You’ve likely been hearing a lot about Remdesivir, so I thought I’d discuss it here and tell you about how with coronavirus multiplying it does interfere! 

SARS-Cov-2, the novel coronavirus that causes COVID-19, is a single-stranded RNA virus – this means that it holds its genetic blueprint in a single strand of RNA (DNA’s “cousin”). Every time the virus wants to copy itself (replicate), it has to make a copy of this RNA. Remdesivir prevents this by giving the viral RNA copier (RdRP) fake RNA letters to use. It’s far from a miracle drug, but it seems to help some people a little, whereas no other drugs have, so in May 2020, it got Emergency Use Authorization from the FDA, meaning doctors can give it to hospitalized patients outside of clinical trials. So, what’d they see? And (cuz it’s me) what’s the underlying biochemistry?⠀

Once SARS-CoV-2 docks on our cells using its Spike protein and sneaks, in, it can get our cells to do a lot of its work for it – like making viral proteins based off of that RNA’s instructions – but our cells don’t have the machinery needed to make copies of RNA from RNA. That task requires an RNA-dependent RNA Polymerase (RdRP) and our cells only have DNA-dependent DNA Polymerases (which we use for copying our own genome) and DNA-Dependent DNA Polymerases (which we use to make mRNA copies of genes to make protein from). So, before the virus can replicate, it has to get our cells to make an RdRP using instructions in the viral genome. 

That’s the “easy part” – once its made, the RdRP has a daunting task ahead of it, traveling along the viral RNA and using it as a template to make more RNAs based on the specific base pairing that’s possible between the 4 RNA letters (nucleotides): A to U and C to C. I like to think of it kinda like a train traveling on a half-track and laying down the other half of the track ahead of it as it goes. And RdRP has a fairly long journey ahead of it. Compared to our genome, which is ~3 billion letters long, the SARS-CoV-2 genome is itty bitty (~30,000 letters (30 kilobases)). But compared to other viruses, it’s pretty damn big. Take influenza (the flu) for example – it has a genome that’s ~14 kb, so SARS-CoV-2 is over 2X as big. ⠀

Being big has some perks – you can hold more genetic information so you can do more things. It’s kinda like having a bigger disk storage size on your phone so you can have more apps, but here you’re directing host cells to make viral proteins instead of directing virtual farmers to do whatever they do in Farmville (if that’s still even a thing?) But being big also has challenges. One of these challenges is making sure that all of those 30,000 letters are copied accurately each time the virus replicates. And, this is no small task. The bigger the genome, the bigger the ask. ⠀

Imagine the copier had an (unfixed) error rate of 1 in 1000. That would mean 30 errors per copy. Viral RNA polymerases are generally kinda sloppy, but usually not quite that bad. Their unfixed error rates range from about 1 in 1000 to 1 in 100,000. So a “typical” RNA polymerase that makes about 1 mutation per 10,000 letters is gonna make about 1 mutation per copy of a ~10kb viral genome. But if your genome is 30kb and you have that same error rate, you’re gonna make 3 errors per copy. So SARS-CoV-2 needs to do better. And it does – coronaviruses (in general, I don’t have the exact number for SARS-CoV-2) mess up less per-base-wise. Still not nearly as good as cellular DNA copying, which can be as accurate as 1 in 100 million.  ⠀

The word “mutation” sounds really scary, but small mutations are fairly common (in everything!) – and most of the time these mutations are neutral, sometimes they make the virus “worse” and sometimes they make the virus “better.” A lot of the time, mutations at the genome level don’t even change the protein at all because there are different “spellings” of amino acids (e.g. GCC & GCA both spell the letter alanine (Ala, A)) (we call such non-letter-changing mutations “synonymous mutations”). Even if the mutations *do* change the protein letter, a lot of the times the swapped out letter is really similar so the protein doesn’t change at all (we call these conservative mutations). But sometimes the letters really do matter, so SARS-CoV-2 has to keep its mutating in check.How’s it do it? ⠀

In order to maintain its genome’s integrity, SARS-CoV-2 polymerase has a built-in proofreader. It has a domain (protein “section”) that is able to sense errors and cut them out (thanks to 3’->5’ exonuclease activity, where 3’ and 5’ represent the starting end and ending end of the growing RNA strand). It works kinda like if you’re paving a road ahead of you with gravel and you accidentally put in a big rock and then, as you keep going you step on the rock (ouch!). Then, instead of keeping going you backtrack and remove the rock before continuing on – when bases mispair or when there’s something else “awkward” about them, the protein senses a “bulge” and shifts the end of the strand from the part of the protein where new bases are added to the part of the protein where bad bases are removed. ⠀

From our perspective, this proofreading ability has pros and cons. The pro is that SARS-CoV-2 isn’t mutating very quickly. And this is really good news for vaccine prospects. The high mutation rate of influenza is one of the reasons it’s so hard to make a “universal” flu vaccine – a vaccine might get your body to recognize one strain of the flu, but, thanks to high mutation rates, the strain of the flu that you’re faced with might be a different strain. And a lot of the strain-to-strain differences take place in the surface proteins of the flu virus that typically serve as antigens (the parts of the virus that the immune system recognizes via antibodies). The tendency of flu viruses to acquire such mutations is referred to as “antigen drift” and it’s no gift…⠀ 

So a pro of proofreading is that there’s less concern about antigen drift. And another related pro is that the coronavirus is less likely to quickly acquire mutations that give it resistance to drugs if we ever find them. Such acquired drug resistance can be a major problem with viruses like HIV.⠀

So a couple pros in the proofreading column from a human perspective. Now for a con…⠀

A common strategy for antivirals is nucleoside analogs – “fake RNA letters.” But because SARS-CoV-2 has this proofreader, it’s harder to trick it. It can often tell if we give it these fake letters. Take for example traditional obligate chain-terminating nucleotides. When a polymerase adds one of these it can’t add another letter because that chain-terminator lacks the “latch-on part” (3’ -OH). So it’s the end of the line – UNLESS you can sense that something’s wrong, cut off the bad letter and keep on going. And RdRp can sense a lot of these.⠀

RdRp is a protein – and it’s one of the coronavirus’ “Nonstructural proteins” – actually it’s several “Nsps” working together as a complex.  RdRp’s polymerase chain (nsp12) forms a complex where it gets help from a couple other little proteins – nsp7 & nsp8. Together, this heterotrimer forms an active polymerase complex and associates with nsp14, whose N-terminal domain contains the ExoNuclease (ExoN) domain, which does the erasing. It’s kinda like a pencil – nsp12,7, & 8 form the pencil part. It can write without the eraser, but it can’t fix its mistakes – on its own it makes about as many mistakes as the flu polymerase. nsp14 is the eraser part – it’s not required for the actual writing, but it is needed if you want to fix your mistakes. ⠀

The ExoN domain is able to remove some fake letters, but RdRP has to sense that there’s a problem first – and it gets tricked by some of the “fakes.” These fakes, in general, are more technically called nucleoside/nucleotide analogs (NAs) – nucleoside refers to the RNA base + sugar and nucleotide refers to base + sugar + phosphate. RNA letters use the 5’ phosphate of the incoming letter to link to the 3’ OH of the previous letter’s sugar ( ‘ is pronounced “prime” and it just refers to an “address” on the sugar). The drugs are often given in a “prodrug” form that has to be processed by the cells into the active form. Such processing commonly involves adding on the phosphates (it’s harder to get charged things into cells) and removing protective groups that keep the prodrugs stable and soluble before they get there. ⠀

Different NAs have different “game plans” (Mechanisms Of Action, MoAs) and they can be grouped into a few main types. ⠀

– Obligate chain terminators lack a 3’ OH so there’s no way to add on letters – the HIV drug AZT is one of these⠀

– Non-obligate chain terminators let you add a couple letters but then the polymerase has to stop because the messed up letters made the RNA structure weird – remdesivir is in this class⠀

– lethal mutagenesis – these have a 3’OH, so that’s no problem, but their base part is sneaky and can’t form normal base pairs, so when the virus goes to use the new strand as a template it makes a TON of errors. Not the few errors that let the virus evolve to evade or efforts to combat it. Instead, we’re talking mutation levels that the virus cannot handle. A couple examples of this type of NA are ribavirin & favipiravir⠀

As I mentioned briefly, remdesivir is one of those non-obligate chain terminators. It’s a trickster – it mimics the RNA letter adenosine (A) so well that RdRP adds it to the growing RNA chain as it makes copies AND so well that the virus’ proofreaders don’t cut it out, BUT not well enough the virus can make functional RNA copies from it. The virus goes to replicate, adds remdesivir thinking it’s an RNA letter, then adds a few more letters before getting stuck and “terminating” – likely due to the “extra” parts of remdesivir (compared to normal A) clashing with the RdRP protein as it gets threaded through the exit. ⠀

That above link is to the science-jargony journal article, and Laurel Oldach wrote a great summary of it – as well as other things about remdesivir, like how it has some sturdy bonds to protect it from getting cut up in our cells here  ⠀

Remdesivir was an early star because it was used to treat one of the first known Covid-19 patients in the US – a 35-year-old man who’d developed pneumonia from Covid-19 was given remdesivir and he got better quicker than expected. It was just a single case report but it was a glimmer of hope – maybe a second chance for the patient and for a pushed-aside drug…⠀

Remdesivir, made by Gilead, was initially found in a screen for hepatitis C and then was provided to the CDC & the U.S. Army Medical Research Institute of Infectious Diseases as part of a library of molecules the government agencies could screen for effectiveness against various infection diseases. The government scientists found it was potentially useful against Ebola, but it was put aside after another treatment for Ebola (monoclonal antibodies) worked better. But, since Gilead had taken it pretty far in the testing process, scientists know a lot about its safety profile – and they know how to make it (thought they’d have to scale things up a lot)⠀

Those case studies and anecdotal stories kicked off numerous clinical trials. Such trials are crucial because they compare patients who receive a drug versus those who get a fake drug (placebo) or some other standard of care. The best trials are double-blind, meaning neither the doctor nor the patient knows if the patient’s on the real drug, and randomized, meaning that patients are randomly assigned one or another. ⠀

Remdesivir was in the news again in the spring when esults from an NIH-funded study, published in the New England Journal of Medicine, , led the FDA to issue  Emergency Use Authorization, meaning that doctors can give it to hospitalized patients outside of clinical trials⠀

What’d the study show? In the trial of 1,063 patients (one of those gold-standard double-blind, randomized, placebo-controlled ones), remdesivir shortened the average hospitalization time from 15 to 11 days. And it lowered the death rate a little, from 11.9% for the placebo to 7.1% in the treated group, but it wasn’t enough to reach statistical significance (i.e. the less death could just be a fluke). ⠀

The results were good enough to lead to the initial EUA – which only authorized it for patients with severe COVID-19 (who needed oxygen, etc.). Then, on August 28, the FDA broadened the scope of the EUA to allow “anyone” who is hospitalized and diagnosed with COVID-19, regardless of how severe their symptoms are. 

This expansion was based in part on an open-label multi-center clinical trial (open-label means that patients and doctors both knew they were or weren’t getting the drug) compared patients treated with Remdesivir for 10 days versus 5 days versus no days (i.e. just standard of care). (note that it was more like “up to 10 days” because only about 3/4 of the 5-day group and 1/2 of the 10-day group actually completed the whole course of treatment, largely due to discharge before the time period was over. So At day 11, the 5-day-treated group had a (statistically-significantly) better “odds of improvement” compared to people who didn’t get it. The 10 day treated group did not do statistically significantly better… And even though the 5-day-treated group’s numbers look better on paper, it’s not clear just how helpful it was on the actual human level – the “key points” section of the article sums it up nicely: “Hospitalized patients with moderate COVID-19 randomized to a 5-day course of remdesivir had a statistically significantly better clinical status compared with those randomized to standard care at 11 days after initiation of treatment, but the difference was of uncertain clinical importance.” 

The results certainly aren’t amazing, and remdesivir definitely isn’t a miracle cure. But it’s the best scientists have found so far for patients who are still early-is in the disease course. And people are really looking for anything. ⠀

If we could reduce hospital stays, that could relieve burden on hospitals, so that’s good. But it wasn’t very helpful for really sick patients, which is obviously not good…Further trials of remdesivir are still underway – for example, Gilead is running separate studies in moderately- and severely-ill patients, the results of which should hopefully be coming out soon. ⠀

Another disadvantage of remdesivir is that it has to be given intravenously (through an IV) which makes delivering it to a ton of people problematic. Especially since, because remdesivir prevents the virus from replicating, it might be more effective early on, before the virus really starts to take hold, so it might be more effective in patients who are pre-symptomatic or just mildly symptomatic – and these patients are unlikely to be the ones in hospitals with IVs.  ⠀

Perhaps remdesivir can serve as a starting point for making a more optimized antiviral – maybe a swallowable form? And in the future it might be administered as part of a multi drug cocktail – similar to how HIV is treated with multiple drugs. ⠀

Even if remdesivir has some benefit, will patients even be able to get it? There’s a limited supply (which has become less limited as production ramps up) and Gilead appears to have no problems price gouging it (despite all the taxpayer money that helped make it possible I might add…). The drug is priced at $3,120 per treatment course 

Last note: a lot of drugs are better known by their brand names, so you might get confused when you hear the generic name (like acetaminophen instead of Tylenol). But, for remdesivir, it’s kinda the other way around. Apparently the trade name is Veklury, which I didn’t know until today when I came across it in an article I was skimming and got confused until I read the intro. 

more Covid-19 resources: ⠀

more on topics mentioned (& others) #365DaysOfScience All (with topics listed) 👉  ⠀

Leave a Reply

Your email address will not be published.