Preprint Molecular Mimicry Between Epstein-Barr Virus and Human Herpesvirus-6 Proteins and Central Nervous System Proteins..., 2025, Almulla, Maes et al

Molecular Mimicry Between Epstein-Barr Virus and Human Herpesvirus-6 Proteins and Central Nervous System Proteins: Implications for T and B Cell Immunogenicity in an In Silico Study.

preprint

Abstract
Background: The Epstein-Barr virus (EBV) and human herpesvirus 6 (HHV-6) are frequently linked to neuropsychiatric illnesses such as multiple sclerosis, depression, and chronic fatigue syndrome/myalgic encephalomyelitis. These viruses may induce autoimmune reactions by molecular mimicry, leading to damage to self-epitopes in the central nervous system (CNS).

Objective: This study seeks to explore the common pentapeptides present in EBV and HHV-6 viral antigens alongside various CNS-related proteins via molecular mimicry. Additionally, it will assess the immunogenicity of these shared pentapeptides in T and B cells.

Method: Sequence alignment was conducted to assess molecular mimicry between 32 EBV and HHV-6 antigens and 10 CNS autoantigens. Protein sequences were obtained from UniProt, structural homology was analyzed using AlphaFold and PyMol, and shared pentapeptides were identified with Alignmentaj. Immunogenicity was assessed via the Immune Epitope Database (IEDB) for potential T- and B-cell activation.

Results: A total of 91 mimicry pentapeptides were identified between viral antigens (EBV and human HHV-6), and CNS proteins. Notably, synapsin (SYN)1 exhibited the highest mimicry, sharing multiple pentapeptides with EBV nuclear antigen (EBNA)1, EBNA6, latent membrane protein (LMP)1, and early antigen diffused (EA-D) and 6 different HHV-6 antigens. Myelin proteins including myelin basic protein, myelin-associated glycoprotein, and myelin-oligodendrocyte glycoprotein also displayed shared pentapeptides with EBV/HHV-6 antigens, indicating potential immune cross-reactivity. EBNA1, EBNA2, EBNA6, LMP1, LMP2, EA-D, and BLLF1 structurally resemble CNS autoantigens and act as immunoreactive epitopes for human T and B cells. Except for EBNA2 and protein U94, all share immunogenic pentapeptide sequences with SYN1.

Conclusion: EBV and HHV-6 antigens mimic CNS proteins, potentially triggering autoimmune responses via T and B cell activation. Shared pentapeptides suggest a link between viral infections and CNS autoimmunity. Further research is needed to clarify molecular mechanisms and explore targeted therapies to mitigate virus-induced neuroinflammation.

https://www.biorxiv.org/content/10.1101/2025.04.02.646883v1

 

What does mimicry actually mean? Does it mean that they different parts can interact with each other? And does that even matter?

 

It would mean that the same antibody binds to both. This is a theory of autoimmune disease that has been around for a while (most recently, in my memory, with some group of endocrinologists claiming that gluten and thyroid tissue exhibit molecular mimicry).

It's already well known in the field of virology that viruses can alter their genomes to be closer to human tissues to evade immune recognition and clearance. The basic idea is that the (healthy) human body constantly produces autoantibodies, but the cells that produce them normally go through a "screening" process that neutralizes them so as to avoid the very thing that happens in autoimmune disease.

There's some interesting work happening on "vaccines" for autoimmune disease that involves trying to exploit this process to essentially (in layman's terminology) tell the immune system to stop freaking out about a particular self-antigen. Viruses that are able to closely mimic human proteins would be able to exploit the same thing to avoid triggering an immune response.

So the theory of autoimmunity goes somewhat in the opposite direction: if a viral protein with high similarity to human tissue does get recognized as a foreign antigen, it might cause those same antibodies to [edit:] be able to bind to similar human tissue, resulting in a new autoimmune disease.

As far as I know, that part of the theory is hardly proven. @Jonathan Edwards would probably have a more insightful answer.

 

Thank you!

Does that imply that the recognition of a human-lookalike virus as a foreign antigen would result in the screening process being overridden in order to be able to produce (or let through) the corresponding antibodies?

 

[edit: removed part of my answer that was just me completely misremembering a part of negative selection so as not to confuse anyone]

Though I know there's definitely other levels of regulation of that "screening" which might be involved, I just haven't read enough about it to know if it could result in what you're describing.

 

The basic idea is that if a microbial protein is sufficiently similar to a host protein then the immune system can make the mistake of mounting a response to the microbial protein that also reacts with self protein. So if the microbial protein has a peptide sequence with letters for amino acids:

AQTGES

and there is a host protein with a peptide

AQGGES

then maybe T cells reacting to AQTGES will help B cells make antibody to AQGGES.

The only problem is that this theory has been around for 60 years and only one serious candidate disease has emerged - Guillain Barré syndrome - and I am a bit sceptical of that as well. Moreover, it doesn't behave like a standard autoimmune disease because it is a one off monphasic illness, not a chronic progressive process.

The idea was invented to explain rheumatic fever, which almost certainly is an immune complex disease that depends on the microbial antigen still being present. And no convincing cross creativity has been found even there.

The real problem with the molecular mimicry theory is that there are millions of microbial antigens that have similar peptides to self that we all meet at some time and we do not get autoimmunity. Moreover, the mimicry theory fails to explain why we see autoimmunity to very specific proteins - IgG Fc, thyroid peroxidase, intrinsic factor, tissue tranglutaminase and the non-protein DNA.

So when Jo Cambridge and I proposed an alternative general mechanism for autoimmunity in 1999 we argued that molecular mimicry really doesn't apply.

The problem with the theory is quite well illustrated by this paper which has an embarrass de richesse in the sense that there are too many peptides. If this was a good theory there would be just one rogue peptide. The fact that there seem to be lots begs more questions than are answered.

Another key factor is that although T cells respond to peptides antibodies respond to shapes of protein surfaces. So T and B cell antigens are completely different things and do not even have to be on the sam molecule in the same response. This is the basis of the idea of antigen co-capture, where the T cell gets presented with a different molecule from the one the B cell is seeing. That was very much the model Jo and I used for working out RA.

 

Cross posted with @jnmaciuch, in agreement.

Yes, the mimicry theory requires the normal veto to be overridden. This to me is one of those explanations of disease by normality. The theory is no theory at all until you have I a theory of why the veto is overriden. That was the basis of our 1999 paper - why you can predict that it could be overriden for certain very specific proteins like IgG Fc, DNA, etc. We made suggestion for about ten autoimmune disease. Some of those turned out wrong but others have been supported - particularly the complex methods of antigen capture likely to generate antinuclear antibodies.

 

@Jonathan Edwards would it be possible to have a response triggered by weak antibody binding to self-antigen from an antibody that was initially generated for a "lookalike" virus that didn't escape? (sorry for the rambly question, couldn't figure out how to phrase it better) I seem to remember discussion of that from an immunology seminar a while ago, but don't know enough about the topic

 

I had thought that the mechanism of mimicry in acute rheumatic fever - the similarity between the epitopes in conserved regions of GAS M protein & myosin and other cardiac antigens - was fairly well stood up; has the theory not stood the test of time?

 

Can you clarify that? I cannot be sure exactly what sequence you are suggesting. Something very interesting that might be what you are saying might occur for reasons I have been thinking about but I think it may be different!! Antibody production is self-fuelling. In a sense you cannot make an antibody to something unless you have already made at least some feeble antibody to it (which may well be an important job of IgM). But things are complicated from there on.

 

I have kept a bit of an eye on rheumatic fever stories for decades and was never convinced in the past. I may have missed something but T cell responses are notoriously easy to misinterpret in vitro. You can get any T cell to respond if you really want to.

But my stronger argument would be that cross reaction with heart cannot explain a disease that has rash, pericarditis and arthritis at the same time. The distribution of lesions is strongly suggestive of small immune complex disease. And once you ave killed the strep these features stop immediately. There is no lingering 'auto reactivity' as far as we know. The chronic nature of the heart problems is due to scarring as far as I can see. Most of it valvular endothelial damage, not cardiomyocytes.

 

The process I have in mind is:
1) Virus infects host, virus has several human-lookalike proteins including viral protein X that resembles human protein Y
2) human-lookalike viral protein X is not lookalike enough to evade recognition, antibodies continue to be produced for X
3) Antibodies for X are still capable of binding to Y, albeit weakly. However, the weak binding is enough to trigger further immune response against cells that produce Y

Do you think this is a possibility, or is the response to antibody binding conditioned on strong antibody binding that is very specific to the "original" antigen-antibody interface?

 

@Jonathan Edwards @jnmaciuch thank you for explaining!

So IgM might have the job of going around and making rough models or moulds/impressions (like pressing a key into putty) (I have no idea about if they make a positive or negative) of the antigens, and then present them to whatever is responsible for making antibodies?

 

The is basically the standard mimicry theory, which has never really been demonstrated as far as I can see, with the possible exception of GB but now that it is clear that a whole range of microbes can precipitate GB the mimicry story wears a bit thin.

If there was mimicry involving several proteins from a microbe, you should get a different clinical syndrome for each protein mimicked and you don't. Similarly, for several microbes, mimicry should be with lots of different host proteins and we don't see lots of clinical pictures.

Basically there are about twenty host proteins to which we produce autoimmune responses commonly. There are about another fifty rare ones. But that is out of thousands of host proteins. This doesn't make sense for mimicry, which should go for just about any protein. There must be special reasons why it is possible to make antibodies to IgG Fc, thyroid proteins, DNA, topoisomerase etc. And there are. For IgG Fc the mechanism was worked out by people like Jacob Native in the 1980s. It is a no-brainer. And DNA is easy with co-capture and TLR involvement, which is now confirmed. The really interesting ones are the sugar residue ones - antibodies to peanut agglutinin receptor in sarcoid and saccharomyces -like glycans in Crohn's.

 

IgM starts things off with a rough and ready binding. But nothing in the immune system can ever instruct. There are no moulds. Everything has to be done by random production and selection.

If the IgM-making B cell binds well enough it is encouraged in a competitive environment in a follicle to mutate its antibody and if the mutated antibody just by chance binds even better the B cell is allowed to switch to making IgG and go on mutating to get even better.

And at the same time loads more new B cells are arriving making IgM and antibodies from B cells already being selected because they are getting good at binding will be increasingly picking up the antigen and getting it brought into follicles for the new cells to sniff out. The antibody of the old cells has to bind to a different end of the antigen for the new ones to see bare antigen to bind to so you end up with antibodies to all different surfaces of the antigen helping each other to mature to high affinity. But that is always done by trial and error, never by taking moulds and making copies. Just as evolution always has to work by trial and error.

 

I obviously have no clue about any of this, but I started thinking about how the screening mechanism works.

Presumably the screening has to contain a list of the protein sequences that are naturally present in the body. And that list probably has to come from the DNA?

What happens if the DNA in a cell mutates enough to cause a physical change but not enough to initiate the regular mechanisms for getting rid of faulty cells? And could such mutations be cause by e.g. a virus? I’m assuming the screening list wouldn’t be updated.

If it’s just one cell it probably won’t cause any problems. But could this happen at a larger scale - and maybe more importantly - in a way that made the cells mutate in the same way so they would be susceptible to the same antibodies?

Is this essentially what happens with transplanted organs?

 

Ah my bad, my (mis)understanding was that molecular mimicry theories always involved some ex-post-facto "override of the veto", as you put it.
[Edit: Nevermind, I answered my own question. I just remembered a detail incorrectly about B cell vs. T cell negative selection.

For some reason I had a vague memory of a discussion of whether affinity for a self antigen could possibly change between negative selection and subsequent antibody production, especially when produced by a viral ‘mimic.’

I can’t find any published source following up on that idea, so either I’m completely misremembering, or the person who gave that talk ended up not finding anything to substantiate that idea enough to publish. Either way, I can safely be ignored.]

Thanks for the clarification!

 

It does come from the DNA to an extent—for example, for T cell negative selection, cells in the thymus are able to actually produce proteins that occur in all other areas of the body. Developing immune cells are subjected to all of them at once, and if they bind to any of these self-antigens, they’re neutralized.

That is why transplanted organs are rejected, but it’s not a phenomenon that’s likely to happen at any large scale endogenously except in the context of tumor growth.

Mutations are happening constantly, but the rate of them occurring is relatively low, and when they do occur, they’re far more likely to have no consequence at all on the final protein (synonymous mutation).

 

It is complicated because of the two recognition systems and the fact that their rules for positive and negative selection are different. If there is T cell bypass - as there is for IgG Fc and DNA - then the B cells can go into positive feedback antibody production for a self antigen once enough antibody is produced - that was the basis of our 1999 paper.

But the false argument is that this needs to be triggered by a foreign antigen like virus. If the self antigen is present in large quantities all the time all you need to start off the process is the random generation of an Ig gene rearrangement that creates an antibody that starts things off. Having another antigen that looks a bit like self does not help. It would only be relevant if it was being recognised by T cells and with T cell bypass that does not apply. Moreover, if T cells could be cheated this way we would all have autoimmunity all the time.

 

T cells use the DNA library thymic cells to screen out all self peptides. But the B cell screening process only depends on the shape of fully folded proteins. Positive and negative responses to proteins are determined by the way they are handled by the immune system already. It is a highly risky strategy but it works if there is cross-checking by the T cells.

Basically B cells work on the basis that if there is no antibody to X already they should not be making antibody to X. But if there is antibody to X and X is around they should make more. They can tell if there is already antibody to X because the antibody picks up X plus complement and complement gives a plus signal. If a B cell that recognises X only sees X without complement it dies.

The point of this system is to have a chain reaction of antibody expansion when a new foreign antigen arrives. As long as there is T cell control it is OK but if there is T cell bypass by antigen co-capture you have a problem. Almost certainly the system has lots of back up check mechanisms for things like IgG FC and DNA because these are uniquely susceptible to co-capture mechanisms.

In the 1980s Jo Cambridge and I puzzled over how all this worked for hours but not enough was known about the way T cells recognise peptide. By about 1990 enough was known but people were miscalculating the role of complement and didn't know much about Toll-like receptors and co-stimulation. By 1996, we had enough information to see how it could all fit into a general theory. Much of that had been laid out already by people like Natvig and Betty Diamond but the importance of the B cell chain reaction seemed new.

When we published the idea for RA everyone assumed we were wrong - including grandees like Av Mitchison. But we had an experiment to do - treat with rituximab - and it proved just how crucial B cells were in a process which at that time was assumed to be mediated by T cells with antibody being irrelevant.