High glucose levels boost the aggressiveness of highly metastatic cholangiocarcinoma cells via O-GlcNAcylation

Increased glucose utilization is a feature of cancer cells to support cell survival, proliferation, and metastasis. An association between diabetes mellitus and cancer progression was previously demonstrated in cancers including cholangiocarcinoma (CCA). This study was aimed to determine the effects of high glucose on protein O-GlcNAcylation and metastatic potentials of CCA cells. Two pairs each of the parental low metastatic and highly metastatic CCA sublines were cultured in normal (5.6 mM) or high (25 mM) glucose media. The migration and invasion abilities were determined and underlying mechanisms were explored. Results revealed that high glucose promoted migration and invasion of CCA cells that were more pronounced in the highly metastatic sublines. Concomitantly, high glucose increased global O-GlcNAcylated proteins, the expressions of vimentin, hexokinase, glucosamine-fructose-6-phosphate amidotransferase (GFAT) and O-GlcNAc transferase of CCA cells. The glucose level that promoted migration/invasion was shown to be potentiated by the induction of GFAT, O-GlcNAcylation and an increase of O-GlcNAcylated vimentin and vimentin expression. Treatment with a GFAT inhibitor reduced global O-GlcNAcylated proteins, vimentin expression, and alleviated cell migration. Altogether, these results suggested the role of high glucose enhanced CCA metastasis via modulation of O-GlcNAcylation, through the expressions of GFAT and vimentin.

Scientific RepoRts | 7:43842 | DOI: 10.1038/srep43842 (UDP-GlcNAc), a substrate for glycosylation, e.g., O-GlcNAcylation. The rate of HBP can be regulated by the concentrations of the substrates, such as glucose and GlcNAc, or controlled by an expression of the rate limiting enzyme; glucosamine-fructose-6-phosphate amidotransfrase (GFAT) 14,15 . Increasing glucose uptake may promote glucose flux through HBP and subsequently increase O-GlcNAcylation. The association between an elevation of global O-GlcNAcylated proteins and tumor progression has been reported 16 . The present authors previously showed that OGT is over-expressed in CCA tissues and increased OGT is correlated with shorter survival of CCA patients 17 . Moreover, knockdown of OGT alleviates the migration/invasion of CCA cells via suppression of NF-κ B nuclear translocation 18 . Nevertheless, the mechanisms by which glucose promotes O-GlcNAcylation and CCA progression remain unclear.
The present study was designed to test the crucial role of high glucose in promoting CCA cell migration/invasion, which, in fact, was found to be more pronounced in the highly metastatic cells. The tests in this study were further designed to indicate if the association between high glucose and HBP activation in CCA cells does occur, which would then subsequently increase O-GlcNAcylation and expression of vimentin, leading to the increased motility of cells. Taken together, the present study shows for the first time in the results, the implications of high glucose on HBP-modulated O-GlcNAcylation and aggressiveness of CCA cells. The findings from this study, not only fulfill the understanding of hyperglycemic conditions promoting CCA progression, but also suggest the possible use of GFAT as a new therapeutic target for CCA treatment.

Results
High glucose promoted migration, invasion, and epithelial-mesenchymal transition (EMT) of CCA cell lines. Two pairs of CCA cells with different metastatic potentials, the parental low metastatic cells, KKU-213 and KKU-214, and the highly metastatic cells designated as L5, KKU-213L5 and KKU-214L5, cultured in normal and high glucose DMEM, were used to investigate the migration and invasion abilities using the Boyden chamber assay. As shown in Fig. 1A, the migrated cell numbers of L5 of both cell lines were significantly higher than those of their parental cells in both normal and high glucose conditions. In addition, high glucose potentiated the migration ability of both parental and L5 cells. High glucose, however, had a more pronounced effect on the L5 cells; approximately 2-3 fold for the L5 cells and 1.5 fold for the parental cells. Similar results were observed for the invasion abilities (Fig. 1B).
Since the effect of high glucose was more obvious for the migration, therefore, the expressions of EMT markers were elucidated. As shown in Fig. 1C, by giving the expression of EMT markers from the parental cells in normal glucose reference as 1, the expressions of epithelial marker, E-cadherin, and the mesenchymal marker, slug, of the parental and L5 cells seemed to be unaltered in the high glucose condition. The expression levels of mesenchymal markers, β -catenin and vimentin, of KKU-213, KKU-214, and their L5 sublines, however, were increased in the high glucose condition. The expression levels of vimentin were of interest because these corresponded with the glucose levels and were consistently observed in both cell lines. Vimentin was increased, 2 fold in KKU-213L5 and 1.6 fold in KKU-214L5.

High glucose increased O-GlcNAcylation in L5 cells more than in the parental cells.
To reveal the effect of glucose on O-GlcNAcylation, the O-GlcNAcylated protein levels of the parental and L5 cells cultured in the normal and high glucose conditions were examined. The global O-GlcNAcylated proteins of the L5 cells were higher than those of the corresponding parental cells in both normal and high glucose conditions ( Fig. 2A,B). Moreover, the effect of glucose-promoting O-GlcNAcylation was more prominent in the L5 cells than the parental cells; 2.5 fold in KKU-213L5 and 1.5 fold in KKU-214L5 (Fig. 2B).    GFAT expression in CCA tissues of patients was correlated with O-GlcNAcylated protein level. Because GFAT expression had a positive relationship with the O-GlcNAcylated protein level of CCA cells, whether or not this association existed in CCA tissues of the patients was next examined. The levels of GFAT and O-GlcNAcylated proteins were investigated in CCA tissues using immunohistochemistry (IHC) and quantitated as IHC scores. Expressions of GFAT and O-GlcNAcylated proteins were divided into low-and high-expressed groups using the median IHC scores. The positive correlations between GFAT expression and O-GlcNAcylated protein levels were observed ( Fig. 6A and B, Fisher's Exact test, P = 0.009). CCA patients (37.7%), who had high GFAT expression in tumor tissues, also had high O-GlcNAcylated protein levels and patients (40%), whose tissues had low GFAT expression, also had low levels of O-GlcNAcylated protein. In addition, expression levels of GFAT from the patients with low O-GlcNAcylated proteins were significantly lower than those with high O-GlcNAcylated proteins (Fig. 6C, Mann-Whitney test, P = 0.0085).

Discussion
EMT is an important step in cancer metastasis. Several studies reported the significance of glucose in promoting EMT and metastasis of cancer. Currently, the linkage of high glucose enhancing metastasis of CCA has been emphasized 12 . In this study, it was demonstrated for the first time that high glucose had a more pronounced effect on the highly metastatic CCA cells than the low metastatic counterparts. The molecular mechanism was shown to Several cancer cells increase glucose uptake to provide energy and metabolic intermediates to support cell survival, proliferation, and metastasis 1 . In this study, high glucose induced metastatic phenotypes of the parental and the highly metastatic derivatives, L5 cells. The highly metastatic cells seemed to respond to high glucose more aggressively than the lower metastatic parental cells. This observation was concomitant with higher Addition or removal of O-GlcNAc modulates protein functions in many ways, e.g., regulating protein phosphorylation, altering the stability/localization of proteins, and modulating protein-protein interactions 15,24 . In the present study, promoting cell migration of the highly metastatic L5 cells by high glucose was concomitantly observed with the induction of vimentin O-GlcNAcylation and with a slower decline of vimentin level in the presence of cycloheximide. Hence, the O-GlcNAcylation of vimentin may promote protein stability and consequently stimulate cell migration. Under high glucose stimulation, over-expression of vimentin was observed in the L5 cells. High glucose promoted O-GlcNAcylation and the stability of vimentin was demonstrated by this study. The significance of O-GlcNAcylation on the stability of proteins, for instance, c-Myc and FOXM1, has been reported 25,26 . In the present study, the stability of vimentin in L5 cells under high glucose was prolonged when compares to those under normal glucose. The O-GlcNAcylation of vimentin was previously reported using mass spectrometry 27 and immunoprecipitation 28 , however, the effects of glucose on O-GlcNAcylation and stability of vimentin were reported for the first time in this study.
Glycolysis, HBP and O-GlcNAc cycling directly and indirectly modulate protein O-GlcNAcylation (Fig. 7). The molecular linkage between high glucose and the expressions of key enzymes in these pathways in CCA cells It is worth mentioning here that high glucose stimulated the expression of OGA in the parental cells of both cell lines. Under the stress of high glucose, the parental cells could possibly respond to reduce the O-GlcNAcylation by increasing OGA; in contrast, the highly metastatic cells were prone to increase O-GlcNAcylation by increasing the expressions of GFAT and OGT. The molecular mechanism by which the parental cells and the highly metastatic cells regulate cellular O-GlcNAcylation is of interest but it is, however, beyond the scope of this study.
The association of OGT and migration of CCA cells was reported previously 18 . In the present study, the role of GFAT in the increase of O-GlcNAcylation and migration in L5 cells under high glucose condition was shown to be significant. Suppression of GFAT action using DON, a GFAT inhibitor, significantly reduced O-GlcNAcylation and migration of L5 cells in both CCA cell lines. As a consequence, vimentin was remarkably reduced in DON treated cells. The use of DON for cancer treatment has been reported in a series of experiments. DON could inhibit cancer growth and systemic metastasis in the VM-M3 murine model 30 . In breast (MCF7) and colorectal (HT29) cancer cell lines, DON and azaserine treatments reduced cell proliferation via the inhibition of O-GlcNAcylation and the reduced stability of β -catenin 31 . Similar effects were reported in endometrial cancer cells, AN3CA and HEC-1-B 32 . Nonetheless, the effect of glucose induced β -catenin expression was not observed in CCA cells in the current study. The association of GFAT and O-GlcNAcylation in CCA patients' tissues was further highlighted herein. High expression of GFAT was correlated with high O-GlcNAcylated protein levels in tumor tissues of CCA patients. All the evidence in the preclinical studies suggested GFAT is a potential target for CCA treatment. Both DON and azaserine, however, have been evaluated in clinical trials and showed adverse effects with variable degrees of gastrointestinal toxicity, myelo-suppression, and neurotoxicity 33,34 . A new potent GFAT inhibitor with less toxicity is challenging for cancer therapy. The in vivo studies using a GFAT inhibitor on growth and metastasis of CCA should be explored.
As summarized in Fig. 7, high glucose induced progression of CCA cells was prominent in the highly metastatic cells. The mechanism was demonstrated partly via the increase of cellular O-GlcNAcylation. GFAT was demonstrated to play a significant role in high glucose induced O-GlcNAcylation and cell migration particularly in the highly metastatic L5 cells. Vimentin was one glucose induced protein in which the stability of this protein may be prolonged, perhaps via GFAT and O-GlcNAcylation. These processes might finally potentiate EMT change and increase cell motility. The association of GFAT and O-GlcNAcylated protein was emphasized not only in the in vitro study but also in the tumor tissues of CCA patients. Altogether, the adverse effects of high glucose on aggressiveness of CCA cells, especially on the highly metastatic cells, was emphasized. The potential role of GFAT as a new therapeutic target, especially for advanced CCA, is suggested. Attempts to control blood glucose and/or reducing GFAT action in CCA patients may minimize the effects of high glucose on the aggressiveness of CCA and may be of benefit in the treatment of CCA patients with diabetes mellitus. The research for a new GFAT inhibitor with safer implications requires urgent attention. The migrated cells were stained by 0.4% sulforhodamine B in 0.1% acetic acid. The migrated cell numbers were determined from at least 5 microscopic fields (100× magnification). For the invasion assay, the transwell was coated with 0.4 mg/ml Matrigel TM (BD biosciences) and the experiment was performed as before for the migration assay. All samples were done in duplicate.

Reagents
Immunoprecipitation. Immunoprecipitation was performed as previously described 18 . Briefly, total cell lysates were incubated with anti-O-GlcNAc, anti-vimentin, or control IgG. The antibody was immobilized using protein-G sepharose beads (GE Healthcare, Buckinghamshire, UK). After centrifugation, precipitated-beads were washed, eluted, and subjected to SDS-PAGE and western blotting.
Western blot. CCA cells were lysed in NP-40 lysis buffer 18 . Protein concentration was determined using the Bradford reagent (Bio-rad laboratories, Hercules, CA). Protein lysate was separated on SDS-PAGE and transferred onto a PVDF membrane. The immunoreactivity was detected using the ECL TM Prime Western Blotting Scientific RepoRts | 7:43842 | DOI: 10.1038/srep43842 Detection System (GE Healthcare). The signals were captured by the ImageQuant ™ LAS 4000 mini-image analyzer and the signal intensities were analyzed by ImageQuant ™ TL analysis software (GE Healthcare).
Statistical analysis. The statistical significance was determined by Student's t-test using GraphPad Prism ® 5.0 software (GraphPad software, Inc., La Jolla, CA). The correlation between the expression levels of GFAT and O-GlcNAcylated proteins in CCA tissues was analyzed using Fisher's Exact test and Mann-Whitney test, IBM SPSS Statistics 22 software (SPSS, Chicago, IL). P < 0.05 was considered statistically significant.