To get ahead of variants, Covid-19 drug makers use evolutionary biology as a guide

Before becoming a Covid-19 drug, each candidate was just a tiny fragment of someone’s immune system, part of a swarm of Y-shaped proteins unleashed to try to keep the coronavirus from invading more cells. If the person recovered, these antibodies might end up in a blood sample in a lab. Some proved more effective than others. Yet even as researchers pinpointed the best of the bunch as possible medications, they knew their power could wane: What worked against the coronavirus as it was last year could falter as the pathogen evolved.

That’s starting to play out, in that some monoclonal antibodies now used to treat patients in the U.S. aren’t great at gumming up the machinery of some new SARS-2 variants. But scientists are betting that those same Darwinian patterns that nudged the virus to become less susceptible to certain treatments can be used to our advantage as well, guiding ongoing development efforts.

“It’s a setback, but what I want to be really clear about is that the class of monoclonal antibodies is not lost,” said Nick Cammack, Covid-19 therapeutics lead at the Wellcome Trust. “There will be others, targeting regions of the virus that the virus doesn’t like changing, because it compromises its ability to grow.”

Cammack isn’t just saying that because he is, in his own description, “a treatments guy.” He and everyone else who works on Covid-19 therapeutics are excited about vaccines, too, as a primary tool for controlling the pandemic. The immunity offered by vaccines lasts longer than the temporary virus-fighting boost of monoclonal antibodies, and although the shots have production and storage complexities, each dose is fairly simple to inject, while an antibody infusion involves having the patient come to a specially equipped clinic.

That poses logistical challenges. The people who seem to benefit most are those recently infected and at high risk of becoming seriously ill, because these drugs can significantly reduce their chances of needing hospitalization — but that means identifying and bringing in those who are very contagious but not yet in dire straits. For an overwhelmed health system, the proposition is tricky enough that, over the last few months, many available doses have gone unused.

But there could still be an important role for these synthetic copies of potent natural defenses. They could, for instance, help keep people from progressing to severe Covid-19 while immunizations are being rolled out, or treat infections that appear among those who’ve refused vaccination — or act as a second line of defense in instances where initial vaccines don’t work quite as well as they might’ve because of immunity-skirting variants.

For that, though, these drugs will have to work against the variants themselves — and labs are now identifying and testing new antibodies that might fit the bill, with some in clinical trials, and others showing promise in lab dishes.

The mutations everyone’s been talking about haven’t evolved specifically to evade these medications. They started out randomly, mistakes sneaking into the genome as the virus copied itself to move from cell to cell, person to person. Many were inconsequential, just tiny blips of genetic chatter that didn’t translate into any clinical difference for us. After all, when it was first spreading among humans, you can imagine that its movement was fairly breezy. “It just had a buffet of susceptible hosts ready to go infect,” said Tyler Starr, a postdoctoral fellow studying evolutionary virology at the Fred Hutchinson Cancer Research Center.

Each of those infections is the work of the virus’ spike proteins, which adorn the pathogen’s surface, allowing it to latch onto human cells and to slip its genes inside. Over time, our immune systems tend to find certain bits of those spike proteins to use as toeholds — crags where our antibodies can cling to stop the inner spread. Some parts of the spike protein give those Y-shaped molecules a better grip than others, and because our immune systems zero in on those spots, so did drug developers. Finding those antibodies that were stickiest meant they might not need as high a dose. The idea behind monoclonal antibodies is that you’re borrowing the molecule of someone whose body has had more time to hone defenses against the virus, and giving it to someone earlier on, like a third-grader getting help from a sixth-grade friend on a math problem.

But targeting those specific bits of the spike protein comes with a tradeoff. “The reason those were selected by many companies is that they’re very immunogenic,” explained Amanda Peppercorn, medicine development leader of GlaxoSmithKline’s Covid-19 monoclonal antibody program. “The antibodies that tend to bind to that area, they tend to bind very tightly, so the potency of those monoclonal antibodies is very good.” The downside, she said, is that it’s “a very plastic region of the virus.”

In other words, some of those random copying errors in the virus’ genetic code can tweak those toeholds, making it harder for antibodies to grab on. It’s tough to say with certainty what has driven some newer lineages to outcompete older ones. But being able to evade some of our bodies’ most potent immune defenses could definitely give some strains a leg up. If our immune systems are in fact putting pressure on the virus in this way, you’d expect to see reinfections, which might be clinically mild but would be a sign that the virus is taking “the next-level step to reopen up people,” Starr said — and there are some preliminary data indicating that that might be happening in South Africa and Brazil.

The mutability of those particular spike-protein crags is bad news for the monoclonal antibodies that bind there — both the natural ones, which might’ve helped drive the trend in the first place, and the synthetic ones used as drugs. Take the handful that have been given emergency authorization in the United States. When a team of virologists grew different virus variants in cells in the lab, and then tried out a series of drugs on them, they reported in a preprint that Eli Lilly’s bamlanivimab appears to work well against the B.1.1.7 lineage first identified in the United Kingdom, but is less effective against the B.1.351 lineage first identified in South Africa. The loss of power against that particular variant was also observed for Lilly’s second monoclonal antibody, which this month was authorized for use in combination with the first. Same goes for one of the two antibodies in Regeneron’s cocktail, though its partner molecule currently holds up well against both those variants — part of the reason Christos Kyratsous, the company’s leader of Covid-19 drug research, likes to refer to a cocktail as “an insurance policy.”

That some drugs don’t work well against some viral lineages doesn’t mean they’re now useless. As far as we can tell from the sequencing data we have, “the good news is, today, bamlanivimab still is effective against the majority of the virus in the U.S.,” said Andrew Adams, scientific leader for Lilly’s anti-Covid-19 platform. “That’s going to change over time, because different strains will circulate,” he added. “The thing we’ve learned is, this will be a cyclical war of attrition with the virus. The virus will change, we’ll come forward with new antibodies.”

How companies are choosing those new antibodies has to do with the way those mutations accumulate. At first, researchers rushed to find the most potent molecules and get them into the clinic as fast as possible — but they kept amassing more and more in their libraries, some from people’s blood, some from mice engineered to express human antibodies. The ones they’re eyeing now target more stable toeholds on the virus, which seem to morph less readily from one lineage to another.

“You can imagine that the virus probably has certain areas that perform absolutely vital functions that can’t afford to be changed,” Adams said. His company is working on an antibody, based on one from a patient who recovered from the initial strain of the virus but that seems to work well against newer variants as well. (“The name will be forthcoming soon,” he said. “Hopefully it’ll be easier to pronounce than bamlanivimab.”)

The idea is to aim for a protein bit so helpful that it keeps getting passed down unchanged from one branch in the tree of virus life to another. That’s why Vir and GlaxoSmithKline are currently in clinical trials for an antibody initially found in the blood of someone who recovered not from the new coronavirus, but from the original SARS outbreak, back in 2003, and seems to work against SARS-2 as well. It’s a little less potent than some of the initially identified antibodies, but the companies have tried to make up for that by tweaking the dose and the chemistry that determines how long it lasts in the body.

“The very conserved sites, they are conserved for a reason, so if there is a mutation, often it is so devastating to the virus that it can’t survive. Certainly that was the thinking that went into picking a highly conserved site,” said Peppercorn, of GSK.

Those more stable gripping places could eventually change, too. Starr, for instance, has found protein bits that are the same in both SARS-1 and SARS-2, but appear to have mutated in some distantly related bat viruses. But he imagines that the tactic overall could be a useful one, especially if we have an arsenal of antibodies all aiming for different spots.

“Evolution is something that we cannot necessarily predict,” Kyratsous, of Regeneron, said. By continuing to collect and analyze antibodies, researchers can find those instances in which our bodies keep pace with new variants, with drugs following not too far behind. While the regulatory details are still up in the air, the aim, he explained, is to have well-chosen backups on hand: “If one of our antibodies completely loses efficacy, we will have options to swap it out.”