Analysis of the Potential for N4-Hydroxycytidine To Inhibit Mitochondrial Replication and Function

N4-Hydroxycytidine (NHC) is an antiviral ribonucleoside analog that acts as a competitive alternative substrate for virally encoded RNA-dependent RNA polymerases. It exhibits measurable levels of cytotoxicity, with 50% cytotoxic concentration values ranging from 7.5 μM in CEM cells and up to >100 μM in other cell lines.

polymerase (POLRMT) or by nuclear RNA polymerase (Pol) I, II, or III, resulting in undesired side effects (7,8). The development of several ribonucleoside analogs has been stopped as a result of toxicity observed during clinical trials. BMS-986094, a prodrug of 2=-C-methylguanosine being investigated for the treatment of HCV, was halted during phase II clinical trials due to cardiac and kidney toxicity (9). Balapiravir, a prodrug of 4=-azidocytidine that was investigated for the treatment of HCV, was stopped in phase II clinical trials due to hematologic toxicity (10). In both cases, later studies suggested the cause of the observed toxicity was mitochondrial dysfunction (8). Zidovudine, the first approved antiretroviral for HIV treatment, was shown to have dose-limiting toxicity resulting from the depletion of mitochondrial DNA (mtDNA) (2). To examine early in the drug development process the potential for nucleoside analogs to cause mitochondrial toxicity, a series of in vitro assays have been developed to characterize the effects of a compound on mitochondrial DNA copy number and mitochondrial function. Key among these are the measurement of mtDNA levels, lactate production, and mitochondrial protein expression (7,(11)(12)(13).
N 4 -Hydroxycytidine (NHC) is a ribonucleoside analog currently in late-stage preclinical development. NHC has shown broad-spectrum antiviral activity against viruses in the Togaviridae, Flaviviridae, Coronaviridae, Pneumoviridae, and Orthomyxoviridae families (14)(15)(16)(17)(18). It has been shown that the 5=-triphosphate of NHC can serve as a competitive alternative substrate and be incorporated into viral RNA (14). Upon incorporation, NHC inhibits viral RNA genome replication, either by disrupting the secondary structure of the genome promoter regions and thus inhibiting replication of the virus (14) or by the introduction of mutations into the viral RNA genome that leads to error catastrophe (15,19). Given that this compound is a ribonucleoside analog with the potential for interfering with mitochondrial function, a series of experiments designed to evaluate the effect of prolonged exposure to NHC on mitochondrial death or mitochondrial dysfunction in vitro was conducted.

RESULTS
Cytotoxicity of NHC in HepG2 and CEM cell lines. Prior to conducting experiments on mitochondrial function, the 50% cytotoxic concentration (CC 50 ) of NHC was determined in each of the cell lines used in this study. The results of these assays are summarized in Table 1 and in Table S1 in the supplemental material. The CC 50 values for the human hepatic origin HepG2 cell line were measured in cells grown in glucose-containing media and in glucose-free media. Cells grown in the presence of glucose are subject to the Crabtree effect, whereby they are able to produce nearly all of their ATP through glycolysis, even though the cells possess functional mitochondria (20). Furthermore, it has been shown that cells are more susceptible to mitochondrial toxins when glucose is replaced with galactose in the incubation media (20). In a side-by-side experiment, the CC 50 values for NHC were similar in HepG2 cells incubated with glucose and in HepG2 cells incubated with galactose instead of glucose. By this measure, NHC does not impair mitochondrial function in this cell line. Among other tested cell lines of different tissue and species origin, the T-lymphoblastoid origin CEM cell line was the most susceptible to NHC treatment, with CC 50 values 3-to 4-fold lower than in other cell lines tested (see Table S1); therefore, the CEM cell line was chosen to evaluate the potential for NHC to cause mitochondrial toxicity. In addition, the HepG2 cell line was also chosen since HepG2 cells have been traditionally used to evaluate mitochondrial toxicity (2,14,20,21) and have demonstrated their utility. NHC is efficiently converted in cells to its 5=-triphosphate. Ribonucleoside analogs must be converted into their 5=-triphosphate metabolites to be utilized as the substrates by RNA polymerases and incorporated into RNA. To confirm that NHC is converted to its 5=-triphosphate form, NHC-TP, HepG2, PC-3, and CEM cells were incubated with NHC and intracellular levels of NHC-TP were quantified by liquid chromatography-tandem mass spectrometry (LC-MS/MS). In HepG2 cells treated with 20 M NHC, intracellular NHC-TP levels reached a maximum concentration of 732.7 Ϯ 2.9 pmol/10 6 cells after a 6-h incubation (Fig. 1A) and remained stable for up to 24 h in the presence of drug in the media. This level appears to be higher than the previously reported value of 71.12 Ϯ 22.66 pmol/10 6 cells, which was determined after a 6-h incubation with 10 M NHC (14). In CEM cells treated with 10 M NHC, the intracellular levels of NHC-TP reached a maximum of 158.4 Ϯ 14.5 pmol/10 6 cells after a 1-h incubation (Fig. 1B). In PC-3 cells treated with 10 M NHC, the intracellular levels of NHC-TP reached a maximum concentration of 819.5 Ϯ 16.8 pmol/10 6 cells after a 6-h incubation (Fig. 1C). These experiments confirm that NHC is efficiently converted to its 5=-triphosphate metabolite in all tested cells.
NHC-5=-triphosphate is a substrate for POLRMT in primer extension assays. A nonradioactive primer extension assay was used to determine whether NHC-TP can be incorporated into RNA by POLRMT as a cytidine analog. The incorporation efficiency of a ribonucleoside analog is usually compared to the natural ribonucleoside; however, in the case of CTP analogs, CTP is incorporated much more efficiently by POLRMT than most CTP analogs which makes it difficult to directly compare their incorporation efficiencies under the same conditions (7). To facilitate the measurement, 3=-dCTP can be used as an intermediary (7). In a single nucleotide incorporation primer extension assay it was determined that POLRMT favors 3=-dCTP as a substrate over NHC-TP with a discrimination value of 12.4 Ϯ 1.2 (Fig. 2). Previously, it has been determined that POLRMT favors CTP over 3=-dCTP with a discrimination factor of 59.7 Ϯ 1.6 (7). By extrapolation, POLRMT favors CTP as a substrate over NHC-TP by a factor of approximately 740. This value suggests that NHC-TP is a relatively efficient substrate for POLRMT (7). It was further determined that NHC-TP does not cause immediate chaintermination upon incorporation into nascent chain RNA in an alternative assay where multiple nucleotides could be incorporated (Fig. 3). As shown in Fig. 3, after NHC-TP was incorporated as a C-analog at the ϩ2 position, NHC-TP can be further incorporated as a G-analog forming a ϩ3 band and as a U-analog at the ϩ4 position in conditions when no competing natural GTP or UTP are present. Partial chain termination occurs at multiple sites further downstream after incorporation of NHC-TP, as demonstrated by accumulation of the unextended bands at the ϩ7 to ϩ9 positions.
NHC causes stronger reduction in mitochondrial DNA-encoded protein expression than in nuclear DNA-encoded protein expression. Since NHC-TP is a substrate for POLRMT and the NHC-modified RNA may have different translation properties, it is important to understand if POLRMT-mediated protein expression is selectively affected in cells incubated with NHC. The effect of NHC on expression of mitochondrial DNAencoded protein cytochrome c oxidase subunit 1 (COX1) was compared to the effect on expression of nuclear DNA-encoded protein succinate dehydrogenase A (SDH-A) in PC-3 cells using an enzyme-linked immunosorbent assay (ELISA)-based quantitative assay. PC-3 cells were chosen because they have been shown to be highly sensitive to mitochondrial toxins in this assay and, therefore, are preferable for such analysis (22). The results of this assay are shown in Fig. 4 and Table 2. In PC-3 cells incubated with increasing concentrations of NHC for 5-days, the measured 50% inhibitory concentration (IC 50 ) value for COX1 protein expression was 2.7-fold lower relative to the IC 50 for SDH-A protein expression, suggesting that NHC has slightly stronger negative effect on mitochondrial mRNA transcription and/or translation. For comparison, chloramphenicol, which was used as a positive control, exhibited stronger effect on mitochondrial protein expression, with IC 50 values of COX1 protein inhibition ϳ12-fold lower compared to that of SDH-A protein (see Table S2). Similarly, dideoxycytidine (ddC) showed selective inhibition of COX1 expression with an SDH-A/COX1 IC 50 ratio of Ͼ49.2.
Long-term incubation with NHC in CEM or HepG2 cells does not cause a reduction in mtDNA levels, or an increase in lactate production. Mitochondrial toxicity can be exhibited by a decrease in mitochondrial number per cell or by a decrease of function in existing mitochondria. The decrease of functional activity of mitochondria should result in less production of ATP by mitochondria even if the mitochondrial number has not changed. In glucose-containing culture medium conditions, the decrease in mitochondrial number or in mitochondrial function in the affected cell puts additional demands on ATP production through mitochondrionindependent anaerobic glycolysis, which is registered by an increase in lactate production. To determine whether any of the mechanisms are involved, i.e., the decrease in The incorporation efficiency was evaluated based on the extension of 22-mer to 23-mer products. The measured K 1/2 values are shown on the right of the graph. The discrimination between the analog and 3=-dCTP, D* analog , was calculated as K 1/2, analog /K 1/2, 3=-dCTP (where K 1/2 is defined as the analog triphosphate concentration resulting in 50% product extension). The D* EIDD-2061 in this experiment is 13.2. In second analogous experiment repeat, the D* EIDD-2061 was measured to be 11.5. mitochondrial quantity or the decrease of mitochondrial function due to decrease in mitochondrial protein quantity or protein function (13,21,23), the effect of long-term incubation with NHC on both mitochondrial number and lactate production in treated cells was investigated.
To measure the long-term effects of exposure to NHC, HepG2 cells were incubated with NHC at 10 and 100 M for 14 days, and CEM cells were incubated with NHC at 10 and 30 M for 14 days. The concentrations were selected near or just above the CC 50 concentrations for the corresponding cell line. Higher treatment concentrations could not be tested since not enough cells survived to perform the assay. After 14 days of treatment, mtDNA copy number in cells and lactate levels in cell culture media were measured relative to the mock-treated cells. ddC, a known inhibitor of DNA polymerase   Tables 3 and  4. There was not a significant difference in relative mtDNA levels in HepG2 cells treated with NHC for 14 days at either concentration. After treatment at 10 M, the level of relative mtDNA in HepG2 cells was 115.0 Ϯ 92.4% compared to a mock-treated control, and after treatment at 100 M, relative level of mtDNA was 68.4 Ϯ 17.6% of the control, but the difference was not statistically significant (P ϭ 0.8). Parallel treatment with positive controls ddC and EtBr resulted in mtDNA decreases to 7.1 Ϯ 2.8% and 19.6 Ϯ 12.1%, respectively, compared to a mock-treated control. The lactate levels in the NHC-treated cells were 103.7 Ϯ 21.2% relative to the mock-control when cells were treated at 10 M and 111.5 Ϯ 38.5% when treated at 100 M, indicating no significant increase in lactate production after 14 days of treatment at either concentration. Treatment with positive controls ddC and EtBr resulted in extracellular lactate increases to 195.9 Ϯ 61.9% and 242.5 Ϯ 47.8%, respectively. In CEM cells treated with NHC at 10 or 30 M, relative mtDNA was measured at 80.0 Ϯ 19.1% and 78.7 Ϯ 15.8%, respectively, compared to the mock-treated controls (100 Ϯ 10.8%). Although the differences of relative mtDNA levels from the controls were statistically significant (P Ͻ 0.05), they were likely not meaningful, particularly considering that no additional decrease was observed at a higher treatment concentration. For comparison, in CEM cells treated with positive controls, the relative mtDNA levels were decreased to 0.15 Ϯ 0.06% after incubation with ddC and to 0.39 Ϯ 0.11% for cells incubated with EtBr. The lactate levels in CEM cells incubated with NHC at 10 M were increased slightly to 118.1 Ϯ 3.4% relative to the mock control but were not increased after incubation with NHC at 30 M (101.8 Ϯ 12.7%). In CEM cells treated with positive controls, the lactate levels were 310.5 Ϯ 71.2% after treatment with ddC and 260.7 Ϯ 39.3% after treatment with EtBr.

DISCUSSION
N 4 -hydroxycytidine, a ribonucleoside analog, is currently in late stage, preclinical development as a broad-spectrum antiviral agent. Since a number of nucleoside analogs have been halted at different stages of development due to mitochondrial  toxicity (2,(8)(9)(10), we conducted an investigation to determine the potential for NHC to interact with cellular processes that could lead to mitochondrial toxicity and dysfunction. A previously reported study has concluded that NHC does not cause mitochondrial toxicity based on the observation that NHC has no effect on mitochondrial DNA or RNA levels after a 7-day treatment (14). Here, we report on a more comprehensive series of experiments designed to thoroughly investigate the potential for NHC to cause mitochondrial toxicity over a longer treatment period, at higher concentrations, and in cell lines that are most sensitive to the cytotoxicity of NHC.
The results indicate that mitochondrial impairment by NHC does not substantially contribute to the observed cytotoxicity. In glucose-free media, where a mitochondrial toxicant can have a more significant impact due to the Crabtree effect, no differences were observed in the cytotoxicity CC 50 value compared to the CC 50 observed in glucose-containing media. While NHC-5=-triphosphate does not inhibit DNA-polymerases ␣, ␤, and ␥ at 1,000 M (Table S3), incorporation into cellular RNA could be a cause of observed cytotoxicity. Incorporation of NHC residues into Pol II RNA could be suggested from an elevated level of mutations found in Pol II-transcribed mRNA in the presence of NHC (26). Cellular RNA Pol II possesses a safeguard activity from misincorporation and is able to excise misincorporated nucleotides (27). However, POLRMT lacks this ability (28), and given that NHC is an efficient substrate for POLRMT, it has the potential to disrupt vital mitochondrial functions such as respirationdependent ATP production (8). POLRMT-synthesized short RNAs serve as primers in mitochondrial DNA synthesis by DNA polymerase ␥, and inhibition of mitochondrial RNA synthesis as a result of NHC incorporation could potentially affect the replication of mtDNA (29). NHC caused partial, delayed chain termination after incorporation by POLRMT, which could potentially cause toxicity due to a decrease in the quantity of mitochondrial RNA. On the other hand, high concentrations of competing CTP, UTP, and GTP present in the cell at physiological concentrations ranging from 200 to 600 M (30) may reduce the likelihood that NHC is misincorporated to a level that would affect RNA synthesis. Stuyver et al. showed that mitochondrial mRNA levels were unchanged in HepG2 cells incubated with NHC at 10 M for 7 days (14), suggesting that NHC-TP is not incorporated at the frequency required to reduce mitochondrial RNA synthesis. NHC was incorporated by POLRMT as a cytidine and uridine analog and, to a lesser extent, as a guanosine analog. Incorporation of NHC as both cytidine and uridine analogs may result in the production of mutagenized mitochondrial mRNAs that are either translated less efficiently or translated into mutated and potentially lessfunctional proteins. In fact, mutations in nuclear and mitochondrial mRNA were observed in tissue culture cells that were treated with NHC at 10 M for 3 days; however, there was no increase in the frequency of mitochondrial or nuclear mRNA mutations in ferrets that were given seven doses of NHC twice daily (b.i.d.) at 200 mg/kg/day, despite the fact that plasma levels of NHC were expected to be above 10 M at any time throughout the treatment course (26). It has been shown that NHC was efficiently anabolized to the triphosphate form in primary hepatocytes and in animal organs in vivo (16,26,31). The decrease in quantity of mitochondrial protein or the potential decrease of functional activity of mitochondrial proteins may result in less production of ATP by the mitochondria even if the mitochondrial number has not changed. If this happens, the affected cell switches its ATP production to mitochondrion-independent anaerobic glycolysis, which can be registered by an increase in lactate production. Although, at some concentrations, there were differences between mitochondrial and nuclear protein expression (see Fig. 4A), these differences are not translated into a decrease in number of mitochondria per cell (as measured by mtDNA/cell), or in an increase in lactate production after a long incubation with NHC. In HepG2 cells treated with NHC for 14 days, there was not a significant difference in mtDNA levels or extracellular lactate levels. In CEM cells, relative mtDNA levels were reduced by about 20% at both 10 and 30 M concentrations of NHC. Extracellular lactate levels also increased by less than 20% in the cells incubated with the lower concentration, but there was no change in extracellular lactate levels in cells incubated with the higher concentration of NHC. It has been previously suggested that increases in extracellular lactate of less than 20% are not biologically relevant (32). The registered effects in CEM cells were likely not indicative of mitochondrial toxicity since there was no dose response in relative mtDNA depletion and there was no increase in extracellular lactate at the higher NHC concentration. It has been suggested that there is a spare capacity in mitochondria to tolerate a reduction in respiratory chain complex activity before there is an impact on ATP synthesis or mitochondrial respiration (33). It was shown that the activity of cytochrome c oxidase, a protein in the mitochondrial respiratory chain complex, in mitochondria isolated from rat muscle and brain could be inhibited by 70 to 80% with KCN without any major changes in mitochondrial respiration or ATP synthesis (34,35). It has also been shown that in cells containing transplanted mtDNA that contained stop codons for the COX1 gene in 85% of its mtDNA, COX1 activity was still measured at 70% relative to cells that contained all nonmutant mtDNA (33). It has been theorized that the existing excess of respiratory chain complexes allows for a "threshold effect," providing mitochondria a spare amount of enzymes necessary to function even when some enzymes have been diminished up to a point (36). It has been confirmed that there is an excess capacity of cytochrome c oxidase for respiration in mitochondria (37), which further suggested that this could be the mechanism for the threshold effect.
NHC inhibits the expression of mitochondrial DNA encoded COX1 protein at lower concentrations compared to expression of nuclear DNA encoded SDH-A protein in PC-3 cells. The IC 50 for COX1 protein expression is 2.7-fold lower than the IC 50 for SDH-A protein expression, which suggests that NHC has a stronger impact on COX1 protein expression, possibly due to the absence of proofreading activity for POLRMT. There has been little discussion on the relevance of these differences in publications. On the other hand, the measured CC 50 for NHC in PC-3 cells, which were based on the quantitation of intracellular ATP levels, was determined to be 261.7 M (see Table 1 and Fig. S1), which suggests that treatment with NHC up to ϳ100 M does not inhibit mitochondrial function above the threshold level necessary to affect ATP synthesis. Current in vivo treatment data with NHC also indicate that there is significant tolerability of NHC in animals, at least for short-duration treatments. It has been reported that NHC is well tolerated in mice and guinea pigs that were given b.i.d. oral doses at 800 mg/kg/day for 6 days in mice and for 3 days in guinea pigs (16). NHC was also well tolerated in ferrets given b.i.d. oral doses of NHC at 200 mg/kg/day for 3.5 days (26). Another group has reported that six daily doses by intraperitoneal (i.p.) injection of NHC at 33 mg/kg/day were also well tolerated in mice, though an elevated i.p. dose of NHC at 100 mg/kg/day apparently caused a weight loss, which was reversible immediately after the end of dosing (14). This quick reversibility would imply that there was no permanent damage to mitochondrial function in any organ.
The 5=-isopropyl ester prodrug of N 4 -hydroxycytidine, designated EIDD-2801 (26), is currently under development for the treatment of a broad range of RNA virus infections. The initial focus will be on the treatment of encephalitic alphavirus infections and influenza. Although extensive work in animal models of both infections indicate the duration of treatment will be short, most likely between three to 5 days of dosing, every effort is being made to determine whether mitochondrial toxicity could still impact utility. To that end, the studies reported here were extended for up to 14 days using very high concentrations of NHC. The results strongly suggest that mitochondrial toxicity will not be a dose or duration of treatment limiting issue. Chronic-toxicity studies (28 days) are under way in rats and dogs, and every effort will be made to monitor for toxicities arising from mitochondrial dysfunction.
For HepG2 cells, the medium was changed on day 6 and day 9 to media with fresh drug. CEM cells were passaged by 1:10 dilution on days 6 and 9 in media with fresh drug. On day 14, the cell-free culture medium was collected. Cells were collected and counted using disposable hemocytometers (INCYTO, Covington, GA). Each experiment was repeated three times with each repeat initiated on different days. The lactate concentration in cell-free medium was measured by using an EnzyChrom L-lactate assay kit (BioAssay System, Hayward, CA). Briefly, all of the samples were diluted 20-fold with water, and the lactate concentration was measured according to the manufacturer's instructions. The lactate concentration was expressed as a percentage of the mock-treated controls. The effect of drugs on mitochondrial DNA quantity per cell was measured by comparing the relative mitochondrial genome copy number (mtDNA copy number divided by nuclear DNA copy number) in cells treated with drug versus mocktreated cells. A nuclear DNA target sequence was used that corresponds to the ␤-actin gene (38), and a mitochondrial target sequence was used that corresponds to mitochondrial DNA nucleotide positions 10620 to 10710 (39). The primers and TaqMan probes (Eurofins MWG Operon, Louisville, KY) used for real-time PCR were ordered were as follows. The primers and probe used for the quantification of nuclear DNA (␤-actin gene) were the sense primer 5=-GCGCGGCTACAGCTTCA-3=, the antisense primer 5=-TCTC CTTAATGTCACGCACGAT-3=, and the probe 5=-(FAM)-CACCACGGCCGAGCGGGA-(BHQ)-3=. The primers and probe for the quantification of mtDNA were the forward primer MH533 (5=-ACCCACTCCCTCTTAGC CAATATT-3=), the reverse primer MH534 (5=-GTAGGGCTAGGCCCACCG-3=), and Mito-Probe [5=-(FAM)-CT AGTCTTTGCCGCCTGCGAAGCA-(BHQ)-3=]. The cells were collected after 14 days of drug treatment and pelleted by centrifugation at 8,600 ϫ g for 2 min. The cell pellets were resuspended in PBS, and the total DNA was isolated by using a DNeasy tissue kit (Qiagen, Hilden, Germany). Cellular DNA from mocktreated cells was serially diluted and used to generate corresponding standard curves for determining a relative copy number of the gene targets. qPCR of each sample was performed in triplicate in a 7500 real-time PCR system (Applied Biosystems, Foster City, CA) in a 15-l total reaction volume containing 1ϫ PCR mix (Life Technologies, Carlsbad, CA), 200 nM ␤-actin probe (for nuclear DNA) or Mito-Probe (for mtDNA), 750 nM ␤-actin sense and antisense primers (for nuclear DNA) or MH533/MH534 primers (for mtDNA), and 1.5 l of a purified, serially diluted DNA sample. The mitochondrial and cellular genome copy numbers were calculated based on the standard curves, and the relative mitochondrial copy number/cell was calculated by dividing the measured mitochondrial copy number by the copy number of the ␤-actin gene. Changes in the mitochondrial copy number/cell were expressed as a percentage of the mock-treated control.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.1 MB.