Artificial intelligence and machine learning have demonstrated their potential role in predictive chemistry and synthetic planning of small molecules; there are at least a few reports of companies employing in silico synthetic planning into their overall approach to accessing target molecules. A data-driven synthesis planning program is one component being developed and evaluated by the Machine Learning for Pharmaceutical Discovery and Synthesis (MLPDS) consortium, comprising MIT and 13 chemical and pharmaceutical company members. Together, we wrote this perspective to share how we think predictive models can be integrated into medicinal chemistry synthesis workflows, how they are currently used within MLPDS member companies, and the outlook for this field.

The serine hydrolase monoacylglycerol lipase (MGLL) converts the endogenous cannabinoid receptor agonist 2-arachidonoylglycerol (2-AG) and other monoacylglycerols into fatty acids and glycerol. Genetic or pharmacological inactivation of MGLL leads to elevation in 2-AG in the central nervous system and corresponding reductions in arachidonic acid and eicosanoids, producing antinociceptive, anxiolytic, and antineuroinflammatory effects without inducing the full spectrum of psychoactive effects of direct cannabinoid receptor agonists. Here, we report the optimization of hexafluoroisopropyl carbamate-based irreversible inhibitors of MGLL, culminating in a highly potent, selective, and orally available, CNS-penetrant MGLL inhibitor, 28 (ABX-1431). Activity-based protein profiling experiments verify the exquisite selectivity of 28 for MGLL versus other members of the serine hydrolase class. In vivo, 28 inhibits MGLL activity in rodent brain (ED = 0.5-1.4 mg/kg), increases brain 2-AG concentrations, and suppresses pain behavior in the rat formalin pain model. ABX-1431 (28) is currently under evaluation in human clinical trials.

Semisynthetic, mechanism-based protein inhibitors of ubiquitin (Ub) and ubiquitin-like modifier (Ubl) activating enzymes (E1s) have been developed to target E1-catalyzed adenylation and thioesterification of the Ub/Ubl C-terminus during the processes of protein SUMOylation and ubiquitination. The inhibitors were generated by intein-mediated expressed protein ligation, using a truncated Ub/Ubl protein (SUMO residues 1-94; Ub residues 1-71) with a C-terminal thioester, and synthetic tripeptides having a C-terminal adenosine analogue and an N-terminal cysteine residue. SUMO-AMSN (4a) and Ub-AMSN (4b) contain a sulfamide group as a non-hydrolyzable mimic of the phosphate group in the cognate Ub/Ubl-AMP adenylate intermediate in the first half-reaction, and these constructs selectively inhibit SUMO E1 and Ub E1, respectively, in a dose-dependent manner. SUMO-AVSN (5a) and Ub-AVSN (5b) contain an electrophilic vinyl sulfonamide designed to trap the incoming E1 cysteine nucleophile (Uba2 Cys173 in SUMO E1; Uba1 Cys593 in Ub E1) in the second half-reaction, and these constructs selectively, covalently, and stably crosslink to SUMO E1 and Ub E1, respectively, in a cysteine nucleophile-dependent manner. These inhibitors limac@mskcc.org or tand@mskcc.org. The Ub/Ubl modifier is coupled by its C-terminal carboxylate to specific lysine sidechains on target proteins via an isopeptide bond. Initial steps in this process are catalyzed by a Ub/ Ubl activating enzyme (E1), which first adenylates the Ub/Ubl C-terminus to form a Ub/Ubl-AMP intermediate, then transfers the Ub/Ubl to a conserved cysteine on the E1 (Figure 1). The Ub/Ubl is then transthioesterified onto a cysteine side chain of a conjugating enzyme (E2), and finally transferred to a lysine side chain of the target protein, often mediated by a ligase (E3). Although structures of several E1s have been reported, 3,4 outstanding questions remain about the mechanisms of these reactions. First, E1s surprisingly crystallize with substrates bound in the active site, rather than the in situ-formed, presumably tighter binding acyl-AMP intermediate, 5 in contrast to other enyzmes that catalyze adenylation reactions. 6 Second, the conserved E1 cysteine that serves as the nucleophile in the thioesterification half-reaction is remote, >30 Å away from the adenylation active site. These observations suggest that additional conformational changes are required in both half-reactions. 3 To investigate these questions, we sought to develop mechanism-based inhibitors of E1s that could then be used in pivotal structural and biochemical studies. 7 Selective inhibitors would also be useful probes for dissecting E1 functions. We and others have used 5′-sulfonyladenosine-based small molecules to inhibit various mechanistically (but not structurally) related enzymes that catalyze adenylation reactions. 8 We envisioned that such inhibitor design strategies might also be effective for E1s and report herein the development of semisynthetic, C-terminally modified Ub/Ubl proteins as novel, ...

To fully exploit the inherent and enduring potential of natural products for fundamental cell biology and drug lead discovery, synthetic methods for functionalizing unique sites are highly desirable. Here we describe a strategy for the derivatization of natural products at ‘unfunctionalized’ positions via Rh(II)-catalyzed amination enabling simultaneous structure-activity relationship (SAR) studies and arming (alkynylation) of natural products. Employing Du Bois C–H amination, allylic and benzylic C–H bonds underwent amination and olefins underwent aziridination. With tertiary amine-containing natural products, amidines were produced via C–H amination/oxidation and unusual N-aminations provided hydrazine sulfamate inner salts. The alkynylated derivatives are readied for subsequent conjugation to access cellular probes for mechanism of action studies. Both chemo- and site-selectivity was studied by application to a diverse set of natural products including the marine-derived anticancer diterpene, eupalmerin acetate (EPA). Quantitative proteome profiling with an alkynyl EPA derivative obtained by site-selective, allylic C–H amination led to identification of several protein targets in HL-60 cells, including several known to be associated with cancer proliferation, suggestive of a polypharmacological mode of action for EPA.