Developing inhibitors of the guanosine triphosphate hydrolysis accelerating activity of Regulator of G protein Signaling-14

RESEARCH ARTICLE COLLECTION: PHARMACOLOGY AND SIGNAL TRANSDUCTION EDITORS' PICK

Developing inhibitors of the guanosine triphosphate hydrolysis accelerating activity of Regulator of G protein Signaling-fourteen

Regulator of G protein Signaling-fourteen, an intracellular inactivator of G protein-coupled receptor signaling, is considered an undruggable protein, given its shallow and relatively featureless protein-protein interaction interface combined with a distal allosteric site prone to nonspecific inhibition by thiol-reactive compounds. Here, we identify and validate a tractable chemotype that selectively and non-covalently inhibits Regulator of G protein Signaling-fourteen GTPase-accelerating protein activity. Combining structure-guided virtual screening, ligand docking across multiple receptor conformers, and enrichment validation, we progressed from a first-generation active, Z nine zero two seven six one nine seven, to over forty second-generation analogs with improved potency. These inhibitors are predicted to engage the solvent-exposed "canyon" in the Regulator of G protein Signaling-fourteen RGS-box that interacts with the G alpha switch one region. Binding pose predictions underscored the importance of non-polar interactions and shape complementarity over polar interactions in engaging this G alpha-binding canyon and revealed an "ambidextrous" pattern of R one- and R two-group orientations. GTPase inhibition was confirmed in fluorescence-based and gold-standard radioactive GTP hydrolysis assays. Two second-generation analogs, Z five five six six zero zero four three and Z five five six two seven eight four four, inhibited Regulator of G protein Signaling-fourteen GTPase activity in both assays and without measurable cytotoxicity. Deep learning-based scoring of predicted docking poses further supported observed affinity gains from R three-group additions. One analog demonstrated favorable in vivo pharmacokinetics and central nervous system penetration. Collectively, our findings establish tractable, non-covalent, small molecule inhibition of a G protein regulatory interface and illustrate how machine learning-enhanced docking can guide ligand optimization for shallow protein surfaces. This work opens the door to future development of Regulator of G protein Signaling-fourteen inhibitors as potential therapeutics for central nervous system and metabolic disorders.

The Regulators of G protein Signaling proteins were first identified nearly thirty years ago as a then "missing

EDITORS' PICK

regulatory component" of the well-established intracellular pathway of G protein-coupled receptor signaling. Specifically, Regulators of G protein Signaling proteins accelerate the otherwise slow intrinsic guanosine triphosphatase activity of heterotrimeric G protein alpha subunits, leading to G protein-coupled receptor signal termination. As G protein-coupled receptors historically constitute a significant fraction of the druggable proteome, it was thought that uncovering a new intracellular component inhibiting G protein-coupled receptor signaling would quickly unveil novel drug discovery opportunities, especially for central nervous system disorders and diseases. For example, soon after discovering Regulator of G protein Signaling-fourteen and identifying shared central nervous system expression patterns across rodents and humans, its genetic ablation in mice revealed a key central nervous system role as a natural suppressor of CA two neuron synaptic plasticity. Loss of mouse Regulator of G protein Signaling-fourteen expression leads to heightened long-term potentiation and enhanced hippocampal-based learning and memory. Regulator of G protein Signaling-fourteen knockout mice also have increased levels of peripheral brown adipose tissue and exhibit enhancements to metabolism, exercise endurance, and longevity, consistent with the idea that Regulator of G protein Signaling-fourteen normally opposes beneficial "beiging" signaling within mouse adipose tissue via the agonist-occupied GPR thirty-five receptor. Such preclinical findings of enhanced learning, memory, metabolism, and longevity highlight Regulator of G protein Signaling-fourteen as a potential anti-aging target in both the periphery and central nervous system. In support of an adipose-centered metabolism function for human Regulator of G protein Signaling-fourteen, population genome-wide association studies have established associations between human Regulator of G protein Signaling-fourteen sequence variations and triglyceride variations; these associations may be direct or instead an indirect result of renal phosphate handling effects that have recently been detailed at the molecular level for Regulator of G protein Signaling-fourteen. However, even with all these justifications, only modest progress has been made to date in developing Regulator of G protein Signaling protein modulators, possessing reversible binding and physicochemical characteristics common to established therapeutic agents, for preclinical and clinical follow-up for these various indications mentioned above.

The biochemical activity characterizing all twenty "canonical" RGS proteins is Ga-directed GTPase-accelerating protein activity. This GAP activity is technically enzymatic-converting a substrate (Ga.GTP) to a product (Ga.GDP.Pi) without affecting the RGS protein "catalyst". However, this enzymatic activity is in fact mediated by a protein-protein interaction between the RGS domain (or "RGS-box") of the RGS protein and the activated Ga subunit. This protein-protein interaction can be stabilized experimentally when the active site of the Ga subunit is made to mimic the leaving group of inorganic phosphate, via substitution with the planar ion aluminum tetrafluoride. Such protein-protein interactions are historically more challenging to inhibit with small molecule interactors than G protein-coupled receptors or enzymes with defined and voluminous active sites, largely because protein-protein interfaces tend to be large, flat, and topologically featureless, lacking the deep hydrophobic pockets sheltering discrete polar interaction hotspots (e.g., for hydrogen bonding or ionic pairing) that typically accommodate high-affinity small-molecule binding. By screening a approximately three thousand compound library using a flow cytometry-based protein-protein interaction assay that quantified RGS four binding to the G alpha subunit Ga., an early study identified four-chloro-N-[methoxy-(four-nitrophenyl)-24-sulfanylidene]benzenesulfonamide ("CCG-4986") as an RGS four GAP activity inhibitor with an in vitro IC fifty of approximately three to five micromolar. However, the aryl sulfonamide CCG-4986 was found to be a covalent modifier of cysteines within the RGS-box of RGS four, hindering subsequent drug development based on this early hit. Reversible binders, rather than reactive compounds, have better long-term success in a traditional pharmaceutical development pipeline. While an attempt has been made to advance a related compound, CCG-63802, as a reversible, allosteric inhibitor of RGS four, its mechanism of action is also dependent on reactive cysteines within the RGS four RGS-box but distant from the presumptive Ga-interacting face. A different chemotype (one, three-diaryl one, two, four-(four H)-triazol-five-ones) was originally under exploration for treating urinary incontinence given suspected activity at maxi-potassium channels; this chemotype was subsequently thought to inhibit RGS proteins directed against Gaq based on indirect evidence from model organism genetics. However, no direct demonstration of these diaryl-triazolones inhibiting RGS-box GAP function in vitro has yet been published in either patent or public literature.

Given the advent and democratization of virtual compound screening using protein docking algorithms on high-performance computing platforms, we chose to revisit the RGS-box target using newly emerging in silico methods and an accumulated volume of RGS-box structural models attained by X-ray crystallography and NMR spectroscopy. We employed the AtomNet deep-learning system of virtual compound screening on

NMR-derived structural models of the RGS-boxes of human RGS fourteen (PDB id two J N U; UniProt O four three five six six) and its closest paralog, human RGS twelve (PDB id two E B Z; UniProt O one four nine two four). AtomNet, a deep convolutional neural network developed for structure-based drug discovery, was recently applied in one of the largest and most diverse virtual high-throughput screening campaigns to date, covering three hundred eighteen projects across four hundred eighty-two labs in thirty countries and identifying novel drug-like scaffolds across all major therapeutic areas and protein classes. In our application toward RGS-box inhibition, out of ninety-six AtomNet-predicted interacting compounds obtained from a two point five-million-element "Screening Collection" Enamine chemical library, two structurally unrelated compounds were reported as valid micromolar inhibitors in vitro. Here, we report on a successful expansion of one of these active compounds into a large class of one, two, four-triazolo[three, four-b][one, three, four]thiadiazine analogs with verified activities, tested both in the Transcreener GDP RGScreen fluorescence polarization assay of steady-state GTP hydrolysis and in the gold standard RGS protein GAP activity assay: namely, radioactive GTP single-turnover hydrolysis. Given the continuing need for potent and selective RGS protein modulators suitable for preclinical evaluation and eventual therapeutic development, the present study advances this goal by expanding the chemical space targeting RGS fourteen's orthosteric site-a shallow Ga-binding canyon.