However, how does Remdesivir intefere with human transcription, or does it?
Via https://www.ncbi.nlm.nih.gov/books/NBK548938/:
Nucleoside analogues can cause liver injury by several mechanisms. Most characteristic is a mitochondrial type of hepatic injury that is probably caused by the nucleoside analogue becoming incorporated into or blocking mitochondrial DNA synthesis by the mitochondrial gamma polymerase, leading to a depletion of mitochondria or decrease in their function. Mitochondrial injury can affect multiple tissues thereby causing myopathy, neuropathy, pancreatitis, bone marrow suppression and/or hepatic injury.
Remdesivir trials for treatment of Ebola virus infections included in vitro studies of hepato- or liver toxicity. Via https://www.who.int/ebola/drc-2018/summaries-of-evidence-experimental-therapeutics.pdf?ua=1:
Remdesivir and the parent nucleoside analog GS-441524 were extensively profiled for in vitro cytotoxicity and mitochondrial toxicity in multiple relevant cell types. Both remdesivir and GS-441524 exhibited > 3.5-fold margins in most in vitro toxicity assays. Data from in vitro studies with liver cell culture systems demonstrated that human hepatocytes are susceptible to remdesivir mediated toxicity, likely due to high cellular permeability and effective intracellular metabolism of the drug. Systemic metabolites of remdesivir detected in vivo in plasma do not exhibit any in vitro hepatotoxicity at pharmacologically relevant levels.
However, treatments in vivo did not show much in the way of negative effects:
Single dose of remdesivir IV infusion from 3 to 225 mg was well tolerated with no dose limiting toxicity observed. No treatment emergent AEs were observed in more than 1 subject per arm. No evidence of renal or liver toxicity was observed. All AEs were Grade 1 or 2.
Multiple-dose IV administration of remdesivir 150 mg once-daily for 7 or 14 days was generally well tolerated. No subjects had a Grade 3 or 4 treatment- emergent laboratory abnormality during the study. Reversible Grade 1 or 2 ALT or AST elevations were observed in several subjects without abnormalities in total bilirubin, alkaline phosphatase (ALP), or albumin. There was no abnormality or clinically significant change in international normalized ratio (INR) in any subjects. Remdesivir did not show any effects on renal function in the multiple-dose study.
Stereochemistry of the active portion of remdesivir may help make it less toxic.
Human DNA polymerase selects for or preferentially incorporates right-handed or "D" nucleosides. Nucleosides that are of the left-handed ("L") enantiomeric form will still be incorporated by RNA polymerase and have clinical effect from blocking viral replication, but will less likely be used by the machinery in human cells.
Here is another paper that explores how DNA polymerase selects for the D-enantiomer: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4121797/
Other nucleoside antiviral drugs appear to have L-stereochemistry (ibid.) for this reason, to reduce toxicity, so I'd assume the same about remdesivir, as well. I tried looking through a paper on the organic chemistry of remdesivir synthesis, but I don't know enough about that subject to see whether there was a specific step or steps along the drug's synthesis, which involve separation and purification of enantiomers. Perhaps someone with more expertise there can comment.