Dendritic Cells Require TMEM176A/B Ion Channels for Optimal MHC Class II Antigen Presentation to Naive CD4⁺ T Cells

Multiple ion channels and transporters are expressed in both innate and adaptive immune cells to control various vital functions, from membrane potential regulation to receptor signaling or migration (1). The role of ion flux has notably been fully appreciated following the molecular characterization of store-operated Ca²⁺ entry (SOCE) through Ca²⁺ release–activated Ca²⁺ (CRAC) channels mediated by the ORAI/STIM complex, best characterized in T cells. However, this system appears dispensable for key functions of macrophages and dendritic cells (DCs) whereas Ca²⁺ signaling remains critical in these cells (2, 3). This observation points to the importance of alternative systems, yet to be discovered, that regulate the intracellular and luminal concentrations of Ca²⁺ but also other ions, including Na⁺, K⁺, Cl^–, Mg²⁺, or Zn²⁺, for the control of immune responses. In comparison with the plasma membrane, there is still a paucity of studies investigating the role of intracellular ion channels and transporters, notably in DCs (4), that could provide major insights into the understanding of new immunomodulatory mechanisms.

We previously showed that the coregulated genes Tmem176a and Tmem176b encode redundant acid-sensitive, non-selective cation channels (5, 6) whose precise functions remain largely unknown in vivo. Transcriptomic, single-nucleotide polymorphism, or epigenetic analysis have associated these homolog genes with different pathologies such as multiple sclerosis (7), chronic obstructive pulmonary disease (8), or age-related macular degeneration (9). These findings suggest an important role of Tmem176a/b in the development of inflammatory diseases, thus emphasizing the need to identify the immune cell types and the functions in which they are predominantly involved.

We initially cloned Tmem176b (originally named Torid) as an overexpressed gene encoding an intracellular four-span transmembrane protein in myeloid cells infiltrating non-rejecting allografts (10). We later demonstrated its contribution to the suppressive function of ex vivo–generated tolerogenic DCs through Ag cross-presentation by allowing cation (Na⁺) counterflux required for progressive endophagosomal acidification (5). However, the function of this ion flux in the homeostasis and physiological response of conventional DCs (cDCs) has not been explored and is likely achieved by both TMEM176A and TMEM176B in a redundant fashion (6).

Unexpectedly, besides myeloid cells, we and others reported a strong expression of Tmem176a and Tmem176b in the retinoic acid-related orphan receptor (ROR)γt⁺ lymphoid cell family, also referred to as type 3 (or type 17) immune cells, producing the prototypical cytokines IL-17 and IL-22 and including Th17 CD4⁺ T cells, γδT17 cells, group 3 innate lymphoid cells (ILC3s), and NKT17 cells (6, 11–14). Moreover, Littman and colleagues (15) included Tmem176a/b in the restricted group of 11 genes whose expression is directly dependent on RORγt in Th17 cells. However, in our previous study, we found no effect or only a modest effect of Tmem176b single deficiency in different models of autoinflammation linked to type 3 immunity (6). We speculated that the absence of Tmem176b could be efficiently compensated by its homolog Tmem176a, located in the same genomic locus, thus masking possible phenotypic alterations.

In this study, we have used germline and conditional (“floxed”) double knockout (DKO) mice to unequivocally determine the importance of Tmem176a and Tmem176b in the biology of RORγt⁺ cells and cDCs in vivo. In that respect, to our knowledge, this is the first study exploring the consequence of Tmem176a/b double deficiency in vivo. Furthermore, we have combined these functional results with the elucidation of the precise trafficking of both proteins using a real-time imaging method. Our findings show that, although Tmem176a/b appear surprisingly dispensable for RORγt⁺ cell functions, these genes are required in the MHC class II (MHC II) pathway in DCs for efficient priming of naive CD4⁺ T cells.

Tmem176a/b DKO mice were generated by a dual targeting approach using the CRISPR-Cas9 system as previously described (16). DKO mice used in this study were generated in the C57BL/6N genetic background (C57BL/6NRj, Janvier Labs). Three consecutive backcrosses with C57BL/6N mice were performed before intercrossing heterozygous mice. To control for cage-dependent microbiota variations, wild-type (WT) and DKO mice were systematically cohoused directly after weaning following sex and age matching.

Conditional DKO mice carrying a floxed Tmem176a/b allele (Tmem176a/b^fl) were generated at the Mouse Clinical Institute (Illkirch, France). Briefly, two consecutive rounds of embryonic stem cell (C57BL/6N genetic background, C57BL/6NCrl, Charles River Laboratories) modifications using two independent selection cassettes were realized to insert LoxP sites on both sides of Tmem176a and Tmem176b first coding exons. F₀ mouse chimeras were crossed to a FlpO deleter mouse (17) (pure C57BL/6N background) to remove the FRT- and F3-flanked neomycin and hygromycin selection cassettes abutted to the two LoxP sites. Allele transmission was verified on F₁ mice before rederivation and housing in a specific pathogen-free mouse facility. Tmem176a/b^fl/WT heterozygous mice were crossed to BAC transgenic Rorc(γt)-Cre mice (generated by Eberl and colleagues (18) and provided by Bernhard Ryffel) or CD11c-Cre mice (Itgax-Cre, generated by Reizis and colleagues (19) and provided by Véronique Godot). Following intercrossing, cohoused sex- and age-matched Tmem176a/b^fl/fl homozygous littermates carrying (or not) a transgenic Cre allele were used for experiments.

OT-I.Ly5.1 homozygous mice were obtained by intercrossing OVA-specific TCR-transgenic OT-I mice (C57BL/6-Tg(TcraTcrb)1100Mjb/Crl) (Charles River Laboratories) with Ly5.1 mice (B6.SJL-PtprcaPepcb/BoyCrl) (Charles River Laboratories).

OT-II.Ly5.1.Foxp3EGFP (enhanced GFP) homozygous mice were obtained by intercrossing OVA-specific TCR-transgenic OT-II mice (C57BL/6-Tg(TcraTcrb)425Cbn/Crl) (Charles River) with Ly5.1 mice (B6.SJL-PtprcaPepcb/BoyCrl) (Charles River Laboratories) and Foxp3EGFP reporter mice [generated by Bernard Malissen (20)].

All mice used for experiments were between 8 and 25 wk of age and kept under specific pathogen-free conditions. Experimental procedures were carried out in strict accordance with the protocols approved by the Commitee on the Ethics of Animal Experiments of the Pays de la Loire and authorized by the French Government’s Ministry of Higher Education and Research.

Mice were given 2% dextran sulfate sodium (DSS) (molecular mass of 36,000–50,000 Da, MP Biomedicals) in drinking water ad libitum for 7 d followed by a recovery period without DSS. Mice were monitored and weighed daily.

C. rodentium (DBS100, ATCC 51459) was cultured aerobically at 37°C overnight at 200 rpm in Luria–Bertani (LB) broth medium (MP Biomedicals) and then diluted 1:10 in fresh LB medium until the concentration of bacteria reached an OD of 600 nm. Mice were pretreated with 750 mg/l metronidazole (Sigma-Aldrich) in 2.5% sucrose drinking water for 4 d as previously described (21), followed by 3 d with regular drinking water. Mice were then fasted 8 h before infection by oral gavage with 2 × 10⁹ CFUs of C. rodentium resuspended in sterile 0.9% NaCl. Bacterial concentration was assessed via serial dilution on LB agar plates to confirm the CFUs administered. Mice were monitored and weighed daily. Feces were collected at days 0 and 6 postinfection for detection of C. rodentium by quantitative PCR (qPCR).

Mice were anesthetized with a mixture of 5% xylazine (Rompun) and 18% ketamine in PBS (170 μl) injected i.p. (8.5 mg/kg xylazine and 76.5 mg/kg ketamine per mouse). Square skin grafts (1 cm²) were prepared from the tail of male donors and transplanted on the shaved left flank of female recipients. The grafts were fixed to the graft bed with 10–12 interrupted sutures and were covered with protective tape. Mice were monitored every other day, and graft rejection was defined as complete sloughing or a dry scab.

EG7, MCA101-sOVA (22) (provided by Clotilde Théry), and B16-OVA tumor cells were recovered from log phase in vitro growth and 1 × 10⁶ cells were injected s.c. in 50 μl of cold PBS into the flank skin of recipient mice. Tumor growth was measured in a blinded fashion with a caliper and expressed as the area based on two perpendicular diameters. Mice were monitored daily and were euthanized when tumor size reached 289 mm².

For experimental autoimmune encephalomyelitis (EAE) induced with myelin oligodendrocyte glycoprotein (MOG) peptide, mice were immunized s.c. at the base of the tail and lower flanks with 200 μg of MOG_35–55 peptide (MEVGWYRSPFSRVVHLYRNGK, GenScript) emulsified in CFA (Sigma-Aldrich) supplemented with Mycobacterium tuberculosis H37Ra (Difco Laboratories) at 8 mg/ml (400 μg/ml per mouse). Pertussis toxin (200 ng, Calbiochem) was injected i.p. on the day of immunization and 2 d later.

For EAE induced with MOG protein (23), mice were immunized s.c. at the base of the tail and lower flanks with 500 μg of mMOGTag protein (mouse MOG_1–125 extracellular domain fused to a tag for stability and purification purposes) provided by Steven Kerfoot and emulsified in CFA. Pertussis toxin (250 ng) was injected i.p. on the day of immunization and 2 d later.

Mice were scored daily for EAE clinical signs on a scale of 0–4: 0, no disease; 1, complete limp tail; 2, limp tail with unilateral hindlimb paralysis; 3, bilateral hindlimb paralysis; 4, bilateral hindlimb paralysis and forelimb weakness (end point). The observer was blinded to the genotype during the scoring.

Mice were immunized s.c. at the base of the tail and lower flanks with 50 μg of whole OVA protein (grade V, Sigma-Aldrich) or OVA_323–339 (ISQAVHAAHAEINEAGR) peptide (GenScript) emulsified in CFA (Sigma-Aldrich). After 7 d, mice were challenged with 250 μg of heat-aggregated OVA (2-min incubation at 100°C) injected (20 μl, s.c.) in the right hind footpad whereas the left hind footpad received 250 μg of heat-aggregated BSA (Sigma-Aldrich) as a control for non-specific inflammation. Footpad thickness was measured prior to and 24 and 48 h after injection with an electronic digital micrometer. The observer was blinded to the genotype during the scoring.

Abs and panels used in this study for FACS analysis and cell sorting are listed in Supplemental Table I. RBCs were lysed with ammonium chloride. Small intestine and colon lamina propria cells were prepared as previously described (6). Before all stainings, dead cells were marked for exclusion using fixable viability dye eFluor 506 (eBioscience) or DAPI (Thermo Fisher Scientific) followed by Fc receptor blocking using CD16/32 Ab (BD Biosciences). Intracellular stainings were realized using an eBioscience Foxp3/transcription factor staining buffer set except for MHC II–CLIP peptide staining where cells were fixed using 4% PFA before permeabilization and staining using a 0.1% saponin, 1% BSA solution in PBS. FACS analyses were performed using BD FACSCanto II or a BD LSRFortessa X-20 (BD Biosciences) and FlowJo (Tree Star) software. For mean fluorescence intensity analysis, values were adjusted by subtracting the basal signal from fluorescence minus one staining for each marker. Absolute cell numbers were determined using CytoCount microspheres (Dako, Agilent Technologies). Total and surface expression analysis of MHC II, H2-M, H2-O, Ii (invariant chain, CD74), and the MHC II–CLIP peptide complex in spleen cDC1 and cDC2 were performed with or without a fixation/permeabilization step, respectively, and by gating on B cells (B220⁺CD11c^–), cDC1 (B220^–CD11c⁺CD11b^–CD8α⁺), or cDC2 (B220^–CD11c⁺CD11b⁺CD8α^–). H2-M (αβ2 dimer) staining was revealed using a FITC-conjugated anti-rat IgG1 Ab. Alexa Fluor 647–conjugated anti–H2-Oβ was provided by Denzin and colleagues (24).

Cells were FACS sorted using a BD FACSAria II (BD Biosciences). For verification of Cre-induced loxP recombination, CD11b^–TCRb⁺CD4⁺CD8^– T cells were FACS sorted from the lamina propria of the small intestine of mice treated as previously described (25) with 20 μg of anti-CD3 (145-2C11, provided by J.A. Bluestone), i.p., at days –3 and –1 before sacrifice. CD11c^highMHC II⁺ cDCs were FACS sorted at high purity (>98%) from the spleen of naive mice after enrichment of CD11c⁺ cells using a PE-conjugated anti CD11c Ab and MACS (Anti-PE MicroBeads, Miltenyi Biotec). For in vitro functional analysis, intestinal CD3⁺CD5⁺CD4⁺ T cells and CD3^–CD5^–CD127⁺ ILCs (first gated on CD45^+/lowCD11b/c^–CD19^–CD90⁺ cells) were FACS sorted from the lamina propria of the small intestine and colon (pooled). For spleen cDCs used in epigenetic analysis and in vitro culture, CD11c^highMHC II⁺ cells were purified as described above. For reverse transcription–qPCR analysis, B cells (CD19⁺B220⁺CD11c^–), plasmacytoid DCs (pDCs; CD19^–B220⁺CD11c⁺) cDC1 (CD19^–B220^–CD11c^highCD11b^–CD8α⁺), and cDC2 (CD19^–B220^–CD11c^highCD11b⁺CD8α^–) were FACS sorted from the spleen of WT mice. OVA-specific CD8⁺ T cells were purified from the spleen of OT-I.Ly5.1 mice using a CD8a⁺ T cell isolation kit II (Miltenyi Biotec). OVA-specific CD4⁺EGFP^– cells and T conventional cells were FACS sorted from the spleen of OT-II.Ly5.1.Foxp3EGFP mice after enrichment using a CD4⁺ T cell isolation kit (Miltenyi Biotec).

OVA-specific naive CD8⁺ T and CD4⁺ T conventional cells from OT-I.Ly5.1 and OT-II.Ly5.1.Foxp3EGFP mice, respectively, were labeled with cell proliferation dye (CPD) eFluor 670 (eBioscience) and coinjected (i.v.) at a 1:1 ratio (total of 1–2 × 10⁶ cells per mouse). One day later, recipient mice were administered (i.p.) 100 μg EndoFit OVA protein (InvivoGen). After 3 days, spleens were harvested and proliferation (CPD dilution) of injected cells (CD45.1⁺CD8⁺ or CD45.1⁺CD4⁺) was assessed by flow cytometry.

Bone marrow cells were cultured (0.5 × 10⁶ cells per mL) in the presence of 20 ng/ml GM-CSF (Miltenyi Biotec) in complete RPMI 1640 medium. By day 8, the cells, referred to as BMDCs, were harvested and incubated in 96-well plate (1 × 10⁴ cells per well) with 250 μg/ml EndoFit OVA protein (InvivoGen) or 10 μg/ml OVA_323–339 (ISQAVHAAHAEINEAGR) peptide (GenScript). After 5 h, the cells were washed three times and CPD-labeled CD4⁺Foxp3^– T cells purified from OT-II.Ly5.1.Foxp3EGFP mice were added and proliferation (CPD dilution) was assessed by flow cytometry 3 d later.

Purified spleen cDCs were plated in 96-well plates at 1 × 10⁵ cells and incubated with or without 0.5 μg/ml LPS (Sigma-Aldrich) for 16 h before the quantification of IL-12p40 and IL-6 in the supernatant by ELISA (BD Biosciences). For phenotypic analysis (MHC II, MHC class I [MHC I], CD80, CD86) by flow cytometry, bulk spleen cells (comprising cDC1 and cDC2 identified using the markers CD11c, CD11b, and CD8α as indicated) or BMDCs were stained freshly or stimulated for 6 h with 0.5 μg/ml LPS.

Purified intestinal CD4⁺ T cells and ILCs were cultured in vitro in 96-well plates (10,000 cells per well in triplicates) in complete medium (Life Technologies, Thermo Fisher Scientific) for 18 h in the presence of anti-CD3/CD28 Dynabeads (Thermo Fisher Scientific) at a ratio of 2:1 or IL-23 with and without IL-7, IL-2, and IL-1β (R&D Systems) at 50 ng/ml (except IL-2 at 50 IU/ml), respectively. IL-17A, IL-17F, IL-22 (R&D Systems), and IFNγ (BD Biosciences) were then measured in the supernatant by ELISA.

Ag endocytosis and degradation in cDCs (CD11c⁺MHC II⁺) were assessed by flow cytometry after incubating 1 million bulk splenocytes in 96-well plates with 50 μg/ml OVA-FITC or DQ-OVA (Invitrogen, Thermo Fisher Scientific), respectively. As controls, cells were incubated at 4°C or treated with bafilomycin A (Sigma-Aldrich).

Highly purified spleen cDCs (1 × 10⁶ cells from two pooled mice for each preparation) were resuspended in 40 μl of PBS. Cells were lysed and chromatin was fragmented with 300 U of micrococcal nuclease (M0247S, New England Biolabs) per well for 10 min at 37°C. After full speed centrifugation, supernatants were collected and filled up to 400 μl. Two micrograms of anti–H3K27 acetylation (39133, Active Motif) was used for immunoprecipitation overnight at 4°C. Twenty five microliters of G protein Dynabeads (Invitrogen) were added for rotation for 4 h at 4°C. Beads were then washed twice with 200 μl of wash buffers with increasing salt concentration. ChIP beads were eluted in 50 μl of ChIP elution buffer (50 mM Tris-HCl [pH 7.5], 10 mM EDTA, 1% SDS). ChIP and input samples were digested with 250 μg/ml proteinase K (GEXPRK006R, Eurobio Scientific) in 50 μl of TE buffer for 1 h at 63°C. ChIP DNA was purified using phenol chloroform. Libraries were then prepared as previously described (26). Libraries were verified and equimolar pools were sequenced on a NextSeq 500 (75-bp single-end reads).

Single-end reads were mapped to the mm10 genome by the Burrows–Wheeler alignment algorithm, and reads mapping to non-canonical and mitochondrial chromosomes were also removed. For each sample, ChIP-seq peaks were detected using DFilter at a p value threshold of 1 × 10⁻⁶. All samples passed the quality controls (fraction of reads in peaks >3% and non-redundancy fraction >0.9). A set of consensus peaks was then obtained by taking the union of all peaks and counting reads for these peaks using bedtools. To perform differential peak calling, differentially acetylated (DA) peaks were determined using edgeR after a cpm normalization. DA peaks were defined with a Benjamini–Hochberg q value of ≤5%. For heatmap representation, peaks were rlog transformed. To determine gene ontology enrichment in upregulated peaks, the GREAT tool was used.

qPCR was performed using a ViiA 7 real-time PCR System and Fast SYBR Green master mix reagent (Applied Biosystems, Thermo Fisher Scientific). Primer sequences are listed in Supplemental Table I.

For C. rodentium bacterial load quantification, genomic DNA (gDNA) from feces was purified using a QIAamp DNA stool mini kit (Qiagen), and the EspB gene was detected using specific primers and normalized with total bacterial gDNA (16S gene) using the 2^–ΔΔCt method.

For mRNA quantification, total RNA was isolated using an RNeasy mini kit or micro kit (Qiagen). Reverse transcription was performed using Moloney murine leukemia virus reverse transcriptase and random primers following the manufacturer’s instructions (Thermo Fisher Scientific). Gene-specific primers were designed over different exons to prevent amplification of gDNA. Gene expression was normalized to Gapdh or 18S and expressed in arbitrary units using the 2^–ΔΔCt method.

Retention using selective hooks (RUSH) experiments were realized as previously described by Perez and colleagues (27). Plasmid constructs are listed in Supplemental Table I. When indicated, 250 nM MitoTracker Deep Red or 100 nM LysoTracker Deep Red (Thermo Fisher Scientific) was added 30 min before imaging. HeLa cells (1.5 × 10⁴) were seeded into a μ-Slide eight-well chamber slide (Ibidi) and transfected using Lipofectamine LTX Plus reagent (Thermo Fisher Scientific). After 24 h, the medium was replaced by medium containing 40 μM d-biotin (Sigma-Aldrich). The initial time-lapse acquisition characterizing TMEM176B was performed at Institut Curie (Paris, France) with a thermostat-controlled chamber using an Eclipse 80i microscope (Nikon) equipped with a spinning disk confocal head (PerkinElmer) and an Ultra897 iXon camera (Andor). Subsequent RUSH analyses were performed at MicroPICell facility (Nantes, France) with a confocal microscope (Nikon A1 RSi). Fluorochromes were excited using solid lasers at 488 and 568 nm, respectively, and sequential fluorescence emission was collected on a GaAsP detector. All experiments were performed with a ×60 objective, numerical aperture of 1.4, Plan Fluor, and pinhole (1 Airy unit). All acquisitions were realized at zoom 2 with a scan size of 1064 × 1064, and these parameters induce a pixel dwell of 0.97 µs/pixel, z optical resolution of 0.26 µm, and x-y resolution of 0.11 µm/pixel according to the Nyquist criterion. All time lapses were realized under 5% CO₂ at 37°C, one frame per minute, and generally seven to eight cells were monitored by multi-point acquisition. Under NIS-Elements (V4.51.00), two-dimensional deconvolution with Richardson–Lucy was performed on each time point for all time lapse. Two-dimensional deconvolution parameters were adjusted for each wavelength. Molecule colocalization analysis was performed using pixel-based methods from the Colocalization Studio plugin in the Icy platform (28). The Pearson coefficient correlation was calculated for each time point and for each cell independently by the thresholding K-mean method (29). HeLa-CIITA [generated by Benaroch and colleagues (30)] were used for the analysis of HLA-DR and HLA-DM.

BMDCs were harvested at day 8 of culture, and 1 × 10⁶ cells were plated on cover glasses (Marienfeld) in 12-well plates for 1.5 h in complete medium at 37°C. The cells were then washed with PBS and fixed in 3% paraformaldehyde, 4% sucrose for 15 min at room temperature (RT), followed by three washes with PBS and permeabilization in 0.05% Triton X-100 for 15 min at RT. H2-M was stained using a purified rat Ab detected by a goat anti-rat Ab conjugated to Alexa Fluor 568 (Supplemental Table I). MHC II Ii (CD74) or LAMP1 were then stained with FITC-conjugated specific Abs (Supplemental Table I). Stainings were performed in PBS for 2 h followed by three washes with PBS (5 min each at RT). DAPI (diluted in PBS) staining was performed for 20 min at RT. Slides were mounted with ProLong Gold antifade mountant (Invitrogen). Images were obtained with A1 R Si confocal microscope (Nikon, Champigny sur Marne, France). Quantification of the distribution of fluorescence intensity per cell was performed under Icy (ROI Statistics) (28). The colocalization studio plugin [correlation methods reviewed in Lagache et al. (29)] was used to calculate values of the Pearson correlation coefficient on each automatically thresholded cells. We performed the same thresholding on the previously filtered images with a median filter of radius 2 to study the coefficient of variation (SD divided by the mean of fluorescence).

Statistical analyses were performed using GraphPad Prism (GraphPad Software, La Jolla, CA) using the Mann–Whitney U test or, for survival rates, using the log-rank (Mantel–Cox) test. The p values <0.05 were considered significant.

Tmem176a and Tmem176b are homolog genes encoding structurally similar four-span transmembrane proteins (6, 10, 31) (Fig. 1A). In the immune system, our previous studies (6, 10, 32) combined with the analysis of Immuno-Navigator (33) and ImmGen (34) public databases (Supplemental Fig. 1) indicate that Tmem176a and Tmem176b are tightly coregulated and highly expressed both in cDCs and in the RORγt⁺ cell family (depicted in (Fig. 1B). Taken together, these observations along with reported evidence of genetic compensation in Tmem176b^−/− single KO mice (6) strongly suggested the need to simultaneously target both genes to decipher their function.

As described previously in a methodological study (16), we generated a germline DKO mouse (Fig. 1C) in the pure C57BL/6N genetic background. Homozygous DKO mice were born in a Mendelian ratio and appeared normal, without any significant difference in weight from control WT littermates, suggesting that Tmem176a/b are not absolutely required, or can be sufficiently compensated, during development.

In order to clearly associate a cell type with a potential phenotypic observation resulting from Tmem176a/b deficiency, we also generated a conditional DKO mouse (Tmem176a/b^flox) (Fig. 1D) that we crossed to Rorc(γt)-Cre and CD11c-Cre (Itgax-Cre) mice to target RORγt⁺ and CD11c⁺ cells, respectively. Verification of differential Cre-driven cell-specific deletion was performed by comparing CD4⁺ T cells from the small intestine and splenic cDCs for their expression of Cre, Tmem176a, and Tmem176b mRNA (Fig. 1E, 1F). To increase the fraction of RORγt⁺ Th17 cells in small intestine CD4 (siCD4)⁺ T cells, mice were injected with anti-CD3 as previously described (25). As anticipated, Cre expression was highest in siCD4⁺ T cells from Rorc-Cre mice and in cDCs from CD11c-Cre mice. Importantly, both Tmem176a and Tmem176b transcriptional levels were specifically decreased in those subsets as compared with Cre-negative littermate mice. Although deletion efficiency appeared lower in siCD4⁺ T cells than in cDCs, this could be explained by the yet relatively low fraction of RORγt⁺ cells (61.9 ± 7.9%) within our preparations of purified siCD4⁺ T cells. Finally, it is interesting to note that cDCs from Rorc-Cre mice exhibited a substantial level of Cre expression, a finding in line with a recent study (35) describing a subset of cDC-expressing Rorc, which, however, did not appear sufficient to allow Cre-mediated allele recombination.

Owing to the high expression of Tmem176a/b in RORγt⁺ cells, we focused our attention on the intestinal mucosa where these cells preferentially localize to exert their sentinel function in response to the gut microbiota (36). Flow cytometry analysis (gating strategy depicted in Supplemental Fig. 2A, 2B) of CD4⁺ T cells and ILCs in the small intestine and colon lamina propria revealed no differences in the proportions of Th17 or ILC3 subsets between WT and DKO mice (Fig. 2A). Furthermore, in vitro cytokine production of intestinal CD4⁺ T cells and ILCs was not affected by Tmem176a/b double deficiency (Fig. 2B). Concordant with this, we did not detect any significant change in the level of expression of Rorc2 (the isoform encoding RORγt), Il23r, Il22, Il17a, or Il17f as well as target genes of the IL-22/IL17 axis in the small intestine and colon of DKO mice (Supplemental Fig. 2C).

To test whether Tmem176a/b could play a role in RORγt⁺ cells in the context of inflammatory responses in the gut mucosa, we used the injury-induced self-resolving model of DSS-induced acute colitis in which Th17/ILC3-derived IL-22 is pivotal to restore barrier integrity upon epithelial damage (37, 38). However, both germline DKO and Tmem176a/b^fl/flRorc-Cre^+/− conditional mice exhibited a similar response to DSS when compared with control mice (Fig. 2C), as well as comparable colon length (data not shown) as an indication of inflammation. Next, we used an infectious colitis model using C. rodentium, a mouse attaching and effacing bacterial pathogen considered as an excellent model of clinically important human gastrointestinal pathogens and against which ILC3s and Th17 have been shown to play a protective and redundant role (11, 39, 40). As shown in (Fig. 2D, Tmem176a/b deficiency did not hamper the mouse resistance to this infection. Moreover, we did not observe a difference in colon length between WT and Tmem176a/b-deficient mice (data not shown). Finally, although fecal C. rodentium bacterial loads detected by qPCR appeared increased in RORγt-specific conditional DKO mice, this difference did not reach statistical significance and was not found in germline DKO mice compared with respective control mice (Fig. 2E).

Thus, these results suggest that Tmem176a/b are not critical for the development and protective functions of ILC3s and Th17 in the intestinal mucosa.

Next, we investigated the effect of Tmem176a/b double deficiency in the biology of cDCs. Because both genes are highly expressed in most hematopoietic precursors, including cDC progenitors (Fig. 1B, Supplemental Fig. 1), we assessed their frequency in the bone marrow but found no alteration in DKO mice (Fig. 3A, Supplemental Fig. 3). Accordingly, the percentages and absolute numbers of CD11c^highMHC II⁺ cells in the spleen and, within this population, the proportions of the two main cDC subsets, namely cDC1 (CD11b^–CD8α⁺) and cDC2 (CD11b⁺CD8α^–), were similar between WT and DKO mice (Fig. 3B). Furthermore, MHC I, MHC II, CD80 (B7-1), and CD86 (B7-2) surface expression levels on cDC1 and cDC2 were not affected by Tmem176a/b deficiency, and LPS stimulation elicited an equally strong upregulation of MHC II, CD80, and CD86 at the surface of both WT and DKO cDC subsets (Fig. 3C). Finally, purified spleen cDCs from WT and DKO mice produced similar basal levels of IL-12 and IL-6 that were both increased with the addition of LPS (Fig. 3D).

In the absence of obvious developmental abnormalities of cDCs, we sought to identify dysregulated cellular pathways resulting from the absence of Tmem176a/b. Chromatin dynamics reflect with great sensitivity gene regulation and they play important roles in immune functions such as trained innate immunity (41) or T cell effector/memory differentiation (42). To investigate whether epigenetic alterations are present in DKO cDCs, we FACS sorted CD11c^highMHC II⁺ cells and performed ChIP by targeting lysine H3K27 acetylation followed by ChIP-seq to detect active promoters and enhancers (epigenomics road map). We found that 962 enhancers were significantly DA between WT and DKO cDCs (Fig. 3E, 3F). Interestingly, one of the most significantly DA gene regulatory region covered the entire Cd74 gene (Fig. 3G), and MHC gene ontology analysis included Ag processing and MHC II protein molecules (Fig. 3H). Taken together, these data suggest that Tmem176a/b deficiency could impact intracellular processes of the MHC II pathway and lead to selective genetic adaptations. Further supporting a role of Tmem176a/b in the MHC II pathway, we found that both homologs were clearly overexpressed in cDC2 when compared with B cells, pDCs, and cDC1 (Fig. 3I), consistent with the propensity of this subset in priming naive CD4⁺ T cells through MHC II Ag presentation (43).

To determine whether Tmem176a/b are required in DCs for Ag presentation in vivo via MHC I or MHC II molecules, we set out to test different models of dominant CD8⁺ or CD4⁺ T cell responses.

First, using a minor histocompatibility transplantation model in which male skin is grafted onto female recipients, we found that WT and DKO mice exhibited the same rate of graft rejection (Fig. 4A), suggesting that CD8⁺ T cell priming and effector functions were not impaired in the absence of Tmem176a/b. Next, we examined whether anti-tumor immune responses could be influenced by Tmem176a/b deficiency. Notably, we hypothesized that DKO mice could exhibit enhanced anti-tumor immunity owing to a recent study proposing that targeting Tmem176b could improve the anti-tumor CD8⁺ T cell response by de-repressing inflammasome activation in myeloid cells (44). Subcutaneous injection of OVA-expressing EG7 thymoma led to detectable tumors within 2 weeks in all mice and, consistently with the relatively immunogenic nature of this cell line in our experimental conditions, tumor regression was then observed in a large fraction of the mice but without a significant difference between WT and DKO mice (Fig. 4B). Furthermore, Tmem176a/b deficiency did not impact tumor growth and mouse survival with two aggressive tumor cell lines, that is, MCA101-sOVA fibrosarcoma and B16-OVA melanoma (Fig. 4B). Thus, our results show that the absence of Tmem176a and Tmem176b does not prevent or enhance CD8⁺ T cell responses. To evaluate CD4⁺ T cell responses in vivo, we first used the model of EAE induced by MOG_35–55 peptide or MOG_1–125 protein immunization (Fig. 4C). Interestingly, whereas EAE developed similarly between WT and DKO mice with MOG peptide immunization (Fig. 4D), DKO mice appeared less susceptible when the MOG protein was used (Fig. 4E), thus pointing to a specific role of Tmem176a/b in Ag processing before peptide–MHC II complex display at the surface. Because of the high variability of this model, we next turned to a model of delayed-type hypersensitivity (DTH) (Fig. 4F). Again, we observed that DKO mice differed from WT mice for the DTH response upon challenge in the footpad only when the whole protein was used for the immunization (Fig. 4G, 4H). Taken together, these data suggest that Tmem176a/b deficiency selectively affects the intracellular processing of exogenous Ags for naive CD4⁺ T cell priming.

To further test this hypothesis, we directly assessed Ag-specific T cell proliferation in WT and DKO mice using naive OVA-specific CD8⁺ and CD4⁺ T cells (Fig. 5A). Importantly, note that cDCs are strictly required in this system to induce activation and proliferation of naive T cells (45). As shown in (Fig. 5B and (5C, CD4⁺ T cell proliferation was markedly diminished in DKO mice in comparison with WT mice, whereas CD8⁺ proliferation was not. Importantly, this alteration was not observed in Tmem176a/b^fl/flRorc-Cre^+/− conditional mice (Fig. 5D) but was replicated, although to a lesser extent than in germline DKO mice, in Tmem176a/b^fl/flCD11c-Cre^+/− mice (Fig. 5E).

In vitro Ag-specific CD4⁺ T cell proliferation was also significantly decreased in the absence of Tmem176a/b using GM-CSF–induced BMDCs only when OVA protein was used (Fig. 5F, 5G), whereas surface expression of MHC II and costimulatory molecules remained unaltered (data not shown).

In conclusion, these results show that Tmem176a/b have an intrinsic function in DCs to allow efficient presentation of exogenous Ags onto MHC II molecules and priming of naive CD4⁺ T cells.

Ag presentation by MHC II molecules is achieved through a series of complex events (depicted in (Fig. 6A) beginning with the uptake and mild degradation of exogenous Ags and including finely regulated processes in the specialized MHC II compartment (MIIC) (46).

Both OVA endocytosis and degradation were similar in WT and DKO cDCs (Fig. 6B, Supplemental Fig. 4), indicating that altered MHC II–mediated Ag presentation by Tmem176a/b DCs cannot be explained by a defect in the initial steps of Ag processing.

To determine whether Tmem176a/b deficiency could influence the surface or intracellular levels of key players in the MHC II pathway, we analyzed the expression of MHC II (I-A^b), H2-M, and H2-O in splenic B cells, cDC1, and cDC2 by flow cytometry (Fig. 6C, Supplemental Fig. 4). As expected, cDC1 exhibited the highest levels of intracellular MHC II, Ii, and H2-O, whereas H2-M was primarily expressed in cDC2, an equilibrium concordant with the intrinsic efficiency of this subset in MHC II processing (43). Although we did not detect aberrant expression of these molecules at the surface of DKO cells, we found that intracellular H2-M was paradoxically and selectively overexpressed in the cDC2 subset of DKO mice compared with WT mice. Additionally, while Ii (invariant chain) expression was unaltered, the intracellular detection of the MHC II–CLIP peptide complex was not increased at the cell surface but was diminished intracellularly in cDC2 of DKO mice (Fig. 6C, lower panels).

Thus, while we expected upregulation of MHC II, H2-M, and Ii (CD74) in DKO cells from our epigenetic analysis (Fig. 3F), this was only the case for H2-M, showing only limited concordance between histone acetylation changes (epigenomic state) and protein levels (proteomic state). However, these data suggest that Tmem176a/b-mediated function is directly involved in the MIIC for optimal MHC II Ag loading or trafficking.

To gain insight into the intracellular function of TMEM176A/B ion channels, we aimed to elucidate their subcellular localization, which has remained elusive, as different studies have reached different conclusions (5, 6, 47). To this end, we used the RUSH system (27), a two-state assay allowing fluorescence-based analysis of intracellular trafficking in living cells at physiological temperature (Fig. 7A). We fused TMEM176A and TMEM176B to EGFP (or mCherry) and streptavidin-binding protein to allow their retention in the endoplasmic reticulum (ER) in the presence of an isoform of the invariant chain (48) fused to the core streptavidin, acting as a hook. Biotin addition can then induce synchronous release and intracellular trafficking analysis. We performed dual-color imaging using multiple organelle-specific proteins or probes (Fig. 7B) to track the intracellular fate of TMEM176A/B from the ER. We used HeLa cells as they allow higher resolution of intracellular compartments compared with immune cells. Addition of biotin triggered a rapid change in the TMEM176B signal from a network of tubular elements characteristic of the ER to a pattern reminiscent of the Golgi apparatus but that rapidly evolved into multiple dynamic vesicles (Fig. 7C, Supplemental Video 1). TMEM176A and TMEM176B exhibited a very similar intracellular dynamic, as measured by strong colocalization throughout the time of acquisition (Fig. 7D, Supplemental Video 2).

Confirming our previous hypothesis that TMEM176B traffics through but does not accumulate in the Golgi apparatus (6), it clearly separated from Golgin-84, a Golgi-resident protein. After 10–15 min of incubation with biotin, Golgin-84 reached the Golgi apparatus where it stayed after longer incubation time in contrast to TMEM176B (Fig. 7E, 7F, Supplemental Video 3). We then examined a variety of markers depicted in (Fig. 7G. We did not observe, or only marginally observed, accumulation at the plasma membrane, a result that was confirmed by co-imaging with CD44. Interestingly, TMEM176B-bearing vesicles could interact with RAB5⁺ early endosomes as well as with RAB7⁺ late endosomes. We generated a LAMP1 (CD107a) RUSH construct to best reveal endolysosomes and also observed a strong association with TMEM176B trafficking during the post-Golgi timeframe. Of interest, TMEM176B could be found colocalized with TNF or TfR (transferrin receptor) during the late events of endocytosis/recycling of these proteins. However, TMEM176B was not associated with RAB4, a marker of recycling endosomes. Although we cannot rule out alternate recycling pathways, these results suggest that TMEM176A/B follow a relatively selective route among the various vesicular compartments of the cell. In this line, TMEM176B did not traffic through mitochondria labeled by the MitoTracker probe but was significantly associated with LysoTracker that preferentially marks acidic vesicles. Moreover, we found some colocalization signals using a cation-dependent mannose-6-phosphate receptor RUSH construct but even stronger association with the monomeric clathrin adaptor GGA1, known to decorate the carrier vesicles budding from the trans-Golgi network and merging toward the endosomes (49).

Taken together, these data show that TMEM176A/B ion channels preferentially localize in the late endosomal compartment and in vesicular vesicles between the Golgi and the endolysosomal system.

Given the requirement of Tmem176a/b in MHC II Ag presentation, the selective alteration of H2-M expression in cDC2 and the preferential trafficking of TMEM176A/B in the late endosomal compartment, we asked whether these ion channels could localize in the MIIC. To recapitulate the MHC II pathway in HeLa cells, we used the HeLa-CIITA cell line that stably expresses the transactivator CIITA (30). To reveal MHC II (HLA-DR) and HLA-DM localization in these cells, we used plasmid constructs expressing yellow fluorescent protein (YFP) tagged to the β- or α-chains of each molecule, respectively, as described in Zwart et al. (50). Colocalized signals were detected between TMEM176B and HLA-DR, mostly in intracellular vesicles (Fig. 7H). However, a more pronounced association was observed with HLA-DM in intracellular compartments presumably highlighting the MIIC.

Thus, these results strongly suggest that TMEM176A/B exert their function directly in the MIIC to contribute to efficient MHC II peptide loading and/or trafficking.

To determine whether the absence of TMEM176A and TMEM176B genes could impact the MIIC, we examined the intracellular localization of MHC II, H2-M, Ii (CD74), and LAMP1 in WT and DKO BMDCs. Interestingly, LAMP1 showed a more homogeneous distribution in DKO compare with WT cells, as measured by a decreased coefficient of variation, whereas MHC II and H2-M stainings were not affected, and Ii appeared only weakly altered (Fig. 8A, 8B). As expected, H2-M exhibited a relatively high colocalization signal with MHC II (Fig. 8C) and a lower association with Ii (Fig. 8D), but without any significant differences between WT and DKO cells. In contrast, we found that the colocalization between H2-M and LAMP1 was significantly low in most DKO cells compared with WT cells (Fig. 8E).

These results suggest that TMEM176A/B could play a role in the intracellular dynamic of the endolysosomal pathway, and more specifically in DCs, in the organization of the MIIC for optimal Ag processing and MCH II–mediated presentation.

Finely tuned ion influx and efflux result from various intricate interplays of multiple channels likely tailored for each type of cell and maturation status. The intriguing high expression of TMEM176A and TMEM176B cation channels in both RORγt⁺ lymphoid cells and DCs logically raises the question of their specific role in these two very distinct immune cell types. Based on expression and functional data, we reasoned that each gene has the potential to compensate for each other and that simultaneous targeting would be requisite to avoid such redundancy.

We present here, to our knowledge for the first time, a functional study of Tmem176a/b DKO mice, either germline or conditional. Given the broad tissue expression of Tmem176a/b, the floxed conditional mouse represents an invaluable tool to achieve Cre-mediated cell-specific deletion and document the role of these ion channels in virtually any tissue or cell of interest. Indeed, there is a growing interest in understanding the role of these homolog genes that are overexpressed in a wide range of cell types other than immune cells, including fibroblast subsets (51), neurons (52), adipocytes (53) or tumor cells (54–56), likely adapting a universal mechanism of intracellular ion flux regulation to their specific needs. Importantly, these Tmem176a/b DKO mice were generated directly in a pure genetic background (C57BL/6N), thus avoiding incorrect interpretations resulting from carryover of gene variants of a different background surrounding the targeted locus (57, 58).

The striking expression of Tmem176a/b in all type 3 immune cells and their dependency on the master transcription factor RORγt (15) make these homologs promising candidates to uncover novel aspects of RORγt⁺ cell biology beyond their cytokine production and could represent a novel therapeutic entry point for treating immune-mediated diseases. Surprisingly, our results indicate that RORγt-dependent intestinal repair and host defense functions are not critically compromised in the absence of Tmem176a/b. However, we cannot exclude that a more detailed analysis of the normal and inflamed intestinal mucosa in DKO mice could unveil a specific role of Tmem176a/b. Furthermore, the normal development MOG_35–55 peptide–induced EAE suggest that these genes are not required for the pathogenicity of Th17 in this model. Investigating the transcriptomic and epigenetic profiles of purified DKO ILC3s or Th17 cells may expose compensatory mechanisms notably involving the regulation of other ion channels that could sufficiently counterbalance the absence of Tmem176a/b. It is also tempting to speculate that these homologs could be required for IL-17/IL-22–independent functions in RORγt⁺ cells, including the regulation of anti-commensal effector CD4⁺ T cells by ILC3s through MHC II–mediated inhibitory presentation (59, 60). Although Tmem176a/b^fl/flRorc-Cre^+/− conditional mice did not exhibit increased CD4⁺ T cell activation and proliferation or neutrophil accumulation in the colonic lamina propria in comparison with control mice (data not shown), the fact that ILC3s selectively share with DCs the expression of MHC II molecules is in favor of a pivotal role of Tmem176a/b in this adaptive function.

In view of the recent study by Segovia et al. (44) using Tmem176b single KO mice or an ion channel inhibitory molecule, our results do not support the hypothesis that inhibiting TMEM176A/B-mediated ion flux could enhance the anti-tumor CD8⁺ T cell response in vivo. Moreover, we did not observe increased IL-1β production by Tmem176a/b-deficient BMDCs (data not shown). Although different experimental conditions could explain this discrepancy, one can speculate that TMEM176B must be targeted alone, leaving TMEM176A function intact, to obtain such phenotype, and that the effect of the inhibitory molecule is therefore achieved in a selective manner. Thus, TMEM176A and TMEM176B could exert both similar and different functions potentially depending on the formation of homodimers or heterodimers leading to the formation of three different identities. This could explain why single (5, 44) and double KO (present study) models would show different biological outcomes. Alternatively, because of the 129 genetic background origin of the Tmem176b single KO mouse (5), confounding genetic factors could be invoked, despite >10 backcrosses onto the C57BL/6 background and the use of littermates (61).

We initially reported that Tmem176a/b were highly expressed in cDCs but not in pDCs (10, 32), likely a consequence of E2-2–mediated repression as revealed by Ghosh et al. (62) for Tmem176a. Remarkably, recent mouse and human single-cell RNA sequencing analysis highlighted these homologs as markers of selective DC subsets, both in mice and humans (63–65). Notably, the association of TMEM176B expression with a subset of cDC2 in Binnewies et al. (65) is concordant with our data in the mouse showing that Tmem176a/b are markedly overexpressed in cDC2 compared with cDC1. cDC2 exhibit an overall dominance in MHC II presentation in vivo resulting from the combination of their intrinsic efficiency (44, 66) and their favorable position within lymphoid tissues for Ag uptake (67). Consistently, we found that Tmem176a/b deficiency selectively affected the capacity of DCs to prime naive CD4⁺ T cells but not CD8⁺ T cells in vivo. Our results point to a defect in the intracellular processing events for exogenous Ag presentation to MHC II molecules. However, this functional alteration is not complete and it is possible that, in the same manner as in ILC3s, compensatory mechanisms develop in DCs in the absence of Tmem176a/b. In support of this hypothesis, the fact that H2-M was found selectively overexpressed in cDC2 of DKO mice suggests an adaptation to alleviate a defect in the MIIC for peptide loading onto MHC II molecules. Alternatively, this expression may also reflect an incorrect intracellular localization of H2-M, thereby disrupting the optimal processes leading to MHC II presentation. Likewise, the reduced levels of the intracellular MHC II/CLIP complex in cDC2 from Tmem176a/b-deficient mice might suggest a general alteration of the MHC II pathway.

The analysis of TMEM176A/B intracellular dynamics enabled us to clearly delineate that they preferentially traffic in the late endolysosomal system in close relationship with the Golgi apparatus. In contrast with the limitations in sensitivity and the non-dynamic nature of classical immunostaining, the RUSH system was instrumental in revealing the route taken by TMEM176A/B from the ER and beyond the Golgi apparatus. However, because we focused on the first hour of trafficking after release in most of our analyses, we cannot exclude that TMEM176A/B can eventually reach other compartments over time. Importantly, it is also noteworthy that this approach using fluorescent protein tagging could lead to erroneous subcellular targeting and cannot completely substitute for an analysis of the intact endogenous proteins.

The strong colocalization found with HLA-DM in HeLa-CIITA cells supports the hypothesis of a direct role in the MIIC. TMEM176A/B-mediated cation (Na⁺) efflux could participate in the regulated acidification of this compartment as a counterion conductance (68). TMEM176A/B could be located on the limiting membrane of the MIIC or on intraluminal vesicles where a direct interaction with HLA-DM would be possible. In this regard, independently of their ion channel function, it is conceivable that these four-span transmembrane proteins act similarly to tetraspanin molecules to stabilize DM-MHC II interaction (69). On the same note, given the reported genetic association between TMEM176A and high-density lipoprotein cholesterol levels in humans (70), the TMEM176A/B function may be connected to cholesterol-containing microdomains for efficient MHC II trafficking (71). High-resolution imaging, fluorescence resonance energy transfer analysis, or the characterization of organelle-specific disruption in DKO DCs could be informative to uncover the precise role of TMEM176A/B in the MHC II pathway. In this regard, our finding showing altered colocalization between H2-M and LAMP1 in DKO BMDCs argues for a potential role of TMEM176A/B in the organization of the endolysosomal system and, more specifically, the MIIC in DCs. However, to definitely demonstrate the role of TMEM176A/B in MHC II Ag presentation, it will be important to assess specific peptide loading onto MHC II molecules in DKO DCs, for example using the YAe Ab to detect surface Eα peptide/I-A^b complexes following Eα-GFP fusion protein uptake (72). Finally, it will be interesting to examine the fate of TMEM176A/B shortly upon phagocytosis or DC stimulation. One can hypothesize that these molecules could also relocate to the plasma membrane, even transiently, to achieve their function(s).

In conclusion, while the intrinsic function of TMEM176A/B in RORγt⁺ cells remains to be further explored, we found that these cation channels play a substantial role in the MHC II pathway to ensure optimal naive CD4⁺ T cell priming by DCs. Remarkably, a recent study by Rudensky and colleagues (35) identified these genes by single-cell RNA sequencing as markers of the T-bet^– cDC2B subset both in mice and humans, a finding that reinforces the hypothesis of a predominant role of TMEM176A/B in cDC2s. Taken together, these results also suggest that the generation of a Tmem176a/b reporter mouse could represent an invaluable tool to study cDC2 subsets in vivo.

We are grateful to Philippe Hulin and Steven Nedellec from the MicroPICell imagery core facility (Nantes, France) for excellent assistance with confocal microscopy and to Claire Usal, Pierre Pajot, and Jean-Marc Merieau for mouse housing and experimental help.

The authors have no financial conflicts of interest.