Targeting UDP‐glucose dehydrogenase inhibits ovarian cancer growth and metastasis

Abstract More than 70% of patients with ovarian cancer are diagnosed in advanced stages. Therefore, it is urgent to identify a promising prognostic marker and understand the mechanism of ovarian cancer metastasis development. By using proteomics approaches, we found that UDP‐glucose dehydrogenase (UGDH) was up‐regulated in highly metastatic ovarian cancer TOV21G cells, characterized by high invasiveness (TOV21GHI), in comparison to its parental control. Previous reports demonstrated that UGDH is involved in cell migration, but its specific role in cancer metastasis remains unclear. By performing immunohistochemical staining with tissue microarray, we found overexpression of UGDH in ovarian cancer tissue, but not in normal adjacent tissue. Silencing using RNA interference (RNAi) was utilized to knockdown UGDH, which resulted in a significant decrease in metastatic ability in transwell migration, transwell invasion and wound healing assays. The knockdown of UGDH caused cell cycle arrest in the G0/G1 phase and induced a massive decrease of tumour formation rate in vivo. Our data showed that UGDH‐depletion led to the down‐regulation of epithelial‐mesenchymal transition (EMT)‐related markers as well as MMP2, and inactivation of the ERK/MAPK pathway. In conclusion, we found that the up‐regulation of UGDH is related to ovarian cancer metastasis and the deficiency of UGDH leads to the decrease of cell migration, cell invasion, wound healing and cell proliferation ability. Our findings reveal that UGDH can serve as a prognostic marker and that the inhibition of UGDH is a promising strategy for ovarian cancer treatment.

mortality rate is due to the low diagnostic options in early stages.
Over 70% of patients are at an advanced stage with regional or distant metastatic sites to the peritoneal cavity at the time of diagnosis. 2 Thus, a detailed understanding of ovarian cancer metastasis is required to develop effective treatments.
Cancer metastasis is a complex process that involves multiple steps. It is caused by the ability of cancer cells to separate from the primary tumour, migrate to distant sites through the bloodstream or lymph system and form secondary tumours. Metastatic tumours are also responsible for higher rates of mortality in most cancers rather than primary tumours. 3,4 In the initial step of the metastasis cascade, cancer cells lose their cell-cell connections and shed from primary tumour sites to invade the circulatory system. This step is triggered by a crucial process named as the epithelial-mesenchymal transition (EMT). In EMT, cancer cells can leave their primary sites and invade the bloodstream. Thus, cancer cells lose their epithelial traits and enhance the mesenchymal phenotype to accomplish metastasis.
During this period, metastatic cells gain the ability to degrade the extracellular matrix. 4 These complex programs are co-ordinated by numerous EMT-inducing factors, such as SNAI1, SIP-1 and Twist. 5 Despite the great efforts, our understanding of metastasis progression is limited, and further studies of the detailed mechanisms of its regulation are required.
Glycosphingolipid (GSL) is a glycolipid on the cell membrane and plays numerous roles in cells, including cell adhesion, development, differentiation, tumour progression and signal transduction. 6 Among all GSLs, Globo H (GH) has been reported to be associated with cancer progression. GH belongs to GSLs with the molecular structure of polysaccharide linked to the ceramide lipid 7 and the carbohydrate structure of GH is Fuca(1-2)Galb(1-3)GalNAcb (1)(2)(3) Gala(1-4)Galb(1-4)Glcb(1). 8 GH is overexpressed in numerous cancers, including breast, colon, gastric, endometrial, lung, ovarian, pancreatic, prostate cancer but shows low expression in normal cells. 9 A cancer-specific therapeutic vaccine composed of synthetic GH conjugated with keyhole limpet haemocyanin was developed for breast cancer. A previous study used a GH-specific antibody MBr1 to for immunohistochemical staining in small cell lung cancer and found that GH-positive tumours are associated with a shorter survival time than GH-negative tumours. 10 Several studies also demonstrated that the expression of GH is related to cancer aggressiveness in breast cancer and small cell lung cancer. Another recent publication revealed that GH cause immunosuppression by reducing Notch-1 signalling in human peripheral blood mononuclear cells and trigger translin-associated factor X-dependent angiogenesis. 9 UDP-glucose dehydrogenase (UGDH) is an enzyme that catalyses NAD + -dependent two-step oxidation of UDP-glucose to generate UDP-glucuronate, but its roles and detailed regulatory mechanisms in cancer progression remain unclear. UGDH participates in tumour formation and cancer migration in breast cancer, colorectal carcinoma, glioblastoma and lung cancer. [11][12][13][14] Previous studies also demonstrated that UGDH is regulated by transforming growth factor-β pathway, including p38, extracellular signal-regulated kinase (ERK), mitogen-activated protein kinase (MAPK). Moreover, UGDH has been reported to regulate cell glycosaminoglycan (GAG) level by catalysing the conversion of UDP-glucose to UDP-glucuronate.
Recently, Wang et al 14 found that inhibition of UGDH led to degradation of SNA1 mRNA and impaired lung cancer migration.
The current clinical condition of ovarian cancer presents high mortality rate. Although there are several improved treatments of ovarian cancer, including the platinum-based chemotherapy and the cytoreductive surgery, the 5-year overall survival of advanced ovarian cancer is approximately 30%. 15 The clinical report implied that the ovarian cancer metastasis leads to the poor prognosis. Hence, the understanding of ovarian cancer metastasis mechanism is an urgent issue for improving the ovarian cancer treatment. To investigate the molecular mechanism of ovarian cancer metastasis, we analysed global protein changes between low invasiveness TOV21G cell line (TOV21G LI ) and highly invasive cell line (TOV21G HI ) by high-through put proteomic approaches. In the obtained protein profile, we observed the overexpression of UGDH in TOV21G HI cells. In the present study, we investigate the role and detailed mechanism of UGDH activity in ovarian cancer metastasis development. Our results provide insights into the identification of both new promising biomarkers and efficient therapeutic targets for ovarian cancer treatment.

| Chemicals and reagents
Lipofectamine ® RNAiMAX transfection reagent was purchased from Invitrogen (Thermo Fisher scientific Inc), and OPTI-MEM was purchased from Gibco (Thermo Fisher scientific Inc). MTT Anti-rabbit and antimouse immunoglobulin (Ig)G FITC-conjugated secondary antibody were purchased from SeraCare KPL. All Biochemicals, chemicals and reagents used in this study were of analytic grade.

| 2D-DIGE analysis and protein identification of MALDI-TOF MS
Prior to perform 2D-DIGE, the cell pellets were solubi-

| Lentiviral constructs for expression of anti-UGDH shRNA and establishing of stable knockdown cell line
The

| Transwell migration and Matrigel TM invasion assay
Transwell permeable inserts with 8 μm pore size PET membrane (COSTAR, Corning Inc) were used to measure cell migration and invasion ability. Briefly, 1 × 10 5 cells with serum-free medium were seeded in the upper chamber and the complete medium with 10% FBS were loaded in the lower chamber for attracting cell migration.

| Scratch wound healing assay
Cell was seeded in two well culture insert (ibidi, Germany) at a den-

| Cell proliferation (doubling time) assay
TOV21G, A2780 and HeyA8 cells were trypsinized and seeded into 96-well plates at a density of 3000 cells/ well. After 24 hours of incubation at 37℃, 5% CO 2 (for Day 0 experiment), medium was removed and cells were incubated with 100 μL of MTT solution (1 mg/ mL) per well for 4 hours at 37℃. The supernatant then was removed, and 100 μL dimethyl sulfoxide (DMSO) was added per well to dissolve the insoluble formazan. The 96-well plate was then shook for 3 minutes before the absorbance was measured by spectrophotometer at 570 nm. The cell growth rates were monitored by the same method in the following time-points, Day1 (24 hours), Day2 (48 hours) and Day3 (72 hours), respectively. The proliferation rates were presented as value relative to Day 0. Cytometry by collecting emission data from FL2 channel at 575 nm.

| Statistical analysis
Data are presented as the mean ± SEM. Differences between the experimental groups were assessed using a paired Student's t test or a one-way ANOVA followed by Tukey's multiple comparison test.
Test results with P < .05 were considered statistically significant.

| Identification of UGDH in highly invasive ovarian cancer cell line via proteomic analysis
To investigate the metastatic mechanism of ovarian cancer, we analysed the expression level of GH, a cancer-specific marker, 18 in TOV21G cells. We isolated two cell groups by BD FACSAria™ III cell sorter according to the expression level of GH. In our flow cytometry data, TOV21G HI cells showed a higher expression level of GH compared to TOV21G LI cells ( Figure 1A). The immunofluorescence results revealed relatively higher expression level of GH in TOV21G HI compared to in TOV21G LI cells ( Figure 1B). Moreover, TOV21G HI cells exhibited significantly increased cell invasion and cell migratory abilities compared to TOV21G LI cells ( Figure 1C,D).
Next, proteomic analysis was applied to elucidate the global protein changes between isogenic TOV21G LI and TOV21G HI cells. We DIGE images and statistic data ( Figure 2B). To further confirm our data of proteomic analysis, we performed immunoblotting to validate the expression level of UGDH between the TOV21G LI and TOV21G HI cell lines. The expression level of UGDH in TOV21G HI cells was significantly higher than that in TOV21G LI cells, suggesting that UGDH is overexpressed in a highly aggressive ovarian cancer cell line.

| Expression of UGDH is correlated to aggressive types of ovarian cancer
In the cell line-based study, we observed overexpression of UGDH  Figure 2D). These results support that overexpression of UGDH is not only related to tumour malignancy but also is a potential prognostic biomarker for ovarian carcinoma and mucinous adenocarcinoma.

| Knockdown of UGDH reduces cell proliferation in ovarian cancer by prompting G1 phase arrest
Previous studies demonstrated that a deficiency of UGDH leads to reduced cell proliferation in colon cancer 12 and glioblastoma. 13 We hypothesized that knockdown of UGDH by short interfering RNA were harvested and examined to determine the cell proliferation rate by MTT assay from starting at days 1-3. As shown in Figure 3A, the relative proliferation rates were significantly lower in siUGDHtransfected TOV21G LI and TOV21G HI cells than in control cells from

| Knockdown of UGDH through siRNA impairs wound healing and migration of ovarian cancer cells
We next focused on the role of UGDH in cell wound healing and cell migration in ovarian cancer by wound healing assay (scratch assay) and transwell migration assay. UGDH was knocked by siRNA in three ovarian cancer cell models, TOV21G, A2780 and HeyA8 cells. The cell migratory ability was significantly decreased when TOV21G HI cells were transfected with siUGDH ( Figure 4A).
Transwell migration assay revealed that knockdown of UGDH significantly decreased the migration ability of TOV21G HI cells compared to control cells. Similar effects were observed in A2780 and HeyA8 cells ( Figure 4C). Our data demonstrate that knockdown of UGDH attenuated the migratory ability of ovarian cancer cells. The wound healing areas in TOV21G HI cells transfected with siUGDH were significantly impaired compared to in TOV21G HI control cells at 12 hours ( Figure 4C), whereas siUGDH did not alter the wound healing ability in TOV21G LI cells. Based on these findings, silencing of UGDH reduced the migration ability in TOV21G HI , A2780 and HeyA8 cells.

| Knockdown of UGDH decreased ovarian cancer tumour growth in xenograft model
The results revealed that knockdown of UGDH decreased ovarian cancer migration, wound healing ability and cell proliferation.
Thus, we hypothesized that a deficiency in UGDH would influ-

| Knockdown of UGDH decreased gh expression level and impaired metastatic ability of ovarian cancer cells by inhibiting mapk signalling pathway, F-actin polymerization and EMT
To understand the regulatory role of UGDH in ovarian cancer metastasis, we focused on monitoring the molecular pathway of UGDH knockdown in TOV21G cells. We investigated the expression level of the EMT markers in UGDH knockdown cell lines. EMT is recognized to have a critical role in cancer metastasis. SIP-1, SNAIL and TWIST were reported as transcription factors with major regulatory roles in cancer metastasis. 19 Our immunoblot analysis showed that SIP-1 and SNAIL were up-regulated in TOV21G HI cells,

F I G U R E 3
Deficiency of UGDH by siRNA inhibited cell proliferation and led to cell cycle arrest. A, Cell proliferation rates of TOV21G LI and TOV21G HI cells were monitored from days 1 to 3 after the treatment with siUGDH by MTT assay. The absorbance at days 2 and 3 was normalized to that at day 1. Data are presented as the mean ± SEM. ***, P < .001 compared to control ovarian cancer cells. B, Representative diagrams of cell cycle of control and siUGDH-treated cells analysed by CFlow software. siUGDH-treated TOV21G LI and TOV21G HI cells and control cells were stained with PI and applied to analyse DNA content by flow cytometry. The distribution of cell cycle phases (G 0 /G 1 , S, G 2 /M) is shown as indicated. C, Numbers of cells at the different cell cycle stages were statistically analysed and graphed. Data derived from three independent experiments are presented as the mean ± SEM. **, P < .01 compared to control cells. D, Expression levels of cell cycle regulatory proteins were monitored in response to siUGDH treatment. Immunoblotting was used to detect the expression level of cyclin D2, cyclin E, p53, p21 and p27 proteins in control and siUGDH-treated TOV21G cells. Immunoblotting results of the indicated proteins were quantified by ImageJ and normalized to LDH. Cell proliferation assay via MTT assay was performed to analyse the effect of siUGDH on A2780 and HeyA8 cells (E), (F). siUGDH-treated cells were monitored to assess the proliferation rates from days 1 to 4 by MTT assay. Data are presented as the mean ± SEM. **, P < .05; ***, P < .001 compared to control ovarian cancer cells. G, H, Representative plots of cell cycle analysis of A2780, HeyA8 and their siUGDH-treated partners. I, J, Immunoblotting was performed to detect the expression levels of cyclin D2, cyclin E1, p21, p27 and p53 in A2780, HeyA8 and UGDH knockdown partners

| D ISCUSS I ON
Cancer metastasis not only accounts for most cancer-related death but also is a major clinical obstacle to cancer therapy. In ovarian cancer, most patients have regional or distant metastasis at the time of diagnosis. 25 to investigate the potential proteins involved in ovarian cancer metastasis. Among the identified proteins, UGDH was found to be elevated in TOV21G HI cells.
UDP-glucose dehydrogenase was reported to induce prostate cancer progression and can serve as a potential biomarker of prostate cancer. 30 In a report from The Human Protein Atlas, overexpressed UGDH is associated with poor survival rates in renal cancer. 31 Oyinlade et al 13  We found that knockdown of UGDH leads to the decrease in the cell proliferation rate in TOV21G, A2780 and HeyA8 cells. Moreover, G 1 phase cell cycle arrest was observed in UGDH knockdown cells.
These results are consistent with previous reports in glioblastoma cells in which knockdown of UGDH by shRNA decreased cell proliferation and displayed a delay in G 0 /G 1 to S phase transition. 13 Our further studies of the expression level of cell cycle regulators showed that p21 and p27 were up-regulated in UGDH knockdown cells. A recent report suggested that expression of p21 and p27 lead to cell cycle arrest in G 0 /G 1 phase, ultimately resulting in cell death in lung cancer. 35 Our results also showed that p21 and p27 were whereas knockdown of UGDH leads to decreases in these markers, indicating that UGDH is involved in regulating ERK, MMP-2 and EMT markers in ovarian cancer. We also revealed that knockdown of UGDH inhibits EMT, possibly through the ERK/MAPK pathway ( Figure 7).
UDP-glucose dehydrogenase serves as a metabolic enzyme and converts UDP-glucose to UDP-glucuronic acid, which is an initial step in the synthesis of GAGs. 50 GAGs including chondroitin sulphate, dermatan sulphate, heparin sulphate and hyaluronan are the major components of extracellular matrix that participate in regulating several cancer cell behaviours, such as invasion, migration and angiogenesis. 51 Chondroitin sulphate binds to fibroblast growth factor-2 and vascular endothelial growth factor and forms complexes with their receptors, which enhances the signalling cascade and promotes cancer progression. 51,52 Moreover, heparin sulphate can interact with fibroblast growth factor-2 (FGF-2) and trigger cell proliferation and EMT by inducing FGF-2 signalling. 53,54 In the current study, we observed overexpression of UGDH in highly invasive ovarian cancer cells, and siRNA-mediated knockdown of UGDH reduced the metastatic abilities of TOV21G, A2780 and HeyA8 cell lines. We predict that knockdown of UGDH decreased the levels of GAGs in ovarian cancer and therefore affected the ovarian cancer metastatic ability.
Previous studies revealed decreased expression levels of GAGs and hyaluronic acid in UGDH-silenced colorectal carcinoma 12 and glioblastoma. 13 Further studies are needed to demonstrate the effect of UGDH on the levels of GAGs in ovarian cancer. Additionally, we found that the expression level of GH was reduced in both UGDH shRNA transduced TOV21G LI and TOV21G HI cells, suggesting that knockdown of UGDH not only affects the amounts of GAGs but also reduces the expression of Globo H ( Figure S1). Furthermore, previous report mentioned that quercetin, a polyphenol, had inhibitory effect on the enzyme activity of UGDH. 55 To comprehensively understand the inhibitory mechanism of quercetin on UGDH, we used quercetin for analysing the effects of quercetin on cell migration and cell invasion ability. Our results demonstrated that quercetin inhibited the cell survival of TOV21G cells ( Figure S2A). Moreover, treatment of quercetin also inhibited the migration and invasive ability of TOV21G cells ( Figure S2B,C).
In conclusion, we demonstrated that UGDH is related to high invasiveness in ovarian cancer and provided evidence supporting that UGDH participates in cancer migration, invasion and cell proliferation in ovarian cancer. Collectively, UGDH can serve as a prognostic marker and a potential therapeutic target for restricting ovarian cancer metastasis. Future studies are likely to develop small molecular inhibitors specific for UGDH and to elucidate the relation between UGDH-related glycan and metastasis.

ACK N OWLED G EM ENTS
We appreciate Dr Chih-Long Chang and Dr Choa-Chih Wu, Taipei MacKay Memorial Hospital, Taiwan, for providing TOV21G cell lines as well as giving experimental comments. We also thank National Laboratory Animal Center (NLAC), NARLabs, Taiwan, for technical support.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.