IL-4-Induced Quiescence of Resting Naive B Cells Is Disrupted in Systemic Lupus Erythematosus Free

This article is featured in Top Reads, p.

The type I IFN signature is a hallmark feature of SLE and is strongly associated with the development of autoantibodies and disease activity (1–4). It is commonly thought that type I IFN production is secondary to uptake of immune complexes containing ribonucleoproteins (RNPs), which then stimulates TLR pathways, especially TLR7, in plasmacytoid dendritic cells (DCs) and B cells (5–8). However, we previously showed that B cell endogenous IFN-β is a prerequisite for efficient TLR7 signaling in response to RNPs (9, 10). Interestingly, elevated B cell endogenous IFN-β persisted throughout different developmental stages in transitional (Tr) stage 1 (T1), mature naive, memory, and germinal center (GC) B cell stages (11, 12), suggesting the possibility that B cell fate decision can be determined at the Tr stage to promote the development of activated B cells from resting B cells.

Much interest has been focused on factors that elevate the B cell response in SLE (13–20). In contrast, the potential role of factors that may contribute to B cell quiescence and that are downregulated in SLE are less well studied (21). It is recognized that the fundamental defect in SLE might be a failure of induction of anergy or tolerance, leading to loss of B cell quiescence; yet, several questions remain unresolved. The interrelationship of transcriptional programs that regulate extracellular cytokines and transcription factors (TFs) that induce B cell quiescence to prevent excessive type I IFN stimulation is unclear. Garaud et al. reported that B cells from patients with SLE with inactive disease have increased expression of genes induced by IL-4 (22). At the cellular level, IL-4 has been shown to suppress DC response to type I IFN (23) and the constitutive and induced production of IFN-β by macrophages (24). IL-4 is well known for its ability to promote the development of anti-inflammatory macrophages (23, 25–28). Such quiescent effects of IL-4 on early-stage B cell development are less well studied, although mouse studies have demonstrated that IL-4 signaling antagonizes T-bet induction in B cells (29).

To determine how the balance of cytokine signaling can regulate the B cell fate decision, the present study analyzed differentially expressed genes (DEGs) at the Tr, resting naive (rNAV), and activated naive (aNAV) stages of B cell development in three healthy control subjects (HC) and three subjects with SLE. The transcriptomic findings were tested using an in vitro stimulation system. B cell phenotypes, auto-Ab profiles, and their association with the expression of IL-4R and IFN-β were analyzed in 47 patients with SLE. The results show that the gene signature at early-stage Tr and rNAV B cells has already established a distinct trajectory for subsequent B cell development in HC and patients with SLE. Our results suggest a model in which the abnormally elevated type I IFN signaling and attenuated IL-4 signaling in SLE B cells shifts the normal B cell developmental transcriptomic program from the follicle-oriented rNAV program to the extrafollicular atypical memory (aMEM) B cell program.

Subjects with SLE were recruited from the University of Alabama at Birmingham (UAB) Lupus Clinic. Clinical data, including Ab seropositivity and renal disease data, were determined by the UAB clinical laboratory and attending physician (Supplemental Table I). Patient race was self-reported. All data were collected in a double-blinded manner until all data collection was completed. All patients with SLE met the American College of Rheumatology 1997 revised criteria (30) and the 2017 American College of Rheumatology/European League Against Rheumatism classification criteria for SLE (31). These studies were conducted in compliance with the Helsinki declaration and approved by the institutional review board at UAB. All participants provided informed consent.

PBMCs from patients with SLE were isolated from heparinized blood by density gradient centrifugation (Lymphoprep/SepMate, STEMCELL Technologies).

For single-cell pan–B cell RNA-sequencing (RNA-seq) analysis, total B cells were enriched by magnetic microparticle purification using the EasySep Human B Cell Enrichment Kit (STEMCELL Technologies). For Tr B cell single-cell RNA-seq (scRNA-seq) analysis, Tr B cells, which contain mainly T1 B cells, were purified using flow cytometry and were gated as live, non-CD27^hiCD38^hi but CD24^hiCD38^hiCD10⁺IgD⁺CD27⁻ B cells. Single cells were captured into Gel Bead-in-Emulsions by using a 10× Genomics Chromium Controller, and single-cell 5′ biased transcriptome libraries and V(D)J libraries were prepared by using the 10× Genomics 5′ Library & Gel Bead Kit (PN-1000014) and Single Cell V(D)J Enrichment Kit, Human B Cell (PN-1000016), according to the company’s manual.

The sequencing raw data were processed with the Cell Ranger pipeline (version 3.1.0). Raw base call files generated by the Illumina NextSeq 500 were demultiplexed into FASTQ files. For single-cell 5′ gene expression data, the Cell Ranger count pipeline was used to perform quality control, sample demultiplexing, barcode processing, alignment against the GRCh38 human reference, and gene counting. For single-cell V(D)J BCR data, the Cell Ranger vdj pipeline was applied to perform quality control, assembly, quantification, and annotation of paired V(D)J transcript sequences. The Seurat (32) package (version 3.2.0) implemented in R was applied to exclude low-quality cells in different single-cell experiments. The cells that expressed fewer than 200 genes were filtered out, and genes that were not detected in at least three single cells were excluded. The filtered contig annotation files generated from single-cell V(D)J BCR data were merged with filtered single-cell gene expression data based on barcode information using the merge function in R. The expression data of individual B cells with the overlapping barcodes were positively selected on the basis of expression of MS4A1 (the gene encoding CD20).

All stages of B cells were divided into different groups based on the expression of Ig genes. The specific groups were classified by the subset function in R. Three subgroups were informatically classified from cells that expressed IGHM and IGHD genes. The expression levels of genes in different subclusters were visualized by histogram plots and bar plots generated by ggplot2 (version 3.3.5). The unsupervised clusters were identified using the Seurat R Package (version 3.2.0). Principal component (PC) analysis was performed using the Seurat function RunPCA, and the k-nearest neighbor graph was constructed using the FindNeighbors function in Seurat with the number of significant PCs identified from PC analysis. Clusters were identified using the FindClusters function with a resolution of 0.8. The clusters were visualized in two dimensions with Uniform Manifold Approximation and Projection. Differential gene expression analyses were carried out using the Seurat function FindMarkers. The Wilcoxon rank-sum test was used with the default threshold of 0.25 for log₂ fold change and a filter for the minimum percentage of cells in a cluster greater than 25%. The DEGs were isolated by comparing significantly upregulated genes and downregulated genes defined as adjusted p value, p_adj < 0.05. Top DEGs were visualized by heatmaps via the ComplexHeatmap (version 2.11.1) package and violin plots using the function VlnPlot in Seurat.

Monocle version 2.10.1 (33, 34) was used to analyze single-cell developmental trajectories. A set of genes that defined B cell development was ordered for supervised trajectories. These genes included IGH genes, aMEM genes, IFN-stimulated genes (ISGs), rNAV, classical memory (cMEM), and GC pathway genes. The expression profiles were reduced to two dimensions using the DDRTree algorithm in the function reduceDimension. The IGHD⁺ B cell group was set as the root_state (the starting point) of the trajectory and IGHG⁺ or IGHA1⁺ B cells as the end_state. Cells were ordered into a trajectory, and any branch points corresponding to cell fate decisions were identified. The plot_cell_trajectory function was used for visualization of the different states of cells through a pseudotime analysis. The function plot_pseudotime_heatmap was used to visualize modules of genes that co-vary across the pseudotime.

The R package ReactomePA (35) (version 1.26.0) was used to perform Reactome pathway-based analysis. The number of selected genes associated with Reactome pathways was assessed via the function enrichPathway. Gene Ontology enrichment analysis was performed using the function enrichGO in clusterProfiler (36) package (version 3.10.1). The top upregulated and downregulated genes significantly differentially expressed from Seurat (false discovery rate, <0.05) were used as target genes. The significant Biological Process terms and genes associated with each term were visualized by the barplots generated by ggplot2 (version 3.3.5). Human Protein Atlas (HPA) pathways were enriched by R package gprofiler2 (version 0.2.1). The significant upregulated and downregulated genes obtained from Seurat (p_adj < 0.05) and the annotation database HPA were used as input datasets. The gost function from gprofiler2 was used to find the biological functions and pathways that are significantly enriched in the gene set (37). The result function was applied to fetch the data frame with the enriched functions and related statistics. The barplots using R package ggplot2 (version 3.3.5) and the network plot using the cnetplot function in clusterProfiler package (version 3.10.1) were used to visualize functional enrichment results. TF-gene co-occurrence and MSIgDE pathway analysis were carried out using the enrichment analysis tool Enrichr program, which applies clustergrams for the visualization of the enrichment results (38).

Analysis of B cell surface proteins was determined by flow cytometry using PE-594 anti-CD19 (HIB19), Brilliant Violet 605 anti-IgD (IA6-2), Brilliant Violet 650 anti-CD27 (O323), PerCP-cyanine 5 anti-CD38 (HIT2), allophycocyanin anti-CD21 (Bu32), and Pacific Blue anti-CD11c (3.9) obtained from BioLegend and PE anti-IL-4R (hIL4R-M57) obtained from BD Biosciences. Dead cells were excluded from analysis with allophycocyanin eFluor 780 Organic Viability Dye (Thermo Fisher Scientific).

For intracellular staining of IFN-β (clone MMHB-3, PBL Assay Science), cells were stained with eFluor 780 viability dye, followed by fixation in 2% paraformaldehyde and 70% ice-cold methanol permeabilization before staining. For intranuclear staining of T-bet (4B10, BioLegend) or IRF7 (12G9A36, BioLegend), permeabilization and fixation were carried out using the eBioscience Transcription Factor Staining Buffer (Thermo Fisher Scientific). Cells (from 300,000 to [1 × 10⁶] per sample) were analyzed by flow cytometry. FACS data were acquired with an LSRII FACS analyzer (S/N 30201809, BD Biosciences) and analyzed with FlowJo software version 10.6.2 (BD Biosciences, Ashland, OR).

B cell differentiation was stimulated using the protocol described by Jenks et al. with slight modifications (39). Briefly, B cells were stimulated in RPMI with 10% FBS supplemented with or without 50 ng/ml IL-4 for 1 h. Cells were then stimulated with CL264, a TLR7 ligand, 1 μg/ml (InvivoGen, San Diego, CA), 10 ng/ml BAFF (R&D Systems, Minneapolis, MN), 10 ng/ml IL-21 (R&D Systems), 50 ng/ml IL-2 (R&D Systems), and 20 ng/ml IFN-γ (R&D Systems), with 10 μg/ml goat F(ab′)₂ anti-human Ig (Thermo Fisher Scientific, Waltham, MA) for 3 d. Cells were washed and resuspended in fresh media with the same media but not anti-IgG/IgM. After 2 additional days, cells were stained for flow cytometry to determine B cell differentiation. In some experiments, the same stimulation was carried out, except that IL-21 was replaced by IFN-β (5 U/ml) (PBL Assay Science).

Results are shown as the mean ± SD. Pearson’s normality test was used to determine the normal distribution of each dataset. A two-tailed, unpaired Student t test was used when two normally distributed groups of datasets were compared for statistical differences. The Mann-Whitney nonparametric test was used when the two datasets for comparison were not normally distributed. A two-tailed, paired Student t test was used to compare the effects of IL-4 in in vitro stimulation. A linear regression analysis or a Pearson correlation test was used to compare the correlation of two different variables. A χ² test (Fisher’s exact test) was used to determine differences in the distribution of a categorical variable. Except for scRNA-seq data, all statistical analyses were performed using Prism software (GraphPad Software, La Jolla, CA). p Values less than 0.05 were considered significant.

The scRNA-seq data used in this study have been deposited in the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE136035) under accession number GSE136035. Raw data files used in the figures that support the findings of this study are available from the corresponding authors upon reasonable request.

The shell and R scripts enabling the main steps of the analysis for this project are available on request.

We performed high-throughput droplet-based scRNA-seq analysis using B cells derived from three HC and three patients with SLE (Supplemental Fig. 1). Transcriptomic analysis revealed that there was an increased frequency of IGHM⁻IGHD⁻ but IGHG⁺ or IGHA1⁺ B cells in patients with SLE compared with HC (Supplemental Fig. 1).

In order to study differential gene expression at the same stage of development, we first analyzed the top 50 DEGs in IGHM⁺IGHD⁺ preswitched B cells in patients with SLE and HC (Fig. 1A, 1B). There was upregulation of ISGs, including IFI44L, IFITM1, MX1, ISG15, IRF7, STAT1, and IFIT3, as well as a double-negative 2 (DN2) signature gene, FCRL5, in SLE B cells. Genes that were downregulated in SLE B cells included IL4R, IL2RG, FCER2, and BACH2 (Fig. 1A, 1B). The Reactome pathway analyses showed that although the ISG node was the top upregulated pathway, the IL-4/IL-13 signaling node was the top downregulated pathway in IGHM⁺IGHD⁺ B cells of patients with SLE compared with HCs (Fig. 1C).

It is unclear if the downregulation of the IL-4 pathway in SLE IGHM⁺IGHD⁺ B cells coevolved with the elevation of ISGs. To address this, we developed a strategy to analyze B cell development at the Tr, rNAV, and aNAV development stages within the IGHM⁺IGHD⁺ B subset. The IGHM⁺IGHD⁺ subpopulation could be clustered into three subsets based on an unsupervised Uniform Manifold Approximation and Projection analysis (Fig. 2A).

Cluster 1 B cells from both HC and patients with SLE expressed higher levels of TCL1A, IGLL5, VPREB3, ACTG1, and PLD4 (Supplemental Fig. 2A) than cluster 2 and cluster 3 subsets of B cells. Analysis of transcriptomic data from flow cytometry sorted Tr stage of B cells from HC1 and SLE2 (Supplemental Fig. 2B, 2C) verified that the same genes that were upregulated in cluster 1 were also upregulated in Tr B cells compared with IGHM⁺IGHD⁺ B cells. Cluster 1 was accordingly determined to be Tr B cells. Cluster 2 B cells were determined to be rNAV B cells based on their expression of characteristic signature genes IL4R, PLPP5, SELL, CXCR4, BACH2, and CCR7 (Supplemental Fig. 2A). Cluster 3 B cells were characterized as aNAV B cells due to the upregulation of IGHG or IGHA1 genes (Supplemental Fig. 2A). There was a higher distribution of Tr and aNAV subsets of B cells but a lower distribution of rNAV B cells in patients with SLE compared with HC (Fig. 2A, lower, and Supplemental Fig. 2A).

Using the strategy described above, we have identified the top 25 genes that were up- or downregulated at the Tr, rNAV, and aNAV stages in patients with SLE compared with HC (Fig. 2B). Gene Ontology Biological Pathway analysis shows that the dominantly upregulated pathways in SLE B cells at all three stages were signaling pathways related to type I and type II IFNs (Fig. 2C). Genes that were upregulated in the aNAV subset of B cells in patients with SLE compared with HC included genes that have been implicated in DN2 and aMEM B cells, including FCRL5, ZEB2, and FCRL3 (39). In contrast, the dominant pathway that was upregulated in HC Tr and rNAV B cells was the cellular response to IL-4 (Fig. 2C). Furthermore, genes that were upregulated in HC aNAV B cells were those related to the regulation of lymphocyte activation and humoral immune response (Fig. 2C).

Fourteen of the top 25 upregulated genes in HC aNAV B cells encode proteins that were identified in GC B cells based on the HPA protein tissue distribution database (Fig. 3A). The pro-DN2 developmental trajectory of SLE naive B cells and the pro-GC developmental trajectory of HC naive B cells were further analyzed by a pseudotime branch point analysis using IGHD^hi B cells as the root and IGHG⁺ or IGHA⁺ B cells as the endpoint (Fig. 3B). We have identified two dominant trajectories in these B cells (Fig. 3B, upper). In SLE, 63.3% of IGHM⁺IGHD⁺ B cells exhibited a developmental trajectory to the DN2 lineage that was positive for aMEM genes (ITGAX, FCRL3, ZEB2, and FCRL5) and ISGs (IRF7, IFI44L, MX1, STAT1, ISG15, and OAS1) (Fig. 3B, middle, and (Fig. 3C). In contrast, 57.6% of IGHM⁺IGHD⁺ B cells in HC remained negative for the expression IGHG or IGHA1 genes (Fig. 3B, lower). As HC rNAV B cells developed into aNAV B cells, they predominantly (20.2% of total) expressed cMEM and GC B cell genes, including CD27, EEF2, ALOX5, ARPC1B, NME2, SPIB, CD44, and LTB (Fig. 3B, lower, and (Fig. 3C). Only 3.3% of B cells from HC expressed aMEM signature genes (Fig. 3B, lower, and (Fig. 3C) (p < 0.0001 comparing patients with SLE and HC). Together, these results suggest that, as opposed to SLE B cells, normal B cell development follows a rNAV path regulated by IL-4R signaling and that activation of these B cells uses a GC-oriented path for further maturation.

Because the SLE and HC B cells use two separate paths for their development, we next determined if separated sets of genes and pathways are involved in each stage of development. A Venn diagram demonstrates identified common genes upregulated or downregulated in Tr, rNAV, and aNAV B cells in patients with SLE and HC (Fig. 4A).

At the Tr stage of B cell development in SLE, there was upregulation of specific ISGs and stress response genes that were not present at subsequent stages of development. Uniquely expressed ISGs and antiviral genes in Tr B cells included IFIT3/ISG60, BST2/tetherin, and HERC5/E3-ISG15 (Fig. 4A) (40, 41). MX2 exhibits antiviral activity against human viruses. DDIT4 encodes DNA damage inducible TF 4, and together with COX7A, which encodes cytochrome c oxidase subunit 7, as well as TRIM22, are p53 target genes induced by DNA damage (42) (Fig. 4A). Thus, the earliest genes uniquely expressed in Tr B cells in SLE are strongly related to cell stress, virus response, DNA damage, and selected ISGs.

We have identified 7 genes that were upregulated in both Tr and rNAV B cells and 11 genes that were consistently upregulated in Tr, rNAV, and aNAV in patients with SLE compared with HC, respectively (Fig. 4A). Seventeen of these 18 genes share common co-occurrence TFs, including HESX1, IRF9, ZNFX1, ZBP1, STAT2, SP110, PARP12, ZC3HAV1, IRF7, and IRF1, based on the Enrichr TF-gene co-occurrence analysis (38) (Fig. 4B). Among these, IRF7 was upregulated in all three subsets of B cells in patients with SLE compared with HC (Fig. 4A).

In SLE, B cells at the aNAV stage of development expressed cell surface receptor signaling pathway genes, which were not expressed by Tr or rNAV B cells. These include FGR, PPP1R14A, CIB1, CEMIP2, HCK, EIF2AK2, and DAPP1. Genes that have previously been reported in DN2 aMEM B cells, including FCRL3, FCRL5, ZEB2, and DAPP1, were also uniquely upregulated in aNAV B cells of patients with SLE (Fig. 4A).

In contrast, 16 genes were upregulated in at least two subsets of B cells in HC compared with patients with SLE (Fig. 4C). Genes that were commonly upregulated in HC Tr B cells and rNAV B cells included IL-4R pathways genes IL4R and IL2RG. Genes that were commonly upregulated in rNAV and aNAV B cells from HC also contain an IL-4R pathway gene, FCER2, and a GC light zone B cell–related gene, CD83 (43). The three genes that were consistently upregulated in all three subsets of B cells in HC were NME2, SPIB, and ARPC1B (Fig. 4C). The TFs that share co-occurrence with 7 of these 16 genes included IKAROS, SCML4, AIOLOS, IRF4, PLEK, STAT4, SPIB, and SPI1 (Fig. 4D).

The transcriptomic data suggest that IL4R⁺ and ISG⁺ naive B cells use two distinctive transcriptomic programs and trajectories for their development. Jenks et al. (39) previously demonstrated that IFN-γ plus anti-Ig, BAFF, IL-21, TLR7, and IL-2 induced DN2 B cell development and that such effect was abrogated when IL-4 was substituted for IFN-γ. Here, we asked if IL-4 could suppress IFN-γ plus anti-Ig + BAFF + IL-21 + TLR7 + IL-2-induced DN2 B cell development. Pretreatment of B cells with IL-4 enhanced CD27⁺CD38⁺ plasmablast development (Fig. 5A, upper). It also suppressed B cell development into the IgD⁻CD27⁻ DN B cells and enhanced the percentages of IgD⁻CD27⁺ cMEM B cells (Fig. 5A, lower). Within the IgD⁺CD27⁻ naive B cell population, the addition of IL-4 resulted in a highly significant increase in the percentages of the CD11c⁻T-bet⁻ rNAV B cells and a significant decrease in the percentages of CD11c⁺T-bet⁺ aNAV B cells (Fig. 5B). Similarly, within the DN B cells, IL-4 treatment resulted in significant inhibition of the development of the CD11c⁺T-bet⁺ DN2 population and preserved B cells as the CD11c⁻T-bet⁻ DN1 population (Fig. 5C), but it did not alter the newly characterized CD11c⁻Tbet^low DN3 population (44–46) (Fig. 5C).

IL-4 has been demonstrated to suppress type I IFN-induced responses, including upregulation of MX1, IRF7, STAT1, STAT2, and IFNB, in DCs (23). In some experiments, we substituted IL-21 with IFN-β in the in vitro culture conditions and further identified that pretreatment of B cells with IL-4 significantly suppressed IFN-β + anti-Ig + TLR7 + IFN-γ + BAFF + IL-2-induced IRF7 and T-bet (Fig. 5D). These results together suggest that the imbalance of type I IFN and IL-4 signaling in patients with SLE has shifted the activated B cell developmental trajectory away from the cMEM pathway but toward the aMEM pathway.

We previously showed that B cell expression of IFN-β is associated with the development of anti-RNP in SLE (9, 10, 47). We now combined this feature with the analysis of IL-4R in naive B cells (Supplemental Fig. 3) derived from 47 subjects with SLE (30 African American subjects [AAs] and 17 White subjects) by flow cytometry. Higher percentages of IL-4R⁻IFN-β⁺ (Fig. 6A, top) or lower percentages of IL-4R⁺IFN-β⁻ (Fig. 6A, bottom) naive B cells positively correlated with the percentage of the CD21⁻IgD⁻ population of B cells. They also correlated negatively with the percentage of CD27⁺IgD⁻ cMEM B cells (Fig. 6B). Patients with SLE who exhibited higher percentages of IL-4R⁻IFN-β⁺ (Fig. 6C, right) and lower percentages of IL-4R⁺IFN-β⁻ naive B cells (Fig. 6C, left) exhibited a higher prevalence of anti-Smith (anti-Sm)/RNP and anti-DNA. There was a positive correlation of Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) in patients with SLE with a higher percentage of IL-4R^-IFN-β⁺ naive B cells, which reached statistical significance for White but not AA patients with SLE (Fig. 6D). Among White patients with SLE, higher percentages of IL-4R⁻IFN-β⁺ naive B cells and lower percentages of IL-4R⁺IFN-β⁻ naive B cells were found in patients who developed anti-Sm or renal disease (Fig. 6E, lower).

Our results suggest that lower expression of IL-4R is not merely a signature of rNAV B cells but IL-4R signaling enhanced CD11c⁻T-bet⁻ DN1 and inhibited the development of a CD11c⁺T-bet⁺ DN2 pathogenic B cell population under an IFN-γ included DN2 polarization condition (39). IL-4R signaling did not alter naive B cell development into the newly recognized CD11c⁻T-bet^low DN3 subpopulation (44). IL-4-stimulated B cells were less sensitive to IFN-β-stimulated expression of IRF7 and T-bet. Interestingly, our results suggest that IL-4R and type I IFN receptor signaling modulate two distinct trajectories of memory B cell development commencing at the Tr stage of B cell development. ISG^hi and IL-4R^lo/− B cells can readily differentiate into pathogenic DN2 B cells seen in SLE. In contrast, B cells that express higher levels of IL-4R are preferentially maintained as CD21⁺IgD⁺ rNAV B cells. When stimulated by IL-4, they further developed into the CD27⁺ cMEM B cells and DN1 B cells. IL-4 was previously shown not to promote DN2 development (39). The present results show that pretreatment with IL-4 acts through the IL-4R on naive B cells to antagonize the pro-DN2 program. Together, these results suggest that IL-4 acts as an important regulator to maintain a subset of B cells as rNAV B cells and that IL-4-stimulated B cells exhibited a developmental trajectory different from that of type I IFN, type II IFN, or TLR7-stimulated B cells.

The opposing role of IL-4 and IFNs in lymphocyte development has been reported previously (29). Four decades ago, Paul et al. originally identified IL-4 as B cell stimulatory factor 1 (48–50), which promoted uncommitted B cells to switch to IgE and IgG1, and this could be inhibited by IFN-γ (51–55). IL-4 was even later shown to act on hematopoietic progenitor cells (56). In T cells, IL-4 promotes the transcription factor GATA3 as a mechanism promoting Th2 development and suppressing Th1 development (57). Our analysis of the transcriptomics of B cells further suggests that the B cell developmental fate decision is committed at the early Tr stage and that the signaling competition between type I IFN versus IL-4 at this stage may alter the developmental trajectories of B cells to all subsequent stages. In this model, B cells that have been stimulated with a stronger type I IFN signal will be committed to developing into ISG^hi cells and thereby aMEM B cells. In contrast, B cells that have been stimulated with a stronger IL-4R signal at the Tr stage will display the transcriptomic imprint of IL-4R response genes such as FCER2 (or CD23). In the spleen, CD23^hi B cells primarily develop into rNAV follicular B cells that can later become the precursors of GC B cells (58). This is consistent with the present finding that genes upregulated in the aNAV B cells of HC encode proteins that were located in the GC. Our results suggest that the lack of IL-4R signaling acts together with the elevated ISG to enforce B cell development into the extrafollicular trajectory and differentiation into aMEM in ISG^hi patients with SLE (59).

The scRNA-seq data were further supported by the finding that B cells with a phenotype of IL-4R⁻IFNβ⁺ at the naive stage developed a higher percentage of CD21⁻IgD⁻ B cells and a lower percentage of CD27⁺IgD⁻ B cells. These results are consistent with previous findings by Jenks et al., who showed that CXCR5⁻CD21⁻CD11c⁺ DN2 are developmentally related to aNAV B cells (39). Complement receptor 2 (CR2/CD21) has been implicated in lupus susceptibility in both human and animal models of disease (60–62). CD21^−/low B cells that are functionally anergic are increased in patients with primary Sjögren syndrome (63) and produce anti-ANA auto-Abs (64). Consistent with these results, we have identified a significantly higher percentage of IL-4R⁻IFNβ⁺ naive B cells in patients who developed anti-DNA. Because there was a significant association of a haplotype formed by the major alleles of three CR2 single-nucleotide polymorphisms with SLE (65), it will be interesting to determine if CR2 single-nucleotide polymorphisms are associated with the development of IL-4R⁻IFNβ⁺ naive B cells in patients with SLE.

We previously showed that B cell endogenous IFN-β and renal disease are higher in AA than in White patients with SLE (10, 66). We would therefore predict that in AAs, disease severity would be predominately determined by type I IFN and less dependent on the dichotomous action of IL-4R. Indeed, there was a positive correlation of SLEDAI with the IL-4R⁻IFNβ⁺ subpopulation in both White and AA subjects, but it reached statistical significance only in the White patients with SLE group. Consistent with this, White patients who with positive results for anti-Sm, anti-DNA, and renal disease had a statistically significant increase in the IL-4R⁻IFNβ⁺ subpopulation of B cells. Thus, the combined analysis of IFN-β and IL-4R is most predictive of SLEDAI and auto-Ab production in White patients with SLE.

A notable observation of the present study is that several of the transcriptomic signature genes, including TFs that are well known to participate in GC and plasma B cell development, were upregulated in Tr and rNAV B cells of HC compared with subjects with SLE. This is consistent with the finding that B cells that are IL4R⁺ and ISG^lo exhibited a developmental trajectory to GC and cMEM B cells. For example, MEF2C and BACH2 were identified as 2 of the top 25 genes in Tr and rNAV B cells in HC compared with SLE, respectively. Although MEF2C has been shown to be required for Ab class switch in a T-dependent response (67), BACH2 was implicated in acting together with BCL6 to orchestrate the GC B cell development (68). SPIB, which was identified to be upregulated in all subsets of B cells in HC compared with SLE, has been found to be associated with the expression of CD23 and follicular B cell survival in mouse spleen (69). Furthermore, several commonly shared TFs by upregulated genes in HC IGHM⁺IGHD⁺ B cells also have been shown to play an important role in GC and plasma B cell development. Among these TFs, IKZF1, which encodes IKAROS, has been identified to regulate B cell tolerance to TLR7 in CD23⁺ B cells (70). Mouse studies have directly confirmed the close connection of Ikaros, Mef2c, and Bach2 (71). Consistent with the present finding, IKZF1 expression was lower in the PBMCs and renal biopsies of patients with SLE who developed lupus nephritis (72). Because IKZF1 polymorphisms have been identified as risk alleles for SLE (73) and IKAROS/AIOLOS degradation-based therapy for SLE is under development (74, 75), it will be important to discern the effects of such therapy in patients and whether auto-Ab are primarily developed through the DN2 or through the GC pathway.

One unresolved question is whether increased ISGs lower IL-4R or vice versa in patients with SLE. Our in vitro culture experiment indicates that the expression of IRF7 induced by IFN-β could be suppressed by IL-4. Because IRF7 was a commonly shared TF by many ISGs in SLE B cells, IL-4 therefore exhibits the ability to control elevated type I IFN and TLR7 signals in B cells. We previously showed that high B cell IFN-β and ISGs at the T1 stage enhanced TLR7 signaling in B cells (9, 10). The present results suggest that upregulation and expansion of IL4R⁺ B cells is the preferred pathway for normal development from the Tr to the naive stage. We propose that the expression of IL-4R is a signature of Tr B cells that have become RNP nonresponsive. This signature enables mature B cells to maintain their quiescent properties until they are further stimulated by neoantigens and additional signals provided by T cells. It is possible that type I IFN disrupts this IL-4R rNAV signaling to break RNP self-Ag tolerance.

In conclusion, our studies suggest a dichotomous action of type I IFN versus IL-4 in shaping B cell developmental trajectories. This model is supported by the findings that infections, especially viral infections, have long been associated with SLE (76, 77). In contrast, infection with parasitic worms has been shown to inhibit SLE through the induction of a Th2 response (78). Our present results suggest that signaling through IL-4R on B cells is a natural antagonizing factor for both type I and type II IFNs. Effective and safe therapy for SLE therefore should allow B cells to restore the homeostatic balance between these two arms of immune responses to achieve a long-term therapeutic effect.

The authors have no financial conflicts of interest.