Extending in Silico Protein Target Prediction Models to Include Functional Effects.

In silico protein target deconvolution is frequently used for mechanism-of-action investigations; however existing protocols usually do not predict compound functional effects, such as activation or inhibition, upon binding to their protein counterparts. This study is hence concerned with including functional effects in target prediction. To this end, we assimilated a bioactivity training set for 332 targets, comprising 817,239 active data points with unknown functional effect (binding data) and 20,761,260 inactive compounds, along with 226,045 activating and 1,032,439 inhibiting data points from functional screens. Chemical space analysis of the data first showed some separation between compound sets (binding and inhibiting compounds were more similar to each other than both binding and activating or activating and inhibiting compounds), providing a rationale for implementing functional prediction models. We employed three different architectures to predict functional response, ranging from simplistic random forest models ('Arch1') to cascaded models which use separate binding and functional effect classification steps ('Arch2' and 'Arch3'), differing in the way training sets were generated. Fivefold stratified cross-validation outlined cascading predictions provides superior precision and recall based on an internal test set. We next prospectively validated the architectures using a temporal set of 153,467 of in-house data points (after a 4-month interim from initial data extraction). Results outlined Arch3 performed with the highest target class averaged precision and recall scores of 71% and 53%, which we attribute to the use of inactive background sets. Distance-based applicability domain (AD) analysis outlined that Arch3 provides superior extrapolation into novel areas of chemical space, and thus based on the results presented here, propose as the most suitable architecture for the functional effect prediction of small molecules. We finally conclude including functional effects could provide vital insight in future studies, to annotate cases of unanticipated functional changeover, as outlined by our CHRM1 case study.

On the Relationship between Molecular Hit Rates in High-Throughput Screening and Molecular Descriptors.

High-throughput screening (HTS) is widely used in the pharmaceutical industry to identify novel chemical starting points for drug discovery projects. The current study focuses on the relationship between molecular hit rate in recent in-house HTS and four common molecular descriptors: lipophilicity (ClogP), size (heavy atom count, HEV), fraction of sp(3)-hybridized carbons (Fsp3), and fraction of molecular framework (f(MF)). The molecular hit rate is defined as the fraction of times the molecule has been assigned as active in the HTS campaigns where it has been screened. Beta-binomial statistical models were built to model the molecular hit rate as a function of these descriptors. The advantage of the beta-binomial statistical models is that the correlation between the descriptors is taken into account. Higher degree polynomial terms of the descriptors were also added into the beta-binomial statistic model to improve the model quality. The relative influence of different molecular descriptors on molecular hit rate has been estimated, taking into account that the descriptors are correlated to each other through applying beta-binomial statistical modeling. The results show that ClogP has the largest influence on the molecular hit rate, followed by Fsp3 and HEV. f(MF) has only a minor influence besides its correlation with the other molecular descriptors.

Development of 7TM receptor-ligand complex models using ligand-biased, semi-empirical helix-bundle repacking in torsion space: application to the agonist interaction of the human dopamine D2 receptor.

Prediction of 3D structures of membrane proteins, and of G-protein coupled receptors (GPCRs) in particular, is motivated by their importance in biological systems and the difficulties associated with experimental structure determination. In the present study, a novel method for the prediction of 3D structures of the membrane-embedded region of helical membrane proteins is presented. A large pool of candidate models are produced by repacking of the helices of a homology model using Monte Carlo sampling in torsion space, followed by ranking based on their geometric and ligand-binding properties. The trajectory is directed by weak initial restraints to orient helices towards the original model to improve computation efficiency, and by a ligand to guide the receptor towards a chosen conformational state. The method was validated by construction of the ß1 adrenergic receptor model in complex with (S)-cyanopindolol using bovine rhodopsin as template. In addition, models of the dopamine D2 receptor were produced with the selective and rigid agonist (R)-N-propylapomorphine ((R)-NPA) present. A second quality assessment was implemented by evaluating the results from docking of a library of 29 ligands with known activity, which further discriminated between receptor models. Agonist binding and recognition by the dopamine D2 receptor is interpreted using the 3D structure model resulting from the approach. This method has a potential for modeling of all types of helical transmembrane proteins for which a structural template with sequence homology sufficient for homology modeling is not available or is in an incorrect conformational state, but for which sufficient empirical information is accessible.

Investigation of D2 receptor-agonist interactions using a combination of pharmacophore and receptor homology modeling.

A combined modeling approach was used to identify structural factors that underlie the structure-activity relationships (SARs) of full dopamine D2 receptor agonists and structurally similar inactive compounds. A 3D structural model of the dopamine D2 receptor was constructed, with the agonist (-)-(R)-2-OH-NPA present in the binding site during the modeling procedure. The 3D model was evaluated and compared with our previously published D2 agonist pharmacophore model. The comparison revealed an inconsistency between the projected hydrogen bonding feature (Ser-TM5) in the pharmacophore model and the TM5 region in the structure model. A new refined pharmacophore model was developed, guided by the shape of the binding site in the receptor model and with less emphasis on TM5 interactions. The combination of receptor and pharmacophore modeling also identified the importance of His3936·55 for agonist binding. This convergent 3D pharmacophore and protein structure modeling strategy is considered to be general and can be highly useful in less well-characterized systems to explore ligand-receptor interactions. The strategy has the potential to identify weaknesses in the individual models and thereby provides an opportunity to improve the discriminating predictivity of both pharmacophore searches and structure-based virtual screens.

Investigation of D1 receptor-agonist interactions and D1/D2 agonist selectivity using a combination of pharmacophore and receptor homology modeling.

The aim of this study was to use a combined structure and pharmacophore modeling approach to extract information regarding dopamine D1 receptor agonism and D1/D2 agonist selectivity. A 3D structure model of the D1 receptor in its agonist-bound state was constructed with a full D1 agonist present in the binding site. Two different binding modes were identified using (+)-doxanthrine or SKF89626 in the modeling procedure. The 3D model was further compared with a selective D1 agonist pharmacophore model. The pharmacophore feature arrangement was found to be in good agreement with the binding site composition of the receptor model, but the excluded volumes did not fully reflect the shape of the agonist binding pocket. A new receptor-based pharmacophore model was developed with forbidden volumes centered on atom positions of amino acids in the binding site. The new pharmacophore model showed a similar ability to discriminate as the previous model. A comparison of the 3D structures and pharmacophore models of D1 and D2 receptors revealed differences in shape and ligand-interacting features that determine selectivity of D1 and D2 receptor agonists. A hydrogen bond pharmacophoric feature (Ser-TM5) was shown to contribute most to the selectivity. Non-conserved residues in the binding pocket that strongly contribute to D1/D2 receptor agonist selectivity were also identified; those were Ser/Cys³·³6, Tyr/Phe5·³8, Ser/Tyr5·4¹, and Asn/His6·55 in the transmembrane (TM) helix region, together with Ser/Ile and Leu/Asn in the second extracellular loop (EC2). This work provides useful information for the design of new selective D1 and D2 agonists. The combined receptor structure and pharmacophore modeling approach is considered to be general, and could therefore be applied to other ligand-protein interactions for which experimental information is limited.

Molecular docking of 2-(benzimidazol-2-ylthio)-N-phenylacetamide-derived small-molecule agonists of human formyl peptide receptor 1.

Human N-formyl peptide receptor 1 (FPR1) is a G protein-coupled receptor (GPCR) involved in host defense and sensing cellular damage. Since structure-based ligand design for many GPCRs, including FPR1, is restricted by the lack of experimental three dimensional structures, homology modeling has been widely used to study GPCR-ligand binding. Indeed, receptor-ligand binding mode predictions can be derived from homology modeling with supporting ligand information. In the present work, we report comparative docking studies of 2-(benzimidazol-2-ylthio)-N-phenylacetamide derived FPR1 agonists, identified here and previously, with several known FPR1 peptide agonists in a FPR1 homology model that is based on the crystal structure of bovine rhodopsin. We found that the binding pocket of the most active molecules shares some common features with high affinity FPR1 peptide agonists, suggesting that they may bind to similar binding sites. Classification tree analysis led to the derivation of a good recognition model based on four amino acid descriptors for distinguishing FPR1 ligands from inactive analogs. Hence, the corresponding residues (Thr199, Arg201, Gly202, and Ala261) can be considered as markers of important areas in the ligand binding site. Concurrently, we identified several unique binding features of benzimidazole derivatives and showed that alkoxy-substituents of the benzimidazole ring are located within a FPR1 hole bounded by Thr199, Thr265, Ile268, and Leu271 or in a groove in the vicinity of Leu198, Arg201, Gly202, and Arg205. The understanding of these molecular features will likely prove beneficial in future design of novel FPR1 agonists based on the benzimidazole scaffold.

Design, synthesis and characterization of a highly effective inhibitor for analog-sensitive (as) kinases.

Highly selective, cell-permeable and fast-acting inhibitors of individual kinases are sought-after as tools for studying the cellular function of kinases in real time. A combination of small molecule synthesis and protein mutagenesis, identified a highly potent inhibitor (1-Isopropyl-3-(phenylethynyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine) of a rationally engineered Hog1 serine/threonine kinase (Hog1(T100G)). This inhibitor has been successfully used to study various aspects of Hog1 signaling, including a transient cell cycle arrest and gene expression changes mediated by Hog1 in response to stress. This study also underscores that the general applicability of this approach depends, in part, on the selectivity of the designed the inhibitor with respect to activity versus the engineered and wild type kinases. To explore this specificity in detail, we used a validated chemogenetic assay to assess the effect of this inhibitor on all gene products in yeast in parallel. The results from this screen emphasize the need for caution and for case-by-case assessment when using the Analog-Sensitive Kinase Allele technology to assess the physiological roles of kinases.

Octopamine receptors from the barnacle balanus improvisus are activated by the alpha2-adrenoceptor agonist medetomidine.

G protein-coupled octopamine receptors of insects and other invertebrates represent counterparts of adrenoceptors in vertebrate animals. The alpha(2)-adrenoceptor agonist medetomidine, which is in clinical use as a veterinary sedative agent, was discovered to inhibit the settling process of barnacles, an important step in the ontogeny of this crustacean species. Settling of barnacles onto ship hulls leads to biofouling that has many harmful practical consequences, and medetomidine is currently under development as a novel type of antifouling agent. We now report that medetomidine induces hyperactivity in the barnacle larvae to disrupt the settling process. To identify the molecular targets of medetomidine, we cloned five octopamine receptors from the barnacle Balanus improvisus. We show by phylogenetic analyses that one receptor (BiOctalpha) belongs to the alpha-adrenoceptor-like subfamily, and the other four (BiOctbeta-R1, BiOctbeta-R2, BiOctbeta-R3, and BiOctbeta-R4) belong to the beta-adrenoceptor-like octopamine receptor subfamily. Phylogenetic analyses also indicated that B. improvisus has a different repertoire of beta-adrenoceptor-like octopamine receptors than insects. When expressed in CHO cells, the cloned receptors were activated by both octopamine and medetomidine, resulting in increased intracellular cAMP or calcium levels. Tyramine activated the receptors but with much lesser potency than octopamine. A hypothesis for receptor discrimination between tyramine and octopamine was generated from a homology three-dimensional model. The characterization of B. improvisus octopamine receptors is important for a better functional understanding of these receptors in crustaceans as well as for practical applications in development of environmentally sustainable antifouling agents.

The annexin I sequence gln(9)-ala(10)-trp(11)-phe(12) is a core structure for interaction with the formyl peptide receptor 1.

The N-terminal part of the calcium-regulated and phospholipid-binding protein annexin AI contains peptide sequences with pro- and anti-inflammatory activities. We have earlier shown that a proinflammatory signal triggered by one of these peptides, Gln(9)-Lys(25), is mediated by FPR1, a member of the formyl peptide receptor family expressed in human neutrophils. To determine the core structure in Gln(9)-Lys(25), smaller peptides were generated, and their capacity to activate neutrophils was determined. A peptide spanning from amino acid Glu(14) to Lys(25) was inactive, whereas the activity was retained in the Gln(9)-Tyr(20) peptide. Removal of amino acids from the C and N terminus of Gln(9)-Tyr(20) revealed that the first amino acid (Gln(9)) was of the utmost importance for activity. The core structure that activated the neutrophil NADPH oxidase to release superoxide anions was Gln(9)-Ala(10)-Trp(11)-Phe(12). This peptide also inhibited the activity induced by N-formyl-Met-Leu-Phe and WKYMVM. A structural model of the peptide agonist-FPR1 complex suggests that the transmembrane part of the binding pocket of the receptor binds optimally to a tetrapeptide. According to the model and the results presented, the N-terminal amino acid glutamine in Gln(9)-Phe(12) is located close to the bottom of the binding cleft, leaving for steric reasons insufficient space to extend the peptide at the N terminus. The addition of amino acids at the C terminus will not affect binding. The model presented may be helpful in developing specific FPR1 ligands.

Selective pharmacophore models of dopamine D(1) and D(2) full agonists based on extended pharmacophore features.

This study is focused on the identification of structural features that determine the selectivity of dopamine receptor agonists toward D(1) and D(2) receptors. Selective pharmacophore models were developed for both receptors. The models were built by using projected pharmacophoric features that represent the main agonist interaction sites in the receptor (the Ser residues in TM5 and the Asp in TM3), a directional aromatic feature in the ligand, a feature with large positional tolerance representing the positively charged nitrogen in the ligand, and sets of excluded volumes reflecting the shapes of the receptors. The sets of D(1) and D(2) ligands used for modeling were carefully selected from published sources and consist of structurally diverse, conformationally rigid full agonists as active ligands together with structurally related inactives. The robustness of the models in discriminating actives from inactives was tested against four ensembles of conformations generated by using different established methods and different force fields. The reasons for the selectivity can be attributed to both geometrical differences in the arrangement of the features, e.g., different tilt angels of the pi system, as well as shape differences covered by the different sets of excluded volumes. This work provides useful information for the design of new D(1) and D(2) agonists and also for comparative homology modeling of D(1) and D(2) receptors. The approach is general and could therefore be applied to other ligand-protein interactions for which no experimental protein structure is available.

Heterodimer formation within universal stress protein classes revealed by an in silico and experimental approach.

Universal stress proteins (Usps) are found in all kingdoms of life and can be divided into four classes by phylogenic analysis. According to available structures, Usps exist as homodimers, and genetic studies show that their cellular assignments are extensive, including functions relating to stress resistance, carbon metabolism, cellular adhesion, motility, and bacterial virulence. We approached the question of how Usps can achieve such a variety of functions in a cell by using a new procedure for statistical analysis of multiple sequence alignments, based on physicochemically related values for each amino acid residue of Usp dimer interfaces. The results predicted that Usp proteins within a class may, in addition to forming homodimers, be able to form heterodimers. Using Escherichia coli Usps as model proteins, we confirmed the existence of such interactions. We especially focused on class I UspA and UspC and demonstrated that they are able to form homo- and heterodimers in vitro and in vivo. We suggest that this ability to form both homo- and heterodimers may allow for an expansion of the functional repertoire of Usps and explains why organisms usually contain multiple usp paralogues.

Mutations in the periplasmic chaperone leading to loss of surface expression of the colonization factor CS6 in enterotoxigenic Escherichia coli (ETEC) clinical isolates.

Enterotoxigenic Escherichia coli (ETEC) cause diarrhoea by adhesion to human enterocytes by one or more colonization factors (CFs) and secretion of heat-labile (LT) and/or heat-stable (ST) enterotoxins. Expression of coli surface antigen 6 (CS6) on the bacterial surface, usually associated with ETEC strains that produce ST alone or in combination with LT, is rarely found in strains expressing only LT. However, a number of LT-only strains which are genotypically positive but phenotypically negative for CS6 have been identified. In this study, eight such strains from India and Guinea-Bissau belonging to different clones were analysed. The CS6 operon cssABCD was transcribed but protein analyses suggested that the structural subunits CssA and CssB of CS6 were absent in the periplasm. Most strains contained truncating mutations within the periplasmic chaperone-encoding gene cssC and protein modelling indicated that this severely affected the substrate-binding capacity of the chaperone. A single-nucleotide polymorphism (SNP) (A-->T) in the 5'-untranslated region of cssC distinguished the eight strains from ETEC strains that do express CS6 on the surface and may be a potential marker for ETEC strains containing phenotypically silent cssABCD. The study emphasizes the importance of using both genotypic and phenotypic methods in epidemiological studies of ETEC, e.g. for vaccine development.

Crystal structure of AcrB in complex with a single transmembrane subunit reveals another twist.

Bacterial drug resistance is a serious concern for human health. Multidrug efflux pumps export a broad variety of substrates out of the cell and thereby convey resistance to the host. In Escherichia coli, the AcrB:AcrA:TolC efflux complex forms a principal transporter for which structures of the individual component proteins have been determined in isolation. Here, we present the X-ray structure of AcrB in complex with a single transmembrane protein, assigned by mass spectrometry as YajC. A specific rotation of the periplasmic porter domain of AcrB is also revealed, consistent with the hypothesized "twist-to-open" mechanism for TolC activation. Growth experiments with yajc-deleted E. coli reveal a modest increase in the organism's susceptibility to beta-lactam antibiotics, but this effect could not conclusively be attributed to the loss of interactions between YajC and AcrB.

Aquaporins in yeasts and filamentous fungi.

Recently, genome sequences from different fungi have become available. This information reveals that yeasts and filamentous fungi possess up to five aquaporins. Functional analyses have mainly been performed in budding yeast, Saccharomyces cerevisiae, which has two orthodox aquaporins and two aquaglyceroporins. Whereas Aqy1 is a spore-specific water channel, Aqy2 is only expressed in proliferating cells and controlled by osmotic signals. Fungal aquaglyceroporins often have long, poorly conserved terminal extensions and differ in the otherwise highly conserved NPA motifs, being NPX and NXA respectively. Three subgroups can be distinguished. Fps1-like proteins seem to be restricted to yeasts. Fps1, the osmogated glycerol export channel in S. cerevisiae, plays a central role in osmoregulation and determination of intracellular glycerol levels. Sequences important for gating have been identified within its termini. Another type of aquaglyceroporin, resembling S. cerevisiae Yfl054, has a long N-terminal extension and its physiological role is currently unknown. The third group of aquaglyceroporins, only found in filamentous fungi, have extensions of variable size. Taken together, yeasts and filamentous fungi are a fruitful resource to study the function, evolution, role and regulation of aquaporins, and the possibility to compare orthologous sequences from a large number of different organisms facilitates functional and structural studies.

A glucocorticoid-responsive mutant androgen receptor exhibits unique ligand specificity: therapeutic implications for androgen-independent prostate cancer.

The cortisol/cortisone-responsive AR (AR(ccr)) has two mutations (L701H and T877A) that were found in the MDA PCa human prostate cancer cell lines established from a castrated patient whose metastatic tumor exhibited androgen-independent growth. Cortisol and cortisone bind to the AR(ccr) with high affinity. In the present study, we characterized the structural determinants for ligand binding to the AR(ccr). Our data revealed that many of the C17, C19, and C21 circulating steroids, at concentrations that are found in vivo, functioned as effective activators of the AR(ccr) but had little or no activity via the wild-type AR or GRalpha. Among the synthetic glucocorticoids tested, dexamethasone activated both GRalpha and AR(ccr), whereas triamcinolone was selective for GRalpha. In MDA PCa 2b cells, growth and prostate-specific antigen production were stimulated by potent AR(ccr) agonists such as cortisol or 9alpha-fluorocortisol but not by triamcinolone (which did not bind to or activate the AR(ccr)). Of the potential antagonists tested, bicalutamide (casodex) and GR antagonist RU38486 showed inhibitory activity. We postulate that corticosteroids provide a growth advantage to prostate cancer cells harboring the promiscuous AR(ccr) in androgen-ablated patients and contribute to their transition to androgen-independence. We predict that triamcinolone, a commonly prescribed glucocorticoid, would be a successful therapeutic agent for men with this form of cancer, perhaps in conjunction with the antagonist casodex. We hypothesize that triamcinolone administration would inhibit the hypothalamic-pituitary-adrenal axis, thus suppressing endogenous corticosteroids, which stimulate tumor growth. Triamcinolone, by itself, would not activate the AR(ccr) or promote tumor growth but would provide glucocorticoid activity essential for survival.