Autoreactive B cells in systemic lupus erythematosus: insights from integrative multi-omics analyses

Abstract

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the emergence of autoantibodies and deposition of immune complexes. SLE presents with heterogeneous multi-organ involvement that varies among patients and its mechanisms have been investigated to facilitate appropriate stratification and treatment selection. Bulk RNA-seq and bulk ATAC-seq have provided important insights into the pathogenesis of SLE; however, these approaches are inherently limited by their reliance on predefined cell subsets and known markers, which can introduce bias and restrict their ability to fully resolve cellular heterogeneity. Recent advances in multi-omics analyses have enabled the investigation of multi-layered information beyond single-cell RNA sequencing (scRNA-seq) alone and have contributed substantially to elucidating the pathogenesis of SLE. Although the emergence of autoreactive B cells and the production of autoantibodies in SLE are well established, the mechanisms of evasion from negative selection remain unclear. Multi-omics analyses have revealed key aspects of SLE pathogenesis, particularly the expansion of atypical B cells (ABCs), an autoreactive population driven by extrafollicular pathways. Furthermore, beyond transcriptional profiling, multi-omics analysis has emerged as an additional investigative method, which combines scRNA-seq with other modalities. Spatial analyses, for instance, have provided critical insights into the tissue localization and persistence of autoreactive B cells within inflammatory niches in lupus nephritis, suggesting that local microenvironments contribute to treatment resistance. Additionally, B cell receptor (BCR) sequencing has revealed distinctive BCR features in SLE, such as reduced somatic hypermutation, increased repertoire naiveness, and increased immunoglobulin variable region heavy chain gene (IGHV4-34) usage in B cells in bone marrow and affected organs (not only in peripheral blood). Furthermore, emerging multimodal approaches integrating spatial, transcriptomic, and epigenomic information further highlight pathogenic cell–cell interactions and inflammatory circuits that cannot be captured by scRNA-seq alone. In this review, we summarize recent multi-omics studies that elucidate the origin, differentiation, and tissue localization of pathogenic B cells in SLE. We provide an overview of the multi-omics analyses focusing on B cells so far, especially at single-cell resolution, and discuss their possible applications in precision medicine.

Background

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the emergence of autoantibodies and deposition of immune complexes in multiple organs [1]. The prevalence of SLE is estimated to be 3–50 cases per 100,000 individuals [2], and the standardized mortality ratio remains high at 2.65 (95% CI 2.13–3.26) [3]. Since the introduction of corticosteroids as a treatment for SLE, patient prognosis has improved markedly. However, corticosteroids are associated with adverse effects such as infections, diabetes, and osteonecrosis, and they cannot be administered indefinitely. Additional therapeutic options have been introduced, including immunosuppressive agents such as tacrolimus and mycophenolate mofetil, as well as biologic agents such as rituximab and belimumab [4]. Nevertheless, a subset of patients remains refractory to these treatments, and treatment resistance is difficult to predict a priori. Therefore, disease stratification based on a more precise understanding of disease pathogenesis is required (Fig. 1) [5].

Fig. 1.

Potential of multi-omics analyses for the stratification of SLE. The introduction of corticosteroids, immunosuppressive agents, or biologic agents as a treatment for SLE improved their prognosis. However, some patients remain refractory to these treatments, and treatment resistance is difficult to predict in advance. Therefore, disease stratification based on a more precise understanding of disease pathogenesis using multi-omics analysis is required. CAR-T, chimeric antigen receptor T-cell

SLE develops through a complex interplay of genetic and environmental factors. For instance, studies have shown that the concordance rate of SLE is approximately 25% in monozygotic twins, whereas it is only about 2% in dizygotic twins [1]. This fact indicates that although genetic factors contribute to the development of SLE, they alone are insufficient to fully explain disease onset. Other approaches for interrogating causative genetic factors include genome-wide association studies (GWAS), which can identify genetic variants that are statistically associated with specific traits or diseases across many genomes [6]. Although GWAS can identify disease-associated genetic variants and robustly demonstrate causal relationships, it provides limited insight into the mechanisms by which these variants contribute to disease pathogenesis. Expression quantitative trait locus (eQTL) analysis is one approach that can overcome this limitation (Fig. 2). eQTL analysis is a method used to determine the effects of genetic variants on gene expression levels. eQTL can serve as a “bridge” to understand how genetic variants influence phenotypes by using mRNA expression as an intermediate phenotype. eQTLs are commonly classified into cis-eQTLs, which affect the expression of genes located near a given single nucleotide polymorphism (SNP), typically within ~ 1 Mb, and trans-eQTLs, which influence genes at more distal loci, including those located outside the cis window or on different chromosomes [7]. By integrating GWAS with bulk RNA-seq data, eQTL analysis can be performed to quantify the extent to which specific genetic variants influence mRNA expression levels. Ota et al. at the University of Tokyo demonstrated this type of analysis when they constructed the Immune Cell Gene Expression Atlas (ImmuNexUT). ImmuNexUT is a large-scale resource comprising bulk RNA-seq profiles from 28 distinct immune cell subsets together with matched whole-genome sequencing data from each individual, encompassing 337 patients with 10 categories of immune-mediated diseases and 79 healthy controls (HCs). By integrating bulk RNA-seq data with whole-genome sequencing data, they were able to perform comprehensive eQTL analyses across individual immune cell subsets. This analysis revealed that cell-type-specific and context-dependent eQTLs showed significant enrichment in immune disease–associated genetic variants and implicated disease-relevant cell types, genes, and environmental factors [8].

Fig. 2.

eQTL analysis quantifies the effect of genetic variants on gene expression levels. eQTLs are commonly classified into cis-eQTLs, which affect the expression of genes located near a given single nucleotide polymorphism (SNP), typically within ~ 1 Mb, and trans-eQTLs, which influence genes at more distal loci, including those located outside the cis window or on different chromosomes. eQTL analysis is a method used to determine the effects of genetic variants on gene expression levels, and it enables the evaluation of how genetic variants influence phenotypes. eQTL, expression quantitative trait locus; SNP, single nucleotide polymorphism; GWAS, genome-wide association studies

However, because bulk RNA-seq uses tissues or cell populations as input material, it is limited by the averaging of heterogeneous cell populations and therefore lacks the resolution to detect differences between individual cells [9]. Single-cell RNA sequencing (scRNA-seq), on the other hand, enables the analysis of mRNA expression at single-cell resolution and can detect cellular heterogeneity by performing unbiased clustering, overcoming the limitations of bulk RNA-seq.

Accordingly, detailed analyses using scRNA-seq have become increasingly widespread in recent years. In this review, we will highlight major scRNA-seq studies focusing on B cells in SLE. It should be noted that scRNA-seq has several limitations. First, it is limited to mRNA expression, lacking epigenomic and proteomic information, and TCR/BCR repertoires cannot be accurately resolved without targeted V(D)J sequencing using constant-region–specific enrichment. Second, samples for scRNA-seq are dissociated into single cells and original spatial information is inevitably lost. To overcome these limitations, several approaches have recently emerged that enable analyses at the single-cell level, including ATAC-seq, spatial transcriptomics, protein profiling, and TCR/BCR profiling. By integrating scRNA-seq with these techniques in a “multimodal” manner, it becomes possible to obtain novel insights that would be difficult to uncover using scRNA-seq alone [10–16].

Although attempts to predict treatment responsiveness and stratify patients have been partially successful and have advanced the treatment of SLE, there remain fundamental problems in SLE pathogenesis [5]. A more precise understanding of disease mechanisms through multi-omics analyses holds the potential to generate new perspectives for immunological stratification, which can help avoid unnecessary use of corticosteroids and immunosuppressive agents and enable the selection of more effective therapies (Fig. 1). In addition, the development of new treatment strategies is expected to address the unmet needs of patients who are unlikely to benefit from existing therapies.

In this review, we provide an overview of the characteristics of pathogenic B cells in SLE that have been elucidated through multi-omics analyses.

Genetic risk for SLE is enriched in B cell-related transcriptional programs

B cells maintain physiological homeostasis and protect against external pathogens through antibody production. On the other hand, B cells play a pivotal role in the pathogenesis of SLE by producing autoantibodies [17]. Autoreactive B cells are normally eliminated through negative selection during differentiation, and it remains unclear why autoreactive B cells evade negative selection and produce autoantibodies in SLE [18].

GWAS are a powerful tool for identifying potential causal relationships between risk factors and health outcomes and have yielded important insights across a wide range of diseases [6]. GWAS have identified multiple genetic loci associated with the development of SLE, including genetic variants in HLA and Fcγ receptor genes, as well as IRF5, STAT4, PTPN22, TNFAIP3, BLK, BANK1, TNFSF4, and ITGAM [19].

By integrating RNA-seq data with GWAS information, eQTL analysis quantifies the effects of genetic polymorphisms on the expression levels of specific genes, thereby enabling investigation of the biological consequences of these variants (Fig. 2). Previously, Ota et al. demonstrated the associations of naïve B cells and unswitched memory B cells with SLE using stratified linkage disequilibrium score regression (LDSC) in the ImmuNexUT database [8]. These data suggest that genetic risk for SLE is already enriched in transcriptional programs of naïve B cells, raising the possibility that predisposition to functional dysregulation may be established prior to overt differentiation and antibody production.

This method, which integrates GWAS and RNA-seq, could be applied to scRNA-seq as well as bulk RNA-seq. For instance, Perez et al. reported a large-scale scRNA-seq analysis using more than 1.2 million PBMCs from 264 unique samples (162 SLE and 99 HCs), in which cells were first classified into their lineages and then integrated with genomic data [11]. This study found a reduction in naïve CD4⁺ T cells and an increase in GZMH⁺CD8⁺ T cells in patients with SLE, as well as elevated expression of interferon-stimulated genes (ISGs) in classical monocytes. In addition, the expression levels of ISGs were inversely correlated with the reduction of naïve CD4⁺ T cells. This study identified 535 cell type–specific cis-eQTLs (cs-eQTLs) and 1207 shared cis-eQTLs (sh-eQTLs). Finally, this study also performed an LDSC analysis, which revealed that the highest enrichment for SLE variants was in classical monocytes and B cells [11, 20].

Emergence and origin of autoreactive B cells

Some naïve B cells are autoreactive even in healthy individuals [21]. The proportion of autoreactive B cells is approximately 50–80% in the bone marrow of healthy individuals, but then gradually decreases according to differentiation to about 10% in naïve B cells in peripheral blood [18]. However, among SLE patients, these autoreactive B cells proceed to differentiate into memory B cells, leading to the production of autoantibodies. This indicates that the processes responsible for eliminating autoreactive B cells are impaired in patients with SLE.

During B cell maturation, in healthy individuals, B cells migrate to germinal centers and mature into memory B cells with class switching and somatic hypermutation (SHM), producing antibodies with a high affinity for foreign antigens. Human memory B cells are commonly identified using CD27 as a marker, and class-switched memory B cells are typically defined as CD27⁺IgD⁻ B cells. This is based on several lines of evidence: CD27⁺ B cells accumulate somatic hypermutations and are therefore considered to have undergone affinity maturation; CD27⁺ B cells represent a subset that plays a central role in memory responses; and CD27⁺ B cells show enhanced differentiation into plasma cells and possess a distinct transcriptional program [22]. This process represents the germinal center reaction within the follicular pathway.

In contrast, autoreactive antibodies are often produced by B cells that have differentiated via an extrafollicular pathway, where they rapidly become plasma cells and exhibit limited clonality [17]. SLE is characterized by the expansion of double-negative (DN) B cells lacking both CD27 and IgD expression, particularly Double-negative 2 (DN2) cells [22]. DN2 cells, defined by a CXCR5⁻CD21⁻CD11c⁺ phenotype in addition to the absence of CD27 and IgD expression, have been suggested to differentiate through the extrafollicular pathway.

The role of T cells in the extrafollicular pathway remains controversial. There is a report demonstrating that B cells receive help from peripheral helper T (Tph) cells [23] and age-associated helper T (ThA) cells, both of which lack CXCR5 expression and are therefore thought to be involved in the extrafollicular pathway [24]. ThA cells are a subset of effector memory CD4⁺ T cells that expands with aging and is increased in autoimmune diseases, including SLE. ThA cells exhibit both a cytotoxic phenotype and B-cell helper functions. Gene expression profiles in ThA cells from SLE patients reflect disease activity and are affected by treatment with a calcineurin inhibitor [24]. Interestingly, the functions of ThA cells are regulated by the transcription factor ZEB2, which has also been reported to be required for the induction of atypical B cells (ABCs) [24, 25]. ThA cells and ABCs are characterized by co-expression of ZEB2 and TBX21 [24, 25]. However, the precise roles of these genes have not yet been fully elucidated. In ABCs, ZEB2 suppresses Myocyte enhancer factor 2b (Mef2b), an essential transcription factor for germinal center development, and plays a crucial role in orchestrating ABC formation by inducing the ABC signature [25]. Inhibition of the JAK–STAT pathway, which is regulated by ZEB2, has been shown to suppress ABC generation [25]. In contrast, in a mouse model, CD11c⁺ B cells were still generated in the absence of TBX21, whereas class switching to IgG2c was impaired [26]. Thus, although ZEB2 and TBX21 are both transcription factors that characterize ABCs, they appear to play distinct roles. ThA cells increase with age and exhibit both a cytotoxic phenotype and B-cell helper functions. In human primary CD4⁺ T cells, increased expression of GZMA and CXCL13 was observed only when ZEB2 and TBX21 were co-expressed. This phenomenon was not observed when either ZEB2 or TBX21 alone was induced, suggesting that cooperation between these two transcription factors is important for ThA function [24]. Taken together, these findings suggest that ZEB2 and TBX21 are transcription factors shared across T-cell and B-cell subsets involved in extrafollicular immune responses. However, the downstream transcriptional and regulatory mechanisms are likely to differ substantially between T cells and B cells, and further studies are needed to clarify their precise functions. On the other hand, Tph cells promote the differentiation of memory B cells into plasma cells via IL-21 [27]. The frequency of Tph cells in peripheral blood from patients with SLE correlates with that of CD11c⁺ B cells, and the frequencies of Tph cells and B cells show a similar pattern in renal tissue. Another study suggests that B cell differentiation occurs in a B cell-intrinsic manner under Toll-like receptor 7 (TLR7) stimulation [28].

Previously, CD11c⁺ B cells that increased in aged female mice and young lupus-prone mice were identified using fluorescence-activated cell sorting (FACS) and designated as age-associated B cells [29]. This population developed in response to TLR7 stimulation. ABCs were later identified in humans, where they were characterized using CD11c expression, as one of the sources of auto-antibody secreting plasma cells driven by interleukin-21 (IL-21) [30]. Notably, these CD11c⁺ B cells increased in patients with SLE in association with disease activity such as Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2 K) and did not necessarily correlate with age [31]. Various naming and annotations were used for this population [29, 30, 32–36] (Table 1). Similar cells referred to as DN2 cells were increased in SLE [30]. Traditionally, CD19, CD20, IgD, CD27, CD38, and CD24 have been used for the classification of human B cells [22]. As mentioned before, DN B cells were defined by the absence of IgD and CD27 expression in B cells, and the DN2 cells were characterized by CXCR5⁻CD11c⁺T-bet⁺ [30]. However, although ABCs and DN2 cells are similar, they cannot be regarded as completely identical subsets. ABCs and DN2 cells share CD11c expression, but DN2 cells represent a subset within the DN B cell compartment. In contrast, among naïve B cells, there is also a subset termed activated naïve B cells (aNAVs), which are characterized by a CD19⁺IgD⁺CD27⁻CXCR5⁻CD11c⁺ phenotype [30]. aNAVs exhibit transcriptomic features similar to those of DN2 cells and are also expanded in SLE [30]. Therefore, ABCs may encompass both DN2 cells and aNAVs, and ABCs and DN2 cells should not be considered completely equivalent.

Table 1.

Alternative naming and annotations of atypical B cells

Method	Species	Naming	Markers	Journal
FCM	Mouse	Age-associated	CD19highCD21−CD11c+	[29]
FCM	Human	DN2	CD19+IgD−CD27−CXCR5−CD21−CD11c+	[30]
scRNA-seq	Human	Atypical memory	CD19+IgD−CD21lowCD27lowCD38−CD71lowCD11c+FcRL5+	[35]
scRNA-seq	Human	Atypical	CD19+CD20+CD10−CD21−CD27−	[34]
scRNA-seq	Human	Atypical	CD19+CD27±CD21lowCD20+CD38−/low	[32]
FCM	Human	Age-associated	CD19+CD21lowCD11c+	[36]
FCM	Human	DN2	CD19+IgD−CD27−CD21−CD11c+	[33]

DN double negative, FCM flow cytometry

Age-associated B cells were originally identified as CD11c⁺ B cells that accumulate in the spleens of aged mice, and this cell population was also expanded in NZB/WF1 mice, a murine model of lupus. Because age-associated B cells secrete autoantibodies and their depletion in vivo leads to a reduction in autoreactive antibodies, they were considered to be involved in autoantibody production [29]. Although the expansion of DN B cells in human peripheral blood has long been recognized in SLE, it was later shown that this expansion reflects an increase in DN2 cells, which are characterized by a CXCR5⁻CD11c⁺ phenotype [30]. Age-associated B cells in mice and DN2 cells in humans share several features, including CD11c expression and a crucial dependence on TLR7 signaling for their development [37]. TLR7 plays a protective role against viral infection by inducing type I interferons (IFNs). However, gain-of-function mutations that cause excessive TLR7 activation have been identified in some SLE patients [28]. Thus, activation of TLR7 signaling may represent a shared mechanism underlying age-associated B cells in mice and DN2 cells in humans induction in both mice and humans.

Recent studies have suggested that ABCs have the potential to serve as a source of autoreactive plasma cells following stimulation through TLR7 or IL-21 [30, 31]. Additionally, naïve B cells were epigenetically distinct in SLE, and ABC accessibility was characteristically programmed by T-BET, AP-1, and EGR transcription factors [38].

These findings suggest that ABCs play a critical role in the production of autoantibodies. However, the origin of ABCs and why they emerge in SLE remains unclear. One proposed possibility is that viral infections such as Epstein–Barr virus (EBV) or cytomegalovirus (CMV) may act as triggers for the development of SLE through the production of antibodies against virus-derived antigens [39, 40]. Recently, an EBV-specific single-cell RNA-sequencing platform was developed [41]. This study found that EBV⁺ B cells exhibited higher expression of genes involved in antigen presentation compared with EBV⁻ B cells, suggesting that EBV⁺ B cells may possess upregulated antigen-presenting capacity. In this study, EBV-positive lymphoblastoid cell lines (LCLs) were generated from samples obtained from patients with SLE who had serum anti-DNA autoantibodies. The EBV-positive LCLs were then co-cultured with T cells and B cells, followed by scRNA-seq, which revealed an expansion of Tph cells and DN2 cells. These findings suggest that EBV⁺ B cells with enhanced antigen-presenting capacity may activate other immune cells and promote the expansion of immune cell populations implicated in SLE pathogenesis. Regarding the association between SLE and EBV, a novel mechanism involving antigen presentation of neoself-antigens has recently been identified. Autoantigens presented by MHC class II molecules in the absence of the invariant chain, which is essential for peptide presentation, are referred to as neoself-antigens, and these antigens induce clonal expansion of autoreactive T cells. In SLE, neoself-antigen-reactive T cells are expanded, and EBV reactivation is thought to play an important role as a major cause of this expansion by downregulating the invariant chain. These findings suggest that EBV reactivation contributes to the pathogenesis of SLE through activation of autoreactive T cells [42]. B-cell activation by neoself-antigen-reactive T cells may contribute to the emergence of ABCs.

mRNA expression often does not directly correspond to protein levels, and proteomic analyses have the potential to address this limitation. Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) is a method that enables protein quantification of cell surface markers at the single-cell level using oligonucleotide-labeled antibodies [43]. Based on previous findings that a plasmablast signature correlates with disease activity [44], Kotliarov et al. identified a shared activation program between vaccination responses in healthy individuals and SLE, characterized by the activation of a plasmacytoid dendritic cell (pDC)–type I interferon–T/B lymphocyte network [45]. In this study, a “TGSig” signal was identified that was associated with the plasmablast signature and was also correlated with the proportion of CD20⁺CD38⁺⁺ B cells in peripheral blood. Furthermore, through CITE-seq profiling of 82 surface proteins in peripheral blood samples from 20 healthy individuals, they showed that a CD40 activation signature in B cells was correlated with the proportion of CD20⁺CD38⁺⁺ B cells, suggesting that baseline alterations in immune cell states exist even in patients with clinically low disease activity.

Spatial analysis has demonstrated the localization of autoreactive B cells

Samples for scRNA-seq are dissociated into single cells, and their original spatial information cannot be analyzed. However, recent spatial analysis technologies such as Xenium, CosMx, or PhenoCycler are able to overcome this limitation [46, 47]. Xenium and CosMx analyze the transcriptome by hybridizing probes to mRNA, whereas PhenoCycler detects proteins through multiplexed immunofluorescence. Spatial transcriptomics and proteomics provide insights into the role of B cells from another perspective.

A previous study performed spatial transcriptomic analysis at single-cell resolution with CosMx in eight cases of childhood-onset lupus nephritis. It showed that B cells resided in tubulointerstitial immune hotspots and exhibited extensive interactions with myeloid dendritic cells (mDC), macrophages, and T cells within the inflamed kidney [10]. Plasma cells in local foci expressed IgA, IgG, and IgM immunoglobulin gene transcripts. B cells persistently reside in tubulointerstitial foci despite therapeutic B cell depletion among treatment-resistant patients. This indicates that the survival of autoreactive B cells in tissues contributes to treatment resistance and therapies targeting only peripheral blood B cells may be insufficient. Regarding the localization of ABCs, interesting findings have recently emerged with the development of spatial analysis technologies. In a study that performed spatial analysis of gut-associated lymphoid tissue (GALT) from healthy individuals using Visium, DN2 cells were found to colocalize with ITGAX⁺CD68⁺ dendritic cells (DCs) in the subepithelial dome (SED), a site where the initial encounter between antigens and the immune system occurs, and in the follicle-associated epithelium (FAE) [48]. This report also noted that the SED was enriched for expression of DNASE1L3, which is a nuclease responsible for the clearance of cell-free DNA [48, 49]. They proposed that DNASE1L3 expression in the SED of healthy GALT may prevent excessive exposure of DN2 B cells to microbial DNA, thereby potentially limiting anti-DNA immune responses. Indeed, DNASE1L3 has been identified as a lupus-associated autoantigen [49], and reduced DNASE1L3 activity in SLE has been associated with increased reactivity against DNASE1L3-sensitive antigens and anti-dsDNA antibody production [49]. In addition, DNASE1L3-deficient mice reportedly develop a lupus-like phenotype [50]. Collectively, these findings suggest that detailed spatial analyses using technologies such as Xenium, Visium, and PhenoCycler to characterize immune cell subsets, including DN2 cells, within inflammatory tissues may provide important insights into the pathogenesis of SLE. These findings underscore the importance of integrative analyses that incorporate spatial information.

In addition, spatial transcriptomics has the advantage of enabling not only the identification of B cell localization but also the characterization of the surrounding inflammatory microenvironment. Multimodal analyses that integrate spatial information with epigenomic data have been reported. In a study that combined spatial transcriptomics with single-cell multiome analysis across eight cases of lupus nephritis, VCAM1-expressing proximal tubule (PT_VCAM1) cells were identified, and these cells were colocalized with myofibroblasts and various types of immune cells, including plasma cells, in the renal cortex, constituting a distinct inflammatory microenvironment. Furthermore, along the differentiation trajectory toward PT_VCAM1 cells, the enhancer landscape revealed progressively increased chromatin accessibility at regulatory regions of proinflammatory genes, exemplified by CSF1 and IL32, as well as in profibrotic genes, including ITGA3 and COL4A1 [13].

It should be noted that multiome analyses require dissociation into single cells, and simultaneous RNA and epigenomic analyses are not conducted in situ on tissue sections with spatial information preserved. However, Russell et al. recently reported Slide-tags, a spatial barcoding approach that enables multimodal analysis at the single-nucleus level while retaining spatial context. This method involves placing tissue sections on arrays of DNA-barcoded beads, allowing spatial barcodes to be assigned to nuclei, then isolating nuclei for downstream sequencing. Importantly, transcriptomic and epigenomic profiling is performed after nuclear recovery rather than directly on the tissue section. This strategy provides sub–10 μm spatial resolution and is compatible with a wide range of single-nucleus assays. Using this framework, the authors acquired combined measurements of RNA expression, chromatin accessibility (ATAC-seq), and TCR-seq within the same nuclei, enabling spatially informed multi-omic profiling [51]. However, in situ spatial analyses that rely on integrated nuclear and cytoplasmic information at single-cell resolution have limitations, including the inability to capture full-length BCR and TCR sequences required for repertoire analysis, as well as difficulties in annotation due to insufficient cell segmentation or limited data depth. Although attempts have been made to overcome these limitations through deconvolution approaches [52], further advances are needed.

Multi-omics analyses uncover dysregulated differentiation and BCR features of autoreactive B cells

Affinity of antibodies is determined by B cell receptors (BCRs) and the diversity of BCRs is generated by V(D)J recombination, somatic hypermutation (SHM), and class switching [53]. Single-cell BCR analysis was initially performed using plate-based approaches. Subsequently, the development of multiple scRNA-seq platforms, including droplet-based and nanowell-based methods, has enabled large-scale, integrated analysis of transcriptomic profiles and BCR sequences at single-cell resolution. Nanowell-based scRNA-seq platforms, which predominantly rely on 3′-barcoded library constructions, capture BCR-derived transcripts and enable repertoire-level analyses. However, they also provide limited coverage of the 5’-located BCR variable regions, precluding direct reconstruction of full-length V(D)J sequences [54].

As can be inferred from the production of autoantibodies, alterations in characteristics of BCRs have been reported among SLE patients. Previous work has shown that immunoglobulin variable region heavy chain gene (IGHV4-34), which is known to code for autoreactive BCR, expands in the peripheral blood of SLE [55]. IGHV4-34–encoded antibodies can react with self-antigens commonly targeted in SLE, such as nuclear antigens, apoptotic cell–derived antigens, and dsDNA [56]. Antibodies expressing the 9G4 idiotype, which is composed of specific amino acid sequences within the framework region (FWR) and complementarity determining region (CDR) of IGHV4-34, have also shown autoreactivity [56]. Ota et al. demonstrated that the 9G4 idiotype of IGHV4-34 showed lower SHM in SLE memory B cells using bulk RNA-seq data from the ImmuNexUT database, which was constructed with the immune cells of peripheral blood [57]. Lower SHM suggests that B cells differentiate, preserving germline sequences, and that the affinity maturation of antibodies is insufficient. This study also showed that the autoreactive germline of IGHV4-34 was preserved in SLE memory B cells. Additionally, it has been reported that the BCRs of memory B cells exhibit increased similarity to those of naïve B cells—so-called “repertoire naiveness”—according to disease activity [57]. These findings indicate that defects exist in the processes responsible for the elimination of autoreactive B cells in SLE.

Despite accumulating insights into BCR features in SLE, reports of BCR-based patient stratification remain limited. Under tacrolimus treatment, a calcineurin inhibitor that suppresses T-cell activation by targeting calcineurin, a calcium- and calmodulin-dependent phosphatase essential for T-cell activation, patients who achieved low lupus disease activity state showed higher flare rates when the usage of IGHV4-34 in unswitched memory (USM) B cells was higher [58]. Thus, tacrolimus-mediated suppression of T-cell activation may not be sufficient to prevent SLE flares in patients with autoreactive B-cell compartments characterized by increased IGHV4-34 usage [58, 59].

Recently, single-cell-level BCR sequencing has also become feasible. For instance, in a study with scRNA-seq/BCR-seq data from blood samples of six patients with active lupus nephritis and four healthy individuals, ABCs in the peripheral blood of patients with lupus nephritis showed an increased IgG1 isotype [14]. Using single-cell BCR sequencing, this study also identified a unique clonal expansion of IGHV4-59⁺ B cells in the kidneys of patients with SLE. In addition, plasma cells exhibited reduced SHM in both light and heavy chains. Namely, the germline was preserved in SLE plasma cells. This finding was consistent with results from bulk RNA-seq analyses [57]. Furthermore, cell–cell communication analysis at the single-cell level suggested potential interactions between GZMK⁺ CD8 T cells and ABCs via the type II IFN pathway in kidneys from SLE patients. Ligand–receptor analysis also suggested potential IFN-γ and type II IFN receptor interaction between intrarenal GZMK⁺ CD8 T cells and ABCs [14]. This observation is consistent with previous findings showing that the proportion of IFN-γ-producing cells among CD8⁺ T cells was increased in SLE and that elevated serum IFN-γ levels correlated with the expansion of DN2 cells [60, 61].

Multi-omics analyses are applicable not only to peripheral blood but also to organs. B cells migrate from the bone marrow to the peripheral blood during differentiation [22]. CD24^hiCD38^hi transitional B cells, derived from bone marrow immature B cells, are increased in SLE and are associated with enhanced type I IFN expression [62]. Moreover, antibodies expressed by early immature B cells have been shown to exhibit autoreactivity [63]. These observations suggest that the migration of autoreactive B cells from the bone marrow into the peripheral blood may contribute to the expansion of autoreactive B cells in SLE. scRNA-seq/BCR-seq of B cells from the bone marrow of ten patients with SLE and five healthy individuals revealed that immature B cell loss in bone marrow is related to severe disease activity proinflammatory status and the overaction of type I interferon signaling [16]. Also, patients with a lower number of B cells in their bone marrow showed enrichment of oxidative phosphorylation, glycolysis/gluconeogenesis, phenylalanine metabolism, and pyruvate metabolism in B cells. SLE patients have an increased proportion of the IgG1 isotype, and IGHV4-34⁺ B cells have lower SHM in bone marrow, like in peripheral blood. In another study, scRNA-seq/BCR-seq using kidney samples and paired blood samples from six SLE patients revealed that B cells have lower SHM in the kidney than in peripheral blood, and ABCs which use IGHV4-59 expand in renal tissues of lupus nephritis [14]. These studies demonstrate changes in BCR profiles within local tissue environments, emphasizing the importance of analyses that are not limited to peripheral blood. Although this study focused on B cells infiltrating cancer tissues, not those in SLE, it reported scRNA-seq data integrated with BCR-seq for tumor-infiltrating B cells across 20 different cancer types, comprising 477 samples from 269 patients [64]. The authors calculated an “extrafollicular index” based on BCR similarity between plasmablasts and memory B-cell subsets to characterize B cells differentiated through the extrafollicular pathway compared with those differentiated through the germinal center pathway. They found that B cells differentiated through the extrafollicular pathway carried fewer SHM. Also, they were biased toward the IgG1 isotype rather than IgA, or differentiated while retaining the IgM isotype. These insights into extrafollicular differentiation may also be applicable to SLE, warranting further investigation.

Collectively, defects in the process of negative selection of B cells in SLE drive aberrant B cell differentiation, allowing autoreactive B cells with abnormal BCR features to persist across peripheral blood and target tissues (Fig. 3). Although genetic risk for SLE is enriched in B cells, its impact on the BCR repertoire remains unclear. Further studies are needed to determine whether BCR abnormalities arise during differentiation through the extrafollicular pathway or whether naïve B cells in SLE patients are intrinsically predisposed to differentiate while retaining potentially autoreactive BCRs.

Fig. 3.

Emergence of autoreactive B cells and BCR abnormalities in SLE. In healthy individuals, autoreactive B cell differentiation into memory B cells is suppressed, leading to apoptosis or anergy. On the other hand, autoreactive B cells differentiate into memory B cells in SLE and produce autoantibodies. ABC, Atypical B cell; EBV, Epstein–Barr virus; IGHV, immunoglobulin variable region heavy chain gene; SHM, somatic hypermutation; ThA, age-associated helper T

Current status and future perspectives of B cell targeting therapy

As described above, the B cell plays a pivotal role in SLE pathogenesis, and B-cell-targeted therapy can modify SLE disease activity to some extent. However, due to insufficient efficacy or side effects, B-cell-targeted therapy still has unresolved problems.

Among B cell targeted therapies for SLE, the most classical agent is rituximab. Rituximab is a chimeric anti-CD20 monoclonal antibody that binds to the CD20 antigen expressed on the surface of human B cells and depletes B cells through antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), complement-dependent cytotoxicity (CDC), and induction of apoptosis [65]. EXPLORER and LUNAR trials, which evaluated the efficacy of rituximab in SLE, failed to meet their primary endpoints, but improvements in serological activities were observed. Therefore, in the 2023 EULAR recommendations, rituximab is positioned as a treatment for SLE patients with severe organ involvement [66]. After rituximab treatment, the number of circulating B cells decreased, while class switched (predominantly IgG1 or IgG2) and clonally expanded B cells remained in SLE patients [67].

Another choice is belimumab, which inhibits B lymphocyte stimulator (BLyS) [68]. BLyS, also known as B cell–activating factor (BAFF) (which belongs to the tumor necrosis factor family), promotes B cell maturation and antibody production by binding to its receptors on the surfaces of B cells [68]. Serum BLyS levels are elevated in SLE patients and correlate with disease activity [68]. In BLISS trials, the belimumab-treated group demonstrated a significantly higher SLE Responder Index (SRI)-4 response rate at week 52 compared with the placebo group and achieved the primary endpoint [69]. Furthermore, in a subsequent BLISS-LN trial, which evaluated the efficacy of belimumab in patients with active lupus nephritis, the belimumab group showed a higher rate of renal response (RR) compared with the placebo group [70]. Finally, bulk RNA-seq showed that belimumab could ameliorate repertoire naiveness in SLE [57].

In addition, several novel B cell–targeted therapies have recently attracted considerable attention. For instance, ianalumab targets B cells through the BAFF receptor (BAFF-R). By binding to BAFF-R, ianalumab exerts two distinct mechanisms of action: inhibition of BAFF:BAFF-R signaling, which is essential for B-cell proliferation and survival, and depletion of B cells through ADCC [71]. Thus, ianalumab exerts dual mechanisms of action, blocking BAFF signaling while inducing B-cell depletion. A phase 2 study was just conducted in patients with SLE who were in moderate or severe disease activity, and ianalumab could modify disease activity with good tolerance [71]. There are also other anti-CD20 therapies in addition to rituximab. For example, obinutuzumab is a fully humanized anti-CD20 mAb [72]. In a randomized, placebo-controlled, phase 2 trial (NOBILITY), obinutuzumab was superior to the placebo in both complete and overall renal responses [72].

Furthermore, CD19 chimeric antigen receptor T-cell (CAR-T) therapy has recently emerged as a novel B-cell-targeted therapeutic approach. In a cohort of eight SLE patients with severe disease activity, CAR-T therapy resulted in drastic clinical responses, with all patients achieving definition of remission in SLE (DORIS) [73]. Although adverse events such as infections and cytokine release syndrome were also reported and careful consideration is required when selecting patients for CD19 CAR-T-cell therapy, this treatment may become a potential therapeutic option for patients with treatment-refractory SLE in the future. An additional CAR-T therapy study compared post-treatment transcriptomic changes in patients with SLE to those observed with other therapies [74]. In this study, scRNA-seq was performed on peripheral blood samples from seven SLE patients who received CD19 CAR-T-cell therapy, and the results were compared with whole-blood transcriptome data from patients treated with other conventional therapies. Although this study is limited by the lack of scRNA-seq data from a control group of patients who did not receive CD19 CAR-T-cell therapy, it demonstrated that CD19-CAR T-cell-induced remission is characterized by suppressed complement and type I interferon pathways and metabolic reprogramming. In addition, this study indicated that CD19 CAR-T-cell therapy could possibly induce greater downregulation of type I and type II interferon pathways, DNA damage response (DDR), and chemokine pathways compared with rituximab and belimumab. These findings not only support the efficacy of CD19 CAR-T-cell therapy but also suggest that current immunosuppressive agents and biologic therapies may not fully suppress pathogenic signaling pathways. Other therapeutic approaches, including allogeneic CD19 CAR-NK therapy and teclistamab, a T-cell engager and CD3 and B-cell maturation antigen (BCMA)-bispecific antibody approved for the treatment of multiple myeloma, have also been reported as potentially effective in SLE [75, 76]. These strategies may provide more accessible and less costly alternatives to CAR-T therapy, although further studies are needed to establish their efficacy.

Although various B cell targeted therapies have been developed and are effective in the treatment of SLE, there are no B cell targeted therapies specifically eliminating autoreactive B cells (Fig. 4). In parallel, recent multi-omics analyses have begun to provide deeper insights into the differential effects and mechanisms of action of these therapies at the cellular and molecular levels. Further development of therapies selectively targeting autoreactive B cells is expected.

Fig. 4.

B cell targeted therapy in SLE. Rituximab, obinutuzumab, ocrelizumab, and ofatumumab target CD20, and epratuzumab targets CD22, which are ubiquitously expressed on mature B cells. Daratumumab targets CD38, which is highly expressed on plasmablasts and plasma cells. Pathways mediated via BAFF and APRIL are inhibited by belimumab and tabalumab (anti-BAFF), and by atacicept and telitacicept (anti-BAFF and APRIL), respectively. Ianalumab targets BAFF-R. CAR-T cell therapy represents an alternative strategy to eliminate B cells

Conclusions

The emergence and increase of autoreactive memory B cells in SLE are well-established phenomena, and it has long been recognized that the genetic background associated with SLE influences B cell function; however, the underlying mechanisms have remained unclear. Recent advances in scRNA-seq and multimodal approaches have substantially deepened our understanding of SLE pathogenesis. GWAS can identify causal variants associated with the SLE phenotype. In addition, integrative analyses of mRNA and GWAS via eQTL analysis can lead to a deeper understanding of disease mechanisms. The differences among these mechanisms remain to be clarified in future studies. Furthermore, spatial analyses have been used to characterize ABC localization within inflammatory microenvironments such as lupus nephritis and the surrounding cellular communities. In addition, BCR-seq demonstrated the change of isotype usage toward IgG1, lower SHM, higher usage of autoreactive IGHV, and increase of repertoire naïveness in SLE. However, many aspects of pathogenic B cell biology, including the mechanisms driving their differentiation and induction, remain to be elucidated.

Although recent advances in B-cell-targeted therapies are remarkable, sometimes treatments are not sufficiently effective and adverse events make it difficult to continue therapy. In particular, the emergence of CAR-T therapy, or CAR-NK therapy and T-cell engagers may provide new therapeutic opportunities for patients with SLE who have been refractory to conventional treatments and difficult to manage. However, from the perspectives of adverse effects and cost, these therapies cannot be applied to all patients, and the importance of precision medicine is likely to increase further. In addition, if therapeutic strategies that specifically target pathogenic B cells can be developed in the future, it may be possible to further reduce treatment-related disadvantages. Further understanding of inflammatory microenvironments through spatial analyses may facilitate the development of therapeutic strategies that selectively target ABCs and autoreactive B cells within inflamed tissues, potentially reducing adverse effects while improving therapeutic efficacy. In the future, these insights are expected to facilitate precision medicine in the treatment of SLE through patient stratification for predicting treatment responsiveness and prognosis. Multi-omics research has the potential to provide a foundation for addressing these challenges.

Abbreviations

ABC

Atypical B cell

ADCC

Antibody-dependent cellular cytotoxicity

ADCP

Antibody-dependent cellular phagocytosis

aNAV

Activated naïve B cells

BAFF

B cell-activating factor

BAFF-R

B-cell-activating factor receptor

BCR

B cell receptor

BLyS

B lymphocyte stimulator

CAR-T

Chimeric antigen receptor T-cell

CDC

Complement-dependent cytotoxicity

cs-eQTL

Cell type-specific cis-eQTL

CMV

Cytomegalovirus

CDR

Complementarity determining region

DC

Dendritic cells

DDR

DNA damage response

DN

Double negative

DORIS

Definition of remission in SLE

EBV

Epstein–Barr virus

eQTL

Expression quantitative trait locus

FACS

Fluorescence-activated cell sorting

FAE

Follicle-associated epithelium

FCM

Flow cytometry

FWR

Framework region

GWAS

Genome-wide association studies

IFN

Interferon

IGHV

Immunoglobulin variable region heavy chain gene

IL-21

Interleukin-21

ImmuNexUT

Immune Cell Gene Expression Atlas from the University of Tokyo

LCL

Lymphoblastoid cell lines

LD

Linkage disequilibrium

LDSC

Linkage disequilibrium score regression

mDC

Myeloid dendritic cells

pDC

Plasmacytoid dendritic cell

PEC

Parietal epithelial cell

RPGN

Rapidly progressive glomerulonephritis

RR

Renal response

scRNA-seq

Single-cell RNA sequencing

sh-eQTL

Shared cis-eQTL

SED

Subepithelial dome

SHM

Somatic hypermutation

SLE

Systemic lupus erythematosus

SLEDAI-2K

Systemic Lupus Erythematosus Disease Activity Index 2000

SNP

Single nucleotide polymorphism

SRI

SLE Responder Index

ThA

Age-associated helper T

TLR7

Toll-like receptor 7

Tph

Peripheral helper T

USM

Unswitched memory

Authors’ contributions

T.U., T.O., and K.F. designed and wrote the manuscript. The authors read and approved the final manuscript.

Funding

This work was supported by the Center of Innovation Program from Japan Science and Technology Agency (JST) (JPMJCE1304 to K.F.); the Ministry of Education, Culture, Sports and the Japan Agency for Medical Research and Development (AMED) (JP21tm0424221, JP21zf0127004, JP22ek0410074, JP223fa627001, JP233fa627001, JP24gm1910005, JP24ek0410109, JP23gm1810005, JP256f0137004j0001, JP266f0137004j0002, JP276f0137004j0003, JPSIP2023A1, JPSIP2023A1_2, and JPSIP2023A1_3 to K.F.; JP23gm1810005 to K.F., and T.O.); the JSPS Grant-in-Aid for Scientific Research (B) (JP22H03110 and JP25K02684 to T.O.); and the Chugai Pharmaceutical Co. Ltd., Tokyo, Japan.

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

Tomohisa Okamura or T.O. belongs to the Social Cooperation Program, Department of Functional Genomics and Immunological Diseases, supported by Chugai Pharmaceutical. Keishi Fujio or K.F. serves as an Associate Editor of this journal and was not involved in the review or decision process for this manuscript.