Alpha-ketoamides as broad-spectrum inhibitors of coronavirus and enterovirus replication

Linlin Zhang; Daizong Lin; Yuri Kusov; Yong Nian; Qingjun Ma; Jiang Wang; Albrecht von Brunn; Pieter Leyssen; Kristina Lanko; Johan Neyts; Adriaan de Wilde; Eric J. Snijder; Hong Liu; Rolf Hilgenfeld

When we introduced a fluoro substituent in the para position of the P2-benzyl group of our lead compound, 11a, we observed good activity against the enterovirus 3C pro s but complete inactivity against the coronavirus M pro s (see Table 1 , compound 11m). This is easily explained on the basis of the crystal structures: In the enterovirus 3C pro s, the fluorine can accept a hydrogen bond from Arg39 (ref. 30) , whereas in the coronavirus M pro s, there would be electrostatic repulsion from the main-chain carbonyls of residues 186 and 188. In agreement with this, rupintrivir (which has P2 = p-fluorobenzyl) is a good inhibitor of the enteroviral 3C pro s, 46 but not of the coronaviral main proteases, as we predicted earlier. 28 In this structure-based inhibitor optimization study, we achieved major improvements over our original lead compound, 11a, by systematically varying the size and the flexibility of the P2-substituent. The compound presenting so far the best compromise between the different requirements of the S2 pockets (SARS-CoV M pro : large and covered, HCoV-NL63 M pro : small and covered, and CVB3 3C pro : large and open) is 11u (P2 = cyclopentylmethyl), which has satisfactory broad-spectrum activity against all proteases tested. However, with regard to its antiviral activities in cell cultures, it is inferior to 11r (P2 = cyclohexylmethyl). The latter compound exhibits very good inhibitory activity against the SARS-CoV M pro as well as the enterovirus 3C pro , and its performance in the SARS-CoV and enterovirus replicons is convincing. Being in the low micromolar range (EV-A71, CVB3), the data for the antiviral activity in cell culture for 11r correlate well with the inhibitory power of the compound against the recombinant proteases as well as in the replicon-based assays. This is not true, though, for the surprisingly good in-cellulo activity of 11r against HCoV 229E in Huh7 cells. Also, the correlation does not seem to hold for LLC-MK2 and CaCo2 cells. We tested the antiviral activity of many of our compounds against HCoV NL63 in these two cell types and found that all of them had low-or submicromolar EC50 values against this virus in LLC-MK2 cells but were largely inactive in CaCo2 cells (not shown). Furthermore, 11r and all other compounds that we synthesized are inactive (EC50 > 87 µM) against CVB3 in Vero cells (not shown), but exhibit good to excellent activities against the same virus in Huh-T7 cells. We have previously observed similar poor antiviral activities in Vero cells not only for aketoamides, but also for Michael acceptors (Zhu et al., unpublished work) . A similar cell-type dependence is seen for the antiviral activity of 11r against MERS-CoV and SARS-CoV. Whereas the inhibitor exhibits excellent activity against MERS-CoV when Huh7 cells are the host cells (400 pM), the inhibitory activity is weaker by a factor of up to 12,500 when Vero cells are used (EC50 = 5 µM). On the other hand, 11r exhibits excellent anti-MERS-CoV activity in human Calu3 lung cells, i.e. in the primary target cells where the compound will have to act in a therapeutic setting (A. Kupke, personal communication). As we tested antiviral activity against SARS-CoV exclusively in Vero cells, the EC50 values determined for our compounds against this virus are in the one-digit micromolar range or higher; the best is again compound 11r with EC50 = 2.1 µM. Interestingly, the relatively weaker activity (or even inactivity) of our inhibitors against