Proteomics-Guided Design of Ester Prodrugs of Antiviral Nucleoside Analogs

To date, three nucleoside phosphoramidate (ProTide) prodrugs have been approved by the FDA: tenofovir alafenamide (TAF), sofosbuvir (SBV), and remdesivir (RDV). ProTides mainly use carboxylesterase 1 (CES1) and cathepsin A (CatA) to cleave their ester moiety to initial intracellular bioactivation. In human lungs, CatA is the primary enzyme for activating ProTides. We found that TAF had the highest susceptibility to CatA and the most rapid hydrolysis rate in lung S9 among these ProTides. Moreover, TAF generated the highest levels of active metabolite in multiple lung cell models and exhibited the highest cell permeability among TAF, RDV, and SBV. Therefore, we hypothesized that high susceptibility to CatA and increased cell permeability would improve ProTides’ bioactivation in the lung. RDV has poor plasma stability and a high susceptivity to CES1, an enzyme highly expressed in the liver. These properties may limit its accumulation and activation in the lung. Therefore, we tested the other two RDV derivates: RDV methylpropyl ester analog (MeRDV) and RDV isopropyl ester analog (IsoRDV). MeRDV and IsoRDV had dramatically greater stability than RDV (P < 0.01), with a 4-5-fold longer half-life in ex vivo human plasma. Moreover, MeRDV was more susceptive to CatA (2561.2 ± 38.0 vs. 1786.5 ± 179.2 pmol/min/µg protein) (P < 0.01) and less susceptive to CES1 than RDV (11.3 ± 8.0 vs. 31.7 ± 1.1 pmol/min/µg protein) (P < 0.05). Accordingly, MeRDV showed a higher hydrolysis rate in human lung tissue homogenates than RDV (P < 0.05). In contrast, RDV generated approximately 1-to 2-fold more active metabolite remdesivir-triphosphate (RDV-TP) than MeRDV in multiple human lung cell models, likely due to RDV exhibiting a greater cell permeability than MeRDV. Nevertheless, MeRDV generated significantly more RDV-TP than Nuc, with the AUC0-24h of RDV-TP being nearly 5-, 10-, and 28-fold larger than Nuc in A549, BEAS-2B, and iPSC-AT2 cells, respectively (P < 0.01). We dosed these drugs on a molar equivalent dose basis via oral gavage or intravascular (IV) bolus injection in Ces1c-/- mice. After oral administration, prodrugs were barely detected in the mouse plasma, with a large amount of downstream metabolite Nuc present instead, suggesting an extensive first-pass effect. After IV administration, RDV plasma concentration declined most quickly, followed by MeRDV and IsoRDV, whereas Nuc plasma concentration decreased most slowly. Overall, the estimated systematic exposure (AUC0-t) of RDV (1570.2 ± 947.82 h⋅nM) and MeRDV(2370.4 ± 1152 h⋅nM) were significantly lower (P < 0.05) than that of IsoRDV (4941.55 ± 626.41 h⋅nM) and Nuc (25123.16 ± 285.38 h⋅nM). Importantly, RDV-TP generated from RDV in the mouse lung (4764.2 ± 753.1 pmol•h/g tissue) was approximately 44% higher than MeRDV (3322.5 ± 331.5 pmol•h/g tissue) (P < 0.05) and 2-fold higher than Nuc (1668.4 ± 276.9 nM•h) (P < 0.001). In addition, RDV and MeRDV generated nearly 2- and 1-fold, respectively, greater exposure of RDV-TP in the lung than IsoRDV (P < 0.05). However, the total RDV-TP and RDV-MP exposure in the liver and kidney was approximately 8-18-fold higher than in the lung after IV administration of MeRDV and RDV. This preferential hepatic and renal distribution could be attributed to the higher expression of Ces1 and Ctsa in the liver and higher expression of Ctsa in the kidney than in other tissues. Further efforts are warranted to optimize the distribution of antiviral active metabolite in the lung.