DSE promotes aggressive glioma cell phenotypes by enhancing HB-EGF/ErbB signaling

Remodeling of the extracellular matrix (ECM) in the tumor microenvironment promotes glioma progression. Chondroitin sulfate (CS) proteoglycans appear in the ECM and on the cell surface, and can be catalyzed by dermatan sulfate epimerase to form chondroitin sulfate/dermatan sulfate (CS/DS) hybrid chains. Dermatan sulfate epimerase 1 (DSE) is overexpressed in many types of cancer, and CS/DS chains mediate several growth factor signals. However, the role of DSE in gliomas has never been explored. In the present study, we determined the expression of DSE in gliomas by consulting a public database and conducting immunohistochemistry on a tissue array. Our investigation revealed that DSE was upregulated in gliomas compared with normal brain tissue. Furthermore, high DSE expression was associated with advanced tumor grade and poor survival. We found high DSE expression in several glioblastoma cell lines, and DSE expression directly mediated DS chain formation in glioblastoma cells. Knockdown of DSE suppressed the proliferation, migration, and invasion of glioblastoma cells. In contrast, overexpression of DSE in GL261 cells enhanced these malignant phenotypes and in vivo tumor growth. Interestingly, we found that DSE selectively regulated heparin-binding EGF-like growth factor (HB-EGF)-induced signaling in glioblastoma cells. Inhibiting epidermal growth factor receptor (EGFR) and ErbB2 with afatinib suppressed DSE-enhanced malignant phenotypes, establishing the critical role of the ErbB pathway in regulating the effects of DSE expression. This evidence indicates that upregulation of DSE in gliomas contributes to malignant behavior in cancer cells. We provide novel insight into the significance of DS chains in ErbB signaling and glioma pathogenesis.


Introduction
High grade gliomas, including grade III anaplastic astrocytomas and grade IV glioblastomas, are among the most aggressive human cancers. They are the third greatest cause of cancer death in people under the age of 35 worldwide [1]. Currently

Tissue array and immunohistochemistry
Paraffin-embedded human glioma tissue microarrays with five normal brain tissue were purchased from Shanghai Outdo Biotech and Pantomics, Inc. UltraVision Quanto Detection System (Thermo Fisher Scientific Inc.) was used for immunohistochemistry. Arrays were incubated with anti-DSE antibody (1:100) in 5% bovine serum albumin/PBS 16 hours at 4˚C. The specific immunostaining was visualized with 3,3-diaminobenzidine and counterstained with hematoxylin (Sigma). Images were obtained by TissueFAX Plus Cytometer. The staining index was calculated as previous descripted [16]. A final score ! 6 were be considered high expression.

Western blotting and DS chains blotting
Western blotting was carried out as reported previously [25]. To analyze DS chains on proteoglycans, 80 μg of total protein lysate were separated by 10% SDS-PAGE and transferred to PVDF membrane. To simply the measurement of the changes to the DS chains in the cell lysate, we used horseradish peroxidase (HRP)-conjugated DS binding protein on western blots, and visualized by ECL. Total protein was measured by stain-free technology (Bio-Rad).

Immunofluorescence microscopy and confocal microscopy
Cells were plated onto coverslips in 24 well culture plates for 24 hours. For DSE staining, cells were fixed in 4% paraformaldehyde and stained with anti-DSE antibody and FITC-conjugated anti-rabbit IgG. Hoechst 33342 was used for nuclear staining. Images were captured by ZEISS Axio Imager A2 microscope. The confocal images of immunofluorescence staining were captured with a confocal microscope, Zeiss LSM 510 META. Five serial scan images within a total thickness of 3.2 μm were obtained for each field.

Dermatan sulfate ELISA
Double antibody ELISA kit for detecting Dermatan sulfate was purchased from MyBio-Source (MBS266153). The assay was performed according to manufacturer's protocol. 0.3 mg of total cell protein lysate in 100 μl lysis buffer was used for each test, and the specimens were triplicate.

CCK-8 assay and colony formation
Cells (2 × 10 3 ) were seeded into 96-well plates with culture medium 16 hours before the experiments. Cell viability was analyzed by CCK-8 reagent at 0, 24, 48, and 72 hours. Four wells per group for each time point were measured following manufacturer's protocol. The experiments were repeated for three times, and relative fold changes of OD 450 nm were shown. For anchorage-dependent colony formation assay, 500 cells were seeded in 6-well plates. Colonies were stained by crystal violet and counted after two weeks incubation.

Cell migration and invasion assay
Transwell inserts with uncoated porous filters (pore size 8 μm) were used to evaluate cell migration, and Matrigel (BD Biosciences) coated porous filters were used to measure cell invasion. 2×10 4 cells in serum-free DMEM were seeded into inserts, DMEM containing 10% FBS was added in lower part of the inserts for 16 hours incubation. Independent experiments were repeated for at least three times. Average number of cells per microscopic field was shown.

Mouse model for in vivo tumor growth
8 weeks male C57BL/6 mice were purchased from National Laboratory Animal Center (Tainan, Taiwan). 2.5 × 10 6 of GL261 mock transfectants and DSE transfectants were injected into right flank subcutaneously (n = 6). Tumor volumes were monitored at day 0, day 4, day 8, and day 14. Excised tumors were weighed and analyzed by Western blotting.
All animal experiments in this study were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of Chung Shan Medical University Experimental Animal Center.

Statistical analysis
All data analysis was performed using GraphPad Prism 6. One way ANOVA multiple comparisons were used for analyzing DSE gene expression in different glioma subtypes. Student t test was used for statistical analyses. Two-sided Fisher exact test was used for comparisons between DSE expression and clinicopathologic features of glioma tissue array. P < 0.05 was considered statistically significant.

Dermatan sulfate epimerase (DSE) is frequently upregulated in gliomas, and correlates with high tumor grade and poor survival
The biosynthesis of the iduronic acid in CS/DS chains is mediated by two epimerases: DS epimerase 1 (encoded by the DSE gene) and DS epimerase 2 (encoded by the DSEL gene) [26]. We first researched the expression of these two epimerases in the ONCOMINE database [27]. Compared with normal brain tissue, 6 out of 26 independent datasets indicated that only DSE is significantly upregulated in brain tumors, especially in grade IV glioblastomas ( Fig 1A). Furthermore, a search of the REpository for Molecular BRAin Neoplasia DaTa (REMBRANDT) database revealed that high expression of DSE is associated with worse overall survival in glioma patients (n = 329), whereas DSEL expression does not correlate with patient outcomes (S1 Fig). In addition, DSE expression in glioblastoma is significantly higher than that in astrocytoma, oligodendroglioma, and normal brain tissue ( Fig 2B). We conducted an immunohistochemical investigation of DSE protein expression in a glioma tissue array containing 77 primary glioma tissue samples and 5 non-tumor brain tissue samples. The immunohistochemical investigation revealed that DSE was expressed in the paranuclear and perinuclear cytoplasm of certain glioma tissue ( Fig 1C). No obvious DSE staining was found in the normal brain tissue. To assess the correlation with clinicopathologic characteristics, DSE expression was classified into two levels: low (staining index < 6) and high (staining index ! 6). Our data revealed that DSE was highly expressed in 73% (56/ 77) of the glioma tumors, whereas all the normal brain tissues exhibited very weak DSE expression (two-sided Fisher exact test, P = 0.0024) ( Fig 1C and Table 1). Furthermore, high DSE expression in glioma tissue was associated with the worst histologic grade (Table 1). These results are consistent with those reported for DSE gene expression in the public databases. In an examination of DSE protein levels in human glioblastoma cell lines, mouse glioblastoma GL261 cells, and mouse brain cortex tissue, we found that 5 out of 7 human glioblastoma cell lines expressed relative high levels of DSE. This anti-DSE antibody also detected endogenous mouse Dse in GL261 cells, and expression of Dse in mouse cortex tissue is very low (Fig 1D). These data suggest that DSE is frequently upregulated in glioma patients, and its expression is correlated with the worst histology grade and poor overall survival.

DSE mediates dermatan sulfate formation in glioma cells
We knocked down DSE in U118 cells and overexpressed it in GL261 cells to investigate the functions of DSE in DS formation. Stable DSE knockdown clones of U118 cells were established by transfection with two short hairpin RNA (shRNA) plasmids (S2 Fig). The DSE knockdown clones were pooled together for further investigation. Western blots and immunofluorescence staining demonstrated that DSE decreased in knockdown cells (Fig 2A). Confocal microscopy was used to reveal subcellular localization of DSE, which mainly expressed in paranuclear and perinuclear cytoplasm (S3 Fig). In addition, we found that DSE knockdown reduced the binding of the DS-binding protein to proteoglycans in the U118 cells (Fig 2B), whereas overexpression of DSE enhanced the DS-binding protein signals in the GL261 cells ( Fig 2C). Furthermore, the DS ELISA assay showed that DSE knockdown decreased DS in cell lysates, whereas DSE overexpression increased DS quantity in cell lysate (Fig 2D). These results indicate that DSE mediates the formation of DS chains on several proteoglycans in glioma cells. Comparison of DSE gene expression in glioma subtypes and normal brain tissue. ÃÃÃÃ P<0.0001, n.s. = not statistically significant. These data are from the REMBRANDT database (http://www.betastasis.com/glioma/rembrandt/). (D) Immunohistochemistry of DSE on a tissue array comprising 77 primary glioma samples and 5 normal human brain tissue samples. All sections were counterstained with hematoxylin. Representative images of normal brain tissue;

DSE regulates the proliferation, migration, invasion, and tumor growth of glioma cells
We investigated the effects of DSE on malignant phenotypes, cell viability, colony formation, and cell migration and invasion in DSE-knockdown and DSE-overexpressed clones. Our data demonstrated that DSE knockdown significantly suppressed cell viability in U118 cells and Ln18 cells, whereas overexpression of DSE enhanced cell viability (Fig 3A and S4 Fig). Consistently, DSE knockdown reduced the number of anchorage-dependent colonies, whereas overexpression of DSE dramatically increased the number of colonies (Fig 3B). Furthermore, DSE knockdown significantly reduced FBS-induced migration and invasion of U118 cells, whereas overexpression of DSE significantly promoted migration and invasion of GL261 cells (Fig 3C  and 3D). To investigate the effects of DSE on tumor growth in vivo, we subcutaneously transplanted mock and DSE-overexpressed GL261 cells into mice. Our results showed that overexpression of DSE significantly enhanced the volume and weight of the tumors (Fig 3E).
one DSE low expression case (upper panel), and two DSE high expression cases (lower panel) are shown. Amplified images are shown at the bottom right of each image. Scale bars, 100 μm. (E) Expression of DSE in glioma cell lines and mouse brain tissue. Protein expression was analyzed by western blotting. Actin was used as an internal control. Relative expression levels in U118 cells from three independent blots are shown at the right.
https://doi.org/10.1371/journal.pone.0198364.g001 Western blots and the immunohistochemistry of the excised tumors confirmed the expression of DSE in tumor tissue ( Fig 3F). These data indicate that DSE can modulate the cell viability, colony formation, cell migration and invasion, and in vivo tumor growth of glioma cells.

DSE mediates heparin-binding EGF-like growth factor (HB-EGF)/ErbB signaling in glioma cells
Aberrations in the ErbB pathway are the most common causes of alterations in human gliomas [28,29], and CS/DS hybrid chains have the ability to interact with several growth factors [7]. We next determined whether DSE regulates the ErbB signaling pathways in glioma cells. We treated glioma cells with the following recombinant ErbB family ligands: EGF, NRG1, and HB-EGF protein. We found that DSE knockdown in U118 cells selectively suppressed HB-EGF-triggered downstream ERK and AKT activation, whereas no obvious differences were observed following NRG1 and EGF treatment (Fig 4A and S5 Fig). HB-EGF can activate epidermal growth factor receptor (EGFR) homodimers or EGFR/ErbB2 heterodimers. We next investigated the phosphorylation of EGFR (Y1068) and ErbB2 (Y1248). DSE knockdown in U118 cells reduced HB-EGF-induced EGFR and ErbB2 activation, whereas it did not significantly alter EGF-induced EGFR activity (Fig 4B). In contrast, the overexpression of DSE in GL261 cells enhanced HB-EGF-induced EGFR and ErbB2 phosphorylation, as well as downstream ERK and AKT signaling, whereas EGF-triggered EGFR activation was still not affected ( Fig 4C). Interestingly, we observed that overexpression of DSE in GL261 cells significantly increased the total protein level of ErbB2 but not of EGFR, and DSE knockdown in the U118 cells attenuated total ErbB2 expression (Fig 4B and 4C). We searched for correlation between DSE and the ErbB family in the REMBRANDT database, and found that the DSE gene was mildly positive associated with ErbB2 expression and mildly negative correlated with ErbB4 ( Fig 4D). These results imply that DSE could involve the expression of ErbB2 and ErbB4 in glioma cells.
We treated glioma cells with an irreversible dual EGFR/ErbB2 inhibitor (afatinib) to evaluate the effects of the EGFR and ErbB2 pathways on DSE-regulated malignant phenotypes. We found that low concentrations of afatinib (0.5 μM) significantly suppressed the cell viability of the control shRNA-transfected U118 cells, but not that of the DSE-knockdown U118 cells. Furthermore, DSE-enhanced cell viability was suppressed by treatment with afatinib in GL261 cells (Fig 5A). A transwell invasion assay revealed that the solvent control (100% of EtOH)   attenuated cell invasion in both control and DSE-knockdown cells, and afatinib treatment suppressed control cells invasion but did not further suppress cell invasion in DSE-knockdown U118 cells (Figs 5B and 3D). In addition, afatinib treatment suppressed DSE-enhanced cell invasion in GL261 cells (Fig 5B). Although we cannot exclude that other pathways may involve in DSE-mediated phenotypes, blocking EGFR/ErbB2 signaling by afatinib can inhibit DSEinduced malignant phenotypes in glioma cells.

Discussion
The present study showed that DSE is frequently upregulated in human glioma tissue and cell lines, and DSE upregulation in glioma tissue is associated with a worse tumor grade and poor overall survival. The formation of DS on proteoglycans is regulated by DSE expression in glioma cells. DSE knockdown suppresses malignant phenotypes, whereas DSE overexpression enhances glioma cell malignancy, both in vitro and in vivo. Mechanically, DSE modulates HB-EGF-induced EGFR/ErbB2 activity and downstream signaling, whereas DSE-induced cell viability and invasion are hindered by EGFR/ErbB2 inhibitors. For the first time, the present study has shown that dermatan sulfate epimerase (DSE) is capable of regulating glioma cell malignancy in vitro and in vivo, and the ErbB pathway is involved in this process. Our findings provide a novel insight into the biological functions of DS chains in the pathogenesis of glioma, and may be useful in the development of a selective marker for glioma treatment.
In the past, DSE was considered a cancer cell antigen, and was called squamous cell carcinoma antigen recognized by T-cells 2 (SART2); it was supposedly highly expressed in most squamous cell carcinomas from various organs, including a subpopulation of colorectal carcinoma [30,31]. Our search of the ONCOMINE database showed that 6 out of 26 analyses (23%) of brain and CNS cancers revealed a greater than two-fold increase in the level of DSE, compared with normal tissue. Among the 20 cancer types classified in ONCOMINE, brain and CNS cancer have the highest ratio of DSE increase, and the 6 datasets mentioned above all refer to gliomas and glioblastomas. The results of the immunohistochemical glioma tissue array investigation showed that normal brain tissue (non-tumor) exhibited no or very low intensity DSE staining, and western blots of normal mouse brain tissue also revealed low expression of DSE. In contrast, there was DSE protein expression in a great portion of the glioma tissue and cell lines, and elevated DSE expression is associated with poor survival. These results indicate that other types of cancer involve elevated DSE expression, and suggest that DSE may participate in the progression of glioma.
CS/DS hybrid chains have greater structural flexibility than pure CS chains, and have greater affinity for various heparin-binding growth factors [23,32,33]. For instance, IdoAcontaining domains interact with fibroblast growth factor 7 (FGF7) and hepatocyte growth factor (HGF) [34,35]. These growth factor pathways are important for tumor development and progression. However, the direct biological functions of DSE and CS/DS chains in cancer cells were not investigated until chondroitin-glucuronate C5 epimerase activity was reported in 2006 [36]. It has been proposed that DSE promotes DS formation, resulting in enhanced HGF binding to the cell surface and the activation of downstream signaling in esophagus cancer cells [21]. Our results also indicated that DSE modulates DS chain formation in glioma cells. Importantly, we report for the first time that DSE selectively regulates HB-EGF-induced signaling, but not EGF and NRG1, implying that changes to DS could modulate the activity of HB-EGF or its affinity for EGFR/ErbB2 receptors. In addition, it would be worth to further investigate whether HB-EGF and HGF signaling pathways cooperate to promote DSE-mediated glioma malignancy.
HB-EGF mRNA expression is two-to five-fold higher in human glioblastomas than in normal brain tissue [37], and the HB-EGF receptors-EGFR homodimer and EGFR/ErbB2 heterodimer-are both commonly amplified in glioblastomas [29,38,39]. Furthermore, HB-EGF is expressed with EGFR in approximately 40% of human glioma tissues [40]. A recent study demonstrated that HB-EGF drives glioma tumorgenesis in mice lacking Ink4a/Arf and Pten through EGFR signaling [41]. These findings suggest that HB-EGF and EGFR/ErbB2 are promising therapeutic targets for glioblastoma treatment. Our data show that DSE regulates EGFR/ErbB2 signaling, and DSE expression may be positively associated with ErbB2 levels in glioblastoma cells. Although the mechanisms underlying the regulation of ErbB2 expression by DSE require further investigation, our inhibitor experiments suggest that EGFR/ErbB2 signaling could involve in the development of DSE-mediated malignant phenotypes in glioblastoma cells.
In conclusion, the results obtain in the present study suggest that DSE regulates the HB-EGF/ErbB pathway, which is involved in the DSE-induced malignant behavior of glioblastoma cells. The present study demonstrates a pathophysiologic role of DSE in glioblastoma cells, and elucidates the biological functions of aberrant CS/DS expression in glioma progression. Because there is no effective single treatment for unselected glioblastoma patients, targeting CS/DS chains with chondroitinase ABC or CSPG inhibitor, combined with other clinical trial agents, may provide novel strategies for treating this fatal disease.