We’re starting to get a clearer picture of how the SARS-CoV-2 coronavirus works when it infects the body, and there are some surprises emerging. This new paper in Cell is an example (here’s a writeup on it at Stat).
We already know the RNA sequence of the virus very well, naturally, and that’s allowing us both to track mutations and to lay out exactly what proteins it forces a cell to make once it gets ahold of the machinery. That post has some background on some of these, and this earlier one mentions a number of them as well, with an eye to existing drugs that might interact with them. There aren’t very many in total – viruses in general are rather stripped-down. Recall that they start off by forcing the expression of a long polypeptide that (with the help of hijacked cellular proteins) starts cleaving itself into many of these necessary viral pieces, an alarmingly compact and efficient “autoloader” mechanism. The limited number of viral proteins means that you can usually assign a clear and necessary function (or more than one) to every one of them – it’s a lot like working at a small startup! There are no associate VPs in charge of facilitation of planning modalities in a viral genome; they are lean and mean.
And that means that they are stark naked examples of evolution in action as well. Recall that the guiding principle of evolution is “Whatever works”. That’s literally it, and nowhere more so than in something as small and as quickly reproducing as a virus. You know that old line about how an oak tree is just an acorn’s way of making more acorns? Viruses live it – the only thing they do is make more virus. There are no added complications for something like feeding behavior, because they don’t eat. There are thus no variations in metabolism, because they don’t have any metabolism. There are no crazy features driven by sexual selection, because they don’t mate. They infect cells and make more virus, and that’s it. And they do it very quickly, over and over. Any change that even slightly assists in infecting cells and making more viral particles that can make more viral particles will be amplified, any change that slightly decreases that efficiency will disappear.
This new paper is concerned with a vital part of that viral business: dealing with the immune defenses of the organisms that they infect. Recall that the human immune system has, broadly speaking, two branches. You have the adaptive part, the one that raises specific neutralizing antibodies and targets T cells at an infection. That one takes a while to get going; there’s a lot to sort through and building up all those targeted weapons doesn’t happen overnight. And you have the innate immune system, which is the “always on” response that recognizes a number of general signs of infection and is ready to act immediately. If that by itself can clear an infection, it certainly will – otherwise it sort of holds the line until the adaptive immune system can range in the artillery and commence firing.
For viruses, the innate immune system is mostly recognizing weirdo RNA species as a sign of infection – these are things that shouldn’t be floating around, and when they show up it sets off the alarm. The receptors that pick these things up (such as the Toll-like receptors, TLRs) set off some serious transcription factor activity, namely NF-kappaB and various interferon regulator factors (IRFs). These head down to the DNA level and alter transcriptional activity, which has a lot of downstream sequels, too: for example, type I and type III interferon proteins (depending on the cell type) start being produced, which in turn set off a list of further interferon-stimulated genes. Over 300 of those are known, so you can see that listing all the effects is a task that gets out of hand really fast, which is a common problem in immunology. These interferons can be secreted to warn neighboring cells, and in addition, a whole list of chemokines are produced and excreted to recruit various types of circulating white blood cells.
Viruses that affect organisms (like us) with such defenses have had plenty of brutal selection pressure, and the pathogens we notice now are the ones that have assembled ways of infecting us anyway. The list of viral countermeasures is a long one – this battle has been going on for a while – and here’s an article that details ten of the most common ones. Many of these come down to hiding as much as possible from the cellular receptors as well as blocking them and their downstream partners with specific viral products. As the article says, this is a sort of arms race, and it has boundary conditions. An immune system varied and powerful enough to immediately wipe out every foreign pathogen could be hard to contain; we have enough problems with autoimmune disease as it stands. And a pathogen that ripped right through its host’s defenses and ran at full speed might well be quickly lethal, which could cut down on the opportunities for spreading. The successful pathogens (looking at the situation from their viewpoint) are the ones that spread in a way that leaves them constant opportunities for growth.
This new paper compares the transcriptional changes that kick in during infection with the current coronavirus to the past SARS and MERS cornaviruses as well as several other (non-corona) respiratory viruses. And it turns out that the SARS-CoV-2 is an unusual one: it manages to block the interfeon-I and III response quite thoroughly, while setting off a larger-than-normal cytokine secretion response. None of the other viruses studied have that profile. If you add IFN-I back to the infected cells in culture, they clear the virus very strongly – the machinery is working, but it’s just not being engaged. Likewise, the overall transcriptional profile of the virus in cells is unique. It’s not that it sets off more changes; in fact, it actually shows fewer transcriptional differences than the other viruses it’s being compared to. But the pattern of genes that are affected is a new one. There’s plenty of expression of a whole list of chemokines, that’s for sure: CCL20, CXCL1, IL-1B, IL-6, CXCL3, CXCL5, CXCL6, CXCL2, CXCL16, and TNF.
These cell culture results carried over quite well to animal models of infection (ferrets, in this case). Looking at nasal epithelial cells from the infected animals over time, the same lack of interferon response and strong cytokine secretion was observed, with what the authors describe as “a unique gene signature enriched for cell death and leukocyte activation“. Compared to influenza A virus infection in the same ferret model, the coronavirus transcriptional response was much less dramatic, but very distinctive. The team was even able to check transcription in human lung tissue (2 post-mortem samples compared to 2 different healthy patients). That’s a very small sample, necessarily, but it showed a very similar profile: no interferon upregulation and plenty of cytokine transcription. They were able to check circulating levels of these in a larger number of patients (24 infected cases versus 24 uninfected controls), and these results were also consistent: they tested negative for interferon, but showed elevation of CXCL9 (which attracts T cells) and CXCL16 (which attracts NK cells), CCL8 and CCL2 (recruiting monocytes and/or macrophages), and CXCL8 (which attracts neutrophils). A sudden oversupply of these cell types might be behind the pathology of the disease, which could be characterized, if these hypotheses are correct, as a uniquely imbalanced response: far too little interferon and far too many cytokines, too early.
I expect we’ll see quite a few other papers in this area; we’ll see if this picture holds up. But it certainly seems consistent with what people have been seeing in the clinic, and it bodes well for the therapies that are aiming to dampen the cytokine response pathways. Does this mean that administration of IFN-I or IFN-III would also be beneficial?