T Cell–Specific Adaptor Protein Regulates Mitochondrial Function and CD4⁺ T Regulatory Cell Activity In Vivo following Transplantation

The T cell–specific adaptor protein (TSAd), encoded by the SH2D2A gene, is an SH2 domain-containing intracellular adaptor molecule that activates the protein tyrosine kinase Lck and elicits intracellular signals resulting in cytokine production (1–3). As its name suggests, TSAd was initially identified based on its expression in T cells (1, 3–6), but more recent studies have demonstrated its expression and function in other cell types (6–8). Nevertheless, targeted disruption of the TSAd gene in mice does not cause any major developmental abnormalities and is rather associated with defects in immune function (3, 9, 10). This finding indicates that its primary function is indeed linked to T cell biology. Furthermore, genetic polymorphisms in the SH2D2A gene in humans that result in lower TSAd protein expression levels are associated with the development of multiple sclerosis and juvenile rheumatoid arthritis (2, 11–13). These collective observations are most suggestive that TSAd plays a major role in immunity and, notably, in human autoimmune disease. However, the cell-intrinsic function of TSAd in regulatory T cells (Tregs) is not known, and its in vivo role in the modulation of T effector cell (Teff) activation has not been reported.

Mechanistically, TSAd is reported to augment upstream TCR-dependent signaling through an interaction that results in the amplification of Lck phosphorylation and Lck-dependent responses (4, 14). TSAd also functions to augment synapse formation between CD4⁺ T cells and APCs (10), and it enhances CD28-dependent costimulation (15). In this manner, TSAd decreases the threshold signal required for Ag-dependent activation of CD4⁺ T cell subsets. In addition, it has been reported to function in T cell migration, notably, in response to the chemokines SDF-1α and RANTES (5, 16, 17). However, knockout (KO) of the Sh2d2a gene in mice is not associated with immunodeficiency but, rather, results in susceptibility for autoimmunity (3, 9). Current models suggest that Treg dysfunction in Sh2d2a^−/− mice (also referred to in the literature as TSAd^−/− or TSAd KO mice) is due to a deficiency in IL-2 production by Teffs and thus IL-2–dependent survival and activity of CD4⁺Foxp3⁺ Tregs (9). However, the autoimmune disease previously reported in TSAd^−/− mice is different from that reported in Il2^−/− or Il2r^−/− mice (18–20). Also, recent studies are suggestive that TSAd may have a cell-intrinsic function in Tregs (21). These observations indicate that its biology has broad implications in Treg-associated disease(s).

In this study, we wished to determine the relative function of TSAd in Teffs and Tregs and the development of immunoregulation in vivo following transplantation. Contrary to expectations, we find that allopriming and Teff responses are sustained in TSAd^−/− recipients of cardiac transplants. Furthermore, we find that TSAd^−/− recipients have accelerated graft failure following MHC class II–mismatched transplantation, and they are resistant to the graft-prolonging effects of costimulatory blockade. Mechanistically, we find that TSAd^−/− Tregs are less efficient in suppressing Teff responses and they possess effector function as evidenced by IFN-γ production. Untargeted transcriptomic RNA-sequencing indicates that the major function of TSAd in Tregs relates to its effects on cellular activation, suggesting that it is potent in modulating CD4⁺Foxp3⁺ Treg biology through cell-intrinsic mechanisms. We also find that mitochondria in TSAd^−/− Tregs lack Lck, and have reduced reactive oxygen species (ROS) production/activation. However, TSAd^−/− Tregs compensate by increasing the numbers of mitochondria per cell, and this effect is not associated with altered Treg metabolism. These collective findings have broad implications and suggest that mutations in the SH2D2A gene in humans will result in generalized CD4⁺Foxp3⁺ Treg dysfunction and susceptibility to chronic inflammatory/autoimmune disease(s) and to graft loss following solid organ transplantation.

Male 6–8-wk-old C57BL/6 (H-2^b), B6.C-H-2^bm12 and BALB/c (H-2^d) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Sh2d2a^−/− mice (also called RIBP^−/− and TSAd^−/− were generated by Jeffrey A. Bluestone [University of California, San Francisco, CA (3)] and gifted to the laboratory by Philip King [University of Michigan, Ann Arbor, MI (9)]. The mice have been backcrossed onto the C57BL/6 (H-2^b) background.

Intra-abdominal heterotopic heart transplantation was performed as previously described (22, 23), and graft survival was monitored by palpation of the heartbeat. In some experiments, BrdU (1 mg; BioLegend, San Diego, CA) was injected i.p. every 12 h for 72 h into recipients, and BrdU incorporation in CD4⁺ T cell subsets was analyzed by flow cytometry using the BD BrdU Flow Kit (BD Pharmingen, San Jose, CA). Anti-mouse CD40L Ab (anti-CD154, clone MR-1), rCTLA4–Ig or control Ig (clone human Fc-G1) were administered on days 0, 2, and 4 posttransplantation (200 μg/d; all from Bio X Cell, West Lebanon, NH). All studies were performed according to an approved Institutional Animal Care and Use Committee protocol at Boston Children’s Hospital.

Following harvest, allografts were divided and either frozen in liquid nitrogen in OCT or fixed in 10% formaldehyde in PBS (Thermo Fisher Scientific, Kalamazoo, MI) overnight at 4°C. Formalin-fixed tissue was paraffin-embedded, sectioned (∼3-μm thick), and mounted on Poly-L-Lysine coated slides (Thermo Fisher Scientific, Tewksbury, MA). The slides were deparaffinized, rehydrated, and stained with Harris H&E (all from Thermo Fisher Scientific). Frozen tissue was cryosectioned (∼4-μm thick), fixed in acetone for 10 min, quenched with 0.3% hydrogen peroxide in PBS for 30 min, blocked for 30 min with 2% BSA in PBS and incubated with primary Ab (diluted in 2% BSA/PBS) overnight at 4°C. After a wash in 0.5% Tween 20 in PBS, sections were incubated with a species-specific peroxidase-labeled secondary Ab (Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature. The slides were washed three times and peroxidase-activity was developed using 0.05% 3-amino-9-ethyl-carbazole and 0.015% hydrogen peroxide in 0.05 M acetate buffer (pH 5.5). Finally, each slide was counterstained in Gill hematoxylin (Sigma-Aldrich, St. Louis, MO) and mounted in glycerol gelatin (Thermo Fisher Scientific). Images were evaluated on a Nikon Eclipse 80i (Nikon, Mississauga, ON, Canada) using a Retiga 2000R CCD camera (QImaging, Surrey, Canada) equipped with NIS Elements software (version 3.22.15; Nikon, Melville, NY). Infiltrates were evaluated by standard grid counting at 40× high-power field magnification of 12 different fields per slide by three independent investigators. Data were expressed as the mean infiltration per allograft.

Murine T cell subsets were isolated from splenocytes using the EasySep mouse CD4⁺, naive CD4⁺, and CD4⁺CD25⁺ T Cell Isolation Kits (STEMCELL Technologies, Vancouver, Canada) or FACS sorted using a FACSAria IIU FACS sorter (BD Biosciences, San Diego, CA). Purity was consistently above 95% for bead-isolated cells and up to 98% for FACS-sorted cells (Supplemental Fig. 1). Graft-infiltrating cells were isolated by enzymatic (1 mg/ml collagenase I; Worthington, Lakewood, NJ) and mechanical disruption and subsequently enriched over a 67–44% Ficoll density gradient. Isolated CD4⁺ T cells were cultured in RPMI 1640 (Lonza, Walkersville, MD) supplemented with 10% FBS (Sigma-Aldrich), 2 mM l-glutamine, 1 mM sodium pyruvate, 0.75 g/l sodium bicarbonate, 100 U/ml penicillin/streptomycin, 0.1 mM nonessential amino acids (all Lonza) and 50 μM 2-ME (Sigma-Aldrich). The cells were stimulated with plate-bound anti-CD3 (clone 145-2C11; Bio X Cell) in the absence or presence of soluble anti-CD28 (clone 37.51; BioLegend), and cytokines were evaluated as indicated by each experiment. In vitro differentiation of CD4⁺ Tregs into induced Tregs (iTregs) was performed by culturing naive CD4⁺CD25⁻ cells for 2 d with plate-bound anti-CD3 (2 μg/ml), anti-CD28 (1 μg/ml), anti–IL-4 (500 ng/ml, clone 11B11), anti–IFN-γ (2 μg/ml, clone XMG1.2), murine TGF-β1 (5 ng/ml), murine IL-2 (50 U/ml) (all BioLegend) and 10 nM rapamycin (Wyeth-Ayerst, Philadelphia, PA) or 100 nM all-trans retinoic acid (Sigma-Aldrich) in DMEM (Lonza) supplemented with 10% FBS (Sigma-Aldrich), 2 mM l-glutamine, 1 mM sodium pyruvate, 0.75 g/l sodium bicarbonate, 100 U/ml penicillin/streptomycin, 0.1 mM nonessential amino acids (all from Lonza) and 50 μM 2-ME (Sigma-Aldrich). On day 2, the cells were transferred to a new cell culture plate in medium supplemented with 50 U/ml IL-2 in the absence of anti-CD3 for an additional 2 d, after which efficiency was evaluated by CD25 and Foxp3 expression.

T cell proliferation was assessed by the addition of 1 μCi of [³H]thymidine per well (PerkinElmer, Boston, MA) during the last 16 h of culture. Incorporated [³H]thymidine was assessed in a Wallac 1450 MicroBeta TriLux scintillation counter (PerkinElmer).

CD4⁺ T cells (3 × 10⁴) were cultured in 96-well polyvinylidene fluoride plates (Immobilon-P; Millipore, Billerica, MA) and stimulated with irradiated (1700 rad) splenocytes and/or with anti-CD3/anti-CD28–coated beads (1:1 ratio beads/cell; Life Technologies, Carlsbad, CA) for 24 h. Subsequently, the cells were stained according to the ELISpot manufacturers protocol (eBioscience, San Diego, CA and BD Biosciences). After staining, the plates were scanned and analyzed on an ImmunoSpot S6 Ultra ELISpot reader (version 5.0; CTL, Shaker Heights, OH).

Nuclear extracts were prepared by lysing cells in hypotonic buffer containing 10 mM HEPES, 10 mM potassium chloride, 1.5 mM magnesium chloride, 1 mM EDTA, 500 μM DTT, and 200 μM PMSF (all Sigma-Aldrich) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific, Tewksbury, MA) for 15 min on ice. After the addition of 1% IGEPAL, the cells were vortexed, and cytosolic and nuclear fractions were separated by centrifugation. Nuclei were resuspended in 20 mM HEPES, 25% glycerol, 4.2 mM sodium chloride, 1.5 mM magnesium chloride, 200 μM EDTA, 0.5 μM DTT, 200 μM PMSF (all Sigma-Aldrich), and protease and phosphatase inhibitors (Thermo Fisher Scientific) for 20 min on ice; subsequently, DNA was removed by centrifugation. Mitochondrial proteins were isolated using the AMRESCO Mitochondrial Protein Isolation Buffer (VWR, Radnor, PA) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). Samples were subjected to electrophoresis on 4–15% gradient SDS-polyacrylamide gels (Bio-Rad Laboratories, Hercules, CA), and separated proteins were transferred to PVDF membranes (Millipore). Membranes were blocked in TBS containing 0.1% Tween 20 and 5% BSA (Thermo Fisher Scientific) for 1 h and incubated with primary Abs overnight at 4°C. Membranes were subsequently washed and incubated with species-specific peroxidase-labeled secondary Abs (Jackson ImmunoResearch) for 1 h at room temperature. After washing, the protein of interest was visualized by chemiluminescence (Thermo Fisher Scientific) in a ChemiDoc MP Imaging System (Bio-Rad Laboratories). The membranes were stripped (Thermo Fisher Scientific) and reprobed with Abs to evaluate protein loading.

Cell culture supernatants were profiled using the Mouse Cytokine/Chemokine 25 Plex Magnetic Bead Panel (MiliporeSigma) on a Luminex LX200 platform equipped with xPONENT software (version 3.1.871.0; Luminex, Austin, TX).

Cells were incubated with conjugated mAbs diluted in 0.5% BSA in PBS for 30 min at 4°C, washed, and fixed with 4% paraformaldehyde. Foxp3 expression was evaluated by fixing and permeabilizing cells using a buffer set prior to staining using an anti-Foxp3 mAb (clone FJK-16s) or isotype (eBR2a) as a control, according to the manufacturer’s instructions (all eBioscience). IRF-4 (clone IRF4.3E4), Helios (clone 22F6), EOS (clone ESB7C2; eBioscience), and Blimp-1 (clone 5E7) staining was performed using the True-Nuclear Transcription Factor Buffer set according to supplier’s instructions (all from BioLegend). Cells were analyzed on a FACSCalibur flow cytometer (BD Biosciences) within 24 h of staining, and at least 50,000 gated events were collected per sample. Data were analyzed using FlowJo software (version X.0.7; Tree Star, Ashland, OR).

In vitro suppression was performed by coculture of CFSE-labeled (2 μM; Life Technologies) CD4⁺CD25⁻ T responder cells with increasing ratios of CD4⁺CD25^high Tregs or iTregs and either irradiated (1700 rad) APC or anti-CD3/anti-CD28 coated beads (1:1 ratio beads/responder cell; Life Technologies). Responder cell activation was analyzed using the standard CFSE-dilution assay, and expansion indices were calculated using the FlowJo proliferation module (version 9.9.3; Tree Star); suppression by KO cells were normalized to the suppressive function of wild-type (WT) Tregs, as previously described (24). Normalized suppressive function was plotted against the corresponding ratio of Tregs: naive CD4⁺ responder T cells (Tresp). Suppressive function was also evaluated using the IFN-γ ELISpot assay.

Treg-specific demethylated region (TSDR) DNA methylation analysis was performed on two male and one female TSAd^−/− and on three age/sex-matched WT controls. FACS-sorted CD4⁺ Treg and Teffs (purity of >99%) were pelleted, frozen, and processed for pyrosequencing by EpigenDx (Hopkinton, MA). Briefly, genomic DNA was isolated, bisulfite-modified, and the 14 CpG islands within the promoter and intron 1 of the Foxp3 gene were amplified by PCR and pyrosequenced. Sequences were aligned to the mouse genome, and the percentage of converted (unmethylated) cytosine to uracil/thymidine for each CpG site was evaluated. All 14 CpG regions in each sample were analyzed, and low DNA methylation in the regions 1–7 was used to indicate a comparable purity between KO and WT cells. Only the male samples are shown, as we observed that the inactivated X chromosome remains methylated within the TSDR (maximum of 50% demethylation in female samples).

RNA-sequencing library preparations were performed as previously described (23, 25). Total RNA was isolated using MyOne SILANE Dynabeads (Thermo Fisher Scientific). RNA was fragmented and barcoded using 8-bp barcodes with standard Illumina adaptors. Libraries were enriched using Agencourt AMPure XP bead cleanup (Beckman Coulter, Brea, CA), PCR amplified, gel purified, and quantified using a Qubit HS (High Sensitivity) DNA Kit (Thermo Fisher Scientific). Libraries were quality tested using Tapestation High Sensitivity DNA tapes (Agilent Technologies, Santa Clara, CA) and sequenced on a HiSeq 2500 (Illumina, San Diego, CA) using 50-bp single-end reads. Reads were aligned to the mouse reference genome GRCm38.p6 using 2-pass STAR aligner version 2.6.0 (26), and gene expression was quantified using the GENCODE gene models (Release M14) using HTseq-count method (27). Counts were normalized, and differential gene expression and principle components were calculated using the DESeq2 method (version 1.14.1) in the R Statistical Computing Environment (version 3.3.2) (28). False discovery rates were calculated, and genes with an adjusted p value (Padj) < 0.001 were considered differentially expressed. Gene Set Enrichment Analysis (GSEA; version 3.0; Broad Institute, Cambridge, MA) was performed using the log₂ ratio of TSAd^−/− to WT CD4⁺CD25^high Tregs at each time point, and the immunologic gene sets (C7) of the Molecular Signatures Database with default parameters of GSEA (29). The transcriptomic dataset has been submitted to the Gene Expression Omnibus database under accession number GSE134515 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE134515).

Total RNA was isolated using the RNeasy Isolation Kit (QIAGEN, Valencia, CA) and reverse-transcribed into cDNA using the qScript Supermix from Quanta Biosciences (Gaithersburg, MD). cDNA was diluted 1:10 in RNase-free water and stored at −20°C until use. Quantitative PCR (qPCR) was performed on a 7300 Real-Time PCR system (Applied Biosystems, Foster City, CA) using 1 μl cDNA, PerfeCTa SYBR Green (Quanta Biosciences) and individual primers (Table I) in a final concentration of 5 μM. Samples were run in duplicate, and the relative gene expression was determined with the comparative cycle threshold (Ct) method and expressed as 2^{−Ct(target gene)−Ct(GAPDH)}.

Statistical analyses were performed using one-way ANOVA or the Student t test (Prism, version 5.03; GraphPad, La Jolla, CA) as appropriate. If the equality test failed, the Kruskal–Wallis test was used. All p values <0.05 were considered statistically significant. Heatmaps were generated using the heatmap.2 function in the gplots package (version 3.0.1.1).

We initially cultured purified populations of CD4⁺ T cells from WT or TSAd^−/− mice with anti-CD3/anti-CD28, and we compared proliferative responses as well as IL-2, IL-4, and IFN-γ production by ELISpot. As illustrated in Fig. 1, TSAd^−/− CD4⁺ T cells, including both naive CD4⁺CD25⁻ as well as CD4⁺CD25^high subsets, are hypoproliferative (p < 0.001) and produce lower levels of cytokines (p < 0.001) versus WT cells following activation with mitogen. However, increasing the concentration of mitogen and/or the addition of IL-2 into cultures overcomes this threshold effect and readily induces activation responses in KO cell subsets (Fig. 1B, 1C). These results confirm previous observations (3, 5, 6, 14) and support a model linking TSAd biology to CD4⁺ Teff function. However, TSAd^−/− mice fail to develop symptoms of immunodeficiency, and the phenotypes of TSAd^−/− CD4⁺ T cells and CD4⁺Foxp3⁺ Tregs are similar to WT mice (Supplemental Fig. 2) even following exposure to pathogens [data not shown and (6)]. Thus, we postulated that these in vitro findings may not be of significance in vivo.

To next evaluate CD4⁺ Teff function in vivo, we used TSAd^−/− mice (H-2^b background) as recipients of fully MHC-mismatched BALB/c (H-2^d) donor heart transplants. Contrary to expectations, we found that graft survival was identical in TSAd^−/− and WT recipients (Fig. 2A). Because this mismatch combination results in the clonal expansion, differentiation, and function of CD4⁺ Teffs (30), our findings indicate that TSAd is redundant for alloimmune effector CD4⁺ T cell activation and acute rejection in vivo.

We next evaluated graft survival following the transplantation of MHC class II–mismatched B6.C-H-2^bm12 donor hearts into either TSAd^−/− or WT mice (both H-2^b). In this model, CD4⁺ Teff activation and rejection is more insidious because of the relative expansion and function of CD4⁺ Tregs (31, 32). As expected, median survival time (MST) was >45 d (Fig. 2B). Consistently, however, we observed that graft failure was markedly accelerated in TSAd^−/− recipients (MST = 22 d) versus WT recipients (p < 0.01, Fig. 2B). Two weeks posttransplantation, inflammatory cell infiltrates were markedly increased within allografts harvested from TSAd^−/− recipients (Fig. 2C–G and data not shown) and the mRNA expression of several proinflammatory cytokines were increased in allografts harvested from TSAd^−/− versus WT recipients (Fig. 2H). Phenotyping demonstrated expanded numbers of CD3⁺ T cells, lower numbers of CD4⁺Foxp3⁺ subsets (p < 0.05), and a higher Teff/Treg ratio (p < 0.01) in TSAd^−/− versus WT recipients (Fig. 3A–E, Supplemental Fig. 3). On day 18 posttransplantation, there was no difference in the frequency of CD44^highCD62L^low effector memory Treg (eTreg) subsets or NRP-1⁺ Tregs (Fig. 3F). However, TSAd^−/− Tregs had a trend for lower levels of expression of Helios, EOS, and Blimp-1 (Fig. 3G) as well as cell surface GITR, CTLA4, and PD-1 (Fig. 3H), suggestive of an overall reduced immunoregulatory phenotype. We also examined the expansion and turnover of Tregs in TSAd^−/− transplant recipients using the in vivo BrdU incorporation assay and found no differences versus WT recipients (Fig. 3I). In contrast, TSAd^−/− Teff proliferation was significantly increased (Fig. 3I), and allopriming of recipient CD4⁺ Teffs was greater in TSAd^−/− versus WT recipients (Fig. 3J). Collectively, these findings are most suggestive that TSAd is primarily functional in the suppressive activity of CD4⁺Foxp3⁺ Tregs.

To further determine the effects of TSAd on alloimmune CD4⁺ T cell responses, we examined the effect of costimulatory blockade on graft survival using either TSAd^−/− or WT mice as recipients of fully MHC-mismatched BALB/c cardiac allografts (Fig. 4A, 4B). Peritransplant administration of anti-CD40L (anti-CD154, on days 0, 2, and 4) resulted in prolonged graft survival in WT recipients (MST > 40 d) versus untreated or control Ig-treated recipients (MST of 8 d; p < 0.01, Fig. 4A), as previously reported (23, 33). However, anti-CD40L was ineffective in prolonging survival in TSAd^−/− recipients (MST 19 d, p < 0.01, Fig. 4A), and all grafts failed (n = 9) by day 22 posttransplant (Fig. 4A). Similarly, CTLA4–Ig treatment prolonged graft survival in WT recipients (MST > 40 d; Fig. 4B) but was ineffective in TSAd^−/− recipient mice (MST = 10 d; p < 0.05).

Our collective findings are suggestive that CD4⁺ Treg activity is reduced in TSAd^−/− mice in vivo, resulting in a failure to suppress allopriming and graft rejection. To confirm this interpretation, we next transplanted fully MHC-mismatched BALB/c hearts into C57BL/6 Rag2^−/− Il2rg^−/− recipients and adoptively transferred either TSAd^−/− or WT Tregs on day 2 posttransplantation. On day 18 posttransplantation, recipient mice received CD4⁺CD25⁻ Teffs to challenge Treg-based immunoregulation. In control mice (without Treg transfer on day 2), we find that day 18 transfer of WT Teffs elicits graft failure within 16 d posttransfer. In contrast, transfer into mice that received WT Tregs (on day 2) resulted in a prolongation of graft survival to a median of 50 d posttransfer (Fig. 4C). However, graft survival was not as prolonged when TSAd^−/− Tregs were used in the initial day 2 adoptive transfer (MST = 30 d; p < 0.05 versus WT Tregs, Fig. 4C).

To understand the molecular basis for the function of TSAd in Tregs, we performed untargeted transcriptomic profiling of FACS-sorted and mitogen-activated TSAd^−/− and WT CD4⁺CD25^high cells using RNA-sequencing (Fig. 5A–D). We find that TSAd regulates (Padj < 0.001) the expression of only a few genes in resting Tregs (no anti-CD3 stimulation; Supplemental Fig. 4A). Whereas WT Tregs respond to anti-CD3 stimulation with the differential expression of >300 genes (Supplemental Fig. 4B), we find that anti-CD3–activated TSAd^−/− Tregs have a minimal response, with the differential expression of only four genes (Supplemental Fig. 4B). Principal component analyses of WT transcriptomes showed activation-induced changes (Fig. 5B). In contrast, TSAd^−/− transcriptomes did not respond to mitogen activation (Fig. 5B). GSEA comparing activated TSAd^−/− to WT transcriptomes showed a high normalized enrichment score for genes expressed in resting Tregs (Fig. 5C). Independent of their activation status, we also noted that TSAd^−/− Treg transcriptomes were enriched for genes that are expressed in Teffs (Fig. 5D, Supplemental Fig. 4C), suggesting that TSAd^−/− Tregs have potential to be unstable and/or possess effector function. Collectively, untargeted transcriptomic analysis indicates that TSAd^−/− Tregs are generally unresponsive to anti-CD3 stimulation, further supporting a critical role for TCR/TSAd–induced signaling in active immunoregulation and/or in Treg phenotype(s).

To next evaluate the effect of TSAd on CD4⁺ Treg stability, we performed DNA methylation assays of CpG motifs within the TSDR of the Foxp3 gene. CD4⁺Foxp3⁺ Tregs were FACS-sorted from the spleens of WT or TSAd^−/− mice, DNA was isolated, and methylation was analyzed, as we described (23). In contrast to WT Tregs, which are heavily demethylated, pyrosequencing demonstrated increased methylation of CpG regions 9–14 of the TSDR in TSAd^−/− Tregs (Fig. 5E, Supplemental Fig. 4D). This observation is further suggestive that TSAd^−/− Tregs are not fully activated and/or may lack suppressive function. In standard in vitro suppression assays (using purified FACS-sorted populations of WT or TSAd^−/− Tregs), we find no major difference in the percentage inhibition of WT CD4⁺CD25⁻ responder proliferation by each Treg population (Fig. 5F, 5G). However, TSAd^−/− Tregs were found to produce high levels of IFN-γ in suppression assays (Fig. 5H). Furthermore, purified populations of CD4⁺CD25^high Tregs isolated from the spleens of TSAd^−/− mice produced significantly higher levels of IFN-γ (p < 0.01), and there was a trend for higher secretion of IL-17 as compared with WT Tregs (Supplemental Fig. 4E). Finally, we evaluated the effect of TSAd on Treg expansion and differentiation by culturing naive TSAd^−/− CD4⁺ T cells in iTreg-inducing conditions in vitro. As illustrated in Supplemental Fig. 4F (and data not shown), we find that TSAd^−/− iTregs (CD4⁺CD25^highFoxp3⁺ cells) expand in a similar manner as that observed using WT cells. Collectively, these observations are suggestive that TSAd promotes activation responses within CD4⁺ Treg subsets and that KO cells have the potential to dedifferentiate and/or to possess effector function.

By Western blot analysis, we find that TSAd is expressed in the cytosol but not within the mitochondria of Tregs (Fig. 6A). We also find that Lck is expressed in both the cytosol and the mitochondria of WT CD4⁺CD25^high Tregs (Fig. 6B) but is notably lacking on the mitochondria of TSAd^−/− Tregs (Fig. 6B). This striking finding is suggestive that TSAd regulates Lck kinase activity within mitochondria. Because the Lck kinase is reported to be functional within the mitochondria of CD4⁺ T cells (34), we next assessed depolarization of mitochondria following treatment with anti-CD3 and found that it is markedly absent in TSAd^−/− versus WT cells (Fig. 6C). We also used a ROS-sensitive probe (CM-H₂DCFDA) to evaluate function and found that activation-induced ROS production is lacking in mitochondria from TSAd^−/− Tregs following stimulation (Fig. 6D). These findings indicate that TSAd is associated with mitochondrial Lck activity and is functional in mitochondrial activation/depolarization within Tregs.

We next isolated CD4⁺CD25^high Tregs from WT and TSAd^−/− mice, and compared metabolic activity using extracellular flux assays in a Seahorse XFe96 bioanalyzer. The measurement of glycolytic activity (as assessed by the extracellular acidification rate [ECAR]) (Fig. 6E, left panel), was identical in each cell type. Also, oxidative phosphorylation, as measured by the oxygen consumption rate (OCR), was identical at basal levels as well as under maximal stress conditions (Fig. 6E, right panel). Thus, metabolism appears to be intact despite the absence of mitchondrial Lck in TSAd^−/− Tregs.

Because Lck-dependent events also regulate mitochondrial morphology (34), we finally performed electron microscopy on WT and TSAd^−/− Tregs. As illustrated in Fig. 6F, we find a prominent increase in the number of mitochondria per TSAd^−/− cell (Fig. 6F; p < 0.0001 versus WT Tregs), but there was no demonstrable difference in morphology. We also used MitoTracker Green uptake and qPCR to confirm the significant increase in mitochondrial mass in TSAd^−/− Tregs versus WT cells (Fig. 6G, 6H, p < 0.05 and p < 0.0001, respectively). We interpret these collective observations to suggest that the increase in mitochondrial mass in TSAd^−/− cells is sufficient to sustain metabolic function at baseline and may account for the normal phenotype of Tregs in unmanipulated mice. Furthermore, our findings demonstrate that the additional effects of TSAd on eTreg activation responses and Foxp3 stability are critical to sustain immunoregulation following transplantation.

In these studies, we show that the Lck adaptor protein TSAd elicits cell-intrinsic signals in CD4⁺Foxp3⁺ Tregs that are associated with enhanced activation and immunoregulatory function. Specifically, we find that graft failure is accelerated in TSAd^−/− recipients of MHC class II–mismatched cardiac transplants and that TSAd^−/− recipients are resistant to the graft-prolonging effects of costimulatory blockade. Furthermore, in adoptive transfer studies, we find that TSAd^−/− Tregs are less efficient than WT Tregs in the suppression of allograft rejection. Untargeted transcriptomic analysis indicates that the major phenotype of TSAd^−/− Tregs is a lack of TCR-dependent activation, and mechanistic studies demonstrate that TSAd^−/− Tregs lack mitochondrial Lck expression as well as ROS production. Collectively, these findings indicate that TSAd has multiple mechanisms of function in Tregs. Because mutations in the SH2D2A gene (encoding TSAd) are associated with autoimmunity in humans, these observations have broad clinical implications.

Previous studies (3, 5, 14) have demonstrated that TSAd regulates the threshold for TCR-dependent activation in CD4⁺ Teffs in vitro. Similarly, we find that mitogen-induced activation of naive TSAd^−/− CD4⁺ T cells in vitro is attenuated as compared with WT controls. Because TSAd is rapidly induced upon T cell activation, this finding is suggestive that its biological effects are important to sustain the continuous activation of effector cells. However, the differentiation of T helper cells is normal in the absence of TSAd (6), and TSAd^−/− mice are not immunodeficient (3, 9) and have an improved tumor clearance compared with their WT littermates (21). In our studies, we find that acute rejection and graft failure is not delayed following fully MHC-mismatched transplantation into TSAd^−/− mice, and alloimmune priming and rejection is enhanced following MHC class II–mismatched cardiac transplantation. Thus, despite in vitro findings suggesting that TSAd promotes and/or sustains effector responses, we find that alloimmune CD4⁺ T cell activation is fully functional in vivo in the absence of TSAd.

In this study, our key findings demonstrated that the primary biological effects of TSAd relate to its function in CD4⁺Foxp3⁺ Tregs. For example, the absence of Lck protein expression on mitochondria within TSAd^−/− Tregs is suggestive that TSAd serves (in part) to shuttle Lck from the cytoplasm to the mitochondria of Tregs. It has however been challenging to link this biological response to a generalized defect in immunometabolism. Whereas activation-induced ROS production is absent in Tregs from TSAd^−/− mice, ECAR and OCR is similar to that observed in WT mice. Because there are increased numbers of mitochondria in TSAd^−/− Tregs, it is possible that there is some compensation to sustain immunometabolism. However, it is also possible that these effects are independent.

Another major mechanistic observation from untargeted transcriptomic profiling of Tregs indicates that the key phenotype of KO Tregs is a profound defect in TCR-dependent activation responses. Thus, it is most likely that TSAd functions to elicit eTreg activation responses and these effects are associated with TCR-independent translocation of Lck to mitochondria.

We also questioned whether the effector phenotype (IFN-γ production) of KO Tregs is interrelated with a lack of Lck and ROS production by mitochondria. In pilot studies (data not shown), we performed iTreg induction assays in the absence or presence of mitochondrial-specific ROS inhibitors or mitochondrial-specific ROS inducers. Both increased and decreased ROS generation during iTreg induction resulted in lower Foxp3 positivity, but activation responses and IFN-γ production did not change in the presence of ROS inhibitors or inducers. Thus, we believe that the effects of TSAd are not linked but are related to its generalized function on eTreg activation. Nevertheless, definitive proof will require additional studies, and, perhaps, the generation of Lck transgenic mice, which are beyond the scope of this report.

Classically, IFN-γ is associated with Teff responses, allograft rejection, and CD4⁺ T helper 1– and CD8⁺-mediated immune responses. However, IFN-γ expression has multiple intrinsic and extrinsic functions that are context dependent in terms of the local tissue microenvironment (35–38). For example, coexpression of the Th1 master transcription factor Tbet in Foxp3⁺ Tregs is important for controlling Th1 and CD8 effector responses (38), but IFN-γ production by unstable Tregs is reported to break immunological tolerance (35, 39). We observed that TSAd^−/− Tregs were efficient in the suppression of CD4⁺ Teff proliferation despite high levels of IFN-γ production. However, analysis of CpG motifs within the TSDR of the Foxp3 gene demonstrated enhanced methylation, suggesting that the absence of TSAd is associated with reduced lineage stability.

In summary, our findings in this report identify novel cell-intrinsic mechanisms whereby TSAd functions to enhance CD4⁺Foxp3⁺ Treg activity. Our observations are most significant for the understanding of immunoregulation following transplantation, especially following treatment with costimulatory blockade (40). In addition, our findings have broad implications (beyond transplantation) because polymorphisms within the SH2D2A gene in humans have been linked to autoimmunity.

We thank Megan Cooper, Kayla MacLeod, and Francesca D’Addio (all from Boston Children’s Hospital) for technical support and Dr. Gary Visner for supervision of this work within the mouse surgical facility. We also thank Dr. Peter Sage (Brigham and Women's Hospital) for guidance on RNA-sequencing, Daniel Brown (Beth Israel Deaconess Medical Center), Dr. Stephen Alexander (The Children’s Hospital at Westmead, Sydney, Australia), and Dr. Anne Spurkland (University of Oslo, Norway) for helpful discussions.

The authors have no financial conflicts of interest.