Restoration of FBP1 suppressed Snail-induced epithelial to mesenchymal transition in hepatocellular carcinoma

Fructose-1,6-bisphosphatase (FBP1), one of the rate-limiting gluconeogenic enzymes, plays critical roles in several cancers and is treated as a tumour suppressor. However, its role in hepatocellular carcinoma (HCC) is unclear. Here, we demonstrated that FBP1 was significantly inhibited during Snail-induced epithelial to mesenchymal transition (EMT) and tissues in HCC. Restoration of FBP1 expression in HCC cancer cells suppressed EMT phenotype, tumour migration and tumour growth induced by Snail overexpression in SMMC-7721 cells. Gene set enrichment analyses revealed significantly enriched terms, including WNT, Notch, ESC, CSR and PDGF, in the group with high Snail and low FBP1 compared with those with low Snail and high FBP1. Low FBP1 expression was significantly correlated with higher AFP level, satellite nodules, portal vein tumour thrombus, and advanced tumour stage. Survival analyses showed that FBP1 was an independent prognostic factor for overall survival and recurrence-free survival. In conclusion, our study revealed a vital role for FBP1 in Snail-induced EMT and prognostic prediction in HCC.


Introduction
Hepatocellular carcinoma (HCC) remains a global public health issue. It is ranked as the fifth leading cancer and the second leading cause of cancer-related mortality 1 . The long-term survival is far from satisfying due to the low rate of curative treatment and the high rate of postcurative recurrence. There is always an urgent demand for a better understanding of the molecular mechanisms underlying HCC 2 .
The ability of differentiated epithelial cells to acquire mesenchymal traits during embryonic development, wound healing, malignant tumour progression and chemoresistance is now termed epithelial-mesenchymal transition (EMT). Via EMT, cells acquire mesenchymal properties, such as expression of Vimentin, but at the same time lose the expression of epithelial markers, such as E-cadherin 3 . Snail is one of the most important transcription factors that drive EMT 4,5 . However, little is known about the role of aberrant glucose metabolism in Snail-induced EMT.
Glucose homeostasis is reciprocally controlled by the catabolic glycolysis/oxidative phosphorylation (OXPHOS) and the anabolic gluconeogenesis pathway. Aberrant glucose metabolism promotes tumourigenesis and progression in many cancers 6 . As first described in 1920s, some tumour cells preferentially rely on glycolysis rather than OXPHOS, even in conditions with ample oxygen ("aerobic glycolysis" or "Warburg effect") 7 . While previous studies have paid much attention to abnormal glycolysis, little effort has been made to understand the role of gluconeogenesis, the reciprocal metabolic process of glycolysis, in cancer. Fructose-1,6-bisphosphatase (FBP1), one of the rate-limiting enzymes in gluconeogenesis, catalyses the hydrolysis of fructose-1,6-bisphosphate (F-1,6-P2) to fructose 6-phosphate (F-6-P) and inorganic phosphate. Recently, FBP1 was reported to play suppressive-suppressive roles in several cancers, including renal cancer 8 , breast cancer 9 , lung cancer 10 , pancreatic cancer 11,12 and gastric cancer 13 . Until now, the most important mechanism underlying FBP1 suppression was reported to be promoter DNA methylation 9,10,14,15 . However, the role and mechanism of dysregulated FBP1 in HCC remain far from clear. We conducted this study to clarify the role of FBP1 in Snail-induced EMT and the prognostic role of FBP1 in HCC.

Materials and methods
Clinical specimens, tissue microarray and immunohistochemistry All patients were enrolled from the First Affiliated Hospital of Sun Yat-sen University between January 2006 and December 2009. The median follow-up for the 242 patients was 31.0 months (range, 3-95 months). Patients were staged according to the seventh edition of the International Union Against Cancer TNM classification system. All protocols were approved by the Ethics and Indications Committee of the First Affiliated Hospital of Sun Yat-sen University. Written informed consent was obtained from all patients.
A tissue microarray (TMA) of 242 pairs of HCCs and corresponding peritumoural tissues was constructed. The tissue samples were incubated with anti-Snail (Abcam, #ab180714, 1:100), anti-FBP1 (Abcam, #ab180714, 1:200) and E-cadherin (Cell Signaling Technology, #9782, 1:400) antibodies overnight at 4°C. Negative controls were performed without primary antibodies. Three representative fields were imaged in a uniform setting for all slides. Image-Pro Plus v6.2 software (Media Cybernetics Inc., Bethesda, MD) was used to measure the density of positive staining. The median values were defined as the cutoff values for high and low Snail or FBP1 expression.

Transwell migration assay
A total of 5 × 10 4 cells/well in serum-free DMEM were seeded into the upper chamber of an 8-μm transwell chamber (Beckton Dickinson, Franklin Lakes, NJ, USA); DMEM with 10% bovine serum albumin (BSA) was added in the lower chamber. After 24-h incubation at 37°C, the cells in the upper chamber were fixed in methanol and then stained with Giemsa solution (Beyotime, Nantong, China). Then, the migrated cells were imaged and quantified.

CCK8 assay
A total of 1 × 10 5 cells/well were cultured in 96-well plates. The CCK8 reagent was incubated into each well for 3 h at 37°C and after 24 h, 48 h, 72 h and 96 h, the Cell Counting Kit-8 was used (Dojindo, Japan). The absorbency of cells at 450 nm was measured using the BioTek Epoch automatic enzyme-labelled instrument (Biotek, USA).

Immunofluorescence assay
Cells were cultured in six-well plates. After incubation for 48 h, cells were washed with phosphate-buffered saline, fixed with 4% paraformaldehyde, blocked with 5% BSA, stained with E-cadherin and Vimentin, and then stained with Fluorescein isothiocyanate (FITC)-labelled secondary antibody and 4',6-diamidino-2-phenylindole (DAPI) as per the manufacturer's instructions (Beyotime, Nantong, China). The images were captured using fluorescence microscope.  h Western blot analysis showed Snail overexpression induced Vimentin but suppressed E-cadherin and FBP1 protein in SMMC-7721. *P < 0.05, **P < 0.01, NS not significant, compared with control. MW molecular weight. All data are based on three independent repeats

Orthotopic transplantation HCC mouse models
Male BALB/c nude mice, aged 4-6 weeks, were used. A total of 2 × 10 6 tumour cells were injected into subcutaneous regions of nude mice. Subcutaneous tumour tissues with a longitudinal diameter of 1 cm were harvested and were cut into approximately 1 mm 3 pieces. Then, tumour pieces were transplanted into the left hepatic lobe of the mice. All mice were sacrificed 6 weeks later and analysed. All mice were obtained from Shanghai Institute of Material Medicine and maintained in a pathogen-free environment. Animal care and experimental protocols were in accordance with the guidelines established by Shanghai Medical Experimental Animal Care Commission and were in accordance with regulations for the Administration of Affairs Concerning Experimental Animals and National Institutes of Health Guidelines. GSEA v3.0 (http://www.broadinstitute.org/gsea/, RRID: SCR_003199) 21,22 was used to find terms predicted to be enriched in C2 (the Kyoto Encyclopaedia of Genes and Genomes pathway, KEGG), in C5 (a gene set that contained genes annotated by the same Gene Ontology (GO) term), and in C6 (oncogenic signatures of gene sets that represent the signatures of cellular pathways that are often dysregulated in cancer). Data from non-HCC patients were excluded. P < 0.05 and false discovery rate (FDR) q-value < 0.25 were considered statistically significant.

Statistical analysis
Statistical analyses were performed using SPSS V19.0 (SPSS Inc., USA), R software V3.5.1 (R Foundation for Statistical Computing, Vienna, Austria) and presented using GraphPad Prism v7.00 (GraphPad Software Inc., . Qualitative variables were analysed using the Pearson χ 2 test or Fisher's exact test. Quantitative variables were compared using t-test, Pearson's correlation test or Spearman rank correlation analysis as appropriate. The survival analyses were plotted using the Kaplan-Meier method and were compared using the logrank test. Multivariate analyses were performed using the Cox regression model method with forward stepwise procedure. P < 0.05 was considered statistically significant.
Next, we tested FBP1 following Snail overexpression. As shown in Fig. 1g, h, Snail upregulation significantly inhibited FBP1 in SMMC-7721 cells. Taken together, these data show that overexpression of Snail expression promoted migration, induced EMT and suppressed FBP1 in HCC. Similar results were found in the Hep3B cell line (Supplemental Fig. 1).

Expression of FBP1 and EMT markers in multiple HCC cell lines
We analysed FBP1 expression by qRT-PCR and western blot in the five HCC cell lines MHCC-97H, MHCC-97L, HepG2, Hep3B and SMMC-7721. We observed that MHCC-97H cells with the highest Snail and Vimentin expressed the lowest FBP1 and E-cadherin, but SMMC-7721 cells with the lowest Snail and Vimentin significantly expressed FBP1 and E-cadherin (Fig. 2a, b).

GSEA of TCGA dataset
To further understand the underlying molecular mechanisms of dysregulated FBP1 and Snail, we performed GSEA analyses using data from TCGA LIHC dataset. The expression data of 24,991 genes from 361 HCC patients were included in our analyses. We aimed to identify terms differing between the high (top 10%) and low (bottom 10%) expressing groups of FBP1 and Snail. As FBP1 was suppressed by Snail expression in HCC, we explored the enriched terms overlapping in the high Snail and low FBP1 groups.
GO terms such as anchoring junction, apoptotic signalling pathway, cell ageing, cytoplasmic region, cell substrate junction, enhancer binding, FC receptor signalling pathway, hippo signalling, positive regulation of cytoplasmic transport, positive regulation of proteolysis, protein dephosphorylation, regulation of defence response to virus, Ras protein signal transduction, KEGG pathways involving WNT signalling pathway, Notch signalling pathway and pathogenic Escherichia coli infection were significantly enriched in the high Snail or low FBP1 group when compared with the low Snail or high FBP1 group, respectively. Moreover, oncological signatures (see figure on previous page) Fig. 5 The expression of FBP1 and prognostic role in HCC patients. a The expression of FBP1 mRNA was suppressed in HCC when compared with non-tumour controls using data from GSE14520, GSE54236, GSE25097 and TCGA HCC cases (T = tumour tissue, N = non-tumour tissue). b The representative images of FBP1 expression in HCC tissues. The magnifications used were ×100 (the upper four) and ×200 (the lower four). c Kaplan-Meier survival curves showed significantly better recurrence-free survival and overall survival in patients with high FBP1 expression when compared with those with low FBP1 expression in our cohort. d Kaplan-Meier survival curves showed significantly better recurrence-free survival and overall survival in patients with high FBP1 expression when compared with those with low FBP1 expression in the GSE14520 cohort. ***P < 0.001, ****P < 0.0001, when compared with control including Early serum response (CSR), Dichloroacetate (DCA), Embryonic stem cells (ESC) and Platelet-derived growth factor (PDGF) were found to be significantly associated with high Snail, as well as low FBP1 (Fig. 4, Supplemental Table 1).

FBP1 was suppressed in HCC patients and indicated prognosis
We analysed the mRNA expression profile of FBP1 in human HCC using data available in the GEO database, including GSE14520, GSE54236, GSE25097 and TCGA HCC. FBP1 mRNA was found to be significantly suppressed in HCC when compared with non-tumour controls (Fig. 5a).

Prognostic role of FBP1 and Snail
We investigated the expression of Snail mRNA in GSE14520, GSE54236, GSE25097 and TCGA HCC datasets. As shown in Fig. 6a, Snail mRNA was lower in tumour than non-tumour tissue, and no significant correlation between FBP1 and Snail mRNA was found. We then examined the expression of Snail in our cohort (Fig. 6b). An almost but not quite significant correlation between FBP1 and Snail mRNA was found (R = -0.091, Pearson correlation P = 0.079). Patients were further divided into four groups: Snail high FBP1 high , Snail low FB-P1 high , Snail high FBP1 low and Snail low FBP1 low . We found that the prognosis of patients in the Snail low FBP1 high group tended to be the best, whereas the prognosis of patients in the Snail high FBP1 low tended to be the worst among the four groups (Fig. 6c, d). Similar results were found in the GSE14520 cohort (Fig. 6e, f).

Discussion
Mounting evidence shows the critical role of aerobic glycolysis in tumourigenesis and progression of cancer 9,13 . In contrast, few studies directly focus on the role of aberrant gluconeogenesis in cancers. Here, we observed suppression of FBP1 during Snail-induced EMT in HCC. We revealed that forced expression of FBP1 repressed the EMT phenotype, HCC migration and tumour growth induced by Snail overexpression. We elucidated several significantly enriched terms, KEGG pathways and oncological signatures in the high Snail and low FBP1 group when compared with the low Snail and high FBP1 group. Last but not least, FBP1 expression was found to be an independent prognostic factor for RFS and OS in HCC.
FBP1 dysregulation provided metabolic advantages and promoted cancer progression in various cancers, including renal cancer 8 , breast cancer 9 , HCC 23,24 , lung cancer 10 , pancreatic cancer 11,12 and gastric cancer 13 . For example, Li et al. reported that FBP1 hindered renal carcinoma progression by inhibiting Hypoxia-inducible factor (HIF)-1α (HIF1α) activity in the nucleus via a direct interaction in an enzyme-activity-independent manner 8 . FBP1 suppression resulted in an increased cancer stem cell (CSC)-like phenotype and tumourigenesis by enhancing the interaction of β-catenin with T-cell factor in basal-like breast cancer cells 9 . FBP1 overcame gemcitabine resistance and inhibited extracellular regulated protein kinases (ERK) activation by blocking IQ motif containing GTPase-activating protein 1 (IQGAP1)-mitogen-activated protein kinase (MAPK) interaction in pancreatic cancer cells 12 . Low FBP1 expression was found to be an independent factor for poorer survival, which were consistent with previous Promoter DNA methylation is the best understood mechanism for FBP1 loss in cancer. The Rat Sarcoma (RAS)/NF-κB: nuclear factor-kappa B (NF-κB) pathway promoted DNA methylation of FBP1 in gastric cancer cells 14 . The Snail-G9a-Dnmt1 complex, which is critical for E-cadherin promoter silencing, was also responsible for the promoter methylation of FBP1 in basal-like breast cancer cells 9 . ZEB1 interacted with the FBP1 promoter to enhance DNA methylation in lung cancer cells 10 . NPM1 bound directly to the FBP1 promoter region to suppress the expression of FBP1 in pancreatic cancer cells 11 . In addition, copy number loss 8,24 and post-translational ubiquitin-mediated degradation 28 were also found to be associated with FBP1 loss. Thus, a complicated and multilayer regulatory network exists for the suppression of FBP1 in cancer, which remains to be clarified.
EMT promotes malignant tumour progression and chemoresistance. As a well-known major driver of EMT, Snail expression has been reported to be correlated with cancer metastasis and poorer survival 29 . Our findings indicated the critical role of FBP1 in Snail-induced EMT and cancer progression in HCC, which is similar to that in breast cancer 9 and gastric cancer 30 . Furthermore, GSEA indicated several enriched pathways deserving to be explored to explain the underlying mechanism of dysregulated FBP1 suppression. For example, the ERK and WNT pathways have been demonstrated to be involved in FBP1 regulation in cancers 10,12 . However, much effort should be made to validate these findings from GSEA.
Several questions remain to be further investigated. First, it is a limitation that we lack biochemical evidence showing that our ectopic FBP1 was enzymatically active. Whether FBP1 enzymatic activity or any other enzymeactivity-independent mechanism plays critical role in Snail-induced EMT remained to be clarified. Second, we found that FBP1 did not affect the expression of Snail, raising the question of how FBP1 suppressed the transcriptional activity of Snail. Two different mechanisms might help explain the phenomenon. One possibility is that FBP1 participates in a Snail subcellular location mechanism, such as nuclear transport, and then modulates Snail stability 31 . Another possibility is that FBP1 directly inhibits the Snail functional domain in the nucleus, similar to HIF in renal cancer 8 . Clarifying the subcellular location of FBP1 in Snail-induced EMT is critical to understanding the detailed interaction of Snail and FBP1. Last but not least, we found no significant correlation between FBP1 and Snail mRNA in HCC patients. The results must be explained with great caution. The difference between in vivo and in vitro experiments might be due to the underlying baseline characteristics (HCC aetiology, patient ethnicity, relatedrelated characteristics) and need to be further validated.

Conclusions
In summary, our study demonstrated that FBP1 was suppressed in Snail-induced EMT and HCC patients. Ectopic FBP1 expression hindered EMT and tumour growth induced by Snail overexpression. Several pathways, such as ESC, CSR and PDGF, could be further explored to explain the underlying mechanisms of FBP1 dysregulation. Loss of FBP1 indicated poorer prognosis and was an independent prognostic factor of HCC. Our study revealed an important role for FBP1 in Snailinduced EMT and prognostic prediction in HCC.