Restoration of Follicular T Regulatory/Helper Cell Balance by OX40L–JAG1 Cotreatment Suppresses Lupus Nephritis in NZBWF1/j Mice

This article is featured in Top Reads, p.

The antinuclear autoantibodies play a key pathogenic role in lupus through the formation of immune complexes that induce systemic inflammation when deposited into multiple organs, such as blood vessels, joints, and kidneys (1). Lupus nephritis (LN) is one of the major risk factors leading to chronic and end-stage renal disease in lupus patients, and it contributes significantly to the morbidity and mortality associated with systemic lupus erythematosus (SLE) (2). T follicular helper (TFH) cell–dependent B cell responses play an important role in promoting humoral immunity against foreign pathogens (3). TFH cells drive the formation and maintenance of germinal centers (GCs), a key anatomical site in the B cell follicles, where TFH and B cells interact and provide mutual help via costimulatory interactions, such as CD40/CD40L and ICOS/ICOSL, and enable Ig class-switch recombination, somatic hypermutation (SHM), affinity maturation, and differentiation of B cells into plasma cells producing high-affinity Abs (3, 4). However, aberrant TFH cell–mediated GC reaction can play a critical role in SLE pathogenesis (5). TFH cells are characterized by the expression of chemokine receptor CXCR5 and lineage-specific transcription factor Bcl6 (6). In contrast, transcription factors Blimp1 and STAT5 negatively regulate TFH cell differentiation (7, 8). The process of TFH cell differentiation occurs in multiple stages: the initial priming of naive CD4⁺ T cells by dendritic cells (DCs) leads to expression of CXCR5, which facilitates migration of early TFH cells along the CXCL13 gradient to the border of the B cell follicles. In the next stage, early TFH cells interact with Ag-specific B cells in the T-B border or within B follicles to undergo further maturation and upregulate CXCR5, PD1, and ICOS expressions. During the third and the final stage, differentiated CXCR5⁺PD1^hi GC-TFH cells help B cells to form GCs. PD1 plays an essential role in the positioning of the TFH/T follicular regulatory (TFR) cells in the GCs and has been used as a marker to define GC-TFH cells (9).

CD4⁺Foxp3⁺ T regulatory cells (Tregs) play a pivotal role in suppressing self-reactive T effector (Teff) cells and thereby help maintain self-tolerance and suppress autoimmunity (10). Although the loss of Treg numbers and defective suppressive functions lead to autoimmunity, restoring the homeostatic balance between Teff cells/Tregs can suppress autoimmune diseases (11). Despite the proven clinical importance of Tregs in many autoimmune diseases, the precise role of Tregs in SLE remains unclear (12). In humans, studies have shown reduced (13, 14), indifferent (15, 16), or increased Tregs (17) in SLE patients compared with healthy control subjects. Similarly, functional defect in Tregs was observed in some studies (18–20), while some reported normal suppressive functions (16, 20). Few other studies have reported the resistant state of Teff cells to Treg-mediated suppression (21). The observed discrepancies could be related to differences in the markers used for defining Tregs, patient recruitment criteria, disease activity, and treatment regimen. Nevertheless, meta-analysis studies suggest that loss of Treg homeostasis might play an active role in SLE pathogenesis, and Treg numbers/functions were often associated with recovery posttreatment in a majority of studies (22).

TFR cells, a specialized subset of effector Tregs, show shared characteristics of both Tregs and TFH cells (4). Although some studies have identified natural Tregs as the precursors for TFR cells (23) and shown similar TCR specificities between Tregs and TFR cells (24), their de novo differentiation from naive T cells has also been reported (24, 25). Coexpression of Treg and TFH lineage-specific transcription factors Foxp3 and Bcl6 impart them with both Treg and TFH-like transcriptional signatures, respectively (26). TFR cells inhibit TFH-dependent B cell responses more robustly than natural Tregs, because of their colocalization in the B cell follicles and GCs (27). Stronger Ab response was seen in Tcrb^−/− mice adoptively transferred with TFR-depleted Tregs, compared with Tregs with intact TFR fraction (28). Impaired TFR response has been shown to exacerbate lupus in both mice and humans (29–31). Reduced circulating TFR frequency and increased TFH/TFR ratio, which correlated with disease severity, have been reported in SLE patients (30). In contrast, another study reported increased CD4⁺CXCR5⁺FOXP3⁺ TFR cells in the peripheral blood of SLE patients. The increased frequency of TFR was found to be positively correlated with autoantibodies and SLE Disease Activity Index scores (32). Although it is conceivable that impaired Treg/TFR cell homeostasis can contribute to lupus pathogenesis, the earlier reports are perplexing and raise questions on the precise role of TFR cells in lupus. However, it should be noted that these human studies analyzed circulating TFR cells from the peripheral blood, which is not an accurate reflection of their numbers or interactions in the GCs in the lymphoid organs where Ab responses are primarily regulated. Thus, we surmised that a detailed study using experimental lupus mice that closely resemble human lupus could provide valuable new insights into the roles of Tregs and TFR cells and whether their selective expansion can suppress GC-B cells response and consequently LN.

Despite the proven clinical relevance of TFR cells in lupus, very limited reports are available on the methods to expand TFR cells and factors regulating their expansion and homeostasis (30, 33). The currently used adoptive Treg therapy uses in vitro–expanded autologous Tregs sorted from patient’s peripheral blood and, therefore, cannot significantly supplement TFR cells because of their infinitesimal numbers in the peripheral blood. Hence an approach that can cause expansion of Tregs and TFR cells in vivo with sustained functions could be used to effectively treat lupus. Recently, we have established a TCR-independent method for the expansion of Tregs with sustained Foxp3 expression using soluble OX40L (a TNF superfamily ligand) and Jagged-1 (JAG1) (a Notch family ligand) (34, 35). Cotreatment with soluble OX40L and JAG1 increased epigenetically stable-functional Tregs, delayed the onset of diabetes in NOD, and alleviated experimental thyroiditis in CBA/j mice (34, 35).

In this study, we investigated the correlation between Tregs/TFH cells/TFR cells and GC-B cells and LN onset using NZBWF1/j mice, in which the pathogenesis of the disease resembles human SLE involving antinuclear Abs, a genetic link to SLE-related genes, and immune complex–mediated LN (36). Further, we evaluated whether soluble OX40L–JAG1 treatment can expand TFR cells and suppress LN. Our results strongly suggest a key role for impaired TFR cell homeostasis, which was associated with aberrant TFH cell expansion and GC-B cell response contributing to LN in NZBWF1/j mice, and demonstrated a, to our knowledge, hitherto unidentified novel method of in vivo TFR expansion using soluble OX40L–JAG1 treatment, which suppressed TFH-mediated GC response and LN.

NZBWF1/j (Stock #100008) mice were purchased from Jackson Laboratories. All animal experiments were approved and performed per the guidelines set forth by the Animal Care and Use Committee at the University of Illinois at Chicago.

Anti-mouse CD4 (clone #GK1.5), anti-mouse CD8a (53.6.7), anti-mouse CD25 (PC61.5), anti-mouse/Rat Foxp3 (FJK16S), anti-PD1 (29F.1A12), anti-CD38 (90), anti-mouse CD134/OX40 (OX86), anti-mouse TIGIT (GIGD7), anti-mouse GITR (DTA1), anti-mouse CTLA-4 (UC10-4B9), and appropriate isotype control Abs were purchased from eBioscience, Thermo Fischer Scientific. Anti-human/mouse Ki67 (11F6), anti-GL7 (GL-7), anti-mouse CXCR5 (L138D7), anti-mouse LAG-3 (C9B7W), and anti-mouse Tim-3 (B8.2C12) were from BioLegend (San Diego, CA). CD4⁺/CD4⁺CD25⁺ EasySep T cell isolation kits were from STEMCELL Technologies. Mouse recombinant OX40L-Fc was provided by Dr. A.L. Epstein (Keck School of Medicine, Los Angeles, CA) (37). Mouse recombinant JAG1-Fc was produced from stable CHO(r) cells expressing soluble JAG1 as described previously (38). T cells were cultured in PRIME-XV T Cell Expansion XSFM medium (Irvine Scientific). Purified F(ab′)₂ goat anti-mouse IgM (μ chain) Ab (clone #Poly21571) and Ultra-LEAF Purified anti-mouse CD40 Ab (1C10) were from BioLegend. Mouse anti-CD3 (clone #2C11), anti-CD28 (clone #PV10), anti-CD25 (clone #PC61), and isotype controls were purchased from Bio X Cell (Lebanon, NH). Mouse recombinant IL-2 was from eBioscience, Thermo Fisher Scientific.

Cells were washed with PBS containing 0.5% BSA, surface stained followed by fixation and permeabilization using Foxp3/Transcription Factor Staining Buffer Kit (Tonbo Biosciences), and stained with appropriate isotype controls and test Abs in the dark at 4°C. Samples were analyzed using CytoFLEX (Beckman Coulter), and data were analyzed using Kaluza v2.1 software (Beckman Coulter) and FlowJo_v10.6.1 (BD Biosciences). FACS was performed using MoFlo Astrios cell sorter (Beckman Coulter). Sort purity was >90% as confirmed by postsort analysis.

Twenty-three-week-old female NZBWF1/j mice were injected (i.p.) with recombinant OX40L (2 mg/kg body weight) and JAG1 (2 mg/kg body weight) once a week for 3 consecutive weeks. The optimal dose was chosen based on our previous studies (34, 35, and unpublished data). Age and sex-matched control mice received PBS. On week 26, mice were sacrificed and analyzed for Treg/TFR/B cells. For periodic treatment after three weekly injections from week 23 to 25, mice received injections at 10-d intervals until week 35. After 37 wk, mice were sacrificed for final analysis. For Treg depletion studies, 22-wk-old mice were treated weekly with either rat IgG1 isotype control or anti-CD25 (200 µg) (39), and mice were monitored for proteinuria until 37 wk.

Urine samples were analyzed for proteinuria using dipstick urine strip test. Proteinuria scoring was done as follows: <30 mg/dl = 1; 30–100 mg/dl = 2; 100–300 mg/dl = 3; 300–2000 mg/dl = 4; and dead mice = 5.

Kidneys were fixed in formalin for 48 h, and paraffin sections (5 μm) were stained with H&E and periodic acid–Schiff reagent. Renal histology was evaluated as previously described (40). Glomerular histology was evaluated by assessing 20 glomerular cross-sections (gcs) per kidney, and each glomerulus was scored on a semiquantitative scale (0–3 ranging from normal to severe disease): 0, normal cellularity (35–40 cells/gcs) and absence of glomerular crescents, sclerosis, inflammation, necrosis; 1, mild disease (glomeruli with mild hypercellularity [41–50 cells/gcs], focal and segmental proliferative changes, crescents, sclerosis, inflammation, necrosis); 2, moderate disease (glomeruli with moderate hypercellularity [50–60 cells/gcs], focal, segmental and/or global diffuse proliferative changes, crescents, sclerosis, inflammation, necrosis); and 3, severe disease (glomeruli with severe hypercellularity [>60 cells/gcs], focal, segmental and/or global diffuse proliferative changes, crescents, sclerosis, inflammation, necrosis). Tubulointerstitial pathology was assessed semiquantitatively on a scale of 0–3 in 10 randomly selected high-power fields. We assessed the interstitial infiltrates and damaged tubules on a scale of 0–3: 0, normal; 1, mild; 2, moderate; and 3, severe disease.

Serum samples were collected and assayed in triplicates for anti-dsDNA IgG (catalog [Cat.] #3031), total IgG (Cat. #3023), anti-dsDNA IgM (Cat. #3032), and total IgM (Cat. #3024) at different dilutions by ELISA using kits from Chondrex as per the manufacturer’s protocol.

Suppression assays were performed by coculturing anti-CD3 (0.5 µg/ml)-stimulated CD4⁺CD25⁻ (T conventional [Tconv]) and CD4⁺CD25⁺ (Treg) cells (50 × 10³) in the presence of 50 × 10³ APCs (from control mice) at Tconv/Treg ratios of 1:0, 1:1, 1:2, 1:4, and 1:8 for 3 d. The extent of suppression was measured using the division index of Tconv cells in no Treg control versus Treg cocultures.

CD4⁺CXCR5⁺PD1^hiGITR⁻ TFH and CD4⁺CXCR5⁺PD1^hiGITR⁺ TFR cell types from control versus OX40L-JAG1–treated mice using GITR as a surrogate marker for Foxp3 expression as described in previous studies (41, 42). For TFR suppression assays, B cells (1 × 10⁵), TFH cells (60 × 10³), and TFR cells (60 × 10³) sorted from control and OX40L-JAG1–treated mice were cocultured in the presence of F(ab′)₂ anti-IgM (µ chain) (5 µg/ml) and anti-CD3 (2 µg/ml) for 6 d. GL-7 expression in B cells was analyzed by flow cytometry.

IgH sequencing of the GC B cells was performed as described previously (43). GC-B cells from control and OX40L-JAG1–treated mice were sorted, and genomic DNA was isolated using DNeasy kit (Qiagen). The IgH V region DNA was PCR amplified using VHJ forward (5′-GCCTGA CATCTGAGGACTCTGC-3′) and IgH intronic enhancer reverse (5′-CCTCTCCAGTTTCGGCTGAATCC-3′) primers. The ∼1.2-kb amplicon spanning rearranged endogenous VH genes cloned into pMiniT 2.0 Vector using PCR cloning kit (New England Biolabs) and transformed into Escherichia coli 10-β competent cells. Colonies from each group were sequenced using forward 5′-ACCTGCCAACCAAAGCGAGAAC-3′ and reverse 5′ TCAGGGTTATTGTCTCATGAGCG-3′ primers. Sequences of all clones were mapped and aligned to the IgH locus. The number of mutations was calculated by performing IMGT (ImMunoGeneTics; http://www.imgt.org/BlastSearch/) blast.

Total RNA was purified from CD4⁺CXCR5⁻GITR⁻ Tconv, CD4⁺CXCR5⁻GITR⁺ Treg, CD4⁺CXCR5⁺PD1^hiGITR⁻ TFH, and CD4⁺CXCR5⁺PD1^hiGITR⁺ TFR cells sorted from control and OX40L-JAG1–treated mice using Qiagen RNeasy kit with the inclusion of on-column DNase treatment step. GITR was used as a surrogate marker for Foxp3 as described in previous studies (41, 42). RNA samples were quantified using NanoDrop One Spectrophotometer (Thermo Fisher Scientific) and analyzed for integrity using Agilent 4200 TapeStation. Sequencing libraries for Illumina sequencing were prepared in one batch in a 96-well plate using CORALL Total RNA-seq Library Prep Kit with Unique Dual Indices (PN M11696; Lexogen) with RiboCop HMR rRNA Depletion Kit V1.3 (PN K03796; Lexogen). In brief, total RNA was used for the first rRNA depletion step, then followed by library generation initiated with random oligonucleotide primer hybridization and reverse transcription. No prior RNA fragmentation was done, because the insert size was determined by the proprietary size restricting method. Next, the 3′ ends of first-strand cDNA fragments were ligated with a linker containing Illumina-compatible P5 sequences and unique molecular identifiers. During the following steps of second-strand cDNA synthesis and double-stranded cDNA amplification, i7 and i5 indices, as well as complete adapter sequences required for cluster generation, were added. Final PCR-amplified libraries were purified and quantified, and average fragment sizes were confirmed to be 254–323 bp by gel electrophoresis using 4200 Tape Station and D5000 Screen Tape (PN 5067-5588; Agilent). The concentration of the final library pool was confirmed by quantitative PCR and then subjected to test sequencing to check sequencing efficiencies and accordingly adjust proportions of individual libraries. Sequencing was carried out on NovaSeq 6000 (Illumina), S4 flow cell, and ∼30 M 2 × 150 bp reads per sample. RNA sequencing (RNA-seq) data have been submitted to NCBI-GEO and can be accessed using GEO accession number GSE181433 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE181433).

Raw reads were aligned with the reference genome mm10 using STAR aligner. ENSEMBL genes were quantified as feature counts (44). Differential expression statistics (fold change [FC] and p value) were computed using edgeR in the generalized linear model framework to test for effects caused by T cell populations and treatments simultaneously, including the interaction between those factors, in addition to pairwise comparisons between different groups. In all cases, p values were adjusted for multiple testing using the false discovery rate correction of Benjamini and Hochberg. Pathway enrichment statistics for differentially expressed genes were obtained using the Core Analysis function in Ingenuity Pathway Analysis (IPA).

Tconv cells from control and OX40L-JAG1–treated mice were cultured with anti-CD3/CD28 stimulation for 24 h, and lactate levels were quantified (Lactate Assay Kit; Millipore Sigma, St. Louis, MO). Lactate levels were normalized to cell numbers and expressed as millimolars per liter (mM/l). From the same experiments, cells were stained with MitoSpy Green and MitoSpy Red CMXRos (BioLegend) and analyzed by flow cytometry.

Tconv cells from control OX40L-JAG1–treated mice were stimulated with anti-CD3/anti-CD28 mAbs (2 µg/ml) for 24 h, and total RNA was isolated using Qiagen RNeasy Plus Kit. mRNA expression levels of cytokines, such as IFN-γ, IL-2, TNF-α, IL-4, IL-6, IL-17, IL-10, and TGF-b, were analyzed by quantitative real-time PCR. Relative expression was calculated by the comparative ΔΔCt method, and results were expressed as relative FC over control mice after normalization with GAPDH internal control.

Statistical analyses were performed using Prism GraphPad (V9.0). Data were expressed as mean ± SEM. Student t test was used to compare two groups, whereas ANOVA with Tukey’s multiple comparisons was used to compare more than two groups. Kaplan–Meier survival analysis and log rank test were performed to determine significant differences in survival. A p value <0.05 was considered significant. Pearson correlation analysis was used to determine the correlation between GC-B cells, TFH cells, and TFR/TFH ratios in control and OX40L-JAG1–treated mice.

At first, we analyzed the differences in Tregs, TFH cells, and TFR cells in NZBWF1/j mice before and after the onset of proteinuria and examined their correlation to GC-B cells and autoantibody production. Although some studies have used CXCR5 as the sole marker for defining TFH cells (30, 32), coexpression of other surface markers such as PD1 and/or ICOS is more widely used to define TFH (45). We preferred to use CXCR5⁺ and PD1 costaining because they could potentially mark TFH/TFR cells present in the GCs.

We followed the gating strategy shown in (Fig. 1A to gate total CD4, total TFH (CD4⁺CXCR5⁺PD1^hi), Tregs (CD4⁺CXCR5⁻Foxp3⁺), and TFR (CD4⁺CXCR5⁺PD1^hiFoxp3⁺) cells in the spleen and renal lymph nodes (RLNs) of female NZBWF1/j mice showing proteinuria score ≥ 2; 23–30 wk of age) versus no proteinuria (<2; 10–15 wk of age). We found a marked reduction in the total CD4⁺ T cells in the RLNs of mice with proteinuria compared with those with no proteinuria (Fig. 1B). In contrast, the frequency of total TFH cells and Tregs was significantly increased, and the proportion of TFR cells in total TFH cells was decreased in both spleen and RLNs of mice with proteinuria (Fig. 1C–E). There was a significant increase in the ratio of % TFH/% CD4 and a reduction in % TFR/total TFH cells observed in both spleen and RLNs (Fig. 1F, 1G). Surprised by the observation of increased Tregs and reduced TFR cells in mice with proteinuria, we analyzed Foxp3 expression in these cell types (Fig. 1H, 1I) and found no significant difference in the Foxp3 expression per se. Further, we analyzed GC B cells (CD3⁻CD19⁺GL-7⁺) using the gating strategy shown in (Fig. 1J and noted a significant increase in GC B cells in the spleen and RLNs of mice with proteinuria (Fig. 1K).

We also observed increased anti-dsDNA IgG levels in mice with proteinuria (Fig. 1L), which correlated with increased GC-B cells and proteinuria levels in these mice. Taken together, our results showed a positive correlation between TFH cells, GC-B cells, autoantibody levels, and proteinuria onset in lupus-prone NZBWF1/j mice. Although there was a systemic increase in Tregs, the TFR subset was reduced in mice with proteinuria. Reduction in TFR cells correlated with increased TFH and GC-B cells, autoantibody levels, and proteinuria onset.

Reduced TFR cells despite increased Tregs seen in mice with proteinuria suggested the impaired ability of Tregs to differentiate into TFR cells that are functionally superior and better localized to suppress GC reactions that could have contributed to the development of LN. Nonetheless, increased Tregs could be correlated with proteinuria onset as well. We depleted Tregs in NZBWF1/j mice at 22 wk of age using an anti-CD25 mAb (Supplemental Fig. 1A, 1B) and found an accelerated onset of severe proteinuria at 26 wk of age when compared with isotype control Ab-treated mice that developed severe proteinuria (proteinuria score ≥ 3) ∼32 wk (Supplemental Fig. 1C). These results strongly suggested that Tregs are critical to preventing the rapid development of LN in NZBWF1/j mice. However, they may be functionally defective or unable to undergo further differentiation into TFR cells required to confer complete protection against LN onset.

Our earlier studies have shown selective expansion of functional Tregs in various strains of mice on treatment with soluble OX40L–JAG1 (34, 35). We analyzed the expression levels of OX40, Notch receptors 1–4, and Rbpj expression in Tconv, Tregs, TFH, and TFR cells sorted from 26-wk-old female NZBWF1/j mice by RNA-seq analysis and found preferential overexpression of OX40 and Notch2 on Foxp3⁺ Tregs and Foxp3⁺ TFR cells over Foxp3⁻ Tconv and Foxp3⁻ TFH cells in NZBWF1/j mice. Rbpj was highly expressed in Tregs, TFH cells, and TFR cells compared with Tconv cells (Fig. 2A). Next, we tested whether OX40L–JAG1 cotreatment can expand Tregs and TFR cells in NZBWF1/j mice. We treated 23-wk-old female NZBWF1/j mice with soluble OX40L and JAG1 once a week for 3 consecutive weeks and analyzed changes in the total CD4 cells, TFH cells, Tregs, and TFR cells in the spleen and RLNs.

We found a significant increase in total CD4⁺ T cells (Fig. 2B), but a reduction in total TFH cells in the spleen (Fig. 2C) of OX40L-JAG1–treated mice. Although there was an increase in total CD4⁺ T cells, total TFH cells were not significantly reduced in RLNs (Fig. 2B, 2C). As shown in (Fig. 2D–F, OX40L–JAG1 treatment caused a 2- to 3-fold increase in Tregs in the spleen and RLNs, and TFR cell frequency was also increased by 1.5-fold in the spleen and RLNs after three treatments with no apparent changes in Foxp3 expression per se. The % total TFH/% CD4 ratio was significantly reduced in the spleen (Fig. 2G) of OX40L-JAG1–treated mice. Although an increase in Tregs was observed in the RLNs, the % TFH/CD4⁺ ratio was not significantly altered (Fig. 2G). Intriguingly, despite a less prominent increase in the frequency of TFR cells in the RLNs, % TFR/% total TFH ratio was significantly increased in both spleen and RLNs of OX40L-JAG1–treated mice (Fig. 2H). More importantly, we found an ∼3-fold increase in Ki67⁺ proliferating cells within Tregs and TFR cells from OX40L-JAG1–treated mice in both spleen and RLNs (Fig. 2I–K), consistent with our previous studies in other strains of mice (35). These results suggested that increased Tregs and TFR cells seen in OX40L-JAG1–treated mice could be caused by increased proliferation. Besides, we also noted an increase in induced Treg (iTreg) differentiation in vitro from Tconv cells isolated from OX40L-JAG1–treated mice compared with control mice (Fig. 2L, 2M). Furthermore, GC-B cells in the RLNs were reduced by 2- to 3-fold in OX40L-JAG1–treated mice (Fig. 3A, 3B). The frequency of plasma B cells in the spleen was moderately reduced in OX40L-JAG1–treated mice (Fig. 3C). We also observed a significant reduction in anti-dsDNA IgG autoantibody levels in OX40L-JAG1–treated mice (Fig. 3D). Anti-dsDNA IgM, total IgG, and IgM levels were slightly increased, but the increases were not statistically significant (Fig. 3E–G). These results suggested that OX40L–JAG1 treatment increased Tregs, TFR cells, and TFR/TFH ratio, which correlated with a reduction in GC-B cells and autoantibody levels in NZBWF1/j mice.

Specific reduction in anti-dsDNA IgG levels with no significant difference in anti-dsDNA IgM levels in OX40L-JAG1–treated mice, along with decreased TFH and GC-B cells, indicated an interruption in the GC events such as IgG class switching, which is essential for high-affinity Ab production. We noted a significant reduction in the frequency of CD40L⁺Foxp3⁻ TFH cells in the spleen and RLNs of OX40L-JAG1–treated mice (Fig. 3H–J). Furthermore, we analyzed the expression levels of other coinhibitory receptors and found a significant increase in the frequency of Foxp3⁻ TFH cells expressing LAG3, TIGIT, and CTLA4 in both spleen and RLNs of OX40L-JAG1–treated mice (Fig. 3H–J). To further determine whether the suppressive phenotype of TFR cells is affected by OX40L–JAG1 treatment, we analyzed markers associated with Treg suppressive functions and lineage stability, such as CD25, CTLA4, GITR, and Helios, in TFR cells (46). We did not see constitutive expression of CD25 and CTLA4 in TFR cells as seen with Tregs. In contrast, GITR and Helios were constitutively expressed on TFR cells in both control and OX40L-JAG1–treated mice. However, we noted a significant increase in CD25⁺ and CTLA4⁺ TFR cells in OX40L-JAG1–treated mice (Fig. 4A–C). To further understand the functional relevance, we cocultured B cells, CD4⁺CXCR5⁺PD1^hiGITR⁻ TFH cells, and CD4⁺CXCR5⁺PD1^hiGITR⁺ TFR cells from control and OX40L-JAG1–treated mice in the presence of anti-IgM/anti-CD3 stimulation and analyzed GL-7 expression in B cells as a marker for activation. As expected, TFH coculture increased GL-7 expression in B cells (Fig. 4D, 4E) compared with B cells cultured alone, and TFR cocultures suppressed TFH-induced GL-7 expression. Intriguingly, TFH-B cells cocultured from OX40L-JAG1–treated mice showed a significant reduction in GL-7 expression compared with that of control mice (Fig. 4D, 4E). But we did not find any significant difference in the TFR cell suppressive activity between control versus OX40L-JAG1–treated mice (Fig. 4E).

Next, we analyzed whether the SHM process was affected in the GC-B cells by sequencing the V region of IgH locus. There was a significant reduction in the average number of mutations per clone in the GC B cells from OX40L-JAG1–treated mice compared with control mice (Fig. 4F). In addition, the proportion of clones showing >15 mutations (∼1% total amplicon length) was more prominent in control mice compared with OX40L-JAG1–treated mice (Fig. 4G). Taken together, these results suggested attenuated TFH cell effector functions in OX40L-JAG1–treated mice that could affect the key GC events such as class switching and SHM that might bring about qualitative changes in Ab production in addition to reduced GC-B cells seen in these mice.

Because OX40L–JAG1 treatment increased TFR cells and reduced GC-B cells and autoantibody levels after the initial three treatments, we tested whether prolonged treatment could ameliorate LN in these mice and determined the effect on Tregs, TFH cells, TFR cells, and GC-B cells. After three initial weekly treatments from 23 to 25 wk of age, mice were treated at 10-d intervals until 35 wk of age, and proteinuria development was monitored every week. By 34 wk of age, 80% of control mice developed severe proteinuria (proteinuria score ≥ 3), whereas none of the OX40L-JAG1–treated mice developed severe proteinuria, as evidenced by respective proteinuria scores (3.7 ± 0.45 [control] versus 1.1 ± 0.1 [OX40L-JAG1]) (Fig. 5A). Furthermore, 40% of the control mice reached a humane endpoint by 37 wk compared with none in the OX40L-JAG1–treated group (Fig. 5B). Histopathological analysis showed increased glomerular hypercellularity and proliferative changes with neutrophilic inflammation, cellular crescents (Fig. 5Cii), necrosis and sclerosis, interstitial lymphocytic infiltrates, and tubular dilatation with luminal casts (Fig. 5Ci, iii) in the control mice as compared with the OX40L-JAG1–treated mice, which had significantly lower glomerular and tubulointerstitial pathology scores (Fig. 5C, 5D). These results suggested that periodic OX40L–JAG1 treatment could suppress LN.

Immunophenotypic characterization showed significantly increased CD4⁺ T cells in the spleen and RLNs of OX40L-JAG1–treated mice at the end of the experiment at 37 wk (Fig. 5E). There was no significant reduction in total TFH cells seen in the RLNs and spleen (Fig. 5F). Intriguingly, there was no significant increase in Tregs seen after prolonged OX40L–JAG1 treatment despite a massive increase seen after the initial three treatments (Fig. 5G). In contrast, TFR cells were significantly increased (>2-fold) in the RLNs of OX40L-JAG1–treated mice (Fig. 5H). The % TFH/CD4⁺ ratio was significantly reduced (Fig. 5I) and % TFR/total TFH ratio was increased (∼3- to 4-fold) in the RLNs of OX40L-JAG1–treated mice with a concomitant reduction in GC-B cells (Fig. 5J, 5K). There was a modest increase in anti-dsDNA-IgM and a decrease in anti-dsDNA IgG levels seen in OX40L-JAG1–treated mice, although the differences were not statistically significant (Fig. 5L, 5M). The Pearson correlation analysis between % TFR versus % GC-B cells (Fig. 5N) and TFR/TFH versus % GC-B cells (Fig. 5O) showed a significant negative correlation of TFR and TFR/TFH ratio to GC-B cells. Altogether, these results revealed the critical importance of TFH and GC-B cells in mediating lupus pathogenesis and a protective role played by TFR cells in controlling GC reactions in lupus. Thus, expansion of TFR cells by OX40L–JAG1 treatment could be beneficial in inhibiting TFH and GC-B cells and ameliorating LN. Moreover, we did not observe any significant differences in the total and differential blood count parameters or liver function parameters in control versus OX40L-JAG1–treated mice except a significant reduction in monocytes in OX40L-JAG1–treated mice (Supplemental Fig. 2A–X). Thus, the long-term treatment did not appear to provoke significant undesirable adverse events.

Next, we sought to investigate how OX40L–JAG1 treatment could affect different Th cell compartments to bring about these changes. We sorted Tconv, Treg, TFH, and TFR cell types from control versus OX40L-JAG1–treated mice using GITR as a surrogate marker for Foxp3 expression, as described in previous studies (41, 42), and performed RNA-seq analysis. We used q < 0.05 and log fold change > 1 (i.e., 2-fold difference) as cutoffs to select differentially expressed transcripts. Compared with control mice, in OX40L-JAG1–treated mice, 707, 1,633, 1,311, and 2,251 transcripts were upregulated, while 2,177, 520, 158, and 344 transcripts were downregulated in Tconv, Tregs, TFH, and TFR cells, respectively. These data showed that OX40L–JAG1 treatment modulated gene expression profiles in all four T cell subtypes. We analyzed the similarity/dissimilarity among different cell types between control and OX40L-JAG1–treated mice by using the Euclidean distance matrix. As shown in Supplemental Fig. 3A–C, the mean Euclidean distance between Tconv cells and Tregs in OX40L-JAG1–treated mice was shorter than that of control mice, suggesting the adaptation of Treg-like gene signature in Tconv cells. This is consistent with an ∼3-fold increase in iTreg differentiation and increased Tregs (∼50% total CD4⁺ T cells) seen in OX40L-JAG1–treated mice (Fig. 2E, 2L, 2M).

IPA identified several canonical pathways related to differentially expressed genes between control and OX40L-JAG1–treated groups. Relevant pathways with −logB-H p > 1.3 (<0.05) and Z scores > +2 or < −2 in each cell type are summarized in Supplemental Table I. The differential expression pattern of genes related to the most relevant pathways in each cell type is summarized in heatmaps (Fig. 6A–D). Intriguingly, many pathways significantly modulated in Tconv cells from OX40L-JAG1–treated mice were primarily inhibited as implied by negative Z scores, except PTEN and Rho GDP-dissociation inhibitor signaling pathways. The major pathways inhibited include IL-15 production, GP-6, IL-8, Tec kinase, STAT3, Sphingosine 1-phosphate, P70S6K, fatty acid α-oxidation, HIF-1α, and NF-κB signaling pathways. Upregulation of PTEN and negative regulation of P70S6K signaling suggest inhibition of the PI3K–AKT–mTOR (mammalian target of rapamycin) axis, which is the major pathway regulating T cell activation (47). Besides, downregulation of STAT3 and HIF-1a signaling, which play a crucial role in Th17 and TFH cell differentiation, suggested inhibition of the differentiation of Tconv cells to these lineages (48–50). In contrast, the vast majority of affected pathways in Tregs from OX40L-JAG1–treated mice showed activation rather than inhibition. Kinetochore metaphase signaling pathway, estrogen-mediated S-phase entry, cyclins and cell-cycle regulation, aryl hydrocarbon receptor, cyclins and cell-cycle regulation, cell-cycle regulation by BTG family proteins, PI3K signaling in B cells, paxillin signaling, cell-cycle control of chromosomal replication, 14-3-3–mediated signaling, glycolysis I, and chemokine signaling pathways were activated in OX40L-JAG1–expanded Tregs. In addition, the role of CHK proteins in cell-cycle checkpoint and G2/M DNA damage checkpoint was inhibited in Tregs. Thus, it is apparent that signaling pathways related to cell-cycle progression were predominantly affected by OX40L–JAG1 treatment, which is consistent with increased Tregs seen in these mice. The major pathways activated in TFH cells by OX40L–JAG1 treatment included integrin linked kinase (ILK) signaling, integrin, GP6, HMGB1, Tec kinase, IL-15 production, IL-6 signaling, 14-3-3–mediated signaling, HIF-1α signaling, and IL-8 signaling pathways. Two pathways, Rho GDP-dissociation inhibitor and PPAR signaling pathways, were significantly inhibited in TFH cells from OX40L-JAG1–treated mice. It should be noted that several pathways like ILK, IL-8, Tec kinase, and HIF-1a that were inhibited in Tconv cells were found to be activated in TFH cells. In the case of TFR cells, the major canonical pathways activated included HIF-1α signaling, cell-cycle control of chromosomal replication, kinetochore metaphase signaling pathway, and ILK signaling, while no pathways were found to be significantly inhibited. Altogether, signaling pathways related to T cell activation and differentiation, such as STAT3, HIF-1a, P70S6K, and NF-κB, were significantly inhibited in Tconv cells from OX40L-JAG1–treated mice, while cyclins and cell-cycle regulation–related pathways were activated in Tregs and TFR cells.

Previous studies have reported chronically activated phenotype, upregulated mTOR activity, and mitochondrial mass in CD4⁺ T cells from SLE patients in contrast with healthy individuals (51, 52). An earlier study reported suppression of lupus in triple-congenic, B6-lpr, and NZBWF1/j strains of mice with dual inhibition of glycolysis and mitochondrial metabolism by the combination of 2DG and metformin, which was associated with markedly decreased mTORC1 activation and autoreactive TFH cells (53). Downmodulation of PI3K–mTOR–P70S6K and HIF-α signaling pathways in Tconv cells from OX40L-JAG1–treated mice, which are major regulators of glycolysis (48), suggested altered metabolic and activation state on OX40L–JAG1 treatment. Therefore, we characterized the expression of multiple coinhibitory receptors, such as PD1, LAG3, TIGIT, and CTLA4, in Tconv cells from control and OX40L-JAG1–treated mice. We found a significant increase in PD1⁺, LAG3⁺, and TIGIT⁺ cells, but not CTLA4⁺ Tconv cells, in the spleen and RLNs of OX40L-JAG1–treated mice (Fig. 7A–C). The expression of multiple coinhibitory receptors is considered a feature for T cell exhaustion (54). Given the downmodulation of T cell activation signaling pathways and coexpression of multiple coinhibitory receptors, we speculated an exhausted phenotype of Tconv cells from OX40L-JAG1–treated mice. One of the salient features of T cell exhaustion is that the cells are refractory to TCR-induced proliferation. In general, CD4⁺Foxp3⁺ Tregs are considered to play a protective role against autoimmunity (55, 56). However, increased Tregs seen in proteinuric NZBWF1/j mice indicated a plausible loss in suppressive functions. Therefore, we analyzed the functional competency of Tregs from control and OX40L-JAG1–treated mice while examining the ability of Tconv cells to proliferate in response to TCR stimulation. We cocultured Tconv cells and Tregs from control and OX40L-JAG1–treated mice in “mix and match” combinations with TCR stimulation. As shown in (Fig. 7D, there was no significant difference in the functional ability of Tregs from control and OX40L-JAG1–treated mice in suppressing Tconv cell proliferation from either group of mice. However, there was a significant reduction in the ability of Tconv cells from OX40L-JAG1–treated mice to proliferate in response to TCR stimulation even in the absence of Tregs, suggesting exhaustion of Tconv cells from OX40L-JAG1–treated mice. Furthermore, we analyzed the mRNA expression of proinflammatory and anti-inflammatory cytokine genes in CD4⁺ T cells after anti-CD3/CD28 stimulation for 24 h. We noted a significant reduction in Ifn-g, Il-4, Il-17, Tnf-a, and Il-2 genes, whereas Il-10, Tgf-b, and Il-6 expressions were not significantly different between control and OX40L-JAG1–treated mice (Fig. 7E). Altogether, these results indicated an exhausted phenotype of Tconv cells in OX40L-JAG1–treated mice.

To further determine whether the metabolic state of the Tconv cells was altered on OX40L–JAG1 treatment, we stimulated Tconv cells from the control and OX40L-JAG1–treated mice with anti-CD3/CD28 and analyzed lactate levels in the cell culture supernatants as a measure of anaerobic glycolysis. As expected, lactate levels were significantly reduced in Tconv cells from OX40L-JAG1–treated mice (Fig. 7F), consistent with the activation of PTEN and inhibition of p70SK and HIF-1a signaling seen in this cell type. Furthermore, we analyzed mitochondrial mass and membrane potential by MitoSpy Green and MitoSpy Red staining. In brief, MitoSpy Green uptake by cells is independent of membrane potential and is directly proportional to mitochondrial mass, whereas MitoSpy Red uptake depends on membrane potential and can act as an indicator of mitochondrial fitness or healthiness. Tconv cells from OX40L-JAG1–treated mice had reduced MitoSpy Green and Red staining (Fig. 7G–I), indicating impaired mitochondrial mass and fitness, which could be associated with mitochondrial dysfunction. Collectively, our results suggested an altered metabolic state and exhausted phenotype in Tconv cells from OX40L-JAG1–treated mice that could suppress their differentiation into TFH lineage.

Although Tregs have been shown to suppress autoreactive T cells in many autoimmune diseases (56), contrasting reports have shown reduced or unchanged or increased Tregs in the peripheral blood from SLE patients compared with healthy control subjects. We have noted increased CD4⁺Foxp3⁺ Tregs in the thymus (data not shown), spleen, and RLNs of NZBWF1/j mice on onset of proteinuria compared with mice that showed no proteinuria. An earlier study observed progressive homeostatic imbalance of Treg/Tconv cells in NZBWF1 mice by comparing young, onset, and diseased NZBWF1/j mice with age-matched BALB/c mice (57). However, similar to our observation, increased Tregs in the thymus, spleen, and peripheral lymph nodes of diseased NZBWF1/j mice compared with young and onset NZBWF1/j mice were seen in this study, suggesting that increased Tregs alone was insufficient to suppress disease progression (57). To determine the precise role of Tregs in lupus pathogenesis, we depleted Tregs using an anti-CD25 Ab and noted accelerated LN mice on depletion of Tregs. Thus, at least a subset of Tregs could be critical to confer protection against LN onset. Independent studies have shown that the administration of IL-2 or IL-2/anti–IL-2 complexes expanded Tregs that suppressed LN in NZBWF1/j mice (57, 58). However, these studies did not consider the possible effects of IL-2 augmentation on other key cell subsets involved in lupus pathogenesis, such as TFH/TFR cells. It should be noted that IL-2–induced STAT5 activation is one of the prominent suppression mechanisms of TFH cell differentiation (8), and IL-2 can expand TFR cells that can negatively regulate autoantibody production (59).

GC reaction has been described as the predominant pathway by which high-affinity IgG-specific autoantibodies are produced after class-switch recombination and SHM, although extrafollicular pathways have been recently implicated (60). A growing body of evidence has demonstrated the crucial role played by TFH cells in the development and preservation of GCs (61–63), and increased TFH and GC-B cell responses have been reported in both experimental and human lupus (64–66). We noted a significant reduction in the TFR cells in mice with proteinuria, which correlated with increased TFH and GC-B cells. Although Tregs present in higher numbers in NZBWF1/j mice with severe proteinuria, they were not fully capable of suppressing these TFH cell responses or LN onset/progression. TFR cells localized in the GCs and B cell zones are better positioned to inhibit TFH cells and GC response than Tregs that are predominantly present in T cell zones. This is evident from the studies showing the superior ability of TFR cells to inhibit Ab production compared with conventional Tregs (28). Because TFR cells have been shown to differentiate mainly from thymus-derived natural Tregs (67), which are present in higher numbers in NZBWF1/j mice during LN onset, it is possible that the Treg-to-TFR differentiation or TFR cell expansion in the periphery could be impaired in lupus-prone NZBWF1/j mice.

Several studies have reported a negative impact of OX40 signaling on Treg/TFR functions (68–72), and its ability to induce TFH cell response during infections and OX40L gene locus is associated with human lupus (73–75). In our earlier studies, we have shown preferential overexpression of OX40 on human and murine thymic Treg precursors and matured thymic Tregs, and OX40L expression on medullary thymic epithelial cells and DCs. Moreover, OX40L promoted IL-2–dependent maturation and proliferation of thymic Tregs. OX40^−/− mice had significantly reduced thymic Treg precursors and matured thymic Tregs. These findings demonstrated the critical role of OX40 signaling in Treg generation in both mice and humans (35, 76). In this study, we noted a preferential overexpression of OX40 on Tregs and TFR cells over Tconv and TFH cells in NZBWF1/j mice. Besides, we did not find a loss of suppressive functions in Tregs isolated from OX40L-JAG1–treated mice relative to control mice in our previous studies (34, 35, 76), as well as in this study. Consistent with our findings, a recent study using IgG1-specific OX40 agonist and Foxp3.GFP reporter mice has demonstrated expansion of Tregs in lymphoid organs without impairing suppressive functions or lineage specificity (77). Nevertheless, another study reported depletion of Tregs by IgG2a isotype–specific anti-OX40 Ab, which was due to FcγR-mediated Ab-dependent cellular cytotoxicity as a result of constitutive expression of OX40 on Tregs rather than direct OX40-mediated signaling (78). Therefore, the divergent effects of OX40 targeting could, at least in part, be attributed to reagents that were used to target OX40 signaling. Earlier, we have shown that adoptive transfer of OX40L⁺JAG1⁺ bone marrow–derived DCs, but neither OX40L⁺JAG1⁻ nor OX40L⁻JAG1⁺ bone marrow–derived DCs, could protect CBA/j mice from experimental autoimmune thyroiditis (79). Similarly, OX40L–JAG1 cotreatment was found to be necessary to delay T1D onset in NOD mice (34). The cotreatment of JAG1 can sustain suppressive phenotype and functions of Tregs through its interaction with Notch3 (35). Interestingly, Notch3 deficiency in lupus-prone B6.Fas^lpr/lpr mice showed an aggravated lupus phenotype with significantly increased mortality, splenomegaly, lymphadenopathy, and nephritis. Furthermore, these mice had reduced Tregs compared with WT B6.Fas^lpr/lpr mice, indicating a protective role for Notch3 signaling in SLE through Treg expansion (80). Besides, Notch signaling on JAG1 binding has also been shown to negatively regulate TCR activation in previous studies (81). In line with these observations, we found increased Tregs, TFR cells, and TFR/TFH ratio in OX40L-JAG1–treated NZBWF1/j mice. More importantly, increased TFR cells and TFR/TFH ratio correlated with reduced GC-B cells and suppression of LN.

A previous study using a NZBWF1/j mouse model has shown exacerbation of renal disease by anti-OX40 agonistic Ab (82). However, similar to our findings, the investigators have reported increased frequency of Foxp3⁺ cells within splenic and kidney-infiltrating OX40⁺ cells in OX40 agonist–treated mice, which could be OX40⁺Foxp3⁺ Tregs, but they did not examine for TFH/TFR cells. Besides, they found no significant difference in the % nonsevere proteinuria (<300 mg/dl) when OX40 agonist was given to 13-wk-old mice, but a significant increase was seen when mice of 21–27 wk of age were treated. In a direct comparison, we used a low dose of OX40L (2 mg/kg) combined with JAG1 (2 mg/kg) (once a week) compared with a higher dose of OX40 agonist (10 mg/kg three times a week) used in their study. In general, dosing for therapeutic studies for established LN in NZBWF1/j mice begins at ∼21–26 wk of age; therefore, we started treatment at 23 wk of age. Taken together, it is possible that differences in the treatment strategy could have contributed to observed differences between the two studies. Besides, JAG1 signaling has also been shown to negatively regulate TCR activation in previous studies (81); thus, the cotreatment of JAG1 with OX40L might have balanced the TCR-dependent costimulatory effects of OX40 on Teff cells while driving the TCR-independent Treg proliferation (35).

In general, the formation of the GCs is influenced by the stimulating Ag encountered by early TFH cells and B cells at the T-B border, followed by migration into the B cell follicles (60, 83). B cells receiving stronger costimulatory and cytokine help from TFH cells are more likely to differentiate into GC-B cells. We noted a significant reduction in CD40L⁺ TFH cells in OX40L-JAG1–treated mice, suggesting an interruption in this major costimulatory pathway. CD40L-CD40 communication between TFH-B cells is critical for GC-B cell response (73). TFH cells from OX40L-JAG1–treated mice showed increased expression of several coinhibitory receptors and were less potent in inducing GL-7 expression in B cells in our in vitro coculture assays. Thus, the attenuated TFH cell functions in OX40L-JAG1–treated mice could be in part because of intrinsic inhibitory signaling as well. Within the GCs, B cells undergo rapid proliferation in the dark zone followed by positive selection in the light zone. During repeated rounds of proliferation, SHMs occur in the BCRs that affect affinity maturation. B cells that receive signals from TFH cells undergo more rounds of proliferation and inherit more mutations and affinity maturation (60). We noted a significant reduction in SHM in the IgH locus of GC-B cells from OX40L-JAG1–treated mice compared with the control mice, which is consistent with attenuated TFH cell help. Increased IgM-to-IgG class switching has been reported in both experimental mouse models of lupus and SLE patients and is one of the key events occurring in the GCs that entails help from TFH cells (84). We also noted reduced anti-dsDNA IgG Ab levels in OX40L-JAG1–treated mice with an insignificant increase in anti-ds DNA IgM. Altogether, our findings suggest interruption of SHM, affinity maturation, and class-switching events occurring in the GCs that reiterate impaired GC-B cell response induced by OX40L–JAG1 treatment.

Previous studies have shown that TFR cells reduced TFH-induced GL-7 expression in B cells in coculture experiments, which is an indication of suppression of B cell activation (85). Consistent with these findings, we noted a reduction in GL-7 expression in TFR cocultures compared with B-TFH cocultures. Nevertheless, the suppressive functions of TFR cells in both control and OX40L-JAG1–treated mice remained comparable in our in vitro assay despite an increase in CD25⁺ and CTLA4⁺ TFR cells in OX40L-JAG1–treated mice. However, under in vivo conditions, an increase in CD25-expressing TFR cells could make activated TFH cells deprived of IL-2, leading to impaired survival similar to the competition for IL-2 between Tregs and Teff cells, which is one of the major mechanisms of Treg-mediated suppression of Teff cells (86).

RNA-seq and IPA analyses showed inhibition of several key pathways involved in T cell activation and differentiation in Tconv cells from OX40L-JAG1–treated mice, such as p70S6K and HIF-1a, which are major regulators of glycolytic metabolism in T cells (47, 48). We noted increased expression of several coinhibitory receptors, reduced lactate levels, mitochondrial mass, and fitness in Tconv cells from OX40L-JAG1–treated mice. Moreover, Tconv cells from OX40L-JAG1–treated mice that were significantly refractory to TCR-induced proliferation and cytokine expression indicated plausible T cell exhaustion. All of these factors substantiate poor Teff cell functions in OX40L-JAG1–treated mice. Given the lack of significant increase in Tregs in 37-wk-old OX40L-JAG1–treated versus control mice despite ∼3-fold increased Tregs seen in 26-wk-old OX40L-JAG1–treated mice, it is likely that Tregs might have undergone apoptosis caused by reduced IL-2 production by exhausted Tconv cells in OX40L-JAG1–treated mice. Moreover, the reduction seen in TFH cells could also be a consequence of reduced HIF-1a and STAT3 signaling in Tconv cells (48) from OX40L-JAG1–treated mice, which are required for their differentiation into TFH lineage. In contrast, reduced glycolysis and mTOR signaling in Tconv cells could aid differentiation of Tconv cells into Treg lineage (52), which is consistent with increased Tregs seen in OX40L-JAG1–treated mice.

Although T cell exhaustion has been deemed as a major evasion mechanism impeding antitumor and antiviral response (87), its implications in autoimmunity are underexplored; therefore, the correlative interpretations need to be derived from viral and tumor immunity studies. Multiple mechanisms could induce and sustain T cell exhaustion, including chronic exposure to Ags with impaired costimulation/coinhibition signals, presence of immunosuppressive cytokines, and suppressor cells, such as Tregs and myeloid-derived suppressor cells (88). In the chronic lymphocytic choriomeningitis virus infection model, it has been demonstrated that Tregs and PD1 signaling synergistically act to induce CD8⁺ T cell exhaustion and simultaneous abrogation of PD1 signaling, and depletion of Tregs overcomes T cell exhaustion (89). Thus, increased Tregs seen in OX40L-JAG1–treated mice could cause T cell exhaustion. Similar to our findings, the presence of a T cell exhaustion–like signature has been noted in new-onset type 1 diabetes patients treated with teplizumab (anti-CD3 Ab), which is known to expand Tregs (90). Taken together, expansion of Tregs and TFR cells at extrafollicular and follicular regions could suppress effector functions of Tconv and TFH cells, which in turn attenuate autoimmune response in lupus.

Our study sheds light on the correlation between GC-TFR/TFH imbalance and GC-B cell response mainly in the lymphoid organs, such as RLNs and spleen, using relevant lupus-prone NZBWF/1j mice, which resolve the confounding findings from human lupus that looked at circulating TFR cells in the peripheral blood. Given the significant resemblance of LN pathogenesis between human and NZBWF1/j mice, it is highly possible that the impaired TFR/TFH homeostatic balance, especially in the GCs of RLNs, could be conserved in human LN. Thus, the identification of novel biomarkers that can accurately reflect GC responses could be of immense help. Moreover, we report a novel mechanism of TFR expansion and inhibition of TFH cell differentiation with OX40L–JAG1 treatment, which will have important implications for future drug development. However, our study has the following limitations as well. Being an F1 hybrid strain, introducing reporter transgenes into these mice is difficult. The use of GITR as a surrogate for Foxp3 allowed us to sort ∼80% Foxp3⁺ Tregs/TFR cells, and a proportion of GITR⁺ cells included non-Treg/TFR cells as well. Due to a lack of lineage-specific depleting Abs to TFR cells, we could not selectively deplete TFR in NZBWF1/j mice to understand the functional relevance of this cell type in controlling GC reactions. Therefore, we could not directly delineate the precise role of TFR/TFH cell imbalance in LN pathogenesis. However, the inverse correlation between TFR or TFR/TFH ratio to GC-B cells in control and OX40L-JAG1–treated mice is strongly suggestive of a key role for TFR/TFH cell imbalance in GC-B cell expansion and LN pathogenesis. In summary, we report a defect in Treg-to-TFR cell conversion, which could lead to reduced TFR/TFH ratio and enhanced TFH-GC-B cell response and progressive LN in NZBWF1/j mice. OX40L–JAG1 cotreatment can suppress TFH-GC-B cell response by expanding functional TFR cells and inhibiting TFH differentiation from Tconv cells, and thereby restoring TFR/TFH balance to suppress LN.

We thank Dr. Shigeru Chiba (Department of Clinical and Experimental Hematology, Institute of Clinical Medicine, University of Tsukuba, Tsukuba, Japan) for sharing JAG1-expressing stable CHO(r) cells. We also acknowledge the technical support and service provided by core research facilities at University of Illinois at Chicago. All sorting experiments were performed at the flow cytometry core facility. RNA-seq experiments were done at the Core Genomics Facility, and bioinformatics analysis was performed by the Core for Research Informatics. Histology experiments were performed by the Research Histology and Tissue Imaging Core Facility.

The authors have no financial conflicts of interest.