Remdesivir increases mtDNA copy number causing mild alterations to oxidative phosphorylation

SARS-CoV-2 causes the severe respiratory disease COVID-19. Remdesivir (RDV) was the first fast-tracked FDA approved treatment drug for COVID-19. RDV acts as an antiviral ribonucleoside (adenosine) analogue that becomes active once it accumulates intracellularly. It then diffuses into the host cell and terminates viral RNA transcription. Previous studies have shown that certain nucleoside analogues unintentionally inhibit mitochondrial RNA or DNA polymerases or cause mutational changes to mitochondrial DNA (mtDNA). These past findings on the mitochondrial toxicity of ribonucleoside analogues motivated us to investigate what effects RDV may have on mitochondrial function. Using in vitro and in vivo rodent models treated with RDV, we observed increases in mtDNA copy number in Mv1Lu cells (35.26% increase ± 11.33%) and liver (100.27% increase ± 32.73%) upon treatment. However, these increases only resulted in mild changes to mitochondrial function. Surprisingly, skeletal muscle and heart were extremely resistant to RDV treatment, tissues that have preferentially been affected by other nucleoside analogues. Although our data suggest that RDV does not greatly impact mitochondrial function, these data are insightful for the treatment of RDV for individuals with mitochondrial disease.

As of December 2022, over half a billion people contracted severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes the upper respiratory disease COVID-19 resulting in approximately, but most likely underestimated, 6.5 million deaths worldwide 1 .Considering the number of infections allowing for the mutagenesis of SARS-CoV-2, treatments for severe disease have been limited 2,3 .Currently, the United States Food and Drug Administration (FDA) has fully approved three medications for severe disease, Veklury™ (remdesivir) and two immunosuppressant therapies, Actemra™ (tocilizumab) and Olumiant™ (baricitinib).The first treatment remdesivir (RDV), a ribonucleoside analogue (adenosine), was first made available under Emergency Use Authorization, May of 2020 4,5 .The FDA directive allowed prescribing physicians to weigh the risks and benefits of this therapeutic treatment for use before its final approval in October 2020.Once approved, physicians in the US were using this ribonucleoside to reduce hospitalization times in efforts to avoid overpopulation and strain on the healthcare system 6 .
RDV was initially in development for the treatment of Ebola virus, but was also shown to be efficacious against SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV), preventing replication and ameliorating disease 7 .Remdesivir is nonspecifically, intracellularly catalyzed into its active form, remdesivirtriphosphate, an analogue of adenosine tri-phosphate 8 .Remdesivir then works by delayed chain termination inhibiting viral RNA synthesis 9,10 .
Mitochondria are of endosymbiotic origin 11 , which uniquely places this organelle at risk for off-target side effects affecting mitochondrial function during both antibiotic and antiviral treatments.Mitochondria contain multiple copies of their own ~ 16.5kB double stranded DNA inside each mitochondrion within the cell.Mitochondrial DNA (mtDNA) encodes for 13 proteins required for oxidative phosphorylation (OXPHOS) and ATP generation for the cell alongwith its own ribosomal and mRNA machinery to transcribe and translate these polypeptides 12 .Mutations, deletions, or mtDNA copy number depletion causes defects in OXPHOS leading to energy failure and tissue dysfunction 13,14 .Antiviral analogs have previously been found to disrupt the function of the mitochondrial DNA-directed RNA polymerase (POLMRT) (transcription, mtDNA replication), the mtDNA polymerase gamma (replication), and thymidine kinase 2 (mtDNA nucleotide recycling) 15 .
Previously, antiviral drugs, specifically nucleoside analogues, demonstrated off-target effects on mitochondrial function.During the US HIV/AIDS epidemic in the 1980 and 1990's, a thymidine analogue designed to treat HIV patients, zidovudine (AZT), was identified, moved into Phase I clinical trials, and was approved by the FDA in a 3-year span from 1984 to 1987 16,17 .AZT was approved in record time with only one 19-week human clinical trial 18 .Undeniably, AZT and newer generation antivirals that came after the development of AZT led the way to better treatment and management strategies for HIV/AIDS, turning a deadly disease into a chronic yet manageable disease.However, research studies years later demonstrated the negative effects AZT exposure had on mitochondrial function 19,20 and was verified to contribute to premature tissue aging and myopathies in AZT-treated patients 21,22 .These effects were also seen with short-term usage impairing the respiratory chain [23][24][25] .AZT has also been shown to cross the placenta in nonhuman primates 26 .Findings in both nonhuman primates and human infants have found AZT may have negative effects on mitochondria and mtDNA causing additional concern for their usage 27,28 .Due to these past findings, we decided to test whether remdesivir inadvertently caused similar off-target effects on mitochondrial function.Here, we report that mild changes to mtDNA occur in response to RDV treatment, but mitochondrial function is largely unperturbed in acute regimens used in this study.

Results
Mv1Lu cells increase mtDNA copy number in response to remdesivir with minimal changes to oxidative phosphorylation.Mv1Lu were chosen to first test whether RDV caused off-target effects in vitro.SARS-CoV-2 uses angiotensin-converting enzyme 2 (ACE2) as an entry receptor into the host cell 29 , which is highly expressed in lung and airway epithelia 30 .These cells are of epithelial origin and have been shown to support the replication of coronaviruses 31 .Mv1Lu cells were treated with either a high (2.5 μM) or low (0.25 μM) dose of remdesivir for 72 h.0.25 μM or 2.5 μM were concentrations that previously showed efficiency in cell culture when clearing SARS-CoV-2 virus 32,33 .While the viability of these cells was unaffected at either dose (Fig. 1a), a significant, slight increase in mtDNA copy number occurred with a high dose of remdesivir with primers targeting the ND6 region (Fig. 1b).However, this result was not recapitulated using another mtDNA primer set against the mtDNA region of the gene ND4 (Fig. 1b).We next set out to test whether mitochondrial function was perturbed by RDV.Oxidative phosphorylation (OXPHOS) protein expression showed mostly no changes except an increase in the nuclear-encoded succinate dehydrogenase [ubiquinone] iron-sulfur (SDHB) subunit of complex II for cells treated with 2.5 μM of RDV (Fig. 1c,e).Other subunits were unaffected, which were also nuclear-encoded, for complexes I (NDUFB8), III (UQCRC2), and V (ATP5A) (Fig. 1c,e).This was also true for the mtDNA-encoded cytochrome c oxidase (COX) 1 subunit (Fig. 1c,e), indicating that changes to mtDNA copy number did not result in a biological effect.Mitochondrial transcription factor A (TFAM), which is responsible for mtDNA replication and transcription, was also unaffected by RDV treatment (Fig. 1d,f).Considering that complex IV activity is sensitive to mtDNA alterations 34,35 , we then tested the activity of citrate synthase (CS) and cytochrome c oxidase (complex IV) finding no effect of remdesivir treatment (Fig. 1g-i).initial reports on RDV also indicated a 10-day treatment period for COVID-19 provided benefit to patients over placebo 6 , so we also chose a 10-day treatment regimen using efficacious doses previously reported in mice 41 .We then measured mtDNA copy number levels using two primer sets targeting the regions encoding the genes for ND1 and COX1 in liver, lung, heart, and skeletal muscle.We found a significant increase in mtDNA copy number 30 days post-treatment in liver using both primer sets (Fig. 2a).However, there were no detectable changes to mtDNA copy number for lung, heart, and skeletal muscle (Fig. 2b-d).With this change in mtDNA copy number, we next decided to profile whether mitochondrial function was altered after treatment in the liver.Surprisingly, western blots probing for oxidative phosphorylation subunits showed no change with RDV treatment compared to controls (Fig. 3a,b), but TFAM protein expression in liver was decreased (Fig. 3c,d).Steady state protein levels can be relatively stable for nuclear and mitochondrial DNA subunits unless a severe mitochondrial defect is detected, so we performed spectrophotometer assays probing CS, an indicator of mitochondrial mass not reliant on OXPHOS, and complex IV activity.CS activity (Fig. 3e) shows a significant increase in activity with RDV treated liver samples, which elevated mtDNA copy number could indicate increased mitochondrial biogenesis to compensate for mitochondrial defects.However, both complex IV activity normalized to protein and the COX/CS ratio remained unaffected by RDV (Fig. 3f-g).
We next profiled liver function to test whether RDV was toxic to the liver, since some changes to mitochondrial function were found.Heightened levels of liver enzymes aspartate transaminase (AST) and alanine transaminase (ALT) in blood serum, as well as the De Ritis ratio are indicative of liver damage.No significant changes in AST, ALT, or the De Ritis ratio were detected in the blood of RDV treated mice (Fig. 3h-i).In line with these findings, no changes in histology of the liver sections were observed (Fig. 3j).Small areas of extramedullary hematopoiesis were seen multifocally within sinusoids of liver section on both groups, which is considered a common incidental finding.
Considering that CS activity was increased as well as mtDNA copy number, we performed next-generation sequencing (NGS) to test whether mutational load was affected after RDV treatment in liver.Mutations in the control region that may affect TFAM binding may explain why TFAM and mtDNA levels did not positively correlate with each other.Sequencing coverage for the mtDNA spanned the whole mitochondrial genome for both groups with a uniform depth of coverage (Fig. 4a).Although the mutational load did not differ between the groups (Fig. 4b), we identified three novel polymorphisms that differed from CD-1 and mouse reference genome publicly available on NCBI.In just one RDV treated mouse, a point mutation, (m.9743C > A) in the ND3 gene region was found that was not present in any of the vehicle samples (Fig. 4c).Overall, RDV did not have the same mutagenic effect as other antiviral ribonucleosides 30 days post treatment.
Lung is unaffected by remdesivir treatment.While no changes were detected in mtDNA copy number in lung (Fig. 2b), we tested whether lung tissue appeared affected by RDV treatment.All of the subunits of oxidative phosphorylation probed for on western blot (Fig. 5a,b) and TFAM expression (Fig. 5a,c) remained unchanged with RDV treatment compared to controls.Pathology of hematoxylin and eosin stained lung tissue sections showed no histological differences as well (Fig. 5d).

Cardiac function is unaffected by remdesivir.
Previous studies have found the nucleoside analogue, AZT, to have a detrimental effect on cardiac and skeletal muscle, but we did not detect any changes to mtDNA copy number in these tissues (Fig. 2c,d).We performed western blotting to probe for subunits of OXPHOS finding no significant changes in either heart (Figure S1a-b) or skeletal muscle (Figure S1c-d) in our treatment groups.
To ensure no change in cardiac function was detected, echocardiograms were measured at three timepoints: 1 day pre-treatment, 1 day post-treatment, and 30 days post-treatment (Fig. 6a).Ejection fraction and fractional shortening did not differ between groups at all the time points tested (Fig. 6b-d).Trichrome staining for histology also shows no cardiac abnormalities in RDV treated mice (Fig. 6e).

Discussion
Remdesivir was originally developed for hepatitis C, repurposed for the Ebola and Marburg viruses, and now is utilized for SARS-CoV-2 5 .Antiviral nucleosides will continue to be recycled and reused against emerging viral threats until researchers generate more targeted combinatory therapies to a specific viral strain.Even then, the fast adaptation and mutational ability of viruses such as SARS-CoV-2 42 will still necessitate the usage of these broad www.nature.com/scientificreports/antiviral treatment strategies.Our data suggests that remdesivir does have the ability to impact mitochondrial DNA and affect mitochondrial function, albeit not to the extent that it greatly impacts tissue physiology under these experimental conditions.Previous in vitro data suggests that mitochondrial RNA polymerase does have the ability to incorporate remdesivir; however, its selectivity for ATP against remdesivir-TP is around 500-fold 43 .Previous studies evaluated whether mitochondrial toxicity occurs in cell culture.Studies using immortalized cell lines found mitochondrial alterations and mtDNA copy number depletion but only at high micromolar concentrations that were cytotoxic 44,45 .However, hiPSC-CMs differentiated into iCMs displayed fragmented mitochondria, depletion of mtDNA encoded RNAs, and defects in respiration and ATP levels when exposed to RDV at 2.5 uM concentrations 41 .The pharmacokinetics of RDV can affect all tissues but brain showing the highest tissue distribution in the liver and kidneys 46 , but side effects reported for RDV such as hypotension and bradycardia have been reported in clinical trials 6,47 .Our data in vivo did not show an effect on the heart (Fig. 2c), but it is possible that time points past 30 days or a longer duration of RDV would have a different effect.Tissue distribution may also explain our findings in liver (Fig. 2a) and future studies should closely evaluate the effect RDV has on the kidneys.Our in vitro work in Mv1Lu cells used concentrations that did not affect cell viability (Fig. 1a) even at micromolar concentrations, and we found only significant effects on mtDNA copy number and SDHB protein levels but those did not significantly impact mitochondrial function.However, it is important to note, that repeated RDV 100 mg therapeutic doses in healthy human subjects do reach micromolar concentrations in the plasma 8 .Previous work has already suggested or shown that SARS-CoV-2/coronaviruses localize to mitochondria in the host cell 48,49 .Other RNA viruses have been shown to affect mtDNA and mitochondrial RNA transcripts 50 .In the case of RNA viruses ZIKA and HSV-1, mitochondrial abnormalities have been found due to the viruses' localization to the mitochondria 51,52 .This raises important issues not addressed in this study or previous studies: In the presence of SARS-CoV-2, will mitochondria be affected and will antivirals compound any dysfunction?It is also possible that with the viral load being reduced with antivirals, they might affect mitochondria to a lesser degree.
In vivo analysis of RDV treatment on mitochondrial function has yet to be extensively profiled.One study treated male, 30-month-old rats for 3 months finding no mtDNA copy number alterations or deletions in heart, kidney, or skeletal muscle 53 .Here we found mtDNA copy number was elevated in liver with concomitant increases in CS activity, but overall liver function appeared unaffected.Our results are largely in agreement; however, the duration between studies differs as well as the follow-up analysis on mitochondrial function and tissue function.We also chose male mice for our study as more men were being hospitalized with SARS-CoV-2 (60.3-39.7%)than women 40 .In addition, mitochondrial function is reported to be higher in multiple tissue and cell types in females as compared to males 54,55 , indicating that female physiology may be more resistant after exposure to nucleoside analogue treatments.One finding that was surprising is that TFAM levels did not positively correlate with mtDNA levels (Figs.2a and 3c,d) considering TFAM is important for mtDNA replication and packaging 56 .NGS sequencing did not find any mutations in the TFAM binding region that could explain this result.We also cannot rule out that although the decrease was significant, it did not reach a threshold required to cause a biological effect.The duration of a more chronic treatment regime could be utilized if RDV is used for other viral infections, but a longer treatment duration was not considered for our study because it did not reflect treatment regimens for COVID-19 38 .Preclinical evaluation is important.Another nucleoside analogue, fialuridine (FIAU), displayed promise in treating chronic hepatitis B infections 57,58 .However, during the Phase II clinical trial, the drug caused hepatic failure, lactic acidosis, and pancreatitis, which was fatal for five of the thirteen enrolled patients 59 .Studies later showed off-target effects included a decrease in mtDNA copy number 60 and the enlargement of mitochondria with abnormal cristae 61 .In conclusion, our data suggests that caution may still be warranted for individuals with mitochondrial disease or mitochondrial defects when choosing RDV as a potential antiviral treatment.

Materials and methods
Cell culture.Mv1Lu cells (ATCC) were grown in high glucose (25 mM) DMEM media (Gibco) with 10% FBS (Gibco), 1% HEPES (Gibco), 1% sodium pyruvate (Sigma-Aldrich), 1% MEM Non-Essential Amino Acids (Gibco) and 1% L-Glutamine (Gibco).Cells were routinely tested for mycoplasma contamination by PCR (Southern Biotech).Western blotting.Homogenized tissue or Mv1Lu cell pellets were resuspended in 1 × RIPA Lysis and Extraction Buffer (Thermo Scientific) with Pierce™ protease and phosphatase inhibitors.Samples incubated on ice for 20 min and then incubated at 4 °C end over end for 20 min.Samples were centrifuged at 16,000 × g for 15 min at 4 °C.Supernatant was collected and protein concentration was determined using the DC™ Protein Assay Kit II (Bio-Rad).

Figure 1 .
Figure 1.Mv1Lu cells increase mtDNA copy number in response to remdesivir with minimal changes to oxidative phosphorylation.(a) Normalized luminescence measurements for viability using CellTiter-Glo for Mv1Lu cells exposed to RDV. n = 4 independent experiments.(b) MtDNA copy number quantification for Mv1Lu cells exposed to RDV using two primer sets against the mtDNA normalized to guide DNA (gDNA) levels.(c) Representative western blots for subunits for oxidative phosphorylation complexes and MTCO1.Vinculin is used as a loading control.Full-length membrane images can be found in FigureS3a-d.(d)Representative western blots for TFAM.Vinculin is used as a loading control.Full-length membrane images can be found in FigureS3e,f.(e) Quantification of western blots in (c) for subunits for oxidative phosphorylation complexes normalized to vinculin.Protein changes in RDV treated cells are normalized to vehicle.(f) Quantification of western blots for TFAM in (d) normalized to vinculin.Protein changes in RDV treated cells are normalized to vehicle.(g,h) Spectrophotometer assays for (g) citrate synthase activity and (h) complex IV activity normalized to protein.(i) Ratio of complex IV activity to citrate synthase activity.Error bars ± SEM. * = p < 0.05; ** = p < 0.01.One dot equals an experimental replicate.

Figure 2 .
Figure 2. MtDNA copy number is increased in the liver of mice treated with remdesivir.(a-d) MtDNA copy number quantification for CD-1 male mice treated with remdesivir for 10 days, 30 days post-treatment, using two primer sets against the mtDNA normalized to gDNA levels.Tissues isolated for analysis were (a) liver (b) lung (c) heart and (d) skeletal muscle.Error bars ± SEM. * = p < 0.05.One dot equals an individual animal.

Figure 3 .
Figure 3. Liver shows mild changes to mitochondrial function when treated with remdesivir.(a) Representative western blots for subunits for oxidative phosphorylation complexes in liver.Vinculin is used as a loading control.Full-length membrane images can be found in Figure S4a-c.(b) Quantification of liver tissue western blots in (a) for subunits for oxidative phosphorylation complexes normalized to vinculin.(c) Representative western blot for TFAM in liver.Vinculin is used as a loading control.Full-length membrane images can be found in Figure S4d,e.(d) Quantification of liver western blots for TFAM in (c) normalized to vinculin.(e-f) Spectrophotometer assays for (e) citrate synthase activity and (f) complex IV activity normalized to protein.(g) Ratio of complex IV activity to citrate synthase activity.(h) Blood serum testing for AST and ALT enzyme levels.(i) De Ritis ratio of AST/ALT levels.(j) Representative hematoxylin and eosin staining images of liver collected 30 days post-treatment.Scale bar = 400 μm for left images.Scale bar = 100 μm for right images.RDV n = 4, vehicle n = 5.Error bars ± SEM. * = p < 0.05; *** = p < 0.001.One dot equals an individual animal.

Figure 4 .
Figure 4. Next-generation sequencing finds no change in mutational load to liver after remdesivir treatment.(a) IGV snapshot of sequencing depth and coverage maps.(b) Quantification of single nucleotide polymorphisms (SNPs) detected per 10 kilo base-pairs per sample.One dot equals an individual animal.(c) Cartoon depiction mapping polymorphisms found in NGS data on the mouse mitochondrial genome.Biorender.comwas used in the generation of this panel.

Figure 5 .
Figure 5. Lung is unaffected by remdesivir treatment.(a) Representative western blots for subunits for oxidative phosphorylation complexes in lung.GAPDH is used as a loading control.Full-length membrane images can be found in Figure S5a-d.(b) Quantification of lung western blots in (a) for subunits for oxidative phosphorylation complexes normalized to GAPDH.(c) Representative western blot for TFAM in lung.GAPDH is used as a loading control.(d) Quantification of lung western blots in (c) for TFAM normalized to GAPDH.(e) Representative hematoxylin and eosin staining images of lung collected 30 days post-treatment.Scale bar = 400 μm for left images.Scale bar = 100 μm for right images.RDV n = 2, vehicle n = 4. Error bars ± SEM. n = 5/group.One dot equals an individual animal.

Figure 6 .
Figure 6.Remdesivir does not cause functional deficits to cardiac function.(a) Timeline diagram showing drug treatment experimental paradigm for echocardiograms performed prior and post-treatment.Biorender.com was used in the generation of this panel.(b,c) Echocardiography showing (b) ejection fraction and (c) fractional shortening.(d) Representative images of echocardiography of each group.Left ventricle inner wall is depicted by yellow dashed lines.(e) Representative trichrome staining images of hearts 30 days post-treatment after echocardiography.Scale bar = 500 μm.Error bars ± SD. n = 10/group.