To elucidate the mechanisms controlling peripheral tolerance, we established two transgenic (Tg) mouse strains expressing different levels of membrane-bound OVA (mOVA) as a skin-associated self-Ag. When we transferred autoreactive TCR-Tg CD8 T cells (OT-I cells), keratin 14 (K14)-mOVAhigh Tg mice developed autoreactive skin disease (graft-vs-host disease (GVHD)-like skin lesions) while K14-mOVAlow Tg mice did not. OT-I cells in K14-mOVAhigh Tg mice were fully activated with full development of effector function. In contrast, OT-I cells in K14-mOVAlow Tg mice proliferated but did not gain effector function. Exogenous IL-15 altered the functional status of OT-I cells and concomitantly induced disease in K14-mOVAlow Tg mice. Conversely, neutralization of endogenous IL-15 activity in K14-mOVAhigh Tg mice attenuated GVHD-like skin lesions induced by OT-I cell transfer. Futhermore, K14-mOVAhigh Tg mice on IL-15 knockout or IL-15Rα knockout backgrounds did not develop skin lesions after adoptive transfer of OT-I cells. These results identify IL-15 as an indispensable costimulator that can determine the functional fate of autoreactive CD8 T cells and whether immunity or tolerance ensues, and they suggest that inhibition of IL-15 function may be efficacious in blocking expression of autoimmunity where a breach in peripheral tolerance is suspected.
Multiple mechanisms contribute to the prevention of autoimmunity to self-Ags. Although self-reactive T cells are efficiently eliminated in the thymus (central tolerance (1, 2, 3)), those that recognize tissue-specific self-Ags escape central tolerance (4, 5) and undergo surveillance by peripheral tolerance mechanisms (6, 7) acting either directly on the self-reactive T cells (ignorance, anergy, phenotype skewing, apoptosis) or indirectly via additional cells (tolerogenic dendritic cells, regulatory T cells) (5).
Two parameters, APC maturation and self-Ag levels, critically control peripheral tolerance (8, 9). APCs, including dendritic cells (DC),3 capture self-Ags from other cells and present them to self-reactive T cells (cross-presentation) to induce tolerance (9). In the absence of proinflammatory signals, DC remain relatively immature, and cross-presentation of Ags by such immature DC leads to tolerance induction rather than activation of self-reactive cells (10, 11, 12, 13). Peripheral T cell tolerance may also reflect low concentrations of self-Ag (14, 15). If efficient cross-presentation is coupled with APC maturation, then autoimmunity may ensue. In addition to these two parameters, recent studies suggest the requirement for cytokine signals such as IL-12 in determining the development of tolerance or autoimmunity (16).
To study mechanisms underlying peripheral tolerance, we established two transgenic mouse strains expressing different levels of membrane-associated OVA (mOVA) controlled by a skin-associated keratin 14 (K14) promoter. Adoptive transfer of OT-I CD8 T cells that recognize OVA as nominal Ag into OVA-high-expressing (K14-mOVAhigh) mice led to rapid development of GVHD-like autoreactive skin disease (17). In contrast, transfer of OT-I cells failed to cause disease in OVA-low-expressing (K14-mOVAlow) mice despite their in vivo expansion. OT-I cells in K14-mOVAlow Tg mice are not fully activated and do not exhibit effector function as reported in some models of peripheral CD8 T cell tolerance (14, 16, 18, 19, 20). We therefore used the K14-mOVAlow Tg strain to elucidate the mechanisms controlling the maintenance or reversal of peripheral tolerance.
We administered several common γ-chain (γc)-using cytokines (e.g., IL-2, IL-7, IL-15, and IL-21) into these mice because they contribute to the homeostasis of CD8 T cells (21, 22, 23, 24, 25, 26, 27). IL-15 converted peripheral tolerance to immunity in K14-OVAlow Tg mice that were adoptively transferred with OT-I cells. The IL-15 acted directly on the OT-I cells not by facilitating CD8 T cell expansion, but by inducing functional changes of the CD8 T cells. Furthermore, neutralization of endogenous IL-15 activity using an anti-IL-2/IL-15Rβ mAb or anti-IL-15 mAb effectively, albeit not completely, inhibited the development of skin lesions in K14-mOVAhigh Tg mice following OT-I transfer. Crossing K14-mOVAhigh Tg mice with IL-15 knockout (KO) or IL-15Rα KO mice also effectively abrogated the development of skin lesions after adoptive transfer of OT-I cells.
Collectively, these results indicate that levels of selected cytokines can determine the outcome of immune responses toward self by affecting the checkpoint that determines whether peripheral tolerance or autoimmunity ensues through modulating the functional status of self-reactive CD8 T cells in vivo. Our findings provide new evidence that cytokines can be a critical costimulator to affect this checkpoint. From a clinical perspective, attempts at blocking IL-15 function may be a promising interventional strategy for some types of graft-vs-host-like reactions or autoimmunity in humans where a breach in peripheral tolerance is suspected.
Materials and Methods
All mice were obtained from the National Cancer Institute Animal Production Program (Frederick, MD), housed in a clean conventional facility, and bred and used in accordance with institutional guidelines. IL-15 KO mice and Rag-1 deficient OT-I mice (17) were purchased from Taconic Farms. IL-15Rα KO mice were purchased from The Jackson Laboratory and were backcrossed to C57BL/6 background for at least 15 generations in house. K14-mOVAhigh Tg mice have been described previously as K14-mOVA Tg mice (17) and are of C57BL/6 background, as are the K14-mOVAlow Tg mice. OT-I mice were crossed onto Thy1.1 mice (The Jackson Laboratory) to generate Thy1.1+ OT-I mice. All animal studies were conducted with prior approval by the Animal Care and Use Committee of the National Institutes of Health.
Total RNA was extracted from various tissues and reverse transcribed with StrataScript First-Strand Synthesis System (Stratagene). Resulting cDNAs were used for real-time PCR using SYBR Green PCR Master Mix (Applied Biosystems) in triplicate. 5′ (GGCATCAATGGCTTCTGAGAA) and 3′ (CCAACATGCTCATTGTCCCA) primers were used to amplify the mOVA fragment. PCR and data collection were performed on ABI Prism 7700 Sequence Detector (PerkinElmer). The data were normalized using actin.
Transfection of 293T cells
Five micrograms of pK14-mOVA was transfected into 5 × 105 293T cells using a Lipofectamine 2000 reagent (Invitrogen). Two days later, cells were harvested and used for Western blot. The pK14-mOVA plasmid, which has a hemagglutinin (HA) tag, was previously described (17). To detect mOVA expression level, a mAb against HA was used.
Preparation of ear specimen and Western blot analysis
Ears were harvested, split, and s.c. tissue was scraped off.
Ears were placed dermis side down into petri dishes containing 5000 U of dispase (BD Biosciences) and incubated at 37°C for 20 min. Epidermis was separated from the dermis, placed in a Dounce homogenizer, and disrupted in a buffer containing 50 mM Tris-HCl (pH 7.4), 1% Nonidet P-40, 0.25% sodium deoxycholate, 0.15 M NaCl, 1 mM EGTA, 1 mM PMSF, and 1 μg/ml each of aprotinin and leupeptin. Cell lysates were blotted to a polyvinylidene difluoride membrane and probed with an anti-HA mAb followed by HRP-conjugated anti-mouse Ig (Amersham). Blots were visualized by incubation with chemiluminescence substrate (Pierce). To verify equal protein loading, blots were stripped of Ab by washing at 55°C for 30 min in 62.5 mM Tris-HCl (pH 7.5), 2% SDS, and 0.1 M 2-ME and reprobed with anti-vinculin Ab.
Abs and flow cytometry
FITC-conjugated anti-Vα2, Thy1.1, PE-conjugated anti-CD62L, CD25, CD44, allophycocyanin-conjugated anti-CD8, and biotin-conjugated anti-Vβ5, Vα2 (BD Pharmingen) were used for cell surface staining. Allophycocyanin-conjugated streptavidin was used for the visualization of biotin-conjugated mAb. Stained cells were analyzed on a FACSCalibur flow cytometer (BD Biosciences).
Preparation of OT-I cells and CD8 T cells
Single-cell suspensions were prepared from pooled lymph nodes (LNs) of Rag-1−/− OT-I mice and used without additional processing. Polyclonal CD8 T cells or Thy1.1+ OT-I cells were prepared from single-cell suspensions using CD8 columns (R&D Systems) from pooled LNs and spleen from C57BL/6 mice or Thy1.1+ OT-I mice, respectively.
Various doses of OT-I cells in 200 μl of PBS were injected i.v. into K14-mOVA Tg mice or C57BL/6 mice. Weight and health status was monitored daily for 14 days after injection. In some experiments, OT-I cells (1 × 107/ml) were labeled with 2 μM CFSE (Molecular Probes) in PBS for 10 min at 37°C. Cells were then washed and resuspended in PBS.
Recombinant murine IL-15, IL-2, and IL-7 were purchased from PeproTech. Two micrograms per dose of each cytokine was administered by i.p. injection twice a day for 5–7 days except for the first and the last 2 days, when they were administered once per day.
In vivo Ab treatment
TMβ1, an anti-CD122 (IL-2/IL-15Rβ subunit) mAb was kindly provided by UCB Pharma. TMβ1 (250 μg/day) or isotype control (rat IgG2a; 250 μg/day) was injected i.p. on days −5, −3, and −1 before and days 1 and 4 after transfer of OT-I cells. K14-mOVAhigh Tg mice were injected with 500 μg/day (i.p.) of rat anti-mouse IL-15 Ab (AIO3) (28) or isotype control Ab for 7 days starting 2 days before transfer of OT-I cells.
Culture of OT-I cells with IL-15
Naive OT-I cells were stimulated with plate-bound anti-CD3 mAb (10 μg/ml) with or without mIL-15 (50 ng/ml) at 2 × 106 cells/well in a 24-well plates in complete RPMI 1640 medium. For culture of OT-I cells with IL-15 alone, 500 ng/ml of mIL-15 was used. Five days later, these pretreated OT-I cells were harvested and transferred into K14-mOVAlow Tg mice.
Intracellular cytokine staining
Five million OT-I cells in 200 μl PBS were injected i.v. into C57BL/6 and K14-mOVA Tg mice. On day 5 after injection, single-cell suspensions from LN and spleens were prepared. Cells were then cultured at a density of 1 × 107 cells in a 24-well plate at 37°C in the presence of 2 μg/ml of OVA peptide (SIINFEKL) and 5 μg/ml of GolgiPlug (BD Pharmingen). Five hours later, cells were harvested and stained for Vα2, Vβ5, Thy1.1, and CD8. Cells were then fixed and permeabilized by incubation with Cytofix/Cytoperm solution (BD Pharmingen). The permeabilized cells were incubated with PE-conjugated anti-IFN-γ mAb and the fluorescence intensities were determined by flow cytometry.
Purification of OT-I cells from K14-mOVAlow Tg mice
Five million OT-I cells or Thy1.1+ OT-I cells were injected into K14-mOVAlow Tg mice. On day 5 following injection, OT-I cells or Thy1.1+ OT-I cells were purified from the pooled LN and spleens of injected K14-mOVAlow Tg mice. The lymphocytes were passed through CD8 columns and the enriched CD8 T cell population was stained with FITC-conjugated anti-Vα2 mAb or anti-Thy1.1 mAb. The cells were then incubated with anti-FITC microbeads (Miltenyi Biotec) and purified by positive selection. The purity of Vα2/CD8+ cells or Vα2/Thy1.1 cells was 85–99%.
Naive or tolerant OT-I cells (1 × 105) were cultured with various concentrations of cytokines and APC (1 × 105). Splenocytes from C57BL/6 were irradiated (3000 rad) and used as APCs. For positive controls, APCs were pulsed with 10 μg/ml of OVA peptide (SIINFEKL) for 1 h, washed twice, and used as peptide-pulsed APCs. In some experiments, naive or tolerant OT-I cells (2 × 105) were cultured with 50 ng/ml of mIL-15. [3H]thymidine (1 μCi) was added during the last 16 h of a 3-day culture.
The cultures established for the proliferation assay described above were used. Supernatants were removed at 48 h and assayed by ELISA (R&D Systems).
Preparation of epidermal sheets and immunofluorescence microscopy
C57BL/6 mice were injected i.p. with PBS or 2 μg/dose of IL-15 twice a day for 3 days. As a positive control for class II MHC and CD86 epidermal cell surface staining, ears were painted with trinitrochlorobenzene (TNCB) (29). TNCB was dissolved in acetone-olive oil (4:1) and C57BL/6 ears were treated with 1% TNCB on both sides 24 h before preparation of epidermal sheets. To obtain epidermal sheets, the ears were split into dorsal and ventral halves and both halves were incubated in 0.5 M ammonium thiocyanate (Sigma-Aldrich) for 20 min at 37°C to allow separation of epidermal sheets from the dermis. The sheets were incubated in cold acetone for 10 min, washed with PBS, and then stained with FITC-MHC class II and PE-CD86 Abs. Images were viewed with a fluorescence microscope (Carl Zeiss MicroImaging).
Tissue samples were fixed in 10% neutral-buffered formalin. Paraffin-embedded tissues were then sectioned and stained with H&E using standard techniques (American HistoLabs).
Clinical severity scores were determined on day 14 after injection of OT-I cells. Zero to 2 points were given for each of the following criteria: 1) skin lesion (erythematous rash, crust formation): no, 0; mild, 1; severe, 2; 2) alopecia (hair loss): no, 0; mild, 1; severe, 2; 3) mucosal involvemant (eye, mouth, nose: scale and crust formation): no, 0; mild, 1; severe, 2; 4) hunched appearance: no, 0; yes, 1; and 5) weight loss: <5%, 0; 5–15%, 1; >15%, 2. The maximal possible clinical severity score for an individual mouse was 9.
Use of this clinical severity score system was validated by several fellows from various laboratories in the Dermatology Branch of the National Cancer Institute. Their assessment of clinical severity of GVHD-like skin lesions differed very little from one another.
The obtained data were compared using a two-tailed Student’s t test (see Fig. 1 A) or a Kruskal-Wallis test. Values of p < 0.05 were referred to as a significant difference.
Levels of OVA dictate the development of autoreactive skin disease following injection of OT-I cells into K14-mOVA Tg mice
To determine mOVA expression levels in various founders of K14-mOVA Tg mice, genomic DNA, mOVA-encoding mRNA, and protein were quantified by PCR (data not shown), real-time PCR (Fig. 1,A), and by Western blotting (Fig. 1,B), respectively. A high expresser (K14-mOVAhigh, Fig. 1,B, column c) and a low expresser (K14-mOVAlow, Fig. 1 B, column b) were identified.
Following adoptive transfer of 5 × 106 naive OT-I cells into these two mice, the K14-mOVAhigh Tg mice developed acute inflammatory skin lesions and weight loss starting on day 4 (Fig. 1, C and E) (17), whereas K14-mOVAlow Tg mice did not develop disease (Fig. 1, C and E). Histology of the ears showed a thickened epidermis, exocytosis, apoptotic epidermal cells, and dermal inflammation in K14-mOVAhigh Tg mice, consistent with GVHD (Fig. 1,D) (17). K14-mOVAlow Tg mice and C57BL/6 mice did not exhibit skin lesions after transfer of OT-I cells (Fig. 1,D). To rule out the possibility that the number of injected OT-I cells determined the occurrence of immune reactions, various numbers of OT-I cells were transferred into K14-mOVAlow Tg mice. As previously reported, as few as 1 × 106 OT-I cells induced disease in K14-mOVAhigh Tg mice (17), whereas K14-mOVAlow Tg mice failed to develop skin lesions even after transfer of 1.5 × 107 OT-I cells (Fig. 1 F). These data suggest that low OVA levels rendered transferred OT-I cells unresponsive and that an increase in the number of injected cells does not overcome the effect of the tolerogenic environment.
OT-I cells injected into K14-mOVAlow Tg mice proliferate and exhibit a CD25lowCD44highCD62Lhigh phenotype without gain of effector function
On day 5 after adoptive transfer of OT-I cells into mice, skin-draining (superficial) LN cells were harvested and analyzed for the presence of OT-I cells and for activation markers on these cells. We observed an increase in the percentage of Vα2Vβ5-positive OT-I cells in both K14-mOVA Tg mice compared with C57BL/6 mice (Fig. 2,A). To completely exclude the recipient-derived Vα2Vβ5 cells, we also transferred Thy1.1+ OT-I cells (Fig. 2,B). We found a similar increase in the percentage of Thy1.1+ OT-I cells in both K14-mOVA Tg mice as in Fig. 2,A. The percentage of OT-I cells in K14-mOVA Tg mice increased in superficial LN on day 4, further expanded by day 7, and decreased by day 14 after transfer (Fig. 2,C). Thus, transferred OVA-specific OT-I cells recognized OVA Ag in both K14-mOVA Tg mice, but more OT-I cells accumulated in K14-mOVAhigh Tg mice than in K14-mOVAlow Tg mice. CFSE dilution analysis showed that transferred OT-I cells did not proliferate in control mice but proliferated to a similar extent in both K14-mOVA Tg mice (Fig. 2,D). Thus, the rate of proliferation did not correlate with the severity of disease. To determine whether disease development correlated with the extent of activation of the OT-I cells in vivo, we recovered OT-I cells from superficial LN on day 4. In control mice, the cells remained phenotypically naive (CD25low, CD44low, and CD62Lhigh expression) (Fig. 2,E). In contrast, the expression pattern of these molecules by OT-I cells was characteristic of fully activated effector CD8 T cells (CD25highCD44highCD62Llow) in K14-mOVAhigh Tg mice (Fig. 2,E). In K14-mOVAlow Tg mice, OT-I cells were not fully activated (CD25lowCD44highCD62Lhigh, Fig. 2 E). Thus, proliferation and activation of OT-I cells are separately regulated in these two different environments.
We next assessed IFN-γ production by transferred OT-I cells as a marker of effector function. In control and K14-mOVAlow Tg mice, IFN-γ production by OT-I cells recovered from LN and spleen at day 5 after transfer was negligible. In contrast, a large percentage of OT-I cells exhibited IFN-γ production in K14-mOVAhigh Tg mice (Fig. 2 F). These findings indicate that OT-I cells adoptively transferred into K14-mOVAlow Tg mice proliferated but were not fully activated and did not gain effector function. We therefore refer to these cells as “tolerant” OT-I cells.
Adoptive transfer of OT-I cells into K14-mOVAlow Tg mice treated with exogenous IL-15 induces GVHD-like skin disease
The importance of γc cytokines in CD8 T cell differentiation, maturation, and homeostasis has been demonstrated (27). IL-15 especially acts on CD8 T cells, and thus we administered IL-15 to determine whether it could modulate the effect of the OT-I cells in the K14-mOVAlow Tg mice. Five million OT-I cells were adoptively transferred into K14-mOVAlow Tg mice, and IL-15 (2 μg per dose) was administered once or twice daily for 5–7 days (Fig. 3,A). In this setting, these K14-mOVAlow Tg mice exhibited weight loss and developed skin lesions beginning on day 4 (Fig. 3,B), consistent with what occurs when OT-I cells are adoptively transferred into K14-mOVAhigh Tg mice. Histology of the ears of the IL-15-treated K14-mOVAlow Tg mice showed basal vacuolar changes, exocytosis, apoptosis of keratinocytes, and intense dermal infiltrates (Fig. 3 C), hallmark changes of GVHD. These results suggest that exogenous IL-15 reverses tolerance of the OT-I cells in K14-mOVAlow Tg mice.
The controls used in these studies helped determine that the effect of the IL-15 on inducing autoreactive skin disease in K14-mOVAlow Tg mice was specific for the OVA Ag. Five different groups of mice were tested: 1) K14-mOVAlow Tg mice injected with OT-I cells, 2) K14-mOVAlow Tg mice injected with OT-I cells and IL-15, 3) C57BL/6 mice injected with OT-I cells and with IL-15, 4) K14-mOVAlow Tg mice injected with CD8 T cells from C57BL/6 mice and with IL-15, and 5) K14-mOVAlow Tg mice with IL-15 alone. Although IL-15 was able to enhance the proliferation of CD8 T cells, only K14-mOVAlow Tg mice receiving OT-I cells and IL-15 developed skin lesions (Fig. 3 B). These results demonstrated that the GVHD-like reaction is specific for OT-I cells and OVA Ag.
Exogenous IL-15 enhances injected OT-I cell function in vivo
Earlier we observed that the activation status of OT-I cells in vivo correlated well with the development of autoreactive skin disease in K14-mOVA Tg mice. Thus, we characterized transferred OT-I cells in the IL-15-treated K14-mOVAlow Tg mice. We observed a significant increase in the percentage of OT-I cells in IL-15-treated K14-mOVAlow Tg mice as compared with nontreated K14-mOVAlow Tg mice (Fig. 3,D). As shown above, in K14-mOVAlow Tg mice, OT-I cells had a CD25lowCD44highCD62Lhigh phenotype (Figs. 2,E and 3,E). Analysis of surface marker expression on OT-I cells in IL-15-treated K14-mOVAlow Tg mice showed a CD25highCD44highCD62Llow phenotype suggestive of full activation (Fig. 3 E).
We next determined whether injected OT-I cells gained effector function in IL-15-treated K14-mOVAlow Tg mice. A significant percentage of OT-I cells in IL-15-treated K14-mOVAlow Tg mice produced IFN-γ (Fig. 3 F), suggesting that administration of IL-15 can covert tolerant OT-I cells into fully autoreactive OT-I cells in vivo.
IL-15 converts tolerant OT-I cells into fully functional OT-I cells
In vitro studies were next conducted to determine the precise mechanism of the effect of IL-15 on OT-I cells in K14-mOVAlow Tg mice. We asked whether IL-15 could alter the functional status of OT-I cells obtained from K14-mOVAlow Tg mice in vitro. We purified OT-I cells from K14-mOVAlow Tg mice on day 5 after adoptive transfer. These tolerant OT-I cells were then cultured with various concentrations of IL-15 and with nonpulsed APC (Fig. 4,A). After 3 days of culture, tolerant OT-I cells proliferated in response to 10 and 50 ng/ml of IL-15 in the absence of peptide (Fig. 4,A). This effect was almost as strong as that seen in the positive controls containing peptide. To rule out the possibility that using anti-Vα2 Abs for purification affect the function of OT-I cells, we also purified Thy1.1+ OT-I cells using anti-Thy1.1 Abs. The result was almost identical with Fig. 4,A (Fig. 4,B), suggesting that it did not have functional effects. To determine whether they gained effector function, production of IFN-γ was also assessed by ELISA (Fig. 4,C). Tolerant OT-I cells without peptide produced IFN-γ in the presence of 50 ng/ml of IL-15. To rule out the possibility that IL-15 might have acted through modulation of APC function, we also performed assays using tolerant OT-I cells in the absence of APC (Fig. 4,D–F). Tolerant OT-I cells were cultured with 50 ng/ml of IL-15 alone for 3 days and examined for expression of activation markers. At the start of the culture, tolerant OT-I cells were not fully activated (Fig. 4,D). However, after incubation with IL-15, they expressed high levels of CD25 and low levels of CD62L, indicating that IL-15 directly induced full activation of tolerant OT-I cells (Fig. 4,D). Cultured OT-I cells proliferated (Fig. 4,E) and produced high levels of IFN-γ (Fig. 4 F) when stimulated with IL-15 in the absence of peptide or APCs. These results indicate that IL-15 per se altered the nature of tolerant OT-I cells.
Because a critical role for IL-7 in homeostatic proliferation and survival of naive T cells has been identified (30, 31, 32), IL-7 was also assessed. When the same doses of IL-7 were injected, K14-mOVAlow Tg mice failed to develop disease (Fig. 4,G), although OT-I cells express IL-7Rα (data not shown). In contrast, K14-mOVAlow Tg mice treated with IL-2 had severe toxic effects and died on day 4 after OT-I transfer (Fig. 4 G). Lower doses of IL-2 (0.5–1 μg/dose) did not affect the ability of OT-I cells to cause skin lesions (data not shown).
We also examined the in vitro effects of IL-7 and IL-2 on tolerant OT-I cells. As shown in Fig. 4,A–C, IL-15 and IL-7 had differential effects on proliferation and production of IFN-γ by tolerant OT-I cells. In contrast to IL-15, IL-7 failed to stimulate proliferation of tolerant OT-I cells, and IL-7 did not induce production of IFN-γ. FACS analysis showed that IL-7Rα was expressed by tolerant OT-I cells but that it was down-regulated compared with the naive OT-I cells (data not shown). This in vitro effect is consistent with the in vivo effects of IL-7 (Fig. 4,G). As in the case with IL-15, IL-2 stimulated tolerant OT-I cells to proliferate (Fig. 4, A and B) and to produce IFN-γ (Fig. 4,C) in the absence of Ag peptide. IL-2 was more potent than IL-15, an effect consistent with the in vivo data (Fig. 4 G). However, as mentioned above, lower doses of IL-2 in vivo did not result in GVHD when OT-I cells were adoptively transferred. There may be a very limited dose range in which IL-2 could simulate the results obtained with IL-15, which was effective across a broad dose range. Taken together, these results indicate that IL-15, but not IL-7, acts as a specific converter of proliferating OT-I cells into autoreactive CD8 T cells. Thus, IL-15 does not simply expand the cell pool, but it alters the nature of the CD8 T cells.
Injected IL-15 may act directly on OT-I cells
We next determined whether IL-15 directly acts on injected OT-I cells. We adoptively transferred OT-I cells that were pretreated with IL-15 into K14-mOVAlow Tg mice to examine whether they could mimic the in vivo transfer of naive OT-I cells into IL-15-treated K14-mOVAlow Tg mice. Naive OT-I cells were cultured with anti-CD3 mAb in the presence of IL-15, and IL-15-primed (pretreated) cells were then transferred. When 1.5 × 107 pretreated OT-I cells were transferred, K14-mOVAlow Tg mice developed GVHD-like disease (Fig. 5,A–C). Furthermore, OT-I cells stimulated with anti-CD3 mAb alone failed to induce disease (Fig. 5, B and C), while OT-I cells pretreated with high doses (500 ng/ml) of IL-15 induced disease in K14-mOVAlow Tg mice (Fig. 5,B), suggesting the importance of IL-15 stimulation. Consistent with observations described above, OT-I cells pretreated with high doses of IL-7 did not induce disease (data not shown). Because K14-mOVAlow Tg mice did not develop disease even when 1.5 × 107 naive OT-I cells were injected (Fig. 1 F), these results indicate that IL-15 pretreatment directly rendered OT-I cells pathogenic in K14-mOVAlow Tg mice and that this cytokine modulates the functional status of OT-I cells to determine the disease outcome.
IL-15 has been implicated in inducing functional maturity of accessory cells upon Ag presentation (33, 34, 35). We therefore determined whether IL-15 induced maturation of DC in our model. We focused on Langerhans cells (LC) because LC are the APC in closest proximity to the keratinocytes that produce OVA. Epidermal sheets were stained for LC after PBS injection, TNCB painting, or IL-15 injection. LC in TNCB-treated ears up-regulated CD86 and formed clusters (Fig. 5,D); however, when 2 μg/dose of IL-15 was injected (total of six doses), LC maintained their immature phenotype (Fig. 5 D), indicating that IL-15 did not activate LC.
Endogenous IL-15 plays a role in autoreactive skin disease in K14-mOVAhigh Tg mice
The action of exogenously administered IL-15 in K14-mOVAlow Tg mice does not necessarily mean that this cytokine has a similar role in controlling the development of autoreactive skin disease under physiological conditions. To this end, blocking experiments were performed with anti-IL-2/IL-15Rβ mAb (TMβ1) and with anti-IL-15 mAb in K14-mOVAhigh Tg mice after transfer of OT-I cells. When OT-I cells were administered into K14-mOVAhigh Tg mice that had been injected with the TMβ1 Ab, recipient mice exhibited slight weight loss, did not develop skin lesions (Fig. 6,A), and ear histology remained normal (Fig. 6,B), confirming that TMβ1 blocked the development of disease (Fig. 6,C). However, to ascertain that this blockade by TMβ1 was not due to Ab-dependent cell-mediated cytotoxicity or by blocking IL-2 action in vivo, we tested anti-IL-15 mAb in K14-mOVAhigh Tg mice. Multiple injections of anti-IL-15 mAbs blocked the development of disease in OT-I-injected K14-mOVAhigh Tg mice (Fig. 6,D). Histology of the ears in anti-IL-15-treated mice revealed minimal to no GVHD (Fig. 6,E). However, it is noteworthy that the effect of this Ab was less complete than that of TMβ1: some K14-mOVAhigh Tg mice treated with anti-IL-15 mAb developed very mild skin lesions, while other mice had no skin lesions at all (Fig. 6, C and F).
Nonetheless, these results with blocking Abs prompted us to generate K14-mOVAhigh Tg mice on an IL-15 KO background to further determine the physiological role of IL-15 in developing autoreactive skin disease in K14-mOVAhigh Tg mice. K14-mOVAhigh Tg/IL-15 KO mice did not manifest skin lesions clinically (Fig. 6,G) or histologically (data not shown) although they transiently lost weight (Fig. 6,I). Elimination of IL-15 by genetic means did not completely abrogate the disease because some mice developed mild mucocutaneous lesions around the eyes or lost >5% of their weight on day 14, giving a clinical severity score of 1–3 (Fig. 6,J). Interestingly, K14-mOVAhigh Tg/IL-15Rα KO mice had a phenotype similar to K14-mOVAhigh Tg/IL-15 KO mice and developed minimal to no skin lesions (Fig. 6 H–J), suggesting that transpresentation of IL-15 plays a role in K14-mOVAhigh Tg mice.
In this study, we have shown that IL-15 regulates the final responses of potentially autoreactive CD8 T cells, that is, whether peripheral tolerance or autoimmunity ensues. The role of IL-15 in this context seems physiological because the abrogation of endogenous IL-15 activity using anti-IL-2/IL-15Rβ Ab or anti-IL-15 Ab prevented the development of GVHD-like skin lesions. This study provides the first evidence of reliable blocking of murine inflammatory disease model using anti-IL-15-neutralizing Ab. Our observations were further strengthened by the results using gene targeting because K14-mOVAhigh Tg mice on an IL-15 KO background did not develop skin disease after transfer of OT-I cells. Interestingly, K14-mOVAhigh Tg mice on an IL-15Rα KO background did not develop skin lesions either, suggesting that IL-15 acted on the injected OT-I cells by way of a trans-presentation mechanism (36, 37) rather than as a soluble factor. Despite the critical demonstration that trans-, rather than cis-, presentation is the mode of action of this cytokine upon development of NK and CD8 T cells in vivo, it remains unclear whether trans-presentation operates during more dynamic immune responses, including graft-vs-host reaction. Thus, our model may be useful in further elucidating the detailed mechanisms underlying the action of IL-15 trans-presentation upon the activation of CD8 T cells.
Outcomes of lymphocyte-mediated immune responses are regulated at many levels. Because T cell activation requires Ag and costimulation, Ag dose and DC maturation primarily determine the outcome of the autoreactive CD8 T cell function and control tolerance or autoimmunity (8, 9). After Ag stimulation by self-Ags, naive CD8 T cells undergo clonal expansion and may develop effector functions to induce immune responses. However, the presence of autoreactive T cells with effector functions is not sufficient for an autoimmune disease to occur (38). Recent studies have demonstrated that cell proliferation can be separated from development of effector functions (16, 19, 20). Thus, signals in addition to Ag and costimulation may be required for the development of CD8 T cell-induced autoimmune disease.
In our model, OT-I cells adoptively transferred into K14-mOVAlow Tg mice proliferated but they were not fully activated and did not gain effector function (14, 16, 18, 19). Thus, OT-I cells failed to cause disease in K14-mOVAlow Tg mice. However, in vivo administration of IL-15 into OT-I-injected K14-mOVAlow Tg mice broke tolerance and caused GVHD-like skin lesions by altering the functional status of the adoptively transferred OT-I cells. Similarly, injection of OT-I cells that had been pretreated with IL-15 ex vivo into K14-mOVAlow Tg mice caused disease. More importantly, neutralizing IL-15 function by in vivo administration of an anti-IL-2/IL-15Rβ or anti-mouse IL-15 Ab effectively blocked the development of GVHD-like skin lesions in these mice, suggesting that, indeed, the levels of IL-15 physiologically control the onset of tolerance or disease in our experimental model. Our observations collectively demonstrate that in addition to Ag and conventional costimulation, a cytokine, in particular IL-15, can be a critical cofactor in the determination of tolerance or autoimmunity.
We also demonstrated the rather nonredundant relevance of IL-15 in our system because other members of the γc-using cytokine, such as IL-7 and IL-21, failed to cause disease when injected into K14-mOVAlow Tg mice with OT-I cells (Fig. 4 G and F. Miyagawa, Y. Tagaya, and S. I. Katz, unpublished observation). In contrast, IL-2 exhibited a similar activity to IL-15 with regard to in vitro activation of OT-I cells. High doses of IL-2 can cause severe, dose-limiting toxicities in humans (39). Similarly, IL-2 induced severe toxic side-effects (death) at doses equivalent to those used for IL-15, whereas lower doses of IL-2 did not enable adoptively transferred OT-I cells to induce GVHD-like skin lesions.
A mechanism for the role of IL-15 on TCR-mediated activation of CD8 T cells has been proposed by Oh et al. (40). They demonstrated that IL-15 induced CD8 T cell maturation by promoting greater survival of high-avidity CTLs and induced higher levels of the surface coreceptor CD8αβ (40). Although this proposition may help explain our data, CD8αβ induction can be a consequence of, rather than the cause of, the acquisition of full CTL activity in our model. In their proposal, the role of IL-15 is defined as a simple enhancer of Ag presentation. However, simultaneous IL-15 and TCR signaling may allow cross-talk between these two distinct pathways, enabling CD8 T cells to undergo diverse events such as maturation, and acquisition of effector functions, as well as the induction of CD8αβ. In a similar context, Curtsinger et al. (16) demonstrated that IL-12 provides a critical signal (which they defined as signal 3) for the acquisition of effector functions in CD8 T cells. Interestingly, IL-15 did not replace IL-12 in their system. This same group demonstrated that IFN-αβ may act as signal 3 (41). One crucial question is whether the cytokine signal simply enhances the TCR signal or cytokines represent a distinct T cell-activating entity by inducing intracellular events that are different from those activated by the TCR signal. The detailed mechanism by which cytokines participate in full CD8 T cell maturation requires further biochemical investigation.
Our experimental system provides a unique and convenient model to study the diverse elements involved in the decision of whether peripheral tolerance or autoimmunity ensues after expansion of autoreactive CD8 T cells. IL-15 is an important factor influencing the development of self-reactive immune effector responses in our system. It appears to directly act on the injected OT-I cells via trans-presentation. Additionally, the interplay between autoreactive CD8 T cells and resident Ag-expressing skin cells may also have a critical influence on the final clinical outcome. Analysis of the behavior of OT-I cells in K14-mOVA Tg mice crossed with various genetically engineered mice should help us uncover the various parameters involved in this process. From a clinical perspective, our model may be appropriate for the study of skin-targeted disease such as erythema multiforme, lichen planus, psoriasis, and systemic lupus erythematosus. Also, we expect that our study would further support ongoing attempts to use IL-15 and anti-IL-15 or anti-IL-15 receptor Abs in reversing tolerance in cancer or in preventing autoimmunity in humans.
We thank Jay Linton for technical assistance and Mark Udey for helpful discussions. The TMβ1 Ab was supplied by UCB Pharma, U.K.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by intramural research funds of the Center for Cancer Research, National Cancer Institute/National Institutes of Health.
Abbreviations used in this paper: DC, dendritic cell; γc, common γ-chain; GVHD, graft-vs-host disease; HA, hemagglutinin; K14, keratin 14; KO, knockout; LC, Langerhans cell; LN, lymph node; MFI, mean fluorescence intensity; mOVA, membrane-bound OVA; Tg, transgenic; TNCB, trinitrochlorobenzene.