Mcu ^fl/fl (Kwong et al., 2015) (strain 029817), Foxp3 ^Cre (strain 016959) and Cd4 ^Cre mice (strain 022071) were purchased from the Jackson laboratories and housed at the University of Würzburg. All animals were on a pure C57BL/6J genetic background and maintained under SPF conditions.

T cells were isolated using MojoSort Mouse T cell isolation kits (BioLegend) and cultured in complete RPMI 1640 medium (Gibco) containing 1 g/L glucose. 24-well plates were pre-coated with 12 μg/ml polyclonal anti-hamster IgG, washed with PBS and activated with either 0.25 μg/ml (for Th17) or 0.5 μg/ml (for Th1 and iTreg subsets) of anti-CD3 (clone 145-2C1) together with 1 μg/ml anti-CD28 (37.51). The following culture conditions were used: Th1 cells: 2.5 μg/ml anti-IL-4 (clone 11B11), 10 ng/ml rhIL-2 and 10 ng/ml rmIL-12. iTreg cells: 2.5 μg/ml anti-IL-4, 2.5 μg/ml anti-IFNγ, 10 ng/ml rhIL-2 and 5 ng/ml rhTGF-beta. Th17 cells: 2.5 μg/ml anti-IL-4, 2.5 μg/ml anti-IFNγ, 20 ng/ml rmIL-6 and 0.5 ng/ml rhTGFβ. Tc1 cells: 2.5 μg/ml anti-IL-4, 10 ng/ml rhIL-2 and 10 ng/ml rmIL-12. Tc17 cells: 2.5 μg/ml anti-IL-4, 2.5 μg/ml anti-IFNγ, 20 ng/ml rmIL-6 and 0.5 ng/ml rhTGFβ. All antibodies and cytokines were from Bio X Cell and Peprotech, respectively. To detect cytokine expression, cells were stimulated with 1 μM ionomycin plus 30 nM phorbol myristate acetate (Calbiochem) for 5 h in the presence of brefeldin A (BioLegend). In some experiments, T cell were treated for 72 h with 2–10 µM Ruthenium red (Sigma).

MOG_35–55 peptide (Synpeptide) was emulsified in CFA with 5 mg/ml M tuberculosis H37Ra (BD Difco). 200 mg MOG_35–55 was injected subcutaneously at the flanks of the mice, followed by intraperitoneal injection of 250 ng pertussis toxin (Enzo) on day 0 and 2. The EAE score was assigned as described before (Stromnes and Goverman, 2006). Mice were sacrificed and spinal cords were isolated, minced into small pieces and digested with 1 mg/ml collagenase D (Roche) and 20 μg/ml DNase I (Sigma) for 40 min at 37°C. CNS-infiltrating lymphocytes were enriched by percoll gradient centrifugation and analyzed by flow cytometry.

The LCMV clone 13 strain was kindly provided by R. Ahmed (Emory University). LCMV was grown in BHK-21 cells and viral titers (PFU) in the supernatant were determined as described (Vaeth et al., 2016; Ataide et al., 2020). For chronic viral infections, mice were injected intravenously with 4 × 10⁶ PFU of LCMV and analyzed 23 days post infection.

Cells were blocked with anti-FcγRII/FcγRIII (Bio X Cell; clone 2.4G2) and debris was excluded using the viability dye eFluor780 (eBioscience). Staining of surface molecules with fluorochrome-conjugated antibodies was performed in PBS containing 0.1% BSA. Intracellular (IC) and transcription factor (TF) staining was carried out with the IC Staining and TF Staining Buffer Kit, respectively (eBioscience). Samples were acquired on a BD FACSCelesta flow cytometer and analyzed with FlowJo Software (Tree Star). The following antibodies were used: anti-mouse CD4 (clones 53–6.7 and GK1.5), anti-mouse IL-17A (C11-18H10.1), anti-mouse IFNγ (XMG1.2), anti-mouse GM-CSF (MP1-22E9), anti-mouse TNFα (MP6-XT22), anti-mouse IL-2 (JES6-5H4), anti-mouse Foxp3 (FJK-16s), anti-mouse T-bet (4B10), anti-mouse RORγt (Q31-378), anti-mouse CD25 (PC61), anti-mouse Ki-67 (B56), anti-CD44 (IM7), anti-CXCR5 (SPRCL5), anti-PD-1 (RMP1-30), anti-CD38 (90), anti-GL.7 (GL7), anti-CD8α (53–6.7), anti-Tim3 (RMT3-23). All antibodies were from BioLegend or eBiosciences. PE-labelled tetramers loaded with the LCMV peptides GP_33-41 and GP_66-77 were provided by the NIH Tetramer Core Facility.

T cells were labelled with CellTrace Violet (Thermo Scientific) according to the manufacturer’s instructions. Cells were blocked with 50% FBS, washed twice with RPMI and stimulated with 1 μg/ml plate-bound anti-CD3 (clone 2C11) and anti-CD28 antibodies (37.51; both Bio X Cell) with 50 U/ml IL-2 (Peprotech). CTV dilution was monitored daily by flow cytometry.

T cells were labelled with 2 μM Fura-2-AM (Thermo Scientific) as described earlier (Vaeth et al., 2017b). Cells were attached to 96-well imaging plates coated with 0.01% poly-L-lysine (Sigma) and washed with Ca²⁺-free Ringer solution. Changes in intracellular Ca²⁺ concentration were analyzed using a FlexStation3 plate reader (Molecular Devices) at 340 and 380 nm. Cells were stimulated with 1 μM thapsigargin (TG) (Sigma) in Ca²⁺-free Ringer solution and SOCE was analyzed after re-addition of 2 mM Ca²⁺. Alternatively, whole splenocytes were loaded with 2 μM Fluo4-AM (Thermo Scientific) and incubated with anti-CD4-APC and anti-CD8a-BV421 antibodies (BioLegend). Measurements of intracellular Ca²⁺ changes were recorded by flow cytometry. Baseline cytosolic Ca²⁺ levels acquired in 0 mM Ca²⁺ Ringer solution, before cells were stimulated with 1 μM thapsigargin and re-addition of 2 mM Ca²⁺. Mitochondrial uptake of free Ca²⁺ was monitored in the presence of digitonin-permeabilized T cells (0.04%; preincubation time 10 min). 1 × 10⁶ cells per well were re-suspended in PTP buffer (125 mM KCl, 2 mM K₂HPO₄, 1 mM MgCl₂, 10 mM HEPES, 20 µM EGTA, pH 7.4) containing 2 mM succinate, 2 µM thapsigargin and 1 μM CalciumGreen-5N (Thermo Scientific). Fluorescence (503/535 nm) was monitored using a Tecan M200 Pro reader in a 96-well plate with injection of 25 µM Ca²⁺ pulses.

CNS specimens were fixed in 4% PFA, embedded in paraffin and stained with Luxol fast blue and Cresyl violet (both Carl Roth) to detect myelin and nuclei, respectively.

RNA was extracted using the RotiPrep RNA mini kit (Carl Roth), followed by cDNA synthesis with the iScript cDNA synthesis Kit (BioRad). qRT-PCR was performed using iTaq Universal SYBR Green SuperMix (Bio-Rad) and the CFX RT-PCR thermocycler (BioRad). For quantitation, C _T values were normalized to 18S gene expression and analyzed using the 2^−ΔCT method. Mcu-for: TTA​GCA​GAA​AAG​CAG​AGA​AGA​G, Mcu-rev: TGA​TGA​AGT​AGG​TGA​CGG​G, Letm1-for: CGG​GGT​AGT​CTG​AGG​GAT​CG, Letm1-rev: TGG​AGT​ACA​GCA​ACG​AGA​CAG, Slc8b1-for: CTG​GAA​GTG​TCA​ACC​AGA​CTG, Slc8b1-rev: AGT​CAC​AGC​GAT​CAG​ATG​TGT, 18S-for: CGG​CGA​CGA​CCC​ATT​CGA​AC, 18S-rev: GAA​TCG​AAC​CCT​GAT​TCC​CCG​T.

Cells were resuspended in RIPA lysis buffer with complete protease inhibitor cocktail (Thermo Scientific). 40 μg of total protein was fractionated by SDS-PAGE and transferred onto a nitrocellulose membrane. Membranes were incubated with antibodies against β-Actin (1:5,000, mouse, clone C4; SCBT), VDAC (1:1,000, mouse, N152B/23; BioLegend), GAPDH (1:200, mouse, 6C5; SCBT) and MCU (1:1,000, rabbit polyclonal; CST). For detection, peroxidase-coupled secondary anti-mouse or anti-rabbit antibodies (BioRad) and ECL Substrate (Thermo Scientific) were used.

Glycolytic proton efflux rate (PER) and oxygen consumption rate (OCR) were measured using a XFe96 extracellular flux analyzer (Seahorse Bioscience) as described before (Vaeth et al., 2017a; Kahlfuss et al., 2020).

All results are means with standard error of the means (SEM). The statistical significance of differences between experimental groups was determined by unpaired Student’s t-test. Differences were considered significant for p values < 0.05.

To investigate the role of MCU (CCDC109A) and mitochondrial Ca²⁺ uptake in primary T cells, we generated mice with T cell-specific deletion of the Mcu gene by crossing Mcu ^fl/fl mice (Kwong et al., 2015) to Cd4 ^Cre animals that express Cre recombinase under the control of the Cd4 promoter. Mcu ^fl/fl Cd4 ^Cre mice were indistinguishable from their littermates and showed a normal composition of thymic T cell populations (Supplementary Figure S1A). The frequencies and numbers of peripheral conventional and regulatory (Treg) T cells were also unaltered (Supplementary Figures S1B–D). Naïve and effector T cell composition was comparable between WT and Mcu ^fl/fl Cd4 ^Cre mice and MCU-deficient mice showed no spontaneous immune dysregulation (Supplementary Figure S1E). We next confirmed that the expression of MCU was completely abolished in T cells at the mRNA (Figure 1A) and protein level (Figure 1B). We next measured mitochondrial Ca²⁺ uptake in digitonin-permeabilized WT and MCU-deficient T cells after extracellular Ca²⁺ addition (Figure 1C). Although we observed a reduced mitochondrial Ca²⁺ uptake in absence of MCU, primary mouse T cells showed only a moderate capacity to buffer extracellular Ca²⁺ elevations. Typically, the extracellular Ca²⁺ pulses are rapidly taken up by the mitochondria until mPTP opening becomes visible by a sudden elevation of free cytosolic Ca²⁺. Our best explanation why we did not measure significant mitochondrial Ca²⁺ uptake in these experiments is that primary T cells have fewer and relatively small mitochondria compared to other immune cells, such as macrophages, which makes it difficult to monitor their Ca²⁺ buffering capacity. In addition, other mitochondrial Ca²⁺ transporters, such as the Na⁺/Ca²⁺/Li⁺ exchanger NCLX and the H⁺/Ca²⁺ exchanger Letm1, may compensate for the loss of MCU. Although expression of NCLX and Letm1 was unchanged (Figure 1D), this data does not exclude adaptational changes in mitochondrial Ca²⁺ handling and other compensatory mechanisms, when MCU is genetically ablated.

Previous reports presented conflicting results regarding MCU’s effect on cytosolic Ca²⁺ elevations. In most studies, genetic or pharmacological inhibition of MCU attenuated ER store depletion and SOCE (Hoth et al., 1997; Gilabert et al., 2001; Samanta et al., 2014; Samanta et al., 2020), presumably due to a negative feedback regulation known as Ca²⁺-dependent inactivation (CDI) of IP₃R and/or CRAC channels (Samanta et al., 2014; Deak et al., 2014; Tang et al., 2015). By contrast, other studies reported unchanged or elevated SOCE upon silencing of MCU (Paupe and Prudent, 2018; Wang P. et al., 2020; Seegren et al., 2020). In our experiments using primary MCU-deficient T cells, we observed a small but reproducible increase in cytosolic Ca²⁺ levels after ER store depletion with the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) inhibitor thapsigargin (TG) (Figures 1E–G). Although unexpected, our results are in line with recent data that also found that deletion of MCU by CRISPR/Cas9-mediated genome editing elevates SOCE and NFAT nuclear translocation in different lymphocytic cell lines and in primary T and B cells (Yoast et al., 2021).

To determine how impaired mitochondrial Ca²⁺ buffering and elevated SOCE affects the activation and effector function of T cells, we isolated naïve CD4⁺ T cells from WT and Mcu ^fl/fl Cd4 ^Cre mice and differentiated these cells into different T helper (Th) subsets. The expression of MCU was similar in all Th cells tested (Supplementary Figure S2A). We focused on Th1, Th17 and iTreg cells for further analyses as these subsets are not only well-defined by their (inflammatory) cytokine profiles but also differ in their dependency on mitochondrial metabolism. However, we did not detect differences in oxygen consumption rate (OCR) or extracellular acidification (glycoPER) as measures of mitochondrial and glycolytic activity, respectively, when we compared MCU-deficient T cells to WT controls (Figures 1H,I). Stimulation of WT and MCU-deficient T cells under Th1, Th17 and iTreg-polarizing conditions revealed also no difference in the expression of activation markers (Figure 2A). Cell cycle entry (Supplementary Figure S2B) and proliferation of MCU-deficient CD4⁺ and CD8⁺ T cells was also comparable to WT controls (Figure 2B), indicating that MCU is dispensable for T cell metabolism and activation. To test whether MCU affects T cell differentiation, we analyzed the expression of T-bet, RORγt and Foxp3 as the “signature” transcription factors of Th1, Th17 and iTreg cells, respectively. We observed, however, no differences in the generation of iTreg (Figure 2C) or inflammatory Th1 and Th17 cells in vitro (Figure 2D, Supplementary Figure S2C), suggesting that mitochondrial Ca²⁺ is not required for T cell differentiation. Finally, we tested whether MCU plays a role in the effector function of T cells by re-stimulating Th1, Th17 and iTreg cells with PMA and ionomycin and analyzed the cytokine expression of WT and MCU-deficient cells by flow cytometry. However, we did not observe significant differences in IFNγ and IL-17A expression by Th1 and Th17 cells, respectively (Figure 2E). Likewise, GM-CSF and IL-2 expression was unaltered in MCU-deficient T cells (Supplementary Figure S2D). Similar as “helper” CD4⁺ T cells, “cytotoxic” CD8⁺ T cells can be also polarized into Tc1 and Tc17 cells that are not only characterized by an inflammatory (Tc1) and memory (Tc17) phenotype but also differ greatly by their mitochondrial activity (Flores-Santibáñez et al., 2018). As in CD4⁺ T cells, ablation of MCU did not affect differentiation and cytokine expression of Tc1 and Tc17 cells (Figure 2F). Collectively, these data suggest that genetic deletion of MCU does not alter T cell metabolism, differentiation and effector function in vitro. By contrast, the widely used, but unspecific, MCU inhibitor Ruthenium Red attenuated T cell metabolism, proliferation and survival (Supplementary Figures S3A–F). Because pharmacological blockade of MCU does not recapitulate the findings of MCU-deficient T cells, this data hints at compensatory mechanisms in MCU knockout mice and/or off-target effects of pharmacological inhibitors.

Although in vitro assays are useful to evaluate basic functions of T cells, the complexity of T cell interaction with lymphoid and non-lymphoid cell types and the varying metabolic conditions in different target tissues cannot be fully emulated in vitro. We therefore employed two complementary animal models of T cell-mediated inflammation; experimental autoimmune encephalomyelitis (EAE) (Figure 3), which resembles aspects of human multiple sclerosis, and persistent viral infection with the lymphocytic choriomeningitis virus (LCMV) strain clone 13 (Figure 4). We first induced EAE in Mcu ^fl/fl Cd4 ^Cre and littermate control mice by immunization with MOG_35-55 peptide emulsified in CFA and monitored the disease progression over the course of 20 days. T cell-specific deletion of MCU did not alter EAE immunopathology, including the paralysis of their extremities (Figure 3A), inflammation-induced weight loss (Figure 3B) and the demyelination of the spinal cord (Figure 3C). The infiltration of lymphocytes into the CNS was also similar in Mcu ^fl/fl Cd4 ^Cre mice compared to control animals (Figure 3D). In line with our in vitro data, the production of IFNγ, IL-17A, GM-CSF, IL-2 and TNFα by encephalitogenic Th1 and Th17 cells following re-stimulation with PMA and ionomycin was not impaired in absence of MCU (Figures 3E,F). Because mitochondrial Ca²⁺ uptake is impaired in all T cells of Mcu ^fl/fl Cd4 ^Cre mice, defects in both inflammatory and regulatory T cells could potentially mask subset-specific functions of MCU. To explore a cell-intrinsic role of MCU in Treg cells, we generated Mcu ^fl/fl Foxp3 ^Cre mice that lack MCU expression specifically in Foxp3⁺ Treg cells. As observed in mice with ablation of MCU in all T cells, Mcu ^fl/fl Foxp3 ^Cre mice were indistinguishable from their littermates and showed normal frequencies of thymocytes and peripheral T cells (Supplementary Figure S4A–D). Importantly, Mcu ^fl/fl Foxp3 ^Cre mice showed no signs of an overt immune activation (Supplementary Figure S4E) and the numbers of Treg cells in peripheral lymphoid organs were unaltered. The differentiation of thymus-derived Treg cells into CD44⁺CD62L^– effector Treg cells was also not perturbed in absence of MCU (Supplementary Figures S4F,G), suggesting that MCU is not required for Treg development and their suppressive function in vivo. In addition to autoimmunity and T cell responses to self-antigens, we tested the anti-viral activity of CD4⁺ and CD8⁺ T cells in WT and Mcu ^fl/fl Cd4 ^Cre mice after infection with the LCMV strain clone 13 (Figure 4). LCMV infection promotes the differentiation of CD4⁺ T cells into T follicular helper (Tfh) cells to support affinity maturation of germinal center (GC) B cells and anti-viral humoral immunity. We did not detect differences in the activation of CD4⁺ T cells and Foxp3⁺ Treg cells in Mcu ^fl/fl Cd4 ^Cre mice compared to WT controls (Figure 4A). The clonal expansion of LCMV-specific CD4⁺ T cells, which were monitored with tetramers loaded with the LCMV-derived GP_66-77 peptide, was also unaltered (Figure 4A). Furthermore, Tfh cell differentiation (Figure 4B) and GC B cell responses (Figure 4C) were intact in Mcu ^fl/fl Cd4 ^Cre mice, demonstrating that MCU is dispensable for ‘helper’ T cell responses during viral infection. Although LCMV induces a strong CD8⁺ T cell-mediated immune response, persistent antigenic stimulation during chronic LCMV infection also promotes T cell ‘exhaustion’ that is characterized by an upregulation of inhibitory receptors (such as PD-1 and Tim3), loss of effector function (e.g., cytokine expression) and apoptosis of CD8⁺ T cells (McLane et al., 2019). Using GP_33-41 tetramers to identify LCMV-specific CD8⁺ T cells, we did not observe differences in the generation and/or expansion of cytotoxic immune responses in Mcu ^fl/fl Cd4 ^Cre mice compared to control animals (Figure 4D). T cell exhaustion (Figure 4E) and cytokine expression (Figure 4F) of MCU-deficient T cells were also comparable to those in WT mice, suggesting that MCU is also expendable for ‘cytotoxic’ anti-viral immune responses.

Collectively, these data demonstrate that, despite attenuating mitochondrial Ca²⁺ uptake and elevating cytosolic Ca²⁺ levels after antigen receptor stimulation, MCU is largely dispensable for T cell-mediated immune responses in vitro and in models of autoimmunity and persistent viral infection.