An ACE2 decoy irreversibly inactivates SARS-CoV-2, even antibody-resistant variants


The ACE2 receptor as it appears at the surface of a cell, with SARS-CoV-2 spike proteins attached to it. The yellow layer is the cell membrane. Two identical ACE2 proteins associate with each other and mostly exist outside the cell. They also associate with two identical proteins called B0AT1, which make the whole thing look a lot more stable.

I have to admit it: I am very tired of COVID-19, stories about it, and everything related to it. Enough already! I thoroughly do not relish this topic. We all want to move on, even if COVID apparently does not.

But I do like to see an innovative approach that sticks it to a disease, and that’s what we’ve got here. Researchers from the Dana Farber Cancer Institute, Harvard University, Boston University, Colorado State University, and Massachusetts General Hospital have devised a therapeutic protein that mimics SARS-CoV-2’s point of attack — the ACE2 receptor — not only distracting the virus from binding real ACE2 receptors, but irreversibly inactivating it. Check it out in the open-access December 7 article in Science Advances.

Vaccines are the first line of defense, of course, so stay up to date! I’ll always remember my first one against COVID, at Gillette Stadium (even though I’m not exactly a Patriots fan). The fact that a huge urban community could come together and direct its resources and facilities for the common good like that was really inspirational to me.


But I’ve thought all along that the best second line of defense against this thing is not so much using antibodies, but tricking it into attaching to something it thinks is the ACE2 receptor, a protein that normally can be found sticking up out of the surface of human cells in the lungs, heart, kidneys, and intestine. This study shows the soundness of that approach but then also takes it a step further.

It turns out that when the SARS-CoV-2 virus fully engages with a real ACE2 receptor, its spike protein — which you see sticking out of the virus in all those endless images of it — undergoes an irreversible change that commits it to invading the cell but eliminates its further ability to attach to an ACE2 receptor. The spike protein actually breaks apart into two pieces, and the one that can attach to ACE2 is forever lost:

The virus’ spike protein is made up of two parts, S1 (red) and S2 (gray). S1 is the part that actually attaches to ACE2 (blue). When it does, the spike protein undergoes a large and irreversible change, where S2 elongates and S1 is kicked out. No more S1 means the virus can’t bind ACE2 anymore.

If we could design an ACE2 decoy the right way, then, maybe the virus would attach to it and be fooled into thinking it has attacked a cell, and the spike protein would undergo the same irreversible breakup and wouldn’t be able to go after cells no longer. The more spike proteins we could hit on a viral particle, the lamer it would become:

On the left, an active virus particle with functional spike proteins sticking out. On the right, what the same particle will ideally look like (ie, toast) after we expose it to an ACE2 decoy

Going into this research, this was an aspect that wasn’t clear. Would an ACE2 decoy be able to not only attach to the virus, but also to trick it into inactivating itself in practice?

If so, we’d have a couple of key advantages over antibodies.

Therapeutic antibodies against SARS-CoV-2 are directed at the spike protein, too, and that totally makes sense because that’s what you want to interfere with, so the virus can’t attack your cells. But antibodies attach to the spike protein in whatever random way they end up doing it, not by mimicking ACE2. So while a garden-variety antibody will stick to the virus just fine, it won’t cause this irreversible and disabling change because the virus doesn’t think it’s found an ACE2; it just thinks it has a big piece of gum stuck to its face.

The other thing is, the spike protein evolves quickly, so an antibody that works great on one variant might do poorly with a new variant. We have definitely seen that in practice. Omicron, for example, is pretty resistant to a number of antibodies that worked on older variants like Delta, because Omicron’s spike protein has evolved a lot; it’s got more than 30 mutations in it! So those old antibodies don’t recognize it anymore. Even Paxlovid, a small-molecule antiviral, is losing its grip on newer variants as well.

But no matter how much a virus changes, its spike protein had better keep its ability to stick to human ACE2. If it doesn’t, that virus goes straight into the evolutionary dustbin. So, if we can design a decoy that looks just like ACE2 to the virus and also has good stability and safety in the body, then we’ll have a weapon that works against all variants of SARS-CoV-2, new and old, no matter how much they evolve, and in fact even against other nasty coronaviruses that might arise in the future.

OK then, so what does a good ACE2 decoy need to have? It’s got to…

  1. Look a lot like ACE2 so that all viral variants recognize it and want to go after it
  2. Be free to float around, not be attached to cells like real ACE2
  3. Have a reasonably long life in the body
  4. Be able to penetrate into tissues where the virus may be hiding
  5. Avoid having the blood-altering functions of real ACE2, so we don’t overdo that
  6. Not cause an immune-response apocalypse in the patient

Part 1 is pretty straightforward. We know what ACE2 looks like when it’s in place and working on the surface of a cell (see the main diary picture). So we’ve got to keep the part that sticks to the spike protein intact. The main question is, how much of it should we keep? The authors tried a few different things there, and truthfully that’s a bit of trial and error. But in the end they found that it worked better when they kept more of the ACE2 protein, even the parts that the spike protein doesn’t attach to.

We can actually hit parts 2, 3, and 4 all at once by combining our ACE2 decoy with the bottom half of an antibody (its “Fc” region). That sets up our decoy to behave like a conventional antibody, except with its business end designed by us. Here’s how a natural antibody compares to this “Fc fusion” we’ll make:

Note the doubled-up structure of the Fc region gives us two ACE2 decoys side by side, just as ACE2 appears on a real cell surface. Bonuses!

Like any other antibody, this Fc-fusion decoy will be soluble, it’ll hang around for a while, and it will be able to penetrate most tissues, even the placenta.

Part 5 isn’t too hard. ACE2’s important job, when it’s not being commandeered by viruses, is to modify hormones that regulate blood pressure. So unless there were some benefit to changing that, which has never been demonstrated, we’d rather not play around with it by adding a ton of active ACE2 all over the place. Luckily, all we have to do is change two amino acids in our ACE2 decoy to make it inactive but still keep its ACE2-like structure.

And Part 6, the immune apocalypse! When an antibody sticks to something, its Fc region can attract a rogues’ gallery of cells from the immune system to attack said thing:

At the center, the hapless target. When antibodies (the green Y-shaped things) with active Fc components bind to it, some monsters show up, and it’s in a bit of trouble

But again, in the spirit of not messing around too much, the authors tweaked Fc (once again with two specific amino acid changes) so it doesn’t have that ability. Not to say that’s necessarily the best choice; others have left Fc alone and let it be active in their Fc-fusion designs. The question is, do we want to encourage an inflammatory response in a patient that’s already got lots of inflammation due to COVID-19? So I’d say I agree with the authors here that we should put active Fc on the backburner. Let’s just hobble the virus particles and leave it at that for the time being.

So, how did the ACE2 decoy perform? First of all, in human cells, it neutralized “original” SARS-CoV-2 (WA01/2020) very well, as did a panel of common anti-COVID therapeutic antibodies (sotrovimab, cilgavimab, tixagevimab, casirivimab, and imdevimab). But against Omicron, all those others lost a lot of their potency, as the FDA has also observed and warned about, but the ACE2 decoy actually gained potency. The decoy was shown to attach effectively to Alpha, Beta, Gamma, Delta, Epsilon, and Omicron variants.

It also had a respectable half-life in the blood serum of hamsters of 52 hours. That’s not as good as a real antibody, but it’s not bad either.

And the answer to the other big question — Can the ACE2 decoy cause the spike protein to change irreversibly like real ACE2 can? — was yes. The decoy was shown to cause the S1 and S2 components of the spike protein to break apart from each other, more so at higher doses, and not at all when no decoy was added.

So clinical trials are up-next, and we do have reason for optimism there. As of 2020, there were 13 FDA-approved drugs of the Fc-fusion type, so it’s not like we’re in uncharted waters. We know this approach can be safe and effective, and hopefully this one follows the same trajectory.

As always, I don’t mean to imply that this is the only group studying this or that it’s going to singlehandedly solve all the world’s problems, but I just want to give a flavor of what is going on in this field, what the thinking is, and where it seems to be going.

But if it’s successful, we’ll have an approach that’s no longer subject to the whims of a mutating virus but instead sets a booby trap at the only door the virus can use to get in. And it’ll give us a blueprint to help us be better prepared against future viral pandemics. Silver lining!

Leave a Reply

Your email address will not be published. Required fields are marked *