Off-Target In Vitro Profiling Demonstrates that Remdesivir Is a Highly Selective Antiviral Agent

Remdesivir (RDV, GS-5734), the first FDA-approved antiviral for the treatment of COVID-19, is a single diastereomer monophosphoramidate prodrug of an adenosine analogue. It is intracellularly metabolized into the active triphosphate form, which in turn acts as a potent and selective inhibitor of multiple viral RNA polymerases.

infected patients (15)(16)(17). These studies demonstrated RDV as an effective antiviral agent with a favorable benefit/risk profile. RDV is intracellularly metabolized to its active triphosphate form (GS-443902) ( Fig. 1), which in turn acts as a potent and selective inhibitor of multiple viral RNA polymerases. Additional metabolites are formed both intracellularly and systemically in plasma, including alanine metabolite (GS-704277), parent nucleoside (GS-441524), and mono-and diphosphate metabolites. To assess the potential off-target toxicity of RDV, the drug and its metabolites, including parent nucleoside analog (GS-441524), intermediate metabolite MetX (GS-704277), and active triphosphate metabolite (GS-443902) (Fig. 1), were characterized in a broad panel of cellular and biochemical assays.

RESULTS
Evaluation of cytotoxicity. Cytotoxicity in human cell lines and primary cells. The cytotoxicity of RDV and GS-441524 was tested in four immortalized human cell lines (laryngeal, hepatoma, prostate, and lymphoblastoid transformed cell lines) and in seven primary human cell types, including primary human hepatocytes (PHH), primary renal proximal tubule epithelial cells (RPTECs), quiescent and stimulated human peripheral blood mononuclear cells (PBMCs), and human hematopoietic progenitor cells, including erythroid, myeloid, and megakaryoid progenitors (Table 1). These cells were chosen for their reported high sensitivity to or potential organ toxicity associated with nucleoside/tide analogs (6). ATP level was used as an indication of cell viability. After 5 to 14 days of continuous exposure to RDV, the CC 50 values ranged from 1.7 to . 20 mM, resulting in selectivity indices (SI, CC 50 /EC 50 ) from .170 to 20,000 with respect to RDV anti-SARS-CoV-2 activity (EC 50 of 9.9 nM in human airway epithelial cells [HAE]) (18).
Among the four cell lines tested, the MT-4 cell line was the most sensitive toward RDV and GS-441524, consistent with reports using other nucleoside/tide analogues (5,19). Among the seven primary cell types tested, PHH showed the lowest CC 50 toward RDV, suggesting that PHHs are susceptible to RDV-mediated toxicity.
The parent nucleoside GS-441524 showed no cytotoxicity at up to 100 mM in any of the cell lines tested, except in MT-4 (CC 50 = 69 6 26 mM). Among the primary human cells, GS-441524 showed no toxicity at the highest concentration tested (100 mM), except in the three human hematopoietic progenitor cells (CC 50 = 9.6 to 13.9 mM after 11 to 14 days of exposure).
Effect on production of reactive oxygen species. Reactive oxygen species (ROS), including peroxides, superoxides, and hydroxyl radicals, are natural by-products of normal cell metabolism and have important roles in cell signaling and homeostasis. However, excessive production of ROS can be induced by drugs and may result in cellular damage and cell death (20). In this study, we evaluated the potential effects of RDV on ROS generation in HepG2 cells, a human hepatic cell line. After a 24-h incubation, RDV-treated cells showed no significant increase in ROS levels at concentrations up to 50 mM, and a 60% increase at 100 mM (P = 0.042 compared with the DMSO control) accompanied by a parallel 60% decrease in cell viability, indicating that ROS is unlikely to be the driver of cytotoxicity (Fig. 2). In comparison, the control compound menadione showed a .3-fold increase in ROS levels after a 30-minute incubation.
Evaluation of mitochondrial toxicity. Cytotoxicity under aerobic metabolic conditions. Some nucleoside analogues have the potential to affect mitochondrial functions via diverse mechanisms. One approach to assess mitochondrial function is a comparison of effects on cell viability in the presence of glucose-favoring glycolysis (i.e., anaerobic metabolism) and galactose-favoring oxidative phosphorylation (i.e., aerobic metabolism). The latter condition may sensitize cells to compounds affecting mitochondrial functions (21). However, the predictability of this assay has not been validated for nucleoside/tide analogues (6). Using intracellular ATP quantification as a readout for cell viability, the effects of glucose and galactose on the cytotoxicity of RDV and GS-441524 were assessed in the HepG2 hepatoma cell line, which was previously identified as a suitable model for testing compounds under aerobic conditions (Table 2) (21). In addition, a similar study was done using the PC-3 prostatederived cell line, a model of rapidly proliferating cells. The CC 50 values of RDV in HepG2 cells were 3.7 and 11.1 mM in the presence of glucose and galactose, respectively ( Table 2), indicating that aerobic conditions did not enhance the cytotoxicity of RDV. In contrast, aerobic conditions in PC-3 cells enhanced RDV cytotoxicity by 6-fold (CC 50 = 1.4 in galactose culture versus 8.9 mM in glucose culture). The divergent results from HepG2 and PC-3 cells suggest that the observed effects are cell line dependent.
The parent nucleoside GS-441524 did not show any cytotoxicity in either PC-3 or HepG2 cells at the highest concentrations tested (100 mM), irrespective of the metabolic conditions (Table 2). Puromycin, used as general cytotoxic control, exhibited similar cytotoxicity in both cell types in the presence of glucose or galactose. The lack of a known nucleoside mitotoxin prevented us from using one as a positive control in this

study.
Effect on mitochondrial DNA. The potential effect of RDV and GS-441524 on mitochondrial DNA (mtDNA) was assessed in vitro by quantitative real-time PCR (qPCR) analysis following continual treatment of HepG2 cells for 10 days. Dideoxycytidine (ddC), a known inhibitor of mtDNA replication, was used as a positive control. HepG2 cells treated with RDV showed a lack of dose response, indicating the lack of a specific effect on mtDNA synthesis. The effect of the parent nucleoside GS-441524 on mtDNA content was overall minimal (Table S1 in the supplemental material).
Effect on mitochondrial protein synthesis. The effect of RDV and GS-441524 on mitochondrial protein synthesis was assessed following a 5-day incubation with PC-3 cells. This particular cell model was chosen for its prior successful use in the studies of mitochondrial toxins (6). The selective effect of a compound on mitochondrial protein synthesis was determined by quantification of the level of cytochrome oxidase subunit 1 (COX-1, encoded by mtDNA) and succinate dehydrogenase A (SDH-A, encoded by nuclear DNA) (5,19). RDV affected the levels of COX-1 and SDH-A to a similar extent, with CC 50 values of 8.9 and 8.6 mM, respectively (Table 3, Fig. 3). These effects manifested in the same range of concentrations as the cytotoxicity measured by cellular ATP levels, indicating a lack of any selective effect of RDV on mitochondrial protein synthesis. GS-441524 showed no effect on protein synthesis up to the highest concentration tested (100 mM) ( Table 3, Fig. 3). Chloramphenicol was used as a positive control, and its specific effect on mitochondrial protein synthesis was consistent with published data (5,19).
Effect on mitochondrial respiration. RDV and GS-441524 were further evaluated for their effects on mitochondrial spare respiratory capacity in the human cell lines PC-3 and HepG2 and the primary cells PHH and RPTEC by measuring the rate of oxygen consumption (OCR) using a Seahorse Extracellular Flux Analyzer (5,(22)(23)(24). Overall, there was a lack of a specific effect on cellular respiration among the four cellular systems tested. The cellular respiration of RDV-treated cells decreased in parallel with decreases in ATP level and total DNA level in HepG2, PHH, and RPTEC, indicating a lack of specific inhibition of cellular respiration (Table 4, Fig. 4). The most profound effect of RDV on spare mitochondrial respiratory capacity was seen in PC-3 cells, where RDV  In Vitro Toxicity Evaluation of Remdesivir Antimicrobial Agents and Chemotherapy inhibited mitochondrial spare respiratory capacity with a CC 50 value (CC 50 = 2.5 mM) lower than those for the inhibition of ATP level and total DNA (CC 50 = 24.0 and 12.5 mM, respectively). Among all four cell systems, GS-441524 showed no effect on mitochondrial respiration at the highest concentration tested (100 mM) ( Table 4, Fig. S1).
In contrast, the positive control chloramphenicol showed specific inhibition of mitochondrial respiration and minimal impact on ATP level and total DNA (Fig. S2). GS-704277, a major intermediate RDV metabolite, was evaluated for its effect on mitochondrial spare respiratory capacity, ATP levels, and total DNA in PHH after both an acute 4-h treatment and a chronic 3-day incubation, the default condition in the experiments described in the previous paragraph. After the 4-h incubation, RDVtreated PHH showed parallel decreases in mitochondrial spare respiratory capacity (21% to 27%) and cellular ATP levels (17% to 31%), indicating a lack of mitochondriaspecific toxicity (Fig. S3). After a 3-day treatment with RDV, PHH showed parallel decreases in mitochondrial spare respiratory capacity, ATP levels, and total DNA, with CC 50 values of 7.6, 7.8, and 13.4 mM, respectively, indicating general toxicity instead of specific mitochondrial toxicity (Table 4, Fig. 4). Neither of the two major systemic metabolites of RDV, GS-704277 nor GS-441524, exhibited any effects on mitochondrial  spare respiratory capacity, ATP levels, or total DNA at the highest concentration tested (100 mM) after either a 4-h or a 3-day treatment ( Table 4, Fig. S1, Fig. S3, Fig. S4).
Interaction with host RNA and DNA polymerases. The active triphosphate metabolite GS-443902 was tested in multiple biochemical assays to assess its interaction with key human DNA and RNA polymerases. The enzymatic activities of human DNA polymerases a and b, as well as that of RNA polymerase II, were unaffected by GS-443902 up to 200 mM, the highest concentration tested (Table 5). In addition, GS-443902 was tested for its incorporation into nucleic acids by host mitochondrial DNA and RNA polymerases using a single nucleotide incorporation assay. GS-443902 was a poor substrate of mtDNA polymerase g, with no detectable incorporation when tested under the supratherapeutic concentration of 50 mM (versus a maximum drug concentration in serum [C max ] of 6mM at a 150-mg clinical dose) (13). It was also a poor substrate for POLRMT, with a rate of incorporation equal to 5.8% relative to ATP when tested under the supratherapeutic concentration of 500 mM (Table 6). This result contrasts with the significantly higher incorporation rates of 92% and 112%, for the triphosphate forms of BMS-986094 and balapiravir, respectively, two anti-HCV nucleosides associated with clinical toxicity (5). Together, these data further support the hypothesis that RDV has low potential to induce mitochondrial toxicity.
Molecular target screen. In addition to the aforementioned known off-targets of nucleoside/tide anlaogues, the RDV-containing diastereomeric mixture GS-466547 and the parent nucleoside GS-441524 were screened against a panel of 87 targets consisting of receptors, ion channels, transporters, and enzymes involved in a wide range of biological processes (Table S3). At 10 mM, none of the compounds showed detectable interaction with any of the 87 targets tested.

DISCUSSION
RDV is a single diastereomer monophosphoramidate prodrug of an adenosine analog with broad-spectrum antiviral activity. Using a prodrug as a way to deliver an active molecule may potentially result in intracellular accumulation of the less cell-permeable metabolites, depending on the activity of the enzymes participating in the metabolism of a particular drug and its conversion into the active form. Therefore, detailed examination of the functional consequences of exposure to high doses is critical for the evaluation of the drug safety.
This study presents a detailed characterization of RDV in a panel of in vitro cytotoxicity and off-target screening assays, undertaken to identify potential safety and toxicity liabilities. In this report, the general cytotoxicity of RDV measured as 50% cytotoxic concentration (CC 50 ) following 5 to 14 days of continuous exposure to the drug ranged from 1.7 mM to .20 mM in selected human cell lines and primary human cells. It is worth noting that the RDV exposure in these cell culture studies was significantly higher than the systemic exposure to RDV with the repeated dosing of 150 mg in humans (plasma C max = 2,720 ng/ml = 4.5 mM; t 1/2 = 1.11 h) (13). The parent nucleoside and major systemic metabolite GS-441524 showed less cytotoxicity than RDV. The differences in toxicity between these two compounds are likely due to the significantly lower level of active triphosphate in GS-441524-treated cells than in RDV-treated cells, as observed in this study as well as other studies (18). However, RDV and GS-441524 showed similar levels of toxicity in three hematopoietic progenitor cells. Cell-dependent cytotoxicity has been widely observed for nucleoside/tide analogs, which can be affected by the duration of compound exposure (5 day versus 14 day), compound permeability, intracellular activation, and differential cellular characteristics, including replication rate and bioenergetic status (5).
Among the primary cells tested, PHH showed the highest sensitivity to RDV, possibly due to high cellular permeability and efficient intracellular metabolism, leading to high levels of the triphosphate metabolite. This observation could shed light on the liver enzyme elevations observed in healthy volunteers treated with repeated doses of RDV (13). In contrast, the two main RDV metabolites in plasma, GS-704277 and GS-441524, are unlikely to contribute to clinical liver toxicity due to their low permeability and ineffective intracellular metabolism or limited systemic exposure. In two phase I studies, RDV was evaluated in healthy subjects with single-dose intravenous (i.v.) administration (across the dose range of 3 to 225 mg; n = 78 RDV and 18 placebo) or multiple doses of 150 mg once daily for 7 or 14 days (n = 16 RDV and 8 placebo) (13). Overall, RDV was well tolerated in both studies, and all adverse events were grade 1 or 2 in severity. In the repeat-dose study, reversible, treatment-emergent grade 1 or 2 alanine aminotransferase (ALT) (50% RDV, 13% placebo) and aspartate aminotransferase (AST) (44% RDV, 0% placebo) elevations were observed. Because transaminase elevations have been observed in patients with COVID-19 (25), the potential of RDV to a The rate of single nucleotide incorporation was measured in the presence of 50 mM nucleotide analog for DNA Pol g and 500 mM for POLRMT and expressed as the % of the natural dNTP or NTP incorporation at the same concentration. Data are presented as the average 6 SD from three or more independent experiments. ND, not done.
exacerbate transaminase elevations in patients with COVID-19 may be difficult to discern. This has been supported by multiple RDV clinical trials. First, an open-label study (n = 397 RDV) without a placebo control arm showed little difference between 5-day and 10-day RDV treatment durations for adverse events of elevations in ALT (6% and 8%, respectively) or AST (5% and 7%, respectively) (15). Second, in a study of RDV in 584 hospitalized patients with moderate COVID-19, patients were randomized at a 1:1:1 ratio to receive 10-day RDV, 5-day RDV, or standard-of-care treatment; there was little difference between the three arms for adverse events of elevations in ALT (32%, 34%, and 39%, respectively) or AST (32%, 32%, and 33%, respectively). Grade 3 or 4 adverse events were reported less frequently for patients who received RDV (ALT: 3%, 2%, and 8%, respectively; AST: 1%, 3%, and 6%, respectively) (17). Finally, in the recently completed double-blind study of RDV versus placebo in 1,062 hospitalized patients with COVID-19, adverse events of increased amino transferase levels, including ALT, AST, or both, were reported in 6% of subjects treated with RDV for up to 10 days compared with 11% of subjects in the placebo group, indicating lack of specific druginduced liver enzyme elevation (16). All of these reports consistently support minimal contribution of RDV to transaminase elevations in patients with COVID-19.
To understand whether the cytotoxicity is associated with mitochondrial toxicity, we measured the effects of RDV on mitochondrial DNA content, protein level, and cellular respiration. In HepG2 cells, RDV showed a lack of a specific dose-dependent effect on mtDNA synthesis. Similarly, RDV did not show specific inhibition of a mitochondrial DNA-encoded protein (CC 50 of 8.9 mM) over a nuclear DNA-encoded protein (CC 50 of 8.6 mM) in PC-3 cells. Additional biochemical assays showed the active triphosphate metabolite GS-443902 was a poor substrate for mitochondrial DNA and RNA polymerases, even when tested at supraphysiological concentrations of 50 and 500 mM, respectively. A recent paper reported the effects of RDV on cellular toxicity through a transcriptomic analysis, and reported effects of RDV on mitochondrial function (26). However, it is not clear whether the results were due to specific mitochondrial toxicity or general cellular toxicity. In addition, the relation between specific gene expression and mitochondrial toxicity has not been fully validated.
Taken together, we conclude that RDV has low potential for the off-target toxicities described for other nucleoside analogs, including mitochondrial toxicity. Furthermore, neither RDV nor its systemic metabolites induced ROS formation in vitro or interacted with any of the 87 targets in the molecular screen. Consistent with the clinical observations of drug-related elevations of liver transaminases following multiple doses of RDV, primary hepatocytes showed high in vitro susceptibility to RDV. In clinical settings of COVID-19 treatment in hospitalized patients, the risk associated with possible RDVrelated liver enzyme elevations is substantially lower than its established benefits in hospitalized COVID-19 patients.

In Vitro Toxicity Evaluation of Remdesivir Antimicrobial Agents and Chemotherapy
Cell cultures. The HepG2 galactose-adapted cell line was established by culturing HepG2 cells in glucose-free Dulbecco's modified Eagle medium (DMEM) (Gibco, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS), 100 units/ml penicillin, 100 units/ml streptomycin, 2 mM glutamine, 1 mM sodium pyruvate, and 10 mM galactose for 3 weeks prior to using in the assays. The PC-3 galactoseadapted cell line was established by culturing PC-3 cells in glucose-free Kaighn's F12 medium supplemented with 10% FBS, 100 units/ml penicillin, 100 units/ml streptomycin, and 7 mM galactose for 3 weeks prior to using in the assays. The cells were passaged twice per week to maintain subconfluent densities. The cell culture procedures for RPTEC and PHH cells followed the vendor's recommendations. PBMCs were isolated from human buffy coats using standard Ficoll separation, stimulated as described elsewhere (19), and tested at both quiescent and stimulated stages. The stimulated PBMCs were obtained from quiescent PBMCs stimulated with 10 units per ml of recombinant human interleukin 2 (hIL-2) and 1 mg/ml phytohemagglutinin P (PHA-P) for 48 h prior to drug treatment. Normal human primary bone marrow (BM) light-density cells were from three different lots obtained from AllCells (Emeryville, CA) or Lonza (Walkersville, MD). All primary human cells were cultured according to the manufacturer's protocols. All cells were cultured at 37°C in a 5% CO 2 incubator with 90% humidity, unless noted otherwise.
General cytotoxicity and cell viability. Cytotoxicity in human cell-lines, PHH, RPTEC, and PBMC. Cells were treated with compounds for 4 h to 14 days depending on assay type. Cultures of freshly isolated PHH cells required fresh medium every 24 to 48 h, so compounds and medium were refreshed on days 0, 2, and 4 or days 0, 1, and 3 during the 5-day culture. After the incubation period, cell viability was measured by the addition of CellTiter Glo viability reagents (Promega, Madison, WI). The luminescence signal was quantified on an EnVision or Victor luminescence plate reader (Perkin-Elmer, Waltham, MA) after the cells were incubated with the reagents for 10 min at room temperature. The compound concentration that caused a 50% decrease in the luminescence signal (CC 50 ), a measure of toxicity, was calculated by nonlinear regression using a sigmoidal dose-response (variable slope) equation as described in the Data Analysis section.
Cytotoxicity in human hematopoietic progenitor cells. The effects of the compounds on the proliferation of human erythroid, myeloid, and megakaryoid progenitors were tested in MethoCult84434, a methylcellulose-based colony assay conducted by StemCell Technology (Vancouver, Canada) (27). After an 11-to 14-day culture, the hematopoietic progenitor colonies (CFU-E, BFU-E, CFU-GM, and CFU-GEMM [E, erythroid; BFU, burstforming unit; GM, granulocyte/macrophage; GEMM, multilineage progenitors]) were enumerated and the CC 50 values were calculated as described in the Data Analysis section.
Detection of intracellular reactive oxygen species level. HepG2 cells were seeded at densities of 12 Â 10 3 cells per well in 96-well plates, with a final volume of 160 ml per well. The cells were incubated overnight at 37°C in a 5% CO 2 and 90% humidity incubator with Eagle's minimum essential medium (EMEM) supplemented with 10% FBS (HyClone, Logan, UT), 100 units/ml penicillin, and 100 mg/ml streptomycin (P/S) (Gibco, Carlsbad, CA). On the next day (day 1), cells in 42 out of the 96 wells per plate were treated with RDV for 24 h. RDV was added to the cell culture plate directly using an HP D300 dispenser (Hewlett-Packard, Palo Alto, CA) with a starting concentration of 100 mM, with 2-fold dilutions to generate a total of 6 concentration points with six replicates for each concentration and 6 wells without RDV as negative controls. On the day of the reactive oxygen species (ROS) assay (day 2), 18 wells out of the remaining untreated 54 wells on the original 96-well plate were incubated with either 0 mM, 100 mM, or 200 mM menadione (Sigma-Aldrich, St. Louis, MO) for 30 min, with 6 replicates for each treatment. A mixture of CellROX Deep Red reagent and NucBlue Hoechst 33342 was added to the cells treated with menadione, RDV, and dimethyl sulfoxide (DMSO). The final DMSO concentration was 0.5% in all wells. The cell culture plate was incubated in a 37°C incubator for 30 min. The cell culture plate was then washed 3 to 4 times with 200 ml/well of 1Â phosphate-buffered saline (PBS) buffer without Ca 21 and Mg 21 (Corning, Corning, New York), followed by a final addition of 100 ml of 1Â PBS in each well. The assay plate was scanned using a Cellomics ARRAYSCAN VTI (Thermo Fisher Scientific, Waltham, MA) high content screening instrument with two channels. One channel measured the intensity of CellROX Deep Red using excitation/emission at 640/665 nm, while the other channel measured the signal intensity from NucBlue/Hoechst 3334 using excitation/emission at 360/460 nm. The ATP level was measured using CellTiter Glo viability reagents (Promega, Madison, WI) in cells treated with RDV for 24 h in a parallel experiment.
Data analysis of calculation of CC 50 and IC 50 values. The CC 50 values were defined as the concentration causing 50% decrease in cell viability, DNA level, protein level, or spare respiratory capacity, in comparison to the DMSO control. The 50% inhibitory concentration (IC 50 ) values were defined as the concentration causing a 50% decrease in product formation in the biochemical assays. Data were analyzed using GraphPad Prism 8.0 (La Jolla, CA). The CC 50 and IC 50 values were calculated by nonlinear regression analysis using sigmoidal dose-response (variable slope) equation (four parameters logistic equation): where X is the log of the concentration of the test compound and Y is the response. The CC 50 or IC 50 values were calculated as an average of three or more independent experiments. Data analysis of measuring mitochondrial DNA. The relative amount of mitochondrial DNA (mtDNA) in treated samples was determined using a relative quantification method based upon the threshold cycle (2 2DDC T ) formula (28). The amount of mtDNA in compound-treated samples relative to the DMSO-treated controls (% mtDNA) was calculated based on the following formulas: where C T , cyt b and C T , b-actin represent the cycle threshold values for the amplification of cytochrome b and b-actin, respectively, as determined by the computational analysis of amplification curves using the ABI Prism software. The final results are presented as the mean % mtDNA 6 SD (standard deviation) from two independent experiments, each performed in triplicate. The 2 2DDC T method was validated for cytochrome b and b-actin genes by determining the DC T values for amplification reaction mixtures containing various amounts of total cellular DNA. Minimal differences were observed in the DC T values in samples containing 5 to 40 ng of total cellular DNA, indicating that neither the amplification nor detection efficiencies of cytochrome b and b-actin were affected by the amount of DNA template within the dilution range relevant for the quantitative analysis performed (Table S5 in the supplemental material).
Measuring mitochondrial DNA. mtDNA was generated by real-time PCR from total DNA isolated from HepG2 cells using the QIAamp DNA minikit (Qiagen, Valencia, CA) according to the manufacturer's protocol. Real-time PCRs were performed using TaqMan universal mastermix (Applied Biosystems, Foster City, CA) in an ABI Prism 7900HT Fast real-time PCR system (Applied Biosystems). Quantification of mtDNA was achieved by amplification of a fragment of the mitochondrial specific cytochrome b gene using the primers and probe. The oligonucleotide sequences of the primer and probe used were cytochrome b forward 59-CCTTCCACCCTTACTACACAATCAA-39; cytochrome b reverse 59-GGTCTGGTGAG AATAGTGTTAATGTCA-39; and cytochrome b FAM-ACGCCCTCGGCTTAC-BHQ1. Chromosomal DNA was quantified by the amplification of a fragment of the b-actin gene using a b-actin Assay-on-Demand kit (Applied Biosystems). Quantification of mitochondrial and chromosomal DNA were performed independently using approximately HepG2 cell culture and compounds' treatment was based on a previously published method (19,29).
Mitochondrial protein synthesis assay. PC-3 cells were treated with the compounds for 5 days and analyzed with the MitoTox MitoBiogenesis in-cell enzyme-linked immunosorbent assay (ELISA) kit (MitoSciences/Abcam, Eugene, Oregon) as described previously (5,19). The assay uses quantitative immunocytochemistry to measure the protein levels of nuclear DNA-encoded succinate dehydrogenase (SDH-A; complex II) and mitochondrial DNA-encoded cytochrome c oxidase (COX-1; complex IV) in cultured cells.
Mitochondrial respiration in cell lines and primary cells. Mitochondrial respiration was monitored by measuring the rate of oxygen consumption (OCR) of the cells after 4-h or 3-day treatments with the compounds, using a Seahorse extracellular flux analyzer (XFe-96) based on published protocols (5,(22)(23)(24) and modified from a previously published method (5). The OCR signals were normalized by cell numbers using DNA content, which method was based on a previously published method (5). The PC-3, HepG2, RPTEC, and PHH cells were seeded at optimized densities of 5 Â 10 3 , 5 Â 10 3 , 7.5 Â 10 3 , and 25 Â 10 3 cells per well, respectively, in XF 96-well plates (Seahorse Bioscience, North Billerica, MA) with a final volume of 160 ml per well. The cells were incubated with compounds for 3 days, except for PHH, which received compound treatment for 4 h or 3 days. Compounds were added to the assay plate directly using an HP D300 dispenser (Hewlett-Packard, Palo Alto, CA) with 2-fold dilutions to give a total of 9 concentration points with six replicates. On the day of the assay, the cell culture medium was replaced with XF assay medium (pH 7.4) containing 10 mM glucose and 1 mM freshly prepared pyruvate. The mitochondrial respiration was monitored by measuring the rate of oxygen consumption (OCR) on a Seahorse Extracellular Flux (XFe-96) analyzer based on published protocols (5,(22)(23)(24). All concentrations were final after mixing unless noted otherwise. Multiple parameters were measured after the sequential injection of the ATP synthase inhibitor oligomycin (2 mM for all cell models), the uncoupler FCCP (0.25 mM for PC-3, 0.5 mM for HepG2 and PHH, and 1.5 mM for RPTEC), and a mixture of the mitochondrial complex I inhibitor rotenone and the complex III inhibitor antimycin A (0.5 mM each for all cell models) (Mito Stress Test kit, Seahorse Biosciences) (22)(23)(24). The spare respiratory capacity was obtained by subtracting the rate of basal respiration from the rate of maximal respiration and normalized to the cell number obtained from DNA level measurements. The data reported for each treatment are the average of the results from six replicates. The methods of detection of measuring ATP level in parallel experiments was based on a previously published method (5).
Biochemical assays. DNA and RNA templates and primers. The DNA polymerase alpha and gamma inhibition assays utilized a 78-mer DNA template and a 19-mer DNA primer: D78 59-ACACA TGATACTACGAATTTTATGCTTCCAATGCCTTACAGTTCTCTAGCGGTGGCGCCCGAACA GGGACCTGAAAGC-39 and D19 59-GTCCCTGTTCGGGCGCCAC-39 were purchased from Integrated DNA Technologies (Coralville, IA). Activated fish sperm DNA was purchased from USB/Affymetrix (Santa Clara, CA) and used as a template for the DNA polymerase beta inhibition assay. The DNA template used in the RNA POLII inhibition assay was a 1,188-bp restriction fragment containing the cytomegalovirus (CMV) immediate early promoter (Promega, Madison, WI). For single nucleotide incorporation, a DNA 19-mer primer and a DNA with numerical aperture of 1.4 at different focal planes to enable computational reconstruction of threedimensional objects with complex shape and estimation of their volume. Calcein was excited with the white laser at 488 nm and the fluorescence signal was detected in the range of 498 to 600 nm using a Leica HyD hybrid detector. The imaging was performed at 37°C to avoid temperature-related changes in shape of the cells. The estimation of the cell volume was verified using the fluorescent beads of known diameter by comparison of the values obtained after calculations based on either the diameter or 3-D reconstruction. This revealed that only minor correction was needed to account for elongation of the point spread function in vertical direction for the objective with high (1.4) numerical aperture, unlike for the objective with low (0.4) numerical aperture. In addition, the results of confocal imaging were confirmed using a cell counter (Cellometer K2, Nexcelom Bioscience, Lawrence, MA, USA) with imaging capabilities, allowing for determining the diameter of the cells in suspension after trypsinization.
Statistical analysis. The differences between compound-treated groups and DMSO-treated groups were analyzed using ordinary one-way ANOVA (Sidak's multiple-comparison test) (GraphPad Prism 8.1) and P values less than 0.05 were defined as statistically significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.3 MB.

ACKNOWLEDGMENTS
Becky Norquist provided medical writing assistance on behalf of Gilead. All authors are employees of Gilead Sciences, Inc. except O.B. and M.P. who were previously employed by Gilead Sciences, Inc. All authors may hold stock or stock options in Gilead Sciences, Inc.
The studies were fully sponsored by Gilead Sciences, Inc.