Altered B cell signalling in autoimmunity

Abstract

Recent work has provided new insights into how altered B cell-intrinsic signals — through the B cell receptor (BCR) and key co-receptors — function together to promote the pathogenesis of autoimmunity. These combined signals affect B cells at two distinct stages: first, in the selection of the naive repertoire; and second, during extrafollicular or germinal centre activation responses. Thus, dysregulated signalling can lead to both an altered naive BCR repertoire and the generation of autoantibody-producing B cells. Strikingly, high-affinity autoantibodies predate and predict disease in several autoimmune disorders, including type 1 diabetes and systemic lupus erythematosus. This Review summarizes how, rather than being a downstream consequence of autoreactive T cell activation, dysregulated B cell signalling can function as a primary driver of many human autoimmune diseases.

Despite the established importance of B cells in the pathogenesis of human autoimmunity, the immune mechanisms that underlie initial breaks in B cell tolerance have not been completely defined. In addition to clonally rearranged B cell receptors (BCRs), B cells express innate pattern recognition receptors (including Toll-like receptors (TLRs)), co-stimulatory molecules (including CD40, CD80 and CD86) and cytokine receptors. Both the establishment of the naive B cell repertoire and B cell activation during an immune response depend on the coordinated, synergistic activation of these receptor families.

Genome-wide association studies (GWAS) have identified hundreds of gene polymorphisms that are associated with an increased risk of developing auto-immunity^1–5. Importantly, the vast majority of these genetic changes are predicted to affect immune function. Most are located in non-coding elements that probably have an effect on gene expression, whereas only a limited number result in altered protein structures. Despite this increasingly robust genetic dataset, there is only a limited amount of mechanistic data with respect to the cell lineage-specific and stage-specific effects of candidate risk variants. Notably, autoimmunity-associated variants identified by GWAS are highly enriched for signalling programmes that may affect B cell function, including in genes that encode receptors, signalling effectors and downstream transcriptional regulators of the BCR, CD40, TLRs or cytokine receptors⁶. Taken together, these data suggest that in an appropriate environmental setting, even modest alterations in B cell signalling may be sufficient to initiate, promote and/or sustain autoimmune disease, particularly diseases that are associated with humoral autoimmunity.

In this Review, we present a model in which dysregulated B cell signalling functions to initiate autoimmunity by modulating the naive BCR repertoire during immature and transitional B cell development, and by promoting the peripheral activation of auto-reactive B cell clones. First, we describe how altered B cell signalling affects the negative and positive selection of B cells during development, skewing the naive B cell repertoire towards self-reactivity or poly-reactivity. Next, we highlight the importance of T cell-independent and T cell-dependent extrafollicular B cell activation in the pathogenesis of humoral autoimmunity. Finally, we discuss how dysregulated B cell-intrinsic BCR, TLR and cytokine signalling can be sufficient to initiate spontaneous, autoimmune germinal centre (GC) responses, resulting in a loss of T cell tolerance, epitope spreading and GC-dependent systemic autoimmunity. In this context, we propose that GWAS-identified risk variants promote autoimmunity by affecting B cell signalling across a continuum of developmental selection and peripheral activation responses.

Receptor crosstalk shapes the naive repertoire

BCRs are generated by the random recombination of germline-encoded variable, diversity and joining gene segments. Although necessary for the generation of receptors that can recognize diverse pathogens, an inherent trade-off of this process is the creation of self-reactive receptors that have the potential to elicit an autoimmune response. Throughout development, immature B cells in the bone marrow (BM) and transitional type 1 (T1) and type 2 (T2) B cells in the periphery are subject to an interplay of positive and negative selection mechanisms to ensure the establishment of a diverse but ‘safe’ repertoire within the follicular mature or marginal zone (MZ) compartments^7,8 (BOX 1). Importantly, although the strength of BCR ligation is the dominant driver of B cell tolerance, recent studies indicate that signalling through the B cell-activating factor receptor (BAFFR; also known as TNFRSF13C), TLRs and CD40 synergizes with BCR activation to define the mature B cell repertoire (FIG. 1). Although the effect of GWAS-identified autoimmunity-associated polymorphisms on this process has not been extensively studied, emerging data indicate that altered signalling downstream of these receptor families can modulate selection, thereby skewing the naive B cell repertoire towards autoreactive B cell specificities.

Box 1. Positive and negative selection of autoreactive B cells.

The majority of autoreactive B cells are removed or segregated from the developing repertoire through the processes of negative selection, which include deletion¹⁷¹, receptor editing¹⁷² and the induction of anergy¹⁷³. In addition to these negative selection mechanisms, positive selection of distinct B cell receptor (BCR) specificities also contributes to the mature B cell repertoire. Provided that it does not surpass a presumed threshold for negative selection, BCR engagement with self-ligands promotes the survival advantage of a limited number of competing B cells during development^174–176. Consistent with an effect of positive selection on B cell development, specific immunoglobulin variable-domain gene families are enriched in the mature B cell compartments^177,178. In addition to BCR engagement, B cell selection is promoted by BAFF-mediated survival signals¹⁷⁹, by engagement with Toll-like receptor ligands⁵² and/or by CD40 signalling^53,55. Notably, although co-receptor signalling has primarily been viewed as restricted to the periphery, accumulating data also suggest earlier roles in the bone marrow^38,46,55. Collectively, these studies support a model in which self-ligand binding to B cells that results in a signal that is below the threshold for negative selection, combined with adequate co-receptor signalling, results in a competitive advantage of B cell clones in entering the mature repertoire.

Figure 1. B cell receptor and co-receptor signalling govern B cell selection and maturation.

Self-reactive B cells are subject to positive or negative selection throughout their development in the bone marrow and periphery (spleen). The selective fate of an individual B cell clone depends on multiple factors, including the location and form of self-antigen encounter, the strength of the B cell receptor (BCR) signal and synergy with co-receptor pathways. Negative selection mechanisms (such as deletion, receptor editing and anergy) are mediated primarily by BCR signalling, with potential input from specific Toll-like receptors (TLRs). By contrast, positive selection through survival and/or clonal expansion occurs primarily in transitional B cells in the periphery, and is driven by a complex interplay between BCR signalling and co-receptor signalling mediated by B cell-activating factor receptor (BAFFR), CD40 and TLRs. As the developing repertoire is fine-tuned, transitional B cells mature and populate the follicular mature (FM) compartment or the marginal zone (MZ) compartment. Although the majority of autoreactive BCR specificities are purged by negative selection, a proportion of mature naive B cells exhibit self-reactivity and/or polyreactivity, particularly within the MZ compartment. The dashed arrows indicate ongoing research regarding nonlinear routes for B cell development. T1, transitional type 1; T2, transitional type 2.

BCR signalling

The BCR is a master regulator of negative and positive selection mechanisms

Sustained BCR signalling is required for the survival of both immature BM B cells and mature B cells in the periphery⁹ through the induction of nuclear factor-κB (NF-κB)-dependent and/or phosphatidylinositol 3-kinase (PI3K)-dependent pro-survival signalling^10–12. Conversely, stronger BCR signalling can promote apoptosis in both immature¹³ and transitional¹⁴ B cells in vitro. Thus, an ‘intermediate’ level of tonic BCR signalling may be optimal for immature B cell survival. However, it is evident that the context of self-antigen encounter, rather than signal strength per se, helps to determine the outcome of these events. As detailed in BOX 1, additional factors — including the location of BCR engagement, the form of self-antigen and synergy with additional co-receptor signals — all affect the developmental fate of individual B cells (FIG. 1).

Recent studies from our laboratory and others have expanded our understanding of ligand-mediated positive selection in the periphery. For example, our group identified a subset of T2 B cells that enter the cell cycle in response to antigen engagement¹⁵. Using the M167 VH1 heavy chain transgenic model, in which B cells that express self-reactive phosphorylcholine-specific BCRs are detected by an idiotype-specific antibody, we demonstrated that M167-idiotype⁺ B cells are enriched within this cycling transitional subset and are further enriched in the MZ compartment, which is consistent with antigen-driven positive selection^15,16. Importantly, the effect of self-ligand engagement on B cell selection is probably effective across a range of antigen specificities. Using the Nur77–GFP reporter strain, in which BCR signalling activates GFP expression under control of the region that regulates Nur77 (also known as Nr4a1), Zikherman et al.¹⁷ demonstrated that BCR signalling occurs as a continuum during development. Consistent with observations in the M167 model¹⁵ and other transgenic models¹⁸, GFP reporter expression initially increases in T2 B cells, implying that the selection of transitional B cells into follicular mature and MZ compartments is refined by antigen stimulation.

The BCR signalling variant PTPN22^R620W promotes greater autoreactivity in mature cells

Although not yet studied in detail, autoimmunity-associated genetic risk variants that affect BCR signalling probably also modulate B cell selection during development. For example, protein tyrosine phosphatase nonreceptor 22 (PTPN22) negatively regulates signalling downstream of both BCR and T cell receptor (TCR) activation. A single-nucleotide polymorphism in PTPN22 (C1858T; R620W) is associated with increased susceptibility to several autoimmune diseases^19,20, including systemic lupus erythematosus (SLE)²¹, type 1 diabetes (T1D)²² and rheumatoid arthritis (RA)²³. Consistent with a role for this polymorphism in modulating B cell tolerance, healthy individuals who carry the autoimmunity-associated allele of PTPN22 exhibit increased autoreactivity in the naive B cell repertoire, together with an expansion of transitional and anergic IgD⁺IgM⁻CD27⁻B cell subsets, relative to individuals who do not carry the allele^24,25.

An unresolved question is whether the PTPN22^R620W polymorphism is a gain-of-function or loss-of-function variant²⁰. Although initial human studies indicated that this polymorphism might increase the phosphatase activity of PTPN22 (resulting in reduced antigen-receptor signalling)^24,26,27, independent mouse knock-in models demonstrated enhanced BCR and TCR signalling^28,29. Although this discrepancy between human and animal studies remains unresolved, T cells from aged Ptpn22 knock-in mice exhibit blunted TCR signalling, which probably reflects chronic antigen stimulation (X. Dai and D.J.R., unpublished observations). On the basis of this observation, we hypothesize that chronic, enhanced antigen-receptor signalling in PTPN22^R620W carriers might explain the observed decreases in BCR and TCR signalling in humans. Importantly, this dichotomy also raises the question as to whether the increase in B cell self-reactivity within the pre-immune repertoire of PTPN22^R620W carriers reflects enhanced positive selection, rather than reduced negative selection, of B cells during development. In support of this concept, we have observed marked alterations in the naive B cell repertoire in mice that have a mutation orthologous to the human PTPN22^R620W variant (G.M. and D.J.R., unpublished observations).

Although the complex mechanisms that underlie the establishment of the mature B cell repertoire require further study, these combined observations suggest that self-antigen-induced BCR signalling shapes the mature naive B cell compartment by subtly altering transitional B cell survival and positive selection. By modulating this signalling — probably in concert with co-receptors, as described below — we propose that a subset of auto-immunity-associated variants skew the pre-immune repertoire towards increased autoreactivity, thus potentiating the risk of humoral autoimmunity.

BAFFR signalling

Coordinate regulation of BCR and BAFFR signalling in naive B cell development

BCR signalling is crucial for B cell survival and tolerance, but B cells also compete for access to limited survival signals during development. Among these signals, the cytokine B cell-activating factor (BAFF; also known as TNFSF13B) has prominent roles in B cell survival and homeostasis³⁰. BAFF can bind to three distinct B cell surface receptors (namely, BAFFR, transmembrane activator and CAML interactor (TACI; also known as TNFRSF13B) and B cell maturation antigen (BCMA; also known as TNFRSF17)); however, the ligation of BAFFR seems to have a dominant role in peripheral B cell maturation.

The BCR pathway regulates BAFFR signalling through several mechanisms^31,32. Strikingly, recent work demonstrates that BAFFR signalling can directly integrate with tonic BCR signalling to promote B cell survival. For example, BAFFR signalling promotes the phosphorylation of proximal BCR signalling components, including spleen tyrosine kinase (SYK) and Igα, and inducible Syk deletion results in reduced BAFF-dependent B cell survival despite intact alternative NF-κB signalling³³. In addition, BAFF seems to co-opt signalling components of the BCR to promote CD19 phosphorylation, resulting in PI3K-dependent B cell survival³⁴. Consistent with this, CD19 is required for the survival of SYK-deficient B cells³⁵. In combination, these studies suggest that a complex interplay between the BCR and BAFFR signalling pathways promotes B cell survival, and that this crosstalk probably modulates transitional B cell selection and the establishment of the mature naive B cell repertoire (FIG. 2a).

Figure 2. Altered B cell receptor and co receptor signalling promotes increased autoreactivity within the naive B cell repertoire.

a | Under homeostatic conditions, self-reactive B cells are subjected to both positive and negative selection mechanisms as they transit into the naive B cell pool and establish the naive repertoire. Whereas tonic B cell receptor (BCR) signalling and BCR engagement with self-antigen primarily regulate these events, synergy between the BCR and co-receptors fine-tunes the tolerance programme within a given B cell. Among these co-receptors, B cell-activating factor receptor (BAFFR) signalling (1) synergizes with BCR signalling during late bone marrow and transitional development through a series of complex events, including proximal biochemical crosstalk and the downstream transcriptional regulation of both receptor and substrate expression. Dual BCR and Toll-like receptor (TLR) signalling (2) is mediated by internalization and delivery of self-antigens that contain TLR ligands to autophagosomes, which contain endosome-resident TLRs. CD40 signalling (3), which is triggered by interaction with CD40 ligand (CD40L) on T cells and possibly other cell types, also integrates with the BCR signalling pathway. Although BCR signalling can modulate CD40 expression, other biochemical or transcriptional events that affect this crosstalk are less well understood. b | In genetic (or environmental) settings that promote an increased risk of developing autoimmunity, the homeostatic signalling thresholds are modulated, and self-reactive B cells exhibit greater positive selection and/or reduced negative selection, leading to a naive repertoire that is skewed towards autoreactivity. For example, excess amounts of B cell-activating factor (BAFF; 1) in the Baff-transgenic mouse model rescue low-affinity self-reactive B cells from negative selection. A similar mechanism has been proposed to exist in individuals with systemic lupus erythematosus (SLE). Similarly, in mouse and human settings of Wiskott–Aldrich syndrome (WAS) deficiency (2), hyper-responsive dual BCR and TLR signalling promotes the positive selection of transitional B cells with BCRs that use a limited subset of genes that encode self-reactive heavy-chain variable (VH) domains. Healthy individuals with the autoimmunity-associated variant PTPN22^R620W (3) exhibit altered BCR and CD40 signalling, and have an enrichment of self-reactive BCR specificities within the naive B cell compartment. Although it has not yet been definitively demonstrated, it is likely that enhanced positive selection, rather than relaxed negative selection, predominantly mediates this change. The thickness of the arrows indicates the strength of pathway activation. MAPK, mitogen-activated protein kinase; NF-κB, nuclear factor-κB; PI3K, phosphatidylinositol 3-kinase.

Given that immature IgM⁺ B cells seem to be normal in Baff^−/− and Baffr^−/− mice, BAFF was not thought to affect BM B cell development and selection. However, new studies challenge this idea and may support a model that is similar to that of BAFFR-driven peripheral selection. BAFFR is expressed by a CD23⁺ immature BM B cell subset, which also expresses B220 (also known as PTPRC), IgM and AA4.1 (also known as CD93). This subset further shares phenotypic and functional characteristics of splenic T2 B cells, including maturation from CD23⁻ counterparts, dependence on both BAFFR and tonic BCR signalling, and proliferation following T cell help^31,36,37. BAFFR-deficient ‘T2-like’ immature cells also exhibit a competitive disadvantage relative to their wild-type counterparts³⁸. Collectively, these data support a revised model of BM B cell tolerance in which BAFFR signalling, in concert with tonic BCR signalling, promotes the differentiation and/or positive selection of certain immature B cells. Additional studies that assess the effects of antigen-mediated BCR signalling are required to better define the potential contribution of this cell population to the establishment of the naive repertoire and/or autoimmune responses.

Excess BAFF promotes the development of an altered repertoire of self-reactive naive B cells

In addition to its effect on B cell survival, BAFF has been implicated in the development of SLE and other autoimmune diseases. For example, BAFF overexpression in Baff-transgenic mice recapitulates several cardinal features of human SLE, including polyclonal B cell proliferation, antinuclear antibody production and the development of immune-complex-mediated glomerulonephritis³⁹. In addition, BAFF levels are increased in a subset of patients with SLE⁴⁰, and a BAFF-inhibiting therapeutic antibody, known as belimumab, demonstrated clinical efficacy in patients with SLE⁴¹, which emphasizes the importance of BAFF in the pathogenesis of the disease.

Although the mechanisms by which BAFF promotes autoimmunity have not been completely defined, excess BAFF probably contributes to disease development through effects on B cell selection during development. Specifically, increased BAFF levels rescue low-affinity self-reactive transitional cells, thereby allowing their maturation and entry into ‘forbidden’ splenic zones. In a model in which developing B cells compete for available BAFF, excess BAFF results in relaxed selection^42–44. Thus, BAFF-dependent survival signals, presumably downstream of BAFFR, contribute to humoral autoimmunity in Baff-transgenic mice by increasing the proportion of autoreactive B cells within the mature naive repertoire (FIG. 2b). In addition, an essential role for B cell-intrinsic TACI signalling in BAFF-driven autoimmunity has recently been identified and is discussed in detail below.

TLR signalling

Dual BCR and TLR signalling orchestrates the selection of self-reactive transitional B cells

Dual BCR and TLR signalling, facilitated by the trafficking of self-antigen to endosomal TLRs, has an important role in the initial activation of mature naive autoreactive B cells⁴⁵. Although biochemical studies of these pathways in transitional B cells have not yet been carried out, human and mouse studies nevertheless support a role for BCR and TLR crosstalk in regulating B cell tolerance, particularly tolerance towards nuclear self-antigens (FIG. 2a).

Interestingly, similarly to BCR signalling, TLR activation also seems to have a dichotomous role in promoting both the positive and negative selection of autoreactive B cells. On the one hand, Myd88^−/− mice (which lack the gene that encodes myeloid differentiation primary response protein 88) have abnormal central tolerance, and patients who lack MYD88 or IRAK4 (which encodes IL-1 receptor-associated kinase 4) exhibit defects in central and peripheral tolerance, implicating TLR-dependent innate signalling in the removal of autoreactive clones during B cell development^46–48. On the other hand, in a mouse model in which lupus is driven by an autoimmunity-associated Cd45 variant, strain-specific variance in TLR9 signal strength in C57BL/6 versus BALB/c mice either facilitated B cell negative selection and central tolerance, or promoted breaks in peripheral B cell tolerance (probably through enhanced positive selection)⁴⁹. Additional evidence for TLR signalling promoting transitional B cell positive selection in both humans and mice comes from the primary immunodeficiency Wiskott–Aldrich syndrome (WAS; FIG. 2b). Mutations in the WAS protein influence actin polymerization and receptor signalling in nearly all haematopoietic cell lineages and, in B cells, WAS mutations result in modestly enhanced signalling downstream of both the BCR and TLRs^50–52. In this context, although central tolerance is maintained in Was^−/− mice, late T2 B cells exhibit enhanced proliferation⁵². High-throughput B cell repertoire sequencing of Was^−/− mice and humans with WAS identified the preferential selection of specific heavy-chain variable (VH) gene families (VH10 in mice; VH4-34 in humans) as late transitional B cells become mature naive B cells⁵². Consistent with the BCR using these VH families to bind to antigenic complexes that also contain TLR ligands, both the increase in transitional B cell cycling and the altered naive repertoire in WAS were MYD88 dependent⁵². Thus, in addition to the known effects on the activation of mature autoreactive B cells, altered BCR and TLR signalling is coordinated to enhance the positive selection of self-reactive specificities into the naive B cell repertoire.

In summary, dual BCR and TLR signalling probably affects both negative and positive selection events during B cell development, with the effects depending predominantly on the developmental stage. TLR signalling in immature B cells primarily promotes negative selection, whereas dual signalling in late transitional B cells facilitates positive selection through clonal expansion. Additional studies that assess BCR crosstalk with specific TLRs during B cell development should improve our understanding for how these coordinated signals shape the naive repertoire.

CD40–CD40 ligand signalling

B cell-intrinsic CD40 signalling affects the naive B cell repertoire

In addition to its well-established role in T cell-dependent B cell responses, recent studies indicate that CD40 signalling can also directly modulate transitional B cell selection (FIG. 2a). CD40 signalling can both rescue transitional B cells from BCR-induced apoptosis in vitro¹⁴ and compensate for reduced BCR signalling during development, as shown by a markedly smaller peripheral B cell compartment in mice that lack both Bruton’s tyrosine kinase (BTK) and CD40 than is found in wild-type mice⁵³. In addition, CD40 ligand (CD40L) on naive CD4⁺ T cells augments the survival of autoreactive B cells⁵⁴. Consistent with this idea, our group demonstrated that CD40 signalling facilitates B cell positive selection, leading to an altered mature B cell repertoire, particularly within the MZ⁵⁵. Thus, although Cd40^−/− mice have normal B cell numbers⁵⁶, CD40 signalling still affects the breadth of specificities in the mature compartment. It is also tempting to speculate about an earlier role for CD40 in central tolerance on the basis of its expression by immature BM cells (G.M. and D.J.R., unpublished observations)⁵⁷ and the competitive disadvantage of Cd40^−/− BM B cells observed by Schwartz et al.⁵⁵. Future mechanistic studies that directly assess the intersecting roles of the CD40 and BCR pathways are likely to provide additional insights into immature and transitional B cell selection.

Enhanced CD40 signalling in the presence of the autoimmunity-associated variant PTPN22^R620W

Whether these CD40-dependent effects on the pre-immune repertoire also contribute to the risk of autoimmunity has not been addressed. However, individuals who carry the autoimmunity-associated R620W variant of PTPN22 exhibit both altered B cell selection and increased CD40 expression in mature naive cells²⁵. These findings raise the possibility that dysregulated BCR signalling in PTPN22^R620W carriers functions, in part, to facilitate CD40 signalling, culminating in events that may promote the increased survival of autoreactive cells (FIG. 2b). Consistent with this idea, although the increased frequency of autoreactive B cells observed in the periphery of patients with immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX syndrome) has been attributed to a lack of regulatory T cell-dependent regulation of B cell tolerance^48,58, an alternative interpretation is that the increased levels of CD40L found on CD4⁺ T cells in these patients instead directly promote the survival and positive selection of autoreactive B cells.

In summary, the studies described above support a paradigm in which BCR signalling coordinates with BAFF family receptors, TLRs and CD40 to shape the mature naive B cell repertoire. Although these signalling pathways are frequently considered separately, crosstalk between them seems to be the norm rather than the exception. Depending on the developmental stage and context, this ongoing signal integration can function to modulate either negative or positive selection. Importantly, we predict that autoimmunity-associated genetic risk variants alter this process by coordinately affecting these B cell-intrinsic pathways, resulting in increased self-reactivity and polyreactivity within the pre-immune repertoire.

Receptor crosstalk shapes B cell activation

Although carriers of GWAS-identified autoimmunity-associated variants frequently exhibit increased auto-reactivity within the naive B cell repertoire, this increase in the proportion of autoreactive naive B cell clones is insufficient for the development of systemic autoimmunity. Indeed, up to 20% of mature B cells in healthy individuals exhibit antinuclear reactivity in the absence of clinical autoimmunity⁵⁹. Therefore, additional immune signals are required to initiate the activation of autoreactive B cell clones and thereby precipitate humoral autoimmunity. During a humoral immune response, antibody-secreting cells (ASCs) are generated through two distinct pathways: namely, the extrafollicular and GC B cell pathways. Importantly, although these pathways are often considered separately, they probably occur in parallel to protect the host from infectious challenge and during autoimmunity. The specific B cell activation pathway that is responsible for the majority of autoantibodies remains undefined, largely because it is difficult to determine whether an ASC is derived from an extrafollicular response or whether it developed within a GC.

Extrafollicular B cell responses

Extrafollicular B cell activation contributes to the production of class-switched autoantibodies

The pathophysiology of SLE is particularly complex, as both human clinical studies and animal models indicate that serum antinuclear autoantibodies can be generated through both extrafollicular and GC B cell activation pathways. This issue is not only of academic interest, as short-lived plasma cells (predominantly derived from extrafollicular responses) are sensitive to anti-proliferative immunosuppression and B cell depletion by CD20-specific antibody, whereas GC-derived long-lived plasma cells are largely resistant to such treatment^60–62.

The importance of extrafollicular B cell activation in autoantibody generation has been most extensively studied in the MRL.Fas^lpr mouse lupus model. Using MRL.Fas^lpr mice that express a transgene that encodes rheumatoid factor (RF), William et al.⁶³ demonstrated that the spontaneous activation of RF⁺ B cells occurs through an extrafollicular pathway, independently of GCs. Microdissection and BCR cloning revealed extensive somatic hypermutation (SHM) in RF⁺ B cells at the T cell–red pulp border, confirming that antigen-driven SHM can occur outside the GC⁶³. The prominent roles of extrafollicular B cell responses in lupus pathogenesis are not unique to the MRL.Fas^lpr model. For example, humoral autoimmunity in transgenic mice that over-express BAFF occurs independently of CD4⁺ T cells⁶⁴. In addition, up to 60% of autoreactive ASCs in NZB/W mice are short-lived plasmablasts, which are sensitive to cyclophosphamide treatment⁶⁰.

Continuous extrafollicular B cell activation is probably also relevant to the pathogenesis of human SLE. Plasmablast-like cells are expanded in the peripheral blood of patients with SLE, and their numbers correlate with titres of double-stranded DNA (dsDNA)-specific antibodies and disease flares⁶⁵. In addition, a peripheral blood plasmablast gene signature identifies a subset of patients with increased disease activity⁶⁶. Moreover, a recent study used deep sequencing and single-cell analysis of peripheral blood ASCs in patients with SLE to characterize the phenotype of autoantibody-producing cells⁶⁷. By studying activated B cells that express the intrinsically self-reactive VH4-34 heavy chain (see BOX 2), this study demonstrated that a considerable proportion of VH4-34⁺ SLE plasmablasts were derived from newly activated B cells that exhibited clonality that persisted for months. In addition, a subset of these autoreactive ASCs lacked BCR mutations, which is most consistent with a non-GC origin.

Box 2. VH4 34⁺ B cells.

VH4-34 is a human heavy chain family that has germline-encoded polyreactivity towards multiple self-antigens, including double-stranded DNA, apoptotic cells and red blood cell carbohydrate antigens^180–183. VH4-34⁺ B cells can be identified in the peripheral blood using the anti-idiotype monoclonal antibody 9G4. In healthy individuals, 5–10% of naive B cells are 9G4⁺, but 9G4⁺ B cells are largely excluded from the germinal centre, memory cell and plasma cell compartments, which suggests the existence of a negative selection checkpoint that prevents autoreactive B cell activation. By contrast, 9G4⁺ B cells are enriched in memory B cell subsets and plasma cells in patients with systemic lupus erythematosus, which is consistent with the observation that the levels of VH4-34⁺ autoantibodies are increased in this disease^184–188.

B cell signals required for extrafollicular responses

Continuously generated, short-lived plasmablasts could be an important therapeutic target in humoral autoimmune diseases, and thus we need to identify the key signals that control their generation. B cells show a unique propensity for activation by integrated signals downstream of the BCR and TLRs^45,68. Although this arrangement has probably evolved to resist viral infection⁶⁹, it carries the risk of autoimmunity as the endosomal TLRs can also respond to RNA-containing and DNA-containing epitopes within apoptotic particles. Consistent with this idea, MYD88-deficient MRL.Fas^lpr mice lack antinuclear antibodies and are protected from systemic autoimmunity⁷⁰. After activation by integrated BCR and TLR signalling, activated B cells migrate to the border of the T cell zone and the follicle, where they interact with CD4⁺ T cells. Although certain auto-antibody specificities can develop in MRL.Fas^lpr mice in a T cell-independent manner, CD4⁺ T cells facilitate extra-follicular B cell responses through interactions between CD40 and CD40L, interactions between inducible T cell co-stimulator (ICOS) and ICOS ligand (ICOSL), and the provision of interleukin-21 (IL-21)⁷¹ (FIG. 3).

Figure 3. T cell-dependent and T cell-independent extrafollicular activation pathways in autoimmunity.

Excess amounts of B cell-activating factor (BAFF) promote the survival of autoreactive B cells (part a), there by increasing the proportion of self-reactive B cells within the transitional and mature naive B cell compartments. Following engagement by self-antigen, autoreactive B cells undergo activation in response to dual B cell receptor (BCR)-dependent and Toll-like receptor (TLR)-dependent signalling. Subsequently, activated B cells differentiate into extrafollicular autoantibody-producing plasmablasts through either T cell-dependent (part b) or T cell-independent (part c) pathways. Activated B cells that engage cognate CD4⁺ T cells at the T cell–B cell border undergo class-switch recombination and differentiation into extrafollicular plasmablasts in response to co-stimulatory signals provided by T cells (part b), including CD40 ligand (CD40L; not shown), inducible T cell co-stimulator (ICOS; not shown) and interleukin-21 (IL-21). Following engagement by self-antigen, autoreactive B cells upregulate their surface expression of transmembrane activator and CAML interactor (TACI) in response to BCR-dependent signalling. TACI engagement by multimeric BAFF 60-mers (part c) promotes the expression of activation-induced cytidine deaminase (AID), resulting in somatic hypermutation, class-switch recombination and the subsequent generation of autoantibody-producing plasmablasts. Although additional B cell subsets may contribute to TACI-dependent autoantibody production in high BAFF settings, transitional B cells make a prominent contribution to the pool of IgG-producing plasmablasts in Baff-transgenic mice. BAFFR, BAFF receptor; TCR, T cell receptor.

Of the known MYD88-dependent receptors, the endosomal receptors TLR7 and TLR9 have prominent effects on the pathogenesis of autoimmunity. As predicted by in vitro studies^68,72, TLR7-deficient and TLR9-deficient MRL.Fas^lpr mice lack RNA-specific and dsDNA-specific autoantibodies, respectively^73,74. However, Tlr7 and Tlr9 deletions result in unexpected opposing effects on disease severity. Specifically, Tlr7 deletion is protective in several mouse lupus models, whereas Tlr9 deletion surprisingly exacerbates disease⁷⁴. Notably, TLR signalling might affect disease severity by facilitating dual BCR-mediated and TLR-mediated activation of autoreactive B cells, or by enhancing type I interferon (IFN) production by plasmacytoid dendritic cells^73,75. However, consistent with a dominant role for B cells, B cell-intrinsic Myd88 deletion abrogates the generation of antinuclear antibodies and nephritis in MRL.Fas^lpr mice⁷⁶.

TACI signalling is required for T cell-independent, BAFF-driven autoimmunity

Baff-transgenic mice develop extrafollicular plasmablasts, high-titre class-switched autoantibodies and systemic autoimmunity in the absence of CD4⁺ T cells⁶⁴. Although BAFF-driven autoimmunity requires B cell-intrinsic MYD88 function⁶⁴, the receptor that binds to BAFF and promotes autoantibody production had, until recently, not been identified. As discussed above, excess BAFF promotes B cell hyperplasia and the survival of autoreactive B cells in the naive repertoire, whereas BAFFR deficiency results in a loss of mature B cells^42,43,77, thereby implicating BAFFR in BAFF-driven autoimmunity. However, TACI signals through MYD88 (REF. 78) and promotes T cell-independent antibody responses⁷⁹, suggesting a role for TACI in this process. Consistent with this idea, two independent studies demonstrated the surprising finding that BAFF-driven autoimmunity is abrogated by Taci deletion^80,81.

Mechanistically, whereas TACI expression is usually limited to mature B cells, we noted a marked upregulation of TACI expression by CD21^lowCD24^hi transitional B cells in Baff-transgenic mice⁸¹. TACI expression identified a subset of activated, proliferating cells that were enriched for autoreactive specificities and that coordinately exhibited the expression of activation-induced cytidine deaminase (AID) and the production of class-switched, somatically mutated autoantibodies⁸¹. On the basis of these findings, we propose a model in which activated transitional B cells and/or naive mature B cells generate a pool of ASCs that produce pathogenic IgG autoantibodies in Baff-transgenic mice (FIG. 3). BAFF levels are increased in many patients with autoimmune disease (including those with SLE, RA and systemic sclerosis), which suggests that BAFF-driven TACI signalling may promote pathogenic plasmablast formation⁴⁰, although this possibility has not yet been examined in humans. In addition, dysregulated transitional B cell activation is likely to be relevant in other clinical scenarios that are characterized by high BAFF levels, including autoimmune disease relapse after B cell-depletion therapy^82,83 and de novo humoral autoimmunity following stem cell transplantation⁸⁴. Notably, as systemic infectious challenges frequently promote local BAFF generation⁸⁵, TACI-dependent activation of transitional B cells may contribute to rapid, T cell-independent antibody responses to blood-borne pathogens.

Autoimmune GC B cell responses

GC B cell responses during autoimmunity

In addition to the extrafollicular B cell responses described above, several lines of evidence indicate that spontaneous GCs are an important site of autoreactive B cell activation in autoimmune diseases^6,86. For example, autoantibody-producing B cells cloned from patients with SLE are frequently class-switched and exhibit extensive SHM⁸⁷. Although SHM can occur within extrafollicular foci in autoimmunity, these findings implicate autoantigen-driven GC selection in the generation of high-affinity, autoantibody-producing B cells. Consistent with this idea, spontaneous GCs are observed in several mouse lupus models, and the B cell-intrinsic signals required for autoantibody production are also necessary for spontaneous GC formation^{50,76,88–90}. In addition, tertiary lymphoid structures and ectopic GCs develop within inflamed tissues in human autoimmunity, including kidneys in lupus nephritis and arthritic joints in rheumatoid arthritis^86,91. Finally, serum autoantibody titres increase years before disease onset and persist despite aggressive immunosuppression, implicating GC-derived long-lived plasma cells in the pathogenesis of autoimmune disease.

Most importantly, in addition to GCs being a site of autoreactive plasma cell generation, emerging evidence demonstrates that autoreactive B cells can direct the formation of spontaneous GCs by initiating the expansion of cognate T follicular helper (T_FH) cells. In this model, rather than being a downstream consequence of self-reactive T cells, dysregulated B cell signalling functions as the primary driver of disease initiation (FIG. 4). On the basis of this paradigm, we review the specific B cell immune mechanisms that underlie the formation of spontaneous autoimmune GCs. Where possible, we focus on the specific signalling pathways that have been shown to promote autoimmunity in a B cell-intrinsic manner.

Figure 4. Self reactive B cells initiate autoimmune germinal centre formation by facilitating breaks in T cell tolerance.

a | After binding to self-antigen (either soluble or bound to antigen-presenting cells) derived from apoptotic particles or other disease-specific targets, autoreactive B cell receptors (BCRs) traffic nuclear antigens to the endosomal receptors Toll-like receptor 7 (TLR7) and TLR9, resulting in initial B cell activation in response to integrated BCR-dependent and TLR-dependent signalling. In parallel, endolysosomal enzymes also process internalized self-antigens (including a broad range of nucleic acid-associated proteins) into peptides for loading onto MHC class II. b | B cells function as antigen-presenting cells to present MHC class II-bound peptides to cognate self-reactive CD4⁺ T cells at the T cell–B cell border of lymphoid follicles. Together with co-stimulatory signals provided by CD80 and/or CD86 and inducible T cell co-stimulator ligand (ICOSL), self-reactive B cells initiate breaks in CD4⁺ T cell tolerance. Activated CD4⁺ T cells subsequently express CXC-chemokine receptor 5 (CXCR5) and B cell lymphoma 6 (BCL-6), resulting in their migration to the B cell follicle as early T follicular helper (T_FH) cells (not shown). Activated B cells also produce interleukin-6 (IL-6), which may facilitate T_FH cell differentiation by inducing BCL-6 expression, although this has not yet been directly tested. c | T_FH cells promote germinal centre (GC) formation through the production of IL-21, which sustains B cell BCL-6 expression and promotes B cell activation, class-switch recombination and plasma cell differentiation. In autoimmune settings, interferon-γ (IFNγ; probably derived from activated CD4⁺ T cells) drives GC formation in a B cell-intrinsic, signal transducer and activator of transcription 1 (STAT1)-dependent manner, in part by enhancing BCL-6 expression. IFNγ also promotes B cell-intrinsic expression of the transcription factor T-bet (encoded by TBX21), which is required for class-switch recombination to pathogenic IgG2a and IgG2c isotypes, but is redundant for IFNγ-driven GC formation. Although not yet directly tested in autoimmune models, this dysregulated GC response is probably affected by additional cytokines, including B cell-activating factor (BAFF), which promotes the selection of high-affinity GC B cell clones, and IL-12, which facilitates T cell IFNγ production and T_FH cell differentiation. d | Iterative interactions between GCB cells and cognate T_FH cells within ongoing autoimmune GCs probably result in epitope spreading and the recruitment of additional autoreactive T cell and B cell clones. BAFFR, BAFF receptor; CD40L, CD40 ligand; ICOS, inducible T cell co-stimulator; IFNγR, IFNγ receptor; IL-6R, IL-6 receptor; TCR, T cell receptor.

Dysregulated BCR signalling promotes GC-dependent systemic autoimmunity

Studies in several independent animal models have clearly shown that dysregulated BCR-dependent signalling is sufficient to promote autoimmunity in a B cell-intrinsic manner. For example, patients with WAS exhibit high rates of humoral autoimmunity^92,93. To examine the cell-intrinsic mechanism that underlies autoimmunity in WAS, we generated BM chimaeras in which B cells, but not other haematopoietic lineages, lacked expression of the WAS protein⁵⁰. In this setting, Was^−/− B cells initiated spontaneous humoral autoimmunity characterized by high-titre class-switched antinuclear antibodies, T_FH cell expansion, spontaneous GC formation and the development of progressive immune-complex-mediated glomerulonephritis. Subsequently, Recher et al.⁹⁴ demonstrated that conditional deletion of Was in B cells also resulted in the formation of spontaneous GCs and the production of class-switched autoantibodies. Mechanistically, Was^−/− B cells are modestly hyper-responsive to both BCR and TLR activation, which suggests that dysregulated dual BCR and TLR signalling may be sufficient to promote autoantibody production.

In support of this idea, B cell-intrinsic deletion of the SRC family tyrosine kinase-encoding gene Lyn also promotes spontaneous autoimmunity⁹⁵. Notably, Lyn^−/− B cells exhibit enhanced BCR-mediated calcium flux⁹⁶, which is probably explained by the role of LYN in driving immunoreceptor tyrosine-based inhibition motif (ITIM)-mediated inhibitory signalling by CD22 or Fcγ receptor IIB (FcγRIIB). These observations are relevant beyond mouse models, as polymorphisms in LYN and the related kinase BLK are associated with SLE^97–99. In addition, rare loss-of-function mutations in SIAE (which encodes sialic acid acetyl esterase) — an enzyme that is required for post-translational modifications that facilitate CD22 activity — increase the risk of an individual developing SLE, T1D and RA¹⁰⁰.

Similarly, the autoimmunity-associated PTPN22^R620W variant also modulates BCR signalling^21–23. To elucidate the mechanisms that drive autoimmune disease, two groups independently generated mouse models in which the gene that encodes the orthologous risk variant in mouse phosphatase (Pep) was knocked in^28,29. Both knock-in strains exhibited enhanced BCR and TCR signalling, spontaneous GC formation and an age-related expansion of effector memory T cells. However, systemic autoimmunity was only observed in mice that had a mixed C57BL/6J×129/Sv background, which potentially reflects the influence of signalling lymphocyte activation molecule (SLAM) family polymorphisms derived from the 129 background on autoimmune GC responses^101,102.

As the autoimmunity-associated PTPN22^R620W variant might promote autoimmunity through T cell-dependent or B cell-dependent mechanisms, we generated a separate strain to evaluate B cell-intrinsic expression of this risk-associated variant. Remarkably, aged animals developed spontaneous GCs and produced high-titre autoantibodies, which indicates that B cell-specific expression of the risk-associated variant was sufficient to promote a break in B cell tolerance²⁸. Notably, consistent with the ability of modest alterations in BCR signal strength to promote an autoimmune GC response, transgenic overexpression of the TEC family kinase BTK, which is essential for BCR-triggered phospholipase Cγ2 (PLCγ 2) activation, was sufficient to trigger analogous events and lead to T cell-dependent autoantibody production¹⁰³.

Together, these studies demonstrate the crucial importance of dysregulated BCR signalling in driving humoral autoimmunity, both by modulating the pre-immune B cell repertoire and by facilitating the activation of autoreactive B cells in the periphery.

B cell antigen presentation in autoimmunity

B cell-targeted therapies are approved by the US Food and Drug Administration for the treatment of RA¹⁰⁴, anti-neutrophil cytoplasmic autoantibody-associated vascu-litis¹⁰⁵ and SLE⁴¹, and the therapeutic benefits of these treatments have also been observed in a diverse range of human autoimmune diseases, including T1D¹⁰⁶, primary membranous nephropathy¹⁰⁷, pemphigus vulgaris¹⁰⁸ and multiple sclerosis¹⁰⁹. However, although clinical improvement correlates with the decrease in autoantibody titres in certain diseases^108,110, B cell depletion frequently results in clinical responses despite persistent serum autoantibody titres^111,112. These observations suggest that additional B cell functions, including MHC class II-dependent antigen presentation or cytokine production, have important roles during the pathogenesis of autoimmune disease.

Initial evidence for antibody-independent B cell effector functions in SLE came from seminal studies using the MRL.Fas^lpr model. Specifically, B cell-deficient MRL.Fas^lpr mice lacked T cell infiltrates in kidneys, whereas MRL.Fas^lpr mice in which B cells express surface but not secreted IgM developed prominent nephritis in the absence of serum autoantibodies^113,114. Although these studies suggested important roles for B cell antigen presentation in the pathogenesis of SLE, direct evidence of B cell antigen-presenting cell (APC) activity promoting autoimmunity was limited to a few models. For example, in the non-obese diabetic mouse (NOD mouse) model of T1D, B cell-intrinsic loss of MHC class II prevented CD4⁺ T cell activation and the development of diabetes; findings that are consistent with studies showing the antibody-independent development of diabetes in secretory IgM-deficient NOD mice^115,116. Similarly, in the experimental autoimmune encephalitis model of multiple sclerosis, B cell-intrinsic MHC class II-dependent antigen presentation, rather than myelin oligodendrocyte glycoprotein-specific antibody production, was required for progressive neurological disease¹¹⁷.

To test whether B cell-intrinsic antigen presentation was similarly required to initiate CD4⁺ T cell activation and spontaneous GC formation in SLE, we generated WAS chimaeras in which all B cells lacked MHC class II⁹⁰. Remarkably, the loss of B cell APC function abrogated the expansion of activated effector memory T cells and the development of spontaneous GCs. These findings were corroborated by an independent study in which B cell-intrinsic Cre-mediated deletion of MHC class II in MRL.Fas^lpr mice prevented CD4⁺ T cell activation and the production of class-switched autoanti-bodies¹¹⁸. Together, these findings confirm that B cells that function as APCs directly initiate CD4⁺ T cell activation in mouse models of SLE, and correspondingly, that a loss of B cell APC activity may explain the antibody-independent benefits of B cell depletion therapy. Consistent with this idea and its relevance beyond SLE, spontaneous anti-insulin GCs are generated in mice that have anti-insulin transgenic T cells and B cells through a process that requires GC B cells to present a repertoire of insulin epitopes derived from circulating insulin¹¹⁹.

B cell-intrinsic TLR signalling in autoimmune GCs

In addition to its role in extrafollicular autoantibody responses, dysregulated TLR signalling is implicated in GC-dependent autoimmunity. For example, B cell-intrinsic MYD88 signalling is crucial for spontaneous GC formation and antinuclear antibody production in several mouse lupus models^50,89,120. Of the MYD88-dependent TLRs, B cell TLR7 signalling is most prominently linked with GC-driven autoimmunity. For example, TLR7 overexpression in lupus-prone mouse strains that carry the Yaa translocation or in Tlr7-transgenic mice results in systemic autoimmunity characterized by spontaneous GC formation^121,122. Moreover, polymorphisms in TLR7 have been linked to the development of SLE in human candidate gene studies^123–125. These effects on autoimmune GCs are most likely explained by B cell-intrinsic TLR7 over-expression, as Tlr7-transgenic B cells are preferentially expanded in the GC relative to wild-type B cells in competitive chimaeras¹²².

However, although B cell-intrinsic TLR9 expression is required for the formation of antinucleosome antibodies¹²⁶, the relative effect of TLR7 signalling versus TLR9 signalling in autoimmune GC responses had not been assessed. For this reason, we generated WAS chimaeras with B cells that were deficient in either TLR7 or TLR9 (REF. 88). As predicted, B cell-intrinsic Tlr7 and Tlr9 deletion prevented the production of RNA-associated and DNA-associated autoantibodies, respectively. However, whereas B cell-intrinsic TLR7 deficiency prevented spontaneous GC formation and systemic autoimmunity, disease severity was notably exacerbated by B cell-intrinsic Tlr9 deletion. Interestingly, spontaneous GCs also develop in non-autoimmune C57BL/6 mice with increasing age in a manner that is dependent on B cell-intrinsic TLR7 signalling but is restrained by TLR9 activation¹²¹. Although the mechanisms by which TLR9 activation restrains autoimmunity remain elusive, these studies demonstrate that the protective roles of TLR9 rely on B cells and not myeloid subsets.

Of relevance to human disease pathogenesis, genetic variants that influence TLR signalling pathways — including TNFAIP3 (which encodes tumour necrosis factor-induced protein 3), TNIP1 (which encodes TNFAIP3-interacting protein 1), ATG5 (which encodes autophagy-related gene 5), IRF5 (which encodes IFN-regulatory factor 5), IRF7 and SLC15A4 (which encodes solute carrier family 15 member 4) — are highly associated with the risk of developing SLE^127,128. Although the effects of individual polymorphisms on downstream signalling and the key cell lineages involved remain to be determined, these findings from GWAS emphasize the crucial importance of dysregulated TLR responses in the pathogenesis of systemic autoimmunity.

IL-6, IL-21 and BAFF in autoimmune GC responses

In addition to cognate interactions between GC B cells and T_FH cells, cytokine signalling markedly influences GC biology, both during infectious responses and in autoimmunity. The pleiotropic cytokine IL-21 facilitates GC formation by promoting the cell-intrinsic activation and differentiation of T_FH cells and GC B cells^129–131. Consistent with important roles for IL-21 in the pathogenesis of autoimmune disease, serum IL-21 levels are increased in mouse and human lupus^132–135. In addition, Il21r deletion or therapeutic IL-21 blockade limits autoimmunity in the MRL.Fas^lpr mice and BXSB-Yaa mice^132,136,137 in a manner that is dependent on B cell-intrinsic IL-21R activation¹³⁸.

In cooperation with IL-21, the pro-inflammatory cytokine IL-6 promotes the formation of GC B cells and T_FH cells. Although an absolute requirement for IL-6 in T_FH cell generation is controversial^{131,139–141}, IL-6 is known to promote early T_FH cell differentiation by transiently inducing B cell lymphoma 6 (BCL-6) expression in CD4⁺ T cells binding cognate antigen¹⁴². In addition, Il6 deletion reduces disease severity in the BXSB-Yaa mouse lupus model, and increased serum IL-6 levels have been observed in human SLE, which suggests potential roles for this cytokine in disease pathogenesis^143,144. Consistent with this idea, IL-6 receptor blockade with tocilizumab decreased the levels of dsDNA-specific antibodies and circulating plasmablasts in early-stage clinical trials in patients with SLE¹⁴⁵. Interestingly, in addition to dendritic cells and other myeloid subsets, activated B cells produce IL-6, which suggests that B cell-derived IL-6 might initiate T_FH cell development and spontaneous GC formation in SLE^146,147, although this hypothesis has not yet been directly tested.

Finally, although systemic autoimmunity in Baff-transgenic mice occurs independently of T cells, T_FH cells are an important source of local BAFF production within GCs^148,149. Although TFH cell-derived BAFF is redundant for GC formation, T cell-derived BAFF promotes the selection of high-affinity GC B cell clones, thereby implicating BAFF in the pathogenesis of GC-dependent autoimmunity.

IFN signalling in autoimmune GC responses

In addition to important functions during antiviral immune responses, the type I IFN cytokine family (comprising IFNα, IFNβ, IFNε and IFNω) has been implicated in the pathogenesis of systemic autoimmunity. For example, a prominent subset of patients with SLE exhibit increased levels of type I IFN-induced transcripts in peripheral blood mononuclear cells, a phenomenon termed the ‘interferon signature’ (REFS 150–152). Mechanistically, TLR-dependent activation of plasmacytoid dendritic cells by nucleic acid-containing immune complexes has been proposed to propagate SLE through increased type I IFN production. Consistent with this idea, human genetic variants that affect type I IFN production or signalling — including variants in IRF5, STAT4 (which encodes signal transducer and activator of transcription 4) and TREX1 (which encodes 3′-repair exonuclease 1) — are associated with an increased risk of developing SLE, whereas type I IFN receptor deficiency ameliorates disease in some mouse lupus models^153,154. Importantly, in addition to type I IFNs, type II IFN (IFNγ) signalling has also been implicated in the pathogenesis of SLE in animal models and human gene expression studies^{150,155–157}.

Recently, human and animal studies have begun to reconcile these overlapping roles for type I and type II IFNs in the pathogenesis of SLE. Whereas B cell-intrinsic type I IFN signalling is redundant for humoral autoimmunity in WAS chimaeras, IFNγ promotes GC-dependent autoantibody production in the Roquin^san/san, B6.Sle1b and WAS chimaera lupus models^90,157,158. Spontaneous GC formation required B cell-intrinsic and T cell-intrinsic IFNγ receptor (IFNγR) signalling for the generation of GC B cells and T_FH cells, respectively. Mechanistically, although IFNγ-driven upregulation of the T-box transcription factor T-bet was required for the production of pathogenic IgG2c autoantibodies, neither complete nor B cell-intrinsic T-bet deficiency affected spontaneous GC formation⁹⁰. Instead, IFNγ synergized with co-stimulatory signals to promote cell-intrinsic expression of the GC master regulator BCL-6 by both CD4⁺ T cells and B cells^90,157. In contrast to these crucial effects of IFNγ on the pathogenesis of mouse lupus, B cell-intrinsic type I IFN signalling accelerates, but is not required for, the production of class-switched autoantibodies and spontaneous GC formation in the WAS model⁹⁰.

Remarkably, these findings precisely model recent longitudinal studies that showed that initial SLE-associated autoantibodies develop years before clinical symptoms and that their appearance correlates with increased serum IFNγ levels¹⁵⁹. By contrast, the type I IFN signature typically develops shortly before the diagnosis of SLE, years after the first detection of antinuclear antibodies^159,160. Thus, increased levels of type I IFNs probably propagate autoimmunity through feed-forward mechanisms before the clinical diagnosis of SLE^150–152, whereas B cell-intrinsic IFNγ signalling may be crucial for initial breaks in B cell tolerance that result in spontaneous GC formation, autoantibody production and a loss of T cell tolerance.

Importantly, the paradigm in which the serial accumulation of disease-associated autoantibodies precedes an increase in pro-inflammatory cytokines and disease onset is not unique to SLE. Rather, antibodies that are specific for islet cell antigens (including 65 kDa glutamic acid decarboxylase (GAD65; also known as glutamate decarboxylase 2), insulinoma-associated protein 2 (IA-2) and insulin) and cyclic citrullinated peptides develop years before clinical symptoms in T1D and RA, respectively^161,162. Although the effect of specific cytokines on disease progression in T1D is less well understood (probably because such data would require local tissue sampling in the setting of organ-specific autoimmunity), an increase in the serum levels of pro-inflammatory cytokines predicts the imminent development of active RA¹⁶³. In this context, there is increasing interest in examining whether immunomodulatory treatments given to at-risk individuals with latent autoimmunity can prevent progression to symptomatic disease. On the basis of the hypothesis that hydroxychloroquine might inhibit endolysosomal TLR signalling or processing of autoantigens for MHC class II-dependent presentation, clinical trials are in progress, or in development, to examine whether hydroxychloroquine treatment of asymptomatic, autoantibody-positive individuals can prevent or delay the development of RA (the StopRA trial¹⁶⁴; ClinicalTrials.gov: NCT02603146) and T1D (run by the T1D TrialNet consortium; C. Greenbaum, personal communication), respectively. Thus, future studies that address the specific immune mechanisms and cytokine signalling pathways that are responsible for the initiation versus the progression of humoral autoimmunity may ultimately hold the promise of establishing treatments to prevent autoimmune disease in at-risk individuals.

Conclusions and future perspectives

Emerging data from animal and human studies demonstrate that even modest alterations in B cell-intrinsic signalling programmes can be sufficient to promote breaks in T cell tolerance and initiate systemic autoimmunity. On the basis of these data, we propose a model in which GWAS-identified autoimmunity-associated variants and/or other genetic alterations in B cell gene products promote humoral autoimmunity through several distinct mechanisms. First, altered B cell signalling programmes increase the risk of autoimmunity by modulating positive and negative selection during B cell development, culminating in an increase in the proportion of mature B cells that exhibit autoreactivity. However, although contributing to disease risk, it is unlikely that increased autoreactivity alone is sufficient for disease development. As a crucial second step, autoantibody-producing B cells are generated from mature or naive B cells through parallel programmes that lead to extrafollicular and GC-dependent B cell activation, respectively. As described here, dual BCR and TLR signalling is crucial for the generation of short-lived, extrafollicular plasmablasts in a process that is markedly facilitated by BAFF-driven TACI signalling. In parallel, this BCR-driven and TLR-driven activation programme can also promote interactions between cognate T cells and B cells, which result in T_FH cell differentiation and the formation of autoimmune GCs, leading to a break in T cell tolerance. As part of this iterative process, rounds of B cell antigen presentation allow for the display of a broad range of self-antigen-derived epitopes, and this probably promotes epitope spreading and the activation of additional autoreactive T cell and B cell clones (FIG. 4d). Finally, autoimmune GCs generate long-lived cell populations that are crucial for the maintenance and propagation of systemic autoimmunity (FIG. 5). Long-lived plasma cells derived from autoimmune GCs secrete pathogenic class-switched autoantibodies. In addition, autoreactive GC B cells can differentiate into IgM⁺ or class-switched memory B cells that can participate in the formation of new autoimmune GCs and/or self-reactive extrafollicular B cell responses. Finally, autoimmune GCs probably promote the generation of self-reactive memory T_FH cells that can initiate the formation of new autoimmune GCs and/or instigate T cell-dependent extrafollicular responses165,166, events that probably no longer require triggering by self-reactive B cells.

Figure 5. A sequential model of the progression of systemic autoimmunity.

After initial breaks in B cell and T cell tolerance, long-lived effector and memory populations are generated within spontaneous, autoimmune germinal centres (GCs) and contribute to the maintenance and progression of systemic autoimmunity. After iterative rounds of B cell proliferation and selection, B cells that express high-affinity B cell receptors (BCRs) exit the GC as either plasma cells or memory B cells. a | In response to CXC-chemokine receptor 4 (CXCR4)-dependent signalling, autoantibody-secreting plasma cells traffic to the bone marrow, where they exhibit long-term survival within bone marrow stromal niches. b | Alternatively, GC B cells can differentiate into autoreactive memory B cells. Long-lived memory B cells probably contribute to the progression of autoimmunity and to disease relapse following treatment owing to an increased propensity for either differentiation into short-lived, autoantibody-secreting, extrafollicular plasmablasts, or recruitment to and activation within secondary autoimmune GCs. c | In addition to memory B cells, emerging evidence suggests that a subset of T follicular helper (T_FH) cells differentiate into memory T_FH cells that can seed secondary GCs and promote the progression of autoimmunity. Importantly, after these initial breaks in T cell tolerance and the establishment of autoimmune T cell memory, additional naive B cell clones that exhibit self-reactivity may be recruited into established GCs in the absence of initial priming signals, thereby broadening the range of high-affinity autoantibodies that are generated. Dashed arrows indicate the movement of long-lived and other plasma cell populations to bone marrow or other lymphoid compartments, respectively.

Key open questions that are relevant to these models include the issue of how specific GWAS variants influence these events and lead to altered tolerance. Rather than a series of linear events, it is likely that variants with greater effect may function across multiple signalling cascades, leading to alterations in the repertoire as well as in cell activation, and some variants may exert this effect across multiple immune cell lineages. It will be important to design studies that specifically define B cell-intrinsic and alternative lineage-intrinsic effects. Moreover, multiple autoimmunity-associated polymorphisms show evidence of selection during human evolution, which may reflect a trade-off between an enhanced immune response to infection and an increased risk of autoimmunity^167,168. For example, the PTPN22^R620W polymorphism may provide increased protection against tuberculosis¹⁶⁹, and SLE-associated polymorphisms in FCGR2B (which encodes FcγRIIB) protect against severe Plasmodium falciparum infection¹⁷⁰. We hypothesize that, by increasing the breadth of the naive B cell repertoire and by lowering B cell activation thresholds, genetic variants associated with autoimmunity also facilitate rapid antibody responses to pathogens and protect against infectious challenge. Gaining a better understanding of the basis of positive selection will provide insights into the autoimmune events that are driven by such ‘friendly fire’.

Importantly, for the vast majority of disease-associated genetic variants the mechanisms that influence immune responses remain undefined. In addition, two or more autoimmunity-associated variants may function in an additive and/or synergistic manner. In this regard, recent advances in gene editing techniques will probably facilitate mechanistic studies by, for example, allowing the introduction of multiple human-relevant risk alleles into embryonic stem cells or zygotes from mouse disease models or the direct editing of primary human haematopoietic cells. Finally, although the BCR and co-receptor signalling pathways are frequently considered individually, as highlighted here, there is extensive crosstalk between these pathways, and these coordinate signals are integral to B cell activation. Therefore, individual genetic polymorphisms will probably affect multiple signalling cascades, potentially across a range of immune cell types. New technologies, including single-cell transcriptome analysis, phospho-mass-spectroscopy and improved bioinformatic tools, have begun to address some of this complexity. In addition to mechanistic insights, these next-generation approaches may facilitate the identification of specific pathways that drive disease in individual patients. Ultimately, this more precise understanding of disease mechanisms may inform both individualized treatment decisions and the development of novel targeted therapies for multiple autoimmune and other disorders.

Glossary

CD40

A tumour necrosis factor receptor superfamily member expressed by antigen-presenting cells that transmits activation signals in response to CD40 ligand (CD40L). Binding of CD40 on transitional B cells to CD40L on naive CD4⁺ T cells activates both the classical and alternative nuclear factor-κB pathways to promote cell survival

Humoral autoimmunity

Causes a broad range of autoimmune diseases that are characterized by the production of pathogenic antibodies. These disorders frequently share risk alleles. For example, the R620W polymorphism in PTPN22 (which encodes protein tyrosine phosphatase nonreceptor 22) is associated with an increased risk of developing several diseases, including rheumatoid arthritis, type 1 diabetes, systemic lupus erythematosus, Hashimoto thyroiditis, Addison disease and others

Extrafollicular B cell activation

In response to infectious challenge, B cell receptor signalling induces the rapid division and differentiation of B cells into short-lived, antibody-secreting plasmablasts at the T cell zone–red pulp border. This extrafollicular B cell response can occur in a T cell-independent or T cell-dependent manner, and is the predominant source of the protective antibodies that are generated early during an infection

Germinal centre

(GC). A specialized lymphoid structure in which activated B cells cycle between anatomically distinct dark zones and light zones as they undergo iterative rounds of affinity selection, which ultimately leads to the generation of both antigen-specific memory B cells and high-affinity, long-lived plasma cells

B cell-activating factor receptor

(BAFFR). A tumour necrosis factor receptor superfamily member that promotes the survival and maturation of B cells. Following ligation by B cell-activating factor (BAFF), BAFFR mainly induces the alternative nuclear factor-κB pathway to promote pro-survival signals. Mice that are deficient in BAFF or BAFFR have a block in B cell development at the transitional type 1 stage

Systemic lupus erythematosus

(SLE). A chronic, multisystem inflammatory disease that is characterized by high-titre, class-switched autoantibodies against nuclear antigens; these autoantibodies are invariably present before disease onset and ultimately lead to the formation of immune complexes that precipitate systemic and/or organ-targeted injury

Type 1 diabetes

(T1D). An autoimmune disease that is characterized by the destruction of insulin-producing pancreatic islet cells. Although T cells are crucial in this process, high-affinity, somatically mutated autoantibodies that target pancreatic islet antigens are present before disease onset, which emphasizes the importance of B cells in the pathogenesis of T1D

Rheumatoid arthritis

(RA). A chronic, autoimmune disease that is characterized by symmetrical polyarticular arthritis and the production of rheumatoid factor and autoantibodies against cyclic citrullinated peptides; the appearance of rheumatoid factor and autoantibodies precedes clinical disease, implicating altered B cell responses early in the disease process

Transmembrane activator and CAML interactor

(TACI). A tumour necrosis factor receptor family member that activates the classical nuclear factor-κB (NF-κB) pathway in response to ligation by multimeric B cell-activating factor (BAFF) or a proliferation-inducing ligand (APRIL; also known as TNFSF13) on developing and mature B cells. TACI signalling promotes T cell-independent antibody responses, and a subset of patients with common variable immunodeficiency have TACI mutations. TACI-deficient animals develop B cell hyperplasia, probably owing to increased serum BAFF levels

Immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome

(IPEX syndrome). A syndrome caused by a lack of functional regulatory T cells that is due to mutations in FOXP3 (which encodes forkhead box P3). It is characterized by severe autoimmunity driven by both activated effector T cells and multiple autoantibodies

MRL.Fas^lpr

A strain of mouse generated by breeding multiple mouse strains to select those that exhibit features of lupus-like disease, including lethal autoimmunity and high-titre antinuclear antibodies. The lpr mutation is an autosomal-recessive mutation in Fas, which results in lymphoproliferation and accelerated autoimmunity

NZB/W mice

F1 mice that are hybrids of NZB (New Zealand black) and NZW (New Zealand white) strains. They develop class-switched antinuclear antibodies, systemic inflammation and immune-complex-mediated glomerulonephritis. Autoimmune susceptibility loci in this strain include polymorphisms in Fcgr2b (which encodes Fcγ receptor IIB) and in signalling lymphocyte activation molecule (SLAM) family genes

Non-obese diabetic mouse

(NOD mouse). A polygenic animal model of type 1 diabetes characterized by spontaneous leukocyte infiltration within the pancreatic islets and by insulin-dependent diabetes, with a bias towards accelerated disease in female mice. Disease in this model depends on both B cell and T cell activation

Tlr7-transgenic mice

A mouse model of lupus generated by transgenic overexpression of the gene that encodes Toll-like receptor 7 (TLR7), which results in systemic autoimmunity and high titres of RNA-specific autoantibodies. The phenotype of Tlr7-transgenic mice partially mimics that of lupus-prone strains that express duplicated Tlr7 through the Y-chromosome-linked autoimmune accelerator (Yaa) mutation

BXSB-Yaa

Lupus-prone male mice that develop accelerated disease due to Y-chromosome-linked autoimmune accelerator (Yaa), a translocation of the telomeric end of the X chromosome to the Y chromosome, which results in the duplication of several genes, including Tlr7 (which encodes Toll-like receptor 7)

Roquin^san/san

A strain of mice with an N-ethyl-N-nitrosourea (ENU)-driven ‘san’ mutation in Roquin (which encodes a RING-type ubiquitin ligase; also known as Rc3h1), resulting in a lupus-like disease that is characterized by spontaneous germinal centre formation, high-affinity double-stranded DNA-specific antibodies, autoimmune thrombocytopenia and immune-complex-mediated glomerulonephritis

B6.Sle1b

A strain of C57BL/6 mice that has the Sle1b sublocus derived from lupus-prone NZW mice. The Sle1b sublocus contains signalling lymphocyte activation molecule (SLAM) family genes with polymorphisms that are involved in antinuclear antibody production, B cell and T cell activation, and spontaneous germinal centre formation.