TPH1 and 5-HT7 Receptor Overexpression Leading to Gemcitabine-Resistance Requires Non-Canonical Permissive Action of EZH2 in Pancreatic Ductal Adenocarcinoma

Simple Summary Most patients with pancreatic cancer initially respond to the first-choice drug gemcitabine, but the cancer cells rapidly acquire drug resistance, resulting in poor survival. In this study, we investigated whether the serotonin (5-hydroxytryptamine, 5-HT) system plays an important role in gemcitabine resistance and the maintenance of pancreatic cancer stem cells (CSCs) in association with an Enhancer of zeste homolog 2 (EZH2), an epigenetic regulator of transcription. Herein, we demonstrate that long-term exposure of PDAC cells to 5-HT leads to enhanced EZH2 expression which, in turn, allows upregulation of TPH1 and 5-HT7, resulting in EZH2-TPH1-5-HT7 axis operating in a feed-forward manner. The results suggest that the EZH2-TPH1-5-HT7 axis may be a highly efficient therapeutic target against drug-resistant pancreatic ductal adenocarcinoma (PDAC). Abstract In the present study, we investigated the regulatory mechanisms underlying overexpression of EZH2, tryptophan hydroxylase 1 (TPH1), and 5-HT7, in relation to gemcitabine resistance and CSC survival in PDAC cells. In aggressive PANC-1 and MIA PaCa-2 cells, knock-down (KD) of EZH2, TPH1, or HTR7 induced a decrease in CSCs and recovery from gemcitabine resistance, while preconditioning of less aggressive Capan-1 cells with 5-HT induced gemcitabine resistance with increased expression of EZH2, TPH1, and 5-HT7. Such effects of the gene KD and 5-HT treatment were mediated through PI3K/Akt and JAK2/STAT3 signaling pathways. EZH2 KD or GSK-126 (an EZH2 inhibitor) inhibited activities of these signaling pathways which altered nuclear level of NF-kB, Sp1, and p-STAT3, accompanied by downregulation of TPH1 and 5-HT7. Co-immunoprecipation with EZH2 and pan-methyl lysine antibodies revealed that auto-methylated EZH2 served as a scaffold for binding with methylated NF-kB and Sp1 as well as unmethylated p-STAT3. Furthermore, the inhibitor of EZH2, TPH1, or 5-HT7 effectively regressed pancreatic tumor growth in a xenografted mouse tumor model. Overall, the results revealed that long-term exposure to 5-HT upregulated EZH2, and the noncanonical action of EZH2 allowed the expression of TPH1-5-HT7 axis leading to gemcitabine resistance and CSC population in PDAC.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) accounts for over 90% of pancreatic cancer cases worldwide and is one of the most aggressive human malignancies. It is the only cancer type in which the five-year survival rate (8% in metastatic PDAC) has not improved over the last several decades despite advances in surgical and other cancer treatments [1]. Gemcitabine is the first line chemotherapeutic drug for PDAC. However, the rate of response to gemcitabine is poor, and PDAC rapidly develop resistance [2,3].
Genomic studies of PDAC have revealed four frequently mutated genes, KRAS, TP53, CDKN2A, and SMAD4. The activating mutation in KRAS is present in 80-90% of pancreatic

Sphere Formation Assay
One thousand cells were seeded on an ultra-low adhering 24-well plate (Corning Incorporated Costar, Corning, NY, USA) in prEGM media (Lonza, Basel, Switzerland) and allowed to form spheres. After four days, the spheres were treated with vehicle or chemical inhibitors (1 and 3 µM). After 11 days of drug treatment, images of spheres were captured using an inverted microscope (IX73, Olympus, Tokyo, Japan). The number of spheroids over 50 µm in diameter was counted by using Image J 1.48v software (National Institute of Health, Bethesda, MD, USA).

Anti-Tumor Activity Measurement Using a Xenograft Tumor Model
Female BALB/c nude mice (OrientBio, Gyeonggi, South Korea) were subcutaneously inoculated with 1 × 10 7 PANC-1 cells/Matrigel (1:1) at the right flank. After tumor volume reached approximately 300 mm 3 , mice in the first set of experiments were administered intraperitoneally (i.p.) with drugs (gemcitabine (50 mg/kg), telotristat (1 or 10 mg/kg), SB-269970 (1 or 10 mg/kg), or GSK-126 (10 mg/kg)) once a day for six days a week (n = 5). In a different set of experiments, a mouse tumor model was made with the same method as before, except the number of mice in each group (n = 6). In this second set of experiments, mice were administered with gemcitabine (50 mg/kg), GSK-126 (10 mg/kg), or gemcitabine plus drugs (telotristat (10 mg/kg), SB-269970 (10 mg/kg), or GSK-126 (10 mg/kg)). Tumor volume was calculated using the equation (l × b 2 )/2, where l and b were the larger and smaller dimensions of each tumor. On the final day of treatment, tumors were excised from sacrificed mice by CO 2 gas inhalation, and tumor weight was measured.
The mouse experiments were performed following the institutional guidelines of the Institute of Laboratory Animal Resources and approved by the Institutional Animal Care and Use Committee of Yeungnam University.

Statistical Analyses
Data from more than three independent experiments were averaged and expressed as mean ± SEM. Statistical significance was determined with the one-way analysis of variance Cancers 2021, 13, 5305 5 of 16 (ANOVA), followed by the Newman-Keul's comparison method, using Graph Pad Prism 5.0 (San Diego, CA, USA). p-values less than 0.05 were considered statistically significant.

EZH2 Supports TPH1-5-HT 7 Axis to Regulate Gemcitabine Resistance and Cancer Stem Cell Population in Pancreatic Cancer Cells
To investigate the relationship between intrinsic gemcitabine resistance and the expression of EZH2 and 5-HT system genes in pancreatic cancer cells, we first compared the levels of resistance and those gene expressions between PDAC cell lines. H6c7 was used as a negative control cell line for gemcitabine resistance (GemR). PANC-1 cells exhibited the strongest gemcitabine resistance, followed by MIA PaCa-2, Capan-1, and Capan-2 cells ( Figure 1A). Corresponding to GemR levels, the level of EZH2, TPH1 and 5-HT 7 receptor expressions was the highest in PANC-1, followed by MIA PaCa-2, Capan-1, and Capan-2 ( Figure 1B and Figure S1A), whereas EZH1, 5-HT 1A , and 5-HT 1B levels did not correlate with GemR levels, and they were higher in Capan-1 and Capan-2 cells than PANC-1 and MIA PaCa-2 cells ( Figure 1B). Treatment of PANC-1 and MIA PaCa-2 cells with GSK-126, an EZH2 inhibitor ( Figure 1C and Figure S1B), or EZH2 knock-down (KD) with siRNA transfection ( Figure 1D and Figure S1C) down-regulated the expression of TPH1 and 5-HT 7 . In addition, KD of TPH1 or 5-HT 7 induced suppression of each other's expression without changes in EZH2 mRNA ( Figure S1D) and protein ( Figure 1E) levels. Down-regulation of these genes with siRNA induced recovery from gemcitabine resistance in PANC-1 and MIA PaCa-2 cells ( Figure 1F). On the other hand, prolonged exposure of H6c7 and Capan-1 cells to 5-HT (10 µM, 96 h) induced gemcitabine resistance ( Figure 1G), accompanied by up-regulation of EZH2, TPH1, and 5-HT 7 expressions ( Figure 1H). The results indicate that EZH2 acted as an upstream regulator of the TPH1-5-HT 7 axis to maintain gemcitabine resistance of PDAC cells. In addition, prolonged exposure of Capan-1 cells to 5-HT significantly enhanced sphere formation in Capan-1 cells ( Figure 1I).
We then examined whether the EZH2-TPH1-5-HT 7 axis was linked to PI3K/Akt and JAK2/STAT3 signaling pathways, which are associated with drug resistance [6,7] and pancreatic CSC characteristics [8]. In PANC-1 and MIA PaCa-2 cells, the phosphorylation of PI3K/Akt and JAK2/STAT3 were suppressed by KD of EZH2-TPH1-5-HT 7 axis gene ( Figure 3A) and GSK-126 treatment ( Figure 3B). Cancers 2021, 13, x 7 of 16 We then examined whether the EZH2-TPH1-5-HT7 axis was linked to PI3K/Akt and JAK2/STAT3 signaling pathways, which are associated with drug resistance [6,7] and pancreatic CSC characteristics [8]. In PANC-1 and MIA PaCa-2 cells, the phosphorylation of PI3K/Akt and JAK2/STAT3 were suppressed by KD of EZH2-TPH1-5-HT7 axis gene (  We further investigated the relative contribution of PI3K/Akt and JAK2/STAT3 pathways in the maintenance of the CSC population by measuring sphere-forming ability of the cells in the presence of the signaling molecule inhibitors. At fixed concentrations (1 and 3 μM) which were selected based on cell viability response to the inhibitors ( Figure  S2A), gemcitabine did not inhibit sphere formation of PANC-1 and MIA PaCa-2 cells, whereas inhibitors of PI3K/Akt, JAK2/STAT3, and EZH2/TPH1/5-HT7 axis significantly suppressed the sphere formation of PANC-1 and MIA PaCa-2 in a concentration-dependent manner ( Figure 4A,B). In both PANC-1 and MIA PaCa-2 spheres, the expressions of CD44, a CSC surface marker, and Nanog, a stemness-associated transcription factor (TF), were most significantly suppressed by GSK-126, followed by telotristat (TPH1 inhibitor), fedratinib (JAK2 inhibitor), stattic (STAT3 inhibitor), wortmannin (PI3K inhibitor), and  We then examined whether the EZH2-TPH1-5-HT7 axis was linked to PI3K/Akt and JAK2/STAT3 signaling pathways, which are associated with drug resistance [6,7] and pancreatic CSC characteristics [8]. In PANC-1 and MIA PaCa-2 cells, the phosphorylation of PI3K/Akt and JAK2/STAT3 were suppressed by KD of EZH2-TPH1-5-HT7 axis gene (  We further investigated the relative contribution of PI3K/Akt and JAK2/STAT3 pathways in the maintenance of the CSC population by measuring sphere-forming ability of the cells in the presence of the signaling molecule inhibitors. At fixed concentrations (1 and 3 μM) which were selected based on cell viability response to the inhibitors ( Figure  S2A), gemcitabine did not inhibit sphere formation of PANC-1 and MIA PaCa-2 cells, whereas inhibitors of PI3K/Akt, JAK2/STAT3, and EZH2/TPH1/5-HT7 axis significantly suppressed the sphere formation of PANC-1 and MIA PaCa-2 in a concentration-dependent manner ( Figure 4A,B). In both PANC-1 and MIA PaCa-2 spheres, the expressions of CD44, a CSC surface marker, and Nanog, a stemness-associated transcription factor (TF), were most significantly suppressed by GSK-126, followed by telotristat (TPH1 inhibitor), fedratinib (JAK2 inhibitor), stattic (STAT3 inhibitor), wortmannin (PI3K inhibitor), and We further investigated the relative contribution of PI3K/Akt and JAK2/STAT3 pathways in the maintenance of the CSC population by measuring sphere-forming ability of the cells in the presence of the signaling molecule inhibitors. At fixed concentrations (1 and 3 µM) which were selected based on cell viability response to the inhibitors ( Figure S2A), gemcitabine did not inhibit sphere formation of PANC-1 and MIA PaCa-2 cells, whereas inhibitors of PI3K/Akt, JAK2/STAT3, and EZH2/TPH1/5-HT 7 axis significantly suppressed the sphere formation of PANC-1 and MIA PaCa-2 in a concentration-dependent manner ( Figure 4A,B). In both PANC-1 and MIA PaCa-2 spheres, the expressions of CD44, a CSC surface marker, and Nanog, a stemness-associated transcription factor (TF), were most significantly suppressed by GSK-126, followed by telotristat (TPH1 inhibitor), fedratinib (JAK2 inhibitor), stattic (STAT3 inhibitor), wortmannin (PI3K inhibitor), and SC-66 (Akt inhibitor) ( Figure 4C and Figure S2B). Interestingly, EZH2 expression in the spheres was down-regulated by the inhibitors except telotristat and SB-269970, whereas TPH1 and 5-HT 7 expressions in the spheres were suppressed by all the inhibitors in both PANC-1 and MIA PaCa-2 ( Figure 4C and Figure S2B). The results indicate that although PI3K/Akt and JAK2/STAT3 signaling pathways were commonly involved in the expression of EZH2, Cancers 2021, 13, 5305 8 of 16 TPH1, and 5-HT 7 , the regulatory mechanism for the expression and action of EZH2 was different from that for the TPH1-5-HT 7 axis in PDAC cells. To reveal this difference, we examined whether the gene transcriptional repression action of EZH2 is linked to these signaling pathways. We found that EZH2 KD increased expression of DAB2IP and PTEN ( Figure 4D), which are known to inhibit Ras and PI3K, respectively [34][35][36][37].
SC-66 (Akt inhibitor) ( Figure 4C and Figure S2B). Interestingly, EZH2 expression in the spheres was down-regulated by the inhibitors except telotristat and SB-269970, whereas TPH1 and 5-HT7 expressions in the spheres were suppressed by all the inhibitors in both PANC-1 and MIA PaCa-2 ( Figure 4C and Figure S2B). The results indicate that although PI3K/Akt and JAK2/STAT3 signaling pathways were commonly involved in the expression of EZH2, TPH1, and 5-HT7, the regulatory mechanism for the expression and action of EZH2 was different from that for the TPH1-5-HT7 axis in PDAC cells. To reveal this difference, we examined whether the gene transcriptional repression action of EZH2 is linked to these signaling pathways. We found that EZH2 KD increased expression of DAB2IP and PTEN ( Figure 4D), which are known to inhibit Ras and PI3K, respectively [34][35][36][37].

EZH2-Regulated Signaling Pathways Potentiate Nuclear Translocation of TFs Linked to TPH1-5-HT 7 Axis in Pancreatic Cancer Cells
We also examined which TFs were responsible for the upregulation of EZH2, TPH1, and 5-HT 7 expressions. In PANC-1 and MIA PaCa-2 cells, EZH2, TPH1, and 5-HT 7 expressions were down-regulated by treatment with mithramycin A (Sp1 inhibitor), PDTC (NF-κB inhibitor), and stattic (STAT3 inhibitor), but not by SR11302 (AP-1 inhibitor) and KG-501 (CREB inhibitor) ( Figure 5A), suggesting Sp1, NF-κB, and STAT3 were involved in these gene expressions. In an inverse proportion to the extent of EZH2 decrease by the TF inhibitors, DAB2IP and PTEN protein levels were increased by treatment with the TF inhibitors ( Figure 5A). The nuclear levels of Sp1, NF-κB, and p-STAT3 were inhibited by gallein (Gβγ inhibitor) ( Figure 5B). However, nuclear Sp1 and NF-κB levels were more suppressed by PI3K/Akt inhibitors than by JAK2/STAT3 inhibitors, while nuclear p-STAT3 level was reduced in the opposite direction by those inhibitors ( Figure 5B). Moreover, telotristat and SB-269970 reduced the nuclear p-STAT3 level, but not the NF-κB or Sp1 level ( Figure 5B). In 5-HT-pretreated Capan-1 cells, similar regulatory actions of the inhibitors of signaling molecule ( Figure 5C) and TFs ( Figure 5D) in the expression of EZH2, TPH1 and 5-HT 7 were observed. these gene expressions. In an inverse proportion to the extent of EZH2 decrease by the TF inhibitors, DAB2IP and PTEN protein levels were increased by treatment with the TF inhibitors ( Figure 5A). The nuclear levels of Sp1, NF-κB, and p-STAT3 were inhibited by gallein (Gβγ inhibitor) ( Figure 5B). However, nuclear Sp1 and NF-κB levels were more suppressed by PI3K/Akt inhibitors than by JAK2/STAT3 inhibitors, while nuclear p-STAT3 level was reduced in the opposite direction by those inhibitors ( Figure 5B). Moreover, telotristat and SB-269970 reduced the nuclear p-STAT3 level, but not the NF-κB or Sp1 level ( Figure 5B). In 5-HT-pretreated Capan-1 cells, similar regulatory actions of the inhibitors of signaling molecule ( Figure 5C) and TFs ( Figure 5D) in the expression of EZH2, TPH1 and 5-HT7 were observed.

Automethylated EZH2 Serves as a Binding Scaffold for Methylated NF-κB and Sp1, and Unmethylated p-STAT3, in a PRC2-Independent Manner
To further identify whether EZH2 acted as a scaffold for NF-κB, STAT3, and Sp1 to up-regulate EZH2-TPH1-5-HT7 gene expressions, we performed co-IP experiment with anti-EZH2 antibody. Co-precipitation of EZH2 with nuclear NF-κB was found in PANC-1 ( Figure 6A), MIA PaCa-2, and Capan-1 cells (Figure 6B), and the binding was further enhanced by 5-HT treatment ( Figure 6B). Similarly, binding of EZH2 with Sp1 and STAT3 was observed, and such binding was further increased by 5-HT treatment ( Figure 6B). However, SUZ12, a component of PRC2, was not precipitated with EZH2 ( Figure 6B), indicating that such scaffold action of EZH2 was PRC2-independent. We also examined

Automethylated EZH2 Serves as a Binding Scaffold for Methylated NF-κB and Sp1, and Unmethylated p-STAT3, in a PRC2-Independent Manner
To further identify whether EZH2 acted as a scaffold for NF-κB, STAT3, and Sp1 to up-regulate EZH2-TPH1-5-HT 7 gene expressions, we performed co-IP experiment with anti-EZH2 antibody. Co-precipitation of EZH2 with nuclear NF-κB was found in PANC-1 ( Figure 6A), MIA PaCa-2, and Capan-1 cells (Figure 6B), and the binding was further enhanced by 5-HT treatment ( Figure 6B). Similarly, binding of EZH2 with Sp1 and STAT3 was observed, and such binding was further increased by 5-HT treatment ( Figure 6B). However, SUZ12, a component of PRC2, was not precipitated with EZH2 ( Figure 6B), indicating that such scaffold action of EZH2 was PRC2-independent. We also examined whether transactivation ability of EZH2 was associated with auto-methylation activity of EZH2, which induces self-activation and methylation of other proteins [38]. In the total protein co-immunoprecipitates with anti-pan methyl lysine antibody, EZH2 and NF-κB were highly methylated, and Sp1 methylation was relatively low level, whereas signaling molecules, PI3K, Akt, JAK2, and STAT3 were not methylated in both PANC-1 and MIA PaCa-2 cells ( Figure 6C). In the nuclear protein precipitates with anti-pan methyl lysine antibody, it was confirmed that nuclear STAT3 and p-STAT3 were not methylated ( Figure 6D). Moreover, EZH2 KD or GSK-126 treatment significantly suppressed the methylation of nuclear NF-κB and Sp1, in addition to blocking methylation of EZH2 itself ( Figure 6D).
Cancers 2021, 13, x 10 of 16 whether transactivation ability of EZH2 was associated with auto-methylation activity of EZH2, which induces self-activation and methylation of other proteins [38]. In the total protein co-immunoprecipitates with anti-pan methyl lysine antibody, EZH2 and NF-κB were highly methylated, and Sp1 methylation was relatively low level, whereas signaling molecules, PI3K, Akt, JAK2, and STAT3 were not methylated in both PANC-1 and MIA PaCa-2 cells ( Figure 6C). In the nuclear protein precipitates with anti-pan methyl lysine antibody, it was confirmed that nuclear STAT3 and p-STAT3 were not methylated ( Figure  6D). Moreover, EZH2 KD or GSK-126 treatment significantly suppressed the methylation of nuclear NF-κB and Sp1, in addition to blocking methylation of EZH2 itself ( Figure 6D).

Antitumor Effects of EZH2-TPH1-5-HT7 Axis Inhibition in PANC-1 Xenograft Tumor Model in Mice
Next, we confirmed that the EZH2-TPH1-5-HT7 axis is a useful therapeutic target against drug-resistant pancreatic cancer, using an in vivo tumor model in which PANC-1 cells were subcutaneously transplanted. Compared to the vehicle-treated control group, treatment with gemcitabine (50 mg/kg) induced a slight decrease in PANC-1 tumor size, whereas telotristat (1 or 10 mg/kg) and SB-269970 (1 or 10 mg/kg) significantly reduced tumor growth in a dose-dependent manner ( Figure 7A,B). In addition, telotristat and SB-

Antitumor Effects of EZH2-TPH1-5-HT 7 Axis Inhibition in PANC-1 Xenograft Tumor Model in Mice
Next, we confirmed that the EZH2-TPH1-5-HT 7 axis is a useful therapeutic target against drug-resistant pancreatic cancer, using an in vivo tumor model in which PANC-1 cells were subcutaneously transplanted. Compared to the vehicle-treated control group, treatment with gemcitabine (50 mg/kg) induced a slight decrease in PANC-1 tumor size, whereas telotristat (1 or 10 mg/kg) and SB-269970 (1 or 10 mg/kg) significantly reduced tumor growth in a dose-dependent manner ( Figure 7A,B). In addition, telotristat and SB-269970 at a dose of 10 mg/kg started regressing tumor-growth on day 31 of drug treatment ( Figure 7A,C). Throughout the treatment period of 42 days, the body weight of drug (telotristat or SB-269970)-treated mice was not significantly different from that of control mice ( Figure 7D). In a separate set of experiments, treatment with GSK-126 (10 mg/kg) showed a similar response, and tumor size started to regress on day 31 of treatment ( Figure 7E,F). The tumor regression effect of GSK-126 plus gemcitabine was not significantly different from that of GSK-126 alone ( Figure 7E). Similarly, the effect of co-administration of telotristat or SB-269970 with gemcitabine was not different from that of combinated treatment of GSK-126 with gemcitabine ( Figure 7E). However, the tumor weight in the group co-administered with GSK-126 and gemcitabine was significantly lower than that of the GSK-126 only group. In addition, the effect of co-administration of other drugs and gemcitabine was not significantly different from that of combinated treatment of GSK-126 and gemcitabine ( Figure 7G). Body weight of mice co-treated with gemcitabine and the other drugs was not significantly different from that of control mice until treatment day 31. Thereafter, the body weight of GSK-126 alone mice or the combinated treatment group was significantly decreased compared to that of control group ( Figure 7H). 269970 at a dose of 10 mg/kg started regressing tumor-growth on day 31 of drug treatment ( Figure 7A,C). Throughout the treatment period of 42 days, the body weight of drug (telotristat or SB-269970)-treated mice was not significantly different from that of control mice ( Figure 7D). In a separate set of experiments, treatment with GSK-126 (10 mg/kg) showed a similar response, and tumor size started to regress on day 31 of treatment (Figure 7E,F). The tumor regression effect of GSK-126 plus gemcitabine was not significantly different from that of GSK-126 alone ( Figure 7E). Similarly, the effect of co-administration of telotristat or SB-269970 with gemcitabine was not different from that of combinated treatment of GSK-126 with gemcitabine ( Figure 7E). However, the tumor weight in the group co-administered with GSK-126 and gemcitabine was significantly lower than that of the GSK-126 only group. In addition, the effect of co-administration of other drugs and gemcitabine was not significantly different from that of combinated treatment of GSK-126 and gemcitabine ( Figure 7G). Body weight of mice co-treated with gemcitabine and the other drugs was not significantly different from that of control mice until treatment day 31. Thereafter, the body weight of GSK-126 alone mice or the combinated treatment group was significantly decreased compared to that of control group ( Figure 7H).

Discussion
It has been reported EZH2 and 5-HT derived from peripheral TPH1 in cancer tissues independently contributes to cancer malignancy by stimulating cancer cell proliferation and invasion [28,[39][40][41]. The present study demonstrated for the first time that EZH2 permits up-regulation of 5-HT system genes, TPH1 and 5-HT7, leading to drug resistance and CSC maintenance in PDAC. In addition, we also revealed that such noncanonical EZH2 action is mediated through the PI3K/Akt and JAK2/STAT3 pathways in a feed-forward manner in association with 5-HT7.
Previous studies on pancreatic cancer cell lines and human tumor tissue microarray have shown the role of 5-HT1B, 5-HT1D, or 5-HT2B in 5-HT-induced cancer cell proliferation

Discussion
It has been reported EZH2 and 5-HT derived from peripheral TPH1 in cancer tissues independently contributes to cancer malignancy by stimulating cancer cell proliferation and invasion [28,[39][40][41]. The present study demonstrated for the first time that EZH2 permits up-regulation of 5-HT system genes, TPH1 and 5-HT 7 , leading to drug resistance and CSC maintenance in PDAC. In addition, we also revealed that such noncanonical EZH2 action is mediated through the PI3K/Akt and JAK2/STAT3 pathways in a feed-forward manner in association with 5-HT 7 .
Previous studies on pancreatic cancer cell lines and human tumor tissue microarray have shown the role of 5-HT 1B , 5-HT 1D , or 5-HT 2B in 5-HT-induced cancer cell proliferation and metabolism [31,39,42]. However, the most recently discovered 5-HT receptor, 5-HT 7 , is increasingly reported as a new therapeutic target to inhibit cancer proliferation, migration, and invasion [28,[42][43][44]. The current study demonstrated an additional important role of 5-HT 7 in inducing drug resistance and CSCs in pancreatic cancer by elucidating the mechanism regulating 5-HT 7 expression: that is, inhibition of 5-HT 7 by its antagonist or removal of 5-HT via TPH1 inhibition suppressed 5-HT 7 expression in PANC-1 and MIA PaCa-2 cells. Similarly, TPH1 expression was also inhibited by a TPH1 inhibitor and 5-HT 7 antagonist. Such mutual regulation of TPH1 and 5-HT 7 expression via 5-HT in PDACs suggests that TPH1-5-HT-5-HT 7 axis operates in a feed-forward manner in PDAC cell lines. In addition, in Capan-1 cells expressing very low level of TPH1 and 5-HT 7 , 5-HT treatment up-regulated TPH1 and 5-HT 7 expressions accompanying gemcitabine resistance. Notably, EZH2 inhibition down-regulated EZH2 itself, TPH1 and 5-HT 7 expression, whereas inhibition of TPH1-5-HT-5-HT 7 axis reduced expression of TPH1 and 5-HT 7 , but not of EZH2. The results indicate that EZH2 acts as a master regulator permitting the expression of TPH1 and 5-HT 7 , forming EZH2-TPH1-5-HT-5-HT 7 axis. EZH1, a paralog of EZH2, is widely expressed in non-proliferating adult cells, whereas EZH2 is preferentially expressed in proliferating cells [45]. EZH1 complements EZH2 in maintaining stem cell identity and executing pluripotency [46]. In the current study, we found EZH2 KD with esiRNA slightly decreased EZH1 expression levels, although we used EZH2 esiRNA with guaranteed specificity and effectiveness. The result may be due to their 63% sequence homology and the 94% identity of their SET domain. This result suggests that it will be important to use a highly specific silencing strategy and to perform rescue experiments in a future study. In addition, a recent study reported that EZH1 is globally distributed in the chromatin of aggressive lymphomas, and both EZH1 and EZH2 play critical roles in the chromatin regulation [47]. Given the report, a future study is also required to determine whether EZH1 also contributed to the modulation of TPH1 and 5-HT 7 pathways in pancreatic cancer cells.
The operation of TPH1-5-HT-5-HT 7 axis utilized PI3K/Akt and JAK2/STAT3 signaling pathways that were activated through Gβγ components of the receptor, consistent with our previous findings in breast cancer cells [28]. There is a report that indicates the JAK2 signaling pathway is linked with TPH1 induction by activation of prolactin receptor, a cytokine receptor, in pancreatic β cells [29,48]. Here, the current study first reported that JAK2/STAT3 is linked to 5-HT 7 receptor in PDAC cells, in addition to the PI3K/Akt signaling pathway. We also demonstrated that 5-HT-enhanced EZH2 also regulated PI3K/Akt and JAK2/STAT3 signaling pathways. In addition, the fact that PI3K/Akt activity was dependent not only on the Gβγ component but also on Ras, which is overactivated by gainof-function mutation in PDAC, was confirmed by the results that 5-HT 7 KD or treatment with inhibitor (SB-269970) maintained PI3K/Akt-dependent NF-κB nuclear translocation and EZH2 expression level. Moreover, our study confirmed that EZH2 KD, but not TPH1-5-HT 7 KD, induced an increase in the expression of DAB2IP and PTEN, which inhibit Ras and PI3K, respectively [34][35][36][37]. These results demonstrate that once EZH2 is expressed, it performs a master regulatory action on intracellular signaling pathways through two directions, (1) activation of PI3K/Akt and JAK2/STAT3 through transactivation of TPH1-5-HT 7 expressions, and (2) supporting PI3K/Akt activity by inhibition of the signals that inhibit PI3K/Akt through downregulation of DAB2IP and PTEN (Figure 8). Despite the present results with PDAC cell lines, future studies will be needed to further evaluate the molecular pathways of the EZH2-TPH1-5-HT 7 axis in vivo.
As confirmed by co-immunoprecipitation with EZH2 and pan methyl lysine antibodies, the trans-activating action of EZH2 was performed in two sequential processes. That is, (1) methylation of EZH2 itself and TFs (NF-kB and Sp1), and (2) methylated EZH2 binds to TFs (NF-kB, Sp1, and p-STAT3), regulating gene expressions. The automethylation of EZH2 is reported as a self-activating mechanism for PRC2 [38]. However, in the current study we found that SUZ12 was not binding with EZH2, indicating the PRC2-independent action of EZH2 in PDAC as reported in breast cancer [49]. Interestingly, although nuclear p-STAT3 was involved in EZH2 transactivating action, p-STAT3 as well as STAT3 were not methylated in PDAC, which is different from other cancer cell types that methylated STAT3 by EZH2 is involved in tumorigenesis of glioblastoma stem-like cells [14].
Overall, EZH2 permits transactivation of TPH1 and 5-HT 7 , allowing the TPH1-5-HT-5-HT 7 axis to stimulate PI3K/Akt and JAK2/STAT3 pathways in a feed-forward manner in PDAC, leading to maintenance of pancreatic CSC populations and drug resistance. The transactivation of TPH1 and 5-HT 7 was further reinforced by removal of the PI3K/Akt inhibitory signals, such as DAB2IP and PTEN. The critical role of the EZH2-TPH1-5-HT-5-HT 7 axis was confirmed in an in vivo mouse tumor model showing similar tumor regressing effects of GSK-126, telotristat, and SB-269970.
Cancers 2021, 13, x 13 of 16 independent action of EZH2 in PDAC as reported in breast cancer [49]. Interestingly, although nuclear p-STAT3 was involved in EZH2 transactivating action, p-STAT3 as well as STAT3 were not methylated in PDAC, which is different from other cancer cell types that methylated STAT3 by EZH2 is involved in tumorigenesis of glioblastoma stem-like cells [14]. Overall, EZH2 permits transactivation of TPH1 and 5-HT7, allowing the TPH1-5-HT-5-HT7 axis to stimulate PI3K/Akt and JAK2/STAT3 pathways in a feed-forward manner in PDAC, leading to maintenance of pancreatic CSC populations and drug resistance. The transactivation of TPH1 and 5-HT7 was further reinforced by removal of the PI3K/Akt inhibitory signals, such as DAB2IP and PTEN. The critical role of the EZH2-TPH1-5-HT-5-HT7 axis was confirmed in an in vivo mouse tumor model showing similar tumor regressing effects of GSK-126, telotristat, and SB-269970. Figure 8. Schematic summary of noncanonical permissive action of EZH2 on TPH1-5-HT7 expression in PDAC. 5-HT activates PI3K/Akt and JAK2/STAT3 through Gβγ linked to 5-HT7. Together with Ras-activated ERK signaling, these signaling pathways enhance the nuclear level of Sp1, NF-κB, and p-STAT3, leading to EZH2 upregulation. Under the permissive action of EZH2, p-STAT3 is required for upregulation of TPH1 and 5-HT7, whereas Sp1 and NF-κB contribute to transcriptional repression of DAB2IP and PTEN. Through its noncanonical action, EZH2 activates intracellular signaling pathways through two ways, (1) activation of PI3K/Akt and JAK2/STAT3 through transactivation of TPH1-5-HT7 expressions, and (2) maintenance of the PI3K/Akt activity by inhibition of its inhibitory signals through downregulation of DAB2IP and PTEN.

Conclusions
In pancreatic cancer, long-term exposure to 5-HT in an autocrine or paracrine manner induced PRC2-independent EZH2 action that supported the TPH1-5-HT7 axis, leading to gemcitabine resistance and a CSC population increase. The inhibition of the EZH2-TPH1-5-HT-5-HT7 axis was effective in regressing gemcitabine-resistant pancreatic cancer growth in vivo, suggesting that the axis may be a potential therapeutic target for the treatment of drug-resistant PDAC.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1. Effect of HTR7, TPH1, or EZH2 KD on the expression of protein or mRNA of EZH2 and 5-HT system, Figure S2. Effects of PI3K/Akt and JAK2/STAT3 inhibitors on cell viability and sphere forming ability of PANC-1 and MIA PaCa-2 cells, Figure S3. Uncropped Western Blot images for Figure 1, Figure S4. Uncropped Western Blot images for Figure 3, Figure S5. Uncropped Western Blot images Figure 8. Schematic summary of noncanonical permissive action of EZH2 on TPH1-5-HT 7 expression in PDAC. 5-HT activates PI3K/Akt and JAK2/STAT3 through Gβγ linked to 5-HT 7 . Together with Ras-activated ERK signaling, these signaling pathways enhance the nuclear level of Sp1, NF-κB, and p-STAT3, leading to EZH2 upregulation. Under the permissive action of EZH2, p-STAT3 is required for upregulation of TPH1 and 5-HT 7 , whereas Sp1 and NF-κB contribute to transcriptional repression of DAB2IP and PTEN. Through its noncanonical action, EZH2 activates intracellular signaling pathways through two ways, (1) activation of PI3K/Akt and JAK2/STAT3 through transactivation of TPH1-5-HT 7 expressions, and (2) maintenance of the PI3K/Akt activity by inhibition of its inhibitory signals through downregulation of DAB2IP and PTEN.

Conclusions
In pancreatic cancer, long-term exposure to 5-HT in an autocrine or paracrine manner induced PRC2-independent EZH2 action that supported the TPH1-5-HT 7 axis, leading to gemcitabine resistance and a CSC population increase. The inhibition of the EZH2-TPH1-5-HT-5-HT 7 axis was effective in regressing gemcitabine-resistant pancreatic cancer growth in vivo, suggesting that the axis may be a potential therapeutic target for the treatment of drug-resistant PDAC.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13215305/s1, Figure S1. Effect of HTR7, TPH1, or EZH2 KD on the expression of protein or mRNA of EZH2 and 5-HT system, Figure S2. Effects of PI3K/Akt and JAK2/STAT3 inhibitors on cell viability and sphere forming ability of PANC-1 and MIA PaCa-2 cells, Figure S3. Uncropped Western Blot images for Figure 1, Figure S4. Uncropped Western Blot images for Figure 3, Figure S5. Uncropped Western Blot images for Figure 4, Figure S6. Uncropped Western Blot images for Figure 5, Figure S7. Uncropped Western Blot images for Figure 6.