Naive tumor-specific CD4⁺ T cells differentiated in vivo eradicate established melanoma

TRP-1–specific CD4⁺ TCR Tg mice were created on a C57BL/6 background that contains TRP-1 as a self-antigen (Muranski et al., 2008). These mice develop vitiligo (depigmentation) slowly with age (Fig. S1 A) but can develop vitiligo rapidly with genetic removal of Foxp3 (denoted Foxp3^sf), which is required for development of T reg cells that suppress autoimmune disease (Kim et al., 2007; Fig. 1 A and Fig. S1 B). TRP-1–specific CD4⁺ TCR Tg mice, which express TRP-1 antigen, have activated T cells that are characterized by low expression of CD62L and high expression of CD44 on TRP-1 CD4⁺ T cells in vivo (Fig. S1 C, top). Because we wanted to study naive CD4⁺ T cells, Tyrp1^+/+ TRP-1 CD4⁺ Tg mice were crossed with white-based brown mutation mice (Tyrp1^B-w) containing a radiation-induced inversion interrupting the tyrp1 gene (Smyth et al., 2006). RT-PCR confirmed that no Tyrp1 messenger RNA (mRNA) was present in Tyrp1^B-w mice, and Tyrp1^B-w mice that express the Tg TCR do not develop vitiligo (unpublished data). These mice have a light-brown coat color (Fig. S1 D) and were further crossed with RAG^−/− mice to generate naive monoclonal TRP-1–specific CD4⁺ T cells (Fig. S1 C, bottom). This mouse model is unique in that it develops T cells specific for TRP-1 that are both Foxp3⁻ and Foxp3⁺ on a RAG^−/− background. The expression of Foxp3 is at the natural physiological ratio of 5–15% (Fig. S1 E).

Once TRP-1–specific CD4⁺ TCR Tg mice on a Tyrp1^B-wRAG^−/− background were obtained (hereafter denoted as TRP-1 CD4⁺ T cells), TRP-1 CD4⁺ T cells were transferred into nontumor-bearing RAG^−/− mice and observed for signs of specific immunity (e.g., depigmentation). Indeed, ∼5–6 wk after transfer, the mice developed progressive depigmentation as indicated by coat color changes from black to white (Fig. 1 B). Depigmentation progressed until the mice became completely white.

To determine if adoptive transfer of TRP-1 CD4⁺ T cells could mediate the regression of established tumors, WT, RAG^−/−, and irradiated WT mice (5 Gy) were given 2 × 10⁵ B16 tumor cells s.c. On day 7, after tumor was visible and palpable, sorted TRP-1 CD4⁺ T cells from 2 × 10⁵ Tyrp1^B-wRAG^−/− mice were transferred i.v. by tail vein injection into tumor-bearing mice. Adoptive transfer of TRP-1 CD4⁺ T cells into lymphopenic mice caused infiltration of tumor-specific T cells into and regression of established tumors (Fig. 1, C and D; and Fig. S1 F). Irradiated WT mice had a transient antitumor response, indicating a temporary lymphodepletion as seen in mice and patients that are lymphodepleted (Dudley et al., 2005; Gattinoni et al., 2005a; Zhang et al., 2005).

Because we wanted to study the role of CD4⁺ T cells in tumor immunity, we focused on RAG^−/− mice because these mice contain no CD8⁺ T cells, B cells, or NKT cells, all of which have been described in tumor immunity (Dougan and Dranoff, 2009). Complete regression of B16 melanoma was observed in RAG^−/− mice (Fig. 1 D). RAG^−/− mice that received 5 Gy irradiation to deplete cytokine sinks (Gattinoni et al., 2005a) and adoptive cell transfer of TRP-1 CD4⁺ T cells also had complete regression. As seen in nontumor-bearing RAG^−/− mice, depigmentation was present in tumor-bearing RAG^−/− mice at ∼5–wk after treatment (Fig. 1 E). Tumor regression was dramatic, with large established tumors regressing to a small scar (Fig. 1 F). RAG^−/− mice do not have endogenous CD8⁺ T, B, or NKT cells; therefore, this therapy was completely independent from these cell types. Because TRP-1 CD4⁺ T cells came from Tyrp1^B-wRAG^−/− mice, there was no concern for contamination from other cells that may contribute to tumor therapy.

To determine whether tumor treatment in lymphopenic hosts was specific, we sorted “open-repertoire” CD4⁺CD25⁻ T cells from WT mice and transferred 10⁶ CD4⁺ T cells into RAG^−/− mice on day 7 after tumor inoculation (Fig. 1 G). CD4⁺CD25⁻ T cells have virtually no Foxp3⁺ T cells and have been shown to help tumor-specific CD8⁺ T cells maintain treatment of established tumors without the addition of exogenous cytokines (Antony et al., 2006). However, by themselves, CD4⁺CD25⁻ open-repertoire T cells did not affect tumor growth in lymphopenic mice. Therefore, we conclude that TRP-1 CD4⁺ T cells are specific for their tumor/self-antigen in vivo and that the antitumor effect is independent of CD8⁺, B, and NKT cells.

Because it has been shown that in vitro differentiated Th17 tumor-specific T cells were superior to other CD4⁺ T cell lineages when treating established tumors (Muranski et al., 2008), we wanted to determine the in vivo default differentiation program that leads to strong tumor regression by TRP-1 CD4⁺ T cells. After 1 wk in vivo, adoptively transferred CD4⁺ T cells were FACS sorted from tumor-draining LNs and the spleens of mice undergoing tumor regression and analyzed by gene microarray. These cells were compared with FACS-sorted naive TRP-1 CD4⁺ T cells. Only genes with a twofold or greater difference were displayed. As shown in Fig. 2 A, TRP-1 CD4⁺ T cells transferred into a lymphopenic environment defaulted to a CD4⁺ Th1 program, as indicated by high expression of Tbet, Ifng, Il18r, Il2r, Il12r (Th1 priming), Il23r (Th1 memory; Robinson and O’Garra, 2002), Il27r (early Th1 priming), Tim3, Cxcr3, Ccr2, and Ccr5 mRNA, with evidence of some Th17-associated genes (Il22, Il23r, and Rora) and Th2 genes (Il10 and gata3) but not Ccr6, Il4, Il5, Il17a, Il17f, or Rorc. Genes for TCR signaling were up-regulated in the spleen (Lck and Zap70) but not in the LN, probably indicating recently activated trafficking T cells, suggested by increased Cd69, Cd5 (strong TCR signals), and Lat expression. Jak3 was up-regulated in TRP-1 T cells, indicating that γ_c signaling was active. Stat4 was highly expressed, demonstrating that IL-12, and possibly IL-23, signals were potentiating IFN-γ production. Co-activation genes, including 41BB and ICOS, were highly up-regulated. Both are involved in enhancing the activation of T cells (Tamada and Chen, 2006; Stephan et al., 2007; McNamara et al., 2008). Genes related to CTL effector functions were increased; Gzmb, Gzmc, Gzmd, and Gzmk were highly up-regulated, with only Gzmb and Gzmc staying at high levels in the spleen. Fasl was also up-regulated.

Genes that regulate the expansion of T cells by controlling T cell activation in a negative manner—TgfβrI, Ctla-4, Cish, Il10, and Socs2—were also up-regulated (Fig. 2 A). High IFN-γ levels were confirmed in the serum of lymphopenic mice undergoing tumor treatment but not in WT mice or untreated lymphopenic RAG^−/− mice (Fig. 2 B). CXCR3 and ICOS, both Th1-associated molecules, were also expressed by TRP-1 CD4⁺ T cells (Fig. 2, C and D).

Next, we looked at effector molecules in TRP-1 CD4⁺ T cells and found perforin, granzyme B, and LAMP-1 (CD107a), all of which are involved in degranulating cytotoxic T cells (Fig. 3). Because TRP-1 CD4⁺ T cells expressed granzyme B on gene array (Fig. 2) and in the spleen (Fig. 3), these data suggest that the antitumor CD4⁺ T cells are cytotoxic. Thus, naive TRP-1 CD4⁺ T cells differentiated into a Th1 phenotype in vivo and expressed cytotoxic T cell–associated genes and molecules that may assist in antitumor immunity.

CXCR3, CCR2, and CCR5 are chemokine receptors that help draw Th1 CD4⁺ T cells into inflamed tissues. Because we saw high chemokine receptor expression on TRP-1 CD4⁺ T cells (Fig. 2, A and C), we checked the serum of lymphopenic and nonlymphopenic mice for the specific ligands and found high expression of IFN-γ–inducible CXCL9 and CXCL10 (the ligands for CXCR3), as well as CCL2 (the ligand for CCR2) and CCL11 (the ligand for CCR2 and CCR5), in lymphopenic mice during tumor regression (Fig. 2 E). CCL3 (the ligand for CCR5) was also moderately high in the serum of lymphopenic mice during tumor regression (unpublished results). CCL2 is known to draw inflammatory monocytes from the blood into LN and activate Th1 CD4⁺ T cells (Nakano et al., 2009). We saw accumulation of inflammatory monocytes (CD11b^highGR-1⁺) in the spleen, LN, and tumors of treated groups (Fig. S2). These cells are class II⁺ and are known to secrete IFN-γ that may help aid Th1 differentiation (Nakano et al., 2009). Other chemokine receptors were up-regulated on TRP-1 CD4⁺ T cells, including CXCR6, which is involved in trafficking to the spleen, and CCR9 (Fig. 2 A). Lastly, a multitude of inflammatory chemokines were differentially expressed by TRP-1 CD4⁺ T cells in the LN and spleen (Fig. 2 A). Therefore, TRP-1 CD4⁺ T cells activated by lymphopenia-induced proliferation may help trigger Th1 chemokines, which could enhance immunotherapy by recruiting inflammatory monocytes and CXCR3⁺CCR2⁺CCR5⁺ TRP-1 CD4⁺ T cells to appropriate sites for activation and tumor infiltration.

It is known that MHC class I is up-regulated on tumor cells in vivo in the presence of IFN-γ⁺ tumor-specific CD8⁺ T cells (Palmer et al., 2008). However, it is not known whether IFN-γ enhances MHC class II expression on established tumors in vivo. Because we saw high levels of IFN-γ expression in the serum and on gene microarray, we evaluated the expression of MHC class II in the tumor microenvironment. 1 wk after adoptive transfer with TRP-1 CD4⁺ T cells, lymphopenic mice were sacrificed and tumors were imaged for MHC class II expression by confocal microscopy. Lymphopenic mice that had received no TRP-1 CD4⁺ T cells had no evident expression of MHC class II on tumor cells in vivo. However, in mice that had received TRP-1 CD4⁺ T cells, MHC class II expression was highly expressed on tumor cells in vivo (Fig. 4 A). To our knowledge, this is the first time class II up-regulation in vivo by B16 melanoma has been reported. This may be similar to the mechanism of induction of MHC class II reported in certain autoimmune diseases (Belfiore et al., 1991).

To determine if host MHC class II was important in activating TRP-1 CD4⁺ T cells in mice with MHC class II^+/+ tumors, tumor-bearing irradiated and nonirradiated MHC class II^−/− mice were treated with naive TRP-1 CD4⁺ T cells. Mice were irradiated to mimic the lymphopenic environment. Tumor treatment in both groups was comparable to the no-treatment groups (Fig. 4 B), suggesting, as previously shown, that MHC class II is needed to prime naive CD4⁺ T cells (Beutner and MacDonald, 1998). This implies that MHC class II expression on the tumor was not sufficient to activate naive CD4⁺ T cells. However, in vitro activated TRP-1 CD4⁺ T cells can treat MHC class II⁺ tumors in MHC class II^−/− mice, and anti-class II antibodies can block this effect (see Quezada et al. in this issue), showing that the function of activated CD4⁺ T cells is independent of host class II.

Next, to determine if tumor antigen alone is enough to activate TRP-1 CD4⁺ T cells, tumor-bearing female Tyrp1^B-wRAG^−/− mice, which express no TRP-1 antigen, were treated with TRP-1 CD4⁺ T cells. In the absence of self-antigen in the periphery, tumor treatment was complete and irradiation did not affect the antitumor response (Fig. 4 C). Even though the tumor may be able to express class II, these experiments demonstrate that T cell activation by host MHC class II⁺ cells may be essential for the success of treatment in the lymphopenic setting. These experiments also show that the tumor/self-antigen, TRP-1, expressed only by the tumor, is adequate to activate naive TRP-1 CD4⁺ T cells transferred into Tyrp1^B-wRAG^−/− mice.

IFN-γ has been attributed to the mechanism of action by which CD4⁺ T cells eradicate tumors (Mumberg et al., 1999; Nishimura et al., 1999; Corthay et al., 2005; Muranski et al., 2008). We administered anti–IFN-γ antibodies (500 µg/mouse) once, at the time of adoptive cell transfer, and found that anti–IFN-γ antibodies were able to attenuate the antitumor response in lymphopenic mice (Fig. 4 D).

To determine why adoptive transfer of T cells resulted in activation of CD4⁺ T cells in RAG^−/− mice, we looked at CD11c⁺ DC activation status because these cells express high levels of MHC class II in vivo. We found, after adoptive cell transfer with TRP-1–specific CD4⁺ T cells, that CD11c^high DC expressed increased levels of MHC class II and CD86 when compared with nontreated groups (Fig. 4 E). Collectively, these experiments suggest that adoptively transferred TRP-1 CD4⁺ T cells become activated on host MHC class II⁺ cells and may eradicate tumors through an IFN-γ–dependent pathway that may be linked to MHC class II expression on the tumor.

To determine how lymphopenia affects long-term maintenance and memory formation of Th1 CD4⁺ T cells, we analyzed TRP-1 CD4⁺ T cells from mice undergoing long-term tumor therapy (∼270 d). It may be possible that in vivo differentiation might bestow upon T cells a signal that allows them to persist and become superior antitumor CD4⁺ T cells. Therefore, we looked at long-term maintenance and found that adoptively transferred TRP-1 CD4⁺ T cells can maintain tumor regression and persist at relatively high levels 270 d after transfer (Fig. 5, A and B). We also looked at memory formation and activation markers that differentiate between effector memory T cells (TEMs) and central memory T cells (TCMs). CD62L and CD44 were high and low, respectively, as expected before transfer of CD4⁺ T cells (Fig. 5 C). Naive T cells were also IL-7Rα^high (Fig. 5 D). After 1 wk, CD4⁺ T cells became activated as shown by the phenotype CD62L^low, CD44^high, and IL-7Rα^low (Fig. 5, C and D). Mice were analyzed for adoptively transferred T cells again 270 d later. CD4⁺ T cells remained at high levels and were mostly TEM (CD62L^low and CD44^high), but ∼12% of CD4⁺ T cells were TCM (CD62L^high and CD44^high; Fig. 5 C).

To assess whether TRP-1 self-reactive T cells converted to T reg cells over this time span, we looked at Foxp3 expression in adoptively transferred cells and found that in long-term responder mice, Foxp3 expression remained stable at 5–15% (Fig. S3). Mice also became completely depigmented (Fig. 5 E). Lastly, granzyme B, IFN-γ, and TNF were evident in TRP-1 CD4⁺ T cells 120 d after transfer, indicating that they were effector T cells (Fig. 5 F). Therefore, tumor-specific CD4⁺ T cells differentiated in vivo can persist for long periods of time and appear to continue to clear antigen based on the progressive depigmentation, granzyme B expression, and TEM phenotype. TRP-1 CD4⁺ TCM cells may also renew the TRP-1 CD4⁺ TEM pool over time.

To determine host mechanisms that affect the activation and persistence of adoptively transferred CD4⁺ T cells, we transferred TRP-1 CD4⁺ T cells into RAG^−/−γ_c^−/− hosts, which lack NK cells in addition to T, B, and NKT cells (Cao et al., 1995). Previously, it was shown that treatment of a 1-d tumor was hindered in RAG^−/−γ_c^−/− mice (Perez-Diez et al., 2007), and this was attributed to a lack of NK cells. In agreement with the previous study, we found that tumor therapy was hindered even when supplemented with irradiation (Fig. 6 A). To examine this further, we depleted NK cells in tumor-bearing RAG^−/− mice starting on days 0 and 7 after tumor challenge, 1 d after ACT, and then weekly. TRP-1 CD4⁺ T cells were transferred into tumor-bearing RAG^−/− mice on day 8. We found that NK1.1 depletion (1 mg of anti-NK1.1/mouse) did not affect tumor treatment in a negative manner (Fig. 6 B). We also irradiated RAG^−/− mice with 5 Gy. The results were similar to those associated with NK cell depletion (Fig. 6 A). Next, we transferred 5 × 10⁶ sorted NK cells into RAG^−/−γ_c^−/− mice along with TRP-1 CD4⁺ T cells. Treatment was similar to RAG^−/−γ_c^−/− mice with TRP-1 CD4⁺ T cells alone (Fig. 6 C).

NK cells are required for efficient priming and activation of CD4⁺ T cells in vivo (Martín-Fontecha et al., 2004). We found that NK cells entered the LN readily in tumor-bearing RAG^−/− mice receiving CD4⁺ T cells but not in RAG^−/−γ_c^−/− mice, untreated RAG^−/− mice, or NK-depleted RAG^−/− mice (Fig. 6 D). These data suggest that NK cells are not required for activation of naive tumor-specific CD4⁺ T cells in vivo.

We were perplexed as to why RAG^−/−γ_c^−/− mice failed to control tumor immunity even when WT NK cells were present. TRP-1 CD4⁺ T cells were analyzed 1 wk after transfer from RAG^−/− and RAG^−/−γ_c^−/− hosts. Expression of CD62L was low on both populations, indicating activation in vivo (Fig. 7 A). However, CD122, ICOS, and CD25 were expressed at lower levels on TRP-1 CD4⁺ T cells from RAG^−/−γ_c^−/− hosts when compared with TRP-1 CD4⁺ T cells from RAG^−/− mice (Fig. 7 A). Consistent with these findings, IFN-γ and the CXCR3 ligand CXCL9, but not CXCL10, were lower in the serum of RAG^−/−γ_c^−/− hosts (Fig. 7 B and not depicted). Surprisingly, 1 wk after therapy, there were more TRP-1 CD4⁺ T cells in RAG^−/−γ_c^−/− hosts than in RAG^−/− hosts (Fig. 7 C), which is in agreement with recent findings by others (Guimond et al., 2009). We therefore looked at other time points in vivo and found that TRP-1 CD4⁺ T cells failed to persist after 4 wk in RAG^−/−γ_c^−/− lymphopenic hosts (Fig. 7 C). Looking at other parameters, we found that TRP-1 CD4⁺ T cells from RAG^−/− mice produced IFN-γ, but not IL-17, when stimulated ex vivo. However, TRP-1 CD4⁺ T cells from RAG^−/−γ_c^−/− mice failed to produce IFN-γ or IL-17 (Fig. 7 D). Because failure to produce IFN-γ may be related to their failure to differentiate, we stained for the Th1 transcription factor Tbet and found that expression was considerably lower in TRP-1 CD4⁺ T cells from RAG^−/−γ_c^−/− hosts (Fig. 7 E). Further exploration showed that CD11c^high DCs expressed less MHC class II and CD80 but similar levels of CD40 in RAG^−/−γ_c^−/− hosts when compared with RAG^−/− hosts (Fig. 7 F). Lastly, it was possible that transferred TRP-1 CD4⁺ T cells could have converted to T reg cells, which lead to tumor progression, but that was not evident beyond 4 wk after therapy (Fig. S4).

Collectively, the data suggest that γ_c signaling on host cells may be required for proper activation of naive CD4⁺ T cells. This may occur through IL-12 signals that increase class II and CD80 expression on host DCs (Ohteki et al., 2001, 2006; Terme et al., 2008). Consistent with this, MHC class II and CD80 were decreased in RAG^−/−γ_c^−/− hosts receiving TRP-1 CD4⁺ T cells. These experiments suggest that lack of γ_c signaling, possibly on host DC, in RAG^−/−γ_c^−/− mice might hinder proper activation of adoptively transferred naive TRP-1 CD4⁺ T cells, which eventually leads to their failure to survive and maintain treatment of established tumors.

ACT using a patient’s own immune cells to treat established cancer has been highly successful therapy against melanoma (Rosenberg et al., 2008) and is increasingly being used against other cancers as well (Stephan et al., 2007; Pule et al., 2008). ACT using CD4⁺ T cells was recently attempted in a patient with metastatic melanoma using NY-ESO-1–specific CD4⁺ T cells expanded in vitro in IL-7 and IL-2 (Hunder et al., 2008). Surprisingly, adoptive transfer therapy with CD4⁺ T cells led to objective cancer regression without prior vaccination or lymphoconditioning. Although this was reported in only one patient, the findings support the possibility of treating established tumors without vaccination or radiation therapies that may cause extended morbidity in patients with prolonged and severe sickness. In the prior study, epitope spreading was reported as the likely cause of the enhanced treatment, indicating that CD4⁺ T cells helped a CD8⁺ T cell response to the established tumor as shown in preclinical models (Marzo et al., 2000; Antony et al., 2006). In this paper, we describe a CD8⁺ T cell– and vaccine-independent therapy in which in vivo differentiated Th1 CD4⁺ T cells recognize self and eradicate established tumors.

Although we focused on a naive T cell– and a vaccine-independent therapy, we do not advocate that vaccines are ineffective or that in vitro differentiation of T cells should not be attempted. Recently, it was shown that a tumor vaccine was required to enhance the immunotherapy of cancer when using in vitro–activated antitumor CD8⁺ T cells and CD4⁺ Th cells (Overwijk et al., 2003; Antony et al., 2005). Vaccine independence was seen only after extreme radiation therapy (9 Gy) before ACT or in vitro programming of TEM cells (Wrzesinski et al., 2007; Klebanoff et al., 2009). The importance of activating CD4⁺ T cell help with a tumor vaccine has also been strengthened of late. A study using a modified TRP-1 vaccine could induce either tumor immunity or autoimmunity depending on the modification in the vaccine (Engelhorn et al., 2006). When the modified vaccine activated CD4⁺ T cell help through enhanced processing of MHC II peptides, tumor immunity was induced (Engelhorn et al., 2006). This new strategy to stimulate CD4⁺ T cells against self emphasizes the need to understand how CD4⁺ T cells become activated in vivo.

In this study, we demonstrate that naive CD4⁺ T cells transferred into lymphopenic hosts differentiated into Th1 cytotoxic T cells expressing the hallmark genes associated with effector T cells—Tbet, IFN-γ, CXCR3, granzyme B, perforin, and LAMP-1—and caused the regression of established tumors without the aid of CD8⁺ T, B, NK, or NKT cells. CD4⁺ T cells have been shown to play a role in cancer immunotherapy through a multitude of mechanisms; CD4⁺ T cells can help cytotoxic CD8⁺ T cells eradicate tumors (Hung et al., 1998; Nishimura et al., 1999; Toes et al., 1999; Sutmuller et al., 2001; Behrens et al., 2004a,b; Antony et al., 2005; Engelhorn et al., 2006; Brandmaier et al., 2009), and CD4⁺ T cells can condition a DC through CD40L to enhance CD8⁺ CTL function (Bennett et al., 1998; Kalams and Walker, 1998; Ridge et al., 1998; Diehl et al., 1999), but the mechanisms by which CD4⁺ T cells cause eradication of established tumors by themselves are still unknown.

Previously it was shown that CD4⁺ T cells could prevent the growth of tumors by antiangiogenesis through the activation of IFN-γR on nonhematopoietic cells. However, this action was not shown for large vascularized tumors in a treatment and adoptive cell transfer model and it was not clear whether CD8⁺ T cells were involved (Qin and Blankenstein, 2000). Rejection of small tumors was also shown to be an indirect process through IFN-γ–dependent activation of macrophages or by eosinophils and NK cells (Hung et al., 1998; Corthay et al., 2005). We noted high levels of CCL11 in the serum during treatment and CCL24 expression by TRP-1 CD4⁺ T cells, which are known to recruit eosinophils, so their role cannot be excluded. Indirect effects of CD4⁺ T cells have also been noted in a study using Tg Marylyn CD4⁺ T cells, which recognize the male antigen H-Y. MHC class II⁻ tumors were amenable to treatment and this was attributed to NK cells (Perez-Diez et al., 2007). In this paper, we do not find a role for NK cells in this model.

Not surprisingly, with adoptive cell transfer therapies using in vitro differentiated CD4⁺ T cells to treat large vascularized tumors, IFN-γR^−/− mice were shown not to respond to treatment (Muranski et al., 2008), and anti–IFN-γ diminished the antitumor effect (Nishimura et al., 1999; Muranski et al., 2008). However, IFN-γ^−/− mice were able to respond to treatment with in vitro differentiated CD4⁺ T cells, indicating that the innate source of IFN-γ (NK cells) was not needed (Qin and Blankenstein, 2000; Muranski et al., 2008). Because we saw increased MHC class II expression in the tumor microenvironment, it is possible that IFN-γ from CD4⁺ T cells may be needed to increase MHC class II expression on tumor cells, which allows a direct interaction between T cells and cancerous cells to occur (Quezada et al., 2010). Direct killing of tumors is possible, as indicated by high granzyme B expression by TRP-1 CD4⁺ T cells on gene microarray and expression of perforin, granzyme B, and LAMP-1 in TRP-1 CD4⁺ T cells isolated from the spleen. IFN-γ may be working indirectly at a local level through macrophages expressing NO, TNF, ROS, and NKG2D receptors (Kahn et al., 1991; Raulet and Guerra, 2009). IFN-γ released by differentiated CD4⁺ T cells may be so extremely high in the tumor microenvironment that the tumor vasculature expressing IFN-γ receptors undergoes anti-angiogenesis. The presence of specific disease, indicated by depigmentation and uveitis (Muranski et al., 2008), supports the theory of a direct killing mechanism against MHC class II⁺ cells. Inflammatory monocytes in the tumor microenvironment may also play an indirect role in the process. Another unappreciated role of IFN-γ may be to induce the chemokines CXCL9 and CXCL10 (Luster et al., 1985), which may be required to attract CXCR3⁺ antitumor CD4⁺ T cells into the tumor site. Most importantly, all of these mechanisms may not be mutually exclusive.

Why then does tumor therapy fail in RAG^−/−γ_c^−/− mice? CD4⁺ T cells need to become properly activated cytotoxic T cells, which is not dependent on NK cells. RAG^−/−γ_c^−/− mice express low levels of MHC class II and CD80 on CD11c^high cells, both of which may be needed to properly activate CD4⁺ T cells in vivo. The availability of MHC class II on APC has already been shown to be critical in regulating the homeostatic proliferation of CD4⁺ T cells (Kassiotis et al., 2003). Therefore, removing DC cells by high-intensity lymphodepleting regimens may harm ACT with CD4⁺ T cells. As seen in other studies, adding these cells back with lymphodepleting regimens may enhance immunotherapy (Dubsky et al., 2007; Palucka et al., 2007; Banchereau et al., 2009).

γ_c signaling is part of the IL-2 cytokine family, which includes IL-4, IL-7, IL-9, IL-15, and IL-21. Increased IL-7 and IL-15 levels in the host have been shown in the clinic and in animal models to be the mechanisms by which lymphodepleting regimens enhance adoptive immunotherapy (Gattinoni et al., 2005a; Dudley et al., 2008; Guimond et al., 2009). However, it has not been shown in a reciprocal manner that the host needs γ_c signaling for efficient activation of tumor-specific T cells. We observed that with loss of γ_c, CD4⁺ T cell differentiation and survival were diminished by wk 4 after transfer. In RAG^−/− γ_c⁺ hosts, we saw expansion and survival of CD4⁺ T cells past 4 wk. We also found higher expression of MHC class II on CD11c^highCD86^high DC and the Th1 transcription factor Tbet in TRP-1 CD4⁺ T cells isolated from RAG^−/− mice when compared with TRP-1 CD4⁺ T cells from RAG^−/−γ_c^−/− mice. Our results differ from those in a recent study showing that γ_c signaling on DC reduces CD4⁺ T cell homeostatic proliferation in vivo (Guimond et al., 2009). We found that TRP-1 CD4⁺ T cells expand initially at 1 wk after adoptive cell transfer but fail to survive in RAG^−/−γ_c^−/− hosts once tumor progresses, which is usually 4 wk or later. Surprisingly, even though there were more tumor-specific T cells in RAG^−/−γ_c^−/− hosts, it was not enough to clear tumor. It may be possible that strong (self-reactive) TCR activation is a prerequisite for survival for self-reactive TRP-1 CD4⁺ T cells because the previous study used polyclonal CD4⁺ T cells or antigen-specific CD4⁺ T cells that do not recognize an antigen in the host (Guimond et al., 2009). This strong activation (indicated by CD5 expression on TRP-1 CD4⁺ T cells) may come from increased expression of CD80 and MHC class II on DC that requires γ_c for their activation to become sensitive to and produce IL-12, which is paramount for Th1 priming in vivo (Ohteki et al., 2001, 2006; Terme et al., 2008). During in vivo differentiation, TRP-1 CD4⁺ T cells expressed high levels of IL12Rβ1 and IL-27R during in vivo differentiation, which are required for priming, and IL-23R, which may be important for Th1 memory formation. The strong up-regulation of Stat4 in TRP-1 T cells in vivo shows that IL-12 and/or IL-23 signals are driving their activation, as indicated by high IFN-γ production and IL-22, respectively. This is further supported by lack of IFN-γ and Tbet expression in TRP-1 CD4⁺ T cells isolated from RAG^−/−γ_c^−/− hosts 4 wk after transfer in addition to lack of IFN-γ and CXCL9, but not CXCL10, in the serum at 1 wk in RAG^−/−γ_c^−/− hosts. Therefore, it appears that lack of proper DC activation leads to low expression of MHC class II and CD80, which subsequently leads to inappropriate activation of naive TRP-1 CD4⁺ T cells in RAG^−/−γ_c^−/− hosts. Loss of IFN-γ expression by CD4⁺ T cells possibly leads to loss of CXCL9 expression in the tissues, and this may lead to failure of recruitment of antitumor CXCR3⁺ CD4⁺ T cells to the tumor site. This pattern of recruitment may be similar to the process that occurs during nephritis or disease of the eye during herpes simplex virus type 1 infection, which requires CXCL9, but not CXCL10, for recruitment of CXCR3⁺ T cells into the tissues (Wuest et al., 2006; Menke et al., 2008).

Why seek to understand how TRP-1 CD4⁺ T cells are activated in vivo? More intense ablation may remove DCs that provide the secondary signals (such as IL-12 or possibly IL-23) that are needed to properly differentiate CD4⁺ T cells in vivo into long-lived memory cells, which is similar to IL-2 for CD8⁺ T cells (Williams et al., 2006). Less space may be the Achilles’ heel of intense lymphodepletion when using CD4⁺ T cells. This may indicate that CD4⁺ T cells are not receiving the proper signals initially for long-lived survival. In mice irradiated with 9 Gy and humans receiving 12 Gy lymphodepleting regimens, IL-2 administration was needed to augment ACT. Transfer of naive T cells has not yet been accomplished in these settings. These cells may require administration of DC for support (Steinman and Banchereau, 2007).

How can this environment be obtained without lymphodepleting chemotherapy or irradiation? Cytokines (such as IL-7 and IL-15) found in patients receiving high-dose radiation therapy and irradiated WT and RAG^−/− mice may be used to mimic lymphopenia in a WT host (Gattinoni et al., 2005a; Zhang et al., 2005; Guimond et al., 2009; Pellegrini et al., 2009). In humans, IL-7 has shown strong promise in recent years, and long administration has been shown to release the inhibitory networks that suppress antitumor immunity (Pellegrini et al., 2009). Cytokine/antibody immune complexes to IL-2, IL-7, or IL-15 may offer another way to mimic lymphopenia, as a much lower amount of cytokines is needed to enhance proliferation of naive and memory phenotype T cells in vivo (Boyman et al., 2006, 2008; Rubinstein et al., 2006; Cho et al., 2007). Lymphopenia-induced activation of CD4⁺ T cells itself may program these cells to become less prone to tolerance induction seen in other systems using CD4⁺ T cells (Sotomayor et al., 1999), and we are currently exploring this.

Administration of antibodies against costimulatory and inhibitory molecules is another method to enhance antitumor immunity. We saw high expression of CTLA-4 and 41BB on TRP-1 CD4⁺ T cells. Anti–CTLA-4 therapy is being used in clinical trials and has been shown to enhance adoptive cell therapies already with CD8⁺ T cells and donor lymphocyte infusions by inhibiting T reg cells and enhancing activation of non–T reg cells (Quezada et al., 2006, 2008; Peggs et al., 2009). With tumor-specific TRP-1 CD4⁺ T cells, this enhances therapy considerably in WT and lymphopenic hosts (Quezada et al., 2010). Stimulation of 41BB also enhances immunotherapy using T cells coexpressing 41BBL (Stephan et al., 2007). It can also break tolerance to a poorly immunogenic tumor (Wilcox et al., 2002). ICOS is being considered as another immunotherapy candidate (Paulos, C. and June, C., personal communication). These therapies could be combined with ACT in the future.

In this study, we focused on naive CD4⁺ T cells, which differentiate properly in their natural state and become long-lived memory T cells. This idea was also demonstrated in another tumor model that involved activation of naive T cells via forced LIGHT expression in the tumor microenvironment (Yu et al., 2004). Although forced expression of LIGHT in human tumors may not be a clinical reality, this led to a profound tumor rejection indicating the importance of activating naive T cells in their natural state.

How do you obtain naive T cells from humans? TCR transduction of naive T cells with CD4⁺ MHC class II tumor-specific TCRs could be attempted in humans, as previously achieved with MHC class I tumor-specific TCRs (Morgan et al., 2006; Hinrichs et al., 2009). It may be possible to transfer naive tumor-specific CD4⁺ T cells into hosts preconditioned with noninvasive cytokine therapies that do not cause harmful side effects. Transduction of high avidity TCRs to tumor antigens into naive or cord PBL is feasible and may be augmented by high doses of IL-7 or IL-15 administration plus anti-CTLA or anti-41BB therapy.

How do you obtain nontolerant highly reactive TCRs as seen in this paper? Mouse TCRs reactive against human cancer antigens can be transfected into human T cells using gene therapy (Johnson et al., 2009). Human HLA Tg mice can be immunized with human cancer antigens to generate highly tumor-reactive TCRs, which can then be used in the immunotherapy of cancer in humans.

In summary, TRP-1 Tg mice represent a new ACT immunotherapy model for the study of ways to induce potent antitumor immunity against established melanoma without the need for cytotoxic CD8⁺ T cells or a vaccine. Understanding the mechanisms of tumor regression in mice with lymphopenia may allow the development of new therapies that work in the absence of chemotherapy and radiation. These data suggest a new role for CD4⁺ T cells as cytotoxic T cells.

Tyrp1^B-wRAG^−/− TRP-1-specific CD4⁺ TCR Tg mice (B6.Cg-Rag1^tm1Mom Tyrp1^B-w Tg(Tcra,Tcrb)9Rest/J; deposited at The Jackson Laboratory) and WT Tyrp1^+/+ TRP-1-specific CD4⁺ TCR Tg mice were created by P.A. Antony and K. Irvine at the National Cancer Institute (Bethesda, MD; Muranski et al., 2008). Recombination-activating gene 1^−/− (Rag1^tm1Mom) mice, Foxp3^sf, C57BL/6 WT, and MHC class II^−/− were purchased from The Jackson Laboratory. RAG^−/−γ_c^−/− mice were purchased from Taconic (Cao et al., 1995). Foxp3^sf mice were crossed with Tyrp1^+/+ TRP-1-specific CD4⁺ TCR Tg mice to generate Tyrp1^+/+ TRP-1-specific CD4⁺ TCR Foxp3^sf Tg mice. All mice were used in accordance with guidelines from the University of Maryland Institutional Animal Care and Use Committee. All experiments were conducted with the approval of the Animal Use and Care Committees of the National Cancer Institute, Johns Hopkins University School of Medicine, and the University of Maryland School of Medicine.

B16.F10 (H-2^b), hereafter called B16, is a TRP-1⁺ spontaneous murine melanoma that was obtained from ATCC and maintained in culture media as previously described (Antony et al., 2006). Tumors were injected subcutaneously at 2 × 10⁵ cells/mouse. Tumors are measured blindly with digital calipers. The perpendicular diameters are determined and multiplied to generate the area in millimeters squared as previously described (Antony et al., 2006).

TRP-1 CD4⁺ T cells were sorted from spleens of donor Tyrp1^B-wRAG-1^−/− TRP-1–specific Tg male mice. Spleens were harvested and made into single-cell suspensions. Cells were made devoid of red blood cells by ACK lysis. Subsequently, cells were counted and enriched for CD4⁺ T cells by magnetic bead sorting using a CD4⁺ T cell enrichment kit from Miltenyi Biotec. Enriched CD4⁺ T cells were counted and resuspended in PBS and used in adoptive transfer studies (2 × 10⁵ cells/mouse). NK cells were isolated from WT mice using NK cell sorting kits from Miltenyi Biotec. When indicated, 5 × 10⁶ NK cells were transferred on the same day as TRP-1–specific T cells. Open-repertoire CD4⁺CD25⁻ T cells from WT mice were sorted as previously described (Antony et al., 2006). Mice were irradiated as previously described (Wrzesinski et al., 2007).

Anti–IFN-γ (XMG1.2, NA/LE) and anti-NK1.1 (NA/LE, PK136) were purchased from BD.

Anti-CD4 (RM4-5), anti-CD25 (PC61), anti-CD122 (TM-β-1), anti-Vβ14 (14–2), anti-NK1.1 (PK136), anti-ICOS (7E.17G9), anti-CD44 (IM7), anti-CD62L (MEL-14), anti-CD11c (HL3), anti-CD86 (GL1), anti–MHC class II (AF6-120.1), anti–IL-17 (TC11-18H10), anti–IFN-γ (XMG1.2), anti-TNF (MP6-XT22), anti-CD49b (DX5), and CD11b (M1/70) were obtained from BD. Anti–IL-7Rα (SB/199), anti-CXCR3 (CXCR3-173), anti-Foxp3 (FJK-16s), anti-Tbet (4B10), anti–MHC class II (I-A/I-E; M5/114.15.2), anti–GR-1 (RB6-8C5), anti-perforin (eBioOMAK-D), anti-NKp46 (29A1.4), anti-LAMP-1 (1D4B), anti–granzyme B (16G6), and anti-NKG2D (CX5) were obtained from eBioscience. All flow cytometry scales are log scales, if not otherwise specified. Intracellular staining for cytokines was done with the Cytofix/Cytoperm intracellular staining kit (BD). For granzyme B, Foxp3, and perforin, the Foxp3 staining buffer set was used as recommended by eBioscience. All samples were run on a FACSCalibur (BD; Department of Surgery, University of Maryland School of Medicine) and analyzed by FlowJo Software (Tree Star, Inc.).

Confocal microscopy was done as previously described (Quezada et al., 2008) except that slides were imaged on an LSM (Olympus) with a 20× oil immersion objective. In brief, tumor was frozen in OCT solution (Sakura). 8-µm sections were cut with a microcryotome (Leica), fixed for 10 min in cold acetone, allowed to air dry, hydrated with PBS for 5 min, and then blocked with PBS with 2% FCS, 50 µg/ml 24G2 antibodies, and 5% rat, hamster, and mouse serum for 15 min. Slides were stained overnight with anti–MHC class II APC and DAPI, washed, and mounted. Samples were scanned on an inverted confocal microscope (LSM; University of Maryland Greenebaum Cancer Center) under a 20× oil immersion objective. Images were analyzed with Image J64 software (National Institutes of Health).

RNA was isolated from FACS-sorted CD4⁺ T cells from LN and spleens of tumor-bearing RAG^−/− mice undergoing treatment with TRP-1 CD4⁺ T cells. RNA was isolated with RNA isolation kits (QIAGEN). Sample amplification and labeling procedures were performed using a Low RNA Input Fluorescent Linear Amplification kit (Agilent Technologies). In brief, 400 ng of total RNA was used as starting material base on available yield of RNA isolation but was kept at the same amount of input in each experiment. Total RNA was reverse transcribed into first-strand and second-strand complementary DNA by MMLV-RT using an oligo-dT primer that incorporates a T7 promoter sequence. The complementary DNA was then used as a template for in vitro transcription in the presence of T7 RNA polymerase and cyanine-labeled CTPs. The labeled complementary RNA was purified using RNeasy micro kit (QIAGEN) and followed by quantification of both concentrations of complementary RNA and dye labeled. RNA spike-in controls (Agilent Technologies) were added to RNA samples before amplification and labeling according to the manufacturer’s protocol. The entire amount of each sample labeled with Cy3 or Cy5 was mixed with control targets (Agilent Technologies). Fragmentation was performed by incubating at 60°C for 30 min and stopped by adding an equal volume of 2× GE Hi-RPM hybridization buffer (Agilent Technologies). Agilent 4X44K whole mouse genome array (G4122F) with 41534 unique probes was used. Fragmented targets were added onto a microarray, assembled into a hybridization chamber (Agilent Technologies), and hybridized at 60°C for 17 h in a hybridization oven with rotation. Hybridized microarrays were washed and dried according to the Agilent microarray processing protocol. Microarrays were scanned using an Agilent G2505B Scanner controlled by Agilent Scan Control 7.0 software. Data were extracted with Agilent Feature Extraction 9.1 software. Differentially expressed targets were identified using the processed data and the log ratio generated by the software. Only values of twofold or higher were reported. Heat maps were hand generated from the gene lists. The GEO microarray data accession no. is GSE19904.

Serum was collected via tail vein using serum collection tubes (BD). Serum was analyzed by MILLIPLEX 32-Plex assay (University of Maryland, Baltimore, Cytokine Core Laboratory, and Millipore).

An unpaired Student’s t test was used to compare the differences between cytokines and chemokines as indicated. Tumor curves were compared using a two-way ANOVA with nonrepeated measures. P-values of ≤0.05 were considered significant. PRISM 5.0b software was used to analyze the data (GraphPad Software, Inc.).

Fig. S1 shows the phenotype of TRP-1 Tg WT, Foxp3^sf, and tyrp1^B-w mice. Fig. S1 also shows that TRP-1 CD4⁺ cells are activated in WT mice but not in tyrp1^B-w mice. Foxp3 expression in naive TRP-1 CD4⁺ T cells from tyrp1^B-w RAG^−/− mice is shown. Lastly, tumor infiltration by TRP-1 CD4⁺ T cells is shown by flow cytometry. Fig. S2 shows accumulation of inflammatory monocytes in the spleen, LN, and tumor after adoptive transfer of TRP-1 CD4⁺ T cells into tumor-bearing lymphopenic mice. Fig. S3 shows stable expression of Foxp3 in TRP-1 CD4⁺ T cells after tumor treatment. Fig. S4 shows that TRP-1 CD4⁺ T cells do not convert to Foxp3⁺ cells in RAG^−/−γ_c^−/− tumor-bearing hosts during tumor progression.

This work is dedicated to the memory of Asher Alexandria Huminsky, Floyd Antony, Kay Antony, Edward Wrzesinski, Earl Norling, and Charles Babinger.

We acknowledge Kari Irvine for creating the TRP-1–specific hybridoma from which the TCR for the Tg mouse was cloned, and Megan Bachinski (Surgery Branch, NCI), Linda J. Kesselring (technical editor, Department of Emergency Medicine, University of Maryland School of Medicine), and Scott E. Strome (Professor and Chairman, Department of Otorhinolaryngology-Head and Neck Surgery Chief, University of Maryland School of Medicine) for helping to edit the manuscript. We appreciate the hard work of the people of Veterinary Resources at the University of Maryland School of Medicine, especially Terry King and Marvin Riddley for their excellent assistance and care for the mice used in these experiments. Lastly, we thank Cano Asher, Cahit Can, and Orhan and Çigdem, and Orcin and Deniz for their support during this project.

This research was initiated at the Surgery Branch of the National Cancer Institute and was supported by the Center for Cancer Research. The work was further supported by a K22 NCI career transition award (5K22CA125200 to P.A. Antony) and the Melanoma Research Foundation (P.A. Antony).

The authors declare that they have no conflicting financial interests.