The short version: The existence of cross-reactive T cells in some patients from prior coronavirus infections has no clear meaning regarding the severity of their COVID-19. They are not sufficient to conclude that the individuals in question are protected, and in fact the possibility exists that they make disease worse. Further research is needed to determine the significance of these cross-reactive T cells, and the idea that the existence of these cells in some individuals can substitute for herd immunity is dangerous and unjustified.

Lately some people have suggested that we do not need a vaccine for COVID-19 because of “T cell immunity.” In some cases this takes the form of the “natural immunity” (disease-acquired immunity) gambit, that thresholds for herd immunity are somehow drastically lowered because of the presence of T cell immunity, that the virus isn’t a big deal at all because of emerging findings regarding crossreactivity of T cell epitopes- basically a collection of soundbites which essentially feel designed to give me a migraine.

T cells and the Immune System

Anyone who has taken a course on the immune system could be forgiven for thinking that it’s a class about the biology of T cells with most of the other actors being ceremonial. Immunology as a field is extremely T cell centric, to a degree that one could argue has resulted in inadequate attention to other critical actors in the immune response, but fortunately it has endowed us with an incredible understanding of T cell biology (and many more questions than answers still, as is the hydra-like nature of scientific inquiry). Let’s focus on T cell memory here.

With respect to the immune system, memory refers to the ability for the immune system to respond both more rapidly and more robustly to threats it has previously encountered. This is a major principle in guiding vaccination and in part why it may frequently be necessary to get more than 1 dose of a vaccine for optimal protection. However, contrary to what the layperson may expect, memory doesn’t function on the level of the entire pathogen i.e. the cells of the immune system don’t see measles virus and think “Aha! I’ve seen that one before!” The real picture is much more biochemistry heavy and less anthropomorphized. The truth is the immune system can only direct responses against specific parts of antigens known as epitopes. Each antibody recognizes a certain epitope, each T cell receptor (more on this a bit later) is specific to a particular epitope. Most of the time we concern ourselves with protein antigens, so epitopes in this case refer to short stretches of amino acid sequences.

Antibodies get all the attention so let’s discuss them for a moment. Antibodies work by recognizing short fragments of folded up proteins, zeroing in on specific amino acids. The part of the antibody responsible for this is known as the idiotope. Antibodies are great but there’s a key limitation here. Generally, antibodies are said to recognize conformational epitopes. When proteins are made, they are not just a linear chain of amino acids, but rather they are driven to fold and assume very complex shapes. As a consequence of this, antibodies recognize sequences that may not ordinarily end up linearly associated with each other (though sometimes they can depending on how the protein ends up folded). The issue of course is that antibodies cannot recognize epitopes that are buried in the inside of the protein. However, that’s generally okay. Antibodies only need to recognize some part of the antigen to perform their effector functions and where that part is is less important. But consider, for instance, a viral infection. Are antibodies going to be enough? In general antibodies work outside of cells, and that means any proteins they recognize must be secreted (like those that are made by the virus as it exits the cell). Proteins destined for secretion undergo processing inside the cell which involves adding complex trees of carbohydrates to them. As it turns out, these can obscure epitopes that antibodies have been designed to recognize though, and viruses can mutate so that their amino acids end up with increasingly complicated carbohydrate trees that hide the epitopes. HIV is particularly good at this process, known as glycan shielding (though it’s by no means the only virus that’s made use of this clever trick). So in essence, we have had to evolve another way to recognize epitopes in conjunction with this to ensure adequate host defense. Fortunately, we have done this with T cells.

You probably aren’t very familiar with the T cell receptor. It doesn’t have the publicity that its cousin, the antibody, does. But it’s one of the most important tools in the immune system’s arsenal. Like the antibody, there are billions of different T cell receptors designed to recognize epitopes, but they are intended to recognize linear epitopes:

In essence, antigen presenting cells will take up antigen and process it, placing short stretches of the proteins onto major histocompatibility (MHC) complex proteins. T cell receptors recognize the complex of MHC proteins and their peptides, and can become activated when supplied with some additional cues. All nucleated cells (every cell except for red blood cells) have MHC class I proteins. MHC Class I ordinarily present self-antigen at all times, which T cells are not supposed to (strongly) respond to, but will present foreign antigen if it is present inside the cell. Specialized antigen presenting cells may also have MHC Class II proteins, which also present linear epitopes, but they can present longer linear epitopes. Killer T cells express a protein called CD8 which is helps recognize MHC Class I proteins specifically. Helper T cells express CD4 which is helps recognize MHC Class II. Because of this, T cells can recognize linear epitopes derived from intracellular pathogens, like viruses, and recruit other machinery to destroy infected cells or kill the cells themselves (in the case of the killer T cells).

Crossreactivity (Heterologous Immunity)

Crossreactivity (probably better called heterologous immunity in this context) refers to the ability to respond to multiple antigens by the ability to recognize a single epitope. T cells are fairly liberal with their recognition of epitopes to a certain degree. They may often recognize epitopes that are not exact matches for the one they are primed against as long as core aspects are similar enough e.g. aspartate and glutamate differ by a single methylene group (CH2) but both are acidic amino acids, so if one is used in place of the other, T cells may respond anyway. There are often evolutionary constraints for which type of amino acid may be used at a given position so this often works quite well. For example, we know that with measles infection, the virus has a very hard time escaping from antibodies that target the H protein used to bind cells because even though like other RNA viruses it readily mutates, mutation at key positions also abolishes its ability to infect cells. Heterologous immunity is perhaps best illustrated with the case of cowpox and smallpox. Cowpox and smallpox are caused by different, but related viruses. However, smallpox has a case fatality ratio of about 30% (though it depends on the kind of smallpox with hemorrhagic being fatal in as many as 95% of cases) while cowpox causes a self-limiting illness in humans. But the epitopes from cowpox are similar enough that one could be protected from smallpox after a cowpox infection, which was the basis for the first smallpox vaccine. This principle is also applied with most vaccines to a certain extent. Live attenuated measles virus is dramatically different from wild type strains, even having different tropism, preferring cells expressing CD46 rather than CD150 (SLAM). Alright so, problem solved right? People have some level of heterologous immunity from common cold coronaviruses, so now they’re protected and everything is fine and the pandemic is over?

Well… immunology is very very complicated (please, for the sake of my sanity, read this article before trying to be an armchair immunologist).

Firstly, why don’t I let Professor Shane Crotty, T cell immunologist and virology expert, tell you what the findings about T cell crossreactivity in COVID-19 mean, since he is the principal investigator for one of the labs that authored the study:

Thread here.

Note that he is very explicit that any possible immunological benefit from crossreactivity is speculative at this time. It is very tempting to assume that the presence of cross reactive T cells in some patients is protective and several studies do demonstrate that patients who have the best outcomes are able to mount rapid and potent T cell responses against the virus. It is, however, entirely possible that this has no actual meaning or correlation with the severity of COVID-19, which Crotty discusses. Ultimately however, the implications of T cell crossreactivity for COVID-19 are not well understood at this time.

There is another darker side to T cell crossreactivity though: immunopathology. T cell mediated immunity (specifically that of killer T cells) revolves around killing infected cells. That’s why there’s ordinarily a high bar for inducing the action of cytotoxic T cells and breaking so called “cross tolerance” but when you already have CD8 memory T cells that have previously been licensed, the bar for setting them off is lower because that epitope is considered safe to respond to based on prior experience. But there are times where this has gone very awry indeed. This was most robustly demonstrated in a mouse model where the mice had recovered and gained immunity to lymphocytic choriomeningitis virus (LCMV) which were then challenged by vaccinia virus (VV). The outcome was extensive necrosis of the visceral fat as part of the immune response (though notably LCMV infection conferred heterologous protection in these mice against several other kinds of viruses).

CD4 T cells can also cause a lot of problems. For example, some patients who received the killed measles vaccine developed a condition called atypical measles syndrome when they encountered the virus causing fever, pneumonitis, pleural effusions, and edema. When this was discovered the vaccine was swiftly taken off the market and this was chalked up to the absence of antibodies that inhibited viral fusion. Except then the issue was reexamined by Griffin et al in 1999 using a macaque model. The results? Measles anti-fusion antibody was produced earlier compared with control macaques, but… CD4 T cells initiated a nonprotective Th2 immune response:Two of the five monkeys immunized with inactivated vaccine developed a petechial rash characteristic of atypical measles over the extremities and lower abdomen (Fig. 2a) and one of three assessed by repeated chest X-rays developed pneumonitis 9 days after challenge (Fig. 3b). This incidence of clinically apparent disease was similar to that reported for atypical measles7,11,12,24,31

Our data show that both immune complex deposition and eosinophil infiltration characterize atypical measles. This is neither a classic Arthus reaction, in which the immune complex-mediated pathology is characterized by infiltration of neutrophils, not eosinophils, nor a classic immediate hypersensitivity reaction. The immunopathology of atypical measles is most suggestive of an anamnestic nonprotective antibody response, leading to deposition of IgG immune complexes and complement activation combined with an exaggerated production of type 2 cytokines, leading to eosinophilia and prolonged increases in IgE. This response is primed by an inactivated vaccine that produces only a transient protective antibody response.

Of course, there is the old adage: mice lie, monkeys exaggerate. What about humans? We have a case study in that too: the RSV vaccine- or more precisely, the candidates that never made it to licensure:

In summary, ERD [(enhanced respiratory disease)] pathogenesis is associated with Th2 polarization of the immune response in the lungs after RSV challenge. RSV vaccines eliciting high levels of IL-4 and/or IL-13 in animal models (compared to the levels in control animals protected by prior wild-type [wt] RSV infection) should be considered prone to priming for ERD and excluded as potential candidates for infant immunization.

Here though there is a small wrinkle:

Formaldehyde, used for virus inactivation in FIRSV [(formaldehyde-inactivated respiratory syncytial virus)], may have contributed to Th2 polarization during ERD by generating carbonyl groups on viral antigens (96).

The killed measles vaccine was also formaldehyde inactivated, so perhaps this is a factor as well.

I think though the point is: we don’t know what these things mean yet. No one does. No one can exclude the possibility right now (at the time of writing this) that the presence of these crossreactive T cells actually worsen pathology. Or that they have any effect in modifying disease at all. So the idea that T cell crossreactivity is going to save us all from COVID-19 and grant herd immunity is at best completely sensational and at worst a blatant lie.