The effect of Curcumin on Multi-Level Immune Checkpoint Blockade and T Cell Dysfunction in Head and Neck Cancer

Despite recent advances in understanding the complex immunologic dysfunction in the tumor microenvironment (TME), fewer than 20% of patients with head and neck squamous cell carcinoma (HNSCC) respond to immune checkpoint blockade (ICB). Thus, it is important to understand how inhibitory IC receptors maintain the suppressed dysfunctional TME, and to develop more effective combination immunotherapy. This study evaluated the immune-modulating effects of Curcumin, which has well-established anti-cancer and chemopreventive properties, and its long-term safety as a phytochemical drug.


Immunohistochemistry
The 4-μm para n-embedded tissue samples were soaked rst in xylene to remove the para n wax and then sequentially in solutions of 100%, 90%, 80%, and 70% ethanol for rehydration, then heated in 1x sodium citrate buffer, PH 6.0 for antigen retrieval. For single immunostaining, endogenous peroxidase activity was blocked in a 1% hydrogen peroxide solution (Sigma-Aldrich, St. Louis, MO, USA) in PBS with 0.3% Triton X-100 for 30 min at room temperature. Nonspeci c binding (2% BSA) was blocked. The sections were incubated with the indicated antibodies at 4°C in a humidi ed box and then incubated with the corresponding horseradish peroxidase-conjugated secondary antibody. Finally, 3,3′ diaminobenzidine (DAB; DAKO) was used to detect these labeled antibodies. The nucleus was stained with hematoxylin.
After rinsing with PBS, the samples were mounted using Permount TM Mounting Medium (Fisher Chemical, Fair Lawn, NJ, USA).

Xenograft tumor models
Six-week-old male nude mice were obtained from Orient Bio (Seongnam, South Korea). Mice were used in accordance with the guidelines of the Institutional Animal Care and Use Committee of Chungnam National University, which approved of the animal research (Daejeon, South Korea). Mice were injected subcutaneously with SCC15 cells (1 × 10 7 in 100μl PBS). After tumors reached 40mm 3 (day 0), Curcumin (0, 50 mg/kg) was administered every other day through intraperitoneal injections. After 22 days, the mice were sacri ced, and the tumor was harvest. Tumor volumes were calculated according to the following formula: tumor volume (mm 3 ) = (length) × (width) 2 × 0.5, and the tumor weights were recorded.

4-NQO-induced oral tumorigenesis model
The carcinogen 4-nitroquinoline 1-oxide (4-NQO) (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in DMSO at 50 mg/ml to create a stock solution, which was stored at -20°C and diluted to a nal concentration of 50 µg/ml. To avoid decomposition of 4-NQO, light was avoided.. For the malignant transformation of the oral mucosa model, a total of 44 female C57BL/6 mice (Narabiotech, Korea), sixweek-old, and weighing 18 to 20 g were used for the studies with 4-NQO. For the drinking water method, the 4-NQO stock solution was diluted in the drinking water for mice. Fresh 4-NQO water was supplied every week. After 16 weeks, the drinking water was switched to distilled water, and then the mice were beginning to treat with corn oil or Curcumin 50 mg/kg by oral gavage for consecutive 6 weeks. The mice were analyzed for oral lesions and weighed at different times for up to 22 weeks. All the experiments were conducted in accordance with the approval of the Institutional Animal Care and Use Committee (IACUC) at Chungnam National University (Daejeon, South Korea).

Results
Curcumin reduces the expression of IC proteins (PD-L1, PD-L2, and Galectin-9) which promote cell viability and cell invasion in HNSCC cell lines To investigate the involvement of IC proteins on the anti-cancer effect of Curcumin, we rst examined the expression of IC proteins in HNSCC cell lines. Cell lysates were prepared from human broblast-cells (hFB) and four HNSCC cell lines (FaDu, SNU1041, SNU1076, and SCC15). The expression of IC proteins PD-L1, PD-L2, and Galectin-9 was upregulated in HNSCC cell lines compared with the human broblasts (Fig. S1A). We initially sought to evaluate the effect of Curcumin on cell viability on HNSCC cell lines. SNU1041 and SCC15 cells were treated with increasing doses of Curcumin for 24h and cell viability was analyzed by WST-1 assay. Data analysis revealed that Curcumin inhibited the cell viability of SNU1041 and SCC15 cell line in a dose-dependent manner ( Supplementary Fig. S1B, C). Next, to determine the in uence of Curcumin on IC proteins, SNU1041 and SCC15 cells were treated with different concentrations of Curcumin for 24h. The treatment of HNSCC cells with Curcumin resulted in dose and time-dependent decreases in IC protein expression in both cell lines ( Fig. 1A-D). The nding that the expression of IC proteins was higher in HNSCC cell lines than in normal cell lines led us to hypothesize that these IC proteins may drive cell growth. To test this hypothesis, we performed a proliferation assay after downregulation of IC proteins in SNU1041 and SCC15 cell lines. PD-L1, PD-L2, and Galectin-9 knockdown cells showed markedly decreased cell viability in both cell lines, indicating the contribution of IC proteins to HNSCC cancer cell viability ( Fig. 1E-G). Cell invasion is crucial step in tumor metastasis. To investigate whether IC proteins affected the malignant behavior of HNSCC, we did the transwell assay. Signi cant inhibition of invasion was observed after knockdown of IC proteins by siRNA compared with the control group ( Fig. 1H-J). These results clearly show that IC proteins PD-L1, PD-L2, and Galectin-9 positively promote the invasion of HNSCC cells. The epithelial-mesenchymal transition (EMT), which refers to changes in cell phenotype from epithelial to mesenchymal morphology, is an essential process during the initiation and progression of tumorigenesis and metastasis [26]. To investigate whether IC proteins regulate the EMT, we examined EMT-related proteins (E-cadherin, N-cadherin, Vimentin, and Slug) by western blotting. Knockdown of IC proteins by siRNA signi cantly suppressed the EMT, as evidenced by upregulating the expression of epithelial marker (E-cadherin) and downregulating the expression of mesenchymal marker (N-cadherin, Vimentin), transcription factor (Slug) (Fig. 1K-M). Together, these data imply that IC proteins PD-L1, PD-L2, and Galectin-9 regulate cancer cell metastasis by affecting the actions of EMT-related genes.
Combination of Curcumin and PD-L1 Ab potentiates the cytotoxic effect of CD8 + T-cells and has an additive effect on IFN-γ and Granzyme B secretion of T cell Curcumin treatment enhances the ability of effector T cells to kill cancer cells [23]. To determine whether Curcumin can enhance the impact of CD8 + T cells on immune-mediated cytotoxicity, CD8 + T cells were isolated from tonsil tissue. CD8 + T-cells were activated by CD3, CD28 and IL-2 ( Fig. 2A) and incubated with SNU1041 cells pretreated with Curcumin for 24 h. Activated CD8 + T-cell and SNU1041 cells were cocultured for another 3 days. Although activated CD8 + T cells could inhibit tumor cell growth in the absence of Curcumin, Curcumin-treated groups showed signi cant inhibitory effects on tumor growth. Simultaneously, the combination of Curcumin with PD-L1 Ab showed additional treatment e cacy compared with control, Curcumin treatment, or PD-L1 Ab group (Fig. 2B). These results showed that the combination of Curcumin with PD-L1 Ab potentiates the effect of CD8 + T-cells-mediated cancer cell killing. Moreover, combined PD-L1 blockade and Curcumin treatment resulted in increase the secretion of IFN-γ, Granzyme B compare with the other three groups (Fig. 2C). To avoid confusion between Curcumin's anti-tumor growth effect, PD-1/PD-L1 mediated tumor immune escape, and CD8 + T cell killing effect, we calculated the difference between these groups. These results show that the combination of Curcumin with PD-L1 Ab potentiates the effect of CD8 + T cells-mediated cancer cell killing and increased the secretion of IFN-γ and Granzyme B.
Administration of 4-NQO induces an aggressive oral squamous carcinoma and overexpression of PD-1 and TIM-3 in peripheral lymphatic tissue in vivo To evaluate the effect of the Curcumin on the anti-tumor immunity response in vivo, we established the 4-NQO-induced carcinogenesis model in immunocompetent C57BL/6 mice. The 4-NQO oral cancer model is well-established and mimics the pathology of human oral cancer [27][28][29]. In this study, to induce tumorigenesis in the mouse oral cavity, C57BL/6 mice were given 4-NQO (50 μg/mL) in drinking water for 16 consecutive weeks (or water in the absence of 4-NQO as a control) and then regular water until week 22 ( To investigate the impact of Curcumin on immunocompetent mice, C57BL/6 mice were exposed to 4-NQO for 16 weeks and then treated with Curcumin 50 mg/kg by oral gavage for consecutive 6 weeks (Fig.  4A). Mice treated with Curcumin had increased weight and overall survival compared to the control group ( Fig. 4B, C). Representative images of the oral lesions in the control and Curcumin-treated groups showed that Curcumin treatment led to a noticeable reduction in tumor size and the number of 4-NQO-induced lesions (Fig. 4D), accompanied by a less malignant and invasive phenotype in the lesions compared with the control group. The tissue sections were stained with H&E for detailed histopathological analysis.
Curcumin-treated mice had less-invasive oral carcinoma lesions, but more low-grade dysplasia in their tongues than the control group (Fig. 4E). At the same time, Curcumin treatment reduced both the number and size of lesions induced by 4-NQO (Fig. 4F, G). Our ndings demonstrated that Curcumin had growth inhibitory effects and led to less tumor formation and a transformative phenotype in the 4-NQO-induced model. Moreover, we established an HNSCC xenograft mouse model using SCC15 cell lines. The tumor volumes were signi cantly decreased after Curcumin treatment ( Supplementary Fig. S2A). The expression of IC proteins PD-L1, PD-L2, and Galectin-9 in tumor tissue from the Curcumin group was signi cantly downregulated compared with the control group ( Supplementary Fig. S2B). We also con rmed the downregulated expression of IC proteins in the Curcumin group by western blotting ( Supplementary Fig. S2C). Our ndings collectively reveal that Curcumin not only regulates tumor sensitivity to immune cell-mediated tumor killing through the suppression of PD-L1 expression but also inhibits the expression of PD-L2 and Galectin-9 in HNSCC cells.
Curcumin restores effector T cells by modulating the expression of PD-1 and TIM-3 on CD4 + or CD8 + T cells or CD4 + CD25 + FoxP3 + Tregs in 4-NQO oral carcinogenesis model To further explore the effects of Curcumin on the T lymphocyte subpopulation in the immune system, we collected cells from the spleen and blood of 4-NQO-induced tumor-bearing mice. Curcumin treatment increased the proportion of CD4 + and CD8 + T cells in the spleen and blood, which increased the immune response to tumors (Fig. 5A, B). PD-1 and TIM-3 on the surface of immune cells are IC molecules that mediate the immune escape of tumor cells. We investigated the expression of PD-1 and TIM-3 on CD4 + T cells and CD8 + T cells by ow cytometry. Curcumin treatment led to a substantial reduction in PD-1 + CD4 + T cells in both the blood and spleen and a decreased percentage of TIM-3 + CD4 + T cells in the blood (Fig.   5C, D). In addition, Curcumin-treated mice also had a signi cantly decreased percentage of PD-1 + CD8 + T and TIM-3 + CD8 + T cells in both blood and the spleen (Fig. 5E, F). These ndings imply that Curcumin can restore effector T cells by modulating the expression of PD-1 and TIM-3 on CD4 + or CD8 + T-cells. Next, we determined the role of Curcumin in the 4-NQO-induced anti-tumor immune response through the regulation of Tregs, which play roles in suppressing the immune response. We examined the population of Tregs in each group by ow cytometry. The population of CD25 + FoxP3 + in CD4 + T cells was signi cantly suppressed in Curcumin-treated mice compared with control mice (Fig. 5G). These results imply that Curcumin can effectively inhibit tumor growth by downregulating Tregs in mice. FoxP3 + Tregs co-expressing PD-1 and TIM-3 are highly immunosuppressive, including a specialized subset of tissue Tregs in breast cancer models [31]. We investigated whether the expression of PD-1 and TIM-3 on Tregs was also reduced by Curcumin treatment. The percentage of PD-1 + and TIM-3 + on Tregs was signi cantly decreased in the Curcumin treatment group (Fig. 5H, I). These data indicate that Curcumin reduces not only Tregs itself but also decreases the expression of IC receptors PD-1 and TIM-3 on Tregs, implying that Curcumin may be an alternative treatment for Treg-mediated immunosuppression in HNSCC.
Curcumin suppresses the expression of IC proteins and promotes the expression of IFN-γ and Granzyme B in 4-NQO oral carcinogenesis model We assessed the expression of IC proteins PD-L1, PD-L2, and Galectin-9 in 4-NQO-induced tongue cancer tissues by western blotting and immunohistochemistry analysis. The expression of PD-L1, PD-L2, and Galectin-9 in the Curcumin group was downregulated compared with the control group in tumor tissues (Fig. 6A, B). Next, we examined the effects of Curcumin treatment on locoregional immunity in vivo. The confocal assay showed that the Curcumin group led to a decreased percentage of CD8 + PD-1 + and CD8 + TIM-3 + in TILs compared with the control group (Fig. 6C). We also analyzed the expression of IFN-γ, Granzyme B, which are produced by cytotoxic T cells. We observed a signi cant increase in the expression of IFN-γ, Granzyme B in Curcumin-treated mice compared with the control group (Fig. 6D). These data show that Curcumin therapy not only effectively reduces the expression of IC molecules on cancer cells but also upregulates T-cell populations in peripheral tissues in vivo, indicating both locoregional and systemic immune activation in mice with carcinogen-induced early lesions.

Discussion
Recurrence is a very challenging issue in cancer treatment, and one of the reasons that cancer treatment is so di cult and can have a poor prognosis [32][33][34]. Recently, the Korean Society of Thyroid-Head and Neck Surgery (KSTHNS) developed the guideline about surgical treatment of HNSCC to improve the patients' survival [35]. The reasons for HNSCC recurrence are mainly due to its high propensity for intrinsic, spatial, and acquired resistance to chemotherapeutic agents, radiotherapy, and anti-epidermal growth factor receptor (EGFR) mAb [36,37]. Although ICB with PD-1/PD-L1 blockade is an alternative for overcoming recurrence and resistance that has shown great promise for the treatment of a variety of advanced cancers including HNSCC, a signi cant durable response is limited to a minority of patients, and eventually, most of these patients experience relapse of the disease [14]. While the mechanism underlying resistance to ICB has not been fully elucidated, accumulating evidence implies that immunosuppressive signaling receptors such as TIM-3, which can impair cytotoxic T-cell functionality, play an important role in acquired resistance [31,38]. Therefore, combination strategies, not only ICB with conventional standard therapies such as cytotoxic or targeted agents and radiotherapy but also multiple IC inhibitors and their ligands in the TME, have been tested in various cancer subtypes [39]. Preliminarily, the combination has increased the response rate in a few clinical trials, but the immune-related adverse events commonly associated with checkpoint blockers, such as diarrhea, colitis, myocarditis and endocrine disorders, are more severe in these patients [40][41][42].
Curcumin is one of the best characterized chemopreventive agents. It has strong anti-oxidative, antiin ammatory and anti-septic properties and has been widely used for a long time in traditional medicine, implying that similar to many phytochemicals, it is safe in the human body and side effects are rare [22].
Recently, Curcumin has emerged as a potent anti-cancer agent that targets several biological pathways and processes in various cancers, including mutagenesis, cell cycle, oncogene expression, angiogenesis, metastasis, and cell death signaling such as apoptosis and autophagy without adverse effects on normal tissue [22]. Moreover, in combination with conventional anti-cancer therapies such as chemotherapy and radiation, Curcumin enhances e cacy by sensitizing cancer cells to their cytocidal effects and reduces treatment-associated side effects including cardio-, hepato-, nephro-and neuroprotective properties by balancing reactive oxygen species or in ammatory reactions [18][19][20][21][22][23]. In addition to its direct effects on cancer cells described above, emerging evidence has shed light on the immune-modulating effects of Curcumin that may play a role in its anti-tumor effects [23]. Although, the immune-modulatory actions of Curcumin have been shown in a wide range of in ammatory and autoimmune diseases such as arthritis, colitis and hepatitis, its immunomodulatory capacity in the TME has emerged relatively recently. Research has shown that Curcumin not only enhances tumor antigen-speci c T cells via reversal of tumor-induced immunosuppression but also enhances cytotoxic T cells by acting directly on immune cell dysfunction, which is one of the major mechanisms of tumor escape from immune surveillance via the signal transducer and activator of transcription 3 and nuclear factor kappa B (NF-kB) signaling pathways [23,43,44]. Lim et al. [45]showed inhibition of in ammation-mediated PD-L1 expression by Curcumin, and Liao et al. [46] showed decreased PD-L1 expression following immunosuppression in cell populations, such as Tregs and myeloid-derived suppressor cells, in a murine oral cancer model after Curcumin treatment. Hayakawa et al. [44] showed that Curcumin increases the induction of tumor antigen-speci c T cells by restoring T-cell stimulation, implying that the combination of PD-1/PD-L1 Ab is attractive for the development of effective ICB. Given the various biologic effects, we hypothesized that Curcumin, as well as PD-L1, may affect the simultaneous expression of other IC proteins on immune cells and its ligands on tumor cells. We hypothesized that Curcumin could be an alternative to overcoming disease progression or relapse due to resistance to immunotherapy. To this end, we evaluated the effects of Curcumin on the expression of multiple IC ligands on tumor cells.
We demonstrated that PD-L1, PD-L2, and Galectin-9 induced HNSCC cell invasion via EMT activation, indicating the intrinsic roles of these IC ligands in HNSCC independent of the interaction with immune cells. In addition, by showing that Curcumin simultaneously inhibits PD-L1, PD-L2, and Galectin-9, for the rst time, we show that Curcumin can simultaneously inhibit IC ligands other than PD-L1. Our data have clinical signi cance in that recent advances have revealed that bidirectional regulation may exist between EMT status and IC ligands, especially PD-L1 expression that ultimately leads to tumor immune escape.
Our data also indicate that PD-L1 signaling plays an important role in the maintenance of EMT status in HNSCC, in accordance with similar reports in solid tumors such as renal cell carcinoma, breast cancer, hepatocellular carcinoma, esophageal cancer, and glioblastoma [47,48]. Although far less investigated than PD-L1, PD-L2 reportedly contributes to T-cell exhaustion by interacting with the PD-1 receptor, implying functional relevance to the TME. A few clinical papers have shown that PD-L2 expression is independently associated with clinical response in anti-PD-1-treated patients, indicating that the effect of ICB may be related partly to blockade of PD-1/PD-L2 interactions. Therefore, targeting both PD-1 ligands may provide additional clinical bene t [49]. The effect of PD-L2 intrinsic signaling on the tumor itself has been much less investigated than that of PD-L1 [50]. However, recently Ren et al. [51] suggested that PD-L2 intrinsically promotes tumor invasion and metastasis via RhoA and autophagy pathways. Consistent with the data, we demonstrated that Curcumin can reduce intrinsic PD-L2 expression independent of immune cells, thereby inhibiting the EMT in HNSCC cells. Although PD-L2 expression in HNSCC tumor tissue has been shown [52], we report for the rst time the association of PD-L2 with the EMT in HNSCC.
Our data also delineate the role of PD-L2 as an immunosuppressive and cancer-promoting signaling molecule. In this study, we showed that suppression of Galectin-9 decreased the EMT in HNSCC cells.
However, because the role of Galectin-9 apart from tumor immune escape has not yet been studied in HNSCC [53], further validation and mechanistic studies about associated molecular pathways are needed.
Our ndings support previous ndings that Galectin-9 expression in solid tumors may be linked to tumor cell adhesion or metastasis in various cancers [54,55], and Galectin-9 is expressed signi cantly more in advanced cancer stages compared to early stages and expressed by tumor-in ltrating lymph nodes [55,56]. Moreover, mechanisms underlying the regulation of multiple IC ligands are need more detailed study. The nude mouse model was used to investigate the effect of Curcumin on the expression of ligand in HNSCC cells independently of immune cells. In this HNSCC xenograft model, we do not think that Curcumin can inhibit tumor progression by targeting these molecules alone, as its anti-cancer properties are through other multiple signaling pathways [22]. However, the in vivo model validates the IC ligandinhibiting effects of Curcumin evident in the in vitro results.
To further evaluate the effects of Curcumin on the TME with regard to ICB, we used the 4NQO-induced syngenic murine tongue squamous cell carcinoma model, which mimics the carcinogenesis of HNSCC.
As expected, the model showed a signi cantly immune-suppressive TME, consisting of increased expression of inhibitory IC proteins such as PD-1 and TIM-3 in lymphocytes. The increase in both CD25 + FoxP3 + Treg cells and PD-1 + and TIM-3 + Treg expression is associated with more suppressive activity [57]. Curcumin treatment not only reduces the size and number of lesions but also may reduce the risk of invasive cancer and increase the proportion of precancerous lesions, implying that Curcumin functions as a chemopreventive agent that inhibits cancer initiation and may also reduce cancer progression[58].
Our study differs from previous studies in that it focused on the effects of inhibiting multiple IC proteins via not only receptors on T cells but also its ligands on tumor cells simultaneously, as a mechanism to restore CD8 + T-cell dysfunction. Given that exhausted T cells exhibit defective proliferative capacities and cytokine production, and inert lytic function [31], Curcumin treatment successfully reinvigorates T cells of the HNSCC TME in that it increases total CD4 + and CD8 + cells in the periphery implying T-cell proliferation; CD8 + TILs are a subset of T cells that directly target tumor cells, and are a good prognostic factor in all HNSCC [59]. Also, Curcumin treatment increases the secretion of cytokines such as IFN-y and Granzyme B, re ecting the effector functions of cytotoxic T cells. Sakuishi et al.[38] reported that TIM-3 + PD-1 + TILs exhibit the most severe exhausted phenotype as de ned by the failure to proliferate and produce IL-2, TNFa, and interferon gamma (IFN-γ); thus, combined targeting of the TIM-3 and PD-1 pathways is more effective in controlling tumor growth than targeting either pathway alone. Some studies have reported inhibition of PD-L1 expression by Curcumin as described above, but this is the rst study of the effect of Curcumin on the Galectin-9/TIM-3 axis in solid tumors.
Moreover, we demonstrated that Curcumin-induced restoration of T-cell function may be due to the effect of Curcumin inhibiting the PD-1 and TIM-3 axis in CD4 + and CD8 + T-cells not only at the tumor site (TILs) but also in the periphery (blood and spleen). These data highlight the clinical usefulness of Curcumin in upregulating TIM-3, which is a well-known regulator of CD8 + T-cell exhaustion [38] and has been shown to mediate adaptive resistance to anti-PD-1 in models of non-small cell lung cancer and HNSCC[60]. Jie et al. [61]showed that the frequency of PD-1 + and TIM-3 + cells was signi cantly increased on CD8 + TILs after cetuximab treatment, which is an mAb to EGFR and the most well-known molecular target agent in HNSCC. The PD-1 and TIM-3 axis is associated with resistance to chemotherapy. The combination of cetuximab with PD-1 or TIM-3 should be considered to improve clinical outcomes for HNSCC patients. In line with these data, clinical trials are currently underway to elucidate the role of combined blockade of TIM-3 and PD-1/PD-L1 in various advanced and/or metastatic solid tumors (Registration Nos. We do not consider Curcumin's inhibitory effects or binding a nity for IC proteins to be superior to those of speci cally engineered mAbs. However, we showed that Curcumin can simultaneously inhibit both PD-1 and TIM-3 expression, which is important for the reinvigoration of exhausted T cells and an important adaptive escape mechanism from anti-PD-1 inhibition, respectively. Another advantage of Curcumin shown in our study is that Curcumin inhibits not only IC proteins on T-cell subsets but also its ligands on tumor cells, whereas existing ICB acts on either the ligand (of a tumor cell or antigen-presenting cell) or IC protein of immune cells. Although additional large-scale randomized clinical trials are needed, our study provides a rationale for combining Curcumin with conventional standard therapeutic modalities including approved IC inhibitors such as nivolumab, pembrolizumab, and durvalumab. This strategy may provide multi-faceted, sustained anti-cancer effects in patients with limited responses to ICB, and even those who fail to respond to ICB and molecular targeted therapy.

Conclusion
In conclusion, we showed that Curcumin restores T-cell dysfunction via multiple and multi-level IC suppression. Given the adverse reactions caused by ICB blockade by speci c mAbs, the well-established safety of Curcumin as a phytochemical and its anti-cancer effects underscore the potential bene ts of

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryFigure.pptx