Resveratrol inhibits bile acid‐induced gastric intestinal metaplasia via the PI3K/AKT/p‐FoxO4 signalling pathway

Gastric intestinal metaplasia (GIM) is the essential pre‐malignancy of gastric cancer. Chronic inflammation and bile acid reflux are major contributing factors. As an intestinal development transcription factor, caudal‐related homeobox 2 (CDX2) is key in GIM. Resveratrol has potential chemopreventive and anti‐tumour effects. The aim of the study is to probe the effect of resveratrol in bile acid‐induced GIM. We demonstrated that resveratrol could reduce CDX2 expression in a time‐ and dose‐dependent manner in gastric cell lines. A Cignal Finder 45‐Pathway Reporter Array and TranSignal Protein/DNA Array Kit verified that resveratrol could increase Forkhead box O4 (FoxO4) activity and that Chenodeoxycholic acid (CDCA) could reduce FoxO4 activity. Furthermore, bioinformatics analysis showed that FoxO4 could bind to the CDX2 promoter, and these conjectures were supported by chromatin‐immunoprecipitation (ChIP) assays. Resveratrol can activate FoxO4 and decrease CDX2 expression by increasing phospho‐FoxO4 nucleus trans‐location. Resveratrol could increase FoxO4 phosphorylation through the PI3K/AKT pathway. Ectopic FoxO4 expression can up‐regulate FoxO4 phosphorylation and suppress CDCA‐induced GIM marker expression. Finally, we found a reverse correlation between p‐FoxO4 and CDX2 in tissue arrays. This study validates that resveratrol could reduce bile acid‐induced GIM through the PI3K/AKT/p‐FoxO4 signalling pathway and has a potential reversing effect on GIM, especially that caused by bile acid reflux.

p-FoxO4 signalling pathway and has a potential reversing effect on GIM, especially that caused by bile acid reflux.

K E Y W O R D S
bile acid, FoxO4, gastric intestinal metaplasia, PI3K/AKT, resveratrol 1 | INTRODUCTION Gastric cancer (GC) is the fifth highest incidence and third highest mortality in cancer-related diseases among the world population (Bray et al., 2018;Etemadi, Safiri, Sepanlou, Ikuta, & Bisignano, 2020;Fitzmaurice et al., 2019), and it is the second most common type of cancer and the third leading cause of cancer-associated mortality in China (Gao & Wu, 2019;Yang et al., 2018). Gastric intestinal metaplasia (GIM), especially incomplete type of intestinal metaplasia (IIM), characterized by caudal-related homeobox 2 (CDX2) expression, is a significant premalignant lesion in the transition from chronic atrophic gastritis to GC (Nakayama et al., 2018;Rodrigues et al., 2018;Savcenko et al., 2019). In fact, many factors promote the development of GIM. Previous studies have shown that long-term bile acid reflux and chronic inflammation are considered as important therapeutic factors (Hegyi, Maléth, Walters, Hofmann, & Keely, 2018;Li et al. 2019;Yu et al., 2019). The transition from IM to GC occurs over an average of 5-10 years (Correa, Haenszel, Cuello, & Ruiz, 1990;Correa, Haenszel, Cuello, Tannenbaum, & Archer, 1975). Therefore, IM stage can be an important broad time window to block the progression from gastritis to GC. Currently, there is no effective progress in the treatment or prevention of IM (Kim, 2019).
Importantly, some natural compounds and small molecules that could be used for the treatment have been discovered.
GC incidence is lower in French than in China, because French people love to drink red wine almost every day among many factors (Bray et al., 2018;Etemadi et al., 2020). Studies have shown that a small amount of red wine can be beneficial for health. Resveratrol is a polyphenol compound in red wine. Previous studies have shown that resveratrol, usually extracted from grapes and vines, has a variety of antioxidant, anti-inflammatory, and anti-tumour effects (Chassot et al., 2018;Huminiecki & Horba nczuk, 2018;J. Xu et al., 2017; Y. Zhang, Cui, Wang, Gong, & Wang, 2019). Since it is easily accessible in nature, resveratrol is widely used in cosmetics and adjuvant drugs for tumours (Huminiecki & Horba nczuk, 2018;B. Xu et al., 2020;Zulueta, 2015). For example, there are many studies on resveratrol in liver cancer, colon cancer and GC (Farhadnejad, Emamat, & Zand, 2019;Fenner, 2017;Nana et al., 2018;Pistollato et al., 2017;Said, Mantawy, & El-Demerdash, 2019). However, there is no research on resveratrol in IM. Therefore, this study investigates whether and how it affects IM and the underlying mechanism.

| Transcription factor array
The identities of the DNA pull-down proteins were determined by an established procedure using a TranSignal Protein/DNA Array Kit (Panomics, USA) according to the manufacturer's protocol. In short, the pulled-down proteins were incubated with TranSignal probe mix (a set of biotin-labelled oligonucleotides corresponding to consensus sequences of 345 known transcription factors as shown in Tables S1 and S3) to allow the formation of DNA/protein complexes, which were then separated from the unbound probes by agarose gel electrophoresis. The probes in the complexes were extracted and hybridized to a TranSignal Array. The hybridized probes signals were visualized using the chemiluminescent imaging system provided with the TranSignal Protein/DNA Array Kit (Panomics, USA) and exposed to X-ray non-fat milk in 1× TBS/0.1% (v/v) Tween-20 at room temperature. Primary antibodies were added and incubated at 4 C overnight. Each primary antibody was used according to the manufacturer's protocol.

| Immunofluorescence
Cells were seeded on a four-well Glass slide (PEZGS16, Millicell, Ger- to the ratio of a control well, and the relative fold-change of the CDCA group relative to the control group was calculated.

| Chromatin immunoprecipitation
Chromatin-immunoprecipitation (ChIP) assays were performed using an EZ ChIP Kit (Millipore, USA). Cells were cross-linked with 1% paraformaldehyde for 10 min at 37 C, and the reaction was quenched with 2.5 M glycine for 5 min at room temperature. DNA was immunoprecipitated from the cell lysates with a FoxO4 antibody (Cat. #720154, Thermo Scientific Invitrogen, USA) and subjected to FoxO4 binding site amplification (Table 4). The amplified fragments were then analysed on an agarose gel. A non-specific IgG antibody was used as a negative control.   Figure 1C).

| Statistical analysis
These results suggest that CDCA could increase CDX2 in GES-1 cells.
Surprisingly, resveratrol was able to inhibit the CDCA-induced increase in CDX2 and the downstream intestinal marker KlF4 ( Figure 1D) at the protein level. In GC cells, resveratrol reduced CDX2 expression ( Figure 1E) at the protein and mRNA levels. Similarly, resveratrol downregulated CDX2 and downstream intestine-specific markers in a time-and dosage-dependent manner at the protein and mRNA levels ( Figure 1F,G). Collectively, these results suggest that resveratrol may be a good candidate for treating or reversing IM.
3.2 | FoxO4 was identified by the Cignal Finder 45-pathway reporter array and TranSignal protein/ DNA array kit and may combine directly with CDX2 promoter, as predicted by bioinformatics analysis To further identify the functional targets involved in resveratrolinduced CDX2 down-regulation in CDCA-stimulated GES-1 and AGS cells, a pathway reporter array and protein/DNA array kit were performed. We found that resveratrol could increase FoxO4 activity upon CDCA stimulation, thus reducing the activity of transcription factors (Figure 2A,B, Tables S1-S4). The red pentagram indicates active FoxO4. Moreover, we used a bioinformatics analysis to predict transcription factors that can bind directly to the CDX2 promoter, and found FoxO4 can adapt to this result ( Figure 2C, Table S5). Then, we determined the changes in active FoxO4 ( Figure 2D). Together, these results suggest that active FoxO4 may play an essential role in the effects of resveratrol on reducing CDX2 expression.

| Resveratrol activated FoxO4 through phosphorylation and nuclear trans-location
Next, we aimed to further investigate the mechanism of FoxO4 activation after resveratrol stimulation. We found that p-FoxO4 Ser262 was increased in a time-and dosage-dependent manner. However, there was no obvious change in regular FoxO4 at the protein level in the AGS cell line ( Figure 3A,B). Similar to this result, the same pattern of p-FoxO4 Ser262 expression was found in CDCA-induced GES-1 cells ( Figure 3C,D). Moreover, FoxO4 expression was significantly reduced. Cell immunofluorescence results revealed that resveratrol increased FoxO4 Ser197 phosphorylation and reduced the increase in CDX2 induced by CDCA, but the FoxO4 changes were not consistent ( Figure 3E-H). Collectively, these results illustrate that FoxO4 works by increasing its phosphorylation and nuclear trans-location.

| FoxO4 inhibits CDX2 expression by directly targeting the CDX2 promoter region which was screened by bioinformatics prediction and ChIP
We further elucidated the connection between FoxO4 and the CDX2 signalling pathway. Reporter genes, containing the CDX2 promoter, were transfected into GES-1 cells, which were then treated with CDCA. This analysis revealed that CDCA-based CDX2 regulation was controlled by potential FoxO4 binding sites located between −2000 and −323 bp ( Figure 4A). ChIP assays further confirmed that FoxO4 These results revealed that FoxO4 regulates CDX2 expression by binding to its promoter in gastric cells.

| CDX2 expression was negatively regulated by p-FoxO4
To identify the relationship between p-FoxO4 and CDX2, GES-1 and BGC823 cells, which have low internal CDX2 expression, were chosen for siFoxO4 transfection to observe the changes in intestine-specific markers. We found that upon the absolute decrease in endogenous p-FoxO4, CDX2 and intestinal markers no longer had inhibitory effects and significantly  3.6 | Resveratrol is able to activate FoxO4 through the PI3K/AKT pathway To investigate the mechanism of FoxO4 activation after resveratrol treatment, the inhibitor PI3K/AKT pathway LY294004, siFoxO4 and Lv-FoxO4 were used. LY294004 blocked the phosphorylation of AKT at Ser473 and Thr308 to block the PI3K/AKT pathway ( Figure 6A). It further reduced the phosphorylation of FoxO4, thereby reducing the decreasing of CDX2 by resveratrol ( Figure 6B). The relative expression changes in p-AKT Ser473, p-AKT Thr308, p-FoxO4 Ser262 and CDX2 clearly indicated the relationship between p-AKT and p-FoxO4 ( Figure 6C). To explore the effect of resveratrol on CDX2 in the absence of p-FoxO4, siFoxO4 was transfected into AGS cells. Resveratrol weakened the reduction in CDX2 levels in the absence of p-FoxO4 ( Figure 6D). The relative expression changes in p-FoxO4 Ser262 and CDX2 clearly indicated the relationship between p-FoxO4 and CDX2 ( Figure 6E). GES-1 cells were transfected with Lv-FoxO4 to observe the relationship between CDCA and p-FoxO4. The absolute increase in p-FoxO4 weakened the effect of CDCA on CDX2 ( Figure 6F,G). These results revealed that resveratrol can activate FoxO4 through PI3K/AKT pathway activation.

| p-FoxO4 and CDX2 showed a reverse correlation in normal and GIM tissue array
Finally, to examine whether the above-described regulation in gastric cell lines is clinically relevant, IHC for p-FoxO4 and CDX2 was applied to 12 normal tissues and 49 IM tissues. Compared with that in normal tissues, CDX2 was significantly increased in IM tissues. The p-FoxO4 levels were decreased in IM tissues compared with normal tissues ( Figure 7A,B; normal: 6.583 ± 0.570, n = 12; 1.417 ± 0.398, n = 12; Wilcoxon matched pairs, p value = .0005; IM: 2.367 ± 0.301, n = 49; 6.959 ± 0.464, n = 49; paired t test, p value <.0001). In addition, the expression of CDX2 and p-FoxO4 showed a reverse correlation ( Figure 7C; n = 61, r = −.5216, p value <.0001). In summary, these results showed that the PI3K/AKT/p-FoxO4/CDX2 pathway is inactive in human gastric IM.

| DISCUSSION
In this study, we discovered a resveratrol-induced pathway involving active AKT and downstream FoxO4 activated by phosphorylation.
The pathway antagonized the induction of CDX2 by bile acids and may contribute to the treatment of IM.
Currently, the function of CDX2 in regulating intestinal differentiation is widely accepted. Furthermore, the previous work of our group has focused on the molecular mechanisms driving CDX2 and its downstream intestine-specific marker expression by bile acids in the oesophagus and stomach, which include the FXR/SHP/NF-κB pathway (Zhou et al., 2018), the miR-92a/FOXD1/NF-κB pathway (Li et al. 2019) Liu et al., 2017;Wood et al., 2004). As a pharmacological agent, resveratrol has a wide spectrum of targets, whose biological F I G U R E 4 FoxO4 inhibits CDX2 expression by directly targeting the CDX2 promoter region. (A) Serially truncated CDX2 promoter constructs were cloned into pGL3-luciferase reporter plasmids and transfected into GES-1 cells. Four hours after transfection, the cells were treated with CDCA (200 μM) for 24 hr, and the relative luciferase activities were determined 72 hr later. (B) A ChIP assay demonstrated the direct binding of FoxO4 to the CDX2 promoter in GES-1 cells. M: Marker. (C,D) qRT-PCR of the ChIP products validated the binding capacity of FoxO4 to the CDX2 promoter. Means ± SEM of a representative experiment (n = 3) performed in triplicate is shown. *p < .05; **p < .01; ***p < .001; ns, not significant activities may thus be dependent on its simultaneous activity on multiple molecular targets (Pirola & Frojdo, 2008). Here, active FoxO4, a common target of resveratrol and bile acids, has been the focus of our research. The Forkhead box O (FoxO) transcription factor family contains four related members: FoxO1, FoxO3, FoxO4 and FoxO6 (Daitoku, Sakamaki, & Fukamizu, 2011;Maiese, Chong, & Shang, 2008;Sykes et al., 2011). We previously discovered the role of Fox4 as a negative regulator of GC and CRC (X. Liu et al., 2011;Su et al., 2014). Phosphorylation by PKB/AKT1 inhibits transcriptional activity and is responsible for nuclear localization and accumulation (Mandai et al., 2018;F. Zhang, Virshup, & Cheong, 2018). In our study, we found that resveratrol could increase p-FoxO4 and nuclear accumulation, which inhibited CDX2 transcription by binding directly to the promoter of CDX2. Cellular immunofluorescence confirmed this result. However, other studies identified that activation of the PI3K/AKT or MAPK pathway induced FoxO4 phosphorylation and resulted in its export from the nucleus into the cytoplasm with a reduction in DNA-binding activity (Roy, Srivastava, & Shankar, 2010;Takaishi et al., 1999). These findings provide evidence that p-FoxO4 may play an essential role in promoting the progression of IM due to bile acid reflux.
Notably, PI3K/AKT and MAPK are oncogenic signalling pathways that are usually activated in various cancers and target FoxO4 in similar ways to inhibit its anti-cancer function. Similarly, isoorientin and momordin Ic induced cell death by up-regulating FoxO4, mediated by inhibition of the PI3K/AKT and MAPK pathways in human hepatoblastoma cancer cells Yuan et al., 2012). In our study, inhibition of the PI3K/AKT cascade by LY294004 decreased FoxO4 phosphorylation and nuclear accumulation, and subsequently increased CDX2 expression. When p-FoxO4 was reduced by 75% or less, the inhibitory effect on CDX2 could be ablated. Conversely, if the absolute content of p-FoxO4 is increased, CDX2 transcription will be inhibited. However, other experimental studies suggest that inhibiting the PI3K/AKT cascade by LY294002 increased FoxO4 transcription, which might be caused by impaired FoxO3 phosphorylation and thus results in increased FoxO3 activity (Franz et al., 2016). Furthermore, acute starvation or caloric restriction has been shown to trigger an increase in FoxO4 mRNA levels in skeletal muscles (Furuyama et al., 2002;Mofarrahi et al., 2014). Together with our results, these findings provide evidence that FoxO4 may be activated by resveratrol via PI3K/AKT activation.
In summary, we propose a schematic model of gastric IM development ( Figure S2). This figure illustrates that in gastric epithelial cells,