unique pharmacological properties. Given the complexity of these new signaling systems, researcher's efforts are turning increasingly to molecular modeling, bioinformatics and computational simulations for mechanistic insights of GPCR functional plasticity. METHODS: We review here current advances in the development and application of computational approaches to improve prediction of GPCR structure and dynamics, thus enhancing current understanding of GPCR signaling. RESULTS/CONCLUSIONS: Models resulting from use of these computational approaches further supported by experiments are expected to help elucidate the complex allosterism that propagates through GPCR complexes, ultimately aiming at successful structure-based rational drug design.

Location of the hydrophobic pocket in the binding site of fentanyl analogs in the µ-opioid receptor

Fentanyl is a highly potent and clinically widely used narcotic analgesic. The synthesis of its analogs remains a challenge in an attempt to develop highly selective ?-opioid receptor agonists with specific pharmacological properties. In this paper, the use of flexible molecular docking of several specific fentanyl analogs to the ?-opioid receptor model, in order to test the hypothesis that the hydrophobic pocket accommodates alkyl groups at position 3 of the fentanyl skeleton, is described. The stereoisomers of the following compounds were studied: cis- and trans-3-methylfentanyl, 3,3-dimethylfentanyl, cis- and trans-3-ethylfentanyl, cis- and trans-3-propylfentanyl, cis-3-isopropylfentanyl and cis-3-benzylfentanyl. The optimal position and orientation of these fentanyl analogs in the binding pocket of the ?-receptor, explaining their enantiospecific potency, were determined. It was found that the 3-alkyl group of cis-3R,4S and trans-3S,4S stereoisomers of all the active compounds occupies the hydrophobic pocket between TM5, TM6 and TM7, made up of the amino acids Trp318 (TM7), Ile322 (TM7), Ile301 (TM6) and Phe237 (TM5). However, the fact that this hydrophobic pocket can also accommodate the bulky 3-alkyl substituents of the two inactive compounds: cis-3-isopropylfentanyl, and cis-3-benzylfentanyl, indicates that this hydrophobic pocket in the employed receptor model is probably too large. .

Modeling GPCRs

Many GPCR models have been built over the years for many different purposes, of which drug-design undoubtedly has been the most frequent one. The release of the structure of bovine rhodopsin in August 2000 enabled us to analyze models built before that period to learn things for the models we build today. We conclude that the GPCR modeling field is riddled with "common knowledge". Several characteristics of the bovine rhodopsin structure came as a big surprise, and had obviously not been predicted, which led to large errors in the models. Some of these surprises, however, could have been predicted if the modelers had more rigidly stuck to the rule that holds for all models, namely that a model should explain all experimental facts, and not just those facts that agree with the modeler's preconceptions.

Steric interactions and the activity of fentanyl analogs at the μ-opioid receptor

Fentanyl is a highly potent and clinically widely used narcotic analgesic. The synthesis of its analogs remains a challenge in the attempt to develop highly selective mu-opioid receptor agonists with specific pharmacological properties. In this paper, the use of flexible molecular docking in a study of the formation of complexes between a series of active fentanyl analogs and the mu-opioid receptor is described. The optimal position and orientation of fourteen fentanyl analogs in the binding pocket of the mu-receptor were determined. The major receptor amino acids and the ligand functional groups participating in the complex formation were identified. Stereochemical effects on the potency and binding are explained. The proposed model of ligand-receptor binding is in agreement with point mutation experiments explaining the role of the amino acids: Asp147, Tyr148, Asn230, His297, Trp318, His319, Cys321, and Tyr326 in the complex formation. In addition, the following amino acids were identified as being important for ligand binding or receptor activation: Ile322, Gly325, Val300, Met203, Leu200, Val143, and Ile144.

Structure modeling of all identified G protein-coupled receptors in the human genome

G protein–coupled receptors (GPCRs), encoded by about 5% of human genes, comprise the largest family of integral membrane proteins and act as cell surface receptors responsible for the transduction of endogenous signal into a cellular response. Although tertiary structural information is crucial for function annotation and drug design, there are few experimentally determined GPCR structures. To address this issue, we employ the recently developed threading assembly refinement (TASSER) method to generate structure predictions for all 907 putative GPCRs in the human genome. Unlike traditional homology modeling approaches, TASSER modeling does not require solved homologous template structures; moreover, it often refines the structures closer to native. These features are essential for the comprehensive modeling of all human GPCRs when close homologous templates are absent. Based on a benchmarked confidence score, approximately 820 predicted models should have the correct folds. The majority of GPCR models share the characteristic seven-transmembrane helix topology, but 45 ORFs are predicted to have different structures. This is due to GPCR fragments that are predominantly from extracellular or intracellular domains as well as database annotation errors. Our preliminary validation includes the automated modeling of bovine rhodopsin, the only solved GPCR in the Protein Data Bank. With homologous templates excluded, the final model built by TASSER has a global C_α root-mean-squared deviation from native of 4.6 Å, with a root-mean-squared deviation in the transmembrane helix region of 2.1 Å. Models of several representative GPCRs are compared with mutagenesis and affinity labeling data, and consistent agreement is demonstrated. Structure clustering of the predicted models shows that GPCRs with similar structures tend to belong to a similar functional class even when their sequences are diverse. These results demonstrate the usefulness and robustness of the in silico models for GPCR functional analysis. All predicted GPCR models are freely available for noncommercial users on our Web site (http://www.bioinformatics.buffalo.edu/GPCR).

G protein–coupled receptors (GPCRs) are a large superfamily of integral membrane proteins that transduce signals across the cell membrane. Because of the breadth and importance of the physiological roles undertaken by the GPCR family, many of its members are important pharmacological targets. Although the knowledge of a protein's native structure can provide important insight into understanding its function and for the design of new drugs, the experimental determination of the three-dimensional structure of GPCR membrane proteins has proved to be very difficult. This is demonstrated by the fact that there is only one solved GPCR structure (from bovine rhodopsin) deposited in the Protein Data Bank library. In contrast, there are no human GPCR structures in the Protein Data Bank. To address the need for the tertiary structures of human GPCRs, using just sequence information, the authors use a newly developed threading-assembly-refinement method to generate models for all 907 registered GPCRs in the human genome. About 820 GPCRs are anticipated to have correct topology and transmembrane helix arrangement. A subset of the resulting models is validated by comparison with mutagenesis experimental data, and consistent agreement is demonstrated.

PREDICT modeling and in-silico screening for G-protein coupled receptors

G-protein coupled receptors (GPCRs) are a major group of drug targets for which only one x-ray structure is known (the nondrugable rhodopsin), limiting the application of structure-based drug discovery to GPCRs. In this paper we present the details of PREDICT, a new algorithmic approach for modeling the 3D structure of GPCRs without relying on homology to rhodopsin. PREDICT, which focuses on the transmembrane domain of GPCRs, starts from the primary sequence of the receptor, simultaneously optimizing multiple 'decoy' conformations of the protein in order to find its most stable structure, culminating in a virtual receptor-ligand complex. In this paper we present a comprehensive analysis of three PREDICT models for the dopamine D2, neurokinin NK1, and neuropeptide Y Y1 receptors. A shorter discussion of the CCR3 receptor model is also included. All models were found to be in good agreement with a large body of experimental data. The quality of the PREDICT models, at least for drug discovery purposes, was evaluated by their successful utilization in in-silico screening. Virtual screening using all three PREDICT models yielded enrichment factors 9-fold to 44-fold better than random screening. Namely, the PREDICT models can be used to identify active small-molecule ligands embedded in large compound libraries with an efficiency comparable to that obtained using crystal structures for non-GPCR targets.

Heavier-than-air flying machines are impossible

Many G protein-coupled receptor (GPCR) models have been built over the years. The release of the structure of bovine rhodopsin in August 2000 enabled us to analyze models built before that period to learn more about the models we build today. We conclude that the GPCR modelling field is riddled with 'common knowledge' similar to Lord Kelvin's remark in 1895 that "heavier-than-air flying machines are impossible", and we summarize what we think are the (im)possibilities of modelling GPCRs using the coordinates of bovine rhodopsin as a template. Associated WWW pages: www.gpcr.org/articles/2003_mod

Mutations at position 125 in transmembrane helix III of rhodopsin affect the structure and signalling of the receptor

Mutation of L125R in trasmembrane helix III of rhodopsin, associated with the retinal degenerative disease retinitis pigmentosa, was previously shown to cause structural misfolding of the mutant protein. Also, conservative mutations at this position were found to cause partial misfolding of the mutant receptors. We report here on a series of mutations at position 125 to further investigate the role of Leu125 in the correct folding and function of rhodopsin. In particular, the effect of the size of the substituted amino-acid side chain in the functionality of the receptor, measured as the ability of the mutant rhodopsins to activate the G protein transducin, has been analysed. The following mutations have been studied: L125G, L125N, L125I, L125H, L125P, L125T, L125D, L125E, L125Y and L125W. Most of the mutant proteins, expressed in COS-1 cells, showed reduced 11-cis-retinal binding, red-shifts in the wavelength of the visible absorbance maximum, and increased reactivity towards hydroxylamine in the dark. Thermal stability in the dark was reduced, particularly for L125P, L125Y and L125W mutants. The ability of the mutant rhodopsins to activate the G protein transducin was significantly reduced in a size dependent manner, especially in the case of the bulkier L125Y and L125W substitutions, suggesting a steric effect of the substituted amino acid. On the basis of the present and previous results, Leu125 in transmembrane helix III of rhodopsin, in the vicinity of the beta-ionone ring of 11-cis-retinal, is proposed to be an important residue in maintaining the correct structure of the chromophore binding pocket. Thus, bulky substitutions at this position may affect the structure and signallling of the receptor by altering the optimal conformation of the retinal binding pocket, rather than by direct interaction with the chromophore, as seen from the recent crystallographic structure of rhodopsin.

Modelling G-protein-coupled receptors for drug design

The G-protein coupled receptors form a large and diverse multi-gene superfamily with many important physiological functions. As such, they have become important targets in pharmaceutical research. Molecular modelling and site-directed mutagenesis have played an important role in our increasing understanding of the structural basis of drug action at these receptors. Aspects of this understanding, how these techniques can be used within a drug-design programme, and remaining challenges for the future are reviewed.

3D modeling, ligand binding and activation studies of the cloned mouse delta, mu; and kappa opioid receptors

Refined 3D models of the transmembrane domains of the cloned delta, mu and kappa opioid receptors belonging to the superfamily of G-protein coupled receptors (GPCRs) were constructed from a multiple sequence alignment using the alpha carbon template of rhodopsin recently reported. Other key steps in the procedure were relaxation of the 3D helix bundle by unconstrained energy optimization and assessment of the stability of the structure by performing unconstrained molecular dynamics simulations of the energy optimized structure. The results were stable ligand-free models of the TM domains of the three opioid receptors. The ligand-free delta receptor was then used to develop a systematic and reliable procedure to identify and assess putative binding sites that would be suitable for similar investigation of the other two receptors and GPCRs in general. To this end, a non-selective, 'universal' antagonist, naltrexone, and agonist, etorphine, were used as probes. These ligands were first docked in all sites of the model delta opioid receptor which were sterically accessible and to which the protonated amine of the ligands could be anchored to a complementary proton-accepting residue. Using these criteria, nine ligand-receptor complexes with different binding pockets were identified and refined by energy minimization. The properties of all these possible ligand-substrate complexes were then examined for consistency with known experimental results of mutations in both opioid and other GPCRs. Using this procedure, the lowest energy agonist-receptor and antagonist-receptor complexes consistent with these experimental results were identified. These complexes were then used to probe the mechanism of receptor activation by identifying differences in receptor conformation between the agonist and the antagonist complex during unconstrained dynamics simulation. The results lent support to a possible activation mechanism of the mouse delta opioid receptor similar to that recently proposed for several other GPCRs. They also allowed the selection of candidate sites for future mutagenesis experiments.

Molecular modeling study of the differential ligand-receptor interaction at the mu, delta and kappa opioid receptors

3D models of the opioid receptors mu, delta and kappa were constructed using BUNDLE, an in-house program to build de novo models of G-protein coupled receptors at the atomic level. Once the three opioid receptors were constructed and before any energy refinement, models were assessed for their compatibility with the results available from point-site mutations carried out on these receptors. In a subsequent step, three selective antagonists to each of three receptors (naltrindole, naltrexone and nor-binaltorphamine) were docked onto each of the three receptors and subsequently energy minimized. The nine resulting complexes were checked for their ability to explain known results of structure-activity studies. Once the models were validated, analysis of the distances between different residues of the receptors and the ligands were computed. This analysis permitted us to identify key residues tentatively involved in direct interaction with the ligand.