Differential Bioactivation Profiles of Different GS-441524 Prodrugs in Cell and Mouse Models: ProTide Prodrugs with High Cell Permeability and Susceptibility to CathepsinA Are More Efficient in Delivering Antiviral Active Metabolites to the Lung

Abstract

We assessed factors determining the tissue-specific bioactivation of ProTide prodrugs by comparing the disposition and activation of remdesivir (RDV), its methylpropyl and isopropyl ester analogs (MeRDV and IsoRDV), the oral prodrug GS-621763, and the parent nucleotide GS-441524 (Nuc). RDV and MeRDV yielded more active metabolite remdesivir-triphosphate (RDV-TP) than IsoRDV, GS-621763, and Nuc in human lung cell models due to superior cell permeability and higher susceptivity to CathepsinA. Intravenous administration in mice showed that RDV and MeRDV delivered significantly more RDV-TP to the lung than other compounds. Nevertheless, all four ester prodrugs exhibited very low oral bioavailability (< 2%), with Nuc being the predominant metabolite in blood. In conclusion, ProTides, such as RDV and MeRDV, are more efficient in delivering active metabolites to the lung than Nuc, driven by high cell permeability and susceptivity to CathepsinA. Optimizing ProTides’ ester structures is an effective strategy to enhance prodrug activation in the lung.

Introduction

Nucleoside analogs are an essential class of antivirals. All nucleoside analogs exert their antiviral activity through their triphosphate nucleoside metabolites (TP-Nuc), which are formed in the intracellular space. Nucleoside analogs usually have two limitations in generating TP-Nuc in cells: limited cell membrane permeability and slow first phosphorylation^1–2. When applying nucleoside analogs for pulmonary infections like COVID-19, how to efficiently deliver antiviral active metabolites into the lung is essential because stopping a pandemic requires an effective antiviral treatment as early as possible, usually within five days after infection³.

There are mainly two prodrug design strategies to boost the cell membrane permeability and bioavailability of nucleoside analogs: one is covering the hydrophilic groups like hydroxyl groups (-OH) on the ribose sugar moiety with ester moieties, which will improve the lipophilicity^4–5. An example employing this strategy is GS-621763, which covers the three −OH with isopropyl esters^6–7. In this case, the parent nucleoside analog GS-441524 (Nuc) itself is the major compound present in the systemic circulation after oral administration because the liver has abundant esterases that can cleave ester moieties and thus convert an ester prodrug back to Nuc^{4–5, 7}. Another strategy is using monophosphate nucleoside (MP-Nuc) as the start point and adding ester and other hydrophobic moieties to cover the −OH on the phosphate group, such as the nucleoside phosphoramidate (ProTide) prodrug. ProTides can load MP-Nuc into cells after ester hydrolysis by intracellular esterases, bypassing the slow and rate-limiting first phosphorylation step that is needed for Nuc. This work will investigate which prodrug form is more favorable for treating pulmonary infections and provide evidence for optimizing the bioactivation of other antiviral nucleoside analogs in the lung.

An example of a ProTide prodrug is remdesivir (RDV), the first antiviral drug approved by the FDA for COVID-19 treatment. RDV loads monophosphate-RDV (RDV-MP) intracellularly after its ester hydrolyzed by carboxylesterase 1 (CES1) and cathepsin A (CTSA), and its P-N bond cleaved by human histidine triad nucleotide binding protein-1 (HINT1)^1–2 (Fig. 1). RDV showed potent antiviral activity against SARS-CoV-2 in cell lines and preclinical species^8–9, and retained potent antiviral activity against later on emerged SARS-CoV-2 variants (e.g., Delta, Lambda, Omicron)¹⁰. RDV showed significantly stronger anti-SARS-CoV-2 activity than its non-prodrug form, GS-441524 (Nuc), in various cell models¹.

Fig. 1.

Bioactivation pathway of RDV^1–2.

Despite these preclinical advantages, the results of RDV clinical trials in treating patients with COVID-19 varied. The clinical study ACTT-1 (NCT 04280705) showed superior effects of RDV than placebo in shortening the time to recovery, symptom improvement, and controlling mortality¹¹. However, several other clinical trials^12–14 failed to find statistically significant clinical benefits of RDV compared with the control group. The reasons why RDV’s potent anti-SARS-CoV-2 activity cannot be reflected in clinical outcomes remain obscure. Several scholars pointed out that the limited efficacy of RDV in patients could result from its premature hydrolysis in human plasma^15–17. RDV has a relatively poor plasma stability with a half-life of 69 min in human plasma ex vivo¹⁸. Previously, the premature hydrolysis of RDV in the human plasma was thought to reduce the plasma concentration of RDV significantly and generate less effective downstream metabolites, the alanine metabolite (GS-704277, Ala) and Nuc, with Nuc being the predominant metabolite detected in human plasma after intravenous (IV) infusion of RDV.¹⁹ Yan et al. assumed that RDV’s hydrolytic metabolite, Nuc, in the plasma drives RDV-TP formation in the lung rather than RDV¹⁷. However, it remains unclear to what degree the plasma stability issue affects the distribution of RDV and the formation of its antiviral active metabolite RDV-TP in the lung.

Two other approved ProTides, tenofovir alafenamide and sofosbuvir, have greater human plasma stability than RDV^20–22. Tenofovir alafenamide and sofosbuvir use isopropyl ester as their prodrug moiety, whereas RDV uses ethylbutyl ester. Thus, we hypothesized that modifying the ester structure of RDV may lower its susceptibility to plasma esterases and improve its plasma stability. Thereby, we evaluated two RDV derivatives with different ester moieties: methylpropyl and isopropyl, namely MeRDV and IsoRDV (Fig. 2). In this work, we compared the bioactivation of these different prodrugs across human cell models. Moreover, we compared their active metabolite distribution profiles in Ces1 knockout (Ces1c^−/−) mice. This work revealed which prodrug form could be more favorable for treating pulmonary infections and provided evidence for optimizing the bioactivation of other antiviral nucleoside analogs in the lung.

Fig. 2.

Chemical structures of RDV, MeRDV, IsoRDV, GS-441524, and GS-621763.

Results

Enzyme susceptibility, tissue homogenate hydrolysis profiles, and plasma stability

To evaluate the susceptibility of ProTides to CES1 and CatA, we performed an in vitro enzymatic activity assay based on the method reported in a previous study²³. After the prodrugs were incubated with recombinant human CatA (rhCatA) or CES1(rhCES1), the concentrations of prodrugs and metabolites were measured by LC-MS/MS. The enzyme activity was calculated by dividing the reduction of the prodrug by the enzyme concentration and incubation time. The activities of rhCatA and rhCES1 on RDV, MeRDV, and IsoRDV hydrolysis are displayed in Fig. 3A. The activity of rhCatA for MeRDV was significantly higher than that for RDV (2,561.2 ± 38.0 vs. 1,786.5 ± 179.2 pmol/min/μg protein, n = 3, P < 0.01, Fig. 3A left), while rhCES1 activity for MeRDV was lower than that for RDV although the difference is not statistically significant (P=0.058, Fig.3A right). The activities of both rhCatA and rhCES1 for IsoRDV were significantly lower than that for RDV (rhCatA: 236.7 ± 172.8 vs. 1,786.5 ± 179.2 pmol/min/μg protein, n = 3, P < 0.001, Fig. 3A left, rhCES1: 7.7 ± 1.4 vs. 15.5 ± 1.8 pmol/min/μg protein, n = 3, P < 0.01, Fig. 3 A right).

Fig. 3. Enzyme susceptibility, ex vivo human plasma stability, and tissue homogenate hydrolysis profiles.

(A) Activity of rhCatA and rhCES1 for hydrolysis of RDV, MeRDV, and IsoRDV. 20 μM prodrug was incubated with 0.5 ng/μL rhCatA for 10 min (left), 40 ng/μL rhCES1 for 20 min (right). Bars show mean values (± S.D.) from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 indicate statistically significant differences between groups by one-way ANOVA with post-hoc Tukey multiple comparison test.

(B) Stability of RDV, MeRDV, IsoRDV, and Nuc in human plasma. The remaining percentages of each drug after incubation for 24 hours (left), and the major metabolites generated following 4 hours of incubation (right). Bars show mean values (± S.D.) from three independent experiments.

(C) Hydrolysis profiles of prodrugs in human tissue homogenates. Prodrug hydrolysis in 1 mg/mL human intestine S9 fractions (HIS9), 0.5 mg/mL human liver S9 fractions (HLS9), 1 mg/mL human lung S9 fractions (HLungS9), and 0.5 mg/mL human kidney S9 fractions (HKS9). Bars show mean values (± S.D.) from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 indicate statistically significant differences from the RDV group by one-way ANOVA with post-hoc Tukey multiple comparison test.

The ex vivo human plasma stability of RDV, MeRDV, IsoRDV, and Nuc are shown in Fig. 3B. Nuc was stable in human plasma with only a 5% reduction after 24 hours of incubation with human plasma. MeRDV and IsoRDV are significantly more stable than RDV in human plasma. After incubation for 4 hours, approximately 70% of RDV was hydrolyzed, whereas MeRDV and IsoRDV were hydrolyzed by about 18% and 15 %, respectively. The major metabolite for these prodrugs in human plasma was Ala. After a 4-hour plasma incubation, RDV generated approximately 3- and 12-fold more Ala than MeRDV and IsoRDV, respectively. Minimal amounts of Nuc and almost no RDV-MP were detected in human plasma after the 4-hour incubation with these prodrugs.

The concentrations of prodrugs and product metabolites of three ProTides after incubation with homogenates (S9 fractions) of four human tissues (lung, liver, intestine, and kidney) were determined by an HPLC-MS/MS assay. The reduction rates of the prodrugs following incubation with the tissue S9 fractions are shown in Fig. 3C. As GS-621763 was designated for oral administration, we only tested its stability in human intestine and liver homogenates, in which GS-621763 could barely be detected after 15 min incubations. Among the tested three ProTides, MeRDV displayed higher hydrolysis rates than RDV in human lung S9 (HlungS9), human intestine S9 (HIS9), and human kidney S9 (HKS9) (P < 0.05). However, in human liver S9 (HLS9), RDV was hydrolyzed at a higher rate than MeRDV (P < 0.05). In all tested tissue S9 fractions, IsoRDV showed the slowest hydrolysis rates among the prodrugs. The metabolites generated after incubation with the human tissue S9 are shown in Fig. S1. Ala was the major metabolite of RDV, MeRDV, and IsoRDV after incubation with human tissue S9. However, Nuc was the major product metabolite of GS-621763 in HIS9 and HLS9. Overall, the metabolic rates of MeRDV were higher in HlungS9 and lower in HLS9 than RDV. These data suggest that a higher susceptibility to CatA could lead to a more efficient ester hydrolysis, the initial activation step, in human lungs.

Intracellular bioactivation and drug entry of prodrugs in various cell lines.

To evaluate the bioactivation of the four prodrugs in human cells, we measured the intracellular concentrations of RDV-TP in human lung carcinoma cells A549, A549 cells expressing ACE2 and TMPRSS2 (A549-ACE2-TMPRSS-2), Calu-3, Caco-2, Vero E6, normal human bronchial epithelial cells BEAS-2B, human liver carcinoma cells Huh-7, renal epithelial cells HEK293 at 6, 12, and 24 hours post incubation. We did not observe any measurable cytotoxicity of the prodrugs and Nuc at the tested concentrations after incubation with the cell lines (data not shown). The tested concentrations of RDV and Nuc in VeroE6 cells (10 μM), iPSC-AT2 (5 μM), and Caco-2 cells (5 μM) were at least five folds lower than the 50% cytotoxic concentration (CC₅₀) values of RDV (101, 32.7, and >100 μM, respectively) and Nuc (>100 μM) in these cell lines based on previously reported data²⁴.

The bioactivation of RDV, MeRDV, IsoRDV, Nuc, and GS-621763 in various cell lines is displayed in Fig. 4. RDV is the most efficient in generating active metabolite RDV-TP in all cell lines tested, followed by MeRDV. Exposure of RDV-TP within 24 hours following incubation (AUC_0–24h) and total drug entry within 6 hours in cell lines are summarized in Table S1 and S2, respectively. RDV-TP formation rates within 6 hours are summarized in Table S3. Compared with Nuc, RDV generated significantly more RDV-TP, with 2-fold higher in Calu-3 cells, 2.3-fold higher in Vero E6 cells, 11-fold higher in A549 cells, 19-fold higher in BEAS-2B cells, and 75-fold higher in iPSC-AT2 cells than Nuc. Moreover, RDV generated approximately 649-fold, 360-fold, and 62-fold more RDV-TP than Nuc in Caco-2, HEK293, and Huh-7 cells, respectively. The intracellular levels of RDV-TP generated from RDV gradually declined after 12 hours in A549, A549-ACE2-TMPRSS-2, BEAS-2B, iPSC-AT2, Caco-2, and HEK293 cells. Of note, in Huh-7 cells, the intracellular levels of RDV-TP generated by RDV and MeRDV declined significantly after 6 and 12 hours, respectively, with minimal RDV-TP being detected at 24 hours after treatment. Interestingly, a considerable amount (~1.3 μM) of Nuc was detected in the culture medium of Huh-7 cells after 24 hours of incubations with RDV and MeRDV, while almost no Nuc was detected in the cell-free culture medium (Fig. S4), suggesting that Huh-7 cells could convert RDV back to Nuc.

Fig. 4. RDV-TP formation by RDV, MeRDV, IsoRDV, Nuc, and GS-621763 in various cell lines.

Ten μM drugs were incubated in A549 (A), A549-ACE2-TMPRSS2 (B), BEAS-2B (C), Calu-3 (D), and Vero E6 (E) cells. 5 μM drugs were incubated in iPSC-AT2 (F), Caco-2 (G), HEK293 (H), and Huh-7 (I) cells. Data are shown as mean ± S.D. from three biological replications.

We also measured the intracellular concentrations of RDV-MP (Fig. S2), prodrugs, Nuc, and Ala (Fig. S3) at 6 hours post-treatment. The sums of the amount of these intracellular compounds represent the total drug entry within 6 hours (Fig. 5A). Of note, we were unable to measure RDV-DP because this intermediate metabolite is unstable, and its concentrations are low, only accounting for a minor portion of the total metabolites. Except for Calu-3 cells, RDV showed significantly higher drug entry in multiple cell lines than MeRDV, IsoRDV, Nuc, and GS-621763 (P < 0.05). Compared with Nuc, the total drug entry of RDV was approximately 5-fold higher in A549 and A549-ACE2-TMPRSS2 cells, 2.5-fold higher in BEAS-2B cells, 2-fold higher in iPSC-AT2 cells, and 2.5- to 3.5-fold higher in Caco-2, HEK293, and Huh-7 cells (Fig. 5A, Table S2). The total drug entry and the intracellular AUC_0–24h of the active metabolite of RDV were approximately 1- to 2-fold higher than those of MeRDV in various cell lines (Fig. 4, Table S1). In addition, MeRDV had significantly higher drug entry and more intracellular active metabolite formation than Nuc in multiple cell lines (P < 0.05) except for Vero E6 and Calu-3 cells.

Fig. 5. Intracellular bioactivation and drug uptake of prodrugs and Nuc in cell lines.

(A) Cell-dependent uptake and bioactivation of prodrugs. The total drug cell entry amount of each drug within six hours in various cell lines (up); The intracellular accumulation of RDV-TP within 24 hours in various cell lines (down). Data are shown as mean ± S.D. from three biological replications. *P < 0.05, **P < 0.01, ***P < 0.001 indicate statistically significant differences between groups by one-way ANOVA with post-hoc Tukey multiple comparison test.

(B) Association between drug entry and intracellular exposure of RDV-TP for RDV, MeRDV, IsoRDV, and Nuc.

(C) Effects of CES1 inhibitor BNPP and CatA inhibitor telaprevir on bioactivation of ProTides. Ten μM RDV (left) or MeRDV (middle) were co-incubated with various concentrations of BNPP or telaprevir in BEAS-2B cells for 24 hours; 10 μM IsoRDV (right) was co-incubated with various concentrations of BNPP or telaprevir in BEAS-2B cells for 48 hours. Data are shown as mean ± S.D. from three biological replications. ***P < 0.001 indicates a statistically significant difference from the control group by one-way ANOVA with post-hoc Tukey multiple comparison tests.

Association between cell drug entry and intracellular bioactivation

To understand the mechanisms underlying differences in activation rates of the ProTides and Nuc, we investigated the relationship between drug entry within six hours and AUC_0–24h of RDV-TP across different cell lines for RDV, MeRDV, IsoRDV, and Nuc. As displayed in Fig. 5A, RDV, MeRDV, and IsoRDV showed a significant cell-dependent pattern in their activations and cell entry: HEK293 and Huh-7 took up more ProTides and generated more RDV-TP than other cell lines. Calu-3 and Vero E6 accumulated significantly less ProTides and generated less RDV-TP than other cell lines. The cell entry rates of RDV, MeRDV, and IsoRDV showed a strong correlation with their intracellular accumulation of RDV-TP, with the R² values being greater than 0.85 (Fig. 5B). In contrast, Nuc and GS-621763 did not correlate with their drug uptake and intracellular accumulation of RDV-TP (Fig. 5B). These results indicated that the higher cell permeability of RDV compared to MeRDV, IsoRDV, and Nuc could be a major contributor to its higher generation of RDV-TP in various cell lines.

Contributions of CES1 and CatA to ProTide activation in BEAS-2B cells

To evaluate the impact of CES1 and CatA activity in the bioactivation of ProTides in BEAS-2B, the intracellular concentrations of RDV-TP were measured following coincubation of each ProTide with various concentrations of the CES1 inhibitor BNPP or the CatA inhibitor telaprevir. As shown in Fig. 5C, BNPP, even at 20 μM, the highest inhibitor concentration tested, did not significantly alter the formation of RDV-TP. In contrast, telaprevir, at 5 μM, the lowest inhibitor concentration tested, reduced the RDV-TP formation by approximately 80%, 70%, and 100% for RDV, MeRDV, and IsoRDV, respectively, indicating that CatA is the major enzyme responsible for bioactivation of ProTides in BEAS-2B cells.

Antiviral activity of RDV and its analogs against SARS-CoV-2

To determine the antiviral efficacy of RDV, MeRDV, IsoRDV, and Nuc against SARS-CoV-2, we infected lung-derived A549 cells expressing the viral receptor ACE2 and protease TMPRSS2 with two strains of SARS-CoV-2 (MOI 1.0), the early strain WA1 and Omicron variant BA.1, in the presence of increasing concentrations of compounds (Fig. 6). DMSO was used as vehicle control. Cell viability remained around 80% or above at the highest concentration used (i.e., 25 μM) (Fig. S5). After 24 hours, infectious virus titers were determined by TCID₅₀. While all analogs had antiviral efficacy against SARS-CoV-2 in a dose-dependent manner, RDV displayed the highest antiviral efficacy against WA1 (EC₅₀ = 0.06 μM), followed by MeRDV (EC₅₀ = 0.15 μM), Nuc (EC₅₀ = 2.2 μM) and IsoRDV (EC₅₀ = 5.2 μM). Similarly, RDV displayed the highest antiviral efficacy against Omicron variant BA.1 (EC₅₀ = 0.0003 μM), followed by MeRDV (EC₅₀ = 0.03 μM), Nuc (EC₅₀ = 0.49 μM) and IsoRDV (EC₅₀ = 0.54 μM). These data are consistent with the more efficient delivery of the active metabolite RDV-TP into lung cell lines from RDV and MeRDV compared to Nuc and IsoRDV.

Fig. 6. Antiviral activity of RDV, MeRDV, IsoRDV, and Nuc against SARS-CoV-2 in A549-ACE2-TMPRSS2 cells.

Data are shown as mean ± S.D. from two biological replicates (i.e., experiments 1 and 2) with four technical replicates each. Viral titers were determined by 50% tissue culture infectious dose (TCID₅₀) assay in A549-ACE2-TMPRSS2 cells.

Mouse Pharmacokinetics

We compared the Pharmacokinetic (PK) profiles of different prodrugs and Nuc after the administration of the compounds at a molar equivalent dose basis in the plasma of Ces1c knockout (Ces1c^−/−) mice. Although several studies reported lung PK profiles of RDV-TP in mice, they used wild-type mice, which have higher expressions of plasma Ces1 than humans^25–26. RDV plasma stability is poorer in mice (half-life, <5 min)²⁷ than in humans (half-life, 69 min)²⁸. Studies based on normal mouse models may underestimate the tissue exposure of RDV due to the extensive Ces1-medicated RDV hydrolysis in mouse plasma. RDV plasma stability was significantly improved (half-life, ~25 min) in Ces1c^−/− mice²⁷. Thus, Ces1c^−/− mice is more appropriate than wild-type mice for the studies of RDV and its ProTide derivatives that are Ces1 substrates. The plasma concentration-time profiles of prodrugs and their major metabolites, Nuc and Ala, are shown in Fig. 7. The major PK parameters for oral and IV administration are displayed in Table 1 and Table 2, respectively. After oral administration with 54.4 μmole/kg of RDV, MeRDV, IsoRDV, or GS-621763, the prodrug forms were barely detected in the plasma, suggesting an extensive first-pass effect after oral administration. The estimated oral bioavailability of RDV, MeRDV, and IsoRDV was less than 2%. Nuc was the major metabolite for all the prodrugs tested. The AUC_0–12h of Nuc was comparable between these prodrugs following oral administration. Ala was the minor metabolite in the plasma for oral ProTides (RDV, MeRDV, and IsoRDV).

Fig. 7. Plasma PK of prodrugs and Nuc in Ces1c^−/− mice.

Plasma PK of the prodrugs (A), Nuc (B), and Ala (C) after oral administration of the prodrugs. Plasma PK of the unchanged prodrugs and Nuc (D) after IV injection of the prodrugs and Nuc. Plasma PK of Nuc (E) and Ala (F) after IV injections of the prodrugs. The data for each time point are shown as mean ± S.D. from three to four mice.

Table 1.

Mean plasma PK parameters of the major metabolite Nuc after oral administration of RDV, MeRDV, and IsoRDV in Ces1c−/− mice.

Parameters	Nuc from RDV	Nuc from MeRDV	Nuc from IsoRDV	Nuc from GS-621763
AUC0-∞ (h⋅nM)	9,528.3 ± 2,263.2	11,600.5 ± 1,612.5	14,878.9 ± 9,899.4	6,355 ± 1,075.2
AUC0-t (h⋅nM)	8,526.0 ± 1,666.0	11,549.1 ± 1,633.5	9,275.6 ± 2,312.6	6,244.9 ± 1,100.7
CL/F (L⋅h−1)	0.15 ± 0.04	0.12 ± 0.02	0.12 ± 0.06	0.22 ± 0.04
Cmax (nM)	1,534.7 ± 276.5	2,474.4 ± 366.4	1,651.6 ± 152.1	4,084.1 ± 763.4
t1/2 (h)	5.97 ± 2.98	3.84 ± 1.81	7.98 ± 2.41	4.38 ± 2.23

Data are shown as mean ± S.D. from three to four mice.

Table 2.

Mean plasma PK parameters after IV administration (bolus) of RDV, MeRDV, IsoRDV, and Nuc in Ces1c−/− mice

Parameters	RDV	MeRDV	IsoRDV	Nuc
AUC0-∞ (h⋅nM)	1,576.2 ± 948	2,382.6 ± 1,154.5	4,950.3 ± 628.2	26,312.3 ± 839.4
AUC0-t (h⋅nM)	1,570.2 ± 947.8	2,370.4 ± 1,152.0	4,941.6 ± 626.4	25,123.2 ± 285.4
CL/F (L⋅h−1)	1.06 ± 0.44	0.79 ± 0.62	0.28 ± 0.03	0.052 ± 0.002
Cmax (nM)	1,371.7 ± 515.1	2,268.9 ± 949.5	5,888.3 ± 587.8	18,132.7 ± 2,656.8
t1/2 (h)	1.25 ± 0.39	0.92 ± 0.22	1.04 ± 0.33	6.34 ± 4.01
Vd (L)	1.91 ± 0.94	0.93 ± 0.48	0.42 ± 0.17	0.47 ± 0.29
	Nuc from RDV	Nuc from MeRDV	Nuc from IsoRDV	-
AUC0-t (h⋅nM)	10,345 ± 1,469.1	7,868.6 ± 2,237.1	8,561.4 ± 3,657.8	-
Cmax (nM)	1,859.5 ± 523.4	2,183.7 ± 947.4	1,810.6 ± 776.0	-
	Ala from RDV	Ala from MeRDV	Ala from IsoRDV	-
AUC0-t (h⋅nM)	5,059 ± 1,151.1	5,329.7 ± 1,920.3	4,237.6 ± 2,453.1	-
Cmax (nM)	8,632.1 ± 1,809.2	8,627.9 ± 3,318.5	8,262.1 ± 4,594.9	-

Data are shown as mean ± S.D. from three to four mice for each time point.

After IV administration, RDV plasma concentration declined at a more rapid rate than MeRDV, followed by IsoRDV. At the first time point (15 min after IV injection), the plasma concentrations of RDV (1,371.7 ± 515.0 nM) and MeRDV (2,268.9 ± 949.5 nM) were significantly lower than those of IsoRDV (5,888.3 ± 587.7 nM) and Nuc (18,132.7 ± 2,656.8 nM). RDV and MeRDV were barely detectable in the plasma 4 hours after IV administration. RDV (1,570.2 ± 947.8 h⋅nM) and MeRDV(2,370.4 ± 1,152 h⋅nM) had significantly lower AUC_0-t(8h) in mouse plasma (P < 0.01) compared with IsoRDV (4,941.6 ± 626.4 h⋅nM) and Nuc (25,123.2 ± 285.4 h⋅nM). The plasma C_max of RDV and MeRDV were significantly lower than IsoRDV’s (5,888.3 ± 587.8 nM, P < 0.05) and Nuc’s (18,132.7 ± 2,656.8 nM, P < 0.01). RDV had a significantly larger apparent distribution volume V_d (1.91 ± 0.94 L) than IsoRDV (0.42 ± 0.17 L, P < 0.05) and Nuc (0.47 ± 0.29 L, P < 0.05), but was not significantly different from MeRDV (0.93 ± 0.48 L, P > 0.05). No significant difference (P > 0.05) was observed in the plasma AUC and C_max for the downstream metabolites Nuc and Ala among RDV, MeRDV, and IsoRDV after IV administration.

Tissue distribution of RDV-MP and RDV-TP after IV administration

We determined the levels of RDV-MP and RDV-TP in six tissues of mice within 8 hours after IV administration of 54.4 μmole/kg of RDV, MeRDV, IsoRDV, and Nuc (Fig. 8 and Fig. 9). In the mouse lung, RDV (AUC_0–8h: 4,764.2 ± 753.1 pmol•h/g tissue) and MeRDV (3,322.5 ± 331.5 pmol•h/g tissue) generated significantly more RDV-TP than Nuc (1,668.4 ± 276.9 nM•h, P < 0.001). The EC₅₀ (358 pmol/g tissue) and EC₉₀ (521 pmol/g tissue) of RDV-TP in mouse lungs were estimated based on the anti-SARS-CoV-2 potency reported in cell models⁸ and the number of cells in each gram of mouse lung²⁹. It is of note that the RDV-TP concentrations measured in the mouse lung after IV administration of RDV were below the estimated EC₉₀ and EC₅₀ levels after 4 and 8 hours, respectively. The RDV-TP levels in the mouse lung after IV administration of Nuc were constantly below EC₅₀. No significant difference was detected between RDV and MeRDV in the AUC_0–8h of RDV-TP and RDV-MP in the liver, kidney, small intestine, and brain. RDV and MeRDV generated significantly more RDV-TP and RDV-MP than Nuc in all the tissues tested. Of note, the amounts of RDV-TP in mouse brain were below the detectable level for all drugs in this study.

Fig. 8. Tissue PK of RDV-MP and RDV-TP after IV administration of RDV, MeRDV, IsoRDV, and Nuc in Ces1c−/− mice.

Data for each time point are shown as mean ± S.D. from three mice. The EC₅₀ (393 pmol/g tissue) and EC₉₀ (679 pmol/g tissue) of RDV-TP in mouse lungs were estimated based on the anti-SARS-CoV-2 potency reported in cell models⁸ and the number of cells in each gram of mouse lung²⁹.

Fig. 9. Tissue exposure of RDV-MP and RDV-TP after IV administration of RDV, MeRDV, and Nuc in Ces1c−/− mice.

(A) RDV-MP and RDV-TP accumulation (AUC_0–8h) in lung, liver, kidney, small intestine, testis, and brain; (B) tissue allocation profiles of RDV-TP alone and RDV-TP plus RDV-MP. Data are shown as mean ± S.D. from three mice.

RDV and MeRDV showed a preferential distribution of RDV-MP and RDV-TP in liver and kidney (Fig. 9): For RDV, the mouse liver and kidney had approximately 10- and 8-fold higher accumulation of RDV-MP + RDV-TP than the lung, respectively; for MeRDV, the mouse liver and kidney had approximately 18- and 16-fold higher accumulation of RDV-MP + RDV-TP than the lung, respectively; for Nuc, the mouse liver and kidney had nearly 13-, and 1.4-fold higher accumulation of RDV-MP + RDV-TP than the lung. These findings suggest a significantly stronger ester hydrolysis capability (mediated by Ces1 and Ctsa) and P-N bond cleavage capability (mediated by Hint1) in the liver and kidney.

Discussion and Conclusions

Unlike CES1 that is specifically highly expressed in the liver, CatA is widely expressed in all types of human lung and bronchus cells at a considerable level. A western blot assay revealed that CatA protein expression was significantly higher than CES1 protein expression (3.3 vs. 0.1 ng/μg total protein) in NHBE cells³⁰. In line with previous evidence³⁰, our inhibitor study demonstrated that CatA is the major enzyme activating RDV, MeRDV, and IsoRDV in BEAS-2B cells (Fig. 5C). Therefore, we hypothesized that a higher susceptibility to CatA would lead to more efficient bioactivation in human lung cells for a ProTide prodrug. To test this hypothesis, in this study, we included two new ProTides, MeRDV and IsoRDV, that replaced the ester moiety of RDV with methylpropyl ester and isopropyl ester, respectively. This change altered the enzymatic susceptibility/selectivity, with MeRDV being more susceptible to CatA and less susceptible to CES1 than RDV and IsoRDV being significantly less susceptible to both CatA and CES1. MeRDV has increased selectivity for CatA over CES1 by threefold when compared to RDV (Fig. 3A). Accordingly, MeRDV showed an approximately 2-fold higher hydrolysis rate in human lung S9 than RDV (Fig. 3C). However, RDV generated at least 2-fold more active metabolites than MeRDV in human lung cell lines, which warrants further investigation to determine whether cell permeability is a key factor. Of note, the cell permeability of RDV and Nuc has not been fully evaluated in previous studies because the intracellular concentrations of prodrug and other intermediate metabolites Ala and Nuc were not determined^31–32. In this study, we simultaneously measured the intracellular levels of prodrugs, Nuc, Ala, RDV-MP, and RDV-TP, enabling estimations of the total amount of drug entry into the cells. We found that the cell permeability of RDV was approximately 1- to 2-fold greater than MeRDV in various cell lines (Table S2). The cell permeability of RDV, MeRDV, or IsoRDV was highly correlated with the intracellular levels of the RDV-TP (Fig. 5B), suggesting that cell permeability is a key factor determining the intracellular generation of active metabolite for a ProTide.

Differential expression of ProTide activating enzymes and transporters may account for the cell-dependent bioactivation. We analyzed the expression of enzymes and transporters involved in the disposition and activation of ProTide prodrugs in the cell lines utilized in our in vitro studies and compared to different types of respiratory tissue cells using data retrieved from the Human Protein Atlas (HPA) database (Fig. S6). A549 cells closely resemble the expression levels of CES1, CatA, HINT1, and P-glycoprotein of AT2 cells and respiratory ciliated cells. Calu-3 and Vero E6 had significantly lower expression of CES1 and HINT1 but higher expression of P-glycoprotein than A549 cells. In an absolute quantitative proteomics study³³, Calu-3 had a significantly higher P-glycoprotein expression(0.44 pmol/mg protein) than A549 and BEAS-2B cells (not detectable). In addition, Vero E6 cells expressed one-fold higher expression of ABCC4 than A549 cells. These findings indicate that VeroE6 and Calu-3 cells may underestimate the bioactivation efficiency of ProTide prodrugs due to their higher expression of drug efflux transporters P-glycoprotein and MRP4.

Tissue expression patterns of activating enzymes in mouse may explain the preferential bioactivation of ProTides in the liver and kidney (Fig.9). We analyzed the protein expressions of relevant enzymes and transporters using data retrieved from the ProteomicsDB (Fig. S7). Mouse liver showed the highest expression of Ces1 among the tissues examined, with more than 10-fold higher than the lung. Mouse kidney and liver exhibited approximately 5- and 2-fold more Ctsa than the lung, respectively. Similarly, the human liver and kidney showed the highest CES1 and CatA expression levels, respectively, suggesting that RDV and MeRDV may also have hepatic and renal preferential bioactivation in humans.

Oral prodrug GS-621763 mainly delivers Nuc to the systemic circulation. Overall, Nuc is significantly less efficient in delivering antiviral active metabolites to the lung because of its lower cell permeability and less intracellular bioactivation than RDV. This study’s findings align with data from the Gilead group that Nuc constantly yielded significantly lower levels of RDV-TP (at least 40-fold lower) than RDV within 48 hours after drug treatment in human bronchial epithelial cells³⁴. Moreover, our mouse lung PK data is in line with Gilead’s data in African green monkeys, which showed that the concentrations of RDV-TP generated from Nuc were approximately 8-fold lower than RDV in the lung 24 hours after IV administration³⁴. The PK profiles of IV injection of Nuc in mice demonstrated that Nuc has a lower distribution rate to tissues and is less efficient in generating RDV-TP than RDV and MeRDV in the lung, even though Nuc has a larger AUC in the plasma. Nuc seemed trapped in the systemic circulation, while RDV and MeRDV can efficiently enter different tissues and cells. IsoRDV showed a similar behavior to Nuc in that it maintained at higher plasma concentrations but generated less amount of RDV-MP and RDV-TP in the lung and other various tissues than RDV and MeRDV. Siegel et al. described the synthesis of RDV (compound 4b) and its analogs, including IsoRDV (compound 4i), and its structure-activity relationship (SAR)²². The performances of IsoRDV observed in our cell study are consistent with the study by Siegel et al.²², in which IsoRDV was more stable in human plasma than RDV (1561 min vs. 69 min), but its antiviral activity was significantly weaker to Ebola virus in various cell lines. Meanwhile, the logD values of Nuc (0.3) and IsoRDV (1.1) are dramatically lower than RDV (2.1)²², indicating that Nuc and IsoRDV had a lower cell membrane permeability than RDV. Our results and the data from Siegel et al. demonstrated that the selection of ester moiety for ProTide prodrugs will determine plasma stability, cell membrane permeability, and intracellular activity. All these data showed that the formation of RDV-MP and RDV-TP in the lung is driven by the cell permeability as well as susceptivity to CatA.

Our animal PK data showed that RDV-TP (Fig. 8) could reach its C_max in the lung within 1 hour and a large portion (about 90%) of plasma RDV disappeared within 30 minutes after IV administration. Considering that the half-life of RDV in a human plasma ex vivo incubation experiment was 69 min²⁸, the rapidly reduced RDV plasma concentrations is very likely caused by its rapid distribution into tissues. Although Nuc was effective against SARS-CoV-2 in monkeys and mice^34–35, Nuc required at least a 10-fold higher dosage than RDV. Collectively, our findings support the use of RDV via IV administration over Nuc due to RDV’s quick tissue distribution and significantly more efficient bioactivation in the lung than Nuc. Moreover, it is favorable for RDV to maintain its intact prodrug form when reaching human lung cells. Therefore, an inhalation formulation could further improve the pulmonary load of RDV and its active metabolite RDV-TP while reducing its distribution in the liver and kidney^{25, 36}.

Our animal PK data suggest that RDV is rapidly distributed in tissues and quick-acting. RDV-TP in the mouse lung increased quickly within one hour and then gradually declined over an eight-hour period after IV bolus injection of RDV (Fig. 8). The concentrations of RDV-TP in the mouse lung may be below the EC₉₀ and EC₅₀ at 4 and 8 hours, respectively, suggesting that the dosage utilized in the mouse study might be inadequate to provide sustainable RDV-TP concentration in the lung during the entire treatment interval. We concur that it could be difficult to correlate the RDV-TP level in gross lung tissue to the in vivo antiviral activity since this concentration number may not reflect its concentration in particular cells infected by the viruses. The RDV dosage in this animal PK study was scaled from the human loading dose on the first day of treatment. The current RDV treatment regimen for COVID-19 patients is IV infusion over 0.5 to 2 hours with 200 mg RDV on Day one, followed by 100 mg once per day³⁷. Thus, whether the pulmonary RDV-TP concentration in patients could remain above its effective level during the interval between two doses (i.e., 24 hours) may warrant further investigations. Given the rapid tissue distribution and intracellular activation of RDV, we assumed that a slow IV infusion rate and a more frequent dosing regimen may help maintain the pulmonary concentration of RDV-TP above the effective level throughout the entire period of treatment. In vivo antiviral study in Ces1c−/− mice showed that 25 mg/kg RDV twice daily significantly reduced the lung virus titers on 2 days post infection while 50 mg/kg RDV single dose daily could not²⁷.

In conclusion, the ProTide prodrugs RDV and MeRDV significantly improved efficiency in delivering the active metabolite to the lung compared to the parent compound (i.e., Nuc). The improved cell permeability is the major factor enabling RDV to deliver more RDV-TP to the lung than other prodrugs and Nuc. CatA is the primary activating enzyme for ProTides in the lung and airway. High susceptibility to CatA is another factor that will boost a ProTide to generate active metabolite in the lung. Data from this study will provide evidence for optimizing the prodrug structure of other nucleoside antivirals for treating pulmonary infections like COVID-19.

Experimental Section

Materials

All compounds are >95% pure. RDV (≥ 98%), MeRDV (≥ 98%), IsoRDV (≥ 98%), GS-621763 (≥ 98%) were purchased from Cayman Chemical (Ann Arbor, MI). TAF (≥ 99%), GS-441524 (Nuc) (≥ 99%), the alanine metabolite GS-704277 (Ala) (≥ 95%), RDV-MP (≥ 98%), and RDV-TP (≥ 98%) were purchased from MedChem Express (Monmouth Junction, NJ). PhosSTOP EASYpack (PhosSTOP) was purchased from Roche (Basel, Switzerland). Bis-(p-nitrophenyl) phosphate (BNPP), dithiolthreitol, 55’ dithiobis (2-nitrobenzoic acid) (DTNB), and Tris-base (Trizma base) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The internal standards, 2-chloroadenosine (2-Cl-adenosine) and adenosine-¹⁵N_’ 5’-triphosphate disodium salt solution, were purchased from Sigma-Aldrich (St. Louis, MO, USA). Qubit^™ protein assay kit was obtained from Thermo Fisher Scientific (Waltham, MA). MTS Assay Kit (ab197010) was purchased from Abcam (Cambridge, MA, USA).

Recombinant human CatA (rhCatA) and CES1 (rhCES1) were purchased from R&D Systems (Minneapolis, MN, USA). Blank human plasma was purchased from Innovative Research (Novi, Michigan, USA). Pooled human tissue S9 fractions, including liver (HLS9), lung (HlungS9), intestine (HIS9), and kidney (HKS9), were products from XenoTech LLC (Kansas City, KS, USA). All other chemicals and reagents were of analytical grade and commercially available. The demographic information about the tissue donors is summarized in Table S4.

Cell lines

Cell lines tested in this study is summarized in Table S5. The African monkey kidney cell line Vero E6 was obtained from ATCC and grown in Eagle’s Minimum Essential Medium (EMEM) supplemented with 10% fetal bovine serum (FBS) (GIBCO, ThermoFisher Scientific). Human hepatocellular carcinoma cell line Huh7 cells were provided by Dr. Lei Yin at the University of Michigan and grown in Dulbe’co’s Modified Eagle Medium (DMEM) supplemented with 10% FBS. Human lung adenocarcinoma epithelial cells Calu-3 were provided by Dr. Daniel Goldstein at the University of Michigan and grown in EMEM supplemented with 10% FBS. Normal human bronchial epithelial cell line BEAS-2B cells were purchased from ATCC (Manassas, VA) and grown in bronchial epithelial cell growth medium (BEGM) (CC-3171) supplemented with BEGM^™ SingleQuots^™ supplements and growth factors (CC-4175) purchased from Lonza (Walkersville, MD). A549-ACE2-TMPRSS2 cells (catalog no. a549-hace2tpsa) were purchased from InvivoGen (San Diego, CA) and grown in F-12K supplemented with 10% FBS. Human pneumocyte type II carcinoma cells A549 (ATCC^® CCL-185^™), human epithelial colorectal adenocarcinoma cell Caco-2 (ATCC^® HTB-37^™), and HEK293 (ATCC CRL-1573^™) were purchased from ATCC (Manassas, VA). A549 cells were propagated in F-12K medium (ATCC) with 10% fetal bovine serum (FBS; Gibco) and ‰1 penicillin-streptomycin-amphotericin B solution (ATCC). Caco-2 cells were cultured in EMEM (ATCC) with 20% FBS and ‰1 penicillin-streptomycin-amphotericin B solution. HEK293 cells were cultured in DMEM with 20% FBS and 1‰ penicillin-streptomycin-amphotericin B solution. All the cells were propagated at 37°C in a humidified atmosphere of 5% CO₂. Cell passages were performed with TrypLE^™ Express Enzyme (Gibco) when cells reached about 80–90% confluence. iPSC-derived alveolar type 2 (iPSC-AT2) cells (SPC2 line, clone SPC2-ST-B2, Boston University) were differentiated as previously described^38–39. After removal of CHIR99021 between days 32 and 37 cells were maintained as three-dimensional alveolospheres embedded in Matrigel (Corning; 356231) in ‘CK + DCI’ media. For experiments alveolospheres were dissociated and seeded onto tissue culture-treated plates coated with Human Collagen Type IV (Sigma: C5533) at 67 μg/mL.

Enzyme, tissue S9, and human plasma activity assay

The activity of rhCatA and rhCES1 for hydrolyzing RDV, MeRDV, and IsoRDV was evaluated using a method similar to a previous study²³ with minor modifications. The optimized buffer systems for rhCatA and rhCES1 were MES assay buffer (pH 5.5) and Tris buffer (pH 7.5), respectively. The CatA assay was carried out in a 50 μL reaction mixture containing 0.5 ng/μL activated rhCatA and 20 μM prodrug (RDV, MeRDV, or IsoRDV) in the MES assay buffer (pH 5.5). The rhCatA assay mixture was incubated at 37 °C for 5 and 10 min with shaking at 1200 rpm. The rhCES1 assay was conducted in a 20 μL reaction mixture containing 40 ng/μL or 20 ng/μL rhCES1 and 20 μM prodrug in the Tris buffer (pH 7.5). The rhCES1 assay mixture was incubated at 37 °C for 20 and 40 min with shaking at 1,200 rpm. Our preliminary study found a linear relationship between the prodrug hydrolysis and the tested enzyme concentrations (CatA: 0.1–0.5 ng/μL; CES1: 20–40 ng/μL) and the incubation times (CatA: 0–10 min; CES1: 0–40 min). We calculated the enzyme activity by dividing the reduction of the prodrug by the enzyme concentration and incubation time.

We incubated 20 μM of each ProTide (RDV, MeRDV, IsoRDV) with each tissue’s S9 fractions (0.5 mg/mL HLS9, or 0.5 mg/mL HKS9, or 1 mg/mL HlungS9, or 1 mg/mL HIS9) in Tris buffer (pH 7.5) at 37 °C for 15 min and 30 min in a shaker at a speed of 1,200 rpm (Benchmark, Multi-Therm^™). For the human plasma stability study, we incubated 10 μM prodrug or Nuc in human plasma at 37 °C for 0.5, 1, 2, 4, 8, and 24 hours with shaking at 1,200 rpm (Benchmark, Multi-Therm^™). The reactions were terminated by adding a two-fold volume of acetonitrile (ACN) containing the internal standards TAF (5 μM) and 2-Cl-adenosine (0.5 μM). The samples were then vortexed for 0.5 min and centrifuged at 21,130 g for 10 min at 4°C to remove the precipitated proteins. The resulting supernatant was collected and diluted with ten volumes of 3% (v/v) ACN containing 0.1% (v/v) formic acid before being injected into an LC-MS/MS system for analysis with an assay method described below.

Cytotoxicity assay

We seeded cells at a density of 15,000 cells/well in a white-wall clear-bottom 96-well plate (Corning) and incubated cells at 37 °C overnight. When cells reached 70–80% confluence, we added serial dilutions (0, 1, 5, 10, 25, 50 μM) of prodrugs or Nuc in medium to each well and incubated the cells for 24 hours at 37 °C. Cytotoxicity was determined using the MTS Assay Kit (ab197010) according to the manufacturer’s instructions.

Bioactivation of prodrugs in various cell lines

The cell incubations and sample preparation for LC-MS/MS analysis were performed based on previously published methods^{26, 40–42} with some minor modifications. First, we prepared the lysis buffer containing 70% methanol (v/v, MeOH: H₂O = 7:3), 0.5 μM 2-chloroadenosine and 5 μM adenosine-¹⁵N_’ 5’-triphosphate as internal standards and 2% formic acid, 10% PhosStop, and 50 μM DTNB as protectors for MP- and TP-Nuc⁴¹. Cells were plated at a density of 0.5 × 10⁶ cells/well in a 6-well plate. After at least 24 hours, when the cells reached 80–90% confluence, we replaced the cell culture media with media containing drugs and incubated at 37 °C. We used 5 μM of prodrugs for HEK293, Huh-7, and iPSC-AT2 cells and 10 μM of prodrugs for Vero E6, A549, A549-ACE2-TMPRSS2, Calu-3, and BEAS-2B cells. We aspirated the medium at 6, 12, and 24 hours after adding drugs. Then, we washed the cells twice with 2 mL ice-cold PBS, scraped cells into 0.3 mL ice-cold lysis buffer, and transferred the lysis buffer to a new 2 mL tube. Then, we washed the well with 0.3 mL water and collected the wash water into the same tube. We shook the collected samples at 1500 rpm on a shaker (Benchmark, Multi-Therm^™) and then stored the samples at −80 °C. Upon analysis, extracts were centrifugated twice at 21,130 × g for 10 min at 4 °C to remove the precipitated proteins. The supernatants were transferred to clean tubes for LC-MS/MS injections and total protein concentration measurement using a Qubit assay. The inhibitors’ effects on the bioactivation of RDV, MeRDV, and IsoRDV in BEAS-2B cells were evaluated by incubating BEAS-2B cells with cell culture medium containing 10 μM of each prodrug with or without various concentrations of telaprevir (5, 10, and 20 μM) and BNPP (5, 10, and 20 μM) at 37 °C for 24 hours (RDV and MeRDV) or 48 hours (IsoRDV). After drug incubations, we aspirated the medium and collected cell lysis samples as described above.

In vitro antiviral activity study of RDV and its analogs

SARS-CoV-2 was obtained through BEI Resources [SARS-related coronavirus 2, isolates USA-WA1/2020, NR-52281 and hCoV-19/USA/GA-EHC-2811C/2021 (omicron variant lineage B.1.1.529), NR-56481]. The lack of genetic drift of our viral stock was confirmed by deep sequencing. Viral titers were determined by 50% tissue culture infectious dose (TCID₅₀) assays in A549-ACE2-TMPRSS2 cells (Reed and Muench method). All experiments using SARS-CoV- 2 were performed at the University of Michigan under biosafety level 3 (BSL3) protocols in compliance with containment procedures in laboratories approved for use by the University of Michigan Institutional Biosafety Committee and Environment, Health and Safety. For infections, A549-ACE2-TMPRSS2 cells were seeded in 24 well plates at 50,000 cells/well and allowed to grow overnight. Cells were infected with SARS-CoV-2 at a MOI of 1 in drug-containing DMEM media with 2% FBS at the final concentration of 0, 1, 5, 10, and 25 uM. The infection was allowed to proceed for 24 hours before harvesting and determination of infectious virus titers by TCID₅₀ on A549-ACE2-TMPRSS2 cells (Reed and Muench method).

Animals

Homogeneous Ces1c−/− mice were obtained from the Jackson Laboratory (JAX stock # 014096). Mice were bred in the University of Michigan animal facility. Adult males (8–12 weeks) with body weight ranging from 23 to 27 g were used for the study. All works were conducted under an animal protocol (PRO00010244) approved by the Institutional Animal Care and Use Committee at the University of Michigan according to the guidelines set by the Association for the Assessment and Accreditation of Laboratory Animal Care and the U.S. Department of Agriculture. The PK study procedure was similar to previous remdesivir mouse PK studies^26–27. Briefly, mice received a single equal mole dose (54.4 μmole/kg) of RDV (32.8 mg/kg), IsoRDV (30.5 mg/kg), MeRDV (31.3 mg/kg) via tail vein injection (IV bolus) and oral gavage. Mice received a single equal mole dose of Nuc (15.8 mg/kg) via tail vein injection (IV bolus). Mice received a single equal mole dose of GS-621763 (27.3 mg/kg) only via oral gavage because GS-621763 is designated as an oral prodrug of Nuc^6–7 and its cell incubations did not show significant improvement in intracellular bioactivation than Nuc in this study. The dose was converted from the recommended human dose (i.e., Day one loading dose: 200 mg) by a factor of 12.3. For each drug treatment by each dose route, 5 × 3 or 4 (replicates) mice were used: specifically, for each complete PK curve, we used five (A, B, C, D, E) mice: mouse A donated its plasma samples at 0.25 and 1 hour and its tissue samples at 1-hour post drug administration; mouse B donated its plasma samples at 0.5 and 2 hour and its tissue samples at 2-hour post drug administration; mouse C donated its plasma samples at 4 and 8 hour and its tissue samples at 8-hour post drug administration; mouse D donated its plasma sample and tissue samples at 12-hour post drug administration; mouse E donated its plasma sample and tissue samples at 24-hour post drug administration. In total, we used approximately 160 mice.

Mouse plasma sample preparation for quantification of prodrugs, Ala, and Nuc

Orbital blood was taken from four mice at 0.25, 0.5, 1, 2, 4, 6, 8, 12, and 24 hours after drug administration. Blood samples were centrifuged under 7000 × g for 5 min at 4 °C. The supernatants (plasma) were collected and stored at −80 °C until analysis. 20 μL of plasma were mixed with 100 μL of methanol containing 0.25 μM 2-chloro-adenosine (2-Cl-adenosine) as an internal standard. The samples were then vortexed for 5 min and centrifuged at 21,130× g for 10 min at 4 °C, and 80 μL supernatants were transferred to clean tubes and evaporated in a SpeedVac SPD1010 vacuum concentrator (Thermo Scientific, Hudson, NH). The dried samples were resuspended in 40 μL of 3% acetonitrile containing 0.1% formic acid for LC-MS/MS analysis.

Mouse tissue sample preparation for quantification of active metabolites

Extraction buffer was prepared according to the formula from a previous study that can stabilize RDV-MP and RDV-TP in the mouse tissue samples^{26, 41}: The PhosSTOP working solution was prepared by dissolving one tablet of PhosSTOP EASY pack in 1 mL water. The extraction buffer consisted of 70% methanol, 10% PhosSTOP working solution, 50 μM DTNB, and 2% formic acid. Lungs, livers, kidneys, testis, and brains were collected 1, 2, 4, and 8 hours after drug administration and snap frozen. On dry ice, frozen tissue samples were pulverized and weighed. Two volumes of ice-cold extraction buffer were immediately added to the powder of mouse lung, kidney, small intestine, testis, and brain. Three volumes of ice-cold extraction buffer were immediately added to the mouse liver powder. The mixture was vortexed thoroughly and then added with an equal volume of water (equal to the volume of extraction buffer). The mixture was vortexed at 1,500 rpm on a shaker (Benchmark, Multi-Therm^™) for 10 min and then centrifugated at 21,130 × g for 10 min at 4 °C to remove the precipitated proteins. The supernatants were transferred to clean tubes and then centrifugated at 21,130 × g for 10 min at 4 °C before the LC-MS/MS analysis.

LC-MS/MS analysis for prodrugs and metabolites

The LC-MS/MS analysis methods for prodrugs RDV, MeRDV, IsoRDV, and GS-621763, and metabolites Ala, Nuc, RDV-MP, and RDV-TP were established based on the methods reported from previous studies^{26, 40–42} with minor modifications. The LC-MS/MS system consisted of a SCIEX API 3000 mass spectrometer and a Shimadzu UFLC system (Shimadzu, Tokyo, Japan). For the quantification of RDV, MeRDV, IsoRDV, Ala, and Nuc, We used a RESTEK Ultra II C18 column (5 μm, 50 cm × 2.1 mm, Bellefonte, PA, USA) with a RESTEK UltraShield UHPLC precolumn filter (0.2 μm frit, Bellefonte, PA, USA) for chromatographic separation at 45 °C. Mobile phase A was water containing 0.1% (v/v) formic acid, and mobile phase B was ACN containing 0.1% (v/v) formic acid. The mobile phase flow rate was 0.5 mL/min, and 1 μL sample was injected for analysis. For the gradient program, mobile phase B was at 5% for the first 0.1 min, increased to 90% during the period of 0.1 min – 2 min and maintained at 90% for 1 min, then returned to 5% at 3.1 min and maintained at 5% until the end of the gradient at 5 min.

For the quantification of RDV-MP and RDV-TP, we used a BioBasic AX column (2.1 × 50 mm, 4.6 μm; ThermoFisher) with a RESTEK UltraShield UHPLC precolumn filter (0.2 μm frit, Bellefonte, PA, USA) for chromatographic separation at 50 °C. Mobile phase A was ACN-H₂O (3:7, v/v) with 10 mM NH₄Ac (pH 6.0), and mobile phase B was ACN-H₂O (3:7, v/v) with 1 mM NH₄Ac (pH 10). The mobile phase flow rate was 0.6 mL/min, and 10 μL sample was injected for analysis. For the gradient program, mobile phase B was at 2% for the first 0.1 min, increased to 60% during the period of 0.1 min – 0.35 min, and maintained at 60% until 0.75 min, then continued to increase to 98% during the period of 0.75 – 3.5 min and returned to 2% at 3.7 min and maintained at 2% until the end of the gradient at 5 min. We operated the MS in a positive mode with turbo electrospray ionization. The following MS parameters were adopted for all compounds: nebulizer gas: 15 psi; curtain gas: 15 psi; collision gas: 6 psi; ion spray voltage: 5500 V; source temperature: 550°C. For each compound, the MS delustering potential (DP), focusing potential (FP), entrance potential (EP), collision energy (CE), and collision cell exit potential (CXP) were summarized in Table S6. For prodrugs, Ala, and Nuc in tissue S9, enzyme reaction samples, and mouse plasma, the analytes were quantified using an 8-point standard curve with concentrations ranging from 31.25 to 20,000 nM, and the calibrators were prepared in extracts from untreated tissue S9, enzyme reaction systems, or mouse plasma. For intracellular metabolites quantification, analytes were quantified using a 5- to 8-point standard curve ranging in concentration from 39.0625 to 10,000 nM prepared in extracts from untreated cells or mouse tissues. All the regression coefficients of calibration curves were greater than 0.99. The peak area ratios of analytes to the internal standards were used for quantification. Quality control samples were used to ensure accuracy and precision and prepared with untreated tissue S9, enzyme reaction systems, and extracts from cells, mouse plasma, or mouse tissues.

Expression of enzymes and transporters affecting the bioactivation of RDV

RDV-activating enzymes and related transporters were selected based on their functions confirmed in previous studies, summarized in Table S7. The mRNA expression data of the enzymes and transporters of interest were extracted from the Human Protein Atlas (HPA) (http://www.proteinatlas.org)⁴³, in which the highest single-cell mRNA expression value found in a particular cell type or cell line was adopted. The mRNA expression of enzymes and transporters in Vero E6 and Calu-3 cells was extracted from a recent study⁴⁴. We selected several human tissue single cells for data analysis, including three types of human lung tissue cells: alveolar epithelial type 1 (AT1) and type 2 (AT2) cells, macrophages, endothelial cells, and two types of human bronchus cells: basal respiratory cells and respiratory ciliated cells. Among these, AT2 and respiratory ciliated cells are the major targets infected by SARS-CoV-2^45–46. The protein expression data in human and mouse tissues were obtained from the ProteomicsDB (https://www.proteomicsdb.org)⁴⁷.

Data analysis

The plasma PK data were analyzed using the software MonolixSuite 2023R1 (Lixoft; Orsay–France). The main PK parameters were calculated using a noncompartmental model. The maximum blood or tissue concentration (C_max) and time to C_max (T_max) were determined from the observed data. The area under the concentration versus time curve up to the last measured point (AUC_0-t) was calculated using the trapezoidal rule. The area under the concentration versus time curve from time zero to infinity (AUC_0-∞) was estimated by extrapolating AUC_0-t to infinity using the trapezoidal rule-extrapolation method. The terminal elimination half-life (t_1/2) was calculated by dividing ln2 by the terminal elimination rate constant λz, which was determined by linear regression of at least three data points from the terminal portion of the concentration versus time plots. The total drug cell entry within 6 hours was estimated by the summation of intracellular concentrations of the prodrug, Ala, Nuc, RDV-MP, and RDV-TP measured 6 hours post-treatment.

GraphPad Prism 9.0 (GraphPad Software, San Diego, CA) was used to generate graphs. IBM SPSS 28.0 (IBM Corp, Armonk, NY) was used for statistical analysis. The analysis of variance (ANOVA) was used to analyze the differences between groups. The linear association between two variances was assessed using Pearson’s correlation analysis. A P-value less than 0.05 was considered statistically significant. Specific tests to determine statistical significance are noted in each figure legend.

Supplementary Material

Supplemental_information

Fig. S1. Major product metabolites in human tissue homogenates after incubation with prodrugs.

Fig. S2. Intracellular concentrations of RDV-MP in various cell lines.

Fig. S3. Intracellular concentrations of prodrugs, Nuc, and Ala at six hours post-treatment in various cell lines.

Fig. S4. Nuc concentrations in the culture medium after 24 hours of drug incubations with or without Huh-7 cells.

Fig. S5. A549-ACE2-TMPRSS2 cell viability at 24 hours after the addition of 25 μM drugs.

Fig. S6. mRNA expression of ProTides activating enzymes and transporters in various cells.

Fig. S7. Protein expression of activating enzymes and transporters in mouse and human tissues.

Table S1. Intracellular accumulation of the active metabolite RDV-TP of prodrugs in various cell lines.

Table S2. Total drug cell entry within six hours of prodrugs in various cell lines.

Table S3. Active metabolite formation rate of prodrugs in various cell lines.

Table S4. Demographic information of the tissue S9 donors.

Table S5. Cell lines selected for this study.

Table S6. Mass spectrometry parameters for RDV analogs and their metabolites.

Table S7. Enzymes and transporters involved in the metabolism and disposition of RDV.

Abbreviations Used:

Ala

GS-704277, alanine metabolite of Remdesivir

AUC

area under the curve

AT1

alveolar epithelial type 1 cells

AT2

alveolar epithelial type 2 cells

ACN

acetonitrile

ANOVA

analysis of variance

BNPP

Bis-(p-nitrophenyl) phosphate

BEGM

bronchial epithelial cell growth medium

COVID-19

Coronavirus disease 2019

CES1

carboxylesterase 1

CatA

Cathepsin A

Ces1c ^−/−

carboxylesterase 1c knockout

CC₅₀

50% cytotoxic concentration

CE

collision energy

CTSA

Cathepsin A

Ctsa

Cathepsin A

CXP

collision cell exit potential

DTNB

55’ dithiobis (2-nitrobenzoic acid)

DP

delustering potential

DMEM

Dulbecco’s Modified Eagle Medium

EC₅₀

half maximal effective concentration

EP

entrance potential

EMEM

Eagle’s Minimum Essential Medium

FBS

fetal bovine serum

FP

focusing potential

HINT1

histidine triad nucleotide binding protein-1

HlungS9

human lung S9

HIS9

human intestine S9

HKS9

human kidney S9

HLS9

human liver S9

HPA

Human Protein Atlas

IsoRDV

isopropyl ester analog of remdesivir

IV

intravenous

iPSC-AT2

induced pluripotent stem cell-derived alveolar epithelial type II-like cells

LC-MS/MS

liquid chromatography with tandem mass spectrometry

MeRDV

methylpropyl ester analog of remdesivir

MOI

multiplicity of infection

MRP4

multidrug resistance-associated protein 4

OATP1B1

Organic anion transporting polypeptide 1B1

P-gp

P-glycoprotein

PK

pharmacokinetic

ProTide

phosphoramidate prodrug

RDV

remdesivir

RDV-TP

remdesivir-triphosphate

RDV-MP

remdesivir-monophosphate

rhCatA

recombinant human Cathepsin A

rhCES1

recombinant human carboxylesterase 1

SARS-CoV-2

severe acute respiratory syndrome related coronavirus

S.D.

standard deviation

SAR

structure-activity relationship

TCID₅₀

50% tissue culture infectious dose