Discussion
First, our findings show that similar intracellular concentrations of remdesivir relative to dATP can effectively inhibit SARS-CoV2 replication. Indeed, Gordon et al. reported high selectivity of the antiviral over incorporation of dATP using purified active RdRp-CoV2 [12]. Further, remdesivir has been reported to inhibit SARS-CoV-2 in lung cells at submicromolar concentrations in vitro [22]. Moreover, similar binding site for remdesivir in the SARS-CoV-2 polymerase have been reported recently by two different approaches, based on a 3D-structural alignment using other virus polymerases [23] and a constrained minimization and conformational search on the RdRp-CoV2 active site [22,23]. In fact, our model provided the specific molecular interactions supporting the binding site (see supplementary material S3 for detailed description). The cyano group in remdesivir, a group not present in the other ligands, allows additional H-bond interactions with the nascent RNA chain, likely being the cause for the strongest ΔGbinding. In sum, our findings on the remdesivir-SARS-CoV2 interaction are robustly aligned with recent studies and validates our approach as a reliable framework to evaluate nucleic acids analogues against SARS-CoV2. We conclude that the partial clinical efficacy observed for remdesivir is not the result of molecular interactions with the RdRp-CoV2, given its strong binding affinity and drug target stability. Alternatively, remdesivir effectiveness might rely on unstudied pharmacokinetic properties and/or time to treatment relative to viral replication.
Regarding tenofovir, our approach is the first to include template-nascent RNA in the ensemble and further improves previous work [24] by using cryo-EM structures instead of homology and by more extensive sampling of the ligand poses using volunteer distributed computations. Consistent with our findings, tenofovir-diphosphate has been shown to permanently terminate polymerase extension of nascent RNA when using recombinant RdRp-CoV2 [25]. However, infusion of tenofovir in Vero cell cultures did not inhibit replication of SARS-CoV-2 [26,27], while the use of 3-90 μM of its prodrug TDF yielded a 15-fold reduction of viral genome release [26] . Further, the use of TDF/FTC for treating SARS-CoV-2 infected ferrets led to better clinical scores and lower virus titers in nasal washes compared to a placebo [28]. Worth noting, the prodrug TDF, formulated to increase tenofovir limited bioavailability [29], is known to diffuse passively across cellular membranes [30,31] and further activate intracellularly, as opposed to tenofovir which requires active transportation for intake before activation [32,33]. Indeed, higher levels of active metabolite after exposure to TDF versus tenofovir has been consistently reported in several cell types [34–36]. In contrast, the prodrug TAF was formulated to reduce drug-adverse events observed for TDF (which distributes body-wide) by being highly HIV-target-cell specific. TAF is well known to selectively activate and present preferential distribution in lymphatic tissues [37].
Our findings support a suboptimal tenofovir-RdRp-CoV2 interaction compared to remdesivir, predicting that the ensemble is more sensitive to the triphosphate form intracellular concentration. Thus, matching sufficient intracellular availability of tenofovir-diphostate with SARS-CoV2 tropism [37,38](such as in the respiratory tract [39]) is essential in drug-driven containment of viral replication, mirroring HIV prophylaxis [35]. Of note, viral replication can occur in different tissues at different clinical stages [40]. Thus, tenofovir efficacy might highly depend on when it is given relative to the within-host SARS-CoV-2 distribution. Together with the available pharmacokinetic/pharmacodynamic evidence, our findings support that TDF, among tenofovir-based compounds, maximizes efficacy at safe clinical dosage especially when taken before or close after exposure, because of high cellular permeation, effective metabolite activation, and low-selective distribution at viral targeted tissues, laying out a plausible molecular interpretation of the apparent inconsistency of tenofovir-based compounds against SARS-CoV-2 [11]. Finally, our findings support further evaluation of remdesivir as treatment against COVID-19 but also for the prodrug TDF [10] (available as a generic worldwide), in particular as prophylaxis, while underscoring the need for understanding the intracellular availability of the drugs in SARS-CoV2 targeted tissues.