Type I and Type II Interferon Stimulated Genes (ISG)
This write-up is meant to be a follow up and a commentary on our publication titled "Dose-dependent IFN programs in myeloid cells after mRNA and adenovirus COVID-19 vaccination". Opinions are my own.
TLDR:
We identified Type I and Type II ISG clusters, in vivo, capitilizing on novel vaccines of a novel antigen. Recaputulated in vitro.
MX1, MX2, IFI44, IFI44L, RIGI, and TENT5A are your Type I ISG. CD169 is a good cell surface protein alternative.
GBP1 to GBP5, WARS1 are good Type II markers. CD64 is a good cell surface protein option.
You can check the paper for more genes, but these are the core of the matter.
Summary of the study
During the Covid-19 pandemic, 3 vaccines were approved for use in the USA. Two new mRNA vaccines, BNT162b2 (Pfizer) and mRNA-1273 (Moderna); and an adenovirus vaccine Ad26.COV2.S (J&J).
12 donors received 2 doses of one of the mRNAs. 3 donors recevied 1 dose adenovirus-based vaccine.
Every donor was a "responder", judging by the traditional metrics of vaccine responsiveness.
We sequenced longitudinal data from each donor. 5 time points for mRNA, 3 for adenovirus (because it's only a single dose.)
Results & Opinion
The core of the paper is based upon the distinct ISG-states we observed in classical monocytes. In the paper, we talked about 2 states ISG-dim and ISG-high. But I would like to expand those a little bit here by splitting ISG-high into ISG-high-mrna and ISG-high-adeno. I will demonstrate that they are different. We only had 3 donors whom received the adenovirus vaccine, but I believe the observation is solid and important.
But first things first, I discovered these ISG-states via unsupervised clustering. The ISG-dim subset is actually present at low levels at the baseline. You likely live with ~5% of your classical monocytes being in this state. And, it roughly doubles (to ~10%) in one day after the first shot of the mRNA vaccines (BNT162b2 and mRNA-1273). Using publicly available data, I showed that that ratio further increases to a ~30% by day 2. Furthermore, this extends to your nonclassical monocytes and dendritic cells (cDC2 to be more precise) [1].
The plot below shows the numbers. A version of this can be found in the paper. The distributions of the 2 mRNA vaccines don't differ, so I merged them here.
We later on showed that this state is a Type I interferon stimulated state. Or at least, it is recaputulated by Type I interferon stimulation. MX1, MX2, RIGI (Used to be called DDX58) are some of it's best markers. Overall, this all makes sense. You do have some basal level interferon activity, called tonic interferon. We put a number on it, ~5% of your circulating APCs are in this state, although there is donor-level heterogenity.
After the booster shot (of the mRNA vaccines), you get an even stronger ISG response. I refer to this state as ISG-high-mrna, and it shows minimal MX1 upregulation. WARS and GBPs (GBP1, 2, 3...) are some of it's best markers. This is a Type II interferon dominant state and this too makese sense. By the time you receive the booster dose, you have preexisting immunity. Antibodies are circulating in the blood, and memory T cells are ready to flood you with interferon gamma. And that's exactly what we see. We detected circulating interferon gamma after the booster of mRNAs. As a results almost all of your circulating APCs are in this ISG-high state. Additionally, two donors who gave somewhat of a lesser response (in terms of AB titers) after the first shot, also had lower levels of ISG-high-mrna activity.
Now, about the adenovirus vaccine. Ad26.COV2.S caused a strong interferon response right away. I call this ISG-high-adeno.
Similar to ISG-high-mrna, there is a strong interferon gamma factor, which I believe is due to preexisting immunity against the adenovirus vector. However, unlike ISG-high-mrna, the effects of Type I interferon is readily available and strong. This, I think, is again due to adenovirus. Look at the plots of MX1 and GBP1 below, my go to markers for Type I and Type II interferon activity. Do you see how J&J (ISG-high-adeno, the green dots) shows both MX1 and GBP1 activity?
Finally, this simple plot below neatly summarizes where each subset stands.
ISG-low and Medium align. So does ISG-high-adeno and IFNa+IFNg stimulation.
I am not suprised to see ISG-high-mrna seems to have more MX1 expression than IFNg stimulation. ISG-dim-mrna is Type II interferon dominant, but it stands to reason that the second shot induced Type I interferon just like the initial dose did.
IFNa stimulation showing more GBP activity than ISG-dim is the most unexpected part for me. If you have an idea why, please let me know!
One way how all this is usefull is, now we have good genesets for distinguishing Type I and Type II interferon activity. Gene lists on Hallmark or MSigDB and others aren't really usefull for that. They are even misleading at times.
Further thoughts:
The mRNA vaccines are designed to not trigger a seismic innate response. Type I Interferon can slow down the metabolism of civilian cells in an effort to slow down the spread of the infecting virus. Therefore, Type I Interferon, could, potentially interfere with the production of the spike protein and dampen vaccine responses. Along with causing inflammation, swelling, and pain, thus making the individuals hate the vaccine. This design seems to have worked, if you compare it with how much Type I activity the adenovirus causes. Yet Type I ISG response still happens, which might be a good thing. Along with reports of antibody responses correlating with the interferon responses and the role of Type I interferon in T cell immunity, Krawczyk et al. showed that TENT5A expression in macrophages plays a in important role for mRNA vaccine efficacy ^2, which happens to be one of the main Type I interferon genes.
References & resources
The processed data is available on cellXgene portal : link to data collection
Code is available on github : link to github
1 - Arunachalam, Prabhu S., et al. "Systems vaccinology of the BNT162b2 mRNA vaccine in humans." Nature 596.7872 (2021): 410-416.
2 - Krawczyk, Paweł S., et al. "Re-adenylation by TENT5A enhances efficacy of SARS-CoV-2 mRNA vaccines." Nature 641.8064 (2025): 984-992.