Costimulation Induces CD4 T Cell Antitumor Immunity via an Innate-like Mechanism

SUMMARY

Chronic exposure to tumor-associated antigens in-activates cognate T cells, restricting the repertoire of tumor-specific effector T cells. This problem was studied here by transferring TCR transgenic CD4 T cells into recipient mice that constitutively express a cognate self-antigen linked to MHC II on CD11c-bearing cells. Immunotherapeutic agonists to CD134 plus CD137, “dual costimulation,” induces specific CD4 T cell expansion and expression of the receptor for the Th2-associated IL-1 family cytokine IL-33. Rather than producing IL-4, however, they express the tumoricidal Th1 cytokine IFNγ when stimulated with IL-33 or IL-36 (a related IL-1 family member) plus IL-12 or IL-2. IL-36, which is induced within B16–F10 melanomas by dual costimulation, reduces tumor growth when injected intratumorally as a monotherapy and boosts the efficacy of tumor-nonspecific dual costimulated CD4 T cells. Dual costimulation thus enables chronic antigen-exposed CD4 T cells, regardless of tumor specificity, to elaborate tumoricidal function in response to tumor-associated cytokines.

In Brief

Morales Del Valle et al. analyze how immunotherapeutic costimulatory agonists specific to CD134 and CD137 overcome peripheral tolerance mechanisms that restrict effector T cell programming. Costimulation enables CD4 helper T cells to secrete the tumoricidal cytokine IFNg via an innate- like TCR-independent process that involves stimulation with IL-1 family cytokines.

Graphical Abstract

INTRODUCTION

Specific T cell antigen, costimulation, and cytokine stimulation act as 3 signals required for productive T cell responses, but in the absence of costimulation (Jenkins and Schwartz, 1987) or when out of sequence (Sckisel et al., 2015; Urban and Welsh, 2014), T cells become poorly primed or dysfunctional. The type of costimulation or cytokine receptor stimulation is critical to inducing effective T cell responses that can resolve infections or reduce tumor burdens. Much attention has been given to the cadre of costimulators in the tumor necrosis factor receptor (TNFR) and immunoglobulin (Ig) superfamilies, and many are noted for their immunotherapeutic potential (Esensten et al., 2016; Sanmamed et al., 2015). Similarly, the third signal provided by cytokines, including interleukin-1 (IL-1) and common γ-chain family members (Netea et al., 2015; Yamane and Paul, 2012), is still expanding, and their importance in programming effector differentiation or T cell receptor (TCR)-independent T cell responses is paramount (Guo et al., 2012). The value of this knowledge is apparent since both the careful consideration of adjuvants for vaccine development and the use of specific costimulators and cytokines in cancer is evident. Thus, “skill acquisition” by effector T cells can be controlled by the type of signal 2 or 3 provided, but the amount of TCR priming through antigen presentation in vivo is the least understood in the context of effector T cell function.

A major impediment to effective antitumor immunity is the chronic exposure of specific T cells to cognate tumor epitopes that can render them dysfunctional, such that the repertoire of functional T cells capable of recognizing tumor epitopes only do so with low avidity (Colella et al., 2000; Drake et al., 2005; Ly-man et al., 2004; Morgan et al., 1998; Shrikant et al., 1999; Sta-veley-O’Carroll et al., 1998; Willimsky and Blankenstein, 2005). For instance, specific T cells that have infiltrated into tumors (TIL) typically become functionally “exhausted” (Baumeister et al., 2016), similar to responses that develop during chronic viral infections (Wherry and Kurachi, 2015). Exhausted T cells are phenotypically marked by the expression of checkpoint receptors such as cytotoxic T lymphocyte-associated protein 4 (CTLA-4), programmed cell death-1 (PD-1), and lymphocyte-activation gene-3 (LAG-3) that transmit inhibitory signals to rein-force the non-functional state (Anderson et al., 2016; Baumeister et al., 2016). Antagonists to these checkpoints provide therapeutic benefit to patients with certain types of cancer; however, checkpoint therapy cures only a minority of patients or elicits partial clinical responses (Brahmer et al., 2012; Hodi et al., 2010; Sharma and Allison, 2015; Topalian et al., 2012).

Since the presentation of tumor epitopes by antigen-presenting cells (APCs) expressing low levels of costimulatory ligands also results in non-productive antitumor T cell responses (Diehl et al., 1999; Sotomayor et al., 1999), a complimentary approach to cancer immunotherapy involves the application of agonists to TNFR superfamily costimulatory receptor members such as CD134 (OX40), CD137 (4–1BB), glucocorticoid-induced TNFR family related protein (GITR), and CD27 (Ascierto et al., 2010; Cohen et al., 2006; Roberts et al., 2010; Weinberg et al., 2011). Nevertheless, even though costimulatory agonists can drive effector differentiation, effector T cells remain highly susceptible to inactivation from persistent antigen (Higgins et al., 2002; Kreuwel et al., 2002). Although TCR-proximal signaling becomes dis-engaged in T-helper 1 (Th1) effector CD4 T cells exposed to persistent antigen, the gene encoding the tumoricidal cytokine interferon g (IFNγ) remains in a transcriptionally competent state (Long et al., 2006), raising the possibility that alternate TCR-independent triggering mechanisms may effectively elicit IFNγ secretion by TIL.

We developed an in vivo model in which specific CD4 T cells are chronically exposed to a specific self-antigen presented by major histocompatibility complex (MHC) class II⁺ CD11c⁺ cells at either a high or a low level. The simplicity of this system allows the analysis of the responding T cells while testing therapeutic costimulation. As expected, persistent antigen induces checkpoint molecule expression similar to what has been observed during chronic infection (Barber et al., 2006). Dual costimulation combination immunotherapy using agonists to CD134 plus CD137 elicits potent therapeutic antitumor immunity in murine tumor models (Adler and Vella, 2013) and is being tested in human clinical trials (e.g., NCT02315066), Dual costimulation programs specific CD4 T cells exposed to persistent antigen to produce IFNγ via a TCR-independent mechanism that involves stimulation with either of the IL-1 family members IL-33 or IL-36, plus either of the Janus kinase/signal transducer and activator of transcription (JAK/STAT) cytokines IL-12 or IL-2. Furthermore, IL-36, which can be induced in the microenvironment of B16-F10 melanomas by dual costimulation (Tsurutani et al., 2016), not only impeded B16 tumor growth when directly injected into tumors as a monotherapy but also boosted the ability of persistent antigen-exposed dual costimulated CD4 T cells to control tumor growth. Hence, an important feature of dual costimulation immunotherapy may be its ability to program TIL to elaborate their tumoricidal function even in the absence of strong TCR ligands.

RESULTS

Differential T Cell Programming Relies on Costimulation and TCR Occupancy

Because peripheral T cell tolerance is induced by chronic exposure to specific antigen, we developed an in vivo model of antigen persistence, permitting the tracking of both the TCR-specific CD4 T cells and the cognate antigen-presenting dendritic cells. A construct encoding the Eα-peptide (Grubin et al., 1997;McNicholas et al., 1982) linked to the IA^b beta chain, as previously described (Ignatowicz et al., 1996), driven by the CD11c promoter (Figure S1A) was used to generate 2 transgenic founder mouse lines on the C57BL/6 (B6) background. One expressed high and the other very low Eα-IA^b on CD11c cells (here- in referred to as High or Low, respectively), as measured by flow cytometry staining using the YAe monoclonal antibody (mAb) that detected the Eα-IA^b complex on splenic CD11c⁺ cells taken from High but not Low mice (Figure S1B; data not shown). We reasoned that this system could thus be used to study the effects of persistent antigen (Eα-peptide) exposure at high or low levels using TEα CD4⁺ T cells that bear a Vα2 and Vβ6 transgenic TCR specific to the Eα-IA^b complex (Grubin et al., 1997). Confirming that TEα CD4 T cells (herein referred to as specific CD4 T cells) respond to the transgenic Eα-IA^b complex in vivo, naive TEα cells (1.7 × 10⁶) underwent robust clonal expansion in the spleens of High but not control B6 recipient mice that do not express Eα. Clonal expansion in the High mice peaked at day 5, and the TEα cell frequency remained elevated relative to B6 controls at day 14 (Figure S1C). Furthermore, this robust expansion of the specific TEα CD4 T cells in the spleens of High mice occurred following the transfer of a range of naive TEa cells (2 3×10⁴–4.9 3×10⁶), as well as in several lymphoid and non-lymphoid sites that included peripheral lymph nodes (PLNs), mesenteric lymph nodes (MLNs), liver, and lung (Table S1).

We also reasoned that this system would be ideal for developing strategies to overcome peripheral CD4 T cell tolerance induced by persistent antigen that could be critical for the development of therapeutics to boost immunity while preventing the suppression of T cell function as is typically important during cancer and infection. T cell costimulation through CD134 and CD137 induces robust CD8 and CD4 T cell immunity and is a powerful immunotherapeutic approach in cancer (Adler et al., 2017). To test whether dual costimulation (anti-CD134 agonist mAb combined with anti-CD137 agonist mAb) is able to prevent persistent antigen-induced CD4 T cell tolerance, specific TEα T cells were transferred (2.5 × 10⁴) into the CD11c-Eα-IA^b High or Low transgenic mice on day 0, and their response in the presence or absence of day 1 dual costimulation was tracked by flow cytometry. Spleens harvested 6 days after transfer and analysis of specific CD4 T cells (Vα2⁺Vβ6⁺) in High mice demonstrated much greater clonal expansion compared to Low mice (Figures 1A and 1B). Dual costimulation (anti-CD134 and anti-CD137) substantially potentiated specific T cell expansion in both High and Low mice, being greater in High mice. Thus, antigen levels and enforced costimulation control the extent of clonal expansion, raising the possibility that they also affect effector differentiation, as seen in other systems using antigen dose or altered peptide ligands (Collins and Frelinger, 1998). Since dual costimulation induces IFNγ potentiation (Lee et al., 2007), T-bet was assessed in the specific CD4 T cells after transfer into High or Low mice. Enforced dual costimulation clearly increased T-bet levels (mean fluorescence intensity [MFI]) in specific CD4 T cells from the High mice, and the percentage was also significantly increased compared to Low counterparts (Figures 1C and 1D). Dual costimulation induces strong cytotoxic programing in both CD8⁺ and CD4⁺ T cells (Cuadros et al., 2005; Gray et al., 2008; Lee et al., 2004; Qui et al., 2011), and thus the level of granzyme B in the specific CD4 T cells was examined and shown to be the greatest in dual costimulated High mice (Figures 1E and 1F). Nevertheless, dual costimulation reciprocally enhanced eomesodermin (Eomes) in the specific CD4 T cells from the Low mice. These data suggest that both antigen level and dual costimulation control effector CD4⁺ T cell differentiation, and dual costimulation with low but persistent levels of antigen was seemingly unable to promote the expression of the cytotoxic effector molecule granzyme B (Figure 1E).

Figure 1. Specific CD4 T Cell Expansion and Differentiation in Response to Chronic Anti-gen and Dual Costimulation.

CD4⁺ TEα cells were adoptively transferred into High or Low mice on day 0, followed by immunotherapy with either dual costimulation (DCo) or control IgG on day 1. Splenocytes were harvested on day 6 for analysis via flow cytometry. Cells are gated on forward scatter (FSC) and side scatter (SSC).

(A) Vα2 and Vβ6 expression on CD4⁺ cells. The percentages of gated regions are indicated.

(B) Quantification of (A). n = 1–3 per experiment across 10–13 experiments, for a total of n = 10–16.

(C) T-bet expression in CD4⁺ Vα2⁺ Vb6⁺ TEα cells from splenocytes (top) and peripheral lymph node (PLN) cells (bottom). Numbers denote MFI.

(D) Quantification of (C). n = 1 per experiment across 5–6 experiments.

(E) GzmB and Eomes expression on TEα cells from spleen (top) and PLN (bottom). Percentage of cells in each quadrant is indicated.

(F) Quantification of (E). n = 1 per experiment across 8–11 experiments for spleen and 3–5 experiments for lymph node.

p values were calculated using Student’s unpaired two-tailed t test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Error bars in graphs depict SD.

Surface Phenotype of CD4 T Cells Programmed by Dual Costimulation and Persistent Antigen

Persistent antigen presentation typically induces T cell check-point receptor expression (Anderson et al., 2016; Baumeister et al., 2016). Dual costimulation substantially increased the expression of the checkpoint receptor LAG-3 (MFI) on spleen-specific CD4 T cells in the High mice, whereas the same CD4 T cells from the Low mice did not express appreciable LAG-3 with or without dual costimulation (Figures 2A and 2C). This is in sharp contrast to the differential upregulation of Eomes (Figures 1E and 1F). PD-1 was upregulated on the specific CD4 T cells in all of the groups after transfer regardless of costimulation, although those from the High mice exhibited the highest PD-1 expression (Figures 2B and 2C). This co-expression of LAG-3 and PD-1 on the dual costimulated CD4 T cells from High mice was consistent with a phenotype of exhausted T cells (Anderson et al., 2016). To assess whether these specific CD4 T cells were functionally exhausted or whether LAG-3 and PD-1 expression was simply indicative of an activated state (Goldberg and Drake, 2011), these cells were analyzed for re-stimulation potential. Specifically, dual costimulated specific CD4 T cells recovered on day 6 from High or Low mice or from B6 mice immunized with Eα-peptide were analyzed for intracellular IFNγ expression following in vitro re-stimulation with specific Eα-peptide, phorbol myristate acetate (PMA) + ionomycin (P+I), or media only (control) (Figure 3A). This media- only culture detected little IFNβ expression regardless of costimulation or the antigen load, suggesting that even in the High mice there was minimal ongoing in vivo stimulus at this time point. When cognate Eα-peptide was added, there was a roughly 10-fold increase of IFNγ synthesis in the High dual costimulation group and a less consistent increase in the High IgG control (Figures 3A and 3B). Overall, the same trend was seen in the Low and Eα-peptide-immunized B6 groups. The day 12 dual costimulated specific CD4 T cells also mounted substantial IFNγ synthesis following re-stimulation Figure 3C), and the recall responses were specific since YAe mAb, which blocks Eα-IA^b (Murphy et al., 1989), inhibited IFNγ production elicited by the Eα-peptide but not PMA + ion- omycin, which by passes TCR-proximal signaling (Figure 3A). In sum, the specific effector CD4 T cells from the dual costimulated primed High and Low groups were just as responsive, if not more responsive, as those from the peptide-immunized mice in which antigen presentation is only transient. This demonstrated that the specific CD4 T cells exposed to persistent antigen plus dual costimulation had become effectors. Nevertheless, the specific effector CD4 T cells must be constrained in vivo since antigen is persistent, which would otherwise lead to fulminant autoimmune disease.

Figure 2. Surface Phenotype of CD4 T Cells Programmed by Dual Costimulation and Persistent Antigen.

Adoptive transfer of TEa cells into High or Low mice on day 0 followed by DCo or IgG immunotherapy on day 1. Flow cytometry of day 6 splenocytes was analyzed. Cells are gated on FSC, SSC, CD4⁺, Vα2⁺, and Vβ6⁺.

(A) LAG-3 expression on TEα cells from spleen (top) and PLNs (bottom). Numbers denote MFI.

(B) PD-1 expression on TEα cells from spleen. The percentages of each gate are noted.

(C) Top: quantification of the percentage of LAG-3⁺ TEα cells from (A). n = 1–2 per experiment across 4 experiments, for a total of n = 4–5. Bottom: quantification of percentage of PD-1⁺ TEα cells from (B) (bottom left), and those from PLNs (bottom right). n = 1–2 per experiment across 5 experiments, for a total of n = 5–6.

p values were calculated using Student’s unpaired two-tailed t test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Error bars in graphs depict SD.

Figure 3. Dual Costimulation Programs Specific CD4 T Cells Exposed to Chronic Antigen to Express IFNγ in Response to TCR Stimulation.

Adoptive transfer of TEα cells into High and Low mice on day 0 was followed by DCo or IgG immunotherapy on day 1. Day 6 splenocytes were stimulated ex vivo with 5 μg/mL Eα -peptide, 1X PMA/I (P+I), or media control for 4 h with brefeldin A (BFA), followed by intracellular staining. Cells were additionally treated with control or YAe blocking antibody and are gated on FSC, SSC, CD4⁺, Vα2⁺, and Vβ6⁺.

(A) IFNγ expression on TEa cells from High mice. n = 1 per experiment across 3 experiments.

(B) Top: quantification of the data from (A) across 3 experiments (Exp). Middle: quantification of the same data, but from Low mice. Bottom: quantification of the same data, but from B6 mice immunized with soluble Eα-peptide. Numbers denote fold difference (percentage of IFNγ⁺ with Eα over that with media).

(C) Quantification of similar data in High and Low mice from day 12 splenocytes. n = 1–3 per experiment across 3–5 experiments, for a total of n = 3–8. p values were calculated using Student’s unpaired two-tailed t test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Error bars in graphs depict SD.

Dual Costimulation with Chronic Antigen Induces Treg Expansion and IL-33R on Specific CD4 T Cells

One explanation for the constraint of T cells is that during exposure to persistent self-antigen presented on CD11c cells, a sub-population of the specific CD4 T cells convert into peripheral regulatory T cells (Tregs). On the contrary, few specific CD4 T cells from High mice had converted to Tregs (Foxp3⁺ and GITR^hi [McHugh et al., 2002; Shimizu et al., 2002]) whether costimulated or not (Figure 4A, top). Nevertheless, Foxp3⁺ non-specific CD4 T cells (Vβ6⁻V α2⁻) expanded in response to dual costimulation compared to control IgG treatment when specific CD4 T cells encountered either Low or High persistent self-antigen (Figure 4A, bottom, and Figure 4B, middle). In fact, dual costimulated High mice contained >7-fold more splenic Tregs compared to IgG High mice (2.3e−6 ± 1.4 versus 0.3e−6 ±0.21; p < 0.05). Thus, although the conversion of specific CD4 T cells into Tregs was not evident, peripheral tolerance may have resulted from the expansion of non-specific Tregs after dual costimulation.

Figure 4. Specific CD4 T Cells Exposed to Chronic Antigen and Dual Costimulation Do Not Differentiate into Foxp3⁺ Tregs.

High and Low mice were treated with DCo or IgG 1 day after TEα adoptive transfer. Day 6 splenocytes were analyzed via flow cytometry. Cells are gated on FSC, SSC, and CD4⁺ cells; FSC, SSC, CD4⁺, Vα2⁺ and Vβ6⁺ (TEα) cells; or FSC, SSC, CD4⁺, and Vα2/Vβ6 negative or single positive (non-Vα2 Vβ6 cells/non-TEα).

(A) Foxp3 and GITR expression on TEα and non-TEα cells. The percentage of cells in each quadrant is indicated.

(B) Top: quantification of spleen data from (A), along with total CD4⁺ quantification. Bottom: quantification of the same experiment on day 6 PLN cells. n = 1–2 per experiment across 5–6 experiments, for a total of n = 6–7.

(C) IL-33R/ST2 expression on day 6 TEα splenocytes from High and Low mice with either DCo or IgG. Left, flow cytometry plots. Right, quantification of flow data. n = 1 per experiment across 8–9 experiments.

p values were calculated using Student’s unpaired two-tailed t test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Error bars in graphs depict SD.

Upon closer inspection it was observed that the specific CD4 T cells from the High mice with dual costimulation expressed greater levels (>2.5-fold MFI) of the costimulatory molecule GITR compared to control IgG High and dual costimulated and IgG-treated Low mice. This high expression level on the effector cells matched Treg levels of GITR (Figure 4A, top left compared to lower left). It was reasoned that this may signify an unknown relation between Tregs and the dual costimulated specific CD4 T cells (Yuan et al., 2014). For example, Tregs can express the IL-33 receptor (ST2; Vasanthakumar et al., 2015), which supports their maintenance, expansion, and insulin resistance (Bapat et al., 2015; Matta et al., 2014; Schiering et al., 2014). Likewise, dual costimulation also boosts IL-33 responses in specific effector CD8 T cells (Ngoi et al., 2012). In the present study, a subpopulation of specific effector CD4 T cells also expressed ST2, which was enhanced by dual costimulation in the High mice (Figure 4C, bar graph). Thus, the similarities between the dual costimulated T cells and Tregs were found only in the specific CD4 T cells primed in the High mice, which may point to an important role for the amount of serial TCR triggering in preparing effector T cells to respond to IL-33 or perhaps other IL-1 family members.

Powerful TCR-Independent Responses in Dual Costimulated Specific CD4 T Cells after Chronic High Antigen Exposure

ST2 signaling can limit IFNγ synthesis by promoting Th2 differentiation (Liew et al., 2016). Furthermore, in the High mice, a sub-population of specific effector CD4 T cells expressed the Th2 transcription factor Gata-3 (Figure S2). To thus address whether the dual costimulated specific effector CD4 T cells had acquired a hidden Th2 phenotype, specific CD4 T cells from High and Low mice were tested for responsiveness to IL-33. IL-1 family members have been shown to trigger TCR-independent responses in Th2 cells, such as IL-33-induced IL-13 synthesis (Guo et al., 2009). It was therefore postulated that the chronically stimulated effector specific CD4 T cells may mount TCR-independent responses.

Whole day 6 dual costimulated spleens from High and Low mice were stimulated with various cytokine combinations and cytokine production by the specific CD4 T cells was subsequently measured (Figure 5B). Combining IL-2 + IL-33 or IL-12 + IL-33 strongly promoted IFNγ production as measured by intracellular cytokine staining in the specific CD4 T cells from the High mice only. The triple combination IL-2 + IL-12 + IL-33 induced IFNγ synthesis even more potently. Substituting IL-36b, another IL-1 family member, for IL-33 was also very potent. The percentage of IFNγ producing effectors was consistently and significantly higher when comparing cytokine combinations used to stimulate the High versus Low groups (Table S2). We detected very little IL-4 in culture supernatants from Eα -stimulated specific effector CD4 T cells (data not shown). This suggested that a TCR-independent response would support IFN γ synthesis over Th2 cytokines, but that dual costimulation primed only the specific CD4 T cells from the High mice for TCR-independent responsiveness.

Figure 5. Dual Costimulation Programs Specific CD4 T Cells Exposed to Chronic Antigen to Express IFNg in Response to Cyto-kine Stimulation in a TCR-Independent Manner.

DCo immunotherapy of High and Low mice was administered 1 day after adoptive transfer of TEα cells. Splenocytes were harvested on day 6.

(A) Vα2⁺Vβ6⁺ TEα and non-Vα2Vβ6/non-TEα cells were sorted by flow cytometry from whole splenocytes. The graphs depict Vα2 and Vβ6 expression of CD4⁺cells before (pre-sort) and after sort. Numbers denote the percentages of gated regions.

(B) IFNγ expression of day 6 TEα cells pre-sort stimulated with the indicated cytokines (50 U/mL IL-2, 2.5 ng/mL IL-12, 1 ng/mL IL-33, and/or 10 ng/mL IL-36β) overnight, followed by 5 μg/mL BFA for 4 h.

(C) IFNγ expression of sorted TEα and non-TEα cells stimulated with the indicated cytokines overnight, followed by BFA for 4 h.

Numbers in (B) and (C) denote MFI. n = 1–2 per experiment across 3–4 experiments, for a total of n = 4–7.

Nevertheless, a possible explanation for these results is the potential to carry over transgenic cognate antigen on APCs into the culture or that IFNγ secretion was induced by an indirect mechanism. To address these caveats, we sorted the specific CD4 T cells (Figure 5A) and then stimulated the highly pure specific CD4 T cells (Vα2⁺Vβ6⁺) versus purified non-specific CD4 T cells (non-Vα2Vβ6) from the same mice. The specific effectors from the High mice responded robustly to the cytokine combinations, whereas those from the Low mice did not, even though they were exposed to dual costimulation (Figure 5C). Both triple combinations significantly induced IFNγ synthesis in the purified specific effectors from the High mice compared to the Low, and the double combination of IL-2 + IL-33 was significantly higher in the pure population from the High versus Low mice (Table S2). In the non-specific population, there was at best a minute induction of IFNγ with IL-2 + IL-12 + IL-36β in cells from the High mice. These data demonstrate that costimulation is necessary but not sufficient to condition effector cells for TCR-independent cytokine responses; the level of specific antigen clearly plays a role by perhaps powering cytokine receptor signaling.

Specific CD4 T Cells Exposed to Chronically High Antigen with Dual Costimulation Retain Therapeutic Helper Function

To evaluate whether the chronically antigen-exposed specific CD4 T cells are capable of mediating a biomedically meaningful response, they were tested in a tumor rejection model. While CD4 T cells can mediate cytotoxicity against tumors (Quezada et al., 2010; Qui et al., 2011; Xie et al., 2010), they are perhaps more likely to help CD8 T cells to attack tumors (Spitzer et al., 2017). Dual costimulation also programs antigen-primed CD4 T cells to help CD8 T cells, albeit through a non-standard pathway. Thus, dual costimulated CD4 T cells specific to an antigen that is unrelated to the tumor can help tumor-specific CD8 T cells through an incompletely understood antigen-nonlinked mechanism that does not require the relevant MHC class I and class II epitopes to be presented by the same APC (Mittal et al., 2015). Using this model, which appears unique to dual costimulation immunotherapy, the impact of chronic High antigen exposure was examined by retransferring day 6 specific CD4 T cells recovered from dual costimulated High mice into B6 recipients, followed by the implantation of the aggressive melanoma B16-F10 (B16) cell line. B16 tumor growth was unrestrained in these B6 recipients of the retransferred CD4⁺ helpers (Figure 6A, left), but the addition of immunotherapy with the Eα-peptide (helper antigen [Ag]) plus dual costimulation (DCo) significantly reduced the tumor burden (Figures 6A–6C). A second experiment tested whether the chronically High anti-gen-exposed, dual costimulated specific CD4 T cells required subsequent dual costimulation treatment to reduce the tumor burden. Thus, 3 groups received either (1) day 6 dual costimulated specific CD4 T cells (CD4⁺ helper cells) + Eα-peptide (helper Ag), (2) helper Ag + DCo, or (3) helper Ag + CD4⁺ helper cells + DCo. The group that received helper Ag + dual costimulation immunotherapy (but without dual costimulated CD4⁺ helper cells), group 2, exhibited a significantly reduced tumor burden compared to group 1, which received dual costimulated CD4⁺ helper cells with helper Ag but without subsequent dual costimulation immunotherapy (Figures 6C–6E). This result is consistent with our previous observations that dual costimulation immunotherapy effectively reduces B16 tumor burden (Mittal et al., 2015; Tsurutani et al., 2016). Nevertheless, group 3, in which dual costimulation immunotherapy was combined with retransfer and re-stimulation of dual costimulated tumor-nonspecific CD4 helper T cells, experienced the greatest reduction in tumor burden (Figures 6C–6E). Help with dual costimulation immunotherapy (group 3) was also the most effective at driving the accumulation of CD8⁺ T cells in tumor-draining lymph nodes, and these cells expressed high levels of PD-1 (Figure 6F), perhaps indicating an activated state. These data clearly show that the chronically stimulated specific CD4 T cells can still provide help for combating tumor growth via cell therapy (Anderson et al., 2017), but there is a requirement for immunotherapeutic costimulation.

Figure 6. Specific CD4 T Cells Exposed to Chronic Antigen and Dual Costimulation Are Capable of Providing Therapeutic Antitumor Help.

(A and B) Day 6 fluorescence-activated cell sorting (FACS) sorted antigen-specific Thy1.1⁺ TCR Tg CD4⁺ T cells (CD4⁺ helper cells, or CD4⁺ help) were transferred from High mice into B16–F10 tumor-bearing (day 3) B6 mice, followed by either no immunization (control, left) or immunization with cognate Eα-peptide (helper Ag) and dual costimulation (DCo, right). (A) Tumor growth curves plotted for each mouse, n = 8 mice/group. (B) Average tumor growth curves calculated from the data in (A). The error bars represent SEMs and the asterisks indicate overall significance between treatment groups by 2-way repeated-measures ANOVA.

(C) Scatterplot of day 12 tumor sizes, with horizontal bars indicating mean values and asterisks indicating significance by 2-tailed unpaired t test.

(D–G) Similar experiment as in (A) and (B), but with the following groups: (1) CD4⁺ helper cells + helper Ag (left), (2) helper Ag + DCo (middle), and (3) CD4⁺ helper cells + helper Ag + DCo (right). (D) Tumor growth curves plotted for each mouse, n = 8–9 mice/group. (E) Average tumor growth curves calculated from the data in (D), with statistics performed similarly to (B). (F) Scatterplot of day 15 tumor sizes, with horizontal bars indicating mean values and asterisks indicating significance by 1-way ANOVA plus Tukey’s post-test. (G) FACS analysis of day 15 tumor-draining lymph nodes (TDLN) showing the percentage of CD8⁺ cells (left) and the PD-1 mean fluorescence intensity (MFI) of these CD8⁺ cells. Asterisks indicate significance by 1-way ANOVA plus Tukey’s post-test.

IL-36 Boosts the Therapeutic Efficacy of Dual Costimulated CD4 T Cells Exposed to Persistent High Antigen

Given that the dual costimulated CD4 T cells exposed to persistent high antigen were able to facilitate an anti-melanoma response (Figure 6) and that IL-1 family members IL-33 and IL-36 can, in conjunction with IL-12 or IL-2, trigger these CD4 T cells to produce the tumoricidal cytokine IFNγ (Figure 5; Table S2), we asked whether increasing intratumoral IL-1 family cytokine availability can boost immunotherapy. We initially focused on IL-36, since the dual costimulation therapeutic response in the B16 melanoma model tracks with the increased intratumoral expression of IL-36 and IL-36R mRNAs (Tsurutani et al., 2016). Mice with established B16 melanomas that received recombinant IL-36β given as a monotherapy via intratumoral injection on days 3, 5, 7, 9, and 11 post-transplantation exhibited significantly slower tumor growth through day 13 compared to controls receiving PBS injections (Figures S3A–S3C).

Intratumoral IL-36β or PBS was next given to dual costimulation-treated mice that received retransferred TEα CD4 T cells prepared from dual costimulated High primary recipients with or without Eα-peptide (helper Ag) to elaborate their helper function. Treatment with intratumoral IL-36β alone or helper Ag alone did not slow tumor growth compared to controls receiving neither, but mice treated with both intratumoral IL-36β and helper Ag exhibited significantly slower tumor growth (Figures 7A–7C). Furthermore, tumor growth was sufficiently slowed in the IL-36β plus helper Ag treatment group that several mice survived beyond day 20 (Figure S4), in contrast to the other 3 treatment groups in which none of the mice survived beyond day 14 due to their large tumor sizes. These data suggest that dual costimulation conditions CD4 T cells to elaborate therapeutic effector function in response to IL-1 family cytokines such as IL-36β that are present within the tumor microenvironment, even when they are unable to receive a tumor-specific TCR trigger.

Figure 7. Intratumoral IL-36β and Activated CD4⁺ TEa Cells from Dual Costimulated High Mice Synergize to Boost Dual Costimulation Melanoma Immunotherapy.

B6 recipients bearing day 3 established B16-F10 tumors received dual costimulation along with 3.3 × 10⁵ retransferred day 6 FACS sorted TEα Thy1.1⁺ TCR Tg CD4 T cells recovered from dual costimulated High recipient mice. These tumor-bearing mice were divided into 4 groups that received (1) a single intraperitoneal (i.p.) injection of 250 μg soluble Eα -peptide to activate the re-transferred specific CD4 helper T cells plus 1 mg intratumorally injected IL-36β on days 5, 7, 9, and 11 (+ Ag + IL-36β); (2) Eα -peptide plus intratumoral PBS injections (+ Ag [C0] IL-36β); (3) control rat IgG plus intratumoral IL-36b injections ([C0] Ag + IL-36β); and (4) control rat IgG plus intratumoral PBS injections ([C0] Ag [C0] IL-36b). n = 9–10 mice/group.

(A) Tumor growth curves plotted for each mouse.

(B) Average tumor growth curves calculated from the data in (A). The error bars represent SEMs and the asterisks indicate the overall significance between treatment groups by 2-way repeated-measures ANOVA.

(C) Scatterplot of day 14 tumor sizes, with horizontal bars indicating mean values and asterisks indicating significance by 1-way ANOVA plus Tukey’s post-test.

DISCUSSION

Transient antigen presentation during vaccination elicits T cell clonal expansion, effector differentiation, and memory formation. In contrast, persistent antigen presentation during chronic infections can drive cognate T cells into a dysfunctional, exhausted state (Wherry and Kurachi, 2015). This tolerogenic response helps to limit autoimmunity directed toward constitutively expressed self-antigens, but it also suppresses tumor-specific CTLs (Baumeister et al., 2016). Immunotherapy using enforced costimulation can boost effector T cell responses and antitumor immunity (Berraondo et al., 2016; Diehl et al., 1999; Maxwell et al., 1999; Sotomayor et al., 1999; Weinberg et al., 2011), although a combination of central and peripheral tolerance mechanisms likely limits the avidities with which responsive T cells can recognize tumor epitopes (Colella et al., 2000; Morgan et al., 1998; Shrikant et al., 1999; Willimsky and Blankenstein, 2005). Hence, the efficacy of costimulation-based cancer immunotherapy may be augmented by strategies to boost weak TCR signals or even trigger TCR-independent effector function intratumorally.

In this report, we demonstrate that dual costimulation using agonists to CD134 plus CD137 preserves effector functionality in CD4 T cells exposed to persistent self-antigen presented on CD11c⁺ dendritic cells. Although these specific CD4 T cells ex-pressed surface markers that are consistent with an exhausted phenotype (PD-1 and LAG-3; Figure 2), they also had differentiated into effectors, as evidenced by their robust clonal expansion and expression of transcription factors and other molecules associated with effector T cells (T-bet, Eomes, and granzyme B [GzmB]; Figure 1). Furthermore, they expressed the tumoricidal effector cytokine IFNγ following stimulation with cognate peptide (Figure 3) and provided therapeutic help in a melanoma model (Figures 6 and 7). Foxp3⁺ Tregs expanded in parallel with the specific CD4 T cells exposed to persistent antigen during CD134 plus CD137 dual costimulation (Figure 4), as we previously observed in a different CD4 T cell transfer system (St Rose et al., 2013). This may have been mediated in part through paracrine IL-2 secreted by the specific CD4 T cells (Long and Adler, 2006; O’Gorman et al., 2009), as well as direct effects of the CD134 and CD137 agonists on the Tregs (Xiao et al., 2012; Zhang et al., 2007). Given that the specific CD4 T cells did not appear to convert into Tregs (Figure 4A), the parallel expansion of pre-existing Tregs may have helped to limit the autoimmune potential of the specific dual costimulated CD4 T cells.

Perhaps the most intriguing observation in this study was that the dual costimulated specific CD4 T cells were also able to produce IFNγ through a TCR-independent process that involves stimulation with cytokine combinations that include at least one JAK/STAT activating cytokine (IL-12 or IL-2) plus an IL-1 family member (IL-33 or IL-36) (Figure 5; Table S2). This innate- like cytokine response can also be triggered in effector CD8 T cells by IL-12 or IL-2 plus one of the IL-1 family members IL-18, IL-33, or IL-36 (Berg et al., 2002; Ngoi et al., 2012; Tsurutani et al., 2016). This TCR-independent pathway is analogous to cytokine responsiveness by innate lymphoid cells (Licona-Limón et al., 2013; Mjösberg et al., 2011) and also by Th2 cells responding to IL-33 with the subsequent release of IL-13 (Guo et al., 2015; Molofsky et al., 2015) and Th1 cells that synthesize IFNγ in response to IL-12 plus IL-18 (Yang et al., 1999; Yoshimoto et al., 1998). The intrinsic mechanism is not completely solved, but strong evidence suggests that IFNγ induced by IL-12 with IL-18 is p38 mitogen-activated protein kinase (MAPK) dependent (Yang et al., 2001). This 2-cytokine CD4 T cell triggering process should not be confused with a role for IL-1 family members in T cell subset effector differentiation (Ben-Sasson et al., 2013; Ngoi et al., 2012; Tsurutani et al., 2016); it rather represents an innate-like TCR-independent response manifested in already differentiated effector or memory T cells, similar to innate lymphoid cells that can respond to certain cytokine mixtures (Bando and Colonna, 2016).

Our current data led us to posit that part of the dual costimulation therapeutic response involves the stimulation of TIL with cytokines present in the tumor microenvironment that trigger TCR-independent elaboration of effector function. Thus, IL-36 and IL-36R mRNAs are elevated in dual costimulated compared to control B16 tumors (Tsurutani et al., 2016). In a separate experiment, the alarmin IL-33 that is readily released from necrotic cells (Haraldsen et al., 2009; Lefrançais and Cayrol, 2012) was detected in both dual costimulated and control B16 tumor lysates at levels R25-fold above background (n = 5 per group, data not shown). IL-12 was also abundant within both dual costimulated and control B16 tumors (13,882 ± 2,419 pμ/mg protein lysate versus 8,343 ± 999, respectively, p < 0.067, n = per/group). To begin testing this idea, we used intratumorally injected recombinant IL-36β as a monotherapy, which dramatically slowed tumor growth compared to PBS controls (Figure S3). Furthermore, intratumoral IL-36b treatment boosted the capacity of the chronically antigen-exposed dual costimulated CD4 T cells to provide therapeutic help (Figure 7).

Because these dual costimulated CD4 T cells recognize an epitope that is tumor unrelated (Mittal et al., 2015), they can only be activated to elaborate their therapeutic helper function within the tumor microenvironment through TCR-independent processes. This raises an intriguing possibility for how adoptive cell therapy (Yang and Rosenberg, 2016) and dual costimulation may be effectively combined. Thus, given our current (Figures 6 and 7) and previous data (Mittal et al., 2015) demonstrating that CD4 T cells can provide therapeutic help without the necessity of using tumor antigen, possibly tetanus or influenza specific CD4 T cells could be stimulated in vitro for growth and differentiation followed by cell therapy and in vivo priming with specific (but tumor unrelated) antigen plus dual costimulation. It may also be beneficial to induce TCR-independent responses at the tumor site, such as those tested here with IL-2 and IL-12, and IL-33 and IL-36 (Figure 5; Table S2). It was recently shown that tumor cells expressing IL-36 are readily controlled by the immune system (Wang et al., 2015), and we found here that providing recombinant IL-36β intratumorally impedes tumor growth when given both as a monotherapy (Figure S3) and in concert with adoptive cell therapy using dual costimulated CD4 T cells and cognate tumor-unrelated antigen (Figure 7). This approach thus adds another tool to inhibit tumor cell growth without purposefully engaging a tumor antigen-specific response. Dual costimulation-based immunotherapy can program effector T cells with the capacity to elaborate their effector functions within the tumor microenvironment, regardless of their tumor specificity.

Dual costimulation robustly augmented CD4 T cell clonal expansion regardless of whether persistent antigen was expressed at high or low levels, albeit the frequency of expanded specific CD4 T cells was greater in the High model (Figures 1A and 1B). These initial results fit a quantitative model as seen previously in which antigen level, TCR occupancy, and costimulation linearly increased T cell activation (Marchingo et al., 2014). Nevertheless, a qualitative difference in CD4 T cell subset phenotyping was also evident in our study. Dual costimulation enhanced T-bet in the specific CD4 T cells derived from the High mice (Figures 1E and 1F), whereas Eomes was greatly increased in the same cells when primed in the Low mice. This is an interesting contrast since our previous data show that dual costimulation and other data using CD137 costimulation alone induce Eomes (Curran et al., 2013; Mittal et al., 2018; Qui et al., 2011); however, in the High mice we detected T-bet and very little Eomes. It is possible that Eomes may have an altered kinetic profile under certain conditions, although the up-regulation of granzyme B in T-bet⁺ Eomes⁻CD4 T cells was unexpected.

Another notable difference was that dual costimulated effector T cells from the High mice were readily capable of TCR-independent IFNγ expression in response to JAK/STAT plus IL-1 family cytokine combinations, but the T cells from the Low mice were not (Figure 5). One possibility is that the specific CD4 T cells primed in the Low mice did not respond to dual costimulation, but as measured by clonal expansion, the response was dramatic (Figures 1A and 1B). In addition, the effector CD4 T cells from the Low mice had the ability to produce substantial amounts of IFNγ in response to recall antigen and PMA + ionomycin stimulation (Figure 3C), consistent with their expression of Eomes (Figures 1E and 1F). Third, the effector CD4 T cells from the High and Low mice both expressed the ST2 (Figure 4C). Hence, greater levels of TCR stimulation (priming in High mice) were required for programming the innate-like cytokine response (Figure 5; Table S2). This demonstrates that TCR stimulation, costimulation, and antigen level are not always additive in determining the outcome of T cell priming, since under these conditions a qualitative difference in functional response was clearly evident. Furthermore, while this report confirms that manipulation of T cell costimulation using biologics critically affects T cell response outcome, it also suggests that the intensity of TCR signaling is high on the “food chain” along with costimulation. Thus, vaccine development and immunotherapy approaches should consider that levels of available specific antigen can yield different effector T cell subset outcomes, even with powerful costimulators.

Ultimately, a question that remains open is how differences in TCR occupancy lead to the different functional outcomes observed here. Enforced costimulation is essential since in its absence we only observed minor differences between the specific CD4 T cells primed in the High versus Low mice. With the exception of clonal expansion, which was greater in the High recipients (Figures 1A and 1B), a minor increase in Eomes-expressing T cells from the Low mice was evident, but this was substantially enhanced after dual costimulation (Figure 1F). Similarly, recall responses to antigen were comparable between the specific CD4 T cells from the High and Low mice. Thus, the TNFR family of costimulators engages pathways in concert with the amount of TCR signaling to skew programming by affecting transcription factor expression and perhaps other factors. One possibility is that triggering TCR and TNFR costimulation may integrate with the other pathways. For instance, TCR and TNFR costimulation may generate a unique pathway emanating from JAK/STAT (IL-2 and IL-12) and MyD88 (IL-33 and IL-36) signaling. This could be in the form of novel transcription factor complexes, which may help to explain the synergistic IFNγ production that is observed in response to these cytokines.

STAR★METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Anthony Vella (vella@uchc.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice were cared for at the UConn Health animal facility and fed ad libitum. All experimental protocols conform to federal regulatory standards, and were approved by the UConn Health Institutional Animal Care Committee (IACUC), Farmington, CT. C57BL/6 (B6) mice were purchased from Jackson Laboratory, Bar Harbor ME. TEα and CD11c-Eα^{hi and low} transgenic mice were bred in-house. Experiments described in Tables S1 and S2, Figures 1, 2, 3, 4, and 5 and S1 and S2 utilized both male and female adult mice that ranged between 6 weeks to 6 months of age. Adult 6-week to 3-month-old females were used exclusively for the tumor immunotherapy studies described in Figures 6 and 7 and S3 and S4.

TRANSGENIC MICE

TEα TCR transgenic mice that produce CD4 T cells that recognize Eα peptide_52–68 were previously described (Grubin et al., 1997). The CD11c-Eα^hi (High) and CD11c-Ea^lo (Low) transgenic mice were generated in the Gene Targeting and Transgenic Facility (UConn Health, Farmington CT). Specifically, the High and Low mice express Eα peptide52–68 genetically linked to the MHC II IA^b beta chain (Ignatowicz et al., 1996) under control of the CD11c promoter (Brocker et al., 1997; Khanna et al., 2010) (depicted in Figure S1). The CD11c⁺ cells surface express the Eα peptide connected to the MHC IA^b b chain in the presence of the MHC II α chain (Figure S1A). Two founder lines were characterized: one with high transgene expression (High) and one with low expression (Low).

TRANSPLANTABLE TUMOR

The transplantable B16-F10 melanoma was purchased from ATCC. The sex of this cell line is unknown.

METHOD DETAILS

Adoptive Transfer, Immunotherapy and Flow Cytometry

Axial, brachial, and inguinal lymph nodes along with spleen were harvested from naive TEα mice, crushed, red blood cells lysed with 155mM NH₄Cl, and the remaining lymphocytes washed. The percentage of CD4⁺Va2⁺Vb6⁺ TEa cells was determined by flow cytometry. Approximately 2.5 × 10⁴ TEα cells were i.v. transferred into High or Low recipient mice and the following day mice were injected i.p. with 50μg of anti-CD134/OX40 and 25 μg of anti-CD137/4–1BB agonist mAbs, or with 75 μg of rat IgG isotype control antibody (Bio X Cell, West Lebanon, NH) similar to our previous work (Lee et al., 2004; Mittal et al., 2015). Figure 3 also included B6 recipients treated the following day with 100 μg Eα peptide_52–68 (Invivogen, San Diego, CA) i.p. together with 50 μg of anti-CD134 (clone OX86, IgG1) and 25 mg anti-CD137 (clone 3H3, IgG2a) similar to our previous work (McAleer et al., 2010).

Similar to previous flow cytometry staining procedures (McAleer et al., 2007), single cell suspensions from spleen and peripheral lymph nodes (PLN; axial, brachial, and inguinal) were washed, resuspended in FACS buffer (HBSS, 0.1% sodium azide, 10mM HEPES, and 20% fetal calf serum) and kept on ice. Cells were treated with FcR blocking antibody (Unkeless, 1979), followed by in-cubation with the following surface staining mAbs for 30 min at 4^∘C: Vα2, Vβ6, and GITR from eBioscience, Waltham, MA; CD4 and LAG-3 from BD PharMingen, Franklin Lakes, NJ; PD1 from BioLegend, San Diego, CA; and IL-33R/ST2 from MD Bioproducts, Oakdale, MN. Cells were then washed with FACS buffer, and either acquired directly for surface stain or further processed for intracellular staining using the Foxp3 fixation/permeabilization buffer set (eBioscience, Waltham, MA) as per the manufacturer’s instructions. The fixed and permeabilized cells were then incubated with the appropriate antibodies (Eomes, Gata3, T-bet, and Foxp3 from eBioscience, Waltham, MA; GzmB from Invitrogen, Carlsbad, CA; and IFNγ from BD PharMingen, Franklin Lakes, NJ) diluted in 1X permeabilization buffer for 30 min at 4^∘C, washed twice with 1X permeabilization buffer, and then resuspended with FACS buffer. Cells were acquired on a LSRII analyzer (BD Biosciences, Franklin Lakes, NJ) using DIVA software, and then data analyzed using FlowJo software (FlowJo, LLC, Ashland, OR).

Cell Sorting

For cytokine stimulations in Figure 5, peripheral lymph nodes (axial, brachial, and inguinal) plus spleen were pooled from Day 6 High or Low mice that received dual costimulation and then CD4⁺Vα2⁺Vβ6⁺ TEα cells as well as CD4⁺Vα2^negVβ6^neg non-TEα cells were sorted using a FACS Aria (BD Biosciences, Franklin Lakes, NJ). For Figures 6 and 7, TEα cells were sorted based on Thy1.1⁺ staining.

Ex vivo Stimulation and ELISA

After RBC lysis, 0.5−1 × 10⁶ splenocytes were seeded in a 96-well flat-bottom plate in a final volume of 200 μl. For peptide stimulation, cells were treated with 5 μg/ml Eα (Invivogen, San Diego, CA) or 1X PMA + I (Fisher Scientific, Hampton, NH) in the presence of 5 μg/ml BFA (Calbiochem). Cells were further treated with Y-Ae – biotin or mouse IgG2b – biotin control antibody (eBioscience, Waltham, MA), and then harvested after 4 h. Cytokine stimulations were performed in the absence of BFA with the following cytokine concentrations: 50 U/ml IL-2, 2.5 ng/ml IL-12, 1 ng/ml IL-33, and 10 ng/ml IL-36β (R&D Systems, Minneapolis, MN). The following morning, cells were treated with 5 μg/ml BFA for 4 h, and then harvested for analysis via flow cytometry.

IL-12 (p70) and IL-33 levels in B16 tumor lysates were measured using ELISA kits from BD Bioscience and R&D Systems, respectively.

Tumor Immunotherapy

Similar to a recent study (Mittal et al., 2015), C57BL/6 mice received 5 × 10⁵ B16F10 tumor cells inoculated intradermally. On day 3 these mice were adoptively transferred with approximately 10⁶ (for Figure 6) or 3.3 × 10⁵ (for Figure 7) FACS sorted TEα cells that were previously stimulated in High mice with dual costimulation for 6 days. As indicated, recipient tumor bearing mice were immunized with or without 250 μg Eα peptide or DCo (both given i.p.), and/or IL-36β (1 μg) or PBS injected intratumorally at the indicated times. Tumors were measured on the indicated days using calipers (in mm²). As indicated, on the day of harvest the frequency of CD8 T cells and PD-1 expression was analyzed in tumor draining lymph nodes (TDLNs).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using Prism (GraphPad Software). P values were calculated using Student’s unpaired two-tailed t test, unless otherwise indicated in the figure legend: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Error bars in graphs depict standard deviations, unless otherwise indicated in the figure legend.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-OX40 (CD134)	Bio X Cell	Cat# BE0031, RRID:AB_1107592
anti-4–1BB (CD137)	Bio X Cell	Cat# BE0239, RRID:AB_2687721
rat IgG	Bio X Cell	Cat# BE0094, RRID:AB_1107795
Chemicals, Peptides, and Recombinant Proteins
recombinant mouse IL-2	Biolegend	Cat# 554504
recombinant mouse IL-36β	Biolegend	Cat# 555404
FoxP3 fixation/permeabilization buffer set	ThermoFisher	Cat# 00–5523–00
Eα peptide 52–68	Invivogen	custom
PMA + ionomycin	ThermoFisher	Cat# 00–4970–03
brefeldin A (BFA)	ThermoFisher	Cat# B7450
Critical Commercial Assays
IL-12 (p70) ELISA kit	BD Biosciences	Cat# 555256
IL-33 ELISA kit	R&D Systems	Cat# M3300
Experimental Models: Cell Lines
B16-F10 melanoma	ATCC	CRL6475
Experimental Models: Organisms/Strains
CD11c-Eαhi transgenic mouse line	This paper	N/A
CD11c-Eαlow transgenic mouse line	This paper	N/A
TEa TCR transgenic mouse line	Grubin et al., 1997	N/A
C57BI76J mice	Jackson Labs	Cat# 000664
Recombinant DNA
CD11c promoter	Brocker et al., 1997	N/A
Eα peptide52–68 linked to MHC II IAb β chain	Ignatowicz et al., 1996	N/A
Software and Algorithms
Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
FlowJo	FlowJo	https://www.flowjo.com/

Highlights.

• Costimulatory agonists enhance dendritic cell-mediated CD4 T cell function

• Costimulated CD4 T cells respond in an innate-like manner to IL-1 family cytokines

• The IL-1 family member IL-36 promotes TCR-independent antitumor CD4⁺ helper function