Molecular mimicry in type 1 diabetes

Molecular mimicry is the concept according to which an external agent (a virus, bacterium or parasite) can carry on its surface short protein fragments (called epitopes). These fragments resemble certain regions of the proteins in human tissues. The immune system recognizes invaders partly by reading these short fragments. Sometimes a microbial epitope and an epitope of the body's own proteins share a sufficiently strong resemblance. The antibodies and T lymphocytes activated against the microbe can then mistakenly attack the body's own cells too — a form of "friendly fire" [1].

The mechanism begins after the acute infection, when the immune system produces clones of T and B lymphocytes specific against the aggressor microbe. If the T-lymphocyte receptor or the antibodies of the B lymphocytes also recognize a fragment of one of the body's own proteins, the attack becomes autoimmune. In essence, the microbe "wears a mask" that resembles one of the body's own proteins. The immune system can then no longer correctly distinguish what belongs to it from what is foreign. Even so, in the development of type 1 diabetes, molecular mimicry is a plausible but not universal mechanism, as the figure below shows [1].

The body has two safety levels that keep the immune system from attacking its own cells. The first is central tolerance, in the thymus, where T lymphocytes "go to school" in childhood. There, under the control of the AIRE protein (Autoimmune Regulator), the epithelial cells of the thymus present to the young T lymphocytes thousands of the body's own proteins, including insulin and other beta-cell antigens. The T lymphocytes that react to the body's own proteins are then eliminated. Mutations in the AIRE gene cause APECED syndrome (autoimmune polyendocrinopathy with candidiasis and ectodermal dystrophy), in which autoimmune diabetes can also appear, proof that central tolerance is essential.

The second level is peripheral tolerance: regulatory T lymphocytes suppress any autoreactive T lymphocyte that has escaped the first safety net in the thymus. In genetically susceptible people, both safety nets can fail. The strongest genetic risk comes from the HLA (human leukocyte antigen) region (the group of genes that shows the immune system which protein fragments to inspect) on chromosome 6, especially the combinations called DR3-DQ2 and DR4-DQ8. Cross-reactivity — the situation in which the same T lymphocytes or the same antibodies recognize both a microbial fragment and one of the body's own — is the bridge between an ordinary infection and autoimmunity.

The four autoantibodies that define autoimmunity in type 1 diabetes are anti-insulin antibodies (IAA), anti-glutamic acid decarboxylase (GAD, the form with a mass of 65 kDa — kilodaltons, the unit in which protein size is measured), anti-tyrosine phosphatase (IA-2) and anti-zinc transporter 8 (ZnT8). These autoantibodies do not attack or destroy the beta cells. The same proteins are, however, the target of the autoreactive T lymphocytes that reach the pancreas and destroy the beta cells. There are sequence similarities between viral epitopes and each of these proteins [2].

The viruses most studied as possible triggers through mimicry are Coxsackievirus B, rotavirus, other enteroviruses (echovirus, enterovirus B), cytomegalovirus, the rubella virus (in congenital infection), the mumps virus and SARS-CoV-2. The idea of epitopes with similar sequences refers to short stretches of amino acids (usually 5–15 amino acids) shared between a viral protein and a pancreatic autoantigen. These resemblances create the phenomenon of molecular mimicry [2].

The Coxsackie B virus (especially CVB4) is the most studied enterovirus in the pathogenesis of type 1 diabetes. Its 2C protein contains a short amino-acid sequence that strikingly resembles an epitope of GAD, the main autoantigen of the beta cell. This sequence is highly conserved across different CVB4 strains and across the entire coxsackie B-like subgroup. Exposure to it is therefore probably repeated several times over a lifetime. In addition, the sequence fits the HLA-DR3 molecule especially well, which favors the activation of autoreactive anti-GAD T lymphocytes [3].

During the infection, antigen-presenting cells process the 2C protein and present the fragment responsible for mimicry, together with HLA-DR3, to CD4+ T lymphocytes (the ones that coordinate the immune response). These multiply and reach the pancreas, where they recognize the body's own GAD65 peptide as an enemy. They then release interferon-gamma and call in cytotoxic CD8+ lymphocytes, which are the ones that destroy the beta cells. Most people exposed to Coxsackie B do not develop type 1 diabetes. The effect depends on having the right HLA subtype (DR3) and on other factors that are still unknown [3].

Rotavirus is one of the most common causes of gastroenteritis in children. On its surface it carries a protein (VP7) with a region that closely resembles a portion of the IA-2 protein in the pancreatic beta cells. Both fragments (from VP7 and from IA-2) bind to the HLA-DR4 molecule and are recognized in exactly the same way by T lymphocytes through their dedicated receptor. In addition, VP7 also has a region similar to GAD, so it simultaneously imitates two pancreatic autoantigens (GAD and IA-2) [4].

Rotavirus has been associated with the appearance of type 1 diabetes-specific autoantibodies in children, and rotavirus gastroenteritis generally occurs in the first years of life. Rotavirus vaccination programs appear to be associated with a lower incidence of type 1 diabetes in children. The results are so far only indicative and not consistent worldwide, so the interpretation must remain cautious. The mechanism remains, for now, an association rather than a proven cause [4] [5].

Yes, at the sequence level. Several computer-based analyses (bioinformatic, also called "in silico") have identified regions shared between certain SARS-CoV-2 proteins (especially the spike protein on the surface of the virus and the nucleocapsid, the protein shell around the genetic material) and human proteins, including endocrine antigens. In the laboratory, human monoclonal anti-spike and anti-nucleoprotein antibodies cross-react with dozens of human tissue antigens, without it being known so far whether this has any consequence for health. Also in the laboratory, SARS-CoV-2 has directly infected human beta cells, which carry the receptors the virus uses to enter the cell (ACE2 and TMPRSS2). The infection alters the cell's morphology, reduces the insulin granules and decreases glucose-stimulated insulin secretion [6] [7].

During the COVID-19 pandemic, several registries reported an increase in new cases of type 1 diabetes. Other studies, however, showed that an important share of these cases were in fact something else. They represented transient stress hyperglycemia, undetected pre-existing diabetes or the effects of corticosteroid therapy administered for the severe forms. SARS-CoV-2 has the theoretical capacity to contribute to type 1 diabetes-specific autoimmunity, through molecular mimicry and direct infection. The evidence that it can trigger autoimmune type 1 diabetes in susceptible people is so far preliminary, still under research [6].

Not yet. Molecular mimicry remains a hypothesis with partial support, not a fully proven cause. The arguments in favor include studies showing structural similarities between portions of viruses and portions of human proteins. Cross-reactive T lymphocytes and antibodies have been detected in patients, reacting both with the virus and with some of the body's own structures. Other arguments come from epidemiological associations: enterovirus infections, congenital rubella and other viruses are linked to a more frequent appearance of type 1 diabetes-specific autoantibodies in children [2].

The limitations, however, are important. Causality is hard to demonstrate, because viral infections occur several years before clinical onset. In addition, most exposed people do not develop type 1 diabetes. For example, Coxsackievirus B and rotavirus infect almost all the children of the world, yet only a minority reach autoimmunity. Molecular mimicry is probably necessary in some cases, but it is not sufficient on its own. It must act together with genetic susceptibility (HLA), immune tolerance defects and other environmental factors, such as the gut microbiota [2] [8].