Irreversible Inhibitors of SARS CoV-1 3CLpro
The warhead reactivity and binding affinity to P1′ are both important design considerations to ensure specificity of irreversible inhibitors. Krantz has demonstrated that the inherent chemical reactivity of acyloxymethylketones can be tuned to inhibit cysteine proteases through the modulation of substituent effects on the carboxylate leaving group.37 The importance of leaving group “strength” was confirmed by the strong dependence between the pKa of the leaving carboxylate and cathepsin B enzyme inhibition. These otherwise weak electrophiles are elegant “quiescent” inhibitors that harness the very same interactions with catalytic residues that lead to proteolysis rate acceleration. Molecular modeling of a 2,6-dichlorobenzoate design with SARS CoV-1 3CLpro indicated that a low strain conformation of the ketone carbonyl was aligned in the oxyanion hole and the substituted benzoate is accommodated within the P1′ site. A series of acyloxymethylketone derivatives were prepared, as depicted in Table 1.
Table 1  SARS CoV-1 3CLpro Inhibition Data for Acyloxymethylketone Inhibitors
      SARS CoV-1 3CLpro FRETa
entry  R  kobs/I (M–1 s–1)  IC50 (nM)
12     283,039 ± 22,586   
13  Me     220 ± 0.5
14  cyc-propyl     182 ± 6
15  tert-butyl     230 ± 5
16  Ph     86 ± 3
17  4-OMe-Ph     79 ± 3
18  4-Me-Ph     87 ± 2
19  4-CN-Ph     53 ± 1
20  4-F-Ph     82 ± 3
21  4-Cl-Ph     97 ± 3
22  2,6-(Cl)2-Ph  62,993 ± 2,501   
23  2,6-(F)2-Ph  12,776 ± 594   
24  2-OH-4-Cl-Ph  11,525 ± 40   
25  2-F, 4-CN-Ph  13,321 ± 2,309   
26  2,6-(Me)2-Ph     74 ± 4
27  2,6-(MeO)2-Ph     205 ± 2
28  2-CN-Ph     17 ± 2
a  See the Experimental Section for details on assay methods; the values were calculated from at least eight data points with at least two independent determinations. As expected, chloromethylketone 12 is a potent irreversible inhibitor of SARS CoV-1 3CLpro. Compounds 22–25 possess the most electron-deficient benzoate leaving groups and display kinetics consistent with irreversible inhibition. The remaining entries appear to be reversible inhibitors in the time scale of the assay, with 28 possessing the most potent IC50. A crystal structure (PDB code 6XHN)38 of 28 in complex with SARS CoV-1 3CLpro at 2.25 Å shows a covalent adduct, which demonstrates the bimodal activity of certain acyloxy inhibitors (Figure 4). In this structure, the electron density for the 2-cyanobenzoate moiety is absent and the 3CLpro active-site cysteine (Cys145) sulfur forms a covalent bond to the methylene carbon (1.8 Å C–S bond length). The ketone carbonyl is positioned within the oxyanion hole and as such engaging in hydrogen bonds with backbone NH groups of Gly143 and Cys145. A detailed analysis of an extended hydrogen bond network from catalytic His41 reveals that the side-chain imidazole serves as a hydrogen bond donor to the interior of the protease. Specifically, the imidazole hydrogen is directed toward a lone pair electron acceptor from a structural water. We can conclude that this water must be an acceptor to His41 as one of the water hydrogens is engaged with acceptor electrons on the side chain of Asp187, and the second hydrogen engages a terminating acceptor backbone carbonyl on Asp176 through a short network that includes the side chains of an internal, neutral His164 and Thr175.
Figure 4  Cocrystal structure of the covalent adduct of 28 bound to SARS CoV-1 3CLpro (PDB code 6XHN). The Connolly surface for the inhibitor binding pocket is shown in gray. The bonds are represented as dashed lines, with the bond length between heavy atoms depicted. The 2,6-dichlorobenzoate derivative 22 displays the highest levels of potency in the SARS CoV-1 3CLpro and in antiviral cytopathic effect assays in Vero 76 cells (EC50 = 0.29 ± 0.19 μM).39 We note here that the assay in Vero 76 cells provided data establishing antiviral activity, but the translation of potency measured in Vero cells, which have high efflux potential, to the potency achievable in human lung cells was unknown (vide infra). As expected, 22 possesses low levels of reactivity toward endogenous nucleophiles such as glutathione (t1/2 > 60 min)40 and exhibits high levels of stability in human plasma (t1/2 > 240 min).41 Although 22 displays a promising activity profile, it possesses very poor solubility in clinically relevant IV vehicle formulations, thus limiting its development as an IV clinical agent.
To improve the solubility characteristics of inhibitors incorporating the lipophilic 2,6-dichlorobenzoate, a limited set of inhibitors containing smaller or more polar amine capping fragments were prepared (Table 2). Replacement of the 4-methoxy indole cap present in 22 with a benzimidazole provided 31. This inhibitor displayed potent irreversible inhibition kinetics but did not improve solubility compared to 22. Reduction in the size of the P2 capping element as represented by 29 and 30 provided derivatives that possessed weak reversible SARS CoV-1 3CLpro inhibition.
Table 2  SARS CoV-1 3CLpro Inhibition Data for P3-Modified Acyloxymethylketone Inhibitors
a  See the Experimental Section for details on assay methods; the values were calculated from at least eight data points with at least two independent determinations. Another design to replace the 4-methoxy indole was the (2R)-tetrahydrofuran-2-carboxylate, which molecular modeling suggested could make additional favorable protein interactions with a Gln189 side chain and led to the preparation of 3. Initially, this derivative appeared to display reversible binding kinetics. However, further kinetic studies of compound 3 with the SARS CoV-1 3CLpro protein demonstrated irreversible time-dependent inactivation of this enzyme (kobs/I = 5834 ± 151 M–1 s–1). Additionally, 3 possessed potent antiviral activity in Vero cells (EC50 = 2.4 μM) and solubility in vehicle formulations almost 20-fold higher than 22. Evaluation of the pharmacokinetic profile of 3 in rat suggested that this inhibitor possessed properties potentially sufficient for an IV continuous infusion dosing paradigm.42
Although irreversible inhibition can provide an extended pharmacodynamic effect when protein resynthesis rate is slow compared to drug clearance, such extended effects for 3CLpro inhibition were not necessarily expected. The 3CLpro-mediated proteolysis of newly expressed viral polyproteins is essential to virus replication; however, this activity only occurs during a single step in the virus life cycle that closely follows each cell infection. Viral particles themselves are not reliant on 3CLpro activity nor are the remaining coronavirus life cycle steps. Each event of cell infection initiates newly synthesized 3CLpro. Because detailed kinetics of this process are not understood, reversible and irreversible inhibitors were investigated equally.