Antibody repertoire and autoimmunity

Reports that autoantibodies are correlated with coronavirus disease 2019 (COVID-19) severity^{1 , 2} shed further light on relationships between infection and autoimmunity and remind us of our need to understand more deeply how antibody specificity is regulated and its role in balancing health and disease.

The antibody variable region responsible for binding to antigen is encoded by an exon that is assembled from gene segments through V(D)J recombination during early B-cell development in bone marrow. Antibody genes can further diversify through somatic hypermutation (SHM) in germinal centers (GCs), leading to improved affinities toward antigens.

Several mechanisms are in place to handle specificities that recognize self at several stages of development, including the early developmental phases and the GC diversification stage.³ The presence of secreted autoantibodies that react to self are generally thought of as breakdowns of tolerance mechanisms.

Strong ligand engagement with cell surface–expressed antibody (the antigen-binding part of the B-cell receptor) during precursor B-cell development in bone marrow encourages antibody light chain exchange, allowing for more innocuous specificities to advance to later stages of B-cell development and exit from bone marrow as transitional naive B cell. Autoreactivity encountered in the periphery can result in B cell anergy or deletion, further eliminating autoreactive specificities. T-cell receptor selection appears to be more stringent; after their own V(D)J-mediated T-cell receptor diversification, they can experience a more complete array of autoantigens through an autoimmune regulator–dependent mechanism to guide thymic T-cell tolerogenesis. In this context, it makes sense that antigen-specific T cells are required to guide selection of SHM-diversifying antibodies in GCs, as they have experienced a more complex selection process.

Despite tolerance filters, there appears to be substantial autoreactivity in the baseline antibody repertoires of healthy controls, and the degree to which this is a “bug” (ie, an unavoidable negative consequence of a complex system) or a “feature” (ie, an aspect that provides perhaps not-so-obvious benefit)—or whether it has aspects of both—is not yet clear. If it is purely a bug, the tolerance filters are leakier than one might expect.

Although a simple breakdown of tolerance is a likely major contributor on its own to autoantibody production, there are some findings that are consistent with an autoreactivity-as-a-feature component to the repertoire (Table I ). The antibody gene assembly process generates specificities that tend to be autoreactive to a level perhaps beyond what might be expected from random sequence generation, particularly in fetal and newborn B-cell development.⁴ The propensity of the agnostically generated early B-lineage cell repertoire toward autoreactivity suggests that there may be some value in an autoreactive repertoire to provide necessary B-cell receptor signals required ontogenetically or as a reservoir for polyreactivity for the potential benefit of on-demand expanded antigen recognition under more pressing infectious stresses.⁵ In certain circumstances, anticytokine antibodies could potentially provide a regulatory effect akin to negative feedback.

Table I.

Autoreactivity in the antibody repertoire

Origins

Polyreactivity is hard-coded in some germline VH gene segments

Latent polyreactivity due to imperfect vs tunable ontogenetic tolerance checkpoints

Affinity-matured autoreactivity due to imperfect vs tunable SHM-related checkpoints

Off-target SHM-mediated crippling of genes regulating B-cell proliferation

Possible benefits

Influence on fate-related BCR signals

Reservoir for expanded antigen recognition

Clearance of apoptotic cells

Negative feedback on cytokines

Negative consequences

Direct self-tissue damage

Driving of chronic clonal expansion

Anti-cytokine antibodies that can prohibit adequate immune response to infection

BCR, B-cell receptor; VH, heavy-chain variable region.

Additional B cell dynamic features that likely influence the probability of autoantibody production include the timing, context, and frequency of production. Antibody secretion occurs through at least 3 different paths with different activation requirements and time scales (Fig 1 ). One is natural antibody secretion—usually IgM spontaneously emitted from innate-like B cells and requires no known exogenous stimulus. The other 2 are antigen-dependent pathways—an extrafollicular pathway and a GC pathway. Both of the latter can lead to antibody-secreting cells and memory cells, but they differ in speed of antibody production, antibody maturation state, and longevity of antibody-secreting cells produced. Extrafollicular antibodies appear quickly, may or may not be (see later) refined by SHM and affinity maturation, and are in general shorter-lived owing to less durable antibody-secreting cells. The GC pathway takes more time due to SHM and T-cell selection, and it is generally believed that GC-derived antibody-secreting plasma cells can be longer-lived.

Fig 1.

Schematic of antibody producing pathways. Three main pathways for antibody production are shown. Innate-like B cells produce natural antibodies in an antigen-independent pathway (top). Antigen-dependent pathways include extrafollicular and germinal center pathways, as indicated with associated indicated features tied to primary (prime) and secondary (boosting) responses. Arrows indicate possible B cell differentiation pathways. Different shades of blue indicate variation due to somatic hypermutation and affinity maturation. ASC, Antibody-secreting cells; B, B cells; FDC, follicular dendritic cells; PC, plasma cells; T, T cells.

An important aspect of antibody production pathways is that the fast-acting extrafollicular response can act on both luck-of-the-draw naive B cells with by-chance specificities from the baseline naive repertoire and matured GC-experienced memory B cells that recognized the same or a similar antigen in the past, making available quick antibodies stored from prior responses. This rapid release of stored information is one of the contributory reasons why repeated exposure to the same or similar pathogen usually results in less disease than primary exposure does.

Autoantibody production from extrafollicular pathways can make sense, as antigen-specific T cells may or may not be involved and the GC autoimmune checkpoint is bypassed. Hence, a cost of quicker antibody production is less time for quality control. Because autoantibodies are often polyreactive, they may offer some benefit to maximize pathogen recognition capacity per B cell clone activated.

The GC pathway incorporates a feature that can reduce autoreactivity during the SHM and T-cell selection process of antibody affinity maturation, but it requires ready access to the autoantigen at play.³ The fact that many pathogenic autoantibodies carry somatic mutations and can lose autoreactivity when reverted to germline⁶ indicates that the GC pathway—either directly or antecedent to an extrafollicular activation of GC-experienced memory B cells—is also involved in the generation of autoantibodies. For chronically activated GCs, another way in which autoantibodies can escape tolerance is through off-target SHM-mediated crippling of genes whose function is to suppress autoreactivity (eg, via mutation of genes regulating cell proliferation that would otherwise suppress expansion of rogue clones).⁷

Anticytokine antibodies, particularly those against type I interferons, can weaken immune defenses and contribute to severe COVID in some individuals.⁸ Assessing their genetic composition and SHM profiles will shed light on their maturation dynamics. The presence of SHM would infer the existence of CD4 T cells that are also specific for the cytokine or another protein to which the cytokine is often linked. Given the transient nature of cytokines, they may not be available in high enough concentrations at locations where B- or T-cell tolerogenic filters occur to delete chance-made specificities. Anticytokine antibodies can influence immunity in multiple ways. They can block neutralizing activity by preventing receptor engagement, influence cytokine half-life, or restrict cytokine distribution. Because anticytokine antibodies can potentially be beneficial in autoimmune disease, their deliberate induction by vaccination has been entertained as a possible therapy.

In severe COVID, the prevalence of individuals with autoantibodies, including those to a large variety of cytokines, occurred at a surprisingly high frequency. More than half of hospitalized patients with COVID-19 developed serum autoantibodies out of proportion to their total serum IgG level.¹ Unlike autoreactivity in newborns, in which case different individuals tend to target similar autoantigen profiles,⁴ autoreactivity in severe COVID can vary between individuals, although there are some recurring themes.

A direct correlation between autoantibodies and antibodies recognizing nonstructural SARS-CoV-2 proteins has been observed.¹ This link may exist because antibody responses against nonstructural proteins would be expected to be associated with greater cellular damage and therefore linked to larger amounts of self-antigen spillage accompanying persistent infection, thus making conditions more favorable for breaking tolerance. It is important to note that autoantibodies are also common in acute infections caused by other viruses as well,⁹ indicating that the COVID correlations may be a feature of a more general phenomenon.

Although case reports are rarely conclusive owing to small case size and uncontrolled variables, there are several reports of autoimmunity associated with rare cases of COVID. These include vasculitis, arthritis, idiopathic inflammatory myopathies, sarcoidosis, systemic sclerosis, idiopathic thrombocytopenic purpura, Guillain-Barré syndrome, SLE, and multisystem inflammatory syndrome in children, among others.

Given the vast number of people who have recovered from COVID vis-à-vis the high prevalence of those with autoantibodies, autoantibodies are probably not sufficient to directly cause severe disease or long-term sequalae. Nor are they necessary, given that people get severe disease and long-COVID without them—at least without those we know about. However, autoantibodies have been found with higher prevalence in cases associated with long COVID,¹⁰ and there is some suggestion that they may contribute to disease pathogenesis² in addition to hindering more effective immunity by blocking type I interferons.⁸ For fatal cases, the consensus appears to be that autoimmunity likely plays a role.

Although autoantibody formation in severe infection may be due to tolerance filter failure, it could also be viewed in part as deployment of a crisis-phase protocol—a desperation-driven last-ditch-effort immune aggression against the pathogen at the expense of targeting precision. In the setting of advanced disease with a small chance of recovery from infection-driven tissue damage, autoimmunity may be a part of an understandable go-for-broke bet by the immune system with its own tissue-damaging effects. This presents a chicken-or-egg scenario with respect to potential drivers and consequences of autoimmunity in such settings. Whether such theoretic Hail Mary play protocols offer any benefit would be difficult to determine, as its association with poor outcomes would be strong either way, analogous to the way in which braking may be associated with vehicle collisions regardless of whether application of brakes served to cause bad outcomes or make them less likely.

Given the exquisite fine-tuning of protein-ligand interactions possible in biologic systems, it seems that autoreactivity in the antibody repertoire could in theory be less than that observed if the optimal conditions for host-pathogen homeostasis would require it that way. This leaves us with the question of whether antibody autoreactivity is an inevitable price to pay for a diverse anticipatory repertoire owing to biophysical limits of specificity or a deliberate feature whose virtues go largely unnoticed. In this light, there is a need to better understand the possible benefits of the autoreactivity that so many tolerance mechanisms seemingly struggle to contain, as well as its relationship to the vast diversity of the antibody repertoire, which as a whole may be providing the greater good.