Curcumin Reverses NNMT-Induced 5-Fluorouracil Resistance via Increasing ROS and Cell Cycle Arrest in Colorectal Cancer Cells

Nicotinamide N-methyltransferase (NNMT) plays multiple roles in improving the aggressiveness of colorectal cancer (CRC) and enhancing resistance to 5-Fluorouracil (5-FU), making it an attractive therapeutic target. Curcumin (Cur) is a promising natural compound, exhibiting multiple antitumor effects and potentiating the effect of 5-FU. The aim of the present study is to explore the effect of Cur on attenuating NNMT-induced resistance to 5-FU in CRC. A panel of CRC cell lines with different NNMT expressions are used to characterize the effect of Cur. Herein, it is observed that Cur can depress the expression of NNMT and p-STAT3 in CRC cells. Furthermore, Cur can induce inhibition of cell proliferation, G2/M phase cell cycle arrest, and reactive oxygen species (ROS) generation, especially in high-NNMT-expression CRC cell lines. Cur can also re-sensitize high-NNMT-expression CRC cells to 5-FU both in vitro and in vivo. In summary, it is proposed that Cur can reverse NNMT-induced cell proliferation and 5-FU resistance through ROS generation and cell cycle arrest. Given that Cur has long been used, we suppose that Cur is a promising anticancer drug candidate with minimal side effects for human CRC therapy and can attenuate NNMT-induced resistance to 5-FU.


Introduction
Statistically, colorectal cancer (CRC) is the second most common cause of cancerrelated death worldwide [1]. Chemotherapy is still a mainstay of CRC treatment, and adjuvant chemotherapy based on 5-fluorouracil (5-FU) can significantly improve the prognosis of CRC [2]. However, 5-FU resistance occurs in certain areas of CRC patients, which leads to poor clinical prognosis [3]. Therefore, the discovery of drugs in possession of a potential synergistic effect with 5-FU is urgently required.

Cells and Cell Culture
Human CRC cell lines SW480, which exhibited low NNMT expression, and HT-29, which exhibited low NNMT expression, were purchased from Cell Bank at the Chinese Academy of Sciences (Shanghai, China). STR authentication of all cell lines was completed. The modified cell lines SW480/NC and SW480/NNMT (NNMT overexpression) were constructed as described in our previous study [6]. All of the cell lines were cultured in RMPI-1640 (cat. No. 11875-093. Gibco, Grand Island, NY, USA) supplemented with 1% penicillin-streptomycin liquid (cat. No. 15140-122. Merck KgaA) and 10% fetal bovine serum (cat. No. 26010066. Thermo Fisher Scientific, Inc., Waltham, MA, USA) at 37 • C with 5% CO 2 .

Cell Viability Assay
Cell viability assay was detected using a Cell-Counting Kit-8 (CCK-8; cat. No. CK04; Dojindo Molecular Technologies, Inc., Kumamoto, Japan) following the manufacturer's instructions, Briefly, 5 × 10 3 cells were seeded in 96-well plates per well and then incubated with different concentrations of Cur and/or 5-FU for 24 h. CCK-8 solution (1:10 v/v) was added to cells and incubated at 37 • C for 2 h, and absorbance at 450 nm was measured using a microplate reader (Bio-Rad, Hercules, CA, USA). The inhibition rate (IR) = [1 − (mean absorbance of experimental wells/mean absorbance of control wells)] × 100%.

Western Blot Analysis
Cells were lysed in RIPA lysis buffer (Beyotime Biotechnology, Shanghai, China) with protease inhibitors to extract total protein. Then, a 40 µg protein sample was separated by SDS-PAGE and transferred to a PVDF membrane (cat. No. ISEQ00010; Millipore, Bedford, MA, USA), followed by immunoblotting with primary antibodies (1:1000 v/v) and horseradish peroxidase-conjugated secondary antibodies (1:5000 v/v). Signals were visualized and captured by Image Lab (Bio-Rad, Hercules, CA, USA).

Cell Apoptosis Analysis
The Annexin V-PE/7-AAD Apoptosis Detection kit (cat. no. 559763, BD Biosciences) and Annexin V-FITC/PI Apoptosis Detection kit (cat. no. 556547, BD Biosciences) were used to detect apoptosis according to the manufacturer's instructions, and signals were analyzed by flow cytometry (FACSCalibur flow cytometer; BD Biosciences, San Jose, CA, USA), as previously described [24].

Cell Cycle Analysis
Cell cycle was analyzed using a cell cycle staining kit (cat. no. CCS01, Multisciences (lianke) Biotech Co., Ltd., Hangzhou, China) according to the manufacturer's instructions. Briefly, cells were harvested and washed with cold PBS twice, stained with binding buffer containing 50 µg/mL propidium iodide in the dark at room temperature for 30 min, and then immediately analyzed by flow cytometry (FACSCalibur flow cytometer; BD Biosciences, San Jose, CA, USA).

ROS Detection
ROS was detected using an ROS assay kit (cat. no. S0033S, Beyotime Biotechnology, Shanghai, China) according to the manufacturer's instructions and analyzed by flow cytometry (FACSCalibur flow cytometer, BD, San Jose, CA, USA), as previously described [24].

Xenograft Experiments
Six-week-old male BALB/c nude mice were purchased from Model Animal Research Center of Nanjing University and housed under pathogen-free conditions (12/12 h light/dark cycle, 24 ± 2 • C; humidity, 50 ± 10%) with free access to food and water. All animals were acclimated for at least 5-7 days before treatment. The xenograft of CRC was constructed using SW480/NC and SW480/NNMT cell lines as previously described [24].
After subcutaneous injection with SW480/NC and SW480/NNMT cells for 10 days, mice were treated with 5-FU (30 mg/kg body weight dissolved in saline, intraperitoneal administration, every 2 days) and/or Cur (100 mg/kg body weight dissolved in 10% PEG, intragastric administration, every other day). Development of the tumors and weight loss were regularly monitored for 3 weeks. All of the mice were intraperitoneally injected with barbiturate (100 mg/kg body weight) and promptly sacrificed by cervical dislocation at the end of the experiment, and all tumors were harvested. Tumor volume was calculated according to V = (length × width 2 )/2.

TUNEL Assay
Tumor tissues were fixed, embedded, and sliced according to a standard protocol. Apoptotic cells in tumor tissue sections was analyzed by TUNEL assay using an in situ Apoptosis Detection kit (cat. No. 11684795910, Roche Diagnostics, Basel, Switzerland) following the manufacturer's instructions, as previously described [7]. Ten random fields of each tumor slide were selected, and the rate of positive stained cells was calculated.

Statistical Analysis
GraphPad Prism v7.0 software was used for statistical analysis. All data are presented as the mean ± SD from three independent experiments. The two-sample t-test was used for two-group comparisons. The ANOVA and Bonferroni method were used for multiple comparisons among more than two groups. p value < 0.05 was considered statistically significant difference.
As previously reported, NNMT can reduce human CRC cells' sensitivity to 5-FU [7]; thus, we aimed to determine whether Cur could reduce the NNMT-induced resistance to 5-FU in CRC cells. When we used the IC 50 concentration of Cur combined with different concentrations of 5-FU, a considerable number of cells died, and it was difficult to evaluate the IC 50 of 5-FU. Then, we used a lower concentration of Cur (15 µM in HT-29 cell lines and 7.5 µM in SW480 cell lines) to co-treat with 5-FU, and the IC 50 of 5-FU was calculated using the CCK-8 assay. Table 1 showed that the IC 50 of 5-FU was dramatically reduced when combined with Cur: from 131.81 to 50.98 mg/L in HT-29/NC (fold decrease of approximately 2.59), from 54.79 to 29.95 mg/L in HT-29/shNNMT (fold decrease of approximately 1.83) ( Figure 1B,C), from 15.52 to 11.64 mg/L in SW480/NC (fold decrease of approximately 1.33), and from 46.32 to 18.2 mg/L in SW480/NNMT (fold decrease of approximately 2.55) ( Figure 1E,F). In summary, it is suggested that Cur could inhibit cell proliferation and attenuate NNMT-induced resistance to 5-FU in CRC cells.
of approximately 2.59), from 54.79 to 29.95 mg/L in HT-29/shNNMT (fold decrease of approximately 1.83) ( Figure 1B,C), from 15.52 to 11.64 mg/L in SW480/NC (fold decrease of approximately 1.33), and from 46.32 to 18.2 mg/L in SW480/NNMT (fold decrease of approximately 2.55) ( Figure 1E,F). In summary, it is suggested that Cur could inhibit cell proliferation and attenuate NNMT-induced resistance to 5-FU in CRC cells.

Cur Inhibits NNMT and p-STAT3 in CRC Cells
p-STAT3 was reported as an upstream transcription factor of NNMT, and Cur could depress p-STAT3 and then depress NNMT expression at the mRNA level [22]. Herein, we treated cells with two concentrations of Cur, 30 μM or 35 μM for HT-29 cell lines, which was approximately IC50 of HT-29/NC or HT-29/shNNMT, and 15 μM or 20 μM for SW480 cell lines, which was approximately IC50 of SW480/NNMT or SW480/NC. Following the treatment of cells with Cur, Western blot showed that NNMT was downregulated in HT-29 cell lines (Figure 2A,B), while in SW480 cell lines, no significant decrease in NNMT was detected ( Figure 2C,D).
We also attempted to verify whether Cur inhibited NNMT expression by depressing p-STAT3. As our study previously reported, p-STAT3 is barely detected in HT-29 cell lines [24]. In SW480 cell lines, Cur could dramatically decrease p-STAT3 ( Figure 2C,E).

Cur Inhibits NNMT and p-STAT3 in CRC Cells
p-STAT3 was reported as an upstream transcription factor of NNMT, and Cur could depress p-STAT3 and then depress NNMT expression at the mRNA level [22]. Herein, we treated cells with two concentrations of Cur, 30 µM or 35 µM for HT-29 cell lines, which was approximately IC 50 of HT-29/NC or HT-29/shNNMT, and 15 µM or 20 µM for SW480 cell lines, which was approximately IC 50 of SW480/NNMT or SW480/NC. Following the treatment of cells with Cur, Western blot showed that NNMT was downregulated in HT-29 cell lines (Figure 2A,B), while in SW480 cell lines, no significant decrease in NNMT was detected ( Figure 2C,D).
Collectedly, it is suggested that Cur could downregulate the endogenous expression of NNMT (in HT-29 cells) and p-STAT3 (in SW480 cells), but we cannot draw a conclusion that the inhibition of NNMT is due to the inhibition of p-STAT3 in CRC cells. The untreated group of HT-29/NC was used as the control group, which was set as 1 in the histogram. (C) NNMT and p-STAT3 proteins were determined by Western blot after treatment with Cur for 24 h in SW480 cells. (D) NNMT proteins were displayed as a histogram after treatment with Cur for 24 h in SW480 cells. The untreated group of SW480/NNMT was used as the control group, which was set as 1 in the histogram. (E) p-STAT3 proteins were displayed as a histogram after treatment with Cur for 24 h in SW480 cells. The untreated group of SW480/NC was used as the control group, which was set as 1 in the histogram. Cur: curcumin. Data are expressed as means ± SD, n = 3, * p < 0.05, ** p < 0.01, *** p < 0.001.

Cur Induces G2/M Phase Cell Cycle Arrest Especially in High-NNMT-Expression CRC Cells
It was investigated whether the anti-proliferative effect of Cur was associated with cell cycle arrest.  The untreated group of HT-29/NC was used as the control group, which was set as 1 in the histogram. (C) NNMT and p-STAT3 proteins were determined by Western blot after treatment with Cur for 24 h in SW480 cells. (D) NNMT proteins were displayed as a histogram after treatment with Cur for 24 h in SW480 cells. The untreated group of SW480/NNMT was used as the control group, which was set as 1 in the histogram. (E) p-STAT3 proteins were displayed as a histogram after treatment with Cur for 24 h in SW480 cells. The untreated group of SW480/NC was used as the control group, which was set as 1 in the histogram. Cur: curcumin. Data are expressed as means ± SD, n = 3, * p < 0.05, ** p < 0.01, *** p < 0.001.
We also attempted to verify whether Cur inhibited NNMT expression by depressing p-STAT3. As our study previously reported, p-STAT3 is barely detected in HT-29 cell lines [24]. In SW480 cell lines, Cur could dramatically decrease p-STAT3 ( Figure 2C,E). Collectedly, it is suggested that Cur could downregulate the endogenous expression of NNMT (in HT-29 cells) and p-STAT3 (in SW480 cells), but we cannot draw a conclusion that the inhibition of NNMT is due to the inhibition of p-STAT3 in CRC cells.

Cur Induces G2/M Phase Cell Cycle Arrest Especially in High-NNMT-Expression CRC Cells
It was investigated whether the anti-proliferative effect of Cur was associated with cell cycle arrest.  The expression of G2/M phase-related proteins was detected after cells were treated with Cur. Western blot shows that the cell cycle inhibitory protein p21 was upregulated, while the cyclin-dependent kinases CDK1 and CDK2 were downregulated in a Cur-dependent manner in HT-29/NC and HT-29/shNNMT cell lines ( Figure 4A,B,D,F). Although slight differences were detected in cyclin A2, cyclin B1, and p-Rb in HT-29/NC, significant decreases were found when NNMT was downregulated in HT-29/shNNMT ( Figure 4A,C,E,G). Consistently, in SW480/NC and SW480/NNMT, p-Rb was significantly downregulated, and CDK1 and CDK2 were slightly decreased ( Figure 4H,K,M,N). However, inconsistently, p21 was reduced, and cyclin A2 and cyclin B1 were increased in SW480 cell lines ( Figure 4H,I,J,L). Thus, it was suggested that Cur could affect the cell cycle-related proteins and, in turn, induce the G2/M phase arrest in CRC cells. The expression of G2/M phase-related proteins was detected after cells were treated with Cur. Western blot shows that the cell cycle inhibitory protein p21 was upregulated, while the cyclin-dependent kinases CDK1 and CDK2 were downregulated in a Cur-dependent manner in HT-29/NC and HT-29/shNNMT cell lines ( Figure 4A,B,D,F). Although slight differences were detected in cyclin A2, cyclin B1, and p-Rb in HT-29/NC, significant decreases were found when NNMT was downregulated in HT-29/shNNMT ( Figure 4A,C,E,G). Consistently, in SW480/NC and SW480/NNMT, p-Rb was significantly downregulated, and CDK1 and CDK2 were slightly decreased ( Figure 4H,K,M,N). However, inconsistently, p21 was reduced, and cyclin A2 and cyclin B1 were increased in SW480 cell lines ( Figure 4H-J,L). Thus, it was suggested that Cur could affect the cell cycle-related proteins and, in turn, induce the G2/M phase arrest in CRC cells. The untreated group of HT-29/NC was used as the control group, which was set as 1. (H-N) Proteins were analyzed by Western blot and were displayed as a histogram in SW480 cell lines. The untreated group of SW480/NC was used as the control group, which was set as 1. Cur: curcumin. Data are expressed as means ± SD, n = 3, * p < 0.05, ** p < 0.01, *** p < 0.001.

Cur Has a Synergistic Effect with 5-FU via Induction of Cell Cycle Arrest in CRC Cells
As previously mentioned, the combination of Cur with 5-FU yielded notable synergy in inhibiting cell proliferation, and it was also determined whether the synergistic effect with 5-FU is associated with cell cycle arrest. As when co-treated with a higher concentration of Cur and 5-FU, a considerable number of cells died, and it was difficult to conduct the following detection, we used 20 mg/L 5-FU in HT-29 cell lines and 10 mg/L 5-FU in SW480 cell lines, representing the half concentration of IC 50 of 5-FU, combined with Cur (30 µM in HT-29 cells and 10µM in SW480 cells). Flow cytometry results show that 5-FU can dramatically induce G1/S phase arrest. The proportion of cells at the G1/S phase increased from 92.81 to 98.58% in HT-29/NC, from 93.83 to 98.23% in HT-29/shNNMT ( Figure 5A,B, respectively), from 88.24 to 97.31% in SW480/NC, and from 91.71 to 96.92% in SW480/NNMT ( Figure 5C,D, respectively). When co-treated with Cur and 5-FU, compared to cells treated with Cur alone, the proportion of cells at the G2/M phase was reduced in HT-29 cell lines, which could have been due to the G1/S phase arrest by 5-FU ( Figure 5A,B). The unstuck cell cycle was also observed in SW480 cell lines following co-treatment with Cur and 5-FU, and it seems that the cells are arrested in both G1/S and G2/M ( Figure 5C,D). We assume that Cur can reinforce the 5-FU-induced disorders on cell cycle in CRC cells.

Cur Has a Synergistic Effect with 5-FU via Induction of Cell Cycle Arrest in CRC Cells
As previously mentioned, the combination of Cur with 5-FU yielded notable synergy in inhibiting cell proliferation, and it was also determined whether the synergistic effect with 5-FU is associated with cell cycle arrest. As when co-treated with a higher concentration of Cur and 5-FU, a considerable number of cells died, and it was difficult to conduct the following detection, we used 20 mg/L 5-FU in HT-29 cell lines and 10 mg/L 5-FU in SW480 cell lines, representing the half concentration of IC50 of 5-FU, combined with Cur (30 μM in HT-29 cells and 10μM in SW480 cells). Flow cytometry results show that 5-FU can dramatically induce G1/S phase arrest. The proportion of cells at the G1/S phase increased from 92.81 to 98.58% in HT-29/NC, from 93.83 to 98.23% in HT-29/shNNMT ( Figure 5A,B, respectively), from 88.24 to 97.31% in SW480/NC, and from 91.71 to 96.92% in SW480/NNMT ( Figure 5C,D, respectively). When co-treated with Cur and 5-FU, compared to cells treated with Cur alone, the proportion of cells at the G2/M phase was reduced in HT-29 cell lines, which could have been due to the G1/S phase arrest by 5-FU ( Figure 5A,B). The unstuck cell cycle was also observed in SW480 cell lines following co-treatment with Cur and 5-FU, and it seems that the cells are arrested in both G1/S and G2/M ( Figure 5C,D). We assume that Cur can reinforce the 5-FU-induced disorders on cell cycle in CRC cells. We further detected the cell cycle-related proteins after treatment with Cur combined with 5-FU. Western blot analysis showed that in HT-29 cells, p21 was more dramatically increased when co-treated with Cur and 5-FU than when treated with Cur or 5-FU alone ( Figure 6A,B), and cyclin A2, cyclin B1, CDK1, and CDK2 were downregulated when co-treated with Cur and 5-FU, although 5-FU could increase them ( Figure  6A,C-F). Both Cur and 5-FU can decrease p-Rb and the G1/S phase-related proteins CDK4 and cyclin D1, moreover, combination of 5-FU and Cur can reinforce the inhibition  We further detected the cell cycle-related proteins after treatment with Cur combined with 5-FU. Western blot analysis showed that in HT-29 cells, p21 was more dramatically increased when co-treated with Cur and 5-FU than when treated with Cur or 5-FU alone ( Figure 6A,B), and cyclin A2, cyclin B1, CDK1, and CDK2 were downregulated when co-treated with Cur and 5-FU, although 5-FU could increase them ( Figure 6A,C-F). Both Cur and 5-FU can decrease p-Rb and the G1/S phase-related proteins CDK4 and cyclin D1, moreover, combination of 5-FU and Cur can reinforce the inhibition of p-Rb, CDK4 and cyclin D1 in HT-29 cells ( Figure 6A,G-I). The cell cycle-related proteins were also detected in SW480 cell lines after treated with 10 µM Cur and/or 10 mg/L 5-FU for 24 h. Consistently, p-Rb and cyclin D1 could be decreased by Cur and 5-FU, and Cur had a synergistic effect with 5-FU on the inhibition of p-Rb and cyclin D1 in SW480 cells ( Figure 6J,P,Q). Interestingly, although 5-FU could upregulate cyclin A2, cyclin B1, CDK1, CDK2, and CDK4 while Cur had a minimal effect on them, when co-treated with Cur and 5-FU, these proteins could be dramatically decreased in SW480 cells ( Figure 6J,L-O,R). p21 decreased when treated with both Cur or 5-FU alone and when co-treated ( Figure 6J,K). Thus we suggest that there is a considerable amount of synergy between Cur and 5-FU in terms of inducing cell cycle arrest in CRC cell lines.

Cur Induces Cell Cycle Arrest by Promoting ROS
ROS is one of the causes of cell cycle arrest. It was determined whether Cur inducing cell cycle arrest through ROS in CRC cells. Herein, we found that Cur could promote ROS generation in SW480 cell lines, and the ROS level induced by Cur was higher in SW480/NNMT cells than in SW480/NC ( Figure 7A). We used NAC, a known scavenger of ROS, to reverse the ROS, and flow cytometry showed that NAC pretreatment could reduce the ROS generation after incubation with 15 µM Cur, but when treated with 20 µM Cur, the ROS level induced by Cur was too high to be reduced ( Figure 7B). Then, we used 15 µM Cur to detect the effect of NAC on cell cycles in SW480 cells. Following pretreatment with NAC, flow cytometry results showed that the G1/S phase was partially rescued in 15 µM Cur-treated SW480 cell lines, from 58.57 to 61.78% in SW480/NC and from 4.57 to 24.02% in SW480/NNMT ( Figure 7E,F, respectively). The significant increase in G1/S by NAC is consistent with the decrease in ROS in SW480/NNMT ( Figure 7B,F). These results show that ROS plays important roles in triggering the G2/M phase cell cycle arrest caused by Cur in SW480 cells. We have reported that NNMT could reduce ROS generation in CRC cells after treatment with 5-FU [7]. It is suggested that Cur could reverse the reduction of ROS by NNMT. When we detected the ROS level induced by Cur in HT-29 cells, a pair of NNMT siRNA was used to knock down NNMT, as the lentiviral vector of NNMT shRNA used to construct the HT-29/NC and HT-29/shNNMT cell lines contained a green fluorescence protein tag that disturbs the green fluorescent signal of ROS. Consistent with the results in SW480 cell lines, the flow cytometry results show that following treatment with 30 µM or 35 µM Cur for 24 h, more ROS was induced in HT-29/siNC cells than in HT-29/siNNMT cells ( Figure 7C). However, ROS promotion caused by Cur was much more mild than that in SW480 cell lines. Meanwhile, pretreatment with NAC could only slightly reduce the ROS promoted by Cur in HT-29 cells (Figure 7D), and the cell cycle arrest caused by Cur could not be reduced by NAC (data not shown). Figure 6. Cur and 5-FU regulate proteins related to cell cycle in CRC cells. Cells were treated with Cur (30 μM for HT-29 cells; 10 μM for SW480 cells) and/or 5-FU (20 mg/L for HT-29 cells; 10 mg/L for SW480 cells) for 24 h. (A-I) Proteins were analyzed by Western blot and are displayed as a histogram in HT-29 cell lines. The untreated groups of HT-29/NC were used as the control group, which was set as 1; (J-R) Proteins were analyzed by Western blot and are displayed as a histogram in SW480 cell lines. The untreated groups of SW480/NC were used as the control group, which was set as 1. 5-FU: 5-Fluorouracil, Cur: curcumin. Data are expressed as means ± SD, n = 3, * p < 0.05, ** p < 0.01, *** p < 0.001. The results indicate that apoptosis induced by Cur is much less in high-NNMT-expression cell lines than in low-NNMT-expression cell lines, and Cur has a limited synergistic effect with 5-FU on increasing cell apoptosis. We assume that NNMT could still protect CRC cells from apoptosis under Cur and/or 5-FU treatment.

Cur Has a Limited Synergetic Effect with 5-FU on Inducing Cell Apoptosis
Given that NNMT could induce resistance to 5-FU through reducing cell apoptosis in CRC cells [7], cell apoptosis was detected to investigate the effect of Cur on CRC cells with different NNMT expression levels. The results indicate that apoptosis induced by Cur is much less in high-NNMTexpression cell lines than in low-NNMT-expression cell lines, and Cur has a limited synergistic effect with 5-FU on increasing cell apoptosis. We assume that NNMT could still protect CRC cells from apoptosis under Cur and/or 5-FU treatment.

Cur Has a Synergetic Effect with 5-FU on Decreasing Tumor Growth and Inducing Apoptosis In Vivo
Nude mice bearing SW480/NC and SW480/NNMT xenografts were used to explore the synergistic effect of Cur with 5-FU in vivo. We used 30 mg/kg 5-FU every 2 days for the 5-FU chemotherapy groups, as described in our previous study [7]. A previous study reported that up to 240 mg/kg/day is considered safe for mice [25]. In the current study, we used 100 mg/kg/day Cur for the 5-FU combination groups. It was observed that the average tumor volume and weight were smaller in the 5-FU combined with Cur treatment groups than in the 5-FU-treated groups, both in SW480/NC and SW480/NNMT tumors ( Figure 8A-C). Moreover, cell apoptosis of the tumor sections was analyzed by TUNEL assay, and the results showed that more apoptosis was induced in groups treated with Cur combined with 5-FU than in groups treated with 5-FU alone ( Figure 8D,E).

Cur Has a Synergetic Effect with 5-FU on Decreasing Tumor Growth and Inducing Apoptosis In Vivo
Nude mice bearing SW480/NC and SW480/NNMT xenografts were used to explore the synergistic effect of Cur with 5-FU in vivo. We used 30 mg/kg 5-FU every 2 days for the 5-FU chemotherapy groups, as described in our previous study [7]. A previous study reported that up to 240 mg/kg/day is considered safe for mice [25]. In the current study, we used 100 mg/kg/day Cur for the 5-FU combination groups. It was observed that the average tumor volume and weight were smaller in the 5-FU combined with Cur treatment groups than in the 5-FU-treated groups, both in SW480/NC and SW480/NNMT tumors ( Figure 8A-C). Moreover, cell apoptosis of the tumor sections was analyzed by TUNEL assay, and the results showed that more apoptosis was induced in groups treated with Cur combined with 5-FU than in groups treated with 5-FU alone ( Figure 8D,E).
In conclusion, Cur is found to enhance sensitivity to 5-FU via downregulating NNMT and ROS-induced cell cycle arrest in CRC. A schematic illustration of Cur downregulating NNMT and attenuating NNMT-induced resistance to 5-FU in CRC cells is shown in Figure 8F.  In conclusion, Cur is found to enhance sensitivity to 5-FU via downregulating NNMT and ROS-induced cell cycle arrest in CRC. A schematic illustration of Cur downregulating NNMT and attenuating NNMT-induced resistance to 5-FU in CRC cells is shown in Figure 8F.

Discussion
Chemo-resistance is a key reason for the poor prognosis of colorectal cancer (CRC). NNMT is an enzyme of nicotinamide metabolism, and we have previously reported that NNMT could enhance resistance to 5-FU in CRC [7]. Therefore, identifying NNMTtargeting drugs and detecting their anticancer efficacies are important for the battle against 5-FU resistance in CRC.
Given that NNMT is an enzyme that promotes cell proliferation and 5-FU resistance, inhibition of NNMT may be one of the ways to reduce 5-FU resistance. We have recently screened a natural products library and found that vanillin can decrease NNMT expression and attenuate NNMT-related resistance to 5-FU in CRC cells [24]. Curcumin (Cur), a promising natural compound of the turmeric plant, has been widely used in food. STAT3 is reported to upregulate NNMT expression, and Cur can inhibit NNMT expression in RNA levels as an inhibitor of p-STAT3 [22]. In the current study, it was also found that Cur could downregulate p-STAT3 and the endogenous expression of NNMT in protein levels. However, as NNMT is endogenously highly expressed and p-STAT3 is barely detected in HT-29 cells, in addition to NNMT being barely detected and p-STAT3 endogenously highly expressed in SW480 cells, we cannot conclude that the inhibition of NNMT by Cur is due to the inhibition of p-STAT3 in CRC cells, and we have not yet found a CRC cell line that possess high expression of both NNMT and p-STAT3. The precise mechanisms by which Cur inhibits NNMT expression require further research.
Cur has been reported to exert multiple anticancer effects [11][12][13][14]. In the present study, we also found that Cur can inhibit cell proliferation in CRC cells especially in high-NNMT-expression CRC cell lines and attenuate NNMT-induced resistance to 5-FU. To find the reasons that Cur caused cell death in CRC cells, especially in high-NNMT-expression cell lines, we detected the change of cell apoptosis and cell cycles following treated with Cur and/or 5-FU. In agreement with previous studies that Cur can induce apoptosis in CRC [17], it is found that Cur can induce cell apoptosis both in HT-29 and SW480 cell lines, and Cur has synergistic effect with 5-FU in increasing cell apoptosis. However, the apoptosis induced by Cur is much less in high-NNMT-expression cell lines than in low-NNMT-expression cell lines. Given that NNMT could induce resistance to 5-FU through reducing cell apoptosis in CRC cells [7], we assume that NNMT could also protect CRC cells from apoptosis by Cur.
In terms of the detection of cell cycles, it is found that Cur can induce more G2/M phase arrest in high-NNMT-expression cell lines than that in low ones, which is consistent with the CCK-8 assay results. Therefore, we deem that Cur reverse the NNMT-induced cell proliferation mainly through inducing cell cycle arrest rather than inducing cell apoptosis. Another important consideration is that subG1 changes is a marker of apoptosis and long exposures (at least 48 h) to some apoptosis-inducible drugs may lead to subG1 accumulation [26][27][28]. However, although Cur could induce apoptosis, no subG1 accumulation was detected following the Cur treatment for 24 h in the present study. We speculated that it might be related to the duration of Cur exposure. This phenomenon of no subG1 but apoptosis status had also been reported in several studies [29,30]. To identify the molecular mechanism by which Cur induces the G2/M phase arrest, we detected the proteins associated with the G2/M phase. Cells express different cyclins during the different stages of the cell cycle. CDK2 is activated by cyclin A2 during the late stages of DNA replication to drive the transition from S phase to G2 phase. Moreover, at the G2 phase, cyclin B complexes with CDK1 to facilitate the onset of mitosis [31]. CDKs phosphorylate and inactivate the retinoblastoma protein (Rb), an adaptor protein that represses transcription, which in turn controls the activity of E2F transcription factors that stimulate proliferation [32]. The cyclin-dependent kinase inhibitor p21 inhibits cell cycle progression primarily through the inhibition of CDK2 activity, and p21 disrupts the interaction between CDK and substrates that bind to CDK-cyclin [33]. Cur markedly decreases CDK1 in human glioma cells and downregulates CDK1 and cyclin B1 in human colon cancer Colo 205 cells [34,35]. In the current study, we found that Cur upregulates p21 and downregulates CDK1, CDK2, cyclin A2, cyclin B1, and p-Rb in HT-29 cells. However, in SW480 cells, Cur causes the G2/M phase arrest via significant inhibition of p-Rb and a slight decrease in CDK1 and CDK2. We assume that Cur plays different roles in different CRC cell lines, as this study lacks observations of a further mechanism.
Given that NNMT has previously noted to promote the cell cycle via the active G1 phase in CRC cells [6], and high expression of NNMT in both HT-29/NC and SW480/NNMT cells can protect CDK1, CDK2, cyclin A2, cyclin B1, and p-Rb against inhibition by Cur, which indicates that NNMT plays a role in promoting G2/M phase. This is inconsistent with the phenotype in which G2/M phase arrest is induced by Cur in high-NNMTexpression cell lines. In terms of inhibiting the cell cycle, in addition to affecting the interphase, Cur has also been reported to inhibit mitosis, including suppression of the spindle [36] and inhibition of mitotic kinesin Eg5 [37]. Besides, Cur has been identified as a DNA topoisomerase II poison involved in DNA damage [38,39]. As no reports have revealed the effect of NNMT on the inhibition of mitosis and DNA topoisomerase, the Cur's impact on them may provide some clues as to why it can inhibit cells with high a NNMT expression. We assumed that NNMT might accelerate the cell cycle and Cur might be more effective at inhibiting cells that have a faster cell cycle. Further research is needed to confirm this.
Moreover, it was found that Cur has a synergistic effect with 5-FU on inducing cell cycle arrest in CRC cells. 5-FU is a pyrimidine analogue and can cause cell cycle arrest at the G1/S phase by inhibiting thymidylate synthetase [40]. In the current study, it was also noted that 5-FU could induce G1/S phase arrest in HT-29 and SW480 cell lines. Although the flow cytometry results show the unstuck cell cycles when co-treated with Cur and 5-FU, the Western bolt results show the dramatic synergistic effect of Cur and 5-FU on the inhibition of p-Rb, cyclins, and CDKs, thus, we assume that Cur can reinforce the disorders induced by 5-FU on the cell cycle in CRC cells.
ROS is one of the causes of cell cycle arrest, and previous studies have reported that Cur could induce ROS generation in CRC cells [41,42]. In this study, we also found that Cur can induce ROS in HT-29 and SW480 cell lines. NNMT is reported to reduce ROS generation in CRC cells after treatment with 5-FU [7], which means that NNMT might act as a ROS scavenger or inhibitor. Interestingly, it was found that Cur can induce more ROS generation in high-NNMT-expression CRC cell lines than in low-NNMT-expression ones. This might explain the more mortality of cells and G2/M phase arrests induced by Cur in high-NNMT-expression cells. A similar phenotype was also found in vanillintreated CRC cells, in which vanillin induced more ROS and cell death in high-NNMTexpression cell lines [24]. However, further research is required to shed light on this seemingly contradictory phenomenon and to explain why the inhibition on ROS of NNMT disappeared. In this study, we also attempted to reduce ROS via NAC to determine whether ROS is one of the key causes of cell cycle arrest induced by Cur. It was found that NAC pretreatment can partially reduce the cell cycle arrest induced by Cur in SW480 cell lines. In HT-29, NAC could reduce the Cur-induced ROS generation, but only a fraction of the cell cycle arrest induced by Cur could be reduced by NAC, which is consistent with previous reports [16]. Moreover, the source of ROS generation and its mechanisms are not clear, and thus, warrant further research.
We constructed a murine xenograft model using SW480/NC and SW480/NNMT cells to assess the synergistic effect of Cur on 5-FU in vivo. It was observed that Cur could enhance the inhibition of 5-FU in tumor proliferation in both SW480/NC and SW480/NNMT groups. Consistent with a previous study [7], tumors in the SW480/NNMT groups exhibited a larger volume and less 5-FU-induced cell apoptosis than those of tumors in the SW480/NC groups when treated with 5-FU alone. However, when treated with Cur combined with 5-FU, it was noted that tumors in the SW480/NNMT were smaller, and there was less 5-FU-induced cell apoptosis when compared with tumors in the SW480/NC groups, which indicates that Cur exerts a stronger synergistic effect with 5-FU on high-NNMT-expression tumors than low-NNMT-expression tumors. This is inconsistent with the results of the in vitro cell experiments and could be due to the limited number of animals, leading to a statistical bias.

Conclusions
In the present study, it is demonstrated that Cur can reverse NNMT-induced resistance to 5-FU in vivo and in vitro, and the mechanism that involved in ROS generation and cell cycle arrest induced by Cur is partially demonstrated. Considering the fact that Cur has long been used, in this study, we propose Cur as a promising anticancer compound with minimal side effects for CRC adjuvant chemotherapy, particularly for CRC with high NNMT expression.   Figure A1B,D). The results indicate that Cur has a limited synergistic effect with 5-FU on increasing cell apoptosis. Given that the apoptosis induced by Cur is much less in high-NNMT-expression cell lines than in low-NNMT-expression cell lines, we assume that NNMT could still protect CRC cells from apoptosis under Cur and/or 5-FU treatment.