The ERM Protein Moesin Regulates CD8⁺ Regulatory T Cell Homeostasis and Self-Tolerance

This article is featured in In This Issue, p.3385

The ezrin–radixin–moesin (ERM) proteins are a family of widely distributed membrane-associated proteins that link plasma membrane proteins with actin filaments in the cell cortex, thereby regulating many fundamental cellular processes, including cell-shape determination, membrane-protein localization, membrane transport, and signal transduction (1–3). Because of their high sequence similarity, ERM proteins are generally thought to be functionally redundant. Lymphocytes predominantly express two ERM proteins, ezrin and moesin (4). Although their functions are mostly redundant during TCR- and BCR-mediated lymphocyte activation (5, 6), we previously found that moesin gene–inactivated mice exhibit lymphopenia in the peripheral blood and lymph nodes (LNs), indicating that moesin has a unique role in lymphocyte homeostasis (7). Recently, whole-exome sequencing of DNA from primary immunodeficiency patients revealed the presence of hemizygous mutations in the moesin gene, which is located on the X chromosome (8). Seven patients with early-onset and persistent lymphopenia were found to harbor one of two different mutations, both of which were predicted to be disease-causing mutations. An additional case with the same mutation was identified by recent newborn screening for primary immunodeficiency (9). Although moesin is ubiquitously expressed in cells of hematopoietic origin (7), in humans and mice with moesin mutations, naive T cells are the most markedly affected lymphocyte subset, which exhibits particularly low numbers of naive CD8⁺ T cells.

T cell lymphopenia is often associated with autoimmunity (10–12). Many patients with lymphopenia-associated primary immunodeficiency, such as Wiskott–Aldrich syndrome and Omenn syndrome, develop autoimmunity (13, 14). A number of mechanisms have been proposed to link lymphopenia and autoimmunity. For example, homeostatic proliferation of lymphocytes in response to lymphopenia is one mechanism that may contribute to autoimmunity, because homeostatic proliferation favors the expansion of autoreactive T cells. Another possible mechanism involves the induction of defective regulatory T (Treg) cells (11).

Treg cells play crucial roles in controlling the balance between effective immune responses and tolerance by regulating the expansion and function of effector T cells. Among effector T cells, follicular helper T (Tfh) cells are a distinct subset of CD4⁺ T cells that provide cognate help to B cells during germinal center (GC) reactions (15, 16). Upon exposure to a foreign Ag, naive CD4⁺ T cells differentiate into Tfh cells, which help B cells differentiate into Ab-producing plasma cells and long-lived memory B cells. Although Tfh cells are required for Ab responses against foreign Ags, their unrestrained accumulation results in the production of autoantibodies, which leads to the development of autoimmune diseases, such as systemic lupus erythematosus (SLE). CD4⁺ and CD8⁺ Treg cells have been identified as negative regulators of GC reactions. CD4⁺ Treg cells expressing Foxp3 are a well-characterized Treg subset that is critical for the maintenance of immune homeostasis (17). A subpopulation of CD4⁺ Treg cells that coexpresses Bcl6 and CXCR5 has been identified as follicular regulatory T (Tfr) cells in humans and mice. These cells limit the expansion of Tfh cells and inhibit GC reactions (18). CD8⁺ Treg cells are characterized by the surface expression of CD44, CD122, and Ly49 in mice, and they regulate the activity of Tfh cells through recognition of the unconventional MHC class I molecule Qa-1 expressed on the Tfh cell surface (19, 20). In humans, HLA-E–restricted CD8⁺ T cells function similarly to murine Qa-1–restricted CD8⁺ Treg cells in maintaining self-tolerance (21). Defective Treg cell homeostasis leads to excessive Tfh cell accumulation and autoimmunity. In a mouse model of Wiskott–Aldrich syndrome, decreased numbers and impaired function of CD4⁺ Treg cells may contribute, in part, to the development of autoimmunity (22). Defective CD8⁺ Treg cell activity also leads to the development of SLE-like autoimmunity (19, 20).

The maintenance of the peripheral T cell pool is regulated by complex homeostatic mechanisms, including TCR signaling from contact with MHC and signaling by the common γ-chain (γc) family of cytokines, such as IL-2, IL-7, and IL-15 (23). Homeostasis of CD4⁺ Treg cells is regulated by contact with IL-2. IL-2 signals are crucial during CD4⁺ Treg cell development in the thymus, as well as homeostatic proliferation and survival in peripheral LNs. In contrast, CD8⁺ Treg cells depend on IL-15 for their proliferation and survival (24), and IL-15–deficient mice exhibit defective CD8⁺ Treg cell function (20).

In this study, we report that moesin deficiency led to an SLE-like autoimmune disease in aging mice. Tfh cells and GC B cells spontaneously accumulated in the spleen and LNs of moesin-deficient mice, leading to the production of anti-dsDNA Abs and the deposition of complement C3 and IgG in the kidneys. In addition, CD8⁺ Treg cells, which inhibit the expansion of Tfh cells, were severely reduced in these mice. Notably, we found that an ex vivo culture of CD8⁺ Treg cells from moesin-deficient mice exhibited a significant attenuation of IL-15–induced proliferation, which was accompanied by defects in STAT5 activation and IL-15Rα internalization. Collectively, our results reveal a novel role for moesin in the regulation of IL-15–dependent CD8⁺ Treg cell homeostasis and the maintenance of self-tolerance.

Moesin-deficient mice were produced as described previously (25) and backcrossed for >10 generations onto the C57BL/6 genetic background. Male moesin-deficient (Msn^−/Y) mice and littermate wild-type (Msn^+/Y) mice were used for most experiments. In some experiments, female heterozygous (Msn^+/−) mice were used. The mice were housed at the Research Center for Animal Life Science at Shiga University of Medical Science. All studies were approved by the Animal Research Committee of Shiga University of Medical Science.

Lymphocytes were isolated from spleens and LNs by mechanical disruption between the frosted surfaces of two glass slides, followed by filtration through 100-μm nylon mesh twice. The collected cells were quantified with a hemocytometer. For proliferation assays, CD4⁺ Treg (CD4⁺CD25⁺) and CD8⁺ Treg (CD8⁺CD44⁺CD122⁺Ly49⁺) cells were sorted using a FACSAria (BD Biosciences) and cultured with plate-bound anti-CD3 (5 μg/ml, 145-2C11) and soluble anti-CD28 (1 μg/ml, 37.51; both from BioLegend) in complete RPMI 1640 for 9 d, with IL-2 (100 ng/ml; R&D Systems) or IL-15 (100 ng/ml; Miltenyi Biotec). The cells were counted with a hemocytometer on the indicated days.

For IL-15Rα–internalization and STAT5-activation assays, isolated splenocytes were starved in serum-free RPMI 1640 for 2 h and then stimulated with IL-15 (100 ng/ml) or IL-2 (100 ng/ml) for the indicated times, and the cell surface IL-15Rα or p-STAT5 levels were determined by flow cytometry. For IL-15Rα staining for immunofluorescence microscopy, sorted CD8⁺ Treg cells were starved in serum-free RPMI 1640 for 1 h. The cells were then stimulated with IL-15 (100 ng/ml) for the indicated times.

The mAbs used for flow cytometric analyses were purchased from BD Biosciences, eBioscience, or BioLegend and included those to CD4 (RM4-5), CD8 (53-6.7), CD21 (7E9), CD23 (B3B4), CD25 (PC61.5), CD44 (IM7), CD45RB (C363.16A), CD45R/B220 (RA3-6B2), CD93 (AA4.1), CD95/Fas (Jo2), CD122 (5H4), CD132 (TUGm2), Bcl6 (BCL-DWN), CXCR5 (2G8), Foxp3 (FJK-16s), GL-7 (GL-7), Ly49 (14B11), ICOS (C398.4A), IgD (11-26c.2a), IgM (RMM-1), IL-15Rα (DNT15Ra), PD-1 (RMP1-30), and p-STAT5 (SRBCZX). Single-cell suspensions were incubated with anti-CD16/CD32 for 10 min, followed by staining with mAbs for 30 min on ice and washing. Data were acquired on a FACSCalibur or FACSCanto II (both from BD Biosciences) and analyzed using FlowJo (TreeStar). For CXCR5 staining, biotinylated anti-CXCR5 was detected with allophycocyanin-labeled streptavidin. For IL-15Rα staining, the cells were incubated with a PE-labeled anti–IL-15Rα mAb and then stained with biotinylated anti-rat IgG (Jackson ImmunoResearch), followed by PE-labeled streptavidin. For intracellular staining, the cells were fixed and permeabilized using a Foxp3/Transcription Factor Staining Buffer Kit (eBioscience) for Bcl6 and Foxp3 and a BD Cytofix/Cytoperm Fixation/Permeabilization Kit for moesin as described previously (26). For p-STAT5 staining, the cells were fixed with 1.5% formaldehyde and permeabilized with methanol.

Paraffin-embedded kidney sections were stained with H&E. For periodic acid–Schiff (PAS) staining, paraffin-embedded sections were stained with 0.5% periodic acid and Schiff solutions, and the nuclei were stained with Mayer’s hematoxylin. Images of H&E- and PAS-stained kidney sections were captured with an Eclipse 90i microscope (Nikon). For C3 and IgG staining, frozen kidney sections were fixed with ice-cold methanol and acetone, respectively, and stained with FITC-labeled anti-mouse C3 (MP Biomedicals) and Alexa Fluor 555–labeled anti-mouse IgG (BioLegend). For staining of spleen sections, the frozen sections were fixed with acetone and stained with biotinylated anti-CD90.2/Thy1.2 (eBioscience), followed by staining with Alexa Fluor 594–labeled streptavidin (Invitrogen) and Alexa Fluor 488–labeled anti-CD45R/B220 (BD Biosciences).

For sorted CD8⁺ Treg cell staining, surface and intracellular IL-15Rα was stained using a successive staining protocol, as described previously (27). IL-15–stimulated CD8⁺ Treg cells were fixed with 4% paraformaldehyde and surface stained with goat Abs against IL-15Rα (R&D Systems) and Alexa Fluor 488–labeled anti-goat IgG (Invitrogen). The cells were then permeabilized with 0.25% Triton-X in PBS and stained with the same goat anti–IL-15Rα Abs, followed by biotinylated anti-goat IgG (Abcam) and Alexa Fluor 555–labeled streptavidin (Invitrogen). Stained cells were cytospinned onto glass slides at 800 rpm for 5 min. The fluorescence microscopy images were captured with an FV-1000D IX81 confocal microscope (Olympus). For the ratio of surface/intracellular IL-15Rα, green and red fluorescence intensities were measured for each cell in captured images using the ImageJ Area measurement tool (National Institutes of Health) (28, 29). The ratio was calculated by dividing the green fluorescence intensity by the red fluorescence intensity.

Total RNA was extracted using TRIzol Reagent (Invitrogen) and reverse transcribed with a High Capacity RNA-to-cDNA Kit (Applied Biosystems). Quantitative real-time PCR was performed using KOD SYBR qPCR Mix (Toyobo) and a LightCycler 480 instrument (Roche). The primer pairs are 5′-CGCCAGATGCAAGTGTTGTAT-3′ and 5′-TCCTGGGGATTATCCAAGTCAAT-3′ for STAT5a, 5′-CGATGCCCTTCACCAGATG-3′ and 5′-AGCTGGGTGGCCTTAATGTTC-3′ for STAT5b, 5′-CCCACAGTTCCAAAATGACGA-3′ and 5′-GCTGCCTTGATTTGATGTACCAG-3′ for IL-15Rα, 5′-TGGAGCCTGTCCCTCTACG-3′ and 5′-TCCACATGCAAGAGACATTGG-3′ for CD122, and 5′-GCAACAGAGATCGAAGCTGGA-3′ and 5′-AGATTGGGTTATAGCGGCTCC-3′ for CD132.

Basal serum titers of the Ig isotypes were analyzed by ELISA. The ELISA plates were coated with goat anti-mouse isotype Abs (Bethyl Laboratories) and blocked with 1% BSA in PBS, followed by the addition of serially diluted serum samples and standards. The plates were incubated and washed, and HRP-conjugated goat anti-mouse isotype Abs (Bethyl Laboratories) were added to the wells. After further incubation and washing, 3,3′,5,5′-tetramethylbenzidine was added as a substrate. The reactions were stopped with 2 M H₂SO₄.

For affinity-maturation assays, nitrophenyl (NP)-coupled chicken γ-globulin (CGG) (NP:CGG conjugation ratio, 39:1; NP₃₉-CGG; Biosource Technologies) was used as an immunogen. The mice were immunized i.p. with 100 μg of NP₃₉-CGG precipitated in alum (Imject Alum; Thermo Fisher Scientific) and boosted with 50 μg of NP₃₉-CGG in PBS on day 42. Sera were collected on days 7, 14, 28, 42, 49, and 63 after the first immunization, and the anti-NP IgM and IgG1 serum titers were measured by ELISA. The plates were coated with NP₅₂-BSA or NP₄-BSA (both from Biosource Technologies) and blocked with 1% BSA in PBS and then serially diluted serum samples were added. The bound Abs were detected with HRP-conjugated goat anti-mouse IgM or IgG1. The plates were developed as described above.

Serum anti-dsDNA Ab titers were measured with a mouse anti-dsDNA ELISA kit (Shibayagi), according to the manufacturer’s instructions.

Statistical analysis was performed using the two-tailed Student t test. To determine survival rates, log-rank analysis was performed.

Although Msn^−/Y mice exhibit T and B cell lymphopenia in the peripheral blood, young Msn^−/Y mice appear healthy (7). In this study, we found that after 16 wk of age, several Msn^−/Y animals became lethargic and died. By 80 wk of age, 28% of Msn^−/Y mice had died, whereas all of the littermate control Msn^+/Y mice survived, indicating that Msn^−/Y mice exhibited a higher mortality than the controls (Fig. 1A). Because lymphopenia is often associated with autoimmunity, we examined the development of autoimmune diseases in aging Msn^−/Y mice. The analysis of sera from 8- to 72 wk-old mice showed that while the anti-dsDNA Ab titers were comparable between Msn^+/Y and Msn^−/Y mice at 8 wk of age, they were significantly elevated in 72 wk-old Msn^−/Y mice (Fig. 1B). High titers of anti-dsDNA Abs are suggestive of autoimmune disease, particularly SLE (30). SLE is a chronic autoimmune disease that is characterized by the generation of autoantibodies and immune complexes, which, together with autoreactive T cells, can cause damage to several organs, including skin, lung, and kidney (31, 32). This type of organ damage leads to serious and fatal complications, such as lupus nephritis (33). A histological analysis of the kidney sections stained with H&E and PAS revealed increased cell numbers, matrix deposition, and thickened basement membranes in the glomeruli of aged Msn^−/Y mice (Fig. 1C). Furthermore, C3 and IgG deposition, which is a common feature of lupus nephritis, was observed in the glomeruli of aged, but not young, Msn^−/Y mice (Fig. 1D). No histopathological changes were observed in the kidney of either young or aged Msn^+/Y mice (Fig. 1C, 1D). These results suggested that moesin deficiency led to an SLE-like autoimmune disease in aged mice.

To investigate the mechanisms underlying the autoimmunity development in Msn^−/Y mice, we first examined the basal serum titers of the various Ig isotypes by ELISA as a parameter of B cell activity. Although young Msn^−/Y mice did not exhibit an autoimmune phenotype, their serum IgM, IgG2b, IgG2c, and IgE levels were significantly higher and their IgA titers were significantly lower than those of age-matched Msn^+/Y mice (Fig. 2A).

We next examined B cell responses after immunization with NP₃₉-CGG. Msn^+/Y and Msn^−/Y mice were immunized with alum-precipitated NP₃₉-CGG on day 0 and boosted on day 42. NP-specific IgM Ab titers, measured with NP₅₂-BSA–coated plates by ELISA, were similar between Msn^+/Y and Msn^−/Y mice in both their primary and secondary immune responses (Fig. 2B). Although the total NP-specific IgG1 titers, measured with NP₅₂-BSA–coated plates, were reduced in Msn^−/Y mice at 4 wk after the primary immunization (Fig. 2C, left panel), high-affinity NP-specific IgG1, measured with NP₄-BSA–coated plates, tended to be higher in these mice after the secondary immunization (Fig. 2C, right panel). The percentage and number of B cells producing NP-specific IgG1 in the spleen were similar between Msn^+/Y and Msn^−/Y mice (Supplemental Fig. 1). Although the ratio of high-affinity/total NP-specific IgG1 gradually increased after immunization in both genotypes, the increase was more prominent in Msn^−/Y mice, especially during the secondary response (Fig. 2D). These results indicated that Ab affinity maturation was enhanced in Msn^−/Y mice.

Because B cell activity was found to be enhanced in Msn^−/Y mice, we next examined whether B cell compartments are expanded in these mice. Although B cell numbers in the blood and LNs of young Msn^−/Y mice were reduced, their numbers in the spleen were slightly increased compared with those in Msn^+/Y mice (7). Thus, we analyzed B cell subsets in the spleen by flow cytometry. Immature B cells are identified as B220⁺CD93⁺ cells, which mature from T1 (B220⁺CD93⁺IgM^hiIgD^lo) to T2 (B220⁺CD93⁺IgM⁺IgD^hi) subsets (34). The number of T1 B cells was slightly higher in Msn^−/Y mice than in Msn^+/Y mice, whereas that of T2 B cells was similar in the two types of mice (Fig. 3A, 3B). The number of mature B cells, including follicular (B220⁺CD93⁻CD21^intCD23^hi) and marginal zone (B220⁺CD93⁻CD21^hiCD23^lo) B cells, was also comparable in both types of mice (Fig. 3A, 3B). Furthermore, immunofluorescence analysis of the spleen revealed that the overall organization of the T cell zones and B cell follicles was normal in Msn^−/Y mice (Fig. 3C).

During humoral immune responses, Ag is transported to the T cell zones and B cell follicles, which initiates the activation and interaction of T and B cells, resulting in GC reactions. We found that the number of GC B (B220⁺Fas⁺GL-7⁺) cells in the spleen of unimmunized young Msn^−/Y mice was significantly increased compared with that in control Msn^+/Y mice (Fig. 4A). The number of GC B cells was also significantly higher in aged Msn^−/Y mice than in age-matched Msn^+/Y mice (Fig. 4B). Spontaneous GC formation and GC cell expansion are hallmarks of many autoimmune-prone mouse strains (35). Because ERM proteins have been implicated in B cell signaling, we examined whether B cell proliferation was affected by the moesin deficiency. Ex vivo stimulation of B cells from Msn^+/Y and Msn^−/Y mice with anti-IgM Abs, which ligate the BCRs, resulted in moderate increases in proliferation of both types of cells (Supplemental Fig. 2). Stimulation via BCR-independent pathways also induced similar increases in the proliferation of Msn^+/Y and Msn^−/Y cells (Supplemental Fig. 2). These results suggested that the moesin deficiency in B cells did not affect their proliferative capacity.

During GC development, B cells receive cognate help from Tfh cells, a subpopulation of effector T cells that migrate into the B cell follicle and GC (36). In this study, we found that the number of Tfh (CD4⁺PD-1⁺CXCR5⁺) cells was higher in the spleen of young Msn^−/Y mice than in that of age-matched Msn^+/Y mice (Fig. 4C). In aged mice, Tfh cells were identified as CD4⁺ICOS⁺CXCR5⁺ cells, because CD4⁺PD-1⁺ cells accumulated in both genotypes, possibly as the result of an age-related expansion of PD-1⁺ T cells (37). We found that the number of Tfh cells was also increased in aged Msn^−/Y mice (Fig. 4D). Taken together, our findings suggested that moesin deficiency led to spontaneous Tfh accumulation and GC expansion that started at a young age.

The expansion of Tfh cells is controlled by several subsets of Treg cells. Foxp3-expressing CD4⁺ Treg cells are a well-characterized subset that regulates various effector T cell subsets. We found that young and aged Msn^−/Y mice had comparable CD4⁺ Treg (CD4⁺CD25⁺Foxp3⁺) cell numbers to those of their age-matched Msn^+/Y counterparts (Fig. 5A, 5B). In addition, the number of Tfr (CD4⁺CD25⁺Foxp3⁺CXCR5⁺Bcl6⁺) cells, a specialized subset of CD4⁺ Treg cells that suppress Tfh and GC B cells, was similar between Msn^+/Y and Msn^−/Y mice (Fig. 5C, 5D). In contrast, the number of CD8⁺ Treg (CD8⁺CD44⁺CD122⁺Ly49⁺) cells, which also inhibit Tfh cell expansion and are essential for self-tolerance (19, 20), were reduced in number and proportion in the spleen and LNs of young Msn^−/Y mice compared with those of Msn^+/Y mice (Fig. 5E, Supplemental Fig. 3A). A more severe reduction in CD8⁺ Treg cells was observed in the spleen and LNs of the aged Msn^−/Y mice compared with the control mice (Fig. 5F, Supplemental Fig. 3B). These results suggested that moesin regulates CD8⁺ Treg cell homeostasis.

The lymphopenic phenotype observed in the blood and LNs of young Msn^−/Y mice is particularly prominent in the naive CD8⁺ T cell subset (7). Patients with mutations in the moesin gene also present with profound lymphopenia, with a marked decrease in naive CD8⁺ T cell counts (8). Naive T cells of the CD4⁺ and CD8⁺ subsets decline with aging, with a concomitant increase in memory T cells. In this study, we found that the number and proportion of naive CD8⁺ T (CD8⁺CD44^loCD45RB^hi) cells were significantly reduced in the spleen of aged Msn^−/Y mice compared with that of Msn^+/Y mice, whereas naive CD4⁺ T (CD4⁺CD44^loCD45RB^hi) cell numbers were comparable in the two types of mice (Fig. 6A, 6B). In contrast, similar numbers of memory CD4⁺ (CD4⁺CD44^hiCD45RB^lo) and CD8⁺ (CD8⁺CD44^hi) T cells were observed in the spleen of Msn^+/Y and Msn^−/Y mice (Fig. 6A, 6B). Thus, CD8⁺ Treg cells, which have a memory phenotype, appear to be a small subset of memory CD8⁺ T cells that is severely affected by the moesin deficiency. The analysis of T cell subsets in female Msn^+/− mice showed that the ratio of moesin-deficient/moesin-expressing cells was decreased more substantially in CD8⁺ T cells than in CD4⁺ T cells and was most profoundly reduced in the CD8⁺ Treg subset (Fig. 6C), indicating that moesin plays a cell-intrinsic role in regulating CD8⁺ Treg cell homeostasis.

CD8⁺ Treg cells depend on IL-15, but not on IL-2, for their maintenance and proliferation in humans and mice (38, 39). To investigate the role of moesin in CD8⁺ Treg cell homeostasis, CD8⁺ Treg cells isolated from the spleen of aged Msn^+/Y or Msn^−/Y mice were cultured in the presence of anti-CD3, anti-CD28, and IL-15 or IL-2. Sorted CD4⁺ Treg cells were cultured as a control. We found that CD8⁺ Treg cells from Msn^+/Y mice proliferated in the presence of IL-15, whereas Msn^−/Y cells did not (Fig. 7A). CD8⁺ Treg cells of either genotype did not proliferate appreciably when cultured with IL-2, although Msn^+/Y cells survived better than Msn^−/Y cells (Supplemental Fig. 4A). In contrast, Msn^−/Y CD4⁺ Treg cells proliferated in response to IL-2, albeit to a slightly lesser degree than Msn^+/Y cells (Fig. 7B). Msn^+/Y and Msn^−/Y CD4⁺ Treg cells failed to proliferate in response to IL-15 (Supplemental Fig. 4B). The defective growth of Msn^−/Y CD8⁺ Treg cells was not due to increased apoptosis, given that the percentage of annexin V⁺ propidium iodide⁻ apoptotic cells was comparable between Msn^+/Y and Msn^−/Y cells after 3 d of culture (Supplemental Fig. 4C). Notably, naive CD8⁺ and CD4⁺ T cells from Msn^−/Y mice showed a similar or even enhanced proliferation rate compared with Msn^+/Y cells when stimulated with anti-CD3, anti-CD28, and IL-2 (Supplemental Fig. 4D). Taken together, these results suggested that the reduced number of CD8⁺ Treg cells in Msn^−/Y mice was due to their attenuated proliferation in response to IL-15.

The receptors for IL-15 and IL-2 consist of three subunits: the shared subunits CD132 (γc) and CD122 (IL-2Rβ) and the unique subunits IL-15Rα and CD25 (IL-2Rα), respectively (40). The trans-presentation model suggests that IL-15 bound to IL-15Rα presents the cytokine to target cells and signals via the CD122/CD132 complex (41, 42). However, IL-15Rα has also been shown to interact with CD122 and CD132 to form a high-affinity heterotrimeric receptor that binds IL-15 and transduces signals via STAT5 phosphorylation (43). Thus, we examined the levels of p-STAT5 as a measure of IL-15 signaling. Upon IL-15 stimulation, Msn^−/Y CD8⁺ Treg cells displayed markedly reduced p-STAT5 levels compared with Msn^+/Y cells (Fig. 7C, left panel), whereas the reduction was not as marked in response to IL-2 (Fig. 7C, right panel). The mRNA expression of STAT5 was similar in the two cell types (Fig. 7D). The mRNAs encoding the three subunits of the IL-15R were also expressed at comparable levels in Msn^+/Y and Msn^−/Y cells (Fig. 7E, left panel). Their cell surface expression levels were also similar in the two genotypes (Fig. 7E, right panels).

Because IL-15–mediated signaling is reported to involve the internalization of IL-15Rα but minimally of CD122 and CD132 (44), we next examined the change in IL-15Rα cell surface expression upon IL-15 stimulation. We observed that IL-15Rα surface levels decreased rapidly in response to IL-15 stimulation in Msn^+/Y and Msn^−/Y CD8⁺ Treg cells, but the rate in decrease was reduced in Msn^−/Y cells (Fig. 7F), suggesting that IL-15Rα internalization is affected by the moesin deficiency. To distinguish between surface and intracellular IL-15Rα, IL-15–stimulated CD8⁺ Treg cells were first stained for surface IL-15Rα, permeabilized, and stained again for intracellular IL-15Rα. Immunofluorescence microscopy of the cells stained in this manner revealed that most of the fluorescence signals obtained after permeabilization did not colocalize with surface staining, thus largely representing intracellular IL-15Rα (Fig. 7G). The ratio of surface/intracellular IL-15Rα was higher in IL-15–stimulated Msn^−/Y cells than in Msn^+/Y cells (Fig. 7H), supporting the view that moesin plays a role in IL-15Rα internalization. Taken together, our results suggest that moesin contributes to the maintenance of CD8⁺ Treg cells by regulating the IL-15/IL-15Rα signaling axis.

In this study, we showed that moesin deficiency led to an SLE-like autoimmune disease in aged mice, which likely increased their mortality. The spleen of moesin-deficient mice exhibited the spontaneous accumulation of Tfh cells and expansion of GC B cells, causing autoantibody production and immune complex deposition in the kidney, two hallmarks of SLE. CD8⁺ Treg cells, which inhibit the expansion of Tfh cells and, thus, are essential for self-tolerance, were severely reduced in these mice. Furthermore, we found that moesin regulated the homeostasis of CD8⁺ Treg cells in a cell-intrinsic manner through its unique role in modulating IL-15–dependent signaling.

The two ERM proteins moesin and ezrin, which are generally regarded as functionally redundant, are abundantly expressed in lymphocytes (5, 6). However, our previous study showed that deficiency of moesin alone in mice leads to impaired lymphocyte homeostasis, which is characterized by peripheral blood lymphopenia (7). In this study, we showed that moesin deficiency led to a systemic autoimmune phenotype, with a marked decrease in CD8⁺ Treg cells, further demonstrating a unique role for moesin in vivo. A unique role for moesin has also been implicated in humans through the study of primary immunodeficiency patients with mutations in the moesin gene (8, 9). These patients present with profound lymphopenia, which closely resembles that in moesin-deficient mice. In humans and mice with moesin mutations, the naive T cell counts are particularly low, with more severe reductions observed in the naive CD8⁺ T cell subset than in the naive CD4⁺ T cell subset.

Animal model studies and clinical observations have shown that lymphopenic conditions are permissive for the development of autoimmunity (10–12). Under normal circumstances, the size of the peripheral T cell pool is remarkably stable. However, under lymphopenia, a homeostatic proliferation of T cells occurs as a compensatory mechanism to restore their numbers. The homeostatic T cell expansion is accompanied by a loss of TCR diversity and emergence of autoreactive T cells and may lead to autoimmunity. Other mechanisms of autoimmunity development involve the loss or dysfunction of Treg cells (45). CD4⁺ Treg cells expressing Foxp3 are primarily produced by the thymus as a functionally mature T cell subpopulation and play key roles in the maintenance of self-tolerance and the control of physiological and pathological immune responses. The depletion of CD4⁺ Treg cells leads to the activation of rare self-reactive T cells, inducing severe autoimmune diseases in humans and mice (46–48). Although the CD8⁺ lineage of Treg cells is less understood, recent studies indicate that Qa-1–restricted CD8⁺ Treg cells contribute to the maintenance of self-tolerance. Qa-1 mutant mice, which are defective in the inhibitory interaction between CD8⁺ Treg cells and their target Qa-1⁺ Tfh cells, exhibit severe autoimmune disease (19). In addition, defective activity of CD8⁺ Treg cells is associated with the development of an SLE-like disease in B6-Yaa mutant mice (20). Although the lymphopenia-induced proliferation of T cells probably precipitated the autoimmunity in Msn^−/Y mice, we speculate that impaired CD8⁺ Treg cell homeostasis further contributed to the development of autoimmunity in these mice. It remains unknown whether the moesin deficiency in CD8⁺ Treg cells alone is sufficient to cause autoimmunity. Additional experiments will be required to clarify the contribution of moesin to CD8⁺ Treg cell function in vivo and to the regulation of autoimmunity.

IL-15 is a homeostatic cytokine that regulates the development and function of NK cells and CD8⁺ T cells; of the latter, CD8⁺ Treg cells are the most sensitive to IL-15 (24, 49). In this study, we provided evidence that moesin regulates the IL-15–dependent proliferation of CD8⁺ Treg cells. Although moesin has been implicated in TCR-mediated T cell activation (5), the observation that the TCR-stimulated proliferation of naive CD4⁺ and CD8⁺ T cells was not reduced suggests that moesin has a unique role in IL-15–dependent pathways. Because IL-15 also plays a role in the survival of naive CD8⁺ T cells (50), our finding that the naive CD8⁺ T cell counts in moesin-deficient mice were substantially reduced, coupled with recent findings that human moesin mutations are associated with reduced naive CD8⁺ T cell numbers, suggests that moesin may play an integral role in IL-15–dependent pathways. In support of this notion, NK cells, which depend on IL-15 for their expansion and survival, are also substantially reduced in patients with moesin mutations (8, 9). Although not as apparent as in IL-15–dependent pathways, moesin may also play a role in IL-15–independent pathways, including those dependent on IL-2, given that moesin deficiency caused some impairment in CD8⁺ Treg cell survival and CD4⁺ Treg cell proliferation in the presence of IL-2. The impact of these defects on CD8⁺ Treg cell homeostasis in vivo remains to be determined.

Signaling initiated by cell surface receptors are often regulated by endocytosis, including receptor internalization and sorting for degradation or recycling. Although G protein–coupled receptors and receptor tyrosine kinases have been well studied for their endocytic regulation of signaling, it is becoming clear that signaling through cytokine receptors, including γc family cytokine receptors, is also regulated by endocytosis (51). IL-15Rα has been shown to be internalized in response to IL-15 in CD8⁺ T cells (44). Because the cytoplasmic domain of IL-15Rα is critical for IL-15–mediated signaling but not for trans-presentation (52), it is possible that the cytoplasmic domain regulates IL-15Rα endocytosis and, thus, signaling. A recent study showed that moesin is associated with early endosomes, which are located close to the membrane, and controls early-to-late endosome transport (53). Although the detailed mechanism by which moesin regulates IL-15–dependent pathways remains unknown, the decreased internalization of IL-15Rα observed in the absence of moesin may explain, in part, the attenuated IL-15–mediated signaling and is consistent with studies describing a role for moesin in cell surface receptor trafficking (26, 54, 55).

In summary, our data establish a novel role for moesin in maintaining CD8⁺ Treg cell homeostasis and self-tolerance. Further research in this area will lead to a greater understanding of how CD8⁺ Treg cells function and influence the development of autoimmunity, along with important insights into the pathophysiology of primary immunodeficiency patients with moesin mutations.

We thank S. Tsukita for providing moesin-deficient mice and M. Yamada for technical assistance.

The authors have no financial conflicts of interest.