γ_c cytokines condition the progressive differentiation of CD4⁺ T cells

Abstract

After their initial antigen encounter in the secondary lymphoid organs, activated T cells must receive additional signals in the peripheral tissues to fully differentiate. Here, we provide evidence that γ_c cytokines are critical during this process. Using the Marilyn (Ml) T cell antigen receptor (TCR) transgenic model, we show that male skin grafts are tolerated in the absence of γ_c, but that Ml CD4⁺ T cells proliferate normally in response to antigen, traffic to the graft site and recruit an inflammatory response [including natural killer (NK) cells, neutrophils, and macrophages] that is independent of T cell γ_c expression. Whereas wild-type T cells demonstrate a progressive differentiation phenotype from the spleen to the tissues, skin-infiltrating effector T cells (CD44^hiCD62L^lo) from γ_c⁻ mice were phenotypically abnormal with reduced ICOS, NKG2D, granzyme B, and IFN-γ expression. These defects could be mapped to deficiencies in IL-2 and, surprisingly, IL-15. These results define a late checkpoint in T cell differentiation in the tissues where γ_c cytokines, including IL-15, authenticate CD4⁺ T cell effector functions.

Common γ chain (γ_c) cytokines (including interleukins IL-2, -4 -7, -9, -15, and -21) are pleiotropic soluble factors that play important roles in the generation and homeostasis of lymphoid cells (reviewed in ref. 1). Humans and mice that are deficient in γ_c, γ_c-coupled receptors, and their associated ligands and signaling pathways have multiple and distinct defects in B, T, and natural killer (NK) cell development, resulting in severe combined immunodeficiency (reviewed in ref. 2). Several studies have unraveled the basis for γ_c dependency of lymphopoiesis (3–6), and two cytokines (IL-7 and IL-15) seem to exert dominant roles in this regard. IL-7 has been shown to play an essential role in maintaining early B and T cell precursors, and in sustaining naive αβ T cells once they exit the thymus and some types of CD4 memory T cells (7–9). In contrast, IL-15 is critical for NK and NKT development, whereas both IL-7 and IL-15 are involved in γδ T cell and CD8 memory cell homeostasis (3, 10–13).

Whereas the requirement for γ_c cytokines for lymphocyte homeostasis in vivo is clearly demonstrated, the evidence for essential roles of γ_c-cytokine signaling during T cell differentiation in vivo is less well defined. Previous studies have reported defective immune responses in mice deficient in some or all γ_c-dependent pathways (3, 14–21). However, unambiguously assigning roles for γ_c cytokines in the elaboration of immune responses is challenging, because these soluble factors have pleiotropic effects on the survival, proliferation and differentiation of several T cell subsets. Thus, γ_c deficiency affects not only the survival of naive T cells but also the function of regulatory T cells (22), and nontransgenic γ_c⁻ mice ultimately develop an autoimmune syndrome secondary to T cell antigen receptor (TCR) reactivity with environmental antigens that severely limits their use for assessing T cell immunity (17, 23) (24). In contrast, certain TCR transgenic mice on the recombinase-activating enzyme (Rag)-2 deficient background harbor “monoclonal” populations of naive T cells that have little environmental cross-reactivity, and thereby provide a means to study roles for γ_c cytokines in immune responses. Marilyn (Ml) mice develop naive CD4⁺ αβ T cells specific for the male peptide Dby/I-A^b complex (25). In the absence of γ_c, female Ml mice efficiently select CD4⁺ single positive thymocytes, and generate a small number of naive γ_c⁻ CD4⁺ T cells in the peripheral lymphoid organs, but show no signs of systemic autoimmunity. Nevertheless, Ml γ_c⁻ CD4⁺ T cells can proliferate normally in response to male antigen (25). Monoclonal Ml γ_c⁻ mice thereby provide a system to define roles for γ_c cytokines in the T cell differentiation process.

Ml CD4⁺ T cells can be activated to proliferate in vitro and in vivo, and differentiate into T helper 1 (T_H1)-like CD4⁺ effector T cells capable of secreting IFN-γ and rapidly rejecting male skin grafts in vivo (26, 27). We wished to assess whether γ_c expression conditioned this immune response, and if so, how. Splenic naive CD4⁺ T cells in Ml γ_c⁻ mice are dramatically reduced and fail to express the anti-apoptotic protein Bcl-2 (25). We rescued the survival defect in Ml γ_c⁻ CD4⁺ T cells using transgenic expression of Bcl-2 and assessed their capacity to differentiate in vivo. We found that whereas γ_c⁺ Ml CD4⁺ T cells progressively differentiate to full effectors that mediate skin graft rejection, skin-infiltrating γ_c⁻ CD4⁺ T cells had an abnormal phenotype and failed to reject the grafts. Further studies revealed a role for IL-15 in this process. These results identify a novel γ_c-dependent checkpoint in the tissues that is required for complete differentiation of effector CD4⁺ T cells.

Results

The γ_c Chain Is Essential for Marilyn CD4⁺ T Cell-Mediated Skin Graft Rejection.

To assess the importance of γ_c cytokines for the functional differentiation antigen-activated CD4⁺ T cells, we compared the fate of full-thickness male skin grafts placed on the flank of female Marilyn (Ml) transgenic mice that were either γ_c-competent or γ_c-deficient. Confirming previous reports (26, 27), we found that female Ml γ_c⁺ transgenic mice could rapidly and completely reject male skin grafts within a two-week period (Fig. 1 A and B). The rejection process was antigen-dependent, as female skin grafts were not rejected (data not shown) and no other lymphocyte subsets (B, CD8⁺ T, NK-T, γδ T) were present (all recipient mice and skin grafts were Rag2-deficient) indicating that the rejection required only CD4⁺ T cells. In contrast, male skin grafts placed on Ml γ_c⁻ female mice were accepted (> 120 days) despite the presence of male antigen-reactive cells in these hosts (Fig. 1A). Because peripheral naive CD4⁺ T cells in Ml γ_c⁻ female mice are markedly reduced (25), we generated Ml γ_c⁻ female mice expressing the human Bcl-2 Eμ2–25 transgene (28) to rescue the survival defect of the peripheral γ_c⁻ CD4⁺ T cells (29). Ml γ_c⁻ Bcl-2⁺ mice harbored similar numbers of naive (CD44^lo, CD62L^hi) CD4⁺ T cells as Ml γ_c⁺ mice (Ml γ_c⁺ mice: 2.52 ± 0.98 × 10⁶ cells; Ml γ_c⁻ Bcl-2⁺ mice: 3.3 ± 2.51 × 10⁶ cells; P = 0.45) that expressed normal levels of Vβ6, CD4 and had a normal mitochondrial potential (51). Still, Ml γ_c⁻ Bcl-2⁺ mice accepted male skin grafts (Fig. 1A) suggesting that graft tolerance was not due to a reduced T cell precursor frequency.

Fig. 1.

Skin graft acceptance in the absence of γ_c despite normal infiltration of activated CD4⁺ T cells. (A) Male skin from alymphoid (Rag2^−/−γ_c^−/−) mice was grafted on female Ml γ_c⁺ mice (Left), Ml γ_c⁻ mice (Center), and Ml γ_c⁻ Bcl-2⁺ mice (Right). Clinical outcome and aspects of the grafts at day 10 and day 120 are shown. Representative examples are shown (n = 10 for Ml γ_c⁺; n = 4 for Ml γ_c⁻; n = 9 for Ml γ_c⁻ Bcl-2⁺). (B) Equal numbers of mature CD4⁺ thymocytes from Ml γ_c⁺ Bcl-2⁺ or Ml γ_c⁻ Bcl-2⁺ mice were adoptively transferred to female Rag2^−/−γ_c^−/− recipients. Male skin was grafted 1 day later. Histological analysis using anti-CD3 antibody at day 10 after transplant is shown. One representative experiment of three is shown. (C) FACS analysis of skin-infiltrating T cells using a combination of CD4, Vβ6, CD44, and CD62L antibodies. CD62L and CD44 expression is shown on gated CD4⁺Vβ6⁺ T cells. One representative experiment of 10 is shown.

The γ_c Chain Is Redundant for Antigen-Specific CD4⁺ T Cell Proliferation and Migration of Activated T Cells into the Tissues.

Ml γ_c⁻ T cells proliferate normally in vivo in response to male antigen (25). Similarly, the absolute numbers of activated splenic CD4⁺ T cells increased in female recipients harboring either Ml γ_c⁺ Bcl-2⁺ or Ml γ_c⁻ Bcl-2⁺ CD4⁺ T cells (Ml γ_c⁺ Bcl-2⁺: 1.14 ± 0.9 × 10⁶ CD44^hi CD4⁺ T cells; Ml γ_c⁻ Bcl-2⁺: 0.48 ± 0.35 × 10⁶ CD44^hi CD4⁺ T cells) after skin grafting. Histological analyses showed that independently of their γ_c status, Ml CD4⁺ T cells progressively infiltrate skin grafts starting at day 6 after graft with a peak at day 10 after graft (Fig. 1B and data not shown). At day 10 after graft, male skin grafts on Ml γ_c⁺ or Ml γ_c⁺ Bcl-2⁺ female mice showed clear clinical (induration, necrotic foci, edema) and histological (inflammatory cell infiltration) signs with rejection by day 15 (Fig. 1 A and B and data not shown). Male grafts placed on Ml γ_c⁻ Bcl-2⁺ female mice showed some clinical signs of edema early in the second week after graft and were also similarly infiltrated by CD4⁺ T cells (Fig. 1B) suggesting that graft acceptance was not the consequence of a failure of γ_c⁻ T cells to migrate and infiltrate the male skin graft. The vast majority of Ml γ_c⁺ or γ_c⁻ CD4⁺ T cells present in the grafts had an activated phenotype (CD44^hiCD62L^lo; Fig. 1C) and no significant difference was observed in the absolute numbers of skin-infiltrating Ml CD4⁺ T cells between recipients harboring either Ml γ_c⁺ Bcl-2⁺ CD4⁺ T cells (9.4 ± 3.1 × 10⁴ CD4⁺ T cells) or Ml γ_c⁻ Bcl-2⁺ CD4⁺ T cells (9.6 ± 3.3 × 10⁴ CD4⁺ T cells; P = 0.2).

Recruitment and Roles of NK Cells and Inflammatory Cells in Ml-Mediated Skin Graft Rejection.

Because efficient differentiation and polarization of T_H1 T cells involves recruitment and activation of NK cells (30, 31), graft acceptance in Ml γ_c⁻ female mice could be secondary to their NK cell deficiency. We therefore adoptively transferred Ml γ_c⁺ Bcl-2⁺ thymocytes to immunodeficient hosts with (Rag2^−/−) or without (Rag2^−/−γ_c^−/−) NK cells and monitored rejection of male skin. Grafts were rejected by day 15 in hosts receiving Ml γ_c⁺ Bcl-2⁺ CD4⁺ T cells (Table 1) irrespective of whether the recipient harbored NK cells. In contrast, skin grafts were accepted in hosts receiving Ml γ_c⁻ Bcl-2⁺ CD4⁺ T cells despite the presence of host NK cells. Thus, host NK cells were neither essential nor sufficient for the rejection of male skin grafts by Ml CD4⁺ T cells.

Table 1.

Role of NK cells in graft rejection by γ_c⁺ or by γ_c⁻ MI CD4⁺ T cells

	Donor T cells
Host	MIγc+Bcl-2+	MIγγc− Bcl-2+
NK+	Rejection	Acceptance
NK−	Rejection	Acceptance

Thymocytes from MI γ_c⁺ Bcl-2⁺ or MI γ_c⁻ Bcl-2⁺ mice were adoptively transferred to NK cell-competent (Rag2^−/−) or -deficient (Rag2^−/−γ_c^−/−) female hosts. One day later, male skin grafts were surgically placed. Graft acceptance or rejection was followed up to 120 days. Four recipient mice of each genotype were analyzed.

Several types of hematopoietic cells are recruited to the inflamed skin grafts during the rejection process. We compared cellular infiltrates in nongrafted female skin (normal skin), in control male skin grafts placed on nontransferred Rag2^−/− recipients (surgical skin) and in skin grafts on Rag2^−/− recipients transferred with Ml γ_c⁺ or γ_c⁻ Bcl-2⁺ CD4⁺ T cells. As expected, normal skin showed no overt histological signs of inflammation and the few tissue-resident cells consisted mainly of CD11b macrophages and a small population of NK cells (Fig. 2 A–C). Whereas the surgical procedure itself induced some inflammation (Fig. 2 A and B), the extent of the inflammatory response paled in comparison with mice that received Ml Bcl-2⁺ CD4⁺ T cells. Remarkably, both γ_c⁺ and γ_c⁻ Ml Bcl-2⁺ CD4⁺ T cells induced a robust recruitment of CD11b⁺Gr-1⁺ granulocytes and CD11b⁺Gr-1⁻ macrophages to the skin grafts suggesting that the difference in rejection outcome was not related to an altered hematopoietic cells recruitment in the absence of γ_c (Fig. 2B). Using CX3CR1-GFP mice (32), we found that resident (Gr1⁻, CX3CR1^hi) and inflammatory (Gr1^hi, CX3CR1^lo) macrophages were similarly recruited in both cases (data not shown). NK cell recruitment (and phenotype; Fig. 2C and data not shown) was similar in recipients receiving Ml γ_c⁺ Bcl-2⁺ CD4⁺ T cells (3.9 ± 2.3 × 10⁴ NK cells) or Ml γ_c⁻ Bcl-2⁺ CD4⁺ T cells (7.3 ± 8.8 × 10⁴ NK cells; P = 0.3). Collectively, these results indicate that Ml γ_c⁻ CD4⁺ T cells are fully competent to recruit diverse inflammatory cells (NK cells, granulocytes, and macrophages) to the site of immune rejection.

Fig. 2.

Normal inflammatory cell recruitment by γ_c⁻ Ml CD4⁺ T cells. (A) Immunohistology of day 10 postgraft skin biopsies using an anti-F4/80 antibody. Analysis was performed on control skin (Left), “surgical skin” (male skin grafted to a control Rag2^−/− recipient; Center Left), male skin graft on Rag2^−/− recipient receiving Ml γ_c⁺ Bcl-²⁺ T cells (Center Right), and male skin graft on Rag2^−/− recipient receiving Ml γ_c⁻ Bcl-2⁺ T cells (Right). One of two independent experiments is shown. (B) FACS analysis of day 10 postgraft skin-infiltrating cells demonstrating presence of macrophages (CD11b⁺Gr1^lo) and neutrophils (CD11b⁺Gr1^hi). No significant difference in the percentage of infiltrating neutrophils (WT, 23 ± 12.6% versus γ_c⁻, 18.2 ± 7.5%; P = NS) was observed (n = 9 for each genotype). (C) FACS analysis showing the infiltration of CD3⁻NK1.1⁺ NK cells. B and C are representative of three independent experiments.

Spurious Effector CD4⁺ T Cell Differentiation in the Absence of γ_c.

T cell activation results in multiple phenotypic changes as naive cells differentiate to effectors (33). We compared the activation profile of splenic and skin-infiltrating T cells in Ml γ_c⁺ mice during the peak of immune rejection (day 10 after graft; Fig. 3A). Activated Ml γ_c⁺ Bcl-2⁺ CD4⁺ T cells (CD44^hiCD62L^lo) in the spleen had the CD27^hiLy6C^loNKG2D^loIL-12Rβ1^lo phenotype (Fig. 3B) consistent with their recent antigen encounter. In contrast, activated CD4⁺ T cells infiltrating the skin were more fully differentiated with a CD27^loLy6C^hiNKG2D⁺IL-12Rβ1⁺ phenotype. This result suggested a progressive differentiation of Ml CD4⁺ effector T cells after their initial antigenic stimulation in the secondary lymphoid organ (spleen) to the end-organ tissue (skin).

Fig. 3.

CD4⁺ T cells that differentiate in the absence of γ_c are spurious effectors. Ml γ_c⁺ Bcl-2⁺ or Ml γ_c⁻ Bcl-2⁺ thymocytes were adoptively transferred to female Rag2^−/−γ_c^−/− recipients, and male skin grafts were placed 1 day later. (A) Comparison of CD44 and CD62L profiles in spleen and skin Ml γ_c⁺ Bcl-2⁺ T cells at day 10 after graft. Activated effector (CD44^hiCD62L^lo) cells were gated as indicated. (B) Phenotype of spleen (dotted lines) and skin-infiltrating (solid lines) Ml γ_c⁺ Bcl-2⁺ effector CD4⁺ T cells. Mean fluorescence intensity is indicated. (C) γ_c expression conditions the phenotype of skin-infiltrating CD4⁺ T cells. Expression of indicated markers on electronically gated γ_c⁺ Bcl-2⁺ (black lines) and γ_c⁻ Bcl-2⁺ (shaded) CD4⁺Vβ6⁺ T cells is shown. Results are representative of three to six independent experiments. (D) Effector functions in skin-infiltrating CD4⁺ T cells were analyzed at day 10 after graft. Granzyme B (grzB) versus CD62L expression on gated CD4⁺ T cells is shown. (E) IFN-γ production from CD4⁺ T cells was analyzed after overnight restimulation of skin-infiltrating T cells with male APC. Results are representative of six independent experiments.

Activated CD44^hiCD62L^lo Ml γ_c⁻ CD4⁺ T cells present in the spleen and skin normally expressed low levels of CD28, CD40L, CD69 and PD-1 (Fig. 3C and data not shown). However, these cells expressed reduced levels of costimulatory molecules (ICOS and NKG2D) suggesting that these activated Ml γ_c⁻ CD4⁺ T cells were not true effector T cells. Moreover, skin-infiltrating Ml γ_c⁻ CD4⁺ T cells were Ly6C^lo and were mostly CD27^hi, whereas the corresponding WT cells had a Ly6C^hiCD27^lo phenotype typical of fully differentiated T cells (Fig. 3 B and C). These results indicate that T cells require γ_c signals to achieve a full effector phenotype.

The molecular mechanisms through which Ml CD4⁺ T cells provoke skin graft rejection could involve distinct effector mechanisms, such as FasL, perforin/granzymes, TNF-related apoptosis-inducing ligand (TRAIL) and/or IFN-γ (34–36). Skin-infiltrating Ml CD4⁺ T cells were FasL^lo and TRAIL^lo (data not shown) and we found that transfer of Ml γ_c⁺ CD4⁺ T cells into female hosts lacking Fas (or TRAIL-R) and grafted with Fas-deficient (or TRAIL-R-deficient) male skin grafts were normally rejected (data not shown). Activated Ml CD4⁺ T cells in the spleen and skin were positive for granzyme B (Fig. 3D), although they lacked inducible CD107 expression in vitro (data not shown). Approximately 20% of activated Ml γ_c⁺ CD4⁺ T cells were able to produce IFN-γ upon restimulation with male APC in vitro [Fig. 3E; supporting information (SI) Fig. 6]. Thus, Ml CD4⁺ T cells harbor an arsenal of inflammatory T_H1 effectors. Strikingly, granzyme B and IFN-γ effector pathways were essentially lacking in skin-infiltrating Ml γ_c⁻ CD4⁺ T cells (Fig. 3 D and E; SI Fig. 6). IFN-γ was important for skin graft rejection as neutralization of IFN-γ activity clearly prolonged skin graft survival and adoptively transferred IFN-γ-deficient Ml γ_c⁺ CD4⁺ T cells showed a markedly reduced capacity to reject male skin grafts (SI Fig. 7).

IL-15 Is Involved in the Differentiation of CD4⁺ T_H1 T Cell Effectors.

Dynamic modulation of γ_c-dependent cytokine receptor expression occurs during T cell activation. Naive CD4⁺ T cells are CD127⁺ (IL-7Rα) and CD122⁻ (IL-2Rβ), but this profile reverses after antigen stimulation so that effector CD4⁺ T cells are CD127⁻ and CD122⁺ (reviewed in ref. 37). Previous reports have suggested that CD127 down-regulation was dependent on an autocrine IL-2 feedback loop (38). We found that naive Ml γ_c⁺ CD4⁺ T cells were as expected CD127⁺ and CD122⁻, and became CD127⁻ and CD122⁺ after exposure to male antigen and activation in vivo (SI Fig. 8). Naive Ml γ_c⁻ CD4⁺ T cells were also CD127⁺ and CD122⁻, and became CD127⁻ and CD122⁺ after activation. Thus, CD127 and CD122 receptor modulation during CD4⁺ T cell differentiation occurs independently of the signaling through γ_c.

We next analyzed graft rejection in Ml female mice lacking CD122. Deficiency in this receptor had no obvious impact on intrathymic generation of Ml CD4⁺ thymocytes, or the peripheral maintenance of Ml CD4⁺ T cells (Fig. 4A; ref. 51). Similar numbers of splenic CD4⁺ CD44^hi cells were recovered in Ml and Ml CD122^−/− mice at day 10 after graft (WT, 1.2 ± 1 × 10⁶ cells versus CD122^−/−, 0.95 ± 0.7 × 10⁶ cells). However, skin graft survival was substantially prolonged in the absence of CD122 (up to 50 days) suggesting that IL-2 and/or IL-15 were required for Ml-mediated graft rejection (Fig. 4B). Comparing the phenotype of skin-infiltrating CD4⁺ T cells with or without CD122 revealed that down-regulation of CD27 and up-regulation of Ly6C were dependent on CD122 expression (Fig. 4C). Similar to γ_c⁻ effectors, skin-infiltrating Ml CD122^−/− CD4⁺ T cells expressed low levels of IFN-γ and granzyme B (Fig. 4 D and E) consistent with the delayed rejection. These observations indicate that a combination of IL-2 and/or IL-15 plays a role in the differentiation of Ml CD4⁺ T cell effectors in vivo.

Fig. 4.

Signaling through CD122 (IL-2Rβ) is required for Ml CD4⁺ T cell differentiation. (A) Absolute numbers of CD4⁺ T cells in thymus and spleen of WT and CD122-deficient Ml female mice. (B) Male skin graft survival is prolonged in the absence of CD122. Male skin grafts were placed on female mice of the indicated genotype (n = 5), and graft survival was monitored. (C) At 10 days after transplant, skin-infiltrating cells were isolated and analyzed by FACS. CD44, CD62L, CD27, Ly6C, and NKG2D expression are shown on electronically gated γ_c⁺ (black lines) and CD122^−/− (shaded histograms) CD4⁺Vβ6⁺ T cells. (D) FACS analysis of granzyme B (grzB) versus CD62L expression on gated CD4⁺ T cells is shown. (E) IFN-γ production from CD4⁺ T cells was analyzed after overnight restimulation of skin-infiltrating T cells with male APC. Results are representative of three independent experiments.

IL-15 is not required for CD4 homeostasis, and a defined role for IL-15 in CD4⁺ T cell differentiation in vivo has not been demonstrated. We found that Ml γ_c⁺ CD4⁺ T cells transferred to hosts lacking IL-15 were compromised in their capacity to reject male skin grafts, although they normally proliferated in the spleen and infiltrated the skin (Fig. 5 A and B). Activated Ml CD4⁺ T cells in this context had a nearly normal effector phenotype (CD44^hiCD62L^loCD27^lo), but showed reduced Ly6C expression and markedly reduced granzyme B levels (Fig. 5C). These results suggest an IL-15-derived signal is involved in the full differentiation of CD4⁺ T cell effectors in vivo.

Fig. 5.

CD4⁺ effector T cell differentiation requires host-derived IL-15 (A) Male skin graft survival was delayed in the absence of IL-15. Ml γ_c⁺ thymocytes were adoptively transferred to female Rag2^−/−γ_c^−/− (filled triangles; n = 5) or female Rag2^−/−IL-15^−/− (open squares; n = 4) recipients, and male skin grafts were placed 1 day later. We observed normal proliferation of Ml T cells in the spleen at day 10 (0.8 ± 0.23 × 10⁶ cells recovered) in the absence of host IL-15. (B) At 10 days after transplant, skin-infiltrating cells were isolated and analyzed by FACS. CD44, CD62L, CD27, Ly6C, and NKG2D expression are shown on electronically gated CD4⁺Vβ6⁺ T cells in the presence (black lines) or absence of host-derived IL-15 (shaded histograms). (C) FACS analysis of granzyme B (grzB) versus CD62L expression on gated CD4⁺ T cells is shown. (D) IFN-γ production from CD4⁺ T cells was analyzed after overnight restimulation of skin-infiltrating T cells with male APC. Between two and four recipient mice of each genotype were analyzed.

Discussion

In this report, we have begun to dissect the biological roles for γ_c cytokines in the differentiation of effector CD4⁺ T cells in vivo. Using the Marilyn TCR transgenic model of male skin graft rejection, we found that many aspects of the antigen-driven CD4⁺ T cell immune response proceed normally in the absence of γ_c, including clonal expansion, classical phenotypic changes associated with T cell activation and maturation, migration of activated T cells to inflamed tissues and T cell-mediated recruitment of hematopoietic cells to the site of the ongoing immune response. Nevertheless, Ml γ_c⁻ CD4⁺ T cells were ineffective in rejecting male skin, and demonstrated abnormal expression of several key costimulatory molecules (ICOS, NKG2D), differentiation antigens (Ly6C, CD27) and effector mediators (IFN-γ, granzyme B) in the tissues. Thus, CD4⁺ T cell differentiation in the absence of γ_c generates “spurious” T cell effectors, activated T cells with classical markers of differentiated T cells (CD44^hiCD62L^lo) but lacking effector function. Recent studies have suggested that differentiation from naive to effector states could occur in a stepwise fashion depending on the signals received in the secondary lymphoid organs or subsequently in the tissues (39). Our comparison of activated CD4⁺ T cells in the spleen versus the skin graft suggests that γ_c cytokines play a critical role in establishing the fully differentiated effector T cell phenotype (CD44^hiCD62L^loCD27^loLy6C^hi) and function (expression of IFN-γ, granzyme B). Together, our results define a quality control point in the process of T cell differentiation where γ_c cytokines are required to authenticate T cell effector functions in the tissues.

Which γ_c cytokines are required for differentiation of Ml effector CD4⁺ T cells? CD122 (a shared receptor for IL-2 and IL-15) is strongly up-regulated as T cells differentiate from the naive to effector states, and previous studies have implied a role for IL-2 and/or IL-15 in this process (15, 20, 21). CD122-deficient Ml T cells failed to normally reject male skin grafts, and the skin-infiltrating CD4⁺ T cells had the CD27^hiLy6C^lo phenotype with poor IFN-γ production capacity and failed to express granzyme B. CD122 was therefore essential for Ml T cells to achieve full functional competence and to avoid “spurious” T cell differentiation. Nevertheless, CD122-deficient Ml CD4⁺ T cells were still capable of skin graft rejection (although with strongly delayed kinetics) suggesting that effector mechanisms other than IFN-γ and granzyme B are active. The γ_c cytokines (other than IL-2/IL-15) that condition graft rejection remain to be identified but might include IL-7.

Previous studies failed to demonstrate defects in the homeostasis or differentiation of CD4⁺ T cells in IL-15-deficient mice (3), although in vitro experiments suggested that IL-15 can modify the function of CD4⁺ T cells activated (40, 41). We found that IL-15 was required for complete CD4⁺ T cell differentiation in vivo. In the absence of IL-15, Ml CD4⁺ T cell differentiation was abnormal (with a failure to up-regulate Ly6C and granzyme B expression) and skin graft rejection was delayed. Because transfer of Ml T cells to Rag2^−/−γ_c^−/− recipients (where all host cells are IL-15 nonresponsive but IL-15 is available to the differentiating CD4 T cell) results in normal Ml-mediated graft rejection, suggesting that IL-15 is provided by the microenvironment to the differentiating CD4⁺ T cell. This finding refocuses the attention on paracrine γ_c cytokine stimulation of differentiating T cells, and suggests that T cell interactions with IL-15-producing stromal cells or APCs might be essential for full T cell differentiation.

IL-2 and IL-15 trigger a similar intracellular signal transduction cascade in target cells because they share CD122 and γ_c receptors. Based on the transient expression of CD25 in activated T cells, it is likely that IL-2 has a role after antigen stimulation when T cells express the high-affinity IL-2 receptor and proliferating T cells are secreting IL-2. At later time points, CD25 expression is reduced (as well as IL-2 secretion), although CD122 expression on activated T cells remains high and cells may respond to IL-15. In this “early IL-2” versus “late IL-15” model, it is difficult to discern whether IL-2 and IL-15 transmit unique or redundant intracellular signals to the cells. Nevertheless, these dynamic changes in receptor expression identify biologically relevant “windows” of IL-2 versus IL-15 responsiveness.

Naive T cells receive primary signals from mature dendritic cells (DCs) in the lymph node, which trigger clonal T cell proliferation. Whereas the length of time necessary to “program” T cell proliferation in this context is not known, CD4⁺ T cells seem to clearly require multiple DC encounters to achieve a fully differentiated phenotype (42). The initial encounter with antigen-bearing DC within the lymph node (LN) might allow for TCR-mediated proliferation and CD127/CD122 receptor modulation. IL-2 could drive early “cell autonomous” T cell differentiation (via cell surface CD25) in the LN, whereas multiple DC interactions could provide an opportunity to receive additional signals. Activated T cells that leave the LN would then enter the tissues where IL-15 would complete their γ_c-cytokine-driven differentiation. The tissular sources of IL-15 [either secreted or “trans” presented through IL-15Rα (ref. 43)] are multiple: activated DC, macrophages, myoblasts, and epithelial cells may produce IL-15 (1, 44, 45) to support CD8⁺ T cell memory homeostasis (46).

The spurious state of T cell differentiation that occurs in the absence of γ_c signaling contrast with previous described states of anergy that occur during nonproliferative T cell receptor triggering (47). γ_c⁻ T cells proliferate normally and differentiate extensively, but fail to mature fully. This γ_c-dependent quality-control checkpoint assures the generation of “useful” T cell effectors. Spurious CD4⁺ Ml γ_c⁻ T cells do not manifest suppressor activity, as mice bearing mixtures of γ_c⁺ and γ_c⁻ Ml CD4⁺ T cells reject male skin grafts normally (data not shown). Ml γ_c⁻ T cells are PD-1^lo and thus differ from “exhausted” PD-1⁺ effectors characterized by chronic viral stimulation (48). The graft tolerance achieved in the absence of γ_c resembles “innocuous inflammation” (49), but differs from that observed using combined costimulatory molecule blockade in which inflammatory infiltrates are strongly reduced (50). Modifying γ_c signaling pathways (including IL-15) during tissue transplantation or infections may lead to therapeutic benefit through a reversible inflammatory reaction.

Experimental Procedures

Animals.

Marilyn (Ml) TCR transgenic (Tg) mice specific for the male antigen HY Dby on the Rag2^−/− or Rag2^−/−/γ_c^−/− (C57BL/6) background have been described (25). Human Bcl-2 Tg mice B6.Cg-Tg^{(BCL2)25Wehi/J} (28) were backcrossed to Ml Rag2^−/− or Rag2^−/−/γ_c^−/− mice to generate Ml Bcl-2⁺ mice with or without the γ_c chain. IFN-γ-deficient mice (B6.129S7^Ifngtm1Ts/J) and CD122-deficient mice (B6^{Il2rbtm1Mak/J}) were used to generate monoclonal Ml IFN-γ-deficient or CD122-deficient mice. B6 mice doubly deficient in Rag2/IL-15 have been described (13). Rag2/Fas^lpr mice, Rag2/TRAIL-R mice, and Rag2/CX3CR1-GFP mice were provided by S. Ezine, F. Rieux-Laucat, and F. Geissmann, respectively (Faculté Necker, Paris, France). Donor and recipient mice were used between 4 and 8 weeks of age. Animals were kept under pathogen-free conditions in the animal facilities at the Institut Pasteur, and all animal experiments were approved by a local committee and in accordance with French law.

Adoptive Transfer Experiments.

Thymocytes from female Ml mice were used as a source of naive CD4⁺ T cells, and the equivalent of 2 × 10⁶ single-positive CD4⁺ thymocytes were transferred i.v. to unconditioned recipient mice via the retro-orbital chamber. Skin grafts were performed 1 day later.

Skin Transplantation.

Full-thickness male skin grafts from alymphoid Rag2^−/−γ_c^−/− or Rag2^−/−IL-15^−/− donors were performed as described (27). Surgical bandages were removed on postoperative day 7, and the clinical condition of the grafts was assessed on a daily basis. Rejection was defined as complete graft necrosis. Histological biopsies and extraction of skin-infiltrating cells were performed at day 10 after graft, which corresponds to the peak of rejection in wild-type mice. Mice were treated with neutralizing antibodies to IFN-γ (six injections of 200 μg of XMG1.2) during the first 2 weeks after graft.

Immunohistochemical Analysis.

Graft biopsies were frozen in OCT compound (Sakura, Zoeterwoude, The Netherlands), and 8-mm cryostat sections were prepared and fixed with 4% paraformaldehyde (Merck, Darnstadt, Germany) for 15 min at room temperature. Sections were permeabilized with 0.1% Triton X-100 (Sigma, Saint-Quentin, France) for 10 min and then incubated in 0.3% H₂O₂ for 5 min, before blocking in 10% FCS. Sections were incubated with anti-CD3 (clone 17A2; BD Biosciences, San Diego, CA) or anti-F4/80 (clone CI-A3-I; Caltag, Burlingame, CA) for 2 h, washed extensively, and then incubated with goat anti-rat HRP-conjugated secondary antibodies (Dako, Glostrup, Denmark) for 30 min before addition of the substrate diaminobenzidine (Dako). Sections were counterstained with Mayer's hematoxylin (Merck) and photographed by using a Zeiss Axiophot at ×100 magnification.

Cell Isolation and Flow Cytometric Analysis.

Single cell suspensions from thymus and spleen were prepared as described (13). Skin graft-infiltrating cells were extracted by using 100 units/ml collagenase (Sigma) and 5 units/ml DNase in RPMI medium 1640/2% FCS with gentle agitation for 1 h at 37°C. During this time, the skin biopsy tissue fragments were dissociated by gentle mechanical disruption. The resulting cell suspension was filtered through a 100-μm mesh, and the recovered cells were analyzed by FACS as described (13). Dead cells were excluded by using TO-PRO3 or Sytox Green (Molecular Probes, Leiden, The Netherlands). FACS acquisitions were performed by using Calibur or Canto (BD Biosciences) analytical flow cytometers, and data sets were analyzed by using Flowjo software.

Intracellular Staining.

Graft-infiltrating cells were stained for surface markers (CD4, CD44, CD62L) and then fixed with Cytofix/Cytoperm solution (BD Biosciences) and permeabilized with Perm/Wash solution (BD Biosciences) according to manufacturer's instructions. Granzyme B was detected by using PE-conjugated GB12 (Caltag), and IFN-γ was detected by using PE-conjugated XMG1.2 (BD Biosciences). For IFN-γ detection, graft-infiltrating cells were precultured overnight with male CD3ε^−/− splenocytes (1/1 ratio) with Brefeldin A addition (10 μg/ml) during the final 4 h of culture.

Abbreviations

TCR

T cell antigen receptor

NK

natural killer

γ_c

common γ chain

Rag

recombinase-activating enzyme

T_H1

T helper 1

LN

lymph node

DC

dendritic cell

TRAIL

TNF-related apoptosis-inducing ligand.