Inhibition of Nr4a Receptors Enhances Antitumor Immunity by Breaking Treg-Mediated Immune Tolerance

CD4⁺CD25⁺ regulatory T cells (Treg), characterized by transcription factor forkhead box P3 (Foxp3) expression, play critical roles in maintaining immunologic self-tolerance and homeostasis (1, 2). Tregs provide dominant regulation over self-reactive conventional T cells by inhibiting their expansion and activation (3). Tumor cells are derived from self-tissues, but they could be recognized as nonself and eradicated by the immune system because of the expression of tumor-specific mutated genes. However, tumor cells with fewer immunogenic mutations are close to “self” and likely evade immune surveillance, resulting in tumor progression. In this case, mechanisms for maintaining self-tolerance, such as Tregs, may play undesired roles by establishing immune tolerance against tumors (4, 5). In addition, tumor tissue itself creates a niche that supports survival and function of Tregs (6). An increased abundance of Tregs and a decreased ratio of intratumoral CD8⁺ cytotoxic T cells (CTL) to Tregs have been shown to predict a poor prognosis in various types of human cancers (7). Tregs behave as major obstacles in clinical application of cancer immunotherapy, such as tumor vaccines or immune checkpoint blockade (4, 8). Therefore, breaking immunosuppression by Tregs is important for successful cancer therapy.

Depletion of Tregs or specific disruption of Treg functions is actually thought to be a promising strategy to enhance antitumor immunity (4, 8, 9). Chemical inhibitors targeting signal pathways or molecules that contribute to the suppression activities of Tregs have been intensively studied (10–13). For example, the master transcription factor Foxp3 is an attractive target, and P60, a peptide inhibitor of Foxp3, was reported to inhibit Treg function and improve tumor vaccine efficiency (14). However, there is not enough evidence to show that Foxp3 alone is sufficient to be targeted for disruption of Treg function because multiple factors work cooperatively with Foxp3 to maintain Treg physiology (15, 16).

We recently discovered that the Nr4a family of nuclear orphan receptors, consisting of three isoforms (Nr4a1, Nr4a2, and Nr4a3), redundantly play essential roles in Treg development and function via their ability to directly transactivate Foxp3 expression (17–19). Thymic Treg development was completely inhibited in mice lacking all three of the Nr4a factors on T cells (CD4-Cre Nr4a1^fl/fl Nr4a2^fl/fl Nr4a3^−/−), and they died within 3 weeks because of systemic multiorgan autoimmunity (18). In addition, Nr4a factors have been shown to be highly expressed on mature Foxp3⁺ Tregs and necessary for maintaining Treg stability and suppressive activities (19). Of note, Nr4a factors not only regulate Foxp3 expression but also globally regulate the Treg-specific transcriptional program. Therefore, Nr4a-deficient Tregs showed not only reduced expression of Foxp3, but also global dysregulation of Treg signature genes, including Foxp3-independent genes such as Ikzf4 (Eos). Thus, we hypothesize that Nr4a factors are a promising target for cancer immunotherapy to disrupt Treg function within tumors.

In this study, using the mouse tumor transplantation model, we found that selective deletion of Nr4a1/Nr4a2 within Foxp3⁺ Tregs significantly suppressed tumor growth and induced potent antitumor immune responses. We then searched for pharmacologic modulators of Nr4a factors and identified two well-known drugs: the classical chemotherapeutic agent camptothecin (CPT) as the inhibitor of Nr4a transcriptional activity and cyclooxygenase (COX)-2 inhibitors, such as the celecoxib analogue SC-236, as the inhibitor of Nr4a transactivation. These drugs synergistically exerted potent antitumor immune responses against mouse tumor models in an Nr4a/Treg-dependent manner. We propose that Nr4a factors play important roles in Treg-mediated suppression of antitumor immunity, and they are attractive therapeutic targets for cancer immunotherapy.

All experiments using mice were approved by the Institutional Animal Care and Use Committee (IACUC; approval number 08004) of Keio University and performed according to IACUC guidelines. The Nr4a1- and Nr4a2-floxed mice and Foxp3^YFP-Cre knockin mice were previously described (19). C57BL/6J mice were purchased from Tokyo Laboratory Animals Science. IFNγ-venus reporter mice were reported previously (20). All mice were maintained on a C57BL/6 genetic background and kept in specific pathogen-free conditions at Keio University.

All cell lines were obtained between 2008 and 2014. Authenticated 3LL (Lewis lung carcinoma) tumor cells were obtained from the Japanese Collection of Research Bioresources Cell Bank. MC38 (colon adenocarcinoma) tumor cells were kindly provided by Dr. James P. Allison (Department of Immunology, MD Anderson Cancer Center, Houston, TX). Human embryonic kidney 293T cells were obtained from the ATCC. 3LL cells were maintained in RPMI1640 supplemented with 10% FBS and 1% penicillin/streptomycin. MC38 and 293T cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. All cell lines were frozen down at early passages (<7) and used in the experiments within five passages after thawing. These cells were not further authenticated by our laboratory; however, routine confirmation of in vitro growth properties, morphology, tumor formation in the C57BL/6 syngeneic mouse strain (3LL, MC38), and transfection efficiency (293T) provided evidence of correct cell identity. They were negative for known mouse pathogens, including Mycoplasma.

(S)-(+)-Camptothecin and 7-Ethyl-10-hydroxycamptothecin (SN-38) were purchased from Tokyo Chemical Industry. Prostaglandin E2 (PGE2) and thymidine were purchased from Nacalai Tesque. Z-VAD-FMK was purchased from Peptide Institute. CPT-11 (irinotecan HCl trihydrate) was purchased from Biochempartner. SC-236 was purchased from Sigma-Aldrich. H-89 was purchased from Cayman Chemical. Mitomycin-C was purchased from Kyowa Hakko Kirin. The Screening Committee of Anticancer Drugs inhibitor kits (http://gantoku-shien.jfcr.or.jp/) containing 363 compounds were kindly provided by a Grant-in-Aid for Scientific Research in the Priority Area “Cancer” from the Ministry of Education, Culture, Sports, Science and Technology (Tokyo, Japan).

The following fluorescently labeled antibodies were purchased from BioLegend, eBioscience, or TONBO biosciences: CD4 (RM4-5), CD8 (53-6.7), CD25 (PC61.5), CD3ϵ (145-2C11), CD11b (M1/70), CD11c (N418), CD45.1 (A20), CD80 (B7-1; 16-10A1), CD86 (B7-2; GL1), Foxp3 (FJK-16s), IFNγ (XMG1.2), TNFα (MP6-XT22), IL4 (11B11), CD107a (1D4B), CD152 (CTLA-4; UC10-4B9), Ki-67 (16A8), and Nr4a1 (12.14). Dead cells were gated-out using Fixable Viability Dye eFluor 780 (eBioscience). For intracellular cytokine staining, cells were stimulated with 50 ng/mL phorbol 12-myristate 13-acetate (Sigma-Aldrich) and 500 ng/mL ionomycin (Sigma-Aldrich) in the presence of Brefeldin A (eBioscience) for 4 hours. Cells were stained for surface antigens, then fixed/permeabilized with Foxp3 fixation/permeabilization concentration and diluent (eBioscience), and stained for nuclear proteins or cytokines. For phospho-CREB (pCREB) staining, cells were fixed/permeabilized with 4% paraformaldehyde and ice-cold 90% methanol, and stained with pCREB (Ser133) (87G3) rabbit monoclonal antibody (Cell Signaling Technology) for 1 hour, followed by incubation with Alexa Fluor 488 goat anti-rabbit IgG (H+L; Thermo Fisher Scientific). Stained samples were measured using a BD Canto II (BD Biosciences), and FlowJo software (Tree Star) was used for data analysis.

Before transplantation into mice, 3LL and MC38 tumor cells were grown to 80% confluence, and then counted and suspended in PBS. Each mouse was subcutaneously injected with 2.5 × 10⁵ 3LL or 5.0 × 10⁵ MC38 tumor cells into the right flank on day 0. Tumor growth was monitored every 3 to 4 days throughout the experiment. Tumor volume was calculated using the following formula: 0.5 × ab² (a, major axis; b, minor axis). For inhibitor treatment experiments, mice were injected intraperitoneally with CPT-11 (75 mg/kg) and/or the COX-2 inhibitor SC-236 (3 mg/kg) on days 6 and 9. For CD8⁺ T-cell depletion, mice received an i.p. injection of 400 μg CD8⁺ T-cell–depleting antibody (TIB211) 3 times per week beginning 6 days after tumor implantation.

At the specified time points after tumor implantation and drug treatment, subcutaneous tumors were excised, minced, and digested with 1 mg/mL collagenase D (Roche) and 100 μg/mL DNase I (Roche) at 37°C for 1 hour with gentle shaking. Cells were passed through a 70-μm filter, following low speed centrifugation (400 rpm, 5 minutes) to remove cell debris and epithelial cells. Red blood cell lysis was then performed if necessary, and cells were washed twice with PBS and used for further experiments.

Spleen and lymph node cell suspensions were prepared from the indicated mice, and CD4⁺ T cells were isolated using a mouse CD4⁺T Cell Isolation Kit and autoMACS Pro (Miltenyi Biotec). CD4⁺CD62L⁺-naïve T cells were prepared as previously described (21). For preparation of CD4⁺CD25⁺ Tregs, isolated CD4⁺ T cells were stained with allophycocyanin (APC)-conjugated anti-mouse CD25 antibody, followed by positive selection with anti-APC microbeads using autoMACS. Over 95% of the sorted CD4⁺CD25⁺ cells expressed Foxp3. When Foxp3^YFP-Cre knockin mice were used, MACS-isolated CD4⁺ cells were stained with PerCP-cy5.5 anti-mouse CD4 antibody, and CD4⁺YFP⁺ cells were sorted using a SONY SH800 cell sorter (Sony).

T-cell receptor (TCR) stimulation was performed by incubating T cells with anti-CD28 Ab (57.31; 2 μg/mL) and plate-coated anti-CD3ϵ Ab (145-2C11; 4 μg/mL). Th-skewing conditions in this study were described previously (17). Retrovirus transduction to primary T cells was performed as described previously (16, 21). All cultures were performed in RPMI1640 supplemented with 10% FBS, 1% penicillin/streptomycin, 100 nmol/L nonessential amino acids, 2 mmol/L glutamine, and 55 μmol/L 2-mercaptoethanol (Invitrogen).

Isolated Tregs were activated using Dynabeads Mouse T-Activator CD3/CD28 (Thermo Fisher Scientific), according to the manufacturer's instructions. At the specified time points, 2.5 × 10⁵ cells were transferred intravenously into the tumor-bearing host.

Responder T cells (CD45.1⁺CD4⁺CD25⁻ cells) were prepared from C57BL/6J CD45.1 congenic mice and labeled with 1 μmol/L carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen). CFSE-labeled responder cells (5 × 10⁴) were cocultured with unlabeled CD45.1⁻ Tregs at the indicated ratio in the presence of Dynabeads Mouse T-activator CD3/CD28 (Thermo Fisher Scientific) in round-bottomed 96-well dishes at a concentration of 1 bead per cell. The CFSE dilution of CD45.1⁺ responder cells was measured using flow cytometry 96 hours later.

Total RNA was extracted using RNAiso Plus (Takara Bio) or ReliaPrep RNA Cell Miniprep System (Promega), and subjected to reverse transcription (RT) using a High Capacity cDNA Synthesis kit (Thermo Fisher Scientific). PCR analysis was performed using an iCycler iQ multicolor real-time PCR detection system (Bio-Rad) and SsoFast EvaGreen Supermix (Bio-Rad). All primer sets yielded a single product of the correct size. Relative expression levels were normalized to 18S rRNA. Sequences for primers used in this study are available upon request.

293T cells were seeded on 24-well plates and transfected with pCMV expression plasmids and pGL4-luciferase plasmids, 200 ng each, using polyethylenimine. Twenty-four hours after transfection, the indicated chemicals were added to the culture and incubated for an additional 18 hours. Cells were lysed in 100 μL lysis buffer, and luciferase activity was measured using a luciferase substrate kit (Promega) and bioluminescence-monitoring apparatus (CL96; Churitsu). For each transfection, 200 ng β-galactosidase was added as an internal control. Luciferase assay–based drug screening was performed as described in Supplementary Methods.

For comparison of two groups, statistical analysis was performed using a Student t test or a Mann–Whitney U test. ANOVA with Tukey post hoc test was used for multiple comparison experiments. P values < 0.05 were considered to indicate statistical significance (*, P < 0.05; **, P < 0.01; and ***, P < 0.001).

Human effector Treg fraction (Foxp3^hiCD45RA⁻CD25^hi; fraction II) has been shown to be highly suppressive and dominant in tumor tissues, among the three fractions of human Tregs (9, 22). Interestingly, we confirmed that Nr4a factors are most highly expressed in the fraction II population of human Tregs (Supplementary Fig. S1A and S1B). This suggests that Nr4a factors can be a target for modulating tumor-associated Tregs.

Then, to investigate the role of Nr4a/Tregs in antitumor immunity, we prepared Treg-specific Nr4a1/Nr4a2 double-conditional knockout mice (Foxp3^YFP-Cre Nr4a1^fl/flNr4a2^fl/flNr4a3^+/+; hereafter called Nr4a-DcKO mice). We used Foxp3^YFP-Cre Nr4a1^+/+Nr4a2^+/+Nr4a3^+/+ as the WT control. Unlike Nr4a-triple KO (Foxp3^YFP-Cre Nr4a1^fl/flNr4a2^fl/flNr4a3^−/−) mice (19), Nr4a-DcKO mice did not show any spontaneous inflammation, suggesting that systemic Nr4a3 sufficiency suppresses autoimmunity. Overall, the T-cell population in the thymus and periphery, including the Treg fraction in CD4⁺ T cells, was normal in Nr4a-DcKO mice (Supplementary Fig. S2A and S2B). However, Tregs from Nr4a-DcKO mice had significantly lower expression levels of Treg-signature genes, such as Foxp3, CD25, and CTLA-4, at both mRNA and protein levels (Supplementary Fig. S2C and S2D). We also confirmed that in vitro suppression activity of Nr4a-DcKO Tregs was attenuated compared with WT (Supplementary Fig. S2E). In addition, at steady state, IFNγ⁺ cell fraction was increased within both CD4⁺ and CD8⁺ conventional T cells in Nr4a-DcKO mice (Supplementary Fig. S2F, top and middle plots). We did not find any IL4-producing CD4⁺ cells in Nr4a-DcKO mice, suggesting the absence of Th2-type inflammation that is characteristic in TKO mice (Supplementary Fig. S2F, bottom plots).

To examine the effect of the Nr4a1/Nr4a2 deletion in Tregs on antitumor immunity, we performed a mouse subcutaneous tumor transplantation model using the syngeneic 3LL tumor cell line. Although tumor growth rates were indistinguishable between WT and Nr4a1- or Nr4a2-single cKO mice, Nr4a-DcKO mice showed a remarkable delay of tumor growth (Fig. 1A). The established tumor was almost completely eradicated in some Nr4a-DcKO mice, but not in WT mice (Fig. 1B). The tumor-resistant phenotype of Nr4a-DcKO mice was also confirmed when the MC38 (colon adenocarcinoma) cell line was transplanted (Fig. 1C–E).

We then assessed the contribution of immune cells to the suppression of tumor growth observed in Nr4a-DcKO mice. As shown in Fig. 2A, transfer of in vitro–activated WT Tregs into 3LL tumor–bearing Nr4a-DcKO mice almost completely abolished tumor growth inhibition, suggesting that delayed tumor growth was due to functional defects of Nr4a-DcKO Tregs. In addition, antibody-mediated depletion of CD8⁺ CTLs in Nr4a-DcKO mice also accelerated tumor growth, suggesting that CD8⁺ CTLs are the main effector population that restricts tumor growth in Nr4a-DcKO mice (Fig. 2B).

Based on these findings, we further examined immune responses that occurred in 3LL tumor–bearing mice. In tumor-draining lymph nodes (TDLN), the fraction of CD8⁺ T cells was significantly increased in Nr4a-DcKO mice compared with WT mice (Fig. 2C), and cells expressing the proliferation marker Ki-67 were also increased among these CD8⁺ T cells, suggesting the active expansion of CD8⁺ CTLs in TDLNs (Fig. 2D). On the other hand, accumulation of Foxp3⁺ Tregs in TDLNs was inhibited in Nr4a-DcKO mice (Fig. 2E). In addition, Foxp3 protein expression was also significantly downregulated in TDLN Tregs of Nr4a-DcKO mice (Fig. 2F). In tumor tissue, enhanced T-cell infiltration was observed in Nr4a-DcKO mice (Fig. 2G, top plots). The fraction of intratumoral Foxp3⁺ Tregs was not affected (Fig. 2G, bottom plots), but the CD8⁺/Treg ratio was significantly higher in Nr4a-DcKO mice compared with the WT mice (Fig. 2H). In addition, the antitumor effector CD8⁺ CTLs, characterized by the expression of effector cytokines (IFNγ and TNFα) and degranulation marker CD107a, as well as IFNγ⁺Th1 CD4⁺ T cells were drastically increased in Nr4a-DcKO mice (Fig. 2I), suggesting the induction of potent antitumor immune responses. Similar results were obtained from the experiments using MC38 tumor cells (Supplementary Fig. S3A–S3C). We also confirmed that enhanced immune responses in Nr4a-DcKO mice occurred in a tumor antigen-specific manner by examining in vitro recall responses (Supplementary Fig. S3D).

We then examined the mechanism of effector CTL enhancement by Nr4a-DcKO Tregs. Previous research demonstrated that intratumoral Tregs suppress costimulatory signals from antigen-presenting cells by CTLA-4–mediated transendocytosis of B7-molecules CD80/CD86, which in turn suppress CD8⁺ CTL effector function (3, 23–25). We observed significant upregulation of CD80 expression on tumor-infiltrating dendritic cells (DC), as well as downregulation of CTLA-4 on tumor-infiltrating Tregs, in Nr4a-DcKO mice (Supplementary Fig. S3E–S3G). Collectively, these results strongly support our hypothesis that inhibition of Nr4a factors in Tregs breaks immune tolerance against tumor cells and facilitates antitumor activities of tumor-infiltrating effector CTLs.

Crystal structure analysis revealed that Nr4a factors function as ligand-independent transcription factors, and their pharmacologic antagonists have not been identified (26). Therefore, we conducted reporter assay-based drug screening to identify functional Nr4a inhibitors from a chemical library (Methods summarized in Supplementary Fig. S4A). We looked for compounds that inhibited the transcriptional activity of Nr4a2 and identified CPT as a potent Nr4a inhibitor (an overview and results of the screening are described in detail in Supplementary Fig. S4B and S4C). CPT is a DNA topoisomerase I inhibitor that was shown to be effective in a broad spectrum of tumors (27). CPT inhibited not only Nr4a2, but also Nr4a1 and decreased Nr4a response element (NBRE) reporter activity in a dose-dependent manner (Fig. 3A). CPT also suppressed Nr4a2-induced transactivation of Foxp3-promoter reporter activity (Fig. 3B). Although the CPT analogue topotecan inhibited Nr4a activity similarly to CPT, other topoisomerase inhibitors, such as daunorubicin or etoposide, did not, suggesting that suppression of Nr4a by CPT is independent of inhibition of topoisomerase activity (Supplementary Fig. S4D). In addition, CPT did not affect the transcriptional activity of other nuclear receptors such as estrogen receptor α, thyroid-hormone receptor β, or peroxisome proliferator-activated receptor γ (Supplementary Fig. S4E), revealing that CPT is not a general transcription inhibitor, but is specific to Nr4a.

To further confirm the screening result, we next examined the effect of CPT on mouse primary T cells. CPT treatment suppressed Foxp3 protein expression induced by ectopic expression of Nr4a2 in naïve CD4⁺ T cells (Fig. 3C), demonstrating that CPT also inhibits Nr4a2 transcriptional activity in T cells. In addition, under in vitro helper T-cell differentiation conditions, CPT potently suppressed induction of Foxp3⁺ induced Treg (iTreg) cells while it promoted induction of IFNγ⁺ Th1 cells in a TGFβ-independent manner, at both mRNA and protein levels (Fig. 3D; Supplementary Fig. S5A and S5B). Because deletion of Nr4a in T cells robustly suppresses iTreg differentiation while promoting Th1 differentiation (17), these results indicate that CPT treatment mimicked the effects of Nr4a knockdown in T cells. Similar results were obtained in the presence of the apoptosis inhibitor Z-VAD-fmk, suggesting that these effects are not a result of cytotoxic effects of CPT (Supplementary Fig. S5C).

We then examined the effect of CPT on the stability of Tregs in vitro. We treated CD4⁺CD25⁺ Tregs (>95% Foxp3⁺ Treg) with CPT in the presence of thymidine and Z-VAD-fmk to inhibit cell proliferation and apoptosis, respectively. CPT significantly reduced the expression of Foxp3 and other Treg-related genes in a dose-dependent manner (Fig. 3E and F). CPT-induced destabilization of Foxp3 expression was not observed in Tregs from Nr4a-DcKO mice, suggesting that CPT reduces Foxp3 through the inhibition of Nr4a factors (Fig. 3G). On the other hand, the expression of Nr4a factors themselves was not affected by CPT treatment (Fig. 3E), which is consistent with our proposal that CPT inhibits transcriptional activity of Nr4a factors, but not their expression.

In vivo effects of CPT on Foxp3⁺ Tregs were also examined using CPT-11 (irinotecan), a water-soluble, less toxic prodrug of CPT (28). Because CPT-11 exerted its effect after being metabolized into SN-38 (7-Ethyl-10-hydroxy-camptothecin) in vivo, we confirmed that SN-38 inhibited Nr4a activity in vitro similarly to CPT (Supplementary Fig. S5D and S5E). We also confirmed that any significant side effects, such as weight loss or immunosuppression, were not observed in naïve mice treated with the dose of CPT-11 used in this study (Supplementary Fig. S5F and S5G). Although CPT-11 treatment did not affect the population of T-cell subsets including Tregs in the thymus and periphery (Supplementary Fig. S5H and S5I), the Foxp3 expression levels in Tregs were significantly decreased (Fig. 3H). These data indicate that CPT destabilizes Foxp3 expression both in vitro and in vivo. Taken together, in addition to the well-defined chemotherapeutic effects, our data revealed a previously unidentified inhibitory activity of CPT on Nr4a function.

COX-2 is often constitutively overexpressed in a variety of tumors, and its enzymatic product PGE2 contributes to tumor progression either by direct effects on tumor cells or through the formation of a tumor-promoting microenvironment (29). Interestingly, in intestinal tumors, PGE2 has been shown to induce Nr4a2, thereby supporting survival and proliferation of tumor cells, and COX-2 inhibitors abolished such effects (30, 31). Therefore, we investigated the effect of PGE2 and COX-2 inhibitors on Nr4a1/Nr4a2 expression in Tregs. Transient stimulation with PGE2 alone for 3 hours upregulated both Nr4a1 and Nr4a2 expressions in Tregs in a dose-dependent manner (Fig. 4A and B). In addition, continuous PGE2 stimulation for 18 hours also enhanced Nr4a1/Nr4a2 expression in Tregs that were maintained by the TCR stimulation (Fig. 4C). Induction of Nr4a expression by PGE2 was dependent on cAMP/PKA (protein kinase A), because PKA inhibitor H-89 abolished these effects (Fig. 4B and C). Of note, PGE2 also enhanced expression of Foxp3 and Ikzf4 (Eos), direct targets of Nr4a, in WT Tregs, but not in Nr4a-DcKO Tregs (Fig. 4D). These data indicate that PGE2 promotes the expression of Nr4a factors and regulates Nr4a-dependent gene expression in Tregs.

Both 3LL and MC38 tumor cells constitutively express COX-2 and produce PGE2 (32, 33). Consistently, treatment of Tregs with MC38 culture supernatant significantly induced Nr4a expression in vitro; however, this was attenuated by pretreating MC38 cells with a COX-2 inhibitor (Fig. 4E). These data indicate that tumor-derived PGE2 actually induces Nr4a expression. In an in vivo model, Tregs from the spleen of 3LL tumor–bearing mice showed higher Nr4a expression than that of tumor-free mice, which was abolished by administration of SC-236, a structural analogue of the clinically used COX-2 inhibitor celecoxib (Fig. 4F). Expression of the Nr4a target genes Foxp3 and Ikzf4 was also upregulated in Tregs in tumor-bearing mice, but was reduced by SC-236 treatment (Fig. 4F). In addition, SC-236 treatment downregulated Nr4a1 expression in tumor-infiltrating Tregs and also suppressed CREB (cAMP response element binding protein) activation evaluated by the phosphorylation levels (Fig. 4G). Taken together, these results suggested that tumor-derived PGE2 induces cAMP/PKA activation, which turns on the transcription of Nr4a factors and their target genes in Tregs, and this regulatory axis can be targeted by a COX-2 inhibitor in vivo.

To examine the effect of the above-identified Nr4a inhibitors on antitumor immunity in vivo, we treated 3LL tumor–bearing mice with CPT-11 and SC-236, according to the experimental outline shown in Fig. 5A. We found that each drug alone significantly suppressed 3LL tumor growth, and they showed more potent antitumor effect when used in combination (Fig. 5B). Depletion of CD8⁺ T cells cancelled the inhibition of tumor growth by the inhibitors (Fig. 5C), indicating that the antitumor effects induced by CPT-11 and SC-236 were dependent on CD8⁺ T cells. A significant increase in the frequency of Ki-67⁺CD8⁺ T cells in TDLNs was observed when the mice were treated with these compounds (Fig. 5D, top plots, and E). Combination treatment synergistically reduced not only the Treg population but also the Foxp3 protein levels in Tregs within TDLNs (Fig. 5D, bottom plots). Of note, combination therapy also synergistically enhanced production of IFNγ from CD8⁺ effector T cells within the tumor (Fig. 5F, top plots). Increase in the frequency of IFNγ⁺Th1 CD4⁺ T cells was also confirmed in the inhibitor-treated mice (Fig. 5F, bottom plots). Similar results were obtained in experiments using MC38 tumor cells (Supplementary Fig. S6A–S6C). These data support our hypothesis that pharmacologic inhibition of Nr4a potently enhances antitumor immune responses.

To gain mechanistic insights, we characterized gene expression of Tregs from TDLNs of 3LL tumor–bearing mice treated with or without Nr4a inhibitors. Expressions of Nr4a1/Nr4a2 and Treg-signature genes, including Foxp3, Il2ra (Cd25), Ikzf4 (Eos), and Ctla4, were significantly downregulated by the treatment (Fig. 6A). The suppression activity of Tregs from tumor-bearing mice was higher than that of Tregs from tumor-free mice, but completely attenuated by treatment with the inhibitors (Fig. 6B; Supplementary Fig. S7A). These results indicate that Nr4a inhibitors could reverse Treg activity that was enhanced by the tumor-bearing conditions.

Then to confirm that antitumor effects of the CPT-11/SC-236 combination depend on the inhibition of Nr4a in Tregs, we performed adoptive Treg transfer experiments. The 3LL tumor–bearing mice were treated with these Nr4a inhibitors, and in vitro–activated Tregs were then adoptively transferred. As shown in Fig. 6C and D, WT Treg transfer significantly diminished antitumor effects evoked by the drugs. Conversely, transfer of Nr4a-DcKO Tregs did not reverse the therapeutic effects in the same situation (Fig. 6C and D). In addition, as observed in Nr4a-DcKO mice, the CPT-11/SC-236 treatment dramatically increased the expression levels of the costimulatory molecule CD80 in tumor-infiltrating DCs, which was also abolished by the transfer of WT Tregs, but not by Nr4a-DcKO Tregs (Supplementary Fig. S7B). Taken together, these data strongly support our conclusion that that the combination of CPT and a COX-2 inhibitor exerts potent antitumor immune responses in an Nr4a/Treg-dependent manner in vivo.

In this study, we revealed that Nr4a receptors are involved in Treg-mediated immune tolerance against tumor cells, and genetic inactivation or pharmacologic inhibition of these factors could evoke CD8⁺ CTL-dominant potent antitumor immune responses. We propose two mechanisms of CTL augmentation by Nr4a inhibition in Tregs. First is the impairment of Treg-mediated suppression of CD8⁺ T-cell proliferation. Tregs have been shown to suppress proliferation of CD8⁺ CTLs by depriving IL2 because Tregs constitutively express CD25, which consists of the high-affinity IL2 receptor (34). Consistently, we observed decreased expression of CD25 in Nr4a-DcKO Tregs (Supplementary Fig. S2C and S2D) and reduced suppression activity compared with WT Tregs (Supplementary Fig. S2E). Therefore, reduced expression of CD25 in Nr4a-DcKO Tregs could be a mechanism for enhanced CD8⁺CTL expansion at the TDLNs of DcKO mice. The second mechanism is the failure of downregulation of costimulatory molecules on DCs through CTLA-4. In a mouse pancreatic cancer model, Jang and colleagues showed that Treg depletion evokes a CD8⁺ CTL-dependent antitumor immune response in a CD11c⁺ DC-dependent manner (25). We also confirmed reduced expression of CTLA-4 in Nr4a-DcKO Tregs and a drastic upregulation of CD80 expression on tumor-infiltrating DCs in Nr4a-DcKO mice (Supplementary Fig. S3E–S3G). These observations suggest that Nr4a inhibition abolishes Treg-mediated functional control of DCs. Taken together, Nr4a factors seem to affect various aspects of tumor immunity regulated by Tregs.

In this study, we identified the classical chemotherapeutic agent CPT as an inhibitor of Nr4a transcriptional activity. It is notable that CPT treatment replicated the phenotype of Nr4a deletion in Tregs both in vitro and in vivo. However, it remains unclear how CPT inhibits Nr4a activity, for example, whether or not CPT physically interacts with Nr4a. Although we could not find structure–activity relationships among the drugs we screened here, amodiaquine (AQ) and chloroquine (CQ), Nr4a agonists that directly bind to the ligand-binding domain of Nr4a2 (35), contain an quinoline skeleton similar to CPT. According to that report, AQ and CQ share a 4-amino-7-chloroquinoline entity, and other tested quinoline compounds without that structure do not affect Nr4a activities. Another report suggested that Nr4a agonists and antagonists may bind to the same site, yet the difference in the form of protein–drug interaction determines whether they work as an activator or as an inhibitor (36). Therefore, we speculate that the quinoline skeleton is important for the binding between the drug and Nr4a protein, but the structure of other skeletons or side chains in the drug may determine its function. Future studies are needed to explain the detailed mechanism of interaction between Nr4a and CPT.

The COX-2/PGE2 pathway is involved in multiple aspects of tumor pathology; they directly promote tumor growth, regulate angiogenesis, and affect tumor-associated immune cells such as macrophages, myeloid-derived suppressor cells, and T cells (29). In this study, we demonstrated that tumor-derived PGE2 induced Nr4a factors and their target genes on Tregs in a cAMP/PKA-dependent manner, and their expression could be modulated by a common COX-2 inhibitor. Previous studies have shown that PGE2 and its downstream cAMP/PKA pathway play important roles in the maintenance of CD4⁺ T-cell biology, including Treg suppression activities (37–40). This is also supported by our data demonstrating that suppressive activities of Treg cells were enhanced by exposure to tumor-bearing conditions, but were attenuated by COX-2 inhibitor (Fig. 6B; Supplementary Fig. S7A). In addition, Nr4a is well identified as a target molecule of cAMP/CREB in the signaling pathways involved in various physiologic and pathologic phenomena, such as hepatic glucose metabolism or the stress response in neurons (41, 42). Collectively, we speculate that Nr4a factors function as a downstream regulator of PGE2–cAMP–PKA signaling, thereby contributing to the formation of an immunosuppressive tumor microenvironment by Tregs, and this regulatory axis can be targeted by COX-2 inhibitors.

The recent success of cancer immunotherapy including immune checkpoint blockade, such as anti–PD-1 and anti–CTLA-4 antibody therapies, triggered the investigation of drugs that activate antitumor immunity from already existing drugs (43, 44). Both CPT and COX-2 inhibitors were well defined as effective antitumor drugs directly acting on tumor cells (27, 45), but here we revealed that they also enhance antitumor immunity by inhibiting Nr4a/Treg function. A similar finding was recently reported for cyclin-dependent kinase 4 and 6 (CDK4/6) inhibitors (13). CDK4/6 inhibitors such as abemaciclib have been shown to be effective against several solid tumors, and their primary mechanism of action is thought to be inhibition of cell-cycle progression in tumor cells. Goel and colleagues (13) revealed that abemaciclib also promotes antitumor immunity by selectively suppressing the proliferation of Tregs and increasing the immunogenicity of tumor cells. Their study and our present study will accelerate the search for anticancer drugs that promote both tumor cell death and antitumor immunity.

Nr4a factors may play important roles in not only Tregs but also CTLs. Mognol and colleagues reported that Nr4a factors are involved in the establishment of cancer-induced exhaustion in tumor-infiltrating CD8⁺ CTLs (46). This study suggested a possibility that Nr4a inhibition may also facilitate the reactivation of exhausted CTLs like anti–PD-1 antibody, in addition to the inhibitory effects on Tregs. Another article showed that Nr4a1 is involved in the function and differentiation of a long-lived subset of tissue-resident memory CD8⁺ (T_RM) T cells (47). Tumor-infiltrating CD8⁺ T cells were reported to exhibit characteristics of T_RM cells (48). Thus, further study is necessary to elucidate the detailed roles of Nr4a factors in antitumor CTLs.

In conclusion, we revealed that Nr4a factors play crucial roles in various parts of Treg function in the tumor microenvironment, and these mechanisms could be applied to tumor therapy using clinically established compounds. Nr4a factors show great promise as effective therapeutic targets of cancer immunotherapy.

No potential conflicts of interest were disclosed.

Conception and design: S. Hibino, A. Yoshimura

Development of methodology: S. Hibino, S. Chikuma, T. Kondo, A. Yoshimura

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Hibino, T. Kondo, M. Ito, S. Omata-Mise, A. Yoshimura

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Hibino, H. Nakatsukasa, S. Omata-Mise, A. Yoshimura

Writing, review, and/or revision of the manuscript: S. Hibino, S. Chikuma, T. Kondo, A. Yoshimura

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Chikuma, A. Yoshimura

Study supervision: A. Yoshimura

We thank N. Shiino, C. Ohkura, Y. Tokifuji, and Y. Hirata for their technical assistance. This work was supported by JSPS KAKENHI (S) 17H06175 (to A. Yoshimura), Advanced Research & Development Programs for Medical Innovation (AMED-CREST) JP17gm0510019 (to A. Yoshimura), the Takeda Science Foundation (to A. Yoshimura), the Uehara Memorial Foundation (to A. Yoshimura), the SENSHIN Medical Research Foundation (to A. Yoshimura), and Grant-in-Aid for Scientific Research on Innovative Areas17H05801 (to S. Chikuma).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.