Balancing between inflammation and regulation, our immune system lives on the sword’s edge. Do adjuvants perturb this balance, contributing to immune response against self? That’s the case being made (by the media) for re-evaluation of adjuvants by the FDA. Let us recall, however, that adjuvants drive inflammation, but are not intrinsically linked to autoimmunity. Rather, they uncover potential cross-reactivity at the B cell and T cell level, to self. Should we stop using adjuvants? Instead, we can use available computational tools to help uncover the potential cross-reactivity, and avoid vaccine components that induce potentially harmful ‘self-responses’. We argue that — better vaccine design, using computational tools, is the solution. Antigens, not adjuvants, deserve higher levels of scrutiny in the process of developing vaccines for human (and animal) use.

New support for this hypothesis was recently published by Sohail Ahmed et al. in Science Translational Medicine, in which they established an immunologic link between influenza vaccine and narcolepsy. To be clear, the immune response was to a specific receptor (the hypocretin receptor) in the brain, and due to cross-reactive antibodies and T cell responses. In fact, T cell cross-reactivity was directly explored and confirmed in this publication. Influenza nucleoprotein (NP) peptide variants were found to bind to the major histocompatibility complex (MHC) allele product that has been tightly associated with narcolepsy development (HLA-DQB1*0602). The generally accepted explanation for the development of narcolepsy, in this case, would be that structural similarities between the hypocretin receptor and influenza H1N1 NP were already present, but the self-reactive T cell and B cell responses were uncovered by the addition of adjuvants to antigens that already had the potential to autoimmune response.

Why, in the first place, do our own T cells respond to self peptides? Like samurai warriors, T cells are extensively trained on ‘self’ in the thymus, but once they move to the periphery, they seek out and destroy infections. Thus, by definition, each T cell receptor does recognize self epitopes. Depending on exposure to re-enforcing T cell epitopes in the periphery, that self-reactive clone may expand and evolve to recognize similar epitopes derived from pathogens, in the periphery.

An evolving hypothesis about the sword’s edge balance between protection from infection and autoimmunity, is that for every T cell that recognizes self and can respond to self in an inflammatory way (protecting us against pathogens that have similar T cell epitopes), there is a matching T cell with identical specificity that can regulate that immune response, suppressing inflammation and protecting “self” from attack. The system swings into attack mode during inflammation, and swings back to quiescence when signals are given that the attack is over. Adjuvants tip that sword’s edge balance, driving these T cells to respond with inflammation, rather than regulation.  

How do T cells recognize ‘self’ and pathogens? They recognize T cell epitopes, short linear strings of amino acids, some of which activate ‘attack’ T cells, while others activate their controllers – regulatory T cells (Tregs). T cell epitopes are now rapidly discoverable by cutting edge computer algorithms.

In addition to finding epitopes, those algorithms can now find cross-reactive epitopes that induce autoimmunity – or immune regulation, by looking at the TCR face of the epitopes. Members of our research group at the iCubed and EpiVax recently made the remarkable observation that Treg epitopes are found in pathogens that have co-evolved with humans. ‘Commensal’ viruses (e.g. Epstein Barr, EBV) carry significantly more of these epitopes than viruses that rapidly emerge, “hit and run” like Ebola and SARS. Different from “attack” epitopes, Treg epitopes have extensive networks of cross-conservation with human proteins on their TCR-face. Epitopes with less TCR-face cross-conservation are associated with autoimmune disease (AID). Pathogens have learned to exploit similarities with the human genome, adopting T cell epitopes that are linked to regulation, so as to suppress human immune responses to the pathogen.

This discovery of TCR networks leads to two obvious and important questions that deserve further exploration: (1) Do these highly TCR-cross-conserved Treg epitopes in self proteins play a key role in self tolerance? and (2) Does more limited TCR-face cross-conservation trigger other adverse events, yet to be as clearly linked to immunization as Narcolepsy? These fundamental questions are directly relevant to human health, and suggests that new tools developed by our team may be of immense value for finding “triggers” for narcolepsy and other autoimmune diseases, as well as “regulators” to suppress those potential problems.

Next – in response to the media quest for an FDA review of adjuvants, one might ask, are adjuvants unsafe? In our opinion, they are not. Inflammation is the culprit in the case of narcolepsy, but inflammation is required for protective immune responses to develop. Adjuvants simply establish the right ‘milieu’ for an activated immune response. What we do think is that vaccine developers should use existing computational tools to scrutinize vaccine antigens, and take cross-reactive T and B cell responses into consideration when developing vaccines.

Lastly, an important message to anti-vaxers. Does the narcolepsy link with H1N1 vaccination mean that individuals should avoid flu vaccines, or vaccines in general, and adjuvants in particular? The answer is no. Vaccines prevent millions of deaths and protect against hospitalizations. However, there are concrete steps that would help avoid the narcolepsy problem, in specific patients, in the future. Factors that would reduce the possibility of narcolepsy following H1N1 vaccination would include avoiding vaccinating patients who have HLA-DQB1*0602 with influenza vaccines that contain NP (NP is found in split and whole influenza vaccines, and is not required for flu vaccine efficacy). Subunit vaccines such as the vaccine made by Protein Sciences in CT, do not carry NP.

The real solution is not to avoid vaccines, or adjuvants. The solution to off target effects is to further refine vaccines, reducing them down to formulations that contain the exact proteins or epitopes that contribute to protective immunity (as we have argued for some time). Eliminating cross-reactive epitopes might also reduce the potential for these adverse, off target effects. Computational tools that make this possible already exist, thanks to the contributions of computational vaccinologists all over the world. Now it is up to vaccine companies to use these tools to make vaccines even better.