Evidence for a direct relationship between impaired coxsackie B virus clearance and subsequent type 1 diabetes
by Adam Burrack, PhD
We return again today to the relationship between viral infection and subsequent autoimmunity. This will be a bit of a deep dive into theory and history of immunology, so grab your pipettes, turn on your flow cytometer, and read on!
The concept of “molecular mimicry” has been debated among immunologists and clinicians for decades. Briefly, the idea is that the immune system “sees” chunks of viruses during an infection – for example by a virus that specifically infects the pancreas – and that these virus-derived proteins are “similar enough” to “self-proteins” – such as proteins derived from beta cells – to incite an immune response against self-proteins (as well as removing – or clearing – the viral infection). What immunologists mean by “similar enough” when referring to these chunks of proteins from the virus and from our own organs are the sequences of amino acids derived from either source that are loaded in the cells of the immune system called antigen-presenting cells and are shown to T cells. This is how T cells communicate with the innate immune system and are activated, through interactions with these cells and local cues including inflammation. In other words, if the molecular mimicry hypothesis is correct, autoimmunity is essentially the result of really bad luck, resulting from the way T cells are activated.
This viral-infection + molecular mimicry hypothesis is attractive, in part, because the viral infection will have caused a large amount of local inflammation, and inflammation is key for “tipping the scales” toward a productive immune response. Inflammation achieves two key bench-marks for starting a T cell response to infection. First, inflammation “matures” the antigen presenting cells to more efficiently activate T cells. In immunology-speak, inflammation increases “co-stimulation” signals (also called “signal 2”, if the chunk of protein shown to T cells is “signal 1”) between the antigen presenting cells and the T cell. Second, inflammation ‘primes’ T cells to be more receptive to cues to respond to the proteins they are currently being shown. In this way inflammation serves as a “signal 3”, telling T cells that they should definitely start a response to what they are currently seeing.
A persistent question in the field – preventing molecular mimicry from wide-spread acceptance as ‘fact’ – have been questions regarding the degree of similarity between viral protein and self-protein required to promote autoimmune responses. In other words, it is not clear whether molecular mimicry requires self-proteins to be released by infections (which wouldn’t really be molecular mimicry, after all, this would be infection leading to excessive damage of tissues), or whether virus-derived proteins actually are similar enough to self-proteins to promote responses. In short, direct evidence for molecular mimicry in human patients with autoimmunity has been lacking.
The broader question – about the immune response in general – raised by molecular mimicry is whether individuals who go on to develop autoimmunity have, in general, hyper-reactive immune systems (ie, they over-respond to infection and autoimmunity is the end result due to collateral damage), or whether these individuals have under-reactive immune systems (ie, they never quite clear viral infections, leaving open the chance for concurrent responses to virus and self, later). Hyper-active immune systems do not necessarily require molecular mimicry to occur: there could be bystander damage to self-tissues which promotes autoimmunity in this setting. If there is damage to self, then the immune system would be seeing both virus and self during the inflammation-causing event, and it would be logical to attack both sources (sort of). On the other hand, under-reactive immune systems (never clearing the infection, so the – ineffective – immune response continues) calls into question whether molecular mimicry is an appropriate term if the virus, in fact, persists long-term. On a per-cell basis, it is critical for the field to address whether the same T cell recognizes both foreign and self-targets – this is the T cell that must be therapeutically addressed – through technology that does not, yet, exist.
A recent publication in Science Reports addresses this question in people with type 1 diabetes. As a take-home message, the authors found that “children who developed early anti-insulin targeted autoimmunity have a compromised humoral immune response in early childhood”. What this means is that children who were not as effective “clearing” a specific type of viral infection in early childhood (coxsackie B virus) were more likely to develop autoimmunity. This suggests that at least some early-onset type 1 diabetic individuals have under-reactive immune systems and that viral infections can persist in these individuals. The authors performed T cell receptor sequencing following stimulation of T cells from these individuals with either pancreas-derived proteins (GAD, a common autoantigen in T1D) or coxsackie B virus proteins. The authors detected T cell receptor sequences that mediated responses to both of these targets. This data is critical. This demonstrates that a single T cell can respond to both the virus and to pancreas-derived targets. So, not only is early-onset T1D associated with poor clearance of a pancreas-specific virus, but single T cells can respond to both virus and pancreas-derived proteins. Presumably, this is due to similarity between the peptides derived from the virus and from the pancreas-specific protein. That’s just bad luck, and nearly impossible statistically, but it appears to be an empirical fact.
Much remains to be ascertained about how this system works. There are several directions work could move forward from this seminal observation. One important next step will be testing whether enhancing the anti-viral T cell response (in mouse models of type 1 diabetes) delays or exacerbates the onset of autoimmunity. Another step could be to specifically remove T cells with the “dual-reactive” T cell receptor (again in mouse models) to test whether likelihood of diabetes onset following CVB infection decreases, or whether other dual-reactive T cells re-emerge over time. Unfortunately for any T cell depletion goal, the thymus produces new T cells in humans until we are at least 20 years old, which means periodic “T cell depletion treatments” would be required. Therefore, a specific depletion treatment for dual-reactive T cells would not be my first choice – especially in people who have not yet developed T1D. The FDA does not generally approve this ‘invasive’ a treatment for someone who has not yet developed the condition being treated.
However, in an adult receiving beta cell replacement therapy (ie transplantation with either isolated islets or whole pancreas), removing these known beta cell-reactive T cells would undoubtedly be useful. This later time-point, islet transplantation, may represent a “more appropriate” time to address this type of “dual-reactive” T cell.