Hepatoma-Derived Growth Factor and DDX5 Promote Carcinogenesis and Progression of Endometrial Cancer by Activating β-Catenin

Background: Our previous work determined the correlation between high nuclear expression of hepatoma-derived growth factor (HDGF) and clinicopathological data of endometrial cancer (EC); however, the modulatory mechanisms and biological role of HDGF in EC have not been reported. Methods: Lentiviral particles carrying human HDGF short hairpin RNA (shHDGF-1, -2, and -3) vector and plasmids for HDGF, DDX5, and β-catenin expression were, respectively introduced into EC cells to evaluate the effects and molecular mechanisms underlying EC cell proliferation, migration, invasion, and metastasis. Quantitative real time reverse transcription polymerase chain reaction (qRT-PCR) and western blotting were used to determine HDGF and DDX5 expression. Co-immunoprecipitation (co-IP), mass spectrometry, and an immunofluorescence co-localization study were conducted to explore the relationship between HDGF, DDX5, and β-catenin. Immunohistochemistry was used to analyze the clinical associations between HDGF and DDX5 in EC. Results: Knocking down HDGF expression significantly decreased EC cellular proliferation, migration, invasion in vitro, as well as tumorigenesis and metastasis in vivo. Conversely, HDGF overexpression reversed these effects. Stable knockdown-based HDGF suppression activated the PI3K/AKT signaling pathway, along with downstream β-catenin-mediated cell cycle and epithelial-mesenchymal transition signaling. Furthermore, co-IP combined with mass spectrometry and an immunofluorescence co-localization study indicated that HDGF interacts with DDX5, whereas β-catenin was associated with DDX5 but not HDGF. Overexpression of DDX5 reversed the suppression of shHDGF. Immunohistochemistry analysis showed that high expression of DDX5 constituted an unfavorable factor with respect to the clinicopathological characteristics of EC tissues and that HDGF and DDX5 high expression (HDGF+/DDX5+) led to a worse prognosis for patients with EC (P < 0.001). In addition, we found that the expression of HDGF and DDX5 was positively correlated in EC tissues (r = 0.475, P < 0.001). Conclusion: Our results provide novel evidence that HDGF interacts with DDX5 and promotes the progression of EC through the induction of β-catenin.


INTRODUCTION
Endometrial cancer (EC) comprises the most common malignancy involving the female genital tract and the fourth most common malignancy in women after breast, lung, and colorectal cancers (1). In 2012, approximately 320,000 new cases of EC were diagnosed worldwide and the incidence is increasing (2). Currently, endometrial carcinogenesis is thought to be a multi-step process involving the coordinated interaction of hormonal regulation, gene mutation, adhesion molecules, and apoptosis; however, the molecular mechanisms underlying the pathogenesis of EC have not been fully elucidated (3). Therefore, a better understanding of the molecular mechanism underlying the progression of EC will likely lead to new insights regarding novel therapeutic targets.
Hepatoma-derived growth factor (HDGF), the gene for which is located on chromosome 1q21-23, is a heparin-binding growth factor that was originally purified from media conditioned with the human hepatoma cell line, HuH7 (4). HDGF is ubiquitously expressed in normal tissues and tumor cell lines that exhibit growth factor properties. The most recent study reported that HDGF acts as a coactivator of SREBP1-mediated transcription of lipogenic genes (5). HDGF is characterized as a mitogen for many cell types and localizes to the nucleus, which is necessary for its mitogenic activity. Characteristics such as promoting growth, suppressing differentiation, and exhibiting angiogenic properties, suggest a role for HDGF in cancer induction and tumor progression (6)(7)(8). Accordingly, a number of studies have focused on the significance of HDGF as a prognostic marker and have demonstrated its clinical value for oral cancer (9), esophageal cancer (10), gastrointestinal stromal tumors (11,12), meningiomas (13), hepatocellular cancer (14), and non-small cell lung carcinoma (15). Consistent with these findings, in our previous study (16), we determined a correlation between high nuclear expression of HDGF and clinicopathological data of EC; however, the functional significance of HDGF in EC remains unknown.
In the current study, we examined and characterized the interaction between HDGF and DDX5, and determined its effect on the activation of PI3K/AKT and downstream βcatenin-mediated cell cycle and EMT signaling to promote the proliferation, invasion, and metastasis of EC.

Cell Culture
The EC cell lines Ishikawa and RL95-2 were purchased from the Chinese Academy of Sciences Cell Bank (Shanghai, China). All cell lines were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum (ExCell, Shanghai, China). Ishikawa and RL95-2 cell lines used in this study were incubated in a humidified chamber with 5% CO 2 at 37 • C.

Immunohistochemistry and Evaluation of Staining
One hundred and twenty two endometrial carcinoma (EC) paraffin sections (3 mm) samples from 2002 to 2008 were obtained in the Third Affiliated Hospital of Guangzhou Medical University, Guangzhou City, China. Detailed information and the IHC of HDGF about the 122 EC tissue specimens was performed in our previous study (16). Immunohistochemistry was performed according to standard procedures. The staining intensity of DDX5 (1:100, Abcam, Cambridge, MA, USA) was scored as previously described (20).

Establishment of EC Cell Lines With Stable Knockdown of HDGF
Lentiviral particles carrying human HDGF short hairpin RNA (shHDGF-1, -2, and -3; Supplementary Table 1) vector and empty vector controls (PLV-Ctr) were constructed by GeneChem (Shanghai, China). Ishikawa and RL95-2 cells were infected with lentiviral vectors, and cells with green fluorescent protein signals (Supplementary Figure 1) were selected for further experiments using qRT-PCR and western blotting analyses. Transient Transfection Using Plasmids or Small Interfering RNAs or PI3K Inhibitor Ly294002 HDGF, DDX5, and β-catenin plasmids were generated by Biosense Technologies (Guangzhou, China). Small interfering RNA (siRNA) for DDX5 and β-catenin (named as siDDX5 and siβ-catenin, respectively) were designed and synthesized by Guangzhou RiboBio (RiboBio Inc., China). At 24 h before transfection, EC cells were plated onto a 6-or 96-well plate (Nest Biotech, China) at 30%-50% confluence. Plasmids were then transfected into cells using Lipofectamine TM 2,000 (Invitrogen Biotechnology, Shanghai, China), according to the manufacturer's protocol. EC cells overexpressing HDGF were treated with or without Ly294002 according to a previous description (30). Cells were collected after 48-72 h for further experiments.

EdU Incorporation Assays
Proliferating EC cells were examined using the Cell-Light EdU Apollo 567 in vitro Imaging Kit (RiboBio, Guangzhou, China), according to the manufacturer's protocol. Briefly, after incubation with 10 mM EdU for 2 h, EC cells were fixed with 4% paraformaldehyde, permeabilized in 0.3% Triton X-100, and stained with Apollo fluorescent dyes. A total of 5 mg/mL of DAPI was used to stain cell nuclei for 10 min. The number of EdU-positive cells was counted under a fluorescence microscope in five random fields. All assays were independently performed three times.

Colony Formation Assay
Cells were plated in 6-well culture plates at 200 cells/well (3 wells/cell group). After incubation at 37 • C in a 5% CO 2 incubator for 15 days, cells were washed twice with phosphate buffered saline (PBS) and stained with hematoxylin solution. The visible colony numbers were counted. All experiments were repeated at least three times.

Cell Cycle Analysis
A total of 5 × 10 6 EC cells were harvested after a 48-h incubation, and then washed with cold PBS. The cells were further fixed with 70% ice-cold ethanol at 4 • C overnight. Fixed cells were washed three times with cold PBS. After incubation with PBS containing 10 mg/mL of propidium iodide and 0.5 mg/mL of RNase A for 15 min at 37 • C. FACS caliber flow cytometry (BD Biosciences, San Jose, CA, USA) was used to obtain the DNA content of the labeled cells.

In vivo Tumorigenesis in Nude Mice
The animal studies were approved by the Animal Ethics Committee of the Southern Medical University. A total of 5 × 10 6 logarithmically growing EC cells transfected with shHDGF or PLV-Ctr (n = 5 per group) in 0.1 mL of RPMI-1640 medium were subcutaneously injected into the left-right symmetric flank of 4-5-week-old male BALB/c-nu mice. The mice were maintained in a barrier facility on HEPA-filtered racks. The animals were fed an autoclaved laboratory rodent diet. After 21 days, the mice were sacrificed and tumor tissues were excised and weighed.

Wound Healing Assay
EC cells were plated in 6-well plates and incubated overnight until 90% confluent. An injury line was made using a 10-µL plastic filter tip to create a wound approximately 10 µm in diameter. Then we removed the culture medium and used PBS to eliminate dislodged cells. Subsequently, the wells were covered with serum-free medium to incubate for 48 h. "Wound closure" was observed at 0, 12, 24, 48 h under an inverted microscope.

Transwell Migration and Invasion Assays
For cell migration assays, 1 × 10 5 cells in 100 µL of RPMI-1640 medium without serum were seeded on a fibronectincoated polycarbonate membrane insert in a Transwell apparatus (Corning, Armonk, NY, USA). In the lower chamber, 500 µL of RPMI-1640 with 10% serum was added as a chemoattractant. After the cells were incubated for 10 h at 37 • C in a 5% CO2 atmosphere, the insert was washed with PBS and cells on the top surface of the insert were removed with a cotton swab. Cells adhering to the lower surface were fixed with methanol, stained with Giemsa solution, and counted under a microscope in 5 pre-determined fields (200×). For the cell invasion assay, the procedure was similar to the Transwell migration assay, except that the Transwell membranes were pre-coated with 24 µg/mL of Matrigel (R&D Systems, Minneapolis, MN, USA) for 4 h. All assays were independently repeated three times.

In vivo Metastasis Assays
In vivo metastasis assays were performed according to a previous study (30). A total of 5 × 10 6 EC-shHDGF and -PLV-Ctr cells were injected into nude mice (n = 5 for each group) through the liver membrane. Whole-body optical images were visualized to monitor primary tumor growth and formation of metastatic lesions. After 2 months, all mice were sacrificed, individual organs were removed, and metastatic tissues were analyzed by hematoxylin and eosin staining.
Frontiers in Oncology | www.frontiersin.org manufacturer's instructions. The cells were lysed and the protein concentrations were measured. Then, 2000 µg of protein in 400 µL of supernatant was incubated with 10 µg anti-HDGF, anti-DDX5, and anti-β-catenin or anti-IgG antibodies coated on beads on a rotator overnight at 4 • C. The beads were washed, eluted in sample buffer, and boiled for 8 min at 95 • C. Immune complexes were subjected to Coomassie brilliant blue staining, mass spectrometry (Geneseed Biotech Co., Ltd, Guangzhou, China) and western blotting analysis. Anti-IgG was used as a negative control.

Immunofluorescence
Immunofluorescent staining was performed according to standard procedures as a previous study (33). Briefly, cells were cultured overnight, then fixed with 3.5% paraformaldehyde and permeabilization by 0.2% Triton X-100 at room temperature. Cells were incubated with mouse anti-β-catenin and rabbit anti-DDX5 antibody overnight at 4 • C. After three washes in PBS, the coverslips were incubated for 1 h with secondary antibody. Then, coverslips were mounted onto slides with mounting solution containing 0.2 mg/mL of DAPI and sealed with nail polish. Slides were stored in a dark box and observed using a scanning confocal microscope (Zeiss LSM 800, Oberkochen, Germany).
Statistical Analysis SPSS 21.0 (Chicago, IL, USA) and Graph Pad Prism 5.0 software (LaJolla, CA, USA) were used to analyze all data for statistical significance. Comparisons between two groups were performed using Student's t-test, one-way analysis of variance (ANOVA) for multiple groups, and a parametric generalized linear model with random effects for tumor growth and the MTT assay. Data are expressed as the mean ± SD from at least three independent experiments. The chi-squared test was applied to determine the relationship between the level of DDX5 expression and clinicopathological characteristics. Analysis of HDGF and DDX5 expression in 122 EC tissues was performed using paired-samples t-tests. The relationships were analyzed using Spearman's correlation analysis. Survival analysis was performed using the Kaplan-Meier method. The multivariate Cox proportional hazards method was used for analyzing the relationship between variables and patient survival time. A prognostic value of < 0.05 (P < 0.05) was considered significant, and all tests were two-sided. Statistical significance was denoted as follows: * P < 0.05, * * P < 0.01, * * * P < 0.001.

Stable Knockdown of HDGF Expression Inhibits Cell Proliferation in vitro and in vivo
To gain insight into the role of HDGF in EC, we first used a lentiviral vector to specifically and stably knockdown the expression of HDGF in the Ishikawa and RL95-2 cells. The levels of HDGF were assessed by western blotting ( Figure 1A) and qRT-PCR ( Figure 1B). The most efficient knockdowns of HDGF expression were shown in shHDGF-3-Ishikawa and shHDGF-1-RL95-2 cells compared to the PLV-Ctr (P < 0.001) (Supplementary Figure 2).
Next, we assessed the effect of decreased HDGF expression on EC cell growth in vitro. The growth curves determined by MTT assays showed that growth of the shHDGF-Ishikawa and shHDGF-RL95-2 cells was significantly slower than that of PLV-Ctr cells (P < 0.05; Figure 1C). Conversely, overexpression of   (Figure 2A) reversed these effects (P < 0.05; Figure 2B). The EdU incorporation assay revealed that the percentage of cells in S phase decreased following the downregulation of HDGF expression (P < 0.001; Figure 1D). Colony formation assays showed that knockdown of HDGF significantly decreased cell proliferation (P < 0.001; Figure 1E). Furthermore, cell cycle analysis demonstrated that HDGF suppression dramatically reduced cell cycle progression from the G1 to S phase (P < 0.05; Figure 1F).
Subsequently, to confirm the growth effect of HDGF in vivo, we performed an in vivo tumorigenesis study by inoculating EC cells into nude mice. Mice in the shHDGF-EC and PLV-Ctr groups were sacrificed 21 days after inoculation, with average tumor weights of 0.187 g and 0.793 g, respectively (P < 0.001; Figure 1G). The mice injected with shHDGF-Ishikawa and shHDGF-RL95-2 cells had smaller tumor burdens ( Figure 1G) and displayed lower expression of HDGF, Ki67 and proliferating cell nuclear antigen (PCNA) in tumor tissues relative to the controls (Figure 1H). These results suggested that HDGF significantly promotes tumorigenesis in vivo.

HDGF Downregulation Suppresses Cell Migration, Invasion, and Intrahepatic Metastasis of EC Cells in vitro and in vivo
To examine the effect of HDGF on cell migration and invasion, a Transwell apparatus, wound healing assay, and Boyden chamber coated with Matrigel were used. After incubation FIGURE 5 | DDX5 interact with both HDGF and β-catenin in EC cells. Venn diagram depicting overlaps between HDGF and β-catenin proteins: the 22 common proteins are listed (A). Co-immunoprecipitation (Co-IP) detected the interaction of exogenous HDGF and DDX5 in RL95-2 cells (B). Co-IP detected the interaction of endogenous HDGF and DDX5 in RL95-2 cells (C). Nuclear co-localization of HDGF and DDX5, as well as β-catenin and DDX5 proteins, in EC cells by immunofluorescence under a scanning confocal microscope (D). DDX5 levels after HDGF knockdown, as assessed by western blotting (E). HDGF and β-catenin levels after DDX5 knockdown (F). Overexpression of β-catenin plasmid in shHDGF-EC cells increased DDX5 levels (G). DDX5 expression after β-catenin knockdown (H). DDX5 mRNA expression after HDGF knockdown, normalized to the expression of ARF5. One-way ANOVA and Dunnett's multiple comparison test ***P < 0.001 (I).
Frontiers in Oncology | www.frontiersin.org for 10 h, a reduced number of migrated cells were observed for shHDGF-Ishikawa and shHDGF-RL9-2 cells compared with PLV-Ctr cells (P < 0.001; Figure 3A). In addition, the wound healing assay demonstrated that shHDGF-Ishikawa and shHDGF-RL95-2 cells inhibited the migration capacity (P < 0.05; Figure 3B). As shown in Figure 3C, the results of the Boyden chamber coated with Matrigel assays were similar to those of the Transwell assays (P < 0.01 for each); however, overexpressing HDGF reversed these effects (P < 0.05; Figure 2C).
To further assess the effect of HDGF on EC intrahepatic metastasis, shHDGF-Ishikawa and control cells were independently injected into the liver capsules of nude mice. Fluorescence imaging was used to identify scattered metastatic nodules in the livers and intestines of nude mice that formed in the mice after 2 months. Only a few scattered metastatic cells were observed following injection of shHDGF-Ishikawa cells, whereas a variety of large clusters were observed in the PLV-Ctr group ( Figure 3D). As can be seen from Figure 3D, control PLV-Ctr cells were more easily transferred to intestinal tissue, while shHDGF was less or not metastasized. The HE and IHC of the metastatic intestinal tissues and numbers of metastatic foci were shown in Figures 3E-G. Taken together, these results suggest that HDGF effectively promotes cell migration, invasion, and intrahepatic metastasis of EC cells in vitro and in vivo.

HDGF Regulates the Expression of Cell Cycle-and EMT-Associated Genes via the PI3K/AKT and β-Catenin Signaling Pathways in EC Cells
To further study the mechanism by which HDGF regulates cell proliferation, migration, invasion, and metastasis, the protein levels of cell cycle-and EMT-associated genes were examined in Ishikawa and RL95-2 cells with stably suppressed HDGF. Knockdown of HDGF inhibited the activation of oncogenic cell cycle regulators, including pRb, E2F1, c-Myc, CCND1, and CDK4, and increased the level of P27 (Figure 4A). Further, we found that EMT markers (N-cadherin, Vimentin, and Snail) were suppressed, whereas the E-cadherin level increased ( Figure 4B). Simultaneously, the levels of p-PI3K, p-AKT, and β-catenin were significantly decreased ( Figure 4C). Further, we observed that knockdown of HDGF significantly suppressed both nuclear and cytosol protein expressions of β-catenin in EC cells ( Figure 4D). In subsequent study, we used specific inhibitor (Ly294002) of PI3K to suppress the expression of p-PI3K and observed that the protein expression of p-PI3k, p-AKT, β-catenin, Snail, N-cadherin, Vimentin was decreased, and E-cadherin was upregulated in overexpressing HDGF EC cells (Figures 4E,F). These results suggest that HDGF regulates the expression of cell cycle-and EMT-associated genes via the PI3K/AKT and β-catenin signaling pathways in EC cells.

DDX5 Interacts With HDGF and β-Catenin
In previous study, our team had shown the interaction of βcatenin and HDGF or DDX5 in Lung Adenocarcinoma (33). To explore the precise molecular mechanisms of HDGF in EC, co-IP, combined with mass spectrometry, was used in Ishikawa cells. This analysis yielded 242 potential HDGF-interacting proteins (Supplementary Table 3), including DDX5 (69 kDa band) and β-catenin (92 kDa band). In addition, we used data sets from public domain data to draw a Venn diagram to show the proteins that interact with HDGF (34) and β-catenin (35) proteins, and observed that there were 22 overlapping proteins ( Figure 5A). Exogenous and endogenous co-IP demonstrated that HDGF and DDX5 interact in Ishikawa cells, whereas β-catenin was associated with DDX5, but not HDGF (Figures 5B,C). Moreover, nuclear co-localization of DDX5 and β-catenin proteins, as well as that of HDGF and DDX5 proteins (33), was observed by immunofluorescence using a scanning confocal microscope ( Figure 5D). Taken together, these results suggest that HDGF is associated with DDX5 in EC, as is the case with β-catenin and DDX5.
Subsequently, western blotting indicated that stable knockdown of HDGF expression resulted in DDX5 protein downregulation ( Figure 5E); however, knockdown of DDX5 did not significantly affect HDGF expression, but resulted in β-catenin protein downregulation (Figure 5F). In addition, we observed that β-catenin suppression decreased the level of the DDX5 protein ( Figure 5G). Furthermore, overexpression of β-catenin from a plasmid in shHDGF-EC cells increased DDX5 expression (Figure 5H). qRT-PCR showed that the expression of DDX5 was reduced after knockdown of HDGF (Figure 5I), which indicated that HDGF affects DDX5 at the level of transcription.
These results indicated that DDX5 overexpression can overcome the EC cell growth suppression induced by shHDGF.

Association of HDGF and DDX5 Expression With the Clinicopathological Characteristics of EC Tissues
Combined with data from our previous work (16), Figures 7A-D displays the expression of HDGF and DDX5. Immunohistochemical staining showed that HDGF and DDX5 positive signals were mostly located in the nuclei of EC cells, with minor cytoplasmic distribution. Our previous results indicated that 25.5% (31/122) and 74.5% (91/122) of cases exhibited high and low nuclear expression of HDGF, respectively (16). Here, we found that DDX5 protein was positively expressed in 32.8% (40/122) and not expressed in 67.2% of tumors (82/122; Table 1). DDX5 expression was significantly related to FIGO stage (I+II vs. III; P < 0.001; Table 1); however, no significant correlation existed with respect to age, menopausal status, histologic grading, depth of myometrial invasion, or lymph node status in patients with EC (P > 0.05; Table 1). The results were similar to those obtained for the HDGF protein.

Correlation Between HDGF and DDX5 Expression in EC
There was a significant positive correlation between HDGF and DDX5 protein levels in EC tissues (r = 0.475, P < 0.001; Table 4).

DISCUSSION
In previous studies, HDGF has been shown to promote the progression of tumors by activating the AKT-MAPK (36), Akt and TGF-β (37), and VEGF signaling pathways (38), and interacting with β-catenin as a positive feedback loop (39); however, the molecular mechanism involved in HDGFassociated EC cell proliferation, invasion, and metastasis has not been elucidated.
In the present study, we observed that HDGF knockdown markedly decreased cell proliferation, migration, invasion, and metastasis in vitro. Furthermore, subcutaneous tumor These findings are consistent with an earlier report by Zhou et al. (40) in which downregulation of HDGF inhibited the proliferation and invasiveness of hepatocellular carcinoma cells.
Multiple studies have shown that PI3K/AKT constitutes a key signal mediator during carcinogenesis (41,42) and that activation of PI3K/AKT may regulate β-catenin signaling through Akt phosphorylation and inactivation of GSK3-β (43). In the current study, we observed that decreased HDGF expression suppressed p-PI3K and p-AKT levels, and the downstream β-catenin-mediated cell cycle and EMT signal molecules, such as pRb, E2F1, c-Myc, CCND1, CDK4, Snail, Vimentin, and Ncadherin, while elevating the expression of P27 and E-cadherin in EC cells. In addition, using Ly294002 to treat overexpressing HDGF EC cells induced a decrease of p-PI3K, p-AKT, βcatenin and metastatic effect related proteins such as Snail, Vimentin, and N-cadherin. Therefore, we hypothesized that HDGF knockdown significantly suppresses EC cell proliferation, migration, invasion, and metastasis through the PI3K/AKT and β-catenin pathways in EC.
To better understand the molecular mechanisms underlying HDGF promotion of EC proliferation and metastasis, we searched public domain data and screened DDX5 as a candidate interaction protein of HDGF. DDX5 is a member of DEAD box family proteins that plays an important role in the progression of many tumors (44)(45)(46)(47)(48). Subsequently, we observed that HDGF not only combined with DDX5 but also induced the expression of DDX5 in EC. Furthermore, DDX5 could reverse the inhibitive effects on cell growth, migration, and invasion in shHDGF EC cells. Previously, HDGF was found to bind the promoter of βcatenin, resulting in β-catenin transcriptional activation (39). Phosphorylated p68 (DDX5) can enter the cytoplasm, which leads to its interaction with β-catenin and displacement of Axin (49). DDX5 (P68) forms a complex with β-catenin and facilitates its transcription activation to regulate both cell adhesion and gene expression (23). Notably, as β-catenin can, in turn, function as a transcription factor to stimulate DDX5 expression by binding to its promoter (50), we thus speculated that HDGF upregulated the expression of DDX5 by inducing β-catenin expression. Consistent with this speculation, we observed that DDX5 was significantly increased in shHDGF-EC cells after transfection of the β-catenin cDNA.
In previous reports, DDX5 had been documented to promote cell proliferation and EMT by interacting with Wnt-β-catenin signaling and inducing the nuclear translocation of β-catenin (23, 26, 28, 51). Consistent with these previous reports, we observed that DDX5 not only directly combined with β-catenin, but also induced the expression of β-catenin via activating PI3K/AKT signaling and stimulated its nuclear translocation, thus inducing cell cycle transition and EMT signaling and the promotion of cell growth and metastasis in EC. These data demonstrated that HDGF interacts with DDX5 to further induce β-catenin to participate in cell growth and metastasis in EC.
In our previous study of HDGF expression in 122 samples of EC, we found that high nuclear expression of HDGF was positively correlated with FIGO stage (P = 0.032) and that patients with high expression of HDGF had poorer overall survival rates compared to those with low expression (P = 0.001) (16). In the current study, we further showed that DDX5 protein was expressed in 32.8% (40/122) of EC samples. Similar to HDGF, DDX5 expression was also significantly associated with FIGO stage (P < 0.001), and patients with high expression of DDX5 had poorer overall survival rates than those with low expression (P < 0.001). Multivariate analyses showed that high expression of DDX5 was an independent predictor of prognosis for patients with EC (P = 0.038). Notably, our data showed a significant positive correlation between HDGF and DDX5 in EC tissues (r = 0.475, P < 0.001). Finally, we observed that high nuclear levels of HDGF and DDX5 led to the worst prognosis for patients with EC.

CONCLUSIONS
In summary, this study provides compelling evidence that HDGF in combination with DDX5 induces β-catenin to form a complex, which significantly promotes EC cell proliferation, migration, invasion, and metastasis in vitro and in vivo. The underlying mechanism likely involves the activation of PI3K/AKT signaling and downstream β-catenin-mediated cell cycle and EMT signaling proteins (Figure 8). Our results suggest that HDGF-DDX5 and β-catenin work together to play important roles in EC carcinogenesis and progression.

ETHICS STATEMENT
For the use of the human tissue specimens for research purposes, prior consent from the patients and approval from the Ethics Committees of the Third Affiliated Hospital of Guangzhou Medical University were obtained. All animal studies were conducted in accordance with the principles and procedures outlined in the Southern Medical University Guide for the Care and Use of Animals.