Tumor-reactive CD4⁺ T cells develop cytotoxic activity and eradicate large established melanoma after transfer into lymphopenic hosts

To determine whether activation and differentiation of small numbers of naive tumor-reactive CD4⁺ T cells in vivo could drive rejection of established tumors, we used naive CD4⁺Trp1⁺ cells isolated from Trp1 Tg mice. C57BL/6 mice were challenged with a lethal dose (2.5 × 10⁵ cells) of the poorly immunogenic B16/BL6 mouse melanoma line. 10–12 d after challenge, when tumors were clearly visible (Fig. 1 b), mice received 5 Gy of irradiation (radiation therapy [RT]) followed by adoptive transfer of naive CD4⁺Trp1⁺ cells (CD4-ACT). To assess the potential combinatorial efficacy of CTLA-4 blockade and CD4-ACT, a cohort of mice also received anti–CTLA-4 mAb. RT by itself failed to induce regression of established tumors (Fig. 1 a). Additional controls including anti–CTLA-4 + RT or anti–CTLA-4 + CD4⁺Trp1⁺ also failed to induce regression (unpublished data). Remarkably, transfer of as few as 50,000 CD4⁺Trp1⁺ T cells into irradiated hosts was sufficient to induce initial regression of large established melanoma tumors and depigmentation in all mice (Fig. 1 a). However, in the absence of anti–CTLA-4, the combination of RT and CD4-ACT failed to eradicate tumors completely, and tumors recrudesced in up to 60% of mice (Fig. 1 a). The addition of anti–CTLA-4 produced long-term protection in all mice (Fig. 1 a), as well as development of disseminated depigmentation (Fig. 1 b, far right). Analysis of T cell dynamics in the blood of treated mice showed rapid expansion of CD4⁺Trp1⁺ cells after transfer into irradiated mice (Fig. 1 c). CTLA-4 blockade increased expansion by two- to threefold, reaching up to 40% of blood lymphocytes (Fig. 1 c), while simultaneously preventing the expansion and accumulation of CD4⁺Trp1⁺Foxp3⁺ cells (Fig. 1 d).

Although mice receiving RT monotherapy showed no changes in serum levels of IFN-γ (Fig. 1 e) and TNF (Fig. 1 f), both inflammatory cytokines were increased significantly after CD4-ACT in irradiated hosts (Fig. 1, e and f). Cytokine production was markedly increased after CTLA-4 blockade (Fig. 1, e and f), correlating with the increase in CD4⁺Trp1⁺ cells (Fig. 1 c) and heightened tumor rejection (Fig. 1 a). These data demonstrate that adoptive transfer of small numbers of naive tumor-reactive CD4⁺ T cells into irradiated recipients can mediate potent rejection of established melanoma. Incorporation of CTLA-4 blockade significantly enhanced T cell expansion, cytokine secretion, and antitumor activity, conferring complete rejection and durable protection in all treated mice.

To further characterize the antitumor response induced by CD4⁺Trp1⁺ T cells, we analyzed their expansion, differentiation, and function in LNs and tumors 8 d after transfer. CD4⁺Trp1⁺ T^eff cells (Foxp3⁻) were detectable in low numbers in LNs of nonirradiated recipients, increasing in number when transferred after RT (Fig. 2 a). CTLA-4 blockade significantly increased expansion of CD4⁺Trp1⁺ T^eff cells transferred into irradiated hosts (Fig. 2 a), while preventing expansion of CD4⁺Trp1⁺Foxp3⁺ T reg cells (Fig. 2 b). The combination of RT and CD4-ACT resulted in a marked Th1-like phenotype, with CD4⁺Trp1⁺ T cells secreting copious amounts of IFN-γ, TNF, and IL-2. (Fig. 2 c and Fig. S1). Contrary to its effect on the absolute number of CD4⁺Trp1⁺ T^eff and T reg cells, CTLA-4 blockade had no impact on the frequency of IL-2, IFN-γ, or TNF single, double, or triple producers in LNs (Fig. 2 c and Fig. S1) or tumors (Fig. 2 f and Fig. S2). Remarkably, in addition to IFN-γ, TNF, and IL-2, in vivo activation of tumor-reactive CD4⁺Trp1⁺ T cell also resulted in their differentiation into granzyme B–producing CD4⁺ T cells in both the periphery and in tumors (Figs. S1 and S2).

Analysis of tumor-infiltrating CD4⁺Trp1⁺ T cells also highlighted the importance of RT for T cell accumulation (Fig. 2 d) and differentiation (Fig. 2 f and Fig. S2). Although CTLA-4 blockade failed to further increase the absolute number of intratumoral CD4⁺Trp1⁺ T^eff cells in irradiated recipients, it markedly reduced the number of tumor-infiltrating CD4⁺Trp1⁺Foxp3⁺ T reg cells, mirroring changes in the LNs and blood. The intratumor ratio of Trp1⁺ T^eff cells to Trp1⁺ T reg cells (Fig. 2 g) and of CD4⁺Trp1⁺ T^eff cells to total T reg cells (Fig. 2 h) increased several fold upon CTLA-4 blockade. Finally, histological analysis of T cell infiltrates confirmed lack of infiltration after monotherapy with RT or CD4-ACT, whereas their combination significantly enhanced CD4⁺Trp1⁺ accumulation within tumors (Fig. 2 I; and Figs. S3 and S4). These data are consistent with those obtained from blood and support a role for RT in the activation and differentiation of tumor-reactive CD4⁺Trp1⁺ T^eff cells and for CTLA-4 blockade in driving further expansion of such cells while preventing T reg cell accumulation.

To determine whether IFN-γ and TNF were directly involved in tumor rejection, we used both neutralizing antibodies and a variety of knockout mice as tumor recipients. TNF blockade during treatment with RT, CD4-ACT, and CTLA-4 blockade (triple therapy) had no impact on tumor rejection (Fig. 3 a). In contrast, neutralization of IFN-γ prevented rejection in four out of five mice, suggesting an important mechanistic role for IFN-γ in tumor rejection (Fig. 3 a). Complete tumor eradication was observed after triple therapy using IFN-γ^−/− recipient mice, indicating that IFN-γ secreted by the transferred CD4⁺Trp1⁺ cells, and not by the host cells, is critical for antitumor activity (Fig. 3 a). IFN-γR^−/− mice also rejected established tumors after triple therapy, suggesting that the target cell for IFN-γ could (at least in an initial phase) be the tumor itself (Fig. 3 a). However, all IFN-γR^−/− recipients regrew pigmented tumors 40–70 d after the primary challenge, suggesting that IFN-γ–sensitive cells other than the tumor are potentially important for complete tumor eradication.

Remarkably, tumor rejection appeared to be independent of endogenous CD4, CD8, and B cells because RAG^−/− recipients were capable of rejecting established melanoma after triple therapy (Fig. 3 b). Tumors were also rejected in perforin^−/− (PFN^−/−) recipient mice treated with triple therapy, suggesting that rejection is independent of endogenous PFN-dependent killing activity (i.e., mediated by CD8 and NK cells). This was further confirmed by depletion experiments with anti-CD8 and anti-NK1.1 mAbs, which failed to prevent tumor rejection (unpublished data). Finally, to determine the cellular targets of anti–CTLA-4, we used recipient mice expressing human CTLA-4 instead of mouse CTLA-4 (Peggs et al., 2009). In these mice, anti–mouse CTLA-4 mAbs block CTLA-4 only on the transferred CD4⁺Trp1⁺ cells. As shown previously (Fig. 1 a), RT and CD4-ACT resulted in partial rejection in wild-type mice, whereas incorporation of anti–mouse CTLA-4 resulted in complete tumor rejection (Fig. 3 c). Similar results were obtained when using human CTLA-4 mice as recipients, suggesting that CTLA-4 blockade on only the transferred cells is sufficient to mediate complete protection (Fig. 3 d). Collectively, the data support a direct role for the transferred CD4⁺Trp1⁺ T cells in tumor rejection, mediated most likely through the action of IFN-γ. This activity is independent of endogenous CD4, CD8, and B cells, or NK cells, suggesting a direct effect of the transferred CD4⁺Trp1⁺ cells on the tumor.

To determine whether the antitumor activity was tumor specific and not mediated through nonspecific mechanisms (i.e., activation of other cellular subsets such as inflammatory myeloid-derived cells; Mumberg et al., 1999; Corthay et al., 2005), we challenged mice with B16/BL6 melanoma in one flank and EL-4 lymphoma in the contralateral flank. Without CD4⁺Trp1⁺ transfer, both EL-4 and B16/BL6 tumors continued to grow (Fig. 4 a), whereas triple therapy resulted in complete rejection of B16/BL6 melanoma without influencing the growth of the unrelated EL-4 tumor (Fig. 4 b). To exclude the possibility that the inflammatory milieu created after triple therapy induces local rejection of tumors regardless of specific recognition by CD4⁺Trp1⁺ cells, we injected a mixture of B16/BL6 and EL-4 in the same site and followed tumor progression after therapy. Triple therapy did not result in rejection of the tumor mixture (Fig. 4 c), arguing against bystander tumor rejection. In the absence of CD4⁺Trp1⁺ transfer, the tumor contained a mixture of B16/BL6 melanoma and EL-4 (black tumors), whereas after CD4⁺Trp1⁺ T cell transfer there was no detectable melanoma within the growing tumor (off-white tumors; Fig. 4 d). To further verify that B16/BL6 melanoma was being rejected in the coinjection setting, mice were challenged with a mixture of EL-4 and B16/BL6-luciferase and treated at day 10 with triple therapy. The luciferase signal was completely eliminated in the triple therapy group but not in control mice lacking CD4-ACT (Fig. 4, e and f). The data argue against a nonspecific or bystander mechanism of tumor rejection and, together with the tumor rejection data in IFN-γR^−/− mice, suggest a direct impact of transferred CD4⁺Trp1⁺ T cells and IFN-γ on melanoma cells.

To determine if the antitumor responses achieved after triple therapy were caused by acquired cytotoxic activity and direct tumor killing by CD4⁺Trp1⁺ T cells, we performed a series of in vivo and in vitro killing experiments. Mice were challenged with tumors and treated at day 10 with or without RT, CD4⁺Trp1⁺ T cells, and anti–CTLA-4. CFSE-labeled B cell targets were injected i.v. 17 d after tumor challenge, and in vivo killing activity was determined 14–16 h later by quantification of the CFSE^high peak (corresponding to target cells loaded with the class II–restricted peptide recognized by CD4⁺Trp1⁺ cells) relative to the unloaded CFSE^low peak. CD4⁺Trp1⁺ cells acquired potent in vivo killing activity after transfer into irradiated recipients (Fig. 5 a). This activity was significantly enhanced after CTLA-4 blockade, whereas little to no activity was detected with CD4-ACT or RT alone (Fig. 5 a, right). Killing activity was confirmed in vitro, where CD4⁺Trp1⁺ cells isolated from triple therapy–treated mice and expanded in vitro were extremely efficient at eliminating peptide-loaded B cell targets (Fig. 5, b and d) or B16/BL6 targets (Fig. 5, c and e). Importantly, although killing activity was largely inhibited by antibodies blocking class II, blockade of FAS–FASL interactions did not significantly impair lysis of B cell targets or tumor cell targets (Fig. 5, d and e). Additional experiments blocking either FASL (MFL4 clone) or TRAIL (N2B2 clone) in vivo also failed to prevent tumor rejection in mice treated with triple therapy (unpublished data). In contrast, and in keeping with the high levels of granzyme B expressed by tumor-reactive CD4⁺Trp1⁺ cells (Figs. S1 and S2), simultaneous blockade of granzyme B and PFN activity in vitro synergized to prevent lysis of tumor targets (Fig. 5 f). Collectively, these data suggest that the cytolytic activity acquired by tumor-reactive CD4⁺Trp1⁺ T cells depends on direct recognition of target cells through class II and degranulation of granzyme-containing lytic granules.

Direct recognition and specific destruction of tumor targets in vitro suggests a requirement for class II expression by the target cells in vivo. To test whether B16/BL6 could express class II in vivo, we challenged mice with a traceable Thy1.1-expressing B16/BL6 melanoma cell line and quantified class II expression on Thy1.1-positive cells after therapy (Fig. 6, a and b). Class II was not detected on the Thy1.1-expressing tumor cells in the untreated group or on tumors from mice treated only with CD4⁺Trp1⁺ and anti–CTLA-4 (Fig. 6, a and b). In contrast, high levels of class II could be found on tumor cells after triple therapy. Furthermore, although RT alone failed to induce class II up-regulation, the combination of RT and CD4⁺Trp1⁺ was as efficient as triple therapy at inducing class II up-regulation on tumor cells (not depicted).

A possible explanation for the high levels of class II expression could be the copious amounts of IFN-γ produced by tumor-reactive CD4⁺Trp1⁺, as IFN-γ has been shown to up-regulate expression of MHC molecules in several different cell types (Boehm et al., 1997; Boss, 1997). In support of this mechanism, IFN-γ was absolutely necessary for class II up-regulation, which was completely ablated by administration of neutralizing antibodies (Fig. 6, a and b).

Finally, to test the direct class II dependence of in vivo recognition and rejection of melanoma tumors, we adoptively transferred CD4⁺Trp1⁺ T cells that had been primed in vivo into tumor-bearing class II knockout (CII^−/−) recipients. Activated cells were used because of a preliminary observation that class II expression on recipient cells is required for priming of tumor-reactive CD4⁺ T cells, as transfer of naive CD4⁺Trp1⁺ T cells into irradiated CII^−/− mice failed to eradicate established tumors (unpublished data). Transfer of activated CD4⁺Trp1⁺ T cells into irradiated CII^−/− mice followed by CTLA-4 blockade resulted in complete tumor eradication. Notably, tumor rejection was completely ablated by antibodies blocking class II MHC (Fig. 6 c). Collectively, the data support a model in which IFN-γ secreted by activated CD4⁺Trp1⁺ T cells induces class II up-regulation on tumor cells, converting them into targets for cytotoxic CD4⁺Trp1⁺ T cells.

To test the efficacy of this new combinatorial therapy in a stringent and more realistic model, we applied our treatment regimen to a mouse model of spontaneous melanoma. Glutamate receptor 1 (Grm-1) Tg mice overexpress the metabotropic Grm-1 driven by the melanocyte-specific promoter of dopachrome tautomerase and develop tumors in the tail and pinnae of the ear between 4 and 6 mo of age (Pollock et al., 2003). Mice received triple therapy after development of large melanoma lesions between 10 and 13 mo of age. Although untreated tumors never regress spontaneously in this model (Pollock et al., 2003), a remarkable regression of tumors in the ears, eyes, and tails was observed after triple therapy (Fig. 7, a and b). CD4⁺Trp1⁺ cells expanded greatly after transfer into irradiated mice (Fig. 7 d), as observed in the transplantable B16/BL6 melanoma model (Fig. 1 c). High levels of IFN-γ and TNF were also detected in the blood of treated mice, occurring with a faster kinetic than with the transplantable melanoma line (Fig. 7, e and f). The data suggest that the efficacy of this therapeutic regimen is not limited to transplantable tumor models but is also evident in a more stringent model of spontaneous and disseminated disease, which is thought to better represent human malignancies. Finally, it illustrates the potency of this therapy, as demonstrated by its ability to mediate regression of many tumor lesions developing for up to 4 mo at different sites in the mouse.

Adoptive transfer of high numbers of tumor-reactive lymphocytes into lymphoablated patients is a promising therapy for late metastatic disease, particularly metastatic melanoma (Dudley et al., 2002, 2008). Although initially focused on CD8⁺ CTL responses (Yee et al., 2002), recent preclinical and clinical studies have redirected attention to CD4⁺ T cells, which are thought to provide the necessary help to the effector CD8⁺ CTL compartment (Hunder et al., 2008). In this paper, we demonstrate that CD4⁺ T cells transferred into lymphopenic mice can mediate rejection of large vascularized melanoma tumors, particularly when combined with CTLA-4 blockade. These data are significant for several reasons: (a) CD4-mediated rejection does not require expansion and manipulation in vitro, but occurs after activation and differentiation of small numbers of naive T cells in vivo in a lymphodepleted host; (b) tumor-reactive CD4⁺ T cells acquire cytotoxic activity and directly reject the tumor in an MHC class II–dependent manner; and (c) CTLA-4 blockade greatly enhances therapeutic efficacy of CD4-ACT, informing a potential strategy for future clinical studies. Finally, this therapy is effective in a mouse model of spontaneous melanoma, reinforcing the potential pathophysiological relevance of our findings. The data offer novel mechanistic insights into the role and function of tumor-reactive CD4⁺ T cells, and have significant clinical relevance for ACT, including approaches incorporating gene therapy. Collectively, they suggest that a small number of naive T cells transduced to express tumor-reactive TCRs might be sufficient to drive tumor rejection after differentiation in vivo, offering an alternative to the extensive ex vivo manipulation that is required to achieve high numbers of in vitro–differentiated tumor-reactive T cells.

The data demonstrate that RT in combination with CD4⁺Trp1⁺ transfer results in expansion, intratumor accumulation, and differentiation of the transferred cells into IFN-γ–, TNF-, IL-2–, and granzyme B–producing effector cells, whereas RT or CD4⁺Trp1⁺ transfer alone fail to mediate these changes (Figs. 1 and 2). Contrary to most studies of ACT using T cells activated in vitro, tumors continued to grow after CD4-ACT, and started to necrose and regress several days thereafter (Fig. 1 b). This may reflect a requirement for priming, expansion in the lymphopenic environment, and differentiation of the transferred cells before acquisition of full antitumor activity. Although slower to manifest than the antitumor effect observed after ACT with cells expanded in vitro (Hanson et al., 2000; Overwijk et al., 2003; Spiotto et al., 2004; Antony et al., 2005), priming in vivo in the lymphopenic hosts is extremely efficient, resulting in rejection of large tumors of up to 600 mm³.

Transfer of CD4⁺Trp1⁺ cells into nonirradiated tumor-bearing RAG^−/− mice, which lack B and T cells, also resulted in extensive proliferation, cytokine production, and tumor rejection (unpublished data), suggesting that the main role of RT in this model is the induction of lymphopenia and/or the elimination of cytokine sinks limiting tumor-reactive T cell responses (Gattinoni et al., 2005a; Muranski et al., 2006). Importantly, in RAG-sufficient hosts, RT may also contribute by sensitizing the tumor stroma (Zhang et al., 2007) and by increasing the expression of adhesion molecules on the tumor vasculature, which would render the tumor susceptible to T cell infiltration (Ganss et al., 2002; Lugade et al., 2005; Quezada et al., 2008).

Although total body irradiation has been successfully used in several major cancer centers across the United States (Muranski et al., 2006; Dudley et al., 2008), clinical application of our findings may benefit from future studies addressing alternatives to lymphodepletion. If the main contribution of RT is to induce lymphopenia, then conditioning of recipients by lymphodepleting chemotherapy regimens should also create an environment favorable to tumor rejection by adoptive transfer of small numbers of tumor-reactive CD4⁺ T cells. Finally, lymphodepletion may not be the only way to fully activate tumor-reactive CD4⁺ T cells; thus, alternatives including activation of the APC compartment (i.e., CD40 and TLR agonists) should be addressed in future studies.

The addition of CTLA-4 blockade to RT and CD4⁺Trp1⁺ transfer further enhanced antitumor responses, inducing higher total numbers of CD4⁺Trp1⁺ cells and higher levels of TNF and IFN-γ in serum samples (Fig. 1). CTLA-4 blockade did not modify cytokine production on a per cell basis but increased the number of tumor-reactive cells, accounting for the overall increase in levels of inflammatory cytokines in the serum. Equally important, CTLA-4 blockade resulted in a significant reduction in the absolute number of CD4⁺Trp1⁺Foxp3⁺ and endogenous T reg cells in the periphery and in the tumors. Collectively, these changes may create a state of combined hyperactivation and reduced regulation, correlating with increased cytotoxicity (Fig. 5 a) and resulting in complete tumor rejection. Notably, restricting CTLA-4 blockade to the transferred cells was sufficient to induce maximal antitumor activity, suggesting a strategy for future combinatorial approaches incorporating TCR transfer and CTLA-4 ablation on the same cell, which may avoid adverse immune events associated with systemic administration of anti–CTLA-4 antibodies (Peggs et al., 2006).

Although previous studies have indicated that Th17 cells polarized in vitro are capable of inducing tumor rejection (Muranski et al., 2008), we did not observe differentiation of CD4⁺Trp1⁺ cells into IL-17 producers in vivo (unpublished data). In the previous study, tumor rejection was dependent mostly on IFN-γ despite the requirement for differentiation into Th17 cells, in accordance with studies suggesting a possible reversion of Th17 into Th1 cells (Shi et al., 2008; Lee et al., 2009; Martin-Orozco et al., 2009). In contrast, we believe that priming of tumor-reactive CD4⁺Trp1⁺ T cells in vivo in a lymphopenic environment leads directly to a Th1-like phenotype. Surprisingly, these cells also expressed high levels of granzyme B in the periphery and tumors (Figs. S1 and S2), which correlated with acquisition of granzyme-dependent cytotoxic activity (Fig. 5) and potent rejection of large established tumors.

The acquisition of cytotoxic activity by transferred tumor-reactive CD4⁺ T cells is particularly striking. This distinguishes our findings from previous work showing that CD4⁺ T cells can help rejection of less well-established tumors through indirect effects of IFN-γ (Mumberg et al., 1999) on NK cells (Perez-Diez et al., 2007) and tumor-infiltrating macrophages (Greenberg et al., 1985; Hung et al., 1998; Corthay et al., 2005; Corthay, 2007). Interestingly, CD4⁺Trp1⁺ cells developed all the hallmarks of CD4⁺ Th cells with the additional benefit of exhibiting cytolytic activity. CD4⁺ CTLs targeting viral antigens (Paludan et al., 2002; Hegde et al., 2005; Heller et al., 2006) and alloantigens (Holloway et al., 2005; Spaapen et al., 2008) have been described previously, but the demonstration of similar activity in a more physiological model for self-/tumor antigen emphasizes the promise of these cells in cancer immunotherapy. Further identification and induction of tumor-reactive CD4⁺ T cells with cytotoxic activities in cancer patients may offer significant advantages for the treatment of human malignancies. These cells could be isolated and minimally expanded before reinfusion into conditioned recipients. The feasibility of such strategies is supported by the potent tumor rejection observed in CII^−/− mice after transfer of CD4⁺Trp1⁺ T cells that had been originally primed in vivo and then minimally expanded before transfer into lymphopenic tumor-bearing mice (Fig. 6 c). Furthermore, similar approaches for the isolation of tumor-reactive lymphocytes after immunotherapy followed by autologous transfer into conditioned recipients have been previously demonstrated in preclinical (Quezada et al., 2008) and clinical settings (Rapoport et al., 2005; June, 2007).

Other aspects of CD4⁺Trp1⁺ T cell function were equally critical for tumor eradication. The Th1-like phenotype that developed after transfer into a lymphopenic environment was characterized by the production of high levels of TNF, IL-2, and IFN-γ. IFN-γ was clearly required for tumor rejection and appeared to directly affect tumor cells in our model, because tumor regression was observed in IFN-γR^−/− recipients. In addition, MHC II expression by tumor cells depended on IFN-γ because its neutralization prevented MHC II up-regulation (Fig. 6, a and b). This is in keeping with previous work on autoimmunity demonstrating that IFN-γ secreted by CD4⁺ T cells can mediate up-regulation of class II on target cells (Wu et al., 1999, 2000). Remarkably, tumors recrudesced in all IFN-γR^−/− recipients after initial regression, suggesting that an IFN-γ–sensitive cell type other than the tumor may have a role in mediating long-term protection, in accordance with previous observations using CD8-ACT in which rejection of the tumor stroma was critical for complete tumor eradication (Spiotto et al., 2004; Zhang et al., 2007).

Based on these data, we propose a model in which tumor-reactive CD4⁺Trp1⁺ T cells transferred into lymphopenic mice expand, differentiate into IFN-γ–secreting, cytotoxic CD4⁺ cells, and accumulate within the tumor. IFN-γ induces up-regulation of MHC II on tumor cells, rendering them targets for the killer activity of CD4⁺Trp1⁺ cells. Finally, cytotoxic activity depends absolutely on class II expression by the tumor and correlates with high levels of granzyme B within CD4⁺Trp1⁺ cells in LNs and particularly in tumors, where it is expressed by 40–50% of infiltrating CD4⁺Trp1⁺ T cells (Figs. S1 and S2). As part of a more global view and in addition to their direct impact on the tumor, IFN-γ–secreting CD4⁺Trp1⁺ cells may also induce a cascade of events involving priming of cytolytic CD8⁺ T cells through activation of DCs and additional tumor destruction through activation of NK cells. Furthermore, high levels of IFN-γ in the tumor may also lead to activation of type I macrophages, which will also favorably affect antitumor activity.

In conclusion, we believe that these data greatly inform our basic understanding of the importance of tumor-reactive CD4⁺ T cells in the context of ACT. Equally importantly, they support development of clinical strategies focusing on exploiting the function of cytotoxic tumor-reactive CD4⁺ T cells generated after transfer and activation in vivo, which may obviate requirements for extensive and possibly detrimental manipulation in vitro before adoptive transfer. Finally, the augmented potency of the transferred tumor-specific T cells observed upon blockade of coinhibitory molecules such as CTLA-4 provides a basis for the improvement of ongoing ACT trials as well as for the development of future combinatorial trials.

6–8-wk-old C57BL/6, B6.SJL, RAG2^−/−, Prf^−/−, IFN-γ^−/−, IFN-γR^−/−, and MHC CII^−/− Δ78 mice were purchased from the Jackson Laboratory. Trp1 Tg mice carry TCR α and β transgenes on a background homozygous for the targeted mutation Rag1^tm1Mom and homozygous for the white-based brown radiation–induced mutation of Trp1, Tyrp1^B-w (Muranski et al., 2008). These mutant mice express an MHC class II–restricted (I-A^b) TCR recognizing the endogenous melanocyte differentiation antigen–minimal TRP-1 epitope corresponding to amino acids 113–127. Trp1 Tg mice were crossed to B6.SJL mice, and their progeny were intercrossed to generate CD45.1^+/+Trp1 Tg mice (also on the Tyrp1^B-WRAG^−/− background). Grm-1 Tg mice (Pollock et al., 2003) were provided by S. Chen (Rutgers, The State University of New Jersey, Piscataway, NJ). Grm-1 Tg mice overexpress the metabotropic Grm-1 driven by the melanocyte-specific promoter of dopachrome tautomerase. Mice develop spontaneous melanoma lesions primarily in the tail and pinnae of the ear between 4 and 6 mo of age. Mice were housed in specific pathogen-free conditions in accordance with institutional guidelines. All animal experiments were approved by the Memorial Sloan-Kettering Cancer Center Institutional Animal Care and Use Committee.

The poorly immunogenic B16/BL6 cell line was originally obtained from I.J. Fidler (M.D. Anderson Cancer Center, Houston, TX). The lymphoma EL-4 was obtained from the American Type Culture Collection. B16/BL6 cells were transduced and selected to constitutively express either Thy1.1 or Renilla luciferase.

Anti–CTLA-4 (9H10), anti-NK1.1 (PK136), anti-CD8 (2.43), anti–IFN-γ (XMG1.2), anti-TNF (XT3.11), and anti–I-A^b (M5/114) were purchased from BioXCell and administered i.p. Antibodies for flow cytometry and immunofluorescence were purchased from eBioscience and BD.

Mice were challenged intradermally (i.d.) with 2.5 × 10⁵ B16/BL6 tumor cells on day 0. In some experiments, mice received an additional i.d. challenge of 2.5–5 × 10⁵ EL-4 lymphoma cells on the opposite flank, or a mix of both tumors in one unique site. On day 10, mice were treated or not with 5 Gy of total body irradiation, 50,000 CD4⁺Trp1⁺ cells i.v., and 200 µg anti–CTLA-4 i.p. Mice received additional anti–CTLA-4 injections of 100 µg on days 13, 16, 20, and 24. For neutralization and blocking experiments, 200 µg anti–IFN-γ, anti-TNF, or anti–I-A^b was administered 1 d before initiation of therapy and every 2 d after that for a total of six injections. CD4⁺Trp1⁺ for adoptive transfers was isolated from Trp1 Tg mice using CD4 magnetic beads (Miltenyi Biotec).

Mice used for functional experiments were sacrificed on day 18 after tumor implantation, and LN cells and tumor-infiltrating lymphocytes were isolated as previously described (Quezada et al., 2006, 2008). Tumor-infiltrating CD4⁺ T cells were purified using CD4 positive selection (FlowComp; Invitrogen) according to the manufacturer’s instructions. Purified CD4⁺ T cells from tumors or bulk cells from LNs were restimulated for 2 h at 37°C with 5 × 10⁴ DCs and 2 µM of Trp1 peptide followed by addition of brefeldin A (BD) for 2 more hours.

Samples were stained with anti-CD45.1–Pacific Blue (A20) and anti-CD4–PerCP-Cy5.5 (GK1.5), fixed, and permeabilized (eBioscience) according to the manufacturer’s instructions, and stained with anti-Foxp3–PE (FJK-16s). Flow cytometry reference beads (PeakFlow blue; Invitrogen) were mixed with the samples before analysis to normalize for the volume of the sample acquired. The absolute number of CD4⁺Trp1⁺ cells per tumor was calculated using the following formula: number of CD4⁺Trp1⁺ cells = (count of CD4⁺Trp1⁺ cells/count of beads)/tumor weight. For functional analysis, restimulated samples were stained as described, fixed, and permeabilized using Cytofix/Cytoperm (BD), followed by anti–IFN-γ–Alexa Fluor 488 (XMG1.2), anti-TNF–PE (MP6-XT22) or anti–IL-2–PE (JES6-5H4), and anti–granzyme B–allophycocyanin (GB11).

Tumors were flash frozen in optimal cutting temperature solution (Sakura). 10-µm sections were cut with a microcryotome (Leica), fixed for 10 min in cold acetone, and stained with anti-CD4–Alexa Fluor 488, anti-CD31–PE, and DAPI, and analyzed on an inverted confocal microscope (LSM; Leica) under a 20× water immersion objective.

Serum from blood samples was analyzed using a Mouse Inflammation Cytometry Bead Array (BD) according to the manufacturer’s instructions.

Splenocytes from B6.SJL mice were labeled with 5 or 0.5 µM CFSE. CFSE^high splenocytes were loaded with 20 µM of Trp1 peptide (Bio-Synthesis) for 2 h at 37°C. On day 17 after tumor challenge (7 d after CD4⁺Trp1⁺ transfer), 5 × 10⁵ cells of a 50:50 mixture of Trp1 peptide–loaded and –unloaded splenocytes were transferred by tail vein injection. 14–16 h later, mice were sacrificed and spleens and LNs were removed, and the percentage of loaded and unloaded B220⁺ splenocytes was analyzed. Cytotoxicity was calculated using the following formula: cytotoxicity = 100% × (1−((unloaded/loaded)_control/(unloaded/loaded)_experimental)).

7 d after adoptive transfer into irradiated tumor-bearing mice, CD4⁺Trp1⁺ cells were purified from LNs using CD4 beads (Miltenyi Biotec) and cultured with equal numbers of DCs and 2 µM of Trp1 peptide. After 3 d, cell cultures were supplemented with DCs, 30 U/ml IL-2, and 2 µM of Trp1 peptide. After 3 d, cells were harvested and used for in vitro cytotoxicity assays or for transfer into tumor-bearing CII^−/− recipients. To determine in vitro killing of tumor targets, B16/BL6 tumor cells were loaded with 0.5 mM CFSE, whereas EL-4 control cells were loaded with 5 mM CFSE. Fluorescent reference beads were mixed with the samples before analysis to normalize for the volume of the sample acquired. Blocking anti–I-A^b and anti-FASL (MFL4 clone) mAbs were used at a final concentration of 50 µg/ml, whereas concanamycin A (Sigma-Aldrich) and the granzyme B inhibitor Z-AAD-CMK (Enzo Biochem, Inc.) were used at 0.5 and 25 µM, respectively.

Data were analyzed using Prism 5.0 (GraphPad Software, Inc.). Experiments were repeated two to three times. Statistical significance was determined by a Student’s t test (between two groups or conditions) or analysis of variance with a post-hoc test (three or more groups or conditions). Tumor survival data were analyzed with the Kaplan-Meier method. The log-rank test was used to compare survival curves for different subgroups on univariate analyses. P < 0.05 was considered statistically significant.

Figs. S1 and S2 show the levels of intracellular cytokines (IFN-γ, TNF, and IL-2) and granzyme B produced by CD4⁺Trp1⁺ T cells isolated from LNs (Fig. S1) and tumors (Fig. S2) of mice receiving different treatments. Fig. S3 depicts histological analysis of whole tumors from mice treated with RT with or without CD4⁺Trp1⁺ T cells and anti–CTLA-4. Fig. S4 illustrates tumor infiltration by CD4⁺Vβ14⁺ T cells after treatment with RT and CD4⁺Trp1⁺ T cells with and without anti–CTLA-4.

We would like to thank Dr. S. Chen for providing the Grm-1 Tg mice, and Dr. A. Jungbluth for critical help with microscopy. We also would like to thank W. Montalvo, J. Geddes, and J. Lu for technical assistance, and E. Corse and P. Savage for critical evaluation of the work.

S.A. Quezada is a Research Fellow funded by the Irvington Institute Fellowship Program of the Cancer Research Institute and a junior member of the Millennium Nucleus on Immunology and Immunotherapy, Pontifícia Universidad Católica de Chile. T.R. Simpson is supported by a Canadian Institutes of Health Doctoral Research Award. K.S. Peggs is currently an investigator at the Department of Haematology, University College London Cancer Institute, University College London, and receives funding from the Leukaemia Research Fund. J.P. Allison is an investigator of the Howard Hughes Medical Institute and holds the David H. Koch Chair in Immunological Studies at the Memorial Sloan-Kettering Cancer Center. This work was also supported by the Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center funded by W.H. Goodwin and A. Goodwin, and the Swim Across America Foundation.

J.P. Allison is an inventor of intellectual property licensed by the University of California to Medarex, and is a consultant for Medarex and Bristol Meyers Squibb. The authors have no further conflicting financial interests.