It’s my impression that the pace of headlines and tweets, etc. about the many COVID-19 variants has increased recently (and it wasn’t exactly an unexplored topic before). Some of the coverage is just horse-race stuff (here comes this one, around the curve comes that one), but some of it is downright alarmist. And while I’m not here to tell you that everything is peachy, I wanted to add my voice to the folks who are trying to tell everyone not to freak out. Here’s why.
Variants in General
First off, it’s important to remember that variant forms of this virus have been happening constantly, all the way through the pandemic. In fact, it’s a safe bet that every single person who has been infected has had many variant coronavirus forms occur in their own bodies during the course of their infection. The huge, overwhelming number of those possible variations, though, are less competent than the variety that infected a person in the first place, so we never even notice them. They vanish before we even know that they were there, because they don’t have any reproductive advantage (or have outright disadvantages compared to the original type). It’s the same with viruses as it is with us: most point-change mutations are silent, and most of the rest make things either somewhat or greatly worse. Beneficial fitness-enhancing mutations are rare indeed.
But of course, with a cycle time as short as a virus (and a reproduction rate as great as one) any such beneficial mutation (or combination of them) has a chance to stand out pretty quickly against the endless melting-snowflake background of mutations that do nothing or worse. That’s how a variant like B.1.1.7 takes over so quickly. Keep in mind, though, that it’s actually a form with a whole list of changes versus “coronavirus classic”, and that’s surely why it’s taken this long to show up.
Maciej Boni makes that point here. The fitness advantage of the new variants we’re seeing has been built up through several mutations that have piled on top of each other – we are, in fact, seeing evolution at the molecular level going on right in front of our eyes. To that point, this is also a pretty good argument that the coronavirus only recently made the jump into humans – it’s facing a whole new set of selection pressures and heading off into directions that it never was forced to explore in bats, pangolins, or what have you. (Of course, if we had historical examples over a long period in those animals, we’d be able to see what adaptations rose to the top in those species, but they surely wouldn’t be the ones that do the best job against us).
Getting Down to the Tiny Details
There’s a lot of good information out there about all this, but this new paper is a good one-stop-shop. It’s from a large multi-center team in the UK, and it concentrates on the B.1.1.7 variant. That one, the B.1.351, and the P.1 variants all have multiple mutations , with several of them in the Spike protein and indeed in its “business end”, the receptor-binding domain (RBD) that interacts with the ACE-2 protein on the surface of human cells. The N501Y mutation is common to all three of them, and that is a very strong indication that it by itself provides some fitness advantage – you have it as part of three expanding variants in completely different parts of the world, each with their own suite of other mutations.
We’ll look at that one in detail, but first, it’s worth thinking about where these multiple-mutation lines come from. As Maciej Boni says, hitting any beneficial point mutation is like winning the lottery in any genetic sequence, so coming up with a whole list of them means that you have to win the lottery several times over. The only way you’re going to do that in a few months is with something with as fast and numerous a turnover as a virus (bacteria can pull it off, too, for the same reasons). The most likely way that these things happen is for the whole process to take place inside the body of a single patient – one with a grinding long-term coronavirus infection that their immune system is having trouble clearing. This report from December of an immune-compromised patient is exactly that situation: this unfortunate person suffered from at least 154 days of coronavirus before dying from it and his underlying conditions. Sequencing of the virus at different points during this infection shows unmistakeable evidence of its evolution under these forcing conditions – constant immune system attack, but never quite enough to kill the infection off. Those are just the conditions you would use in a lab to generate resistant forms of a virus or bacterium (and it’s just what is done to probe the weak points of a potential new antiviral or antibiotic drug). It’s entirely plausible that these multiple-mutation variants are coming (at least partially) from such patients. In the case of B.1.1.7, it looks like strains with the N501Y mutation and others were already out there last fall, but the addition of a further deletion mutation at residues 69-70 pushed it along a bit further, and that’s the form we’re seeing today.
So what it is about N501Y? The illustration above is from the current Cell paper, and shows (at the bottom of each pane) the N501 form (where the residue is asparagine) and the Y501 form where it’s been mutated to tyrosine. It appears that the new tyrosine of the coronavirus RBD has a chance to form better interactions with two of the amino acid side chains of the human ACE2 receptor, the tyrosine at 41 and the lysine at position 53. You don’t get much more zoomed-in than this – this, in fact, is the level that medicinal chemistry works at (or tries to work at!) all the time. All of our drug potency and selectivity advances, all of our screening hits and promising candidates. . .in the end, they come down to single side chains, functional groups, sometimes single atoms on those groups making slightly better or slightly worse contacts with some protein target. That slight increase in stickiness, the slightly slower off-rate afforded by the interactions with the pi-electrons of the tyrosine ring and the polarity of its phenolic OH group: these are the sorts of things that can alter the course of the pandemic. In fact, that off-rate effect is explicitly demonstrated in this paper through binding experiments using an SPR assay (surface plasmon resonance, the gold standard for this sort of thing), and these show clearly that the tyrosine form parts with the ACE2 protein more reluctantly.
At this level, you’re getting close to what happens when you zoom in so closely on an image that you start seeing the pixels (or the film grain, if you’re around my age or older!) These sorts of molecular interactions are the pixels of medicinal chemistry and of biochemistry in general, added up in their billions and summed up every instant for their positive and negative effects. And if you start to think about how any of those work by themselves, you end up talking about quantum mechanics and the behavior of the actual electrons around the individual atoms and bonds, and that really is a glimpse of the pixel level of reality itself. What a shock it was, about a hundred years ago or a bit more, when it started dawning on people that the universe did indeed have such a granular texture, both in terms of its matter and in terms of its forms of energy, and that indeed the two of those were actually just two sides of the same (almost incomprehensible) coin. It’s worth the occasional reminder that everything, pandemics included, comes down to such things if you pay close enough attention.
In this mutation, the advantage of N501Y is probably not only that it gives slightly better binding to the human cellular target, but that it simultaneously gives slightly worse binding to several types of antibody that were raised to a form of the virus that had an asparagine there rather than that newfangled tyrosine. This change is going to be through the same sorts of interactions, and in this fallen world one should not be surprised to see sins both of omission and of commission. My old biochemistry professor, Bob Schideler (who was indeed old, even in 1982) referred to functional groups on molecules as “peculiar localizations of charge”, and there are surely favorable interactions with the electrons around the asparagine side chain that are no longer there after the switch to tyrosine, and quite likely new unfavorable ones that appear as well (one cloud of electron density banging into another in a negative-charges-repel manner, for example). The Cell paper demonstrates these changes against a large panel of such antibodies, and does a thorough job of showing the binding landscape. It seems very likely that this two-for-one sale (better binding to the target, worse binding to the antibodies that are trying to shut things down) is what’s led to N501Y being a component of so many new strains.
Variants Out in the Real World
But here’s the good news from this paper: there are a lot of neutralizing antibodies out there (present in people both after infection and after vaccination) and not all of them are affected. Even though the paper shows (as does work from many other labs) that overall neutralization is decreased with serum from recovered patients or from vaccinated ones, there is plenty left: “Though much is taken, much abides“. The paper notes explicitly that with B.1.1.7 there is no evidence of vaccine escape. (This is versus the AstraZeneca/Oxford vaccine, and the same statement holds for all the others).
That’s not to say that there can’t be such escape, in some future variant, as I was saying here the other day. It may be that the B.1.351 variant is a step towards this – but on the other hand, it may have gone down a cul-de-sac from which it will have trouble stumbling into further useful changes. That sort of thing happens constantly. The landscape of possible mutations is incomprehensibly vast, and their effects on binding (both to the viral target (ACE-2) and to the antibodies we’re raising against them) is similarly beyond calculation. But we know from observation and from our own experiments in molecular evolution that there are far more blind alleys than ways out of any particular maze.
The key is for us not to find out the hard way. Every single person that gets infected is another chance for the coronavirus to try its mutational landscape against another challenge, and as the B.1.17, B.1.351, and P.1 variants all spread, they’ll be trying their luck against us too. Relentlessly. And remember, those are just the ones we know about. The Cell paper above refers to global sequencing and surveillance for this sort of thing as “wholly inadequate”, and sadly, the authors are right about that.
I was quoting Angela Rasmussen on Twitter the other day, to the effect that if we’re tired of the virus winning evolutionary Powerball drawings, then we should damn well stop selling it so many tickets. We do that by all that stuff everyone’s tired of hearing about, about masks and indoor gatherings and all the rest of it, and also by rolling out effective vaccines as quickly as possible. I am very, very glad to see the way that the case numbers (and hospitalizations) are going down across the US and in many other regions of the world. As best I (or anyone) can tell, this seems to be due to several factors working at once – I would not have predicted such a steep decline, and anyone who tells you that they absolutely saw it coming should be regarded with suspicion and invited to forecast where we go from here. But let’s take it and run with it. The harder these numbers drop, the fewer lottery tickets we sell to the coronavirus, and the better the chance we have to keep beating it down with further vaccination. I would love for us, as in “humanity” us, to actually get out ahead of this virus for once.