Building a QSAR model of a new biological target for which few screening data are available is a statistical challenge. However, the new target may be part of a bigger family, for which we have more screening data. Collaborative filtering or, more generally, multi-task learning, is a machine learning approach that improves the generalization performance of an algorithm by using information from related tasks as an inductive bias. We use collaborative filtering techniques for building predictive models that link multiple targets to multiple examples. The more commonalities between the targets, the better the multi-target model that can be built. We show an example of a multi-target neural network that can use family information to produce a predictive model of an undersampled target. We evaluate JRank, a kernel-based method designed for collaborative filtering. We show their performance on compound prioritization for an HTS campaign and the underlying shared representation between targets. JRank outperformed the neural network both in the single-and multi-target models.
Given the high homology in amino acid sequence between the ␦-opioid receptor and the two other types ( and ), distinct residues in this receptor may confer its selectivity to some ligands. In order to identify molecular determinants in the human ␦ receptor responsible for the selectivity of ␦-selective ligands, two different ␦/ chimeras were constructed. In the first one, the ␦ sequence from the top of transmembrane 5 to the C terminus was replaced by the equivalent sequence, and in the second one, 13 consecutive residues in the third extracellular loop region of the ␦ receptor were replaced by the counterpart. These two chimeras retained the ability to bind the nonselective bremazocine but completely lost the ability to bind different ␦-selective ligands. These results suggested that the region of the third extracellular loop of the ␦ receptor is crucial for the type selectivity. Opioid receptors are cell surface glycoproteins that constitute specific binding sites for a variety of compounds used for treating pain. Extensive pharmacological studies led to the definition of three opioid receptor types, , , and ␦ (1, 2). Indeed, the availability of highly selective ligands permitted a better characterization of these three types of receptors. In the past few years, cDNAs encoding the , , and ␦ of different species have been cloned (reviewed in Ref.3). However, human receptors represent the ultimate therapeutic targets. Thus, the cloning of human , , and ␦ cDNAs (4 -6) provided particularly relevant tools for opioid drug discovery.Analysis of the predicted amino acid sequences of the three human opioid receptors has shown that these receptors have characteristics common to the guanine nucleotide-binding regulatory protein-coupled receptors with an extracellular N-terminal domain, a cytoplasmic C-terminal domain, and seven putative transmembrane domains (7,8). Moreover, given their high degree of identity (approximately 60%) with the highest similarity in the transmembrane domains and intracellular loops, distinct residues in these receptors may confer their selectivity to some ligands. Binding sites of selective ligands most probably reside, at least in part, in the divergent regions that are the extracellular loops and the N-terminal domain. The C-terminal domain is also a divergent region, but its involvement in binding of ligands is unlikely.All of the opioid analgesics currently used clinically interact with the receptor and are known to induce severe side effects and addiction (9, 10). These major disadvantages were less pronounced with the use of a number of agonists selective for the receptor which also display analgesic properties (11-13), suggesting that some effective analgesics with minimal side effects, selectively interacting with one of the three types of opioid receptors, may be developed. Particularly, ␦ receptors that bind enkephalin-like peptides with high affinity have been proposed to mediate analgesia (14 -16) and to induce weak opiate physical dependence (17, 18), making them an interesting t...
A new type of thrombin exo-site inhibitor has been designed with enhanced inhibitory potency and increased metabolic stability. With the aid of the model of the structure of the thrombin-hirudin fragment complex [Yue, S.-Y., DiMaio, J., Szewczuk, Z., Purisima, E. O., Ni, F., & Konishi, Y. (1992) Protein Eng. 5, 77-85], cyclic analogs of the hirudin fragment (hirudin55-65) were designed and synthesized. In these analogs, the side chains of appropriately substituted residues, 58 and 61, were joined in order to restrict the conformation of the inhibitor. An analog with an 18-membered lactam ring showed higher antithrombin activity (IC50 = 0.57 microM) than the corresponding analogs with 17- or 16-membered rings and was 2-fold more potent than its linear counterpart. Even 4-fold greater enhancement was obtained when a shorter fragment, hirudin 55-62, was cyclized. This cyclization not only improved the potency but, more importantly, dramatically increased the resistance to proteolytic digestion. Remarkable enhancement of stability to proteolysis was observed for peptide bonds located in the exocyclic linear peptide segments. These results are discussed using molecular modeling.
Nonpeptide delta opioid agonists are analgesics with a potentially improved side-effect and abuse liability profile, compared to classical opioids. Andrews analysis of the NIH nonpeptide lead SNC-80 suggested the removal of substituents not predicted to contribute to binding. This approach led to a simplified lead, N, N-diethyl-4-[phenyl(1-piperazinyl)methyl]benzamide (1), which retained potent binding affinity and selectivity to the human delta receptor (IC(50) = 11 nM, mu/delta = 740, kappa/delta > 900) and potency as a full agonist (EC(50) = 36 nM) but had a markedly reduced molecular weight, only one chiral center, and increased in vitro metabolic stability. From this lead, the key pharmacophore groups for delta receptor affinity and activation were more clearly defined by SAR and mutagenesis studies. Further structural modifications on the basis of 1 confirmed the importance of the N, N-diethylbenzamide group and the piperazine lower basic nitrogen for delta binding, in agreement with mutagenesis data. A number of piperazine N-alkyl substituents were tolerated. In contrast, modifications of the phenyl group led to the discovery of a series of diarylmethylpiperazines exemplified by N, N-diethyl-4-[1-piperazinyl(8-quinolinyl)methyl]benzamide (56) which had an improved in vitro binding profile (IC(50) = 0.5 nM, mu/delta = 1239, EC(50) = 3.6 nM) and increased in vitro metabolic stability compared to SNC-80.
The design of low molecular weight thrombin inhibitors IIa-d (hirutonins) that bind concurrently with the enzyme's catalytic site and auxiliary "anion-binding exosite" for fibrinogen recognition is reported. A practical synthesis of the required homologous ketomethylene arginyl dipeptide inserts [Arg psi CO(CH2)nCO] (n = 1-4) corresponding to the P1-P1' scissile position of hirutonins is described. The substitution of the scissile amide function by a ketomethylene group is compatible with the enzyme active site and conferred complete plasma proteolytic stability. This modification also enhanced enzyme affinity up to 20-fold with hirutonin-4 (IIb, n = 4) displaying highest affinity (Ki = 140 +/- 20 pM). Hirutonins 1-4 exhibited potent inhibition of plasma prothrombin time (PT) and activated partial thromboplastin time (aPTT). The inhibition was biphasic and showed good correlation with the corresponding Ki. Hirutonin-2 inhibited thrombin-mediated platelet aggregation and exhibited a strong antithrombotic effect comparable to r-hirudin in an in vivo rat arteriovenous shunt model (ED15 = 1.20 mg/kg for hirutonin-2 and 1.14 mg/kg for r-hirudin). Lower molecular weight inhibitors were obtained by substituting the six native amino acid residues (Q-S-H-N-D-G), connecting the active site and the auxiliary exosite binding elements with a variable number of interening omega-aminopentenoyl units. In addition, the exosite component was reduced to seven amino acid residues (D-F-E-P-I-P-L). Incorporation of these modifications into the bifunctional format resulted in nanomolar thrombin inhibitory peptides (IIIa-c). The resulting inhibitors were studied by molecular modeling with alpha-thrombin, and the bimolecular interactions served to explain the retention of high enzyme affinity.
N alpha-Acetyl[D-Phe45,Arg47]hirudin45-65 (P53) is a bivalent thrombin inhibitor (Ki = 5.6 nM) that consists of an active site inhibitor segment, [N alpha-acetyl-(dF)PRP]; a fibrinogen recognition exo site inhibitor segment, hirudin55-65 (DFEEIPEEYLQ-OH); and a linker, hirudin49-54 (QSHNDG), connecting these inhibitor segments (DiMaio et al., 1990). The structure-function relationships of the linker were studied using a combination of various omega-amino acids, which modified the length of the linker as well as the number and the locations of peptide bonds. Linkers with 14-18 atoms (counting only the atoms contributing to the length of the linker) showed a competitive inhibition with Ki = 1.7-3.4 nM. The potency of the inhibitors with 12-13-atom linkers was sensitive to the chemical structure of the linker. The high-potency inhibitors showed a competitive inhibition, while the low-potency inhibitors showed a hyperbolic inhibition. Among them, an inhibitor with a 13-atom linker showed the highest potency (Ki = 0.51 nM, an 11-fold improvement from that of P53 above), indicating that this is an optimal linker length. Since linkers with 6-10 atoms failed to bridge the active site and exo site inhibitor segments, a minimum of 11 atoms was required to bridge them, even though the potency of the inhibitor with an 11-atom linker was weak (Ki = 26 nM). Molecular dynamics simulation of the inhibitors with 13-atom linkers suggested that some linkers serve as a functional domain with the amide bond of the linker interacting with thrombin through hydrogen bonds.(ABSTRACT TRUNCATED AT 250 WORDS)
A potent thrombin inhibitor, [D-Phe45, Arg47] hirudin 45-65, that contains an active site-directed sequence D-Phe-Pro-Arg-Pro, an exosite specific fragment hirudin 55-65 (H55-65) and a linker portion hirudin 49-54, was designed based on the hirudin sequence [DiMaio et al. (1990) J. Biol. Chem., 265, 21698-21798]. A three-dimensional model of the complex between the B-chain of human thrombin and the inhibitor [D-Phe45, Arg47] hirudin 45-65 was constructed using molecular modelling starting from the X-ray C alpha coordinates of the thrombin-hirudin complex and the NMR-derived structure of the thrombin-bound hirudin 55-65. The contribution of the H49-54 fragment to the thrombin-inhibitor interaction was deduced by examining a series of analogs containing single glycine substitution and analogs with reduced number of residues within the linker. The results were consistent with the molecular modelling observations i.e. the H49-54 fragment serves the role of a spacer in the binding interaction and could be replaced by four glycine residues. The studies on the interaction of the exosite-directed portion of the inhibitor with thrombin using a series of synthetic H55-65 analogs demonstrated that residues AspH55 to ProH60 play a major role in binding to human thrombin where the side chains of PheH56, IleH59 and GluH57 showed critical contributions. Molecular modelling suggested that these side chains may contribute to inter- and intramolecular hydrophobic and electrostatic interactions, respectively.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.