Read on Twitter
Very, very excited to share our @TU_Muenchen @DZIF_ pre-print on 'reverse phenotyping' of SARS-CoV-2-specific T cells. A thread. 1/n https://www.medrxiv.org/content/10.1101/2020.12.07.20245274v1
One important difficulty with analyses of antigen-specific T cell responses is that you usually need to stimulate T cells with their antigen to see if they are antigen-reactive. 2/n
But when you do that, you also change the phenotype, so it's basically too late to see which phenotype they *would have had* if they *had not been* stimulated. I like to see this as 'Schrödinger's cat of antigen-specific T cell analyses' 
3/n
3/n
(of course the T cell geeks among you know there's pMHC multimers for CD8 T cells, but antigenic stimulation is still the gold standard for assessing antigen reactivity of CD4 T cells and also for CD8s when the epitope is not so well defined.) 4/n.
we wanted to not only see which T cells are SARS-CoV-2 reactive after re-stimulation, but also characterize their original in vivo phenotype. We took blood from COVID-19 patients, stimulated half of the sample with SARS-CoV-2 spike antigen and left the other half untouched. 5/n
We then performed scRNA seq and saw a cluster of cells that specifically reacted to the stimulus. Are these antigen-reactive T cells? 6/n
Indeed they are. Using the T cell receptor as a natural barcode to link cells belonging to the same clonotype, we could show that some clonotypes reacted in an antigen-specific manner. We also validated these SARS-CoV-2 reactive TCRs by orthotopic TCR replacement via CRISPR. 7/n
We could then look at antigen-reactive or un-reactive clonotypes, and see which phenotypes they have with or without stimulation. This revealed on the one hand phenotypic effects (i.e. biases) you generate through in vitro re-stimulation... 8/n
...and on the other hand allows to precisely characterize the original in vivo phenotype these cells have (or would have had if they had not been stimulated in the other condition). 9/n
We also analyzed respiratory T cells from the same and further patients with COVID-19. Our freaks from @HelmholtzMunich integrated additional data from Berlin, Shenzhen and Chicago, resulting in analysis of 279,663 single cells from 50 patients with different disease states. 10/n
Finding: Signatures of antigen-reactive T cells are generally enriched in the respiratory tract, but it is the signature of in vitro re-stimulated or unstimulated cells which reflects 'hot' or 'cold' environments from patients with or without virus transcript. 11/n.
Speculation: Interaction with viral antigen at the site of infection could explain this. Neatly shows how the power of the @humancellatlas can be leveraged even when no dedicated analyses (in this case of antigen-reactive T cells) are performed in a given data set. 12/n 
It should be mentioned that this can only be a starting point to look at the phenotype of currently and previously activated SARS-CoV-2 reactive T cells. We show the power of the tool, which we can't wait to apply to all kinds of T cell analyses also for other contexts. 13/n.
Fantastic and very fun collaboration with @SchillerLab Dirk Busch @fabian_theis @schubert_benni @davidsebfischer @MeshAnsari09 Karolin Wagner and many more from @TU_Muenchen @HelmholtzMunich @LMU_Muenchen @DZIF_ @humancellatlas Thank you all! 14/n.
