Rev-erbα inhibits proliferation by reducing glycolytic flux and pentose phosphate pathway in human gastric cancer cells

Rev-erbα is a nuclear receptor, which regulates circadian rhythm, inflammatory responses and lipid metabolism. We previously showed Rev-erbα reduction in human gastric cancer, which is associated with TMN stages and poor prognosis. We hypothesized that Rev-erbα modulates proliferation via glycolytic flux and the pentose phosphate pathway (PPP) in gastric cancer. Knockdown of Rev-erbα significantly increased proliferation as well as glycolytic flux and the PPP in human gastric cancer cells. These effects were reduced by a Rev-erbα agonist GSK4112 in a dose-dependent manner. Furthermore, Rev-erbα was recruited on the promoters of PFKFB3 and G6PD genes, thereby inhibiting their gene transcription. GSK4112 treatment reduced PFKFB3 and G6PD gene expression, which was not affected by BMAL1 knockdown. Pharmacological inhibition of glycolysis and the PPP using corresponding PFKFB3 and G6PD inhibitors attenuated Rev-erbα knockdown-induced proliferation in gastric cancer cells. GSK4112 treatment was not able to reduce proliferation in SGC-7901 overexpressing both PFKFB3 and G6PD genes. Both PFKFB3 and G6PD were overexpressed in patients with gastric cancer, and positively correlated with the TMN stages. The PPP and glycolysis were enhanced in gastric cancer tissues of patients with low expression of Rev-erbα compared to the patients with high expression of Rev-erbα. In conclusion, Rev-erbα reduction causes gastric cancer progression by augmenting the PPP and glycolysis.


Introduction
Gastric cancer is the second most common cause of cancer-related death worldwide 1,2 . Gastric cancer cells rely on aerobic glycolysis instead of oxidative phosphorylation providing bioenergetics and macromolecule synthesis for their proliferation, which is termed 'Warburg effect'. It has been shown that the levels and/or enzymatic activities of hexokinase II (HKII), phosphosphofructokinase (PFK), M2 type of pyruvate kinase (PKM2), 6-phosphofructokinase-2/ fructose-2,6-bisphosphatase (PFKFB3), and glucose-6phosphate dehydrogenase (G6PD) involved in glycolysis and pentose phosphate pathway are increased, which results in the progression and/or poor prognosis of gastric cancer [3][4][5][6][7][8][9][10] . In fact, reoperative metabolic syndrome is predictive of significant gastric cancer mortality after gastrectomy 11 . However, the mechanisms underlying these increased enzymes in gastric cancer are not clear.
Rev-erb is a nuclear receptor, which has two members, Rev-erbα and Rev-erbβ, and serves as receptor for heme. Rev-erb plays a number of diverse and important roles in regulation of circadian rhythm, inflammatory responses, senescence, and lipid in central nervous system and peripheral organs 12,13 . There are limited studies regarding the role of Rev-erb in tumors. Rev-erbα mRNA levels are significantly reduced in breast cancer cells, while Rev-erbβ mRNA expression is significantly increased 14 . Activation of Rev-erbα/β inhibits breast cancer cell proliferation, and is cytotoxic in cancer cells derived from different tumor types, namely brain, leukemia, breast, colon and melanoma 13,15 . Our recent study showed that the expression of Rev-erbα was decreased in human gastric cancer, which was associated with TMN stages and poor prognosis 16 . However, it is unknown whether Rev-erbα modulates glycolysis and subsequent proliferation in gastric cancer cells. We, therefore, hypothesize that Rev-erbα inhibits glycolysis and pentose phosphate pathway, thereby reduces proliferation in gastric cancer cells.

Cell synchronization
Once reaching 80-90% confluence, SGC-7901 cells were cultured with DMEM medium containing 50% FBS for 2 h for synchronization, which was considered as zeitgeber time 0 (ZT0) 17,18 . Cells were then kept in serum-free medium, and were harvested every 6 h. Total RNA was extracted for PFKFB3 and G6PD gene expression.

Patients and tissues collection
All samples were obtained from 74 patients with diagnosed gastric cancer who were subjected to surgical operation at the First Affiliated Hospital of Anhui Medical University in 2014, as previously described 16,22 . The study population median age was 63.4 years (range: 33-84) and the gender distribution was 58 males and 16 females. The detailed clinical characteristics were described in our previous publications 16,22 . None of the patients had received any other therapies, such as radiotherapy or medical chemotherapy before surgery. All patients' hepatic, renal and bone marrow function were normal. Eastern Cooperative Oncology Group (ECOG) performance status between 0 and 2. Patients were excluded with serious imbalance, pregnancy, and breast-feeding. Informed consent was obtained from all individual participants included in the study. The research protocol was approved by the Clinical Research Ethics Committee of Anhui Medical University. All methods were in accordance with the accurate guidelines to carry out.

EdU-incorporation cell proliferation assay
Click-iT EdU Flow Cytometry kit was purchased from Thermo Fisher Scientific, Inc. (Waltham, MA, USA). Cell proliferation was performed according to the manufacturer's instructions 23 . In brief, the cells were incubated with 5′-ethynyl-2′-deoxyuridine (EdU, 10 µM) for 2 h. Cells were harvested and mixed with 3 ml of PBS containing 1% bovine serum albumin (GE Healthcare Life Sciences), which was centrifuged at 1500 × g for 10 min at 4°C and fixed with 100 µl 4% formaldehyde for 15 min. Cells were then washed and incubated with 100 µl saponin-based permeabilization buffer for 15 min. After permeabilization, the samples were incubated with CuSO4 and Alexa Flour® 488 coupled to azide for 30 min, and measured by the flow cytometry with 50,000 events.

Glycolytic analysis
XF-24 Extracellular Flux Analyzer (Seahorse Bioscience) was used for real-time analysis of extracellular acidification rate (ECAR). Cells were seeded in Seahorse XF-24 cell culture microplates at the density of 20,000 cells/well. ECAR was recorded followed by sequential injections with 10 mM glucose, 1.0 μM oligomycin, and 50 mM 2-deoxy-D-glucose according to the manufacturer's manual 24 .

Lactate, glucose and NADPH measurement
Intracellular levels of lactate were determined using lactate assay kit (BioVision, Milpitas, CA) according to the manufacturer's instructions. Glucose levels in the supernatants were determined using Glucose Colorimetric/ Fluorometric Assay Kit (Biovision) according to the manufacturer's instructions. Glucose consumption was calculated as fold change relative to untreated controls. Intracellular NADPH levels were determined using the NADP/NADPH Quantification Kit (Biovision) according to the manufacturer's instructions.

Western blot
Cells were lysed in lysis buffer (25 mM HEPES, 2 mM MgCl2, 2 mM DTT, 1 mM EDTA, 1 mM PSMF, 5 µg/ml leupeptin, pH 7.4). Freeze-thawing the suspension liquid containing the extracted protein 3 times subsequently, the lysates were centrifuged at 10,000 rpm at 4°C for 10 min 25 . The concentration of extracted protein in supernatants was determined by the BCA assay. The protein extracts (10-20 µg) were separated by 4-12% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. Following non-specific blocking with 5% skimmed milk at room temperature for 2 h, PVDF membrane was washed 3 times with TBST (TBS contained 0.1% Tween-20) for 10 min each time. The membranes were then incubated with antibodies against REV-ERBα (Cat#. ab174309; Abcam), PFKFB3 (Cat#: ab181861, Abcam) and G6PD (Cat#: ab993, Abcam) overnight at 4°C , and then washed 3 times with TBST for 10 min each time. The membranes were incubated with the corresponding rabbit anti-goat horseradish peroxidaseconjugated secondary antibody (rabbit; cat no. AP106P; 1:5,000 dilution; EMD Millipore) for 2 h at 20°C, after that washed 3 times with TBST again for 10 min each time. The detection of the molecules of interest was carried out using enhanced chemiluminescence (Beyotime Institute of Biotechnology, Haimen, China). The bands were quantified to calculate relative protein expression levels using Quantity one software version 4.99.5.2.0 (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted from human tissues and cells using TRIzol® (Life Technologies; Thermo Fisher Scientific, Inc.). Total RNA was reverse transcribed by cDNA synthesis using a PrimeScript RT Reagent kit with gDNA Eraser (Perfect Real Time; Takara Bio, Inc., Otsu, Japan) at 37°C for 30 min and 85°C for 5 s. qPCR was operated using a 7900 Thermal Cycler (ABI, Applied Biosystems; Thermo Fisher Scientific, Inc.) with GoTaq® Green Master Mix (Promega Corporation, Madison, WI USA) at an initial denaturation at 95°C for 30 s, followed by 40 cycles of denaturation for 5 s at 95°C, annealing for 30 s at  (10). β-actin was used as an internal gene for normalization.

ChIP assay
Cells were fixed with 1% formaldehyde and terminated with 2.5 mM glycine. The scraped cells were sonicated for lysis in PBS with sodium thiosulfate. The lysates were divided into three aliquots, one of which was a positive control and received no treatment and one of which was a negative control, which was incubated with target protein.
One-third of the cell lysate was served as the test group, and was incubated with antibody against MIST1 (1:100) and Protein G PLUS-Agarose. After removal of RNA and protein, DNA was extracted with phenol-chloroform, respectively. Next, the degree of enrichment on gene promoters was detected using real-time quantitative PCR.

Statistical analyses
In vitro experiments were performed with at least three biological replicates based on our preliminary studies. Data are expressed as the mean ± standard deviation. Comparison between different groups was performed using analysis of variance. The Student-Newman-Keuls test was the post-hoc test used following analysis of variance. Data analysis was performed using SPSS 17.0 software (SPSS, Inc., Chicago, IL, USA). All reported p values were two-sided, and a value of p < 0.05 was considered statistically significant.

Rev-erbα inhibits human gastric cancer cell proliferation
To determine the role of Rev-erbα on the proliferation in human gastric cancer cells, we transfected undifferentiated (BGC-823) and moderately differentiated (SGC-7901) gastric cancer cells with Rev-erbα siRNAs (Fig. 1a). As shown in Fig. 1b, c, knockdown of Rev-erbα gene increased the proliferation of both SGC-7901 and BGC-823 cells. Treatment with a Rev-erbα agonist GSK4112 (0.5 μM and 2 μM) for 48 h reduced the proliferation of SGC-7901, BGC-823, and AGS cells in a concentrationdependent manner. Compared to SGC-7901 cells, BGC-823, and AGS cells were not sensitive to GSK4112 (0.5 μM) to reduce cell proliferation (Fig. 1b-d).
Altogether, Rev-erbα inhibits proliferation in human gastric cancer cells.

Rev-erbα reduces glycolysis in human gastric cancer cells
To determine the role of Rev-erbα on glycolysis in gastric cancer cells, we measured extracellular acidification rate (ECAR) using Seahorse XF Analyzer in SGC-7901 cells transfected with Rev-erbα siRNA. Transfection with Rev-erbα siRNA significantly reduced the protein levels of Rev-erbα (Fig. 2a). As shown in Fig. 2b, knocking down the Rev-erbα gene increased basal glycolysis and glycolytic capacity in SGC-7901 cells. Treatment with a Rev-erbα antagonist SR8278 (1 μM and 3 μM) for 48 h significantly increased intracellular levels of lactate in both SGC-7901 and BGC-823 (Fig. 2c). In contrast, treatment with a Rev-erbα agonist GSK4112 (0.5 μM and 2 μM, 48 h) reduced glucose consumption in both SGC-7901 and BGC-823 cells (Fig. 2d). Altogether, Rev-erbα reduces glycolysis in human gastric cancer cells.

Rev-erbα inhibits the expression of genes encoding ratelimiting enzymes in glycolysis
To investigate whether Rev-erbα reduces glycolysis by inhibiting the gene encoding rate-limiting enzymes, we transfected Rev-erbα siRNA into SGC-7901 cells and detected the expression of HKII, PFK1, PFK2 (PFKFB1, PFKFB2, and PFKFB3 subtype), and PK using quantitative real-time PCR. As shown in Supplementary  Fig. 1A, the mRNA levels of PFK1, PFKFB1, PFKFB2, or PK were not altered in SGC-7901 cells transfected with Rev-erbα siRNA. However, knockdown of Rev-erbα significantly increased the levels of HKII and PFKFB3 genes. Furthermore, chromatin immunoprecipitation was utilized to determine whether Rev-erbα is recruited on the promoters of these genes. It was found that Rev- erbα was recruited to the promoters of human PFKFB3 but not HKII gene ( Supplementary Fig. 1B). These results demonstrate that Rev-erbα is recruited on the promoters of PFKFB3 gene, and suppresses its expression.

Rev-erbα inhibits G6PD gene expression and NADPH generation
To investigate whether Rev-erbα regulates the PPP, we transfected Rev-erbα siRNA into SGC-7901 cells and detected the expression of G6PD gene using quantitative real-time PCR. As shown in Supplementary Fig. 2A, knockdown of Rev-erbα significantly increased the levels of G6PD gene. Furthermore, the levels of NADPH were reduced in SGC-7901 cells treated with GSK4112 (0.5 μM and 2 μM, 48 h) compared to vehicle-treated cells (Supplementary Fig. 2B). Chromatin immunoprecipitation was utilized to determine whether Rev-erbα is recruited on the promoters of this gene. It was found that Rev-erbα was recruited to the promoters of human G6PD gene (Supplementary Fig. 2C). These results demonstrate that Reverbα is recruited on the promoters of G6PD gene, and suppress the PPP.

Rev-erbα modulates PFKFB3 and G6PD gene expression independent of BMAL1
It has been shown that PFKFB3 gene expression and G6PD enzymatic activity are rhythmic in certain cells and organs 26,27 . Thus, we determined the expression of PFKFB3 and G6PD genes at different zeitgeber times. The expression of both PFKFB3 and G6PD genes revealed the presence of multiple peaks every 12 h. Therefore, the expression of both PFKFB3 and G6PD was not circadian (approximately 24-hour cycles) in SGC-7901 cells (Supplementary Figure 3A, B).
Rev-erbα has been shown to regulate BMAL1 thereby controlling circadian rhythm 28 . Hence, we determined whether Rev-erbα modulates PFKFB3 and G6PD gene expression by modulating BMAL1. As shown in Supplementary Fig. 3C, BMAL1 gene expression was reduced in SGC-7901 cells when BMAL1 siRNA was transfected. As expected, GSK4112 treatment (2 μM) significantly reduced expression of PFKFB3 and G6PD genes in SGC-7901 cells (Supplementary Fig. 3D, E). However, knockdown of BMAL1 had no effects on PFKFB3 or G6PD gene expression in SGC-7901 cells treated with GSK4112 (2 μM) ( Supplementary Fig. 3D and 13E). These results demonstrate that Rev-erbα modulates PFKFB3 and G6PD gene expression independent of BMAL1.

Rev-erbα protein inhibits PFKFB3 and G6PD gene expression depending on its DNA binding domain (DBD)
To determine whether the Rev-erbα protein binds directly to the PFKFB3 and G6PD genes, the DBD of Reverbα protein was mutated so as to detect PFKFB3 and G6PD gene expression. First, we generated Rev-erbα KO SGC-7901 cells using the CRISPR/Cas9 system (Supplementary Fig. 4A). Next, we transfected the wild-type and mutant DBD plasmids of Rev-erbα in these KO cells, and determined PFKFB3 and G6PD mRNA levels. As shown in Supplementary Fig. 4B, mutation of Rev-erbα DBD significantly increased PFKFB3 and G6PD gene expression in SGC-7901 cells. Hence, Rev-erbα binds to PFKFB3 and G6PD DNA through its DBD.

Rev-erbα inhibits gastric cancer cell proliferation by PFKFB3 and G6PD
Although Rev-erbα inhibited PFKFB3 and G6PD gene expression, it is unclear whether PFKFB3 and G6PD mediate the effect of Rev-erbα on proliferation of human gastric cancer cells. First, Rev-erbα siRNA was transfected into SGC-7901 cells and treated them with PFKFB3 and G6PD enzyme inhibitors 3-PO (25 μM) and DHEA (250 μM) for 48 h, respectively. Click-iT EdU cell proliferation assay and flow cytometry were used to detect the incorporation of EdU into DNA. As shown in Fig. 3a, both 3-PO and DHEA significantly reduced Rev-erbα knockdown-induced proliferation in SGC-7901 cells. Combined 3-PO and DHEA treatments further reduced proliferation induced by Rev-erbα knockdown in SGC-7901 cells compared to treatment alone. In contrast, overexpression of PFKFB3 and G6PD increased proliferation in SGC-7901 cells (Fig. 3b). Treatments with a Rev-erbα agonist GSK4112 (2 μM, 48 h) significantly reduced PFKFB3 and G6PD overexpression-induced proliferation in SGC-7901 cells (Fig. 3b). However, GSK4112 treatment did not affect the proliferation of SGC-7901 cells induced by dual overexpression of PFKFB3 and G6PD (Fig. 3b). These results suggest that Rev-erbα inhibits proliferation of human gastric cancer cells via downstream PFKFB3 and G6PD signaling.

Glycolysis and the PPP are enhanced in patients with low expression of Rev-erbα in gastric cancer tissues
To extrapolate the in vitro findings into human samples, we determine the expression of PFKFB3 and G6PD as well as the metabolites, including lactate, pyruvate and NADPH in patients with gastric cancer. As shown in Supplementary Fig. 5A, the protein abundance of PFKFB3 and G6PD was increased in gastric cancer tissues, which was associated with increased TNM stage. This was confirmed by immunohistochemistry ( Supplementary Fig. 5B).
We next grouped the patients with gastric cancer into low and high expression of Rev-erbα, as we previously described 16 . We found that the mRNA levels of PFKFB3 and G6PD were augmented in patients with low abundance of Rev-erbα (Fig. 4a). Furthermore, the levels of lactate and NADPH were also increased in patients with low expression of Rev-erbα compared to the patients with high expression of Rev-erbα (Fig. 4b). The levels of pyruvate were not changed in patients with gastric cancer between low and high expression of Rev-erbα (Fig. 4c). Altogether, these results suggest that low expression of Rev-erbα associates with increased glycolysis and the PPP in gastric cancer.

Discussion
We have shown that Rev-erbα was reduced in human gastric cancer, which was associated with poor differentiation, TMN stages and poor prognosis 16 . However, the mechanisms underlying these findings are not clear. In the present study, we found that Rev-erbα inhibits glycolysis and the PPP, thereby reducing proliferation in gastric cancer cells. This was due to increased expression of PFKFB3 and G6PD in human gastric cancer cells. Moreover, reduced Rev-erbα in patients with gastric cancer correlated with increased glycolysis and the PPP in cancer tissues. These findings provide potential therapeutic approaches by activating Rev-erbα to inhibit the proliferation of gastric cancer cells.
Rev-erbα is highly expressed in adipose tissue, skeletal muscle and brain, which plays important roles in lipid metabolism, adipocyte differentiation, inflammatory response, and circadian rhythm 28,29 . Although heme, a ligand of Rev-erbα, inhibits the expression of phosphoenolpyruvate carboxykinase involved in gluconeogenesis 30 , there are no reports regarding the role of Rev-erbα in glycolysis and the PPP. There are three rate-limiting enzymes, including HKII, PFK, and PK, in glycolysis and a rate-limiting enzyme G6PD in the PPP. We found that Rev-erbα could be recruited on the promoters of PFKFB3, an isoform of PFK, and G6PD genes, and directly inhibited their gene transcription in human gastric cancer cells. Along with the data on ECAR, lactate, and glucose consumption, Rev-erbα suppresses glycolytic flux and the PPP in gastric cancer cells by inhibiting PFKFB3 and G6PD gene expression. There are two pathways for Reverbα-mediated transcription repression. Once in nucleus, Rev-erbα competes with the retinoic acid-associated orphan receptor alpha (RORα) to bind with the ROR response element (RORE) of targeting genes, such as circadian gene Bmal1, inhibiting their transcription 28,31,32 . However, we found that knockdown of Bmal1 did not influence PFKFB3 or G6PD gene expression when cells were treated with GSK4112. These findings suggest that BMAL1 is not involved in Rev-erbα-mediated modulation on PFKFB3 or G6PD gene expression in gastric cancer cells. In the present study, we showed that PFKFB3 and G6PD genes did not exhibit rhythmic expression. This is not in agreement with the finding that the expression of these genes is rhythmic in tongue cancer cells 26,27 . This discrepancy may be due to cell-specific (gastric vs tongue cancer cells). The second pathway is that that Rev-erbα inhibits gene transcription by directly binding to target response elements (RevREs) and by recruiting the corepressor N-CoR and histone deacetylases on targeting gene promoters 33 . It remains elusive whether Rev-erbα inhibits PFKFB3 and G6PD gene expression using similar mechanisms in human gastric cancer cells.
It has been shown that Rev-erbα binds to the genome using both DBD-dependent and DBD-independent mechanisms 21 . We found that mutation of Rev-erbα DBD domain increased the mRNA slevels of PFKFB3 and G6PD in human gastric cancer cell. This suggest that Reverbα relies on its DBD, thereby inhibiting PFKFB3 and G6PD gene expression. Further study is required to determine on the mechanisms of increased HKII gene expression once Rev-erbα was knockdown, despite Reverbα was not recruited on HKII gene promoters.
The glycolysis and its parallel shunt PPP not only generate bioenergetics for meeting the demands of malignant cell growth, but also provide a variety of biosynthesis of macromolecules for their anabolism. This was corroborated by our findings that pharmacological inhibition of glycolysis and the PPP reduced proliferation in gastric cancer cells, whereas overexpression of PFKFB3 and G6PD augmented proliferation in these cells. In addition, inhibition of PFKFB3 and G6PD significantly attenuated Rev-erbα knockdown-induced increase in proliferation. These finding suggest that Rev-erbα inhibits proliferation in gastric cancer cells by suppressing PFKFB3 and G6PD gene expression. It is interesting to note that GSK4412mediated inhibition of proliferation was abolished in dual PFKFB3 and G6PD overexpressing cells but not in individual overexpressing cells. This suggests that both PFKFB3 and G6PD are required for proliferation when Rev-erbα was reduced in human gastric cancer cells.
Utilizing clinical human samples, we found that the protein abundance of PFKFB3 and G6PD was increased in patients with gastric cancer. Furthermore, this was Fig. 4 Glycolysis and the PPP are increased in human gastric cancer tissues with low Rev-erbα expression. G6PD and PFKFB3 mRNAs a, NADPH b, lactate c, and pyruvate d were measured in human gastric cancer tissues based on high and low Rev-erbα expression. Data are represented as the mean ± SD. N = 6-15. ***P < 0.001 versus group with high Rev-erbα expression associated with increased TNM stage of human gastric cancer. This is corroborated by the findings that both PFKFB3 and G6PD are overexpressed in patients with gastric cancer, and promoted the proliferation and migration of gastric cancer cells 3,7 . It is interesting to note that the PPP and glycolysis were enhanced in human gastric cancer with low expression of Rev-erbα compared to gastric cancer tissues with high expression of Rev-erbα. These findings further conformed that Rev-erbα is reduced in gastric cancer, which leads to proliferation by promoting the PPP and glycolysis in gastric cancer cells.
In summary, we demonstrated for the first time that Rev-erbα was recruited to promoters of PFKFB3 and G6PD genes, thereby inhibiting glycolytic flux and the PPP, and subsequent proliferation in in vitro human gastric cancer cells. Furthermore, the PPP and glycolysis were enhanced in gastric cancer patients with low expression of Rev-erbα compared to gastric cancer tissues with high expression of Rev-erbα. Therefore, our findings provide a new biomarker and promising therapeutic strategy for the treatment of gastric carcinoma by targeting Rev-erbα-mediated metabolic reprogramming.