Dissecting autoimmune encephalitis through the lens of intrathecal B cells

Autoimmune diseases result from attacks of the immune system against the body itself. It is the adaptive immune system with its T cells, B cells, and antibodies produced by effector B cells (i.e., plasma cells) that is mainly engaged in autoimmune reactions. When the immune system attacks brain structures, autoimmune encephalitis may ensue. Autoimmune encephalitides represent a variety of severe conditions hallmarked by autoantibodies to antigens expressed either on the surface of or within brain cells (1). Three of the commonest forms of autoimmune encephalitis are associated with antibodies against the cell surface proteins N-methyl-D-aspartate receptor (NMDAR), leucine-rich glioma inactivated-1 (LGI1), or contactin-associated protein like 2 (CASPR2). In PNAS, Theorell et al. (2) have studied intrathecal immune cells from two patients with anti-LGI1 encephalitis and one patient with anti-CASPR2 encephalitis. The authors report a striking dominance of clonally expanded plasma cells that produce the characteristic autoantibodies.

In autoimmune encephalitides, autoantibodies against cell surface antigens are pathogenic, because they interfere with the physiological roles of neurotransmitters or ion channels. The autoantibodies arise from activated B cells that differentiate into antibody-secreting plasma cells. This process requires that the B cells sense antigen via their B-cell receptors (BCRs). In addition, they need a secondary signal, which in most cases will come from a cognate T cell. B cells that bind antigen to their BCR will internalize the resulting complex. Subsequent presentation of antigen-derived peptides on surface HLA II molecules allows interaction with T cells that express an antigen-specific T-cell receptor (TCR). Because this process directly links B-cell specificity to antigen presentation, antigen-specific B cells are extremely efficient antigen-presenting cells for T cells (Fig. 1). The T cell–B cell crosstalk results in expansion of B cells and the generation of antibody-producing plasma cells. It will also lead to activation and clonal expansion of T cells. Since an antibody is the soluble version of a BCR, antibody responses indirectly report on the key event of antigen-dependent crosstalk between T cells and B cells.

Fig. 1.

T cell–B cell crosstalk leading to the formation of autoantibodies. B cells capture antigen with their B cell receptor (BCR), leading to internalization, processing and presentation of antigen in the form of peptide fragments bound to HLA II molecules. T cells that recognize the presented antigen via their T cell receptor (TCR) can give activation help to the B cells. Interacting T cells and B cells may recognize the same autoantigen. Alternatively, complex formation between self and foreign antigen allows T cells specific for the foreign antigen to provide help to self-reactive B cells. In both cases, B cells differentiate into plasma cells that secrete soluble versions of their BCRs as autoantibodies. Created with https://BioRender.com.

Together, cognate T cells and B cells can form structures known as germinal centers. They typically occur in secondary lymphoid organs such as lymph nodes, but some conditions are associated with so-called ectopic germinal centers in tissues. Within germinal centers, B cells undergo rapid proliferation accompanied by somatic mutation of their BCR genes. With the help of T cells, mutations that increase the BCR affinity for antigen are selected for. This process is called affinity maturation, and it is essential for the formation of high-affinity antibodies. While B cells under some circumstances can undergo T cell-independent activation, T cells appear necessary for affinity maturation and generation of highly mutated antibodies in germinal centers.

“In PNAS, Theorell et al. have studied intrathecal immune cells from two patients with anti-LGI1 encephalitis and one patient with anti-CASPR2 encephalitis.”

A variety of mechanisms exist to prevent autoimmune attacks by the immune system. As TCRs and BCRs are generated by stochastic gene rearrangements, self-reactive antigen receptors are inevitably generated. Resulting autoreactive cells can be removed during T-cell and B-cell development. This process is known as central tolerance, and it works more stringently for T cells than for B cells. Hence, mature self-reactive B cells are common. The presence of self-reactive B cells is not necessarily a problem, since such cells do not become autoantibody-producing plasma cells without T-cell help. In addition to central tolerance, there are peripheral tolerance mechanisms, including action of regulatory T cells, which help prevent activation of autoreactive T cells and B cells. When an autoimmune disease develops, one or more of these immune tolerance mechanisms have failed.

By performing single-cell sequencing of immune cells in the cerebrospinal fluid (CSF) of three encephalitis patients, Theorell et al. (2) were able to get information about transcriptomic profiles as well as antigen receptor sequences of T cells and B cells. The most interesting insight comes from the analysis of cells of the B-cell lineage. Altogether, 381 IgG BCR sequences were obtained of which 166 were expressed as recombinant monoclonal antibodies (mAbs). The authors found that of the mAbs expressed from singleton BCR clones, 62% reacted with either LGI1 or CASPR2. This autoreactivity increased to 100% of cells in clonal groups with four or more members. Most of the B-cell-lineage cells had transcriptomic profiles of plasmablasts or plasma cells, thus being antibody-secreting cells. The BCRs carried somatic mutations suggestive of affinity maturation. When reverting the sequences to presumed germline configuration, 65% of the mAbs lost reactivity to either LGI1 or CASPR2.

Together with the reported HLA association of anti-LGI1 and anti-CASPR2 encephalitis (1), the importance of somatic BCR mutations for antigen reactivity strongly suggests that B-cell activation is dependent on T-cell help. The nature of this T-cell help, however, remains elusive. Theorell et al. (2) did not find evidence of clonally expanded T cells in the CSF of their patients. Moreover, affinity-enhancing antibody mutations did not appear to arise intrathecally, suggesting that T cell-mediated affinity maturation occurs outside of the CNS (2). Hence, B cells undergoing T cell-dependent activation in peripheral lymph nodes could subsequently cross the blood–brain barrier. This notion is also supported by the apparent lack of ectopic germinal centers in brain tissue of autoimmune encephalitis patients (3) and the observation that autoantibody levels are higher in serum than in the CSF (1). Despite the implication of T cells in autoantibody production, circulating T cells with reactivity to LGI1 are not more frequent in patients with anti-LGI1 encephalitis than in healthy donors (4).

As B-cell activation requires presence of antigen, initiation of anti-LGI1 and anti-CASPR2 responses outside of the CNS would imply that the neuronal antigens are present in the periphery. Alternatively, autoreactive B cells could potentially be activated in response to a different, unknown antigen as a result of cross-reactivity. In another autoimmune disease of the brain, multiple sclerosis, Epstein–Barr virus (EBV) infection has been linked to disease initiation (5), and it has been demonstrated that some EBV-reactive antibodies can cross-react with the brain antigen GlialCAM (6). Interestingly, patients with herpes simplex virus encephalitis frequently develop autoimmune encephalitis with autoantibodies (7), suggesting that viral infection can lead to subsequent CNS autoimmunity. However, a role of molecular mimicry between viral epitopes and self-antigens has not been demonstrated, and the mechanisms underlying virus-induced autoantibody formation are so far undefined.

One prominent example where foreign antigen is involved in driving autoantibody formation is the gluten-sensitive enteropathy celiac disease. In this case, the mechanisms of T cell–B cell crosstalk, leading to antibody production, have been dissected (8). Autoreactive B cells recognizing the enzyme transglutaminase 2 (TG2) can receive activation help from T cells specific for dietary gluten proteins. Because many gluten-derived peptides are excellent substrates for TG2, the formation of TG2-gluten enzyme–substrate complexes can facilitate BCR-mediated uptake of gluten by TG2-specific B cells and, thus, interaction with gluten-specific T cells. The outcome is mutual activation of gluten-reactive T cells and TG2-reactive B cells, accompanied by the production of disease-specific anti-TG2 autoantibodies. Common for celiac disease and anti-LGI1/anti-CASPR2 encephalitis is accumulation of autoreactive plasma cells in the affected target organs. However, while anti-LGI1 and anti-CASPR2 target cell surface proteins, anti-TG2 antibodies in celiac disease are directed against an intracellular enzyme, and the soluble antibodies are not known to have a direct pathogenic effect. In the form of membrane-bound BCRs, however, anti-TG2 antibodies are deeply implicated in the pathogenesis of celiac disease, as they allow efficient presentation of gluten peptides on disease-associated HLA II molecules. A similar role of B cells as antigen-presenting cells might be possible in autoimmune encephalitides. A requirement for interaction between T cells recognizing a foreign antigen and autoreactive B cells is that the foreign antigen is physically linked to the self-antigen (Fig. 1).

Characteristic features of anti-TG2 autoantibodies are low mutation levels and preserved TG2 reactivity in germline configuration. Similar observations have been made for anti-NMDAR encephalitis antibodies, thus separating them from the anti-LGI1 and anti-CASPR2 encephalitis antibodies (9). Due to the central role of T cells in antibody affinity maturation, the low mutation levels in anti-NMDAR antibodies could be interpreted as a sign of T cell-independent B-cell activation. In celiac disease, however, we know that autoantibody production depends on gluten-reactive T cells. In this case, the low mutation levels may be a result of T cell–B cell crosstalk mainly occurring outside of germinal centers. A similar scenario could thus be relevant for anti-NMDAR encephalitis. Importantly, the reported lack of HLA-association in anti-NMDAR encephalitis (1) does not exclude the involvement of T cells in autoantibody formation. Strong HLA associations indicate that a limited set of peptides originating from one or few proteins are recognized by disease-relevant T cells, while no HLA association can be expected if many peptides binding to different HLA allotypes are implicated.

Given the central role of B cells, either in production of pathogenic autoantibodies or in presentation of antigen to pathogenic T cells, B cell-directed therapies hold great potential for treatment of many autoimmune diseases (10). The therapies include B cell-depleting mAbs against CD20 and CD19, chimeric antigen receptor (CAR) T cells against CD19, bispecific T-cell engagers (BiTEs) as well as depleting therapies aiming at antigen-specific B cells with chimeric autoantibody receptor (CAAR) T cells. Preclinical studies of the latter approach have been reported for anti-NMDAR encephalitis (11).

The results presented by Theorell et al. strongly indicate that intrathecal production of anti-LGI1 and anti-CASPR2 autoantibodies depends on antigen-specific interactions between T cells and B cells. A big remaining question is what the T cells recognize. The implicated T cells can be autoreactive T cells, but it remains a possibility that they are specific to (still unidentified) foreign antigen(s).