I’ve been getting a lot of questions in the last few days about several Spike-protein-related (and vaccine-related) topics, so I thought this would be a good time to go into them. There’s been a recent report about the vascular effects of the Spike protein alone (not coronavirus infection per se), and another presentation on similar effects in lung tissue. These are almost certainly looking at the same phenomena – the lungs are of course full of vascular tissue, and what’s being seen in both cases is very likely mediated by effects on the vascular endothelium.
In the first study, hamsters were injected with a pseudovirus was created that expressed surface Spike protein, while in the second the researchers just injected the protein directly into mice. The pseudovirus team went on to compare endothelial cells with different mutational forms of the ACE2 surface protein (S680D, with increased stability and S680L, with decreased stability). The response to the pseudovirus was quite different in these two, suggesting that it is indeed the binding of the Spike protein to ACE2 that’s a key part of this process. That happens as the coronvirus infects vascular tissue, of course, but this work shows that it’s not the whole process of viral infection that’s responsible for all the trouble: it starts with the initial binding event.
So I’ve been getting questions about what this means for vaccination: if we’re causing people to express Spike protein via mRNA or adenovirus vectors, are we damaging them just as if they’d been infected with coronavirus? Fortunately, the answer definitely seems to be “no” – in fact, the pseudovirus paper notes near the end that the antibody response generated by vaccination against the Spike protein will be beneficial in two ways, against infection and against the Spike-mediated endothelial damage as well. There are several reasons why the situation is different.
Consider what happens when you’re infected by the actual coronavirus. We know now that the huge majority of such infections are spread by inhalation of virus-laden droplets from other infected people, so the route of administration is via the nose and/or lungs, and the cells lining your airway are thus the first ones to get infected. The viral infection process leads at the end to lysis of the the host cell and subsequent dumping of a load of new viral particles – and these get dumped into the cellular neighborhood and into the bloodstream. They then have a clear shot at the endothelial cells lining the airway vasculature, which are the very focus of these two new papers.
Compare this, though, to what happens in vaccination. The injection is intramuscular, not into the bloodstream. That’s why a muscle like the deltoid is preferred, because it’s a good target of thicker muscle tissue without any easily hit veins or arteries at the site of injection. The big surface vein in that region is the cephalic vein, and it’s down along where the deltoid and pectoral muscles meet, not high up in the shoulder. In earlier animal model studies of mRNA vaccines, such administration was clearly preferred over a straight i.v. injection; the effects were much stronger. So the muscle cells around the injection are hit by the vaccine (whether mRNA-containing lipid nanoparticles or adenovirus vectors) while a good portion of the remaining dose is in the intercellular fluid and thus drains through the lymphatic system, not the bloodstream. That’s what you want, since the lymph nodes are a major site of immune response. The draining lymph nodes for the deltoid are going to be the deltoid/pectoral ones where those two muscles meet, and the larger axillary lymph nodes down in the armpit on that side.
Now we get to a key difference: when a cell gets the effect of an mRNA nanoparticle or an adenovirus vector, it of course starts to express the Spike protein. But instead of that being assembled into more infectious viral particles, as would happen in a real coronavirus infection, this protein gets moved up to the surface of the cell, where it stays. That’s where it’s presented to the immune system, as an abnormal intruding protein on a cell surface. The Spike protein is not released to wander freely through the bloodstream by itself, because it has a transmembrane anchor region that (as the name implies) leaves it stuck. That’s how it sits in the virus itself, and it does the same in human cells. See the discussion in this paper on the development of the Moderna vaccine, and the same applies to all the mRNA and vector vaccines that produce the Spike. You certainly don’t have the real-infection situation of Spike-covered viruses washing along everywhere through the circulation. The Spike protein produced by vaccination is not released in a way that it gets to encounter the ACE2 proteins on the surface of other human cells at all: it’s sitting on the surface of muscle and lymphatic cells up in your shoulder, not wandering through your lungs causing trouble.
Some of the vaccine dose is going to make it into the bloodstream, of course. But keep in mind, when the mRNA or adenovirus particles do hit cells outside of the liver or the site of injection, they’re still causing them to express Spike protein anchored on their surfaces, not dumping it into the circulation. Here’s the EMA briefing document for the Pfizer/BioNTech vaccine – on pages 46 and 47, you can read the results of distribution studies. These were done two ways – by using an mRNA for luciferase (and thus looking at the resulting light emission from the various rodent regions!) and by using a radioactive label (which is a more sensitive technique). The great majority of the radioactivty stays in and around the injection site. In the first hours, there’s also some circulating in the plasma. But almost all of that ended up in the liver, and no other tissue was much over 1% of the total. That’s exactly what you’d expect, and what you see with drug dosing in general: your entire blood volume goes sluicing through the liver again and again, because that’s what the liver is for. But when things like this hit the hepatic tissue, they stay there and eventually get chewed up by various destructive enzymes (that’s also a big part of what the liver is for). It’s a one-way ticket.
So the reports of Spike protein trouble are interesting and important for coronavirus infection, but they do not mean that the vaccines themselves are going to cause similar problems. In fact, as mentioned above, the fact that these vaccines are aimed at the Spike means that they’re protective in more ways than we even realized.
Update: there’s another level of difference that I didn’t mention. In the Moderna, Pfizer/BioNTech, J&J, and Novavax vaccines, the Spike protein has some proline mutations introduced to try to hold it in its “prefusion” conformation, rather than the shape it adopts when it binds to ACE2. So that should cut down even more on the ability of the Spike protein produced by these vaccines to bind and produce the effects noted in the recent papers. That comes in particularly handy for the Novavax one, since it’s an injection of Spike protein itself, rather than a vaccine that has it produced inside the cells. Notably, the AstraZeneca/Oxford vaccine is producing wild-type Spike (although that’s still going to be membrane-anchored as discussed above!)