Let’s dive into some details. But while doing so, I need to note up front that not all of these details are completely known, immunology being what it is. Still, over 25 years of work on the idea of mRNA vaccines have provided a lot of information, which (never forget!) is the only reason that the current vaccines could be developed so quickly. If you remember Tina Turner’s 1986 song “Overnight Sensation”, you’ll have the right idea: after years (decades) of hard work, false starts, and expensive lessons learned, mRNA vaccines for infectious disease were finally ready to come out of nowhere. I’ve linked to this review before (open access), but it’ll give you an idea of how long all this has been in the works.
One thing to note is that such vaccines can have rather different effects if administered through different routes. Here’s a mouse study in 2015 from researchers at Penn and Acuitas (the Vancouver lipid company that’s now partnered with Pfizer and BioNTech for their vaccine), looking at lipid nanoparticle mRNA injected several different ways (intradermally, intraperitoneally, subcutaneously, intramuscularly, and intravenously). They just used the RNA for firefly luciferase as a marker for convenience, because you can inject the mice later on with luciferin (the partner for the enzyme) and just look to see where they light up. For those of you outside the biomedical sciences, I am speaking completely literally.
What they found was that the mRNA injected i.p or i.v. went straight down to the liver, which is not a surprise. You might remember Alnylam, the company developing siRNA therapies – their big success has been targeting rare disorders in the liver, because they’re well aware that that’s where most of their RNA constructs are going to park, anyway. In this mouse model, the mRNA hit the hepatocytes and caused them to make plenty of luciferase, but not for long: at Day 1, the livers were lit up like a used car lot, but by Day 3, everything was gone. At that point, though, there was still some light coming from the sites of injection.
And that’s what you see with the intramuscular dose as well: a good part of the dose goes to the liver, but in that case there’s a substantial effect in the muscle tissue itself, and it’s longer-lasting (out to about a week of protein expression). The subcutaneous and intradermal injections, though, didn’t really show up in the liver at all: the sites of injection light up and keep on going, again, for about a week. This paper didn’t note it, but you would also expect all three of these injections to have some drainage into the lymphatic system as well, which is also important for setting off the immune system. Here’s a 2017 paper from CureVac and Acuitas that demonstrates this in monkeys, though, in the context of experimental vaccines for influenza and for rabies – they could see activation of the innate immune system at the injection site after an intramuscular dose and at the corresponding draining lymph nodes. That’s an important effect with vaccines in general, as you’d imagine, since that’s one of the big sites for antibody maturation and B-cell selection, as mentioned in the last post here. This 2017 paper from the Karolinska Institute, working with GSK, demonstrates this with GSK’s well-known vaccine adjuvant.
And here’s a 2017 paper from the same team at Karolinska along with Moderna, looking at LNP-mRNA influenza vaccines in a primate model as well. They didn’t even bother with intravenous dosing by this point – it’s a comparison between intramuscular and intradermal. It looks like the intradermal dose comes on more quickly in antibody production, but in the end, the two are pretty similar. The paper goes into detail on the rise in the number of germinal centers in the draining lymph nodes, the exact location of all that B-cell selection and the corresponding antibody changes, which argues strongly that the antibody improvements mentioned in yesterday’s post will occur after mRNA vaccination as well. Another 2017 paper from Moderna has more influenza mRNA vaccine data in several animal models, with consistent results – these were the studies leading up to their human trials in this area. In this case, you see protein production at the injection site, in the downstream lymph nodes, and (coming in third), in the spleen (also good news from the immunologic view), and then the liver. All other tissues were much further down the list.
The total amount of protein produced was also estimated in the 2015 paper, and it turns out that the i.v. route actually makes a bit more than the others, shorter duration and all. But overall, you’re almost certainly better going intramuscular – it still has good protein production, and the multiday duration and lymph node involvement are both good for the immune response you’re trying to induce.
Both the Pfizer/BioNTech and Moderna vaccines are being given i.m., so if you’d like to know what parts of your body are producing the coronavirus Spike protein antigen, the answer seems to be the muscle tissue at the site of injection, the lymphatic tissue downstream in your armpit on that side, your spleen, and (for the first day or two) your liver. The bulk of the Spike that you’re going to make is probably made in the first two or three days, anyway, from what we can see from the animal models. My wife was just saying that it’s too bad that we don’t have a luciferase-style readout built in for us humans as well – she’s enthusiastic about being able to watch a green glow coming from under her skin to know that the vaccine is doing its job, but freely admits that this would probably set off a lot of lunatic conspiracy theories as well.
Let me finish off with another paean to experimentation. If you’ve had a chance to look at those papers referenced along the way, you’ll see that many of them make comparisons to other vaccine technologies. This 2018 one from Penn, Duke, Acuitas and others is especially clear on that point. What you can see is that it was already becoming apparent that the mRNA platform had great promise for inducing strong, wide-ranging immune responses – stronger, in fact, than many comparator techniques. The work that had been done over the years on formulations, RNA modifications and other techniques was paying off just in time for the current pandemic. mRNA vaccination was in exactly the right stage for things to take off with a good expectation of success.
It wasn’t always that way. I mentioned the innate immune system before, and without getting too far into the weeds, it has to be noted that getting the balance right between that innate response and the adaptive response is a key for any vaccine technology. You’d like to have the innate system in play, but if you set it off too strongly with either an mRNA or viral vector vaccine, you can actually damage your total antigen production (and the subsequent adaptive antibody response) due to attacks on the vaccine species themselves.
It is not totally obvious how you strike that balance, though: the innate immune system works through a whole army of receptors – a long list of toll-like receptors (TLRs), RIG-I and the other pattern-recognition receptors and proteins, such as MDA5, LGP2 and more. This is all an elaborate sentry system that is watching for weirdo DNA and RNA species as a sign of viral infection, and is ready to bat them down through a whole list of counterattacks. So to get a good RNA vaccine, you frankly have to make it work like a particularly stealthy virus, and not trip every single alarm before your payload gets a chance to enter the cells. But as mentioned, some activation of the innate system is needed to get the adjuvant boost. A lot of that work has to be done empirically, which is why we’ve seen so many RNA and DNA formulation ideas over the years. None of these have been stupid ideas (far from it) but some of them work better than others, and the current lipid nanoparticle ones are the current state of the art – the LNPs themselves activate the innate immune system, but in a way that doesn’t seem to trigger too much of a self-defeating cytokine response. It’s a useful enough effect that they’re being proposed as adjuvant additions to other, more traditional vaccines for that effect alone. This effect had to be discovered by hard work and repeated testing; you’re not going to whiteboard your way past mammalian immunology, not for a long time yet. . .
Update: if you’d like to know more about the supply chain for the lipids and RNA in making these vaccines, here is an excellent place to find out the details.
