Identification of CD3+CD4-CD8- T cells as potential regulatory cells in an experimental murine model of graft-versus-host skin disease (GVHD).

We have developed K14-mOVA transgenic (Tg) mice that express membrane-associated ovalbumin (mOVA) under the control of a K14 promoter, as well as double Tg mice, by crossing them with OT-I mice that have a TCR recognizing the OVA peptide. When injected with CD8(+) OT-I cells, K14-mOVA Tg mice develop graft-versus-host disease (GVHD), whereas double Tg mice are protected. This suggests that, in double Tg mice, regulatory mechanisms may prevent infused OT-I cells from inducing GVHD. We demonstrated that, after adoptive transfer, TCRαβ(+)CD3(+)CD4(-)CD8(-)NK1.1(-) double-negative (DN) T cells are increased in the peripheral lymphoid organs and skin of double Tg mice and exhibit a Vα2(+)Vβ5(+)TCR that has the same TCR specificity as OT-I cells. These DN T cells isolated from tolerant double Tg mice proliferated in response to OVA peptide and produced IFN-γ in the presence of IL-2. These cells could also suppress the proliferation of OT-I cells and were able to specifically kill activated OT-I cells through Fas/Fas ligand interaction. These findings suggest that DN T cells that accumulate in double Tg mice have regulatory functions and may have a role in the maintenance of peripheral tolerance in vivo.

Introduction

Multiple mechanisms of immune tolerance to self-antigen are required to prevent autoimmunity. Some self-reactive T cells are deleted during thymic differentiation (Kappler ; Kisielow ; Pircher ), but others escape clonal deletion and go on to populate the periphery (Bouneaud ; Walker ). These self-reactive T cells are controlled by peripheral tolerance mechanisms (Rocha ; Jones et al., 1991) acting either directly on the self-reactive T cell (ignorance, anergy, phenotype skewing, apoptosis), or indirectly via additional cells (tolerogenic dendritic cells, regulatory T cells) (Walker ). There is growing evidence that these regulatory cells play important roles in the maintenance of immune tolerance to self and foreign antigens. A variety of regulatory T cells (Treg) have been described, including CD4+ (Gonzalez ; Sakaguchi ; Shevach, 2011), CD8+ (Jiang ), CD4-CD8- double negative (DN) TCRαβ+ cells (Zhang ; Priatel ), γδTCR+ cells (Hayday ), and NKT cells (Sharif ; Duarte ). To date, CD4+CD25+ T cells are the most extensively studied Treg that play an important role in preventing the development of autoimmune diseases and allograft rejection (Sakaguchi ; Shevach, 2011).

A novel subset of antigen-specific TCRαβ+CD3+CD4-CD8-NK1.1- DN regulatory T cell has been identified (Zhang ) and shown to be able to specifically suppress the activity of auto-, allo-, or xeno-reactive CD8+ T cells (Zhang ; Ford ; Chen ), CD4+ T cells (Chen ; Voelkl ; Zhang ; Chen ), B cells (Hillhouse et al., 2006; Zhang et al., 2006), or DCs (Gao ) in several animal models and humans. Unlike CD4+CD25+ T cells that execute their function in an antigen nonspecific manner (Sakaguchi ; Shevach, 2011), DN Treg can down-regulate immune responses in an antigen specific manner both in vitro and in vivo (Zhang ; Ford et al., 2001; Zhang ; Chen ). This population of DN Treg differs from other Treg in its surface marker expression, cytokine profile, and mechanism of suppression, and can specifically suppress CD8+ and CD4+ T cells that are primed against the same alloantigen (Zhang ; Ford ; Chen ; Young ).

We developed transgenic (Tg) mice that express ovalbumin (OVA) under the control of the keratin 14 promoter (K14-mOVA). After passive transfer of OT-I cells, CD8+ T cells transfected with a T cell receptor (TCR) that recognizes OVA-peptide in association with MHC class I, these mice develop GvHD-like disease including skin lesions and weight loss (Shibaki ; Miyagawa ). However, double Tg mice that have both K14-mOVA and OT-I TCR (Vα2 and Vβ5) transgenes do not develop GvHD after transfer of OT-I cells, suggesting that peripheral tolerance may be involved.

In this study, we demonstrate that the number of Vα2+Vβ5+CD3+CD4-CD8- DN T cells is significantly increased in peripheral lymphoid organs as well as in skin of double Tg mice. We further demonstrate that the DN T cells from the peripheral lymphoid organs of double Tg mice are able to specifically suppress and kill OT-I cells, CD8+ T cells carrying the same TCR specificity. Thus, these DN T cells are able to suppress and kill autoreactive CD8+ T cells in this model and may participate in the immunologic tolerance exhibited by the double Tg mice when OT-I cells are adoptively transferred.

Results

Double Tg mice do not develop GvHD after transfer of OT-I cells

When injected with OT-I cells, K14-mOVATg mice regularly develop GvHD-like disease, including weight loss and inflammatory skin lesions with histological changes, while double Tg mice generated by crossing K14-mOVATg mice with OT-I mice do not develop GvHD (Figure 1a-c). When we transferred green fluorescent protein (GFP)+OT-I cells to these double Tg mice, GFP+OT-I cells did not proliferate in the skin-draining lymph nodes (SDLNs) or spleens of double Tg mice compared with those transferred to K14-mOVA mice (Figure 1d, e). Although the adoptively transferred GFP+OT-I cells migrated to the ear skin of K14-mOVA Tg mice, the percentages of GFP+OT-I cells in the skin of double Tg mice were significantly decreased compared to those transferred to K14-mOVA Tg mice (Figure 1f). Flow cytometric analysis demonstrated that the adoptively transferred GFP+OT-I cells did not show an activated phenotype in the SDLNs of double Tg mice (unpublished data). These results suggest that regulatory mechanisms prevent these OT-I cells from inducing GvHD in double Tg mice. Even adoptive transfer of CD44highCD62Llow OT-I cells activated in vitro with OVA-peptides and interleukin (IL)-2 that cannot migrate to LNs, did not cause GvHD in the double Tg mice (Figure S1).

Figure 1

Double Tg mice do not develop GvHD

(a) Weight course graph. Five million OT-I cells were adoptively transferred into K14-mOVA and double Tg mice. The mice were weighed daily for two weeks. (b) Clinical photos and (c) H & E-stained ear tissues of mice 14 days after transfer of 5 × 106 OT-I cells. **, p<0.01 and *, p<0.05. “scale bars = 50 μm”. (d, e) The numbers of GFP+OT-I cells in the SDLNs (cervical LNs, inguinal LNs and axillary LNs) (d) and spleens (e) of 3 mice each 2, 5, 7 and 14 days after adoptive transfer of 5 × 106 GFP+OT-I cells. (f) The percentages of infiltrating GFP+OT-I cells in cell suspensions from the ear skin of 3 mice each at 5 and 14 days after transfer of 5 × 106 GFP+OT-I cells.

CD3+CD4-CD8- cells (DN T cells) expressing Vα2+Vβ5+ TCR are significantly increased in double Tg mice

When percentages of CD4+, CD8+ and double negative (DN) cells were compared in SDLNs of C57BL/6, K14-mOVA and double Tg mice. DN cells were significantly increased in SDLNs of double Tg mice (>80%) compared with DN cells in C57BL/6 and K14-mOVA mice (25%) (Figure 2a). The DN cells from C57BL/6 and K14-mOVA Tg mice are mainly B cells expressing B220. In contrast, an additional population of DN cells not expressing B220 was detected in LNs of double Tg mice (Figure 2b). These DN cells included a large population expressing CD3 (DN T cells; 20.2%) (Figure 2b). We also identified a population of DN T cells in the spleens of double Tg mice (data not shown).

Figure 2

Increase in DN T cells in SDLNs and skin of double Tg mice

(a) SDLN cells were harvested from the mice and stained with CD4 and CD8. The numbers in the lower left quadrants indicate the percentages of DN cells in SDLNs. (b) The graphs show expression of B220 or CD3 in the gated CD4-CD8- DN populations. (c) DN T cells were purified from pooled LNs (inguinal LNs, axillary LNs, mesenteric LNs) and spleens of double Tg mice and stained with CD3, Vα2 and Vβ5. Most of the gated CD3+ cells were Vα2+Vβ5+. (d) Vα2+Vβ5+ gated cells stained with various antibodies. The Vα2+Vβ5+ gated cells were stained positively for Ly6A. Thin lines represent isotype controls. (e) Percentages of infiltrating GFP-Vα2+Vβ5+CD4-CD8-CD3+ cells in the cell suspensions from ears of 3 mice each at 5 and 14 days after transfer of 5 × 106 GFP+OT-I cells.

DN T cells purified from the LNs and spleens of double Tg mice exhibited the Vα2+Vβ5+ TCR as is present on OT-I cells (Figure 2c) but not NK1.1, DX5 or γδ TCR (Figure 2d). Consistent with Zhang who demonstrated that Ly-6A (Sca-1) is highly expressed on DN Tregs, we also found that Ly-6A was highly expressed on DN T cells from double Tg mice (Figure 2d). Thus, adoptively transferred OT-I cells may be regulated by the DN T cells in LNs and spleens. Interestingly, we found that the percentage of DN T cells was significantly increased in the skin of double Tg mice compared with cell suspensions from skin of K14-mOVA mice after adoptive transfer of GFP+OT-I cells (Figure 2e). These findings indicate that migrated OT-I cells may be regulated by DN T cells localized in the skin.

DN T cells proliferate and produce IFN-γ in the presence of OVA peptide and IL-2

To analyze the capacity of the peripheral DN T cells to proliferate and produce cytokines after antigen stimulation, they were purified from double Tg mice. As shown in Figure 3a, they did not proliferate in response to OVA peptide without IL-2. However, strong proliferation was observed in response to the antigen when IL-2 was added at the start of culture. This response was augmented by the addition of IL-4 to the cultures. Cytokine production by these cells was also assessed (Figure 3b and Table S1) but no cytokine was detected from culture supernatants of the DN T cells stimulated by the antigen in the absence of IL-2. When IL-2 was added, the DN T cells produced interferon (IFN)-γ (Figure 3b). IL-2, IL-4 and IL-10 production was also assessed using DN T cells as well as OT-I cells. The latter, when stimulated with IL-2 and IL-4 produced significant amounts of IL-10, suggesting that DN T cells may exert their suppressive effects by producing IL-10 (Table S1). These results suggest that DN T cells can respond in an antigen-specific manner but require an exogenous source of IL-2 to maintain their viability and their capacity to produce cytokines.

Figure 3

DN T cells from LNs and spleen of double Tg mice respond to OVA-peptide

(a) DN T cells (1 × 105) were stimulated with APCs (1 × 105) in the presence or absence of 5 μg/ml OVA-peptide. The DN T cells proliferate when cultured with OVA-peptide in the presence of IL-2 alone or IL-2 plus IL-4. (b) Production of IFN-γ from OT-I cells and DN T cells was assessed by ELISA. IFN-γ is produced by DN T cells with OVA-peptide only in the presence of IL-2 alone or IL-2 plus IL-4. Data are representative of three independent experiments with triplicates in each experiment (error bars, SD) (a,b) .

Activated DN T cells do not up-regulate cell surface CD8 when cultured with IL-2

To further characterize the DN T cells, the expression of cell surface markers was examined. Purified DN T cells from double Tg mice expressed high levels of CD62L, and they did not express the acute activation markers, CD25 and CD69. The cells also expressed low amounts of CD44 (Figure 4a), indicating that they have a naïve phenotype. After stimulation with OVA peptide, the number of DN T cells expanded over 20-fold during 10 days of culture (data not shown). The expanded DN T cells expressed CD25, CD69 and CD44, and exhibited reduced expression of CD62L, indicating their activated status after specific antigen stimulation (Figure 4b). The activated DN T cells did not up-regulate CD8 and retained the CD4-CD8- double negative phenotype with partially down-regulated expressions of CD3, Vα2 and Vβ5 10 days after stimulation (Figure 4b) as well as at 2 and 7 days after stimulation (data not shown). We also examined OVA-specific cytotoxity of the activated DN T cells with a killing assay. DN T cells could not kill EG7 cells, EL-4 thymoma cells transfected to endogenously produce the SIINFEKL peptide of OVA (Figure 4c). However, CD8+ OT-I cells were able to kill EG7 cells using the same experimental conditions. These results indicate that the DN T cells do not become cytotoxic CD8+ lymphocytes (CTLs) even after stimulation with OVA peptide.

Figure 4

Cultured DN T cells with IL-2 are activated and retain the double negative phenotype

After 10 days of culture, the DN T cells were harvested and analyzed. Figures illustrate expression of various markers on DN T cells gated on Vα2+Vβ5+ cells in the absence (a) or presence (b) of peptide. Thin lines represent isotype control staining. (c) OT-I or DN T cells, activated by 5 day in vitro culture with OVA-peptide, IL-2 and IL-4 were used as effector cells. OT-I target cells were activated with ConA and IL-2 for 2 days. Target cells (EL4, EG7 or OT-I cells) were labeled with calcein and incubated with effector cells. OT-I cells exhibited cytotoxicity in a dose-dependent manner.

DN T cells are able to suppress proliferation of OT-I cells

It has been demonstrated that DN T cells possess regulatory function and can suppress immune responses mediated by CD8+ or CD4+ T cells that are syngeneic to the DN T cells (Zhang ). To determine whether the DN T cells from our double Tg mice have the ability to regulate syngeneic CD8+ T cells (OT-I cells), CFSE-labeled OT-I cells were stimulated with OVA peptides, and cultured with or without DN T cells that were isolated from double Tg mice and activated in vitro. CFSE-labeled OT-I cells were reduced in the wells containing DN T cells (Figure 5). This reduction of OT-I cells indicated that the DN T cells may kill OT-I cells. When the number of DN T cells in the culture wells was increased, suppression of proliferation of OT-I cells was also observed. These results suggest that the DN T cells are effective regulators of OT-I cells. The inability of DN T cells, activated in vitro, to suppress GvHD (Figure S2) may be due to their undergoing apoptosis shortly after adoptive transfer.

Figure 5

Regulatory function(s) of DN T cells from double Tg mice

Naïve OT-I cells were labeled with CFSE and cultured with activated DN T cells, and proliferative responses in the presence of antigen were assessed by flow cytometry. Figures show gated CD8+CFSE+ cells. Ratios of DN T cells to OT-I cells are indicated as DN ×20 - ×2.5.

DN T cells specifically kill syngeneic CD8+ T cells

To determine whether DN T cells isolated from double Tg mice could kill OT-I cells, DN T cells and OT-I cells were used as effector cells and target cells, respectively, in a calcein release killing assay to detect a perforin-dependent cytolytic pathway. However, DN T cells did not kill OT-I cells (Figure 4c). Next, we performed a JAM test using activated OT-I cells labeled with [3H] thymidine as target cells to detect a Fas-dependent pathway. Retention of [3H] thymidine by target OT-I cells was inhibited by effector DN T cells in an effector/target cell (E/T) ratio-dependent manner (Figure 6a). The killing of OT-I cells by DN T cells was blocked by the addition of Fas-Fc fusion protein before and during the JAM test (Figure 6b). These results indicate that a Fas-FasL interaction is involved in DN T cell-mediated cytotoxicity of OT-I cells. We next determined the antigen-specificity of DN T cell-mediated cytotoxicity. On the other hand, when activated Matahari cells that express a TCR with a different antigen-specificity from DN T cells were used as targets, DN cells were not cytotoxic (Figure 6c). Consistent with a previous report (Zhang ), our data demonstrate that DN T cells are able to kill activated CD8+T cells through a Fas-dependent pathway and that sharing of the TCR-specificity by both DN T cells and target cells is required for cytotoxicity to occur.

Figure 6

DN T cells specifically kill activated CD8+T cells through the Fas/FasL pathway

(a) OT-I cells were activated by anti-CD3 mAb, labeled with [3H]thymidine, and used as target cells (1.5 × 104/well). They were pulsed with OVA-peptide or left untreated. DN T cells were incubated with labeled target cells for 18hr and exhibit cytotoxicity in a dose-dependent manner. (b) Cytotoxicity is inhibited in the presence of a Fas-Fc fusion protein at an E:T ratio of 20:1. (c) OT-I cells and Matahari cells were stimulated independently with anti-CD3 mAb. DN T cells were stimulated with OVA-peptide. Activated DN T cells were cytotoxic to OT-I cells but not to H-Y peptide-pulsed Matahari cells. Data are representative of three independent experiments with triplicates in each experiment (error bars, SD) (a-c).

Discussion

To understand mechanisms of autoimmunity and tolerance against skin-associated self-antigen, we developed transgenic mice that express membrane-tethered or soluble form of a model self-antigen, OVA, under the control of a K14 promoter (Miyagawa ). Using these mice, we have identified several factors that induce tolerance or attenuate autoimmune responses (Miyagawa ; Gutermuth ; Paek ; Miyagawa ). In this study, we report that TCRαβ+CD3+CD4-CD8-DN Treg that are present in the LNs, spleens and skin may be potential regulatory cells to control the tolerance status in double Tg mice. We have demonstrated that double Tg mice expressing the transgenes for both OT-I TCR and K14-mOVA develop tolerance to OVA antigen, and that DN T cells which are increased in double Tg mice have Ag-specific regulatory functions. DN T cells from double Tg mice suppress the proliferation of CD8+ OT-I cells that have the same TCR specificity as DN T cells. We also demonstrated that DN T cells selectively kill activated OT-I cells via a Fas-Fas ligand-dependent mechanism. In addition, DN T cells may exert their suppressive effects by producing IL-10. Ideally, the ability of DN T cells by in vivo depletion from tolerant double Tg mice or by adoptive transfer into K14-mOVA Tg mice could be functionally assessed. However, due to the lack of specific markers on DN T cells, selective depletion of DN T cells is currently not possible in vivo without affecting other T cell subsets. Likewise, the purification of a sufficient number of DN T cells from double Tg mice for adoptive transfer is not possible because only small numbers of these cells can be purified. Instead we transferred DN T cells expanded in vitro together with OT-I cells into K14-mOVA Tg mice to determine whether DN T cells suppress GvHD. The results showed that the activated DN T cells did not suppress GvHD (Figure S2), a not surprising finding since injection of activated cells may not mimic injection of non-activated DN T cells. Moreover, it is likely that most activated DN T cells die after adoptive transfer into K14-mOVATg mice.

Antigen-specific DN Treg were first identified and characterized by Zhang . They demonstrated that DN Treg play an important role in the induction of tolerance in transplantation murine models of allogeneic skin (Zhang ; Ford ) and of xenogeneic heart (Chen ). Since then several studies demonstrated that DN Treg exhibit immune suppressive effects in an antigen-specific manner in multiple disease models both in vitro and in vivo (Ford et al., 2007; Zhang ; Chen ; Hillhouse ; Chen ; Young ; Mclver et al., 2008; Dugas et al., 2010). DN Treg cells have been reported to be present in humans (Fischer ) and comprise 1-3% of peripheral T cells in normal mice and humans (Zhang ; Fischer ).

DN Treg can exhibit cytotoxicity through various mechanisms depending on the type of target cells. They have been demonstrated to kill syngeneic CD8+ and CD4+ T cells in an antigen-specific manner through Fas-FasL interaction (Zhang ; Ford ). Other studies have shown that DN Treg kill autologous B cells via the perforin/granzyme pathway (Zhang et al., 2006), while they kill alloantigen-expressing DCs mainly through Fas-FasL interaction (Gao ). Our study also demonstrated that DN T cells kill syngeneic CD8+OT-I cells through the Fas-FasL dependent pathway but not through perforin/granzyme pathway.

Although the role of DN Treg in preventing allo- or xeno-immune responses is well studied, their role in preventing T cell-mediated autoimmune disease is not well known. Few studies have addressed this by using self-antigen Tg murine models. Priatel suggested that DN T cells bearing the alloreactive H-2b 2C TCR in antigen-expressing mice had regulatory function and could suppress autoreactive CD8+ T cell proliferation by using a Fas/FasL-dependent cytolytic mechanism. Subsequently, Ford et al. demonstrated (2007) that injection of gp33 peptide prevented development of type-1 diabetes in gp33 peptide-specific TCR Tg (P14) mice with increased DN T cells and decreased CD8+ T cells. These gp33 peptide-activated DN T cells suppressed proliferation of syngeneic CD8+ T cells via their cytotoxicity in an antigen-specific fashion. Our K14-mOVA Tg murine model shows that the proportion of DN T cells is increased in peripheral lymphoid organs as well as in skin in tolerant double Tg mice. The presence of OT-I cells and DN T cells in skin of double Tg mice indicates that some OT-I cells escape from regulation in the peripheral lymphoid organs and migrate to the skin, but may be further regulated by local DN T cells. It is unclear which regulatory mechanism dominates in these tolerant double Tg mice. The data suggest that DN T cells may provide a novel therapeutic approach to antigen-specific T cell-mediated autoimmune diseases.

Materials and Methods

Mice

All mice were obtained from the National Cancer Institute Animal Production Program (Frederick, MD), housed in a clean facility, bred and used in accordance with institutional guidelines. OT-I mice were obtained from Dr. Judy Kapp, Emory University, Atlanta, Georgia. Rag-1 deficient OT-I mice were purchased from Taconic (Germantown, New York). GFP+OT-I Tg mice were generated by crossing human ubiquitin C promoter-GFP Tg mice (Jackson, Maine) with OT-I mice. K14-mOVA Tg mice have been described previously (Shibaki ). Double Tg mice were generated by crossing K14-mOVA Tg mice with OT-I mice.

Adoptive Transfer

Five million OT-I cells or GFP+OT-I cells in 200 μl of phosphate buffered saline (PBS) were injected intravenously into K14-mOVA mice or double Tg mice. Weight and health status were monitored daily for 14 consecutive days after injection.

Antibodies and flow cytometry

FITC-conjugated anti-CD3, B220, NK1.1, γδTCR, Vα2, PE-conjugated anti-CD4, CD8, CD3, CD62L, CD25, CD69, CD44, DX5, Ly6A, Vβ5, PerCP5.5-conjugated anti-CD8, APC-conjugated anti-CD8, Alexa700-conjugated anti-Vα2, APC-Cy7-conjugated anti-CD4, and biotin-conjugated anti-Vβ5 (BD bioscience, San Jose, Calfornia) monoclonal antibodies (mAbs) were used for cell surface analysis. APC-conjugated streptavidin was used for the visualization of biotin-conjugated mAb. Stained cells were analyzed on a FACSCalibur and LSR II flow cytometers (BD bioscience).

Preparation of OT-I cells and Matahari cells

Single cell suspensions were prepared from pooled LNs and spleens of OT-I mice, GFP/OT-I mice. CD8+ cells were purified by Mouse CD8 T cell columns (R&D system, Minneapolis, Minnesota). In some experiments, single cell suspensions from LNs of Rag-1 deficient OT-I mice were used as CD8+OT-I cells.

Purification of CD3+CD4-CD8- DN T cells

DN T cells were purified from pooled LNs (inguinal LNs, axillary LNs, and mesenteric LNs) and spleens of 6- to 12-wk-old double Tg mice as follows. The lymphocytes were passed through Mouse T cell enrichment columns (R&D systems) and stained with FITC-conjugated anti-CD4, CD8, and B220 mAb. The cells were then incubated with anti-FITC MicroBeads (Miltenyi Biotec, Auburn, Calfornia) and purified by negative selection. The purity of DN T cells was routinely 90-98%.

Stimulation of T cells

Purified DN T cells were cultured at 1 × 106 cells/well in a 24-well plate in the presence of 2 × 106 of irradiated (3000 rad) and sex-matched splenocytes from C57BL/6 mice pulsed with 5 μg/ml OVA-peptide (SIINFEKL) as antigen-presenting cells (APCs), 50 ng/ml recombinant mouse IL-2 (rmIL-2) and 30 ng/ml recombinant mouse IL-4 (rmIL-4; Peprotech, Rocky Hill, New Jersey) in 10 % fetal bovine serum, 10 mM MEM non-essential amino acids, 1 mM sodium piruvate, 10 mM HEPES, penicillin/streptomycin RPMI 1640 culture medium (complete RPMI) for various times. Activated cells were then used for experiments. One million GFP+OT-I cells/well in 24-well plates were co-cultured with 1 × 106 of splenocytes from C57BL/6 mice treated with 1 ng/ml SIINFEKL and 50 mg/ml mitomycin C (Sigma-Aldrich, MO) in the presence of 20 ng/ml rmIL-2 in complete RPMI for 5 days. The activated GFP+OT-I cells were used for adoptive transfer experiments.

Collection of skin infiltrating cells

Ears were minced and digested in RPMI 1640 containing 2 mg/ml Liberase TL (Roche Diagnostics, Indianapolis, Indiana) at 4°C overnight, then were incubated at 37°C for 30 minutes in RPMI 1640 containing 2 mg/ml Liberase TL and 0.04 mg/ml DNase I (Sigma-Aldrich, St. Louis, Missouri) under gentle agitation. The suspensions were dissociated by pipetting and sequentially passed through 70 μm and 100 μm filters.

Proliferation assays

DN T cells (1 × 105) were stimulated with APCs (1 × 105) in the presence or absence of 5 μg/ml OVA-peptide in a 96-well flat-bottom plate. [3H] thymidine was added during the last 16 hours of a 3.5-day culture.

ELISA

DN T cells (1 × 105) or OT-I cells (1 × 105) were stimulated with 5 μg/ml OVA-peptide and APCs (1 × 105). Supernatants were removed at 48 hours after and assayed by ELISA kits (R&D systems).

CFSE labeling

OT-I cells (1 × 107/ml) were labeled with 2 μM CFSE (Molecular Probes, Eugene, Oregon) in PBS for 10 minutes at 37°C. Cells were then washed and resuspended in complete RPMI.

Suppression assay

DN T cells were cultured for 5 days and harvested. Varying numbers of DN T cells (DN T cells: CFSE-labeled OT-I cells was 2.5-20:1.) were mixed with CFSE-labeled OT-I cells. After a 2-day incubation, all the cells in each well were harvested and analyzed by flow cytometry. Cultures without DN T cells were used as controls.

Killing assays

A calcein AM release assay was used to determine perforin-dependent cytotoxicity. OT-I cells or DN T cells stimulated for 3-5 days were used as effector cells. As target cells, OT-I cells were activated with Concanavalin A (ConA) and IL-2 for 2 days. A thymoma cell line, EL4, and OVA-expressing EL4 (EG7) were treated with 500 U/ml IFN-γ (Peprotech) for 2 days and used as target cells. Target cells were then labeled with calcein AM (Molecular Probes) and added to 96-well round-bottom plates (1.5 × 104/well) with specific ratios of effector cells for 3 hours at 37°C. The supernatants were recovered, and calcein AM release was measured using a CytoFluor 2350 plate reader (Millipore). Specific lysis = ([experimental-spontaneous]/[maximal-spontaneous]) × 100. Maximal lysis was achieved with 0.1% TritonX-100, and spontaneous lysis was obtained by incubating target cells alone. The JAM test was used to determine Fas-dependent cytotoxicity of DN T cells. DN T cells stimulated for 3 days were harvested and seeded into 96-well round-bottom plates as effector cells. The target cells (OT-I cells and Matahari cells) were stimulated with plate-bound purified anti-CD3 mAb (10 μg/ml; BD Bioscience) and 10 μCi/ml of [3H] thymidine was added during the last 18 hours of a 2-day culture. [3H] thymidine-labeled target cells were pulsed with peptides for 1 hour, and then co-cultured with the effector cells at 37°C for 18 hours. The cells were harvested and counted in a beta counter. Specific cell lysis was calculated using the following equation: percentage of specific killing = (S-E)/S ×100, in which E (experimental) is cpm of retained DNA in the presence of effector cells, and S (spontaneous) is cpm of retained DNA in the absence of effector cells. In some experiments, DN T cells were treated with 10 μg/ml Fas-Fc fusion protein (R&D systems) to block FasL on DN T cells. Fas-Fc fusion protein was added 1 hour before and during the killing assays.

Histological analysis

Tissue samples were fixed in 10% neutral-buffered formalin. Paraffin-embedded tissues were sectioned and stained with hematoxylin and eosin (H & E) staining using standard techniques (American HistoLabs, Rockville, Maryland).

Statistical analysis

Data were compared using a Student's t test. Values of p < 0.05 were referred to as a significant difference.