Murine Lupus Susceptibility Locus Sle1c2 Mediates CD4⁺ T cell Activation and Maps to Estrogen-Related Receptor Gamma Esrrg

Abstract

Sle1c is a sublocus of the NZM2410-derived Sle1 major lupus susceptibility locus. We have previously shown that Sle1c contributes to lupus pathogenesis by conferring increased CD4⁺ T cell activation and increased susceptibility to chronic graft-versus-host disease (cGVHD), which mapped to the centromeric portion of the locus. In this study, we have refined the centromeric sublocus to a 675Kb interval, termed Sle1c2. Mice from recombinant congenic strains expressing Sle1c2 exhibited increased CD4⁺ T cell intrinsic activation and cGVHD susceptibility, similar to mice with the parental Sle1c. In addition, B6.Sle1c2 mice displayed a robust expansion of IFNγ expressing T cells. NZB complementation studies showed that Sle1c2 expression exacerbated B cell activation, autoAb production, and renal pathology, verifying that Sle1c2 contributes to lupus pathogenesis. The Sle1c2 interval contains two genes, only one of which, Esrrg, is expressed in T cells. B6.Sle1c2 CD4⁺ T cells expressed less Esrrg than B6 CD4⁺ T cells, and Esrrg expression was negatively correlated to CD4⁺ T cell activation. Esrrg encodes for an orphan nuclear receptor that regulates oxidative metabolism and mitochondrial functions. In accordance with a reduced Esrrg expression, B6.Sle1c2 CD4⁺ T cells present reduced mitochondrial mass and altered mitochondrial functions, as well as altered metabolic pathway utilization when compared to B6. Taken together, we propose Esrrg as a novel lupus susceptibility gene regulating CD4⁺ T cell function through their mitochondrial metabolism.

Introduction

The murine NZM2410 strain spontaneously develops an autoimmune disease that mimics systemic lupus erythematosus (SLE), including the presence of anti-nuclear autoAb (ANA), immune activation, and immune-complex induced glomerulonephritis (GN). Derived from the classic (NZB × NZW)F1 (NZB/W F1) lupus model, it has an advantage over its parental strains in that it is homozygous, making it an ideal model to identify novel genetic determinants of lupus (1). Linkage analysis of NZM2410 to GN identified the major lupus susceptibility locus, Sle1, as an NZW-derived interval on telomeric chromosome 1 (2). This region overlaps with syntenic human SLE QTLs, 1q22–23 and 1q41–42, suggesting that similar genetic factors may mediate pathogenesis in both species (3). Subsequent studies using congenic mice demonstrated distinct functional requirements that Sle1 imparted in the induction of murine lupus. Specifically, B6.Sle1 mice display B and T cell intrinsic loss of tolerance to chromatin (4–6). Furthermore, complementation analyses with the other NZM2410-derived SLE susceptibility loci (7) and with the NZW genome (8) demonstrated that Sle1 expression was necessary for disease to develop in this model. Still, the identification of the underlying genetic determinants of SLE pathogenesis in this 62 Mb region, which contains an estimated 350 genes, remained a daunting task.

Three Sle1 subloci, Sle1a, Sle1b, and Sle1c, contribute to the production anti-chromatin autoAbs, revealing the complexity of this locus (9). Further phenotypic characterization of these subloci revealed increased activation of CD4⁺ T cells by Sle1a and Sle1c and defective B cells tolerance by Sle1b (9–11). Using congenic recombinants, Sle1c was determined to correspond to at least two subloci, Sle1c1 and Sle1c2 (12). Complement receptor 2 (Cr2) was identified as a candidate gene for Sle1c (3), and subsequently found to co-segregate with the telomeric Sle1c1 (12). Ensuing human association studies validated these findings by identifying a haplotype that alters CR2 splicing that was associated with SLE (13, 14). Additionally, Sle1b has been attributed to polymorphisms in the SLAM gene cluster, with direct evidence for one SLAM family member Slamf6 (11, 15, 16). More recently, evidence has shown that expression of Slamf1, Slamf2 and Slamf4, other members of the SLAM family located within Sle1b, are also involved in controlling ANA production (17, 18). Finally, Sle1a1 corresponds to a novel splice isoform of Pbx1, which is found more frequently in SLE patients than in healthy controls (19). Thus, a variety of novel lupus susceptibility genes have been identified so far in the Sle1 locus affecting both B and T cell functions.

We have previously reported that Sle1c2 is associated with increased activation and proliferation of CD4⁺ T cells (12). In the current study, we mapped Sle1c2 to estrogen-related receptor gamma (Esrrg), which encodes for the orphan nuclear receptor, ERRγ. The expression of this gene, which regulates oxidative metabolism (20), has not been previously been reported in T cells. We showed that the NZW allele is associated with a reduced Esrrg expression in CD4⁺ T cells, which strongly correlates with increased cell activation, and the expansion of IFNγ secreting T cells. In addition, B6.Sle1c2 CD4⁺ T cells showed a reduced mitochondrial mass and hyperpolarization consistent with their reduced Esrrg expression. Finally, we demonstrated that Sle1c2 contributes to lupus phenotypes in two disease models. These results suggest that Esrrg is a novel lupus susceptibility gene that regulates CD4⁺ T cell function and activation through their mitochondrial metabolism.

Materials and Methods

Mice

B6.Sle1c mice that contain a NZW-derived interval at the telomeric end of chromosome 1 have been described previously (9). The loci previously referred to as Sle1c.Cr2^w-1 on the telomeric end and Sle1c.Cr2^b-1 on the centromeric end (12) have been renamed Sle1c1 and Sle1c2 respectively to be more consistent with the terminology of the other loci. To generate additional recombinant subcongenic strains, (B6 × B6.Sle1c)F1 × B6 backcross progeny were genotyped for recombinations in the Sle1c interval with microsatellites that are polymorphic between NZW and B6. Recombinants were bred to B6 and the progeny of this expansion backcross were then bred to homozygosity. To fine-map the ends of the recombinant congenic intervals, single nucleotide polymorphisms (SNPs) that are polymorphic for B6 and NZW were selected from the Mouse Phenome Database (http://phenome.jax.org/SNP), and alleles were determined by sequencing. C57BL/6 (B6), B6.Cg-Tg(TcraTcrb)425Cbn/J (B6.OTII), B6(C)-H2-Ab1^bm12/KhEgJ (B6.bm12), B6.Cg-Igh^aThy1^aGpi1^a/J (B6.Thy1^a), B6.129P2-Tcrb^tm1MomTcrd^tm1Mom/J (B6.Tcrbd^−/−), and NZB mice were purchased from the Jackson Laboratory. B6.FoxP3-eGFP mice (21) were a kind gift from Dr. Vijay Kuchroo (Harvard Medical School). All mice were bred and maintained at the University of Florida in specific pathogen-free conditions and experiments were performed using age and sex-matched cohorts at the ages indicated in the text. All experiments were conducted according to protocols approved by the University of Florida Institutional Animal Care and Use Committee.

Cell isolation and culture

Single cell suspensions were prepared from spleen and red blood cells were lysed. Cells were then washed in ice cold 5% FCS in PBS and passed through 30μm nylon mesh to remove debris. Splenocyte suspensions were enriched for CD4⁺ T cells by negative selection with magnetic beads (Miltenyi). RPMI supplemented with 10% FCS, HEPES, 2-Mercaptoethanol, and penicillin-streptomycin was used as culture medium. Cells were stimulated either with plate-bound 5μl/ml anti-CD3e (145-2C11) and 2.5μl/ml anti-CD28 (37.51) (BD Biosciences), or with 0.5μg/ml PMA and 1μM ionomycin (Sigma). For Ag-specific proliferation assays, irradiated (2000 Rad) B6 splenocytes were pulsed for 2 h at 37°C with either OVA_323–339 or histone H4_76–90 at the indicated concentrations. The Ag-loaded splenocytes were then co-cultured 1:1 with CD4⁺ T cells from B6.OTII or B6.Sle1c2.OTII mice in triplicate. ³H-thymidine was added at 1 μCi/ 200μl for the last 18 h of 72 h cultures to measure proliferation. Cells were then harvested onto glass filter paper to measure thymidine incorporation. For T_H17 polarization, CD4⁺ T cells were cultured with anti-CD3e and anti-CD28, 2.5 ng/ml TGF-β (Peprotech), and 25 ng/ml IL-6 (Peprotech) for 48 h. For CD4⁺ FoxP3⁺ regulatory T cell (Treg) induction, FACS-sorted CD4⁺GFP⁻ T cells from B6.FoxP3-eGFP or B6.Sle1c2.FoxP3-eGFP mice were cultured with anti-CD3e and anti-CD28, and 2.5 ng/ml TGF-β with or without 10 nM all-trans-retinoic acid (atRA, Sigma).

Flow cytometry

Cell suspensions were blocked with 10% normal rabbit serum and anti-CD16/32 (2.4G2) in staining buffer (5% FCS, 0.05% sodium azide in PBS) and incubated on ice for 30 min. Biotinylated or fluorophore-conjugated Abs specific for CD3 Molecular Complex (17A2), CD4 (RM4-5), CD8a (53–6.7), CD44 (IM7), CD62L (MEL-14), CD25 (7D4), CD69 (H1.2F3), CD90.1 (OX-7), CD90.2 (53–2.1), B220 (RA3-6B2), CD19 (1D3), IgM (II/41), CD80 (16-10A1), CD86 (GL1), I-Ab (AF6-120.1), and isotype controls were used in predetermined amounts. SA-PerCP-Cy5.5 (BD Biosciences) was used to detect biotinylated Abs. IFNγ (XMG1.2), IL-4 (11B11), IL-17A (TC11-18H10), and FoxP3 (FJK-16s) were detected using Fixation/Permeabilization kits (eBiosciences) according to manufacturer's protocol. When cytokine profiles were analyzed, cells were treated with leukocyte activation cocktail (BD Biosciences) prior to staining. All Abs were from BD Biosciences except for anti-FoxP3, which was from eBiosciences. Analysis was performed on a FACSCalibur cytometer (BD Biosciences) with at least 50,000 events per sample collected, and lymphocyte populations were gated based on forward and side scatter characteristics.

For metabolic labeling, splenocytes derived from 2–3 month old B6, B6.Thy1^a, and B6.Sle1c2 mice were stained in RPMI 1640 medium at a density of 1 × 10⁶ cells/ ml with cell-permeable metabolic dyes at 37°C for 30–120 min, followed by surface staining with PE Cy7 conjugated CD3 (17A2), PerCP conjugated CD4 (GK1.5), APC Cy7 conjugated CD8a (53–6.7), PE conjugated CD11b (M1/70), APC conjugated CD11c (N418), and Alexa Fluor 700 conjugated CD19 (6D5) Abs for 30 min at 4°C. All Abs for this experiment were obtained from Biolegend. Metabolic indicators were used for measurement of NO, mitochondrial transmembrane potential, mitochondrial mass, Ca²⁺ stores, superoxide production, apoptosis, and necrosis (27,28). Diaminorhodamine-4M (DAR-4M) was used to evaluate peroxynitrite production, a byproduct of increased NO production. Mitochondrial transmembrane potential was measured using potentiometric indicator tetramethylrhodamine methyl ester (TMRM). Mitochondrial mass was measured by fluorescence of MitoTracker Green-FM (MTG). Calcium stores were measured by fluorescence of Fluo-3 AM (1 μM). Hydroethidine (HE), dihydrorhodamine 123 (DHR) and 2′,7′-dichlorofluorescin (DCF) were used as an indicator for superoxide production. Apoptosis and necrosis were evaluated by co-staining with FITC conjugated Annexin V and propidium iodide (PI). TMRM, MTG, HE, Fluo-3M and PI were obtained from Invitrogen-Molecular Probes, DAR-4M from Calbiochem, and FITC conjugated Annexin V was from BD Biosciences. The metabolic profiling was evaluated on a LSR-II flow cytometer (BD Biosciences), with 50,000 events per sample collected. Data was analyzed using Flow Jo cytometry analysis software (TreeStar, Ashland, OR). MFI values of B6.Sle1c2 mice were normalized to the means obtained for B6 mice studied in parallel and set at 1.0.

Gene expression analyses

Total RNA was isolated using RNeasy mini kits, Qiashredders, and RNase-free DNase sets (Qiagen). cDNA was then synthesized using the ImProm-II Reverse Transcription System (Promega). Primers for ERRγ target genes (Supplemental Table 1) were designed with the Primer3 software (http://frodo.wi.mit.edu/primer3/) and used in Sybr® Green (Applied Biosystems) based qRT-PCR. Taqman Gene Expression Assays (Applied Biosystems) were used to measure Esrrg (Mm00516269_mH, spans exons 3 and 4 of transcript 1) and Gapdh control expression. Relative quantity of gene expression was calculated using the comparative C_T method (RQ=2^−ΔΔCt) normalized to the average ΔC_T of the B6 samples. Global gene expression was compared between negatively-bead selected CD4⁺ T cells from 6 month old B6 and B6.Sle1c2 mice (n=5 per strain). Their cDNA was synthesized, fragmented, and biotin-labeled using the Ovation Biotin RNA Amplification and Labeling System (NuGEN Technologies), then hybridized to Affymetrix Mouse Genome 430 2.0 arrays. Data analysis was based on the use of “internal standards” and generalization of the “Error Model” (22) as presented elsewhere (23).

ELISA

Anti-dsDNA and anti-chromatin IgG were measured by ELISAs as previously described (4). Sera were tested in duplicate in a 1:100 dilution. Relative units were standardized using serial dilutions of a positive serum from a B6.Sle1.Sle2.Sle3 (BcN/LmoJ) mouse, arbitrarily setting the 1:100 dilution reactivity to 100 units.

Western blots

Cell lysates prepared from splenic CD4⁺ T cells purified by negative selection were probed with rabbit polyclonal ERRγ Ab (Z-21, Santa Cruz Biotechnology) and revealed with horseradish peroxidase-conjugated anti-rabbit IgG (Sigma Aldrich). Horseradish peroxidase-conjugated anti-GAPDH Ab (Sigma Aldrich) was used as control. The membranes were developed by enhanced Supersignal West Pico Chemiluminescent substrate (Thermo Scientific, Rockford, IL). The bands were quantified using Alphaview software (Alpha Innotech).

For VDAC (voltage-dependent anion channel) expression, protein lysates were prepared from splenocytes and negatively-isolated T cells (Dynal mouse T cell isolation kit, Invitrogen) through lysis in radio-immunoprecipitation assay buffer (150 mM NaCl, 2% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris pH 8.0, 1 mM PMSF, 1 μg/ml aproptinin, 1 μg/ml pepstatin, 1 g/ml leupeptin, 1 mM NaF, 1 mM sodium orthovanadate, 0.1 mM sodium molybdate, 10 mM sodium pyrophosphate) at a density of 4 × 10⁷ cells/ml on ice, followed by addition of equal volumes of Laemmli buffer and were heated to 95°C for 5 min prior to separation on SDS-PAGE gels and transfer to 0.45 μm nitrocellulose membranes. VDAC / porin was detected with rabbit polyclonal Ab (Abcam #ab34726), and β-actin was detected with mouse monoclonal Ab (Millipore #MAB1501R). Western blots were imaged using a Kodak Image Station 440CF and quantified by automated densitometry using Kodak1D software (Eastman Kodak, Rochester NY).

Mixed bone marrow chimera

Chimeras were prepared as previously described (5). Briefly, 4–5 month old B6.Tcrbd^−/− recipients were lethally irradiated with two doses of 525 Rad 4–6 h apart the day before reconstitution. Bone marrow (BM) cell suspensions from B6.Thy1^a and B6.Sle1c2 were mixed 1:1 after depleting T cells using anti-CD5 magnetic beads (Miltenyi). Recipients received 10⁷ BM cells from sex-matched donors by i.v. injections and grafts were allowed to reconstitute for 8 weeks.

Chronic Graft versus host disease (cGVHD)

cGVHD was induced as previously described (24). Briefly, B6 and B6.Sle1c2 hosts received 8 × 10⁷ B6.bm12 splenocytes via i.p. injections. Sera were collected weekly after induction and screened for anti-dsDNA and anti-chromatin IgG. Hosts were sacrificed 3 weeks after transfer, kidneys were prepared for histology, and splenocytes were analyzed by flow cytometry. The presence of immune complexes in the kidneys was evaluated on frozen tissue sections stained with FITC-conjugated anti-C3 (Cappel) and anti-IgG H+L chains (Jackson Immunoresearch). Staining intensity was evaluated in a blind manner on a semi-quantitative 0–3 scale and averaged on at least 10 glomeruli per section. Two separate sets of cGVHD induction were performed with at least five mice per strain, yielding similar results.

Experimental autoimmune encephalomyelitis (EAE)

EAE was induced in 4-month-old male B6 or B6.Sle1c2 mice. On d 0, mice received an emulsion of 50μg myelin oligodendrocyte glycoprotein (MOG) peptide sequence 35–55 (MOG_35–55) and 500 μg of Mycobacterium tuberculosis (Difco) in incomplete Freund's adjuvant (Sigma) via subcutaneous injections at the base of the tail. Additionally, 500 ng of pertussis toxin (List Biologicals) was administered i.p. on d 0 and 2. Daily clinical scores were assessed by the following criteria: 0, no disease, 1, flaccid tail, 2, hind limb paraparesis, 3, hind limb paralysis, 4, quadriplegia. Mice were euthanized at a score of 4, or after a 40 d after induction.

Statistical analysis

Statistical analyses were performed using the GraphPad Prism 5 software. Unless indicated, graphs show mean and SEM for each group. For comparisons between two groups, two-tailed Mann-Whitney were used when n ≥ 5 and Student's t-tests were used either when n < 5, or when data sets passed D'Agostino and Pearson omnibus normality tests as indicated. ANOVA with Dunn's post-test and 2-way ANOVA with Bonferroni post-test corrections were used for multiple comparisons as appropriate. Each in vitro experiment was performed at least twice.

Results

Increased CD4⁺ T cell activation maps to the centromeric end of Sle1c

In order to narrow down the number of candidate genes responsible for the increased CD4⁺ T cell activation displayed by Sle1c congenic mice, six subcongenic strains were generated in which recombinant screening was targeted to the centromeric end of the interval (Fig. 1A). Phenotypic screening showed that two strains, REC2b and REC5, displayed increased spleen weight and CD4:CD8 T cell ratio in aged mice as compared to B6 (Fig. 2A). Additionally, these two strains displayed increased CD4⁺ T cell activation with a significantly higher percentage of CD69⁺ T cells and of CD44^hi CD62L⁻ T effector memory cells (Tem). Since the strain with the shortest interval necessary for increased activation is the REC5 subcongenic, and REC1, 2, 3, and 8 were B6-like, Sle1c2 was then defined as the 675Kb region between SNPs rs30920616 and rs32528185 (Fig. 1A and B). Except where noted, the REC5 strain was used as B6.Sle1c2 for the remainder of this study.

FIGURE 1.

Physical map of Sle1c. A, The Sle1c and its recombinant intervals (REC1 to REC8) are shown on the telomeric end of chromosome 1 with NZW-derived regions in white and B6-derived regions in black. All known polymorphic MIT microsatellite markers, as well as SNPs that define recombination intervals are depicted. B, Sle1c2 is defined by the non-overlapping centrometric end of Sle1c that carries the NZW alleles in REC5 and REC2b (Sle1c2^w), and the B6 alleles in the other strains (Sle1c2^b). The SNPs defining the areas of recombination on both ends are shown, along with the exonintron structure of the two known protein coding genes located in Sle1c2. Scale is in Mb and all positions are current with Ensemble release 67 (www.ensembl.org/Mus_musculus/), which is based on NCBI m37.

FIGURE 2.

Phenotypic mapping of Sle1c2. A, Comparison of spleen weights, CD4:CD8 ratios, CD69 expression and Tem percentages of CD4⁺ T cells from spleens of 10–14 month old mice. Significance levels indicate ANOVA with Dunn's multiple comparison analysis to B6. Ag-specific (B), polyclonal (C), and PMA-induced (D) proliferation of CD4⁺ T cells from 3–5 month old mice was measured by ³H thymidine incorporation, and stimulation indexes were calculated relative to cultures with medium only. Both REC2b.OTII and REC5.OTII strains were used as Sle1c2.OTII. Two-way ANOVA (‡‡) was used to measure strain effect on Ag-specific proliferation with Bonferroni's post-test indicating significance at each concentration in B. Mann-Whitney was used in C and t test was used in D. Absolute numbers of splenic CD4⁺ T cells (E) and of B and CD8⁺ T cells (F) were compared between B6 and B6.Sle1c2 at 5–8 and 10–14 month old cohorts. Naïve T cells were defined as CD4⁺ CD44^lo CD62L⁺ and Tem cells were defined as CD4⁺ CD44^hi CD62L⁻. Significance levels indicate Mann-Whitney comparisons to B6 mice of the same age group. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, ‡‡ P ≤ 0.001.

As opposed to the splenomegaly phenotype, which is age-dependent, CD4⁺ increased activation was detectable as early as 2–3 months of age (Supplemental Fig. 1A), suggesting that it is a primary consequence of Sle1c2 expression. Importantly, as previously shown for the entire Sle1c interval (12), a mixed BM chimera experiment showed that increased CD4⁺ T cell activation was intrinsic to Sle1c2-expressing T cells (Supplemental Fig. 1B). Since we have previously reported an increased proliferation of B6.Sle1c CD4⁺ T cells (12), we screened the congenic recombinants to discern whether this phenotype also mapped to the refined Sle1c2 locus. To assess whether the increased proliferation was antigen-specific, B6.Sle1c2.OTII double-congenic mice were generated for comparison to B6.OTII mice, which carry a transgenic TCR that is specific for the OVA323-339 ovalbumin peptide. Increased proliferation was observed with Ag-specific OVA (Fig. 2B), polyclonal TCR (Fig. 2C) and PMA/ionomycin (Fig. 2D) stimulation of B6.Sle1c2.OTII as compared to B6.OTII T cells. This increased activation and proliferation resulted in an expansion of the CD4⁺ T cell compartment in older mice, specifically Tem to the expense of naïve CD4⁺ T cells (Fig. 2E). Age-dependent expansion of the B and CD8⁺ T cell compartments were also observed in B6.Sle1c2 mice (Fig 2F). However, surface marker analysis found no evidence of increased activation in these cells (not shown). Taken together, these results mapped the CD4⁺ T cell activation phenotype observed in Sle1c mice to the 675 Kb Sle1c2 interval. This drastically reduced the set of candidate genes from 48 to 2, namely Esrrg, encoding Estrogen-related receptor gamma, and Ush2a, encoding for Usher syndrome 2A homolog (Fig. 1B).

B6.Sle1c2 CD4⁺ T cells exhibit a marked TH1 skewing

A gene expression and pathway analysis of CD4⁺ splenocytes revealed that a large number of genes related to IFNγ expression were upregulated in B6.Sle1c2 mice, suggesting a TH1 skewing (Supplemental Fig. 2A, 2B). Intracellular staining confirmed that a larger percentage of Sle1c2 CD4⁺ splenocytes expressed IFNγ as compared to B6 (Fig. 3A). Importantly, CD4⁺ T cells from the parental B6.Sle1c strain showed a similar percentage of IFNγ⁺ as the B6.Sle1c2 subcongenic strain. No difference was observed for IL-4 production (data not shown).

FIGURE 3.

Sle1c2 upregulates IFNγ expression. A, Representative histograms and quantification of ex vivo IFNγ intracellular staining in CD4⁺ T cells were from the spleens of 2–3 month old mice. Significance levels indicate ANOVA with Dunn's multiple comparison analysis to B6. B, Representative plots and quantitation of FoxP3⁺ IFNγ⁺ double positive CD4⁺ T cells after 3 d of culture with plate-bound anti-CD3/CD28 and TGFβ. Cells were collected from 1–3 and 8–10 month old mice. C, D, Representative plots (C) and quantification (D) of intracellular staining for IFNγ, FoxP3, and IL-17 in CD4⁺ T cells from 8–10 month old mice after 3 d of TH0 (anti-CD3 and anti-CD28 only), Treg (stimulation plus TGFβ), and TH17 (stimulation plus TGFβ and IL-6) polarizing conditions. In B and D, t tests compare B6 to Sle1c2 mice for each age or treatment group. : * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

The microarray analysis also revealed that several genes involved in TH17/Treg homeostasis were upregulated in B6.Sle1c2 CD4⁺ T cells (Supplemental Fig. 2C). These included FoxP3, Il2ra, and IL10 for Treg cells, and Irf4, Rora, Il17a, IL21, and Il22, for TH17 cells. Additionally, the Ahr and Hif1a pathways, which have been implicated in TH17/Treg homeostasis (25–27) were also upregulated (Supplemental Fig. 2B–C). Given this altered gene expression profile, we examined whether Sle1c2 expression augmented Treg and TH17 differentiation. Using FoxP3-eGFP reporter mice, we showed that TGFβ with or without atRA equally induced FoxP3 expression in B6 and B6.Sle1c2 CD4⁺ GFP⁻ T cells (Supplemental Fig. 1C). Moreover, contrary to our previous reports that described a decrease in the percentage of Treg defined as CD4⁺ CD62L⁺ CD25⁺ (10, 12), the size of the CD4⁺ FoxP3⁺ Treg compartment was similar between B6 and B6.Sle1c2 mice, with an expansion of FoxP3⁺ Treg in older B6.Sle1c2 mice (Supplemental Fig. 1D). Interestingly, in vitro Treg polarization produced significantly more FoxP3⁺ IFNγ⁺ cells in B6.Sle1c2 mice at 8–10 month old, but not 1–3 month old (Fig. 3B and D). This not only confirms the strong TH1 skewing associated with Sle1c2, but also suggests that B6.Sle1c2 mice may accumulate Tregs corresponding to a plastic population of IFNγ⁺ adaptive Tregs that have been linked to autoimmunity (28).

TH17 in vitro polarization did not reveal differences in IL-17 production between the two strains (Fig. 3C and D). However, as with Treg polarization, IFNγ production was increased in the B6.Sle1c2 TH17 polarized CD4⁺ T cells (Fig. 3C and D), highlighting again the TH1 skewing induced by Sle1c2 expression. Finally, EAE was used as a model to test for differences in TH17/Treg homeostasis in vivo, and no differences were observed in day of onset or severity between the two strains (Supplemental Fig. 1E). Hence, while the global gene expression analysis predicted increased commitment to TH1, Treg, and TH17 lineages, cellular assays only revealed a strong TH1 skewing.

Sle1c2 exacerbates lupus in the induced cGVHD model and via epistatic interactions with the NZB genome

To determine whether the CD4⁺ T cell phenotypes that segregate with Sle1c2 are relevant to lupus pathogenesis, we used the cGVHD-induced lupus model (29). Preliminary results mapped an increased cGVHD susceptibility to the centromeric portion of Sle1c that includes Sle1c2 (12). Here we showed that CD4⁺ T cells were significantly more activated in B6.Sle1c2 recipients of B6.bm12 splenocytes as compared to B6 recipients (Fig. 4A). In addition, B6.Sle1c2 CD11c⁺ dendritic cells and B cells displayed altered activation (Fig. 4B and C) that was concurrent with increased serum autoAb production (Fig. 4D). Finally, the kidneys of cGVHD-induced B6.Sle1c2 mice showed an increased deposition of IgG2a immune complexes as compared to B6 (Fig. 4E). These results demonstrate that the activated phenotype of Sle1c2 CD4⁺ T cells contributes to induced humoral autoimmunity.

FIGURE 4.

Sle1c2 enhances induced lupus phenotypes in the cGVHD model. cGVHD was induced in B6 (filled circles) and B6.Sle1c2 (open circles) and splenic compartments were compared by flow cytometry 3 weeks later. Age-matched non-induced mice are also shown as control. A, CD4⁺ T cell activation was assessed by high forward scatter (indicative of blasting cells), CD69 expression and by expansion of T_em cells. B, Dendritic cell activation was indicated by B7–2 (CD86) expression. C, B cell activation was measured by CD22, CD69, B7–1 (CD80) and B7–2 expression. D, Glomerular IgG2a and C3 deposition was detected in frozen kidney sections. Sections were separately scored for each Ab and additive values (IgG2a score + C3 score) are shown. E, Weekly serum anti-chromatin and anti-dsDNA IgG were measured by ELISA. A, B, C, and E were compared by Mann-Whitney tests and D was compared by 2-way ANOVA with Bonferroni's post-test. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

We have previously shown that Sle1c expression increases lupus phenotypes presented by (NZB × B6)F1 hybrids (30). Using the same model, we compared the phenotypes of (NZB × B6)F1 and (NZB × B6.Sle1c2)F1 mice, in which the NZB hemi-genome interacts with either the B6 or the NZW Sle1c2 alleles, respectively. Importantly, the NZB and NZW share the same haplotype at the chromosome 1: 189.80–190.050 Mb interval that includes the Sle1c2 portion of Esrrg, as determined for all 228 published SNPs that are polymorphic between B6 and NZW, with an NZB allele known or imputed with confidence ≥ 0.9 (data not shown). Therefore, the Sle1c2 locus is predicted to be homozygous for the NZB/NZW allele in (NZB × B6.Sle1c2)F1 mice, and heterozygous in (NZB × B6)F1 mice. This is relevant because the Sle1c2 CD4⁺ T cell phenotype is recessive (data not shown). Our findings showed that in 12 month old NZB hybrids, splenomegaly, CD4:CD8 T cell ratio, percentage of Tem, and B7-2 expression on B cells were significantly increased in mice with the homozygous NZW Sle1c2 allele (Fig. 5A). Moreover, the expression of Sle1c2 significantly enhanced the production of anti-dsDNA IgG triggered by the NZB hemi-genome as the mice aged (Fig. 5B). Finally, IgG immune complex deposition was greatly exacerbated by Sle1c2 (Fig. 5C–D), but did not result in clinical disease. Taken together, this data establishes that Sle1c2 can contribute to lupus through interactions with the NZB genome.

FIGURE 5.

Sle1c2 exacerbates spontaneous lupus phenotypes induced by the NZB genome. A, Spleen weight, CD4:CD8 ratios, percentage of CD62L⁻ CD44^hi CD4⁺ and B7–2⁺ B220⁺ splenocytes in 12 month old (NZB × B6)F1 and (NZB × B6.Sle1c2)F1 mice (n=14 and 25, respectively). B, Serum anti-dsDNA IgG is shown as unit values at 5 and 12 months of age, and as relative individual increase between these two ages for each strain. C and D, Glomerular IgG and C3 deposition was detected in frozen kidney sections. Sections were separately scored for each Ab and additive values (IgG score + C3 score) are shown in C. Representative sections (X 100) are shown in D. Mann-Whitney tests: * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Esrrg expression is downregulated in B6.Sle1c2 CD4⁺ T cells

The 675Kb Sle1c2 interval contains 2 protein-coding genes, Esrrg and Ush2a (Fig. 1B), with the latter having no detectable expression in CD4⁺ cells (data not shown). Esrrg message expression was detected in CD4⁺ splenocytes, and at a lower level in CD4⁻ splenocytes and in thymocytes (Fig. 6A). Furthermore, Esrrg expression was significantly lower in the B6.Sle1c2 CD4⁺ as compared to B6 CD4⁺ fraction (40.6 ± 11.4% less). No difference between strains was observed in CD4⁻ splenocytes. Esrrg expression in CD4⁺ T cells is low compared to liver, heart, brain and kidney, highly metabolic tissues in which Esrrg expression has been reported (31–33), going from about 4 times lower than in liver to about 50 times lower than in kidney (Supplemental Fig. 3A). In our microarray analysis, five probes were located within the Esrrg locus (with only one actually probed the ERRγ coding sequence). However, the signal for all 5 probes was below the detectable threshold. CD4⁺ T cell ERRγ protein expression was detected in CD4⁺ T cells (Fig. 6B). We were not able, however, to show a difference between B6.Sle1c2 and B6. Esrrg overexpression in cell lines showed that messages and protein levels were not tightly correlated, suggesting an important post-transcriptional regulation (data not shown). Regardless of age or strain, Esrrg message expression in splenic CD4⁺ T cells showed a strong negative correlation with both CD69⁺ and Tem percentages (Fig. 6C), suggesting that Esrrg negatively regulates CD4⁺ T cell activation.

FIGURE 6.

Esrrg expression is decreased in Sle1c2 CD4⁺ T cells. A, qPCR comparing Esrrg expression in thymocytes, splenocytes, CD4⁺ T cells, and CD4⁻ fractions between B6 and B6.Sle1c2 (8–10 month old, n=7 per strain). t tests were used to compare B6.Sle1c2 Esrrg expression to B6 for each cell population. B, ERRγ expression was analyzed in B6 and B6.Sle1c2 CD4⁺ T cells by semi-quantitative Western blotting, with cell lysates undiluted, diluted 1:2 (0.5) and 1:4 (0.25). The ERRγ protein abundance was normalized to the expression of GAPDH. The numbers in Y axis indicate the ratio of normalized ERRγ protein abundance in B6.Sle1c2 CD4⁺ cells to that in B6 CD4⁺ cells. C, Correlation between Esrrg expression and CD69⁺ and Tem percentages of splenic CD4⁺ T cells. The young and old cohorts were 2–3 and 7–8 months old, respectively. The significance that slopes do not equal zero (P) and correlation coefficients for linear regressions (R²) are shown. D, qPCR analysis of Hif1a, Slc16a3, and Pkm2 expression in B6 and B6.Sle1c2 CD4⁺ splenocytes, compared with Mann-Whitney tests. E, Percent change in expression of genes in metabolic pathways from microarray analysis comparing B6.Sle1c2 to B6 CD4⁺ T cells. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Esrrg expression was compared between B6.Sle1c2 and B6 adult brain, heart, kidney, and liver. Esrrg expression was significantly decreased in B6.Sle1c2 brain, and a trend for decreased expression was observed in kidney and liver (Supplemental Fig. 3B). Interestingly, an opposite trend was observed in heart. Esrrg-deficient mice die postnatally due to a defective switch from glycolytic to oxidative metabolism (34). Consistent with this result and a reduced Essrg expression in B6.Sle2c2 mice, we have observed a high mortality in B6.Sle1c2 neonates (69% total litter loss, corresponding to 34 litters lost out of 49 born) as compared to B6 (18% litter loss corresponding to 13 out 71 born during the same period of time and in the same room). It is possible that in response to reduced Esrrg expression, a compensatory mechanism may increase Esrrg expression in the myocardium of a percentage of B6.Sle1c2 neonates, resulting in their survival.

To determine if decreased Esrrg expression in CD4⁺ T cells had transcriptional consequences on its putative targets, we selected 12 genes (Supplemental Table I) that were differentially expressed by B6.Sle1c2 CD4⁺ T cells in the microarray analysis and that were also known to have ERRγ bound to their promoters by chip-on-chip analyses (34, 35). The expression of five of these genes, transcriptional regulators c-myc binding protein (Mycbp) and retinoic acid receptor alpha (Rara), a subunit of electron transport complex I (Ndusf1), a mitochondrial protein modifier (Ppif), and a mitochondrial oxidoreductase (Rtn4ip1), was significantly decreased in Sle1c2 CD4⁺ T cells (Supplemental Fig. 3C). This data suggest that the decreased Esrrg expression induced by Sle1c2 affects Esrrg target gene expression.

Since Esrrg is known to regulate glycolytic and oxidative metabolic programs (34), we analyzed the expression of several genes involved in these pathways by quantitative PCR. Hif1α and Slc16a3, encoding for monocarboxylic acid transporter 4 (MCT4), which are both mediators of glycolysis (27), were expressed at a higher level in B6.Sle1c2 CD4⁺ T cells (Fig. 6D). In addition to glycolysis, a microarray pathway analysis revealed an influence of Sle1c2 in several other metabolic pathways, including glutaminolysis, glycogenesis, and fatty acid oxidation (Fig. 6E). Furthermore, genes regulating electron transport chain were also differentially regulated, indicating altered mitochondrial function. Finally, out of the 7 metabolic genes that were analyzed in both the Esrrg^−/− heart (34) and the Sle1c2 CD4⁺ T cells in this study, five of them (Eno2, Gsk3, Cox6a2, Ndufa10 and Ndufs1) showed an expression change in the same direction. Overall, these data provide evidence that Sle1c2 results in an altered metabolic profile in CD4⁺ T cells, due to decreased Esrrg expression.

B6.Sle1c2 CD4⁺ T cells exhibit mitochondrial dysfunction

Esrrg expression has been reported to directly correlate with mitochondrial biogenesis and functions (33). Accordingly, a decreased Esrrg expression is predicted to reduce mitochondrial mass and functions in Sle1c2 CD4⁺ T cells. This was indeed verified in the spleen of 2–3 month old mice, in which the difference in T cell activation is still minimal, and therefore not likely to induce secondary metabolic changes. The mitochondrial membrane potential and mass were significantly decreased in Sle1c2 CD4⁺ T cells as compared to B6 (Fig. 7A and B), and this was associated with decreased Ca²⁺ and NO levels (Fig. 7D and E). In turn, the production of reactive oxygen intermediates (ROI), as measured by HE fluorescence, was increased in Sle1c2 CD4⁺ T cells. However, no difference in ROI was found between the two strains when measured with the DCF or DHR dyes (data not shown). These findings are suggestive of mitochondrial dysfunction, characterized by an inability to maintain transmembrane potential while leaking ROI. Consistent with impaired mitochondrial functions, a significant increase of necrosis was observed in Sle1c2 splenocytes (Fig. 7F). This corresponds to a predisposition to pro-inflammatory death by necrosis that has been previously reported in lupus T cells (36).

FIGURE 7.

Mitochondrial function is impaired in the CD4⁺ T cells of B6.Sle1c2 mice. A–E, Splenic CD3⁺ CD4⁺ T cells were assessed by flow cytometry for mitochondrial mass (A), mitochondrial membrane potential (B), oxidation (C), Ca (D) and NO (E) contents. The graphs show mean + SEM mean fluorescence intensity (MFI) normalized to B6 mean values. F, viability analysis of ex-vivo splenocytes based on Annexin V (Ann V) and PI staining. Live cells were Ann V⁻ PI⁻. G, VDAC expression in CD4⁺ T cells analyzed by Western blot. β-actin expression is shown as a control. The graph on the right shows the densitometry analysis of the VDAC/ β-actin ratios normalized to one B6 sample. All data were collected from the same 2–3 months old mice. B6 and B6.Thy1^a (B6 ^a) values were pooled in the graphs under B6 and compared to B6.Sle1c2 values with Mann-Whitney tests. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Interestingly, VDAC levels were significantly increased in Sle1c2 CD4⁺ T cells (Fig. 7G), which may correspond to a compensatory mechanism to maintain mitochondrial membrane potential in spite of reduced Esrrg expression. A reduced mitochondrial mass was not observed in other Sle1c2 immune cell populations, except CD11c⁺ cells (Table 1). This may be related to the fact that Esrrg expression is very low in CD4⁻ splenocytes (Fig. 6A), therefore not affecting mitochondria in these cell types. Other parameters showed significant differences between the two strains in non-CD4⁺ cell types, such as NO levels which were significantly reduced in CD8⁺ T cells, B cells, CD11b⁺ and CD11c⁺ splenocytes (Table 1), indicating that the metabolic consequences of Esrrg deficiency may not be confined to T cells.

Table 1.

Comparison of mitochondrial function in the major splenic cell subsets between B6.Sle1c2 and B6 mice. MFI values were normalized to B6 means set at 1.0.

Cell type	MTG		TRMR		HE		DAR-4M		Fluo-3AM
	B6	Sle1c2	B6	Sle1c2	B6	Sle1c2	B6	Sle1c2	B6	Sle1c2
CD4+ T	1.04 + 0.09	0.65 + 0.80	0.96 + 0.18
				0.31 + 0.03	0.98 + 0.08	1.16 + 0.07	1.03 + 0.07	0.65 + 0.03	0.96 + 0.12	0.62 + 0.06
CD8+ T	0.97 + 0.09	0.68+ 0.08	0.98 + 0.34
				0.32 + 0.03	1.02 + 0.06	1.35 + 0.14	1.04+ 0.08	0.42 + 0.08	0.93 + 0.24	0.45 + 0.04
CD19+	1.04 + 0.10	0.98 + 0.09	0.98 + 0.11
				0.48 + 0.03	1.00 + 0.03	1.19 + 0.06	1.00 + 0.04	0.69 + 0.03	1.00 + 0.04	0.98 + 0.06
CD11b+	0.98 + 0.20	0.97 + 0.13	1.43 + 0.92
				0.35 + 0.03	0.98 + 0.05	1.18+ .02	0.99 + 0.04	0.54+ 0.06	1.00 + 0.28	0.79 + 0.04
CD11c+	0.98 + 0.09	0.70 + 0.08	1.05 + 0.40
				0.69 + 0.05	0.97 + 0.11	1.66 + 0.17	0.99 + 0.02	0.45 + 0.03	0.95+ 0.19	0.45 + 0.04

The values indicate mean ± SEM for 5 mice per strain.

Values were compared between the strains for each parameter.

Bold indicates P < 0.05,

underline indicates P < 0.01, and

double underline indicates P < 0.001.

Discussion

Within the major NZM2410-lupus susceptibility Sle1 locus, increased CD4⁺ T cell activation mapped to both Sle1a and Sle1c (10), with each of these loci corresponding to two subloci (12, 37). The current study focused on the functional characterization and mapping of the CD4⁺ T cell phenotypes associated with Sle1c. Originally defined as the 7.72 Mb at the telomeric end of chromosome 1, Sle1c contains 48 protein-coding genes and 4 microRNAs. Previous analyses have identified Cr2 on the temoleric end of Sle1c as the candidate gene associated with impaired B cell responses (3). In this study, we confirmed that centromeric Sle1c2 enhances CD4⁺ T cell activation, and have established that Sle1c2 CD4⁺ T cells are characterized by a strong TH1 skewing. In addition, we have demonstrated the contribution of Sle1c2 to autoimmune pathology by showing that its expression enhances both induced and spontaneous lupus. Further, congenic mapping and gene expression analysis have identified Esrrg, one of the two genes located in the critical interval, as the candidate gene for Sle1c2. Finally, consistent with published works that have established that Esrrg is a potent regulator of mitochondrial metabolism (33), the decreased expression of Esrrg in Sle1c2 CD4⁺ T cells is associated with a decreased mitochondrial mass and with significant alterations of mitochondrial functions.

ERRγ belongs to the family of nuclear receptors, which are transcription factors whose activities are mediated by endogenous ligands and other coregulatory proteins. Several nuclear receptors have already been implicated in T cell biology, including RARα, RORγ, PPARγ, the vitamin D and the glucocorticoid receptors, and the estrogen receptors (38–40). ERRγ, belonging to the estrogen-related receptor (ERR) family, is expressed at low levels in human spleen and thymus (41), but does not yet have an established role in T cell biology. ERRs are structurally related to estrogen receptors, but do not bind natural estrogens. In fact, ERR endogenous ligands have not been identified and they are thus referred to as orphan nuclear receptors. It has been shown, however, that ERRs are constitutively active due to properties of their ligand binding domains (42). Regardless, several synthetic compounds augment ERRγ-mediated transactivation of its target genes, demonstrating its potential as a therapeutic target (43). Interestingly, bisphenol A (BPA), an industrial component known to be an endocrine disruptor, binds to and alters the activity of ERRγ (44). BPA exposure has been linked to autoimmunity, and, relevant to this study, it has been shown to strongly downregulate IFNγ production (45). This raises the possibility that ERRγ may mediate environmental triggers of autoimmunity.

ERRs are important regulators of metabolism and energy homeostasis and Esrrg expression is highly restricted to metabolically active tissues (42). Esrrg transactivates genes involved in mitochondrial biogenesis, lipid transport and metabolism, tricarboxylic acid cycle, electron transport chain, and oxidative phosphorylation, allowing for energy production by efficient fatty acid oxidation. This vital function is demonstrated in Esrrg null mice, where the lack of a critical switch from glycolytic to lipid based metabolism in the myocardium results in perinatal lethality (34). Conversely, Esrrg overexpression in skeletal muscle resulted in increased mitochondrial biogenesis and function, as well as exercise capacity (33). Activation and proliferation by T cells is also metabolically demanding (46). However, contrary to other metabolically demanding tissues, activated T cells employ aerobic glycolysis to meet energy requirements. Known as the Warburg effect, this form of metabolism generates glucose metabolites at the expense of ATP production (47). Remarkably, when ERRγ expression is suppressed by microRNA, the Warburg effect is induced (48). The rate of glycolysis has direct consequences on CD4⁺ T cell activation and differentiation. GLUT1 transgenic mice, which have increased glucose uptake in their T cells, show T cell activation phenotypes very similar to B6.Sle1c2 mice, namely, age-dependent accumulation of CD69⁺ and Tem cells, increased proliferation, and increased IFNγ production (49). Consequently, aged mice suffered from hypergammaglobulinemia and GN, exhibiting a direct connection between T cell metabolism and autoimmunity. Conversely, as recently reported in Myc-deficient mice, T cells display impaired proliferation with an altered activation phenotype when the switch to aerobic glycolysis is obstructed (50).

We hypothesize that Sle1c2 contributes to the Warburg effect in CD4⁺ T cells as decreased Esrrg expression would limit transcription of target genes that regulate oxidative lipid based metabolism, skewing toward glucose based programs. This hypothesis is supported first by the decreased expression of Esrrg target genes, Ndufs1, Ppif, and Rtn4ip1 that regulate metabolism and mitochondrial function (51–53), and second by the decreased mitochondrial biogenesis and functions in Sle1c2 CD4⁺ T cells. Recently, Rathmell et al. reported that ERRα functions as a regulator of T cell metabolism (54). This isoform, encoded by a separate gene, Esrra, binds to the same DNA consensus sequence as ERRγ, and shares the same general function as a metabolic regulator. However, contrary to the increased T cell activation observed in B6.Sle1c2 mice, which have reduced Esrrg expression in CD4⁺ T cells, Esrra^−/− mice lacked an expansion of Tem as they aged. Interestingly, Esrra^−/− CD4⁺ T cells were not defective in their proliferative and glycolytic capacities in response to TCR activation whereas wild-type T cells in which ERRα was acutely inhibited were deficient in these capacities. This suggested that a developmental compensatory mechanism was offsetting the effect of Esrra deficiency, and indeed the authors observed alterations in mTOR and AMPK pathways. At this point it is not clear why chronic ERRγ deficiency leads to increased activation, while ERRα deficiency results in a more naïve phenotype. It is conceivable that these two isoforms utilize different cofactors to dictate their function. Overall these findings validate our observations that deficiencies in ERRγ cause globally augmented metabolism in CD4⁺ T cells and results in altered T cell function.

Dysregulated T cells are a major contributor to SLE pathogenesis, and targeting them is an important focus for therapeutic intervention (55). Mitochondrial dysfunctions have specifically been identified as a characteristic of human lupus T cells, which may offer additional therapeutic targets (56). Recent studies have highlighted the regulation of metabolism as an important checkpoint of T cell activation and effector functions. Our study associates, for the first time, a gene known as a major regulator of metabolism, mitochondrial dysfunctions, and CD4⁺ T cell activation TH1 polarization in a lupus model. A better understanding of the role of ERRγ in CD4⁺ T cells will provide a better understanding of how metabolism regulates immune functions in the context of autoimmunity.

Abbreviations

SLE

systemic lupus erythematosus

ANA

anti-nuclear autoAb

GN

glomerulonephritis

NZB/W F1

(NZB × NZW) F1

SNP

single nucleotide polymorphism

atRA

all-trans retinoic acid

B6

C57Bl/6J

B6.OTII

B6.Cg-Tg(TcraTcrb)425Cbn/J

B6.bm12

B6(C)-H2-Ab1^bm12/KhEgJ

B6.Thy1^a

B6.Cg-Igh^aThy1^aGpi1^a/J

B6.Tcrbd^−/−

B6.129P2-Tcrb^tm1Mom Tcrd^tm1Mom/J

T_reg

regulatory T cells

BM

bone marrow

Tem

T effector memory cells

VDAC

voltage-dependent anion channel

DAR-4M

Diaminorhodamine-4M

MTG

MitoTracker Green

HE

Hydroethidine

DIP

Deletion/Insertion polymorphism

ERR

estrogen – related receptor

BPA

bisphenol A