Machine learning-aided search for ligands of P2Y6 and other P2Y receptors.

The P2Y6 receptor, activated by uridine diphosphate (UDP), is a target for antagonists in inflammatory, neurodegenerative, and metabolic disorders, yet few potent and selective antagonists are known to date. This prompted us to use machine learning as a novel approach to aid ligand discovery, with pharmacological evaluation at three P2YR subtypes: initially P2Y6 and subsequently P2Y1 and P2Y14. Relying on extensive published data for P2Y6R agonists, we generated and validated an array of classification machine learning model using the algorithms deep learning (DL), adaboost classifier (ada), Bernoulli NB (bnb), k-nearest neighbors (kNN) classifier, logistic regression (lreg), random forest classifier (rf), support vector classification (SVC), and XGBoost (XGB) classifier models, and the common consensus was applied to molecular selection of 21 diverse structures. Compounds were screened using human P2Y6R-induced functional calcium transients in transfected 1321N1 astrocytoma cells and fluorescent binding inhibition at closely related hP2Y14R expressed in CHO cells. The hit compound ABBV-744, an experimental anticancer drug with a 6-methyl-7-oxo-6,7-dihydro-1H-pyrrolo[2,3-c]pyridine scaffold, had multifaceted interactions with the P2YR family: hP2Y6R inhibition in a non-surmountable fashion, suggesting that noncompetitive antagonism, and hP2Y1R enhancement, but not hP2Y14R binding inhibition. Other machine learning-selected compounds were either weak (experimental anti-asthmatic drug AZD5423 with a phenyl-1H-indazole scaffold) or inactive in inhibiting the hP2Y6R. Experimental drugs TAK-593 and GSK1070916 (100 µM) inhibited P2Y14R fluorescent binding by 50% and 38%, respectively, and all other compounds by < 20%. Thus, machine learning has led the way toward revealing previously unknown modulators of several P2YR subtypes that have varied effects.

Stereocontrolled access to thioisosteres of nucleoside di- and triphosphates.

Nucleoside diphosphates and triphosphates impact nearly every aspect of biochemistry; however, the use of such compounds as tools or medicinal leads for nucleotide-dependent enzymes and receptors is hampered by their rapid in vivo metabolism. Although a successful strategy to address the instability of the monophosphate moiety in oligonucleotide therapeutics has been accomplished by their isosteric replacement with phosphorothioates, no practical methods exist to rapidly and controllably access stereopure di- and triphosphate thioisosteres of both natural and unnatural nucleosides. Here we show how a modular, reagent-based platform can enable the stereocontrolled and scalable synthesis of a library of such molecules. This operationally simple approach provides access to pure stereoisomers of nucleoside α-thiodiphosphates and α-thiotriphosphates, as well as symmetrical or unsymmetrical dinucleoside thiodiphosphates and thiotriphosphates (including RNA cap reagents). We demonstrate that ligand-receptor interactions can be dramatically influenced by P-stereochemistry, showing that such thioisosteric replacements can have profound effects on the potency and stability of lead candidates.

Structure activity relationships of 5-HT2B and 5-HT2C serotonin receptor antagonists: N6, C2 and 5'-Modified (N)-methanocarba-adenosine derivatives.

(N)-Methanocarba adenosine derivatives were structurally modified to target 5-HT2B serotonin receptors as antagonists, predominantly containing branched N6-alkyl groups. N6-Dicycloalkyl-methyl groups, including their asymmetric variations, as well as 2-iodo, were found to generally favor 5-HT2Rs, while only N6-dicyclohexyl-methyl derivative 35 showed weak 5-HT2AR affinity (Ki 3.6 µM). The highest 5-HT2BR affinities were Ki 11-23 nM (N6-dicyclopropyl-methyl-2-iodo 11, 2-chloro-5'-deoxy-5'-methylthio 15 and N6-((R)-cyclobuty-cyclopropyl-methyl)-2-iodo 43), and Ki 73 nM at 5-HT2CR for 36. Direct comparison of adenine ribosides and their corresponding rigid (N)-methanocarba derivatives (cf. 51 and MRS8099 45) indicated a multifold affinity enhancement with the bicyclic ring system. Compounds 43, 45 and 48 were functional 5-HT2BR (KB 2-3 nM) and 5-HT2CR (KB 79-328 nM) antagonists in a Gq-mediated calcium flux assay, with 5-HT2BR functional selectivity ranging from 45- (48) to 113-fold (43). Substantial adenosine receptor (AR) affinity (Ki, A1AR < Ki, A3AR < Ki, A2AAR) was still present in this series, suggestive of dual acting compounds: 5-HT2B antagonist and A1AR agonist, potentially useful for treating chronic conditions (fibrosis; pain). Given its affinity (17 nM) and moderate 5-HT2BR binding selectivity (32-fold vs. 5-HT2CR, 4-fold vs. A1AR), 43 (MRS7925) could potentially be useful for anti-fibrotic therapy.

Alicyclic Ring Size Variation of 4-Phenyl-2-naphthoic Acid Derivatives as P2Y14 Receptor Antagonists.

P2Y14 receptor (P2Y14R) is activated by extracellular UDP-glucose, a damage-associated molecular pattern that promotes inflammation in the kidney, lung, fat tissue, and elsewhere. Thus, selective P2Y14R antagonists are potentially useful for inflammatory and metabolic diseases. The piperidine ring size of potent, competitive P2Y14R antagonist (4-phenyl-2-naphthoic acid derivative) PPTN 1 was varied from 4- to 8-membered rings, with bridging/functional substitution. Conformationally and sterically modified isosteres included N-containing spirocyclic (6-9), fused (11-13), and bridged (14, 15) or large (16-20) ring systems, either saturated or containing alkene or hydroxy/methoxy groups. The alicyclic amines displayed structural preference. An α-hydroxyl group increased the affinity of 4-(4-((1R,5S,6r)-6-hydroxy-3-azabicyclo[3.1.1]heptan-6-yl)phenyl)-7-(4-(trifluoromethyl)phenyl)-2-naphthoic acid 15 (MRS4833) compared to 14 by 89-fold. 15 but not its double prodrug 50 reduced airway eosinophilia in a protease-mediated asthma model, and orally administered 15 and prodrugs reversed chronic neuropathic pain (mouse CCI model). Thus, we identified novel drug leads having in vivo efficacy.

In vivo phenotypic validation of adenosine receptor-dependent activity of non-adenosine drugs.

Some non-adenosinergic drugs are reported to also act through adenosine receptors (ARs). We used mouse hypothermia, which can be induced by agonism at any of the four ARs, as an in vivo screen for adenosinergic effects. An AR contribution was identified when a drug caused hypothermia in wild type mice that was diminished in mice lacking all four ARs (quadruple knockout, QKO). Alternatively, an adenosinergic effect was identified if a drug potentiated adenosine-induced hypothermia. Four drugs (dipyridamole, nimodipine, cilostazol, cyclosporin A) increased the hypothermia caused by adenosine. Dipyridamole and nimodipine probably achieved this by inhibition of adenosine clearance via ENT1. Two drugs (cannabidiol, canrenoate) did not cause hypothermia in wild type mice. Four other drugs (nifedipine, ranolazine, ketamine, ethanol) caused hypothermia, but the hypothermia was unchanged in QKO mice indicating non-adenosinergic mechanisms. Zinc chloride caused hypothermia and hypoactivity; the hypoactivity was blunted in the QKO mice. Interestingly, the antidepressant amitriptyline caused hypothermia in wild type mice that was amplified in the QKO mice. Thus, we have identified adenosine-related effects for some drugs, while other candidates do not affect adenosine signaling by this in vivo assay. The adenosine-modulating drugs could be considered for repurposing based on predicted effects on AR activation.

Structure-Activity Relationship and Neuroprotective Activity of 1,5-Dihydro-2H-naphtho[1,2-b][1,4]diazepine-2,4(3H)-diones as P2X4 Receptor Antagonists.

We analyzed the P2X4 receptor structure-activity relationship of a known antagonist 5, a 1,5-dihydro-2H-naphtho[1,2-b][1,4]diazepine-2,4(3H)-dione. Following extensive modification of the reported synthetic route, 4-pyridyl 21u (MRS4719) and 6-methyl 22c (MRS4596) analogues were most potent at human (h) P2X4R (IC50 0.503 and 1.38 µM, respectively, and selective versus hP2X1R, hP2X2/3R, hP2X3R). Thus, the naphthalene 6-, but not 7-position was amenable to substitution, and an N-phenyl ring aza-scan identified 21u with 3-fold higher activity than 5. Compounds 21u and 22c showed neuroprotective and learning- and memory-enhancing activities in a mouse middle cerebral artery occlusion (MCAO) model of ischemic stroke, with potency of 21u > 22c. 21u dose-dependently reduced infarct volume and reduced brain atrophy at 3 and 35 days post-stroke, respectively. Relevant to clinical implication, 21u also reduced ATP-induced [Ca2+]i influx in primary human monocyte-derived macrophages. This study indicates the translational potential of P2X4R antagonists for treating ischemic stroke, including in aging populations.

Synthesis and pharmacological characterization of multiply substituted 2H-chromene derivatives as P2Y6 receptor antagonists.

P2Y6 receptor (P2Y6R) antagonists represent potential drugs for treating cancer, pain, neurodegeneration, asthma, diabetes, colitis and other disorders. However, there are few chemical classes of known competitive antagonists. We recently explored the structure activity relationship (SAR) of 2H-chromene derivatives as P2Y6R antagonists of moderate affinity. New analogues in this series modified at five positions were synthesized and shown to antagonize Ca2+ transients induced by the native agonist UDP in human (h) P2Y6R-expressing (but not turkey P2Y1R-, hP2Y2R- or hP2Y4R-expressing) astrocytoma cells. Alternatives to the reported 2-(trifluoromethyl)- and 3-nitro- substitutions of this scaffold were not identified. However, 6­fluoro 11 and 6­chloro 12 analogues displayed enhanced potency compared to other halogens, although still in the 1 - 2 µM range. Similar halogen substitution at 5, 7 or 8 positions reduced affinity. 5- or 8­Triethylsilylethynyl extension maintained hP2Y6R affinity, with IC50 0.46 µM for 26 (MRS4853). The 6,8­difluoro analogue 27 (IC50 2.99 µM) lacked off-target activities among 45 sites examined, unlike earlier analogues that bound to biogenic amine receptors. 11 displayed only one weak off-target activity (σ2). Mouse P2Y6R IC50s of 5, 25, 26 and 27 were 4.94, 17.6, 6.15 and 17.8 µM, respectively, but most other analogues had reduced affinity (>20 µM) compared to the hP2Y6R. These analogues are suitable for evaluation in in vivo inflammation and cancer models, which will be performed in the future studies.