γδ T Cells Protect the Liver and Lungs of Mice from Autoimmunity Induced by Scurfy Lymphocytes Free

The γδ T cell subpopulation of lymphocytes plays multiple modulatory roles during the course of immune responses, including the production of proinflammatory cytokines (TNF-α, IFN-γ, IL-17), as well as granzymes capable of lysing infected or stressed cells (1). γδ T cells have also been shown to provide B cell help, trigger dendritic cell maturation, and facilitate the priming of αβ T cells by presenting Ag. In addition to their effects in augmenting immune responses, γδ T cells have also been reported to exert immunoregulatory roles in several different experimental systems (2). In a mouse model of doxorubicin-induced nephropathy, Vγ6/Vδ1 T cells exerted a protective function (3) and Vγ6/Vδ1 T cells were also reported to have an inhibitory effect in a model of pulmonary fibrosis induced by chronic inhalation of Bacillus subtilis microorganisms (4). In one model of experimental autoimmune encephalomyelitis in B10.PL mice, γδ T cells controlled encephalitogenic T cells by Fas/Fas ligand–dependent apoptosis of T effector cells (5). Decidual γδ T cells play a protective role in pregnancy by secreting IL-10 (6), whereas γδ T cells can prevent type I diabetes induced in neonatally thymectomized NOD mice by producing TGF-β1 (7). No single immunosuppressive mechanism can account for the regulatory activity attributed to γδ T cells. The potential contribution of γδ T cells to self-tolerance is still largely unknown.

Foxp3⁺ regulatory T cells (Tregs) are potent suppressors of immune activation and play a crucial role in the maintenance of self-tolerance (8). Mutations of Foxp3 result in the fatal immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome in humans (9), whereas a 2-bp insertion in Foxp3 leads to the premature termination of translation, resulting in the lymphoproliferative syndrome of the scurfy mice (10, 11). Scurfy mice develop multiorgan inflammation in skin, lungs, pancreas, small intestine, and liver associated with splenomegaly, as well as lymphadenopathy, resulting in death at the age of 3–5 wk. In 3- to 4-wk-old scurfy mice, disease is limited to ear skin, tail skin, liver, and lung. Transfer of scurfy lymphocytes to RAG deficient (RAG^−/−) recipients reproduces the inflammatory phenotype of the scurfy donor, but inflammation is observed in many more organs in the recipient than in the scurfy donor (12). The major reason for the enhanced disease phenotype in the RAG^−/− recipients is that scurfy mice only live to an age of 4–5 wk and some organ-specific autoreactive clones may not have had sufficient time to expand in the scurfy donor. Cotransfer of Foxp3⁺ Tregs (12) or in vitro–induced Tregs completely suppress the activation and expansion of the scurfy T cells in this transfer model (13). Suppression was mediated by a mechanism that inhibited the expansion of the scurfy cells in secondary lymphoid organs (12).

The development of multiorgan autoimmune disease in the scurfy transfer model offers the opportunity to determine the antigenic targets recognized by autoreactive Abs and T cells. We previously described a protocol to identify Ags recognized by scurfy T cells in mouse skin by first examining the Ags recognized by scurfy autoantibodies (14). Transfer of total lymphocytes from scurfy mice to RAG^−/− mice resulted in the development of a pool of autoantibodies that resembled the Ab repertoire found in the scurfy donor. Scurfy sera were screened for reactivity to skin proteins, and we identified several keratins as antigenic targets. As only a small number of scurfy B cells are transferred by this procedure, we attempted to optimize the induction of Ab-producing B cells by transferring the scurfy lymphocytes into TCRα^−/− mice. In this study, we demonstrate that γδ T cells present in the TCRα^−/− mice inhibited the development of some of the organ-specific diseases seen after transfer of scurfy cells. We characterize several of the potential immunosuppressive mechanisms used by these regulatory γδ T cells and demonstrate that they enhance the percentage of scurfy T cells capable of producing IL-10. Thus, γδ T cells can regulate highly activated scurfy T cells and maintain self-tolerance in vivo.

C57BL/6 mice were obtained from the National Cancer Institute Mouse Repository (Frederick, MD). C57BL/6 RAG1^−/− and C57BL/6 TCRα^−/− mice were obtained from Taconic Farms (Germantown, NY). C57BL/6 TCRβδ^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Female heterozygous B6.Cg-Foxp3^sf/J (scurfy) mice were purchased from The Jackson Laboratory and bred to C57BL/6 wild-type (WT) male mice to generate hemizygous male B6.Cg-Foxp3^sf/Y) (scurfy) offspring. All animal protocols were approved by the National Institute of Allergy and Infectious Diseases Animal Care and Use Committee.

Spleen cells were isolated from scurfy mice between 19 and 24 d of age and transferred (1 × 10⁶) to RAG^−/−, TCRα^−/−, or TCRβδ^−/− mice by retroorbital injection. B cells were isolated from spleens of TCRα^−/− mice by using anti-CD19 beads and autoMACS cell separator (Miltenyi Biotec) and cotransferred with scurfy spleen cells at a ratio of 1:15. γδ T cells were isolated from spleens of TCRα^−/− mice that received scurfy spleen cells 5 wk previously by using a TCRγ/δ T cell isolation kit (Miltenyi Biotec) and cotransferred with scurfy spleen cells to RAG^−/− mice at a ratio of 1:5.

To isolate lymphocytes from mouse liver, mice were anesthetized, and liver tissues were perfused in situ via the superior vena cava with PBS. Livers were dissected and placed into dishes with complete RPMI media (RPMI 1640 supplemented with 10% FBS, 50 μM 2-ME [Sigma-Aldrich], 1% sodium pyruvate, 1% nonessential amino acids, 1% HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine). The tissues were pressed against a 70-μm cell strainer with the plunger from a 5-cc syringe. The cells were centrifuged at 30 × g for 3 min to pellet hepatocytes. The supernatants were transferred to new tubes and then centrifuged at 320 × g for 5 min. The cells were then resuspended in 33% Percoll (Sigma-Aldrich) in PBS and centrifuged at 500 × g for 15 min. The cell pellets were resuspended in ACK lysis buffer and incubated for 5 min. To isolate lymphocytes from mouse lung, mice were anesthetized, and lungs were perfused in situ with PBS. Lungs were dissected and placed into tubes with 0.5 mg/ml collagenase from Clostridium histolyticum (Sigma-Aldrich) in complete RPMI media. The tissues were dissected into small pieces and incubated for 1 h at 37°C. After digestion, the tissues were pressed against a 70-μm cell and washed. The cells were resuspended in 33% Percoll in PBS and centrifuged at 500 × g for 15 min. The cell pellet was resuspended in ACK lysis buffer and incubated for 5 min.

Anti–NK1.1-allophycocyanin-Cy7 (PK136), anti–CD19-allophycocyanin (1D3), anti–CD44-Alexa Fluor 700 (IM7), anti–TCRβ-V450 (H57-597), anti–Ki-67-PE (MOPC-21), anti–IL-4-PE (11B11), and anti–IL-17A-PE (TC11-18H10) were purchased from BD Biosciences. Anti–TCRδ-FITC (GL3), anti–CD8-PE (53-6.7), anti–CD62L-eFluor 605NC (MEL-14), anti–NKG2D-allophycocyanin (CX5), anti– retinoic acid early transcript 1 (Rae-1d)-PE (RD-41), anti– murine UL16-binding protein transcript 1 (MULT1)-PE (5D10), anti–CD25-PE-Cy7 (PC61.5), anti-Foxp3- Alexa Flour 700 (FJK-16s), anti-CTLA-4-PE (UC10-4B9), anti–GITR-PE (DTA-1), anti–GARP-PE (YGIC86), anti–CD39-PE (24DMS1), anti–IL-10-PE (JES5-16E3), and anti–IFN-γ-PE-Cy7 (XMG1.2) were purchased from eBioscience. Anti–TCRVγ1-PE (2.11), anti–TCRVγ4-allophycocyanin (UC3-10A6), and anti–CD73-allophycocyanin (TY/11.8) were purchased from BioLegend. Anti–CD4-Qdot605 (RM4-5) was purchased from Life Technologies. Purified mouse anti-mouse latency-associated peptide (LAP) clone TW7-16B4 was provided by H. Weiner (Harvard Medical School, Boston, MA). The LAP Ab was labeled with SureLight-allophycocyanin at Columbia Biosciences (Columbia, MD). To separate dead cells, cells were stained with Live/Dead fixable aqua dead cell stain kit (Life Technologies) before surface staining. For intracellular staining of Foxp3, CTLA-4, Ki-67, granzyme B, and perforin, cells were fixed and permeabilized using the Foxp3 fixation/permeabilization staining kit (eBioscience). For staining of cytokines, cells were stimulated with PMA/ionomycin plus protein transport inhibitors (eBioscience) and then fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences). For intracellular staining of IL-10 in B cells, LPS (10 μg/ml) was also added for the stimulation of B cells. For FACS analysis, the data were acquired on a FACS LSR II (BD Biosciences) and analyzed using FlowJo software.

Spleens of TCRα^−/− mice were first depleted of B cells using anti-CD19 beads, and γδ T cells were purified using the Pan T cell isolation kit II (Miltenyi Biotec). Isolated naive γδ T cells were expanded as described previously (15) with minor modification. In brief, cells were cultured in complete RPMI media with IL-2 (200 IU/ml) on dishes precoated (5 μg/ml) with the γδ T cell–specific mAb GL4 (BD Biosciences) for 3 d. The recovered cells were then expanded in medium containing IL-2 (100 IU/ml) for an additional 6 d.

The severity of autoimmunity on the liver, lung, and skin was scored as described previously (16).

NKG2D-specific mAbs HMG2D (Bio X Cell) and CX5 (provided by Dr. Lewis L. Lanier, University of California San Francisco, San Francisco, CA) were administered to TCRα^−/− recipients of scurfy spleen cells via i.p. injection (400 and 250 μg per mouse, respectively).

POM-1 (Santa Cruz Biotechnology), a chemical inhibitor of NTPDase, was administered to TCRα^−/− recipients of scurfy spleen cells via i.p. injection at 10 μg/g daily.

Statistical analysis was performed with Prism software. Multiple groups were compared by ANOVA followed by a Tukey multiple comparisons test. Two groups were compared by an unpaired two-tailed Student t test. Comparisons in Fig. 1B and 1C were made by a Kruskal–Wallis test followed by a Dunn multiple comparisons test. Comparisons in Figs. 2C and 4B were made by two-tailed Mann–Whitney tests. The data are expressed as means ± SEM. A p value <0.05 was considered significant.

When we examined the three groups of recipients 5 wk after transfer, we were surprised to find that the absolute number of scurfy CD4⁺ and CD8⁺ TCR αβ⁺ T cells was significantly suppressed in the liver and lung, but not in the spleen, in TCRα^−/− recipients when compared with RAG^−/− and TCRβδ^−/− recipients (Fig. 1A). Histologically, liver and lung inflammation was less severe in TCRα^−/− recipients than that in RAG^−/− and TCRβδ^−/− recipients, although the differences in histology score between lungs of TCRα^−/− and TCRβδ^−/− recipients were not statistically significant (Fig. 1B, 1C). Elevated levels of liver enzymes were observed in RAG^−/− and TCRβδ^−/− recipients, but not in TCRα^−/− recipients, compared with control WT mice at 5 wk after transfer (Fig. 1D). Interestingly, inflammation in the ear skin was not suppressed in either of TCRα^−/− and TCRβδ^−/− recipients (Fig. 1E). These results demonstrate that TCRα^−/− recipients contain an immunoregulatory cell population that specifically protects against the organ-specific autoimmune pathology induced by scurfy T cells in the liver and lung, but not in the skin.

We first checked the different recipient mice for the presence of γδ T cells (TCRδ⁺NK1.1⁻) and B cells (CD19⁺). The absolute number of γδ T cells was higher in the skin draining lymph nodes (sLN) of TCRα^−/− mice than in WT mice, and γδ T cells as expected could not be detected in the sLN of TCRβδ^−/− mice (Supplemental Fig. 1A). The absolute number of B cells in the three strains was similar. Following transfer of the scurfy cells, expansion of γδ T cells was only observed in TCRα^−/− mice, indicating that the γδ T cells were of recipient origin and not derived from the scurfy donor (Supplemental Fig. 1B). No B cells could be detected in the RAG^−/− recipients, but similar numbers of B cells were present in the organs of the TCRα^−/− and TCRβδ^−/− recipients, indicating that the B cells are of recipient origin (Supplemental Fig. 1C).

Although both TCRα^−/− and TCRβδ^−/− mice have similar B cell pools, only the TCRα^−/− mice were protected from liver and lung pathology. However, it remained possible that IL-10–producing regulatory B cells (17) mediated protection in the TCRα^−/− recipients. There was no difference in the capacity of B cells from TCRα^−/− and TCRβδ^−/− to produce IL-10 either before or after transfer of scurfy cells (Fig. 2A). We also directly examined the suppressive capacity of Β cells in this model by cotransferring scurfy spleen cells and Β cells from untreated TCRα^−/− mice at a ratio of 1:15 (1 × 10⁶ and 15 × 10⁶) into RAG^−/− recipients. Cotransfer of Β cells did not suppress the expansion of scurfy T cells or pathology in liver and lung in the recipients 5 wk after transfer despite the presence of significantly higher numbers of B cells in the cotransferred recipients (Fig. 2B–D). It is thus unlikely that B cells play an immunoregulatory role in this model.

We therefore focused our studies on the endogenous γδ T cells in recipient mice and compared the absolute number of γδ T cells in recipient mice before and after the scurfy cell transfer. γδ T cells markedly expanded in the liver, lung, spleen, and sLN in the TCRα^−/− recipients (Fig. 3A). The percentage of CD62L^loCD44^hi activated γδ T cells significantly increased in TCRα^−/− recipients after transfer (Fig. 3B, 3C). Of note, most γδ T cells in the liver were CD62L^loCD44^hi before transfer (Fig. 3C). The percentage of Ki-67⁺ γδ T cells also increased 5 wk after transfer (Fig. 3D, 3E), consistent with an ongoing expansion process. The expansion of γδ T cells was not restricted to a particular γδ T cell subset, as Vγ1- and Vγ4-expressing γδ T cells were almost equally expanded in TCRα^−/− recipients after transfer except for a preferential expansion of Vγ1-expressing γδ T cells in the lung (Fig. 3F, 3G). This may indicate that Vγ1-expressing cells have a unique role in suppressing autoimmune pathology in the lung. These results indicate that endogenous γδ T cells are activated in response to the scurfy cell transfer and polyclonally expanded in the liver, lung, spleen, and sLN of TCRα^−/− recipients. As scurfy mice also contain γδ T cells, we enumerated the number of γδ T cells in 20- to 23-d-old scurfy mice. However, we did not observe an increase in the number of γδ T cells compared with WT littermates even though the scurfy mice exhibited marked expansion of conventional T cells as this age (Supplemental Fig. 2).

To directly examine the suppressive capacity of γδ T cells in this model, scurfy spleen cells were cotransferred into RAG^−/− recipients with γδ T cells (1:5) from TCRα^−/− recipients that received scurfy spleen cells 5 wk previously. Cotransfer of γδ T cells suppressed the expansion of scurfy T cells and decreased inflammation in the liver and lung, but not in the spleen or sLN 5 wk after transfer (Fig. 4A, 4B). In addition to freshly explanted γδ T cells from treated mice, we expanded γδ T cells from untreated TCRα^−/− mice by stimulation with anti-TCRδ mAb and IL-2. Cotransfer of in vitro–cultured γδ T cells also suppressed the expansion of scurfy T cells in the lung, but not the liver (Fig. 4A). Marked expansion of the cotransferred freshly explanted or in vitro–expanded γδ T cells was seen when they were cotransferred with the scurfy cells, but no expansion was seen when the in vitro–cultured γδ T cells were transferred alone (Fig. 4C).

As the suppressive function of the γδ T cells in this model closely resembled that of Foxp3⁺ Tregs, we first examined the expression of a number of markers that are preferentially expressed on Tregs. Very low to undetectable levels of CD25, Foxp3, CTLA-4, LAP, or GARP were detected on γδ T cells in TCRα^−/− mice, and these levels were unchanged in the spleens of TCRα^−/− mice following transfer of the scurfy cells. The GITR was expressed at high levels on γδ T cells and its expression was slightly decreased after transfer of the scurfy cells (Fig. 5A). The immunosuppressive cytokine IL-10 was expressed by a low percentage of γδ T cells in TCRα^−/− mice, and the percentage of IL-10–producing γδ T cells did not increase after scurfy cell transfer (Fig. 5B). The expression of granzyme B was increased in γδ T cells following scurfy cell transfer, but perforin expression could not be detected either before or after transfer (Fig. 5C). We cannot exclude the possibility that some of the suppressive effects of the γδ T cells may be secomdary to granzyme B–mediated cytolysis.

One other potential pathway that has been postulated to be involved in the immunosuppressive function of Tregs and other cell types is the ectoenzyme-mediated degradation of ATP to adenosine by CD39 and CD73 (18). CD73 was constitutively expressed at high levels on most γδ T cells in TCRα^−/− mice before transfer (Fig. 6A). In contrast, CD39 was expressed at low levels on splenic and lung γδ T cells and markedly upregulated after scurfy cell transfer, whereas CD39 was expressed on liver γδ T cells at high levels constitutively and its expression was not changed after scurfy cell transfer (Fig. 6A, 6B). Thus, most γδ T cells in the liver of untreated TCRα^−/− mice expressed both CD39 and CD73 in a pattern that resembled that seen on Foxp3⁺ Tregs (18), whereas most γδ T cells in the lung expressed CD73 alone (Fig. 6C).

The expression of CD39 in both liver and lung was correlated with an activated phenotype (CD62L^loCD44^hi) even before transfer (Fig. 6D), and this prompted us to examine the expression of CD39 on γδ T cells in WT mice. In WT mice, γδ T cells expressed high levels of CD39 in the liver, but not in the lung, spleen, or sLN (Supplemental Fig. 3A). As seen in the TCRα^−/− mice, high levels of CD39 expression correlated with the activation state of the cells (Supplemental Fig. 3B). It is possible that γδ T cells may be involved in the maintenance of immune homeostasis in mouse liver in the steady-state.

To determine whether the CD39/CD73 pathway contributed to the immunosuppressive function of γδ T cells, we treated TCRα^−/− recipients of scurfy T cells with the nucleoside triphosphate diphosphatase inhibitor POM-1 (19). POM-1 was administered i.p. (10 mg/kg daily for 5 wk), but it did not reverse the γδ T cell–mediated suppression of liver and lung inflammation in the recipients (Fig. 6E).

Another potential pathway by which γδ T cells might mediate suppression in this model is via the lectin-type activating receptor NKG2D that is expressed by most NK cells and NKT cells, by many γδ T cells, and by Ag-experienced CD8⁺ αβ T cells (1). NKG2D recognizes a diverse family of MHC class I–related proteins such as Rae-1, murine MULT1, and histocompatibility 60 protein in mice, which are expressed on the surface of infected, stressed, or transformed cells (1). We hypothesized that NKG2D on γδ T cells in TCRα^−/− recipients might recognize NKG2D ligands on scurfy T cells or on damaged cells in affected organs. As expected, most NK cells expressed NKG2D in untreated TCRα^−/− mice (Fig. 7). Some γδ T cells expressed NKG2D in the liver, lung, and spleen of untreated TCRα^−/− mice and the expression level was upregulated after the scurfy cell transfer (Fig. 7). About a half of the γδ T cells and a few CD8⁺ αβ T cells also expressed NKG2D in untreated WT mice (Supplemental Fig. 4). Our attempts to examine the expression of NKG2D ligands (Rae-1δ and MULT1) on scurfy T cells by flow cytometry and on liver tissues from scurfy mice by immunofluorescence were unsuccessful (data not shown). Lastly, we treated TCRα^−/− recipients of scurfy cells with a blocking NKG2D-specific mAb (CX5 at 250 μg per mouse 1 d prior to the scurfy cell transfer and twice weekly starting on day 3) (20), which resulted in partial reduction of NKG2D expression on γδ T cells, but failed to reverse the suppressive effects of the γδT cells on the liver and lung inflammation in TCRα^−/− recipients (data not shown).

CD4⁺ and CD8⁺ T cells from the spleen and lymph nodes of symptomatic scurfy mice primarily represent Th1 cells with high percentages of IFN-γ–producing cells (Fig. 8A), but they also contained a significant percentage of IL-10–producing CD4⁺ T cells. Scurfy CD4⁺ and CD8⁺ T cells from RAG^−/−, TCRα^−/−, and TCRβδ^−/− recipients contained even higher percentages of IFN-γ–producing cells and very few IL-4– and IL-17–producing CD4⁺ T cells Fig. 8A, 8C–E.). Surprisingly, CD4⁺ and CD8⁺ scurfy cells from spleen, liver, and lung recovered from the TCRα^−/− recipients contained a 2- to 3-fold increase in the percentage of cells producing IL-10 (Fig. 8B). Almost all the IL-10–producing CD4⁺ and CD8⁺ cells also produced IFN-γ (Fig. 8A). These results suggest that γδ T cells may protect TCRα^−/− recipients by enhancing IL-10 production from scurfy T cells.

In this study we have demonstrated that γδ T cells can exert potent immunoregulatory effects and protect the liver and lung of TCRα^−/− mice from the immunopathology seen when scurfy T cells are transferred to immunodeficient recipients. The scurfy defect, together with loss of TGF-β1 (21) and CTLA-4 (22), represents one of the most severe defects resulting in the development of systemic and organ-specific autoimmune disease. The scurfy cells used to induce disease were obtained from animals older than 21 d of age, at which time almost all the CD4⁺ and CD8⁺ cells exhibit a highly activated phenotype and a high percentage of IFN-γ–producing T cells. In general, Foxp3⁺ Tregs, with rare exception (23), are the only suppressor T cell population capable of suppressing scurfy T effector cells. We were therefore surprised at the potent suppressive capacity of the γδ T cells for liver and lung inflammation that we observed by comparing the spectrum of disease induced by transfer of scurfy cells to RAG^−/−, TCRα^−/−, and TCRβδ^−/− mice. We then confirmed these results in cotransfer studies with purified γδ T cells isolated from TCRα^−/− mice that had previously received scurfy cells and with in vitro–expanded γδ T cells from naive mice. It remains possible that the development of γδ T cells is abnormal in TCRα^−/− mice and that suppression might not be a property of γδ T cells that develop in normal mice. It is also possible that differences in the microbiome of the recipient strains may have played some role in augmenting the function of the γδ T cells, as the TCRα^−/− and TCRβδ^−/− strains were obtained from different sources.

We do not think that B cells in the TCRα^−/− mice played a significant immunoregulatory role in this model, as B cells were recently reported to be critical for autoimmune pathology (24) rather than for autoimmune regulation in scurfy mice. Cotransfer of B cells failed to suppress the expansion of scurfy T cells in RAG^−/− recipients. Furthermore, no increase in IL-10–producing B cells was observed in the TCRα^−/− recipients. It is also not likely that NK cells played significant roles in suppression of the transferred scurfy cells, as the numbers of NK cells in livers and lungs were less in TCRα^−/− recipients than in RAG^−/− or TCRβδ^−/− recipients (data not shown).

We observed marked expansion of the γδ Τ cells in the liver and lung, but only when cotransferred with scurfy T cells. No homeostatic expansion of γδ Τ cells was observed when they were transferred alone. We do not understand why the γδ T cells failed to expand when transferred alone in the lymphopenic environment, but the γδ T cells used in these studies were previously expanded in vitro by stimulation with anti-γδ TCR and IL-2. However, the marked expansion when cotransferred with the scurfy cells raised the possibility that the γδ T cells were activated by a ligand expressed on the scurfy cells. NKG2D (25) is expressed by γδ T cells and enables them to promptly respond to stressed or damaged cells that express NKG2D ligands. We found that about a half of γδ T cells in the liver and lung of untreated TCRα^−/− mice and WT mice expressed NKG2D and observed marked upregulation of NKG2D expression in response to the scurfy cell transfer. This suggested a possible contribution of NKG2D to the activation of γδ T cells in our experimental system although both TCR and NKG2D might be involved as reported previously (26). The levels of NKG2D were positively correlated with the expression of CD44 on γδ T cells (data not shown). We failed in our attempts to identify NKG2D ligands on scurfy T cells by flow cytometry or in the inflamed liver by immunofluorescence. Although NKG2D ligands were detected on hepatocytes in a chronic hepatitis B virus infection model after stimulation with Con A (27), their expression was transient. Similar transient expression of NKG2D ligands was observed on epidermal cells in a tape-stripping model (28). These findings suggest that the timing after cell stress is important for the detection of NKG2D ligands, and we cannot rule out the possibility that NKG2D ligands were expressed at some time point after scurfy cell transfer. However, we also attempted to use a blocking Ab to NKG2D (20) to reverse the suppressive effects of the γδ T cells, but we did not observe an increase in the immunopathology in the liver.

Numerous other suppressive mechanisms (29–33) have been proposed to account for the immunoregulatory effects of γδ T cells, and some studies have suggested that immunoregulatory γδ T cells resembled Foxp3⁺ Tregs. The activated γδ T cells in this model did not express Foxp3, Fas ligand, cell surface TGF-β, or elevated levels of IL-4. Interestingly, both CD39 and CD73, which are frequently expressed on Tregs (18, 34), were markedly upregulated on γδ T cells in liver and lung in response to the scurfy cell transfer. CD39 has recently been reported as a surface marker of mouse regulatory γδ T cells, and CD39⁺ γδ T cells have been shown to suppress contact hypersensitivity, although the mechanism used by these cells was unclear (35). The major function of CD39 is to convert ATP into AMP. CD73 then dephosphorylates AMP into adenosine that acts via A2a receptors to potently inhibit T cell responses (36, 37). Because the CD39/CD73 pathway seemed to be a likely mediator of suppression in our model, we attempted to block the function of CD39 with the nucleoside triphosphate diphosphatase inhibitor POM-1 (19). However, we did not observe a decrease in the suppressive function of the γδ T cells in treated mice at the doses of drug we used.

It is unclear why γδ T cells could suppress scurfy induced disease in TCR αβ^−/− recipients, but failed to expand or exert suppression in scurfy mice. One explanation for the failure to observe any effects of γδ T cells in the scurfy donor is that in the scurfy mouse γδ T cells are overwhelmed by the early and extensive influx of T effector cells. One of the more puzzling aspects of our results was the selective suppression of effector T cell expansion and immunopathology in liver and lung, but not in skin, spleen, or other secondary lymphoid tissues. It was previously noted that T cells doubly deficient for the scurfy gene and IL-2 also produced a disease phenotype distinct from that of T cells with only the scurfy mutation in that the doubly deficient T cells failed to induce skin and lung pathology. This difference in organ specificity of disease induction was later shown to be secondary to the IL-2–mediated regulation of genes controlling trafficking/chemotaxis/retention and Th2 differentiation (38). It is unlikely that the pattern of organ specificity observed in the present study was similarly regulated, as we observed regulation of liver and lung immunopathology, but not skin pathology, by γδ T cells. It remains possible that the organ specificity of protection by γδ T cells may be related to the normal physiologic function of γδ T cells in those organs. In WT mice, γδ T cells in the liver had an activated phenotype and expressed high levels of CD39 and NKG2D. γδ T cells, particularly in the liver, may contribute to what has been termed liver tolerance, an environment promoting T cell tolerance to Ags delivered via the portal vein (39).

CD4⁺ T cells from the scurfy donors contained an elevated percentage of IL-10–producing T cells. One of the most surprising findings in this study was the marked upregulation of IL-10 production by both CD4⁺ and CD8⁺ cells by the scurfy T cells following transfer to TCR αβ^−/− recipients. These results suggest that γδ T cells modulate IL-10 production by scurfy effectors and that the enhanced production of IL-10 may be responsible for the suppressor function of the γδ T cells. The mechanism by which γδ T cells modulate cytokine production by effector T cells and the role of the IL-10–producing scurfy cells in mediating protection from immunopathology remain to be determined.

We thank Dr. L. Lanier for the gift of the anti-NKG2D blocking mAb.

The authors have no financial conflicts of interest.