Integrated single-cell analysis unveils diverging immune features of COVID-19, influenza, and other community-acquired pneumonia

This is not specific for the above paper, however, serves to make an important point.

When you want to study SARS-CoV-2: you study the virus, ist near 30kb genome, its 29 proteins. You can alter them, you can cut them out and study them in isolation, it informs about that protein, that virus.
If you want to know about the immune response, you expose model animals, or look at people post infection. This informs about the level of immunity, the robustness, the type of antibodies, the epitopes targeted, the length of protection and waning against infection. All great information.

BUT
If you want to say something about how the virus behaves and want to make a point it behaves different or is in anyway special. Now you need controls. The more the better, but at least some relevant infection controls. Why?

SARS-CoV-2 is a virus, so we know, from many detailed experiments in the past century, it will invoke an immune response. We know, for any pathogen, that antibodies will be made, that T cells will be activated, that a memory compartment will be filled with memory B and T cells specific for this pathogen. This really is immunity 101. 
We also know there is an acute infection phase, that innate immunity will play a very important role at the start: much IFN will be produced, cells will be activated! Oh yes, many cells will be activated, they will be recruited from the bone marrow, into the blood stream, and subsequently into the tissues. Proportions of cells can dramatically change!
Dendritic cells and monocytes will eat pieces of the pathogen, and change their function! Yes, they do change their function; they eat less, become less responsive; their job is now to take the pieces of the pathogen to the lymph nodes.

T and B cells will be tested in the lymph nodes to select those specific for the pathogen pieces. They will be expanded: express many markers that immunologist know well. Some are mistakenly used as "exhaustion markers": like PD-1, TIM-3, LAG-3: they are activation markers, also expressed on so called exhausted cells. But, you can also call cells exhausted if you functionally test them!
T cells move from the blood into the tissues: there may be a drop in total T cells at that moment.

When the pathogen is cleared; T cells die!! Oh, that sounds bad; but it is normal: always happens. You do not need millions of them any more; you cannot store and maintain them. You do keep a few 1000s: memory T cells for a re-encounter. But not all will be immediately back to base! Your immune system is not a tick tock movie of 30 seconds! T cells numbers will keep patrolling the tissues for longer. Monocytes, neutrophils, macrophages, etc will be needed to clean up any debris, and play important roles in tissue repair. Their presence is not something scary!

Now, why this paper? If you would actually use infection controls: you would immediately notice that all the above is normal. Important is to compare similar time points in infection, compare similar types of infection, and be clear if you compare immune naive (any SARS-CoV-2 infection prior to vaccines) or those after a level of immunity has been acquired. Single-cell RNA-sequencing generates enormous amounts of data. What is physiologically important? How many "populations" are really populations, how many are virtual?


This paper show that "The immune response in COVID-19 is characterized by an expansion of CD8 EMRA-like T cells and type I interferon-stimulated NK cells, both demonstrating high cytotoxic potential." That makes perfect sense. Then they properly compare SARS-CoV-2 with Influenza: be careful: kinetics can be a bit difference, SARS-CoV-2 patients are immune naive (prior to June 2020), Important as well; these are hospitalized patients; the authors are aware that this may have shortcomings and may show an aggerated response: they highlight that in the discussion. 

Flu patients will not be. So you expect some differences to be found: more expansion and more robust inflammation in the SARS-CoV-2 cohort compared with a more memory reactivation response in influenza. 
This means in SARS-CoV-2 infection you notice more differences in the naive T cells compartment, stronger activated CD8 T cell response: in the influenza cohort the response is more subdued. The authors very appropriately are careful with their conclusions:

"together, these findings indicate an important role for cytotoxic T cells with EMRA-like features in controlling the virus both in influenza and COVID-19 pneumonia, although it would be informative to contrast these findings with the T cell response in infections that do not require hospitalization."
"Overall, the proportional expansion and activation of CD8 T and NK cells, together with a potent interferon response could be indicative of a homeostatic response to a viral infection. While in this regard it would be of interest to compare patients with different disease outcomes, we deemed the number of patients with a complicated outcome in our cohort too small to justify such an analysis."
"non-classical monocytes preponderated, characterized by enhanced expression of genes involved in MHC-II signaling. This, together with the differential lymphocyte responses discussed above, clearly suggests that the peripheral immune response in CAP depends at least in part on the type of causative pathogen."

No scaremongering, no excaturated conclusions: stating the observations, being careful with the context. This is how it should be done.