Galectin 1 Proangiogenic and Promigratory Effects in the Hs683 Oligodendroglioma Model Are Partly Mediated through the Control of BEX2 Expression

Abstract

We have previously reported that galectin 1 (Gal-1) plays important biological roles in astroglial as well as in oligodendroglial cancer cells. As an oligodendroglioma model, we make use of the Hs683 cell line that has been previously extensively characterized at cell biology, molecular biology, and genetic levels. Galectin 1 has been shown to be involved in Hs683 oligodendroglioma chemoresistance, neoangiogenesis, and migration. Down-regulating Gal-1 expression in Hs683 cells through targeted small interfering RNA provokes a marked decrease in the expression of the brain-expressed X-linked gene: BEX2. Accordingly, the potential role of BEX2 in Hs683 oligodendroglioma cell biology has been investigated. The data presented here reveal that decreasing BEX2 expression in Hs683 cells increases the survival of Hs683 orthotopic xenograft-bearing mice. Furthermore, this decrease in BEX2 expression impairs vasculogenic mimicry channel formation in vitro and angiogenesis in vivo, and modulates glioma cell adhesion and invasive features through the modification of several genes previously reported to play a role in cancer cell migration, including MAP2, plexin C1, SWAP70, and integrin β₆. We thus conclude that BEX2 is implicated in oligodendroglioma biology.

Introduction

Gliomas are the most common primary brain malignancy found in adults [1,2]. High-grade gliomas of astrocytic origin are characterized by the diffuse invasion of distant brain tissue by a myriad of single migrating cells with reduced susceptibility to apoptosis and consequent resistance to the cytotoxic insults of proapoptotic drugs [2]. Current recommendations are that patients with high-grade astrogliomas (including glioblastoma multiforme, GBM) should undergo maximum surgical resection, followed by concurrent radiation and chemotherapy with the alkylating drug temozolomide (TMZ) [3]. Oligodendrogliomas are rare and diffusely infiltrating tumors arising in the white matter of cerebral hemispheres and display better sensitivity to treatment and a better prognosis than other gliomas [4].

Chronic platelet-derived growth factor signaling in glial progenitors leads to the formation of oligodendrogliomas in mice, whereas chronic combined Ras and Akt signaling leads to astrocytomas [5]. The different histologies of these tumors imply that the pathways activated by these two oncogenic stimulations are different and that the apparent lineage of the tumor cells may result from specific signaling activity [5]. Combined activation of Ras and AKT leads to the formation of astrocytic GBM in mice [6]. In human GBMs, AKT is not mutated but is activated in approximately 70% of these tumors, in association with loss of PTEN and/or activation of receptor tyrosine kinases [6]. Oligodendrogliomas with 1p19q codeletion harbor a gene expression profile that partially resembles the gene expression of normal brain samples, whereas gliomas with epidermal growth factor receptor amplification express many genes in common with glioblastoma cancer stem cells [7]. Neuronal differentiation could occur in oligodendrogliomas with 1p19q codeletion and the cell of origin for gliomas with 1p19q codeletion could be a bipotential progenitor cell, able to give rise to both neurons and oligodendrocytes [7].

Galectin 1 (Gal-1) belongs to a family of 15 carbohydrate-binding proteins with affinity for β-galactosides [8,9] and has been shown to play an important role in astroglioma and oligodendroglioma cell biology [9]. For example, Gal-1 is a key player in astroglioma and oligodendroglioma cell migration [10–12]. Galectin 1 expression is increased under hypoxic conditions [13], and radiotherapy stimulates Gal-1 expression in glioma cells [14].

We have recently demonstrated that TMZ stimulates Gal-1 expression in the Hs683 oligodendroglioma model [15] and that decreasing Gal-1 expression in these Hs683 cells increases the antitumor effects contributed by TMZ both in vitro and in vivo [15,16].

Galectin 1 is a key player in tumor neoangiogenesis [17], and we have recently demonstrated that Gal-1 exerts its proangiogenic effects in human Hs683 oligodendrogliomas through the endoplasmic reticulum transmembrane kinase/ribonuclease inositol-requiring 1α (IRE-1α)-mediated control of the expression of oxygen-related protein 150, which, in turn, controls vascular endothelial growth factor (VEGF) maturation [16].

We observed that down-regulating Gal-1 expression in Hs683 human oligodendroglioma cells through targeted small interfering RNA (siRNA) provokes a marked decrease in the expression of the brain-expressed X-linked gene. BEX2, which belongs to a small family of six genes with a wide distribution among tissue types [18,19], is found highly expressed in the embryonic brain [20]. BEX2 is described as a tumor suppressor gene in astrogliomas [21] and has been implicated in apoptotic features of breast cancer [22]. The role of BEX2 in oligodendroglioma biology has never been investigated, at least to the best of our knowledge, and therefore represents the aim of the present work. As a human oligodendroglioma model, we made use of the Hs683 cell line, whose oligodendroglial origin has been extensively characterized. Indeed, Hs683 tumor cells are 1p19q co-deleted [23] and are highly sensitive to chemotherapy under in vivo orthotopic brain xenograft conditions [23]. These cells display very weak (if any) vimentin expression and high levels of integrin β₄ expression [24] as do oligodendrogliomas [24,25]. Hs683 cells do not express the human 1p-distal ATAD 3B gene, which is highly expressed in astroglioma cells [26]. However, they contain only one Notch2 gene copy per diploid genome as do oligodendrogliomas [27], in which loss of the 1p centromeric marker within intron 12 of the Notch2 gene is associated with a favorable prognosis in oligodendroglioma patients [28,29].

Materials and Methods

Cell Cultures and Compounds

The Hs683 (ATCC code HTB-138), U87 (ATCC code HTB-14), U373 (ATCC code HTB-17), and T98G (ATCC code CRL-1690) human glioma cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and maintained in our laboratory as detailed previously [15,16,23,24].

We also made use of four primocultures established in our facilities as detailed previously [30]. Briefly, glioma tumor specimens were transferred during surgery into culture medium (RPMI-1640 with 20% fetal calf serum [FCS], 1% glutamine, 1% penicillin/streptomycin). The tissue was gently disrupted and washed with physiological saline to remove red blood cells. After passage 5, cells were grown in a culture medium containing 10% FCS without antibiotics. These primocultures have been labeled GL-5, GL-16, GL-17, and GL-19 in the current study. They are part of a collection of several hundred glioma primocultures established in the Department of Neurosurgery of the Wagner Jauregg Hospital (Linz, Austria).

Anti-Gal-1 siRNA and Anti-BEX2 siRNA

The sequence of the siRNA (Eurogentec, Seraing, Belgium) used in the current work are listed below. For Gal-1, the sense sequence was 5′-gcugccagauggauacgaadtdt-3′ and the antisense sequence was 5′-uucguauccaucuggcagcdtdt-3′ [15,16]. For BEX2, the sense sequence was 5′-gcaggagaguuuuaccuaudtdt-3′ and the antisense sequence was 5′-auagguaaaacucuccugcdtdt-3′. A corresponding nonspecific sequence siRNA (designated throughout as CT siRNA) was used as a process control (sense, 5′-ggaaaucccccaacagugadtdt-3′; antisense, 5′-ucacuguugggggauuuccdtdt-3′). The antisense and sense strands of the siRNA were annealed by the manufacturer in 50 mM Tris, pH 7.5 to 8.0, and 100 mM NaCl in diethylpyrocarbonate-treated water. The final concentration of siRNA duplexes was 100 µM.

Hs683 glioma cells were either left untreated (wild type: wt) or were transfected with calcium phosphate (Kit ProFection mammalian transfection system; Promega, Leiden, the Netherlands) containing the specified siRNA for 16 hours (day 0). On day 1, the transfection procedure was repeated. On day 2, each group of cells was pooled and replated for subsequent experiments. On days 5, 7, 9, 12, and 14, Hs683 cells were either scraped into cold PBS buffer (for RNA extraction) or lysed directly in boiling lysis buffer (10 mM Tris pH 7.4, 1 mM Na₃ VO₄, 1% SDS, pH 7.4) for protein extraction. Hs683 cells were also cultured on glass coverslips for immunofluorescence (IF) analyses as detailed previously [15,16]. The efficiency of the anti-Gal-1 siRNA was evaluated by means of Western blot (WB) analysis and/or IF and the efficiency of the anti-BEX2 siRNA was evaluated by means of IF and/or reverse transcription-polymerase chain reaction (RT-PCR).

In Vitro Transfection of Anti-BEX2 siRNA into Hs683 Cells and Their Grafting into the Brains of Immunocompromised Mice

Hs683 glioma cells were transfected in vitro with anti-BEX2 siRNA or CT siRNA as detailed above. Hs683 cells were grafted orthotopically into the brains of nude mice as described previously [31]. In each experiment, all mice (8-week-old female nu/nu mice, 21–23 g; Iffa Credo, Charles Rivers, Arbresle, France) had 100,000 Hs683 cells (wt, CT siRNA-transfected, or anti-BEX2 siRNA-transfected) stereotactically implanted on the same day (day 5 after siRNA transfection). Each experimental group contained 11 mice. All the in vivo experiments described in the present study were performed on the basis of Authorization No. LA1230509 of the Animal Ethics Committee of the Federal Department of Health, Nutritional Safety and the Environment (Belgium).

In Vivo Determination of Tumor Neoangiogenesis

Each mouse bearing a brain Hs683 glioma underwent euthanasia (in a CO₂ atmosphere) for ethical reasons when it had lost 20% of its body weight compared with that recorded on the day of tumor grafting. The brain was removed from the skull, fixed in buffered formalin for 48 hours, embedded in paraffin, and cut into 5-µm-thick sections.

After heat-induced antigen retrieval in citrate buffer (pH 6.0), sections were exposed to 0.4% hydrogen peroxide for 30 minutes to quench endogenous peroxydase. The sections were then incubated with an anti-CD34 antibody (diluted 1:25; Abcam, Cambridge, United Kingdom) for 1 hour at room temperature. For visualization of antigen location, the Poly-HRP IHC Detection Systems (ImmunoLogics, Duiven, the Netherlands) method and 3,3′-diaminobenzidine (Biogenex, Klinipath, Olen Belgium) were used. Finally, the sections were conterstained with hematoxylin.

The stained blood vessels were counted using a grid as indicated previously [31,32]. Typical blood vessels analyzed are illustrated in Figure 3B. Fifteen fields at a Gx40 magnification were analyzed per histological slide, giving thus a total of 165 histological fields per group.

Figure 3.

(A) Survival of immunocompromised mice with brain xenografts of wild-type nontransfected human Hs683 cells (red circles), Hs683 cells transfected in vitro with CT siRNA (black circles), or anti-BEX2 siRNA (blue circles). (Ba) Morphological illustration of the invasive patterns of Hs683 after orthotopic xenografts into the brains of immunocompromised mice (hematoxylin and eosin-stained; magnification, x10). NB indicates normal brain tissue; T, tumor. (Bb) Illustration of the blood vessels (black arrows) stained with an anti-CD34 antibody in Hs683 xenografts (magnification, x40). In this picture (wt condition), we can observe four blood vessels on one microscopic field. (C) Distribution of the blood vessel counts on 165 microscopic fields in wild type (wt: Ca), CT siRNA- (Cb), or anti-BEX2 siRNA- (Cc) transfected Hs683 orthotopic xenografts.

In Vitro Determination of the Capacity of Hs683 Cells to Undergo Vasculogenic Mimicry

Petri dishes, 35 mm in diameter, were precoated with Matrigel and then incubated at 37°C for 10 minutes. Hs683 wild-type cells, transfected with CT siRNA or transfected with anti-BEX2 siRNA (400,000 cells per Petri dish) were seeded on the coated Petri dishes at day 5 after transfection and cultivated for 48 hours. The ability of these cells to perform vasculogenic mimicry-like processes was assessed qualitatively at 24 and 48 hours after seeding as described previously [16].

In Vitro Determination of Anchorage-Dependent Cell Growth

Anchorage-dependent cell growth was assessed using the 3-[4,5-dimethylthiazol-2yl]-diphenyltetrazolium bromide (MTT; Sigma, Bornem, Belgium) colorimetric assay, as detailed elsewhere [33]. All determinations were carried out in sextuplicate.

In Vitro Determination of Anchorage-Independent Cell Growth

Anchorage-independent cell growth was assessed using the soft agar clonogenic assay as previously described [34]. All determinations were carried out in sextuplicate.

In Vitro Quantitative Determination of Cell Migration

Cell motility was characterized using computer-assisted phasecontrast microscopy (quantitative videomicroscopy), which enables the trajectories of living cells maintained in culture to be quantified [33,35]. The greatest linear distance migrated by each cell (out of hundreds of cells analyzed per experimental condition) was calculated from these trajectories [33,35]. All experiments were performed for 72 hours with one image recorded every 4 minutes. All determinations were carried out in triplicate.

In Vitro Determination of Hs683 Cell-Invasive Features

Invasive features of Hs683 cells in vitro were assessed by means of the Boyden transwell invasion system (BD BioCoat Matrigel invasion chambers; BD Biosciences Discovery Labware, Bedford, MA) as detailed elsewhere [35].

In Vitro Adhesion Assay

The adhesion assay was performed as described previously [36] with modifications. Six-well plates were precoated with Matrigel (diluted 1:3 in RPMI culture medium with 10% FCS) or with 10 µg/ml of the 120-kDa fragment of fibronectin (Chemicon, Biognost, Waterven, Belgium) for 1 hour at 37°C. The wells were washed, and nonspecific binding was then blocked with 0.1% BSA in PBS for 30 minutes. Hs683 cells (100,000 cells per well) were allowed to adhere for 1 hour at 37°C, after which nonadherent cells were gently washed away with warm PBS. Adherent cells were formalin-fixed, hematoxylin-stained, and counted in 10 fields per well at a Gx10 magnification using an Olympus microscope (Olympus, Antwerp, Belgium).

Determination of BEX1 and BEX2 Genomic Expression in a Clinical Series of 153 Gliomas and 23 Normal Brain Samples

The genomic expression of BEX1 and BEX2 was analyzed in a series of 176 human brain samples, including 23 normal brain tissues, 7 grade 2 astrocytomas, 19 grade 3 astrocytomas, 77 GBMs, 38 grade 2 oligodendrogliomas, and 12 grade 3 oligodendrogliomas. Microarray data were generated by the Henry Ford Hospital (Detroit, MI) from the Affymetrix Array Series GSE4290: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE4290.

Genomic and Proteomic Analyses

Genomic analyses

RNA was extracted from wt Hs683 cells or from those transfected with CT siRNA or specific siRNA directed against Gal-1 or BEX2, and the quality and integrity of the extracted RNA was assessed as detailed previously [33,35]. Full genome analyses were performed on day 5 of transfection at the VIB Microarray Facility (UZ Gasthuisberg, Catholic University of Leuven, Leuven, Belgium) using the Affymetrix Human Genome U133 set Plus 2.0 (High Wycome, UK). Microarray data analyses were undertaken as detailed previously [15,16,32].

Protein expression measurements

Western blot and IF analyses were performed as detailed previously [16,31–33]. Control experiments included the omission of the incubation step with the primary antibodies (negative control). Equal loading was verified by the bright Ponceau red coloration of the membranes. The integrity and quantity of the extracts were assessed by means of tubulin immunoblot analysis. The following primary antibodies were used for WB analysis: anti-Gal-1 (dilution 1:1000; Preprotech TebuBio, Boechout, Belgium), anti-integrin β₆ (dilution 1:100; Calbiochem, VWR International, Leuven, Belgium), anti-plexinC1 (dilution 1:1000; R&D Systems, Abingdon, United Kingdom), anti-VEGF (dilution 1:500; Santa Cruz; TebuBio), anti-switch-associated protein 70 (SWAP70; dilution 1:500; Novus Biologicals; DivBioscience, Breda, the Netherlands), and anti-IRE-1a (dilution 1:500; Cell Signaling, Bioké, Leiden, the Netherlands). Secondary antibodies were purchased from Pierce (PerbioScience, Erembodegem, Belgium) for the WB s and from Molecular Probes (Invitrogen, Merelbeke, Belgium) for fluorescent detection (Alexa Fluor-conjugated antibodies).

Western blots were developed using the Pierce Supersignal Chemiluminescence system (Pierce, Perbio Science N.V., Aalst, Belgium). The staining patterns were also assessed by immunofluorescence for BEX2 (dilution 1:50; Abcam) and integrin β₆ (dilution 1:50) and were analyzed by means of a computer-assisted fluorescent Olympus AX70 microscope (Omnilabo, Antwerp, Belgium) equipped with a MegaView2 digital camera and analysis software (Soft Imaging System, Munster, Germany), as detailed previously [16,33,35].

Reverse Transcription-Polymerase Chain Reaction

The procedure used for the standard RT-PCR was identical to that described previously [33,35]. The following primers were used: BEX2 forward 5′-gtgatcggccaacact-3′ and reverse 5′-tccaagcctatgcattatgtc-3′, BEX1 forward 5′-tagatgggaatatgatgcatagg-3′ and reverse 5′-caggtaaaggtttacaacaagg- 3′, actin forward 5′-ctaagtcatagtccgcctag-3′ and reverse 5′-aaatcgtgcgtgacattaagg-3′, and Gal-1 forward 5′-agcaacctgaatctcaaacc-3′ and reverse 5′-cttgaattcgtatccatctgg-3′. The PCR conditions for all primer pairs, except actin, were predenaturation: 4 minutes, 94°C, PCR amplification: 35 cycles at 94°C, 30 seconds (denaturation), 59°C, 30 seconds (annealing), 72°C, 1 minute (extension); and final extension: 1 cycle at 72°C, 10 minutes. For actin, the PCR conditions were predenaturation: 4 minutes, 94°C, PCR amplification: 25 cycles at 94°C, 30 seconds (denaturation), 62°C, 30 seconds (annealing), 72°C, 1 minute (extension); and final extension: 1 cycle at 72°C, 10 minutes.

Statistical Analyses

Survival analyses were carried out by means of Kaplan-Meier curves, which were compared with the log-rank test. Statistical comparisons between control and treated groups were established by carrying out the Kruskal-Wallis test (a nonparametric one-way analysis of variance). When this multigroup test was significant, post hoc tests (Dunn procedure) were used to compare the group pairs of interest, thus avoiding multiple comparison effects. Data correlation was analyzed by means of the nonparametric Spearman correlation index and its associated statistical test. All the statistical analyses were realized using Statistica (Statsoft, Tulsa, OK).

Results

BEX1 and BEX2 Expression in Human Malignant Gliomas and Human Normal Brain Tissues

We first analyzed the genomic expression of BEX2 and its closely related family member BEX1 in a series of glioma and normal brain tissue samples, and we observed that BEX1 genomic expression appreciably correlated to BEX2 genomic expression in gliomas (r = 0.71; P < .001 Spearman test; Figure 1A). However, the patterns of BEX1 and BEX2 expression differed as illustrated in Figure 1, B and C. Indeed, whereas BEX1 expression gradually decreased according to the sequence normal > oligodendrogliomas > astrocytomas > highly malignant astrogliomas (Figure 1B), BEX2 expression was globally lower in all glioma groups when compared with normal brain tissues (Figure 1C), with the exception that expression in the oligodendroglioma grade 2 group was higher than that in glioblastomas (Figure 1C).

Figure 1.

(A) Comparison of BEX1 versus BEX2 genomic expression in a series of 153 human glioma samples (see A and B). BEX1 (B) and BEX2 (C) genomic expression analyzed in 23 normal brain tissue samples (NB), 7 grade 2 astrocytomas (A2), 19 grade 3 astrocytomas (A3), 77 GBMs, 38 grade 2 oligodendrogliomas (OD2), and 12 grade 3 oligodendrogliomas (OD3). Each case is symbolized by a black dot, and the bars represent median values.

Decreasing Gal-1 Expression in Hs683 Oligodendroglioma Cells Induces a Decrease in BEX2 Expression

BEX1 and BEX2 are 85% identical [18], and therefore, to ensure that we pursued our studies solely on BEX2, BEX1 versus BEX2 expression was analyzed (by means of an RT-PCR technique) in four human permanent cell lines from the ATCC (including the Hs683 oligodendroglioma model) and in four astroglioma primocultures. Figure 2A shows that the oligodendroglioma Hs683 cell line and the U373 and T98G astroglioma cell lines do not express BEX1. When performing genome-wide analysis of Hs683 cells transiently transfected with siRNA targeting Gal-1 [15,16], it was observed that BEX2 was the gene whose expression was the most modified of a series of approximately 140 genes. Indeed, the levels of BEX2 mRNA were decreased by >20-fold in anti-Gal-1-transfected Hs683 cells compared with wt or CT siRNA-transfected cells as revealed in the whole-genome Affymetrix analysis or here by means of PCR analysis (Figure 2B). The decrease of BEX2 expression in anti-Gal-1-transfected cells was also confirmed by means of quantitative RT-PCR (data not shown).

Figure 2.

(A) Expression of BEX1 and BEX2 mRNA in eight different human glioma cell lines (four established cell lines, namely, Hs683, U373, T98G, and U87, and four primocultures, namely, GL-5, GL-16, GL-17, and GL-19) assessed by RT-PCR. (B) Expression of BEX2 mRNA in human Hs683 cells: wild-type non-transfected (wt), CT siRNA-transfected (CT siRNA), or anti-Gal-1 siRNA-transfected (aGal-1-siRNA) on days 7 and 9 after transfection, assessed by RT-PCR. (C) Expression of BEX2 mRNA in Hs683 cells: wild-type nontransfected (wt), CT siRNA-transfected (CT siRNA), or anti-BEX2 siRNA-transfected (aBEX2 siRNA) on days 5 and 6 after transfection, assessed by RT-PCR. (D) Immunofluorescence analysis of BEX2 expression in Hs683 cells on day 7 after transfection in wild-type nontransfected (Da) or CT siRNA-transfected (Db) or anti-BEX2 siRNA-transfected (Dc). (Dd) BEX2 expression quantitatively determined by means of computer-assisted fluorescence microscopy in wild type (wt) and transfected Hs683 cells.

Confirmation That the siRNA Approach Decreases BEX2 Expression in Hs683 Cells

Figure 2C shows that the anti-BEX2 siRNA we designed successfully decreased BEX2 mRNA levels in Hs683 oligodendroglioma cells up to days 5 and 6 after transfection, whereas the levels of Gal-1 mRNA remained unchanged. The levels of BEX2 protein (quantitatively determined by means of computer-assisted fluorescence microscopy) were also significantly decreased up to day 7 after transfection (Figure 2D). Additional experiments showed that the decrease in BEX2 mRNA levels in Hs683 cells lasted for 8 to 9 days (data not shown).

Decreasing BEX2 Expression in Hs683 Cells Increases the Survival of Hs683 Orthotopic Xenograft-Bearing Mice

Immunocompromised mice grafted with Hs683 oligodendroglioma cells transfected with anti-BEX2 siRNA revealed a significant (P < .05) but modest increase in survival (blue circles) compared with mice grafted with wt (black circles) or CT siRNA-transfected Hs683 cells (red circles; Figure 3A). This experiment has been conducted twice with similar results (data for the second experiment are not shown).

BEX2 Is Implicated in Hs683 Oligodendroglioma Neoangiogenesis In Vivo

As Gal-1 has been shown to play an important role in angiogenesis, we investigated whether decreasing BEX2 expression in Hs683 oligodendroglioma orthotopic xenografts in immunocompromised mice (Figure 3Ba) impaired neoangiogenesis features. For this purpose, 11 mice per group orthotopically grafted with wt, control, or anti-BEX2 siRNA-transfected Hs683 cells underwent euthanasia on day 17 after tumor grafting. The blood vessels in the tumor were stained by immunohistochemistry using an anti-CD34 antibody. The typical blood vessels taken into account to assess angiogenesis in Hs683 tumors are illustrated in Figure 3Bb. Decreasing BEX2 expression in Hs683 xenografts significantly (P < .01) decreased the number of blood vessels within these tumors (Figure 3Cc) compared with controls (Figure 3, Ca and Cb).

Ability to Elicit Vasculogenic Mimicry Differs in Wild-type, CT siRNA-, and Anti-BEX2 siRNA-Transfected Hs683 Cells In Vitro

Gross histological examination as performed and illustrated in Figure 3B cannot distinguish between true angiogenesis and vasculogenic mimicry processes. Aggressive tumors, including, for example, melanomas, may generate tumor cells that form vascular channels that facilitate perfusion independent of tumor angiogenesis [37]. Vasculogenic mimicry has also been shown to be a feature of gliomas [38]. In addition, we have reported previously that Gal-1 is implicated in vitro in vasculogenic mimicry performed by Hs683 oligodendroglioma cells [16]. Accordingly, differences in the patterns of vasculogenic mimicry between wt, CT siRNA-, and anti-BEX2 siRNA-transfected Hs683 cells were investigated. The data reveal that when wt (Figure 4A) and control-transfected (Figure 4B) Hs683 cells were cultured on Matrigel, networking processes resembling vasculogenic mimicry were observed. In contrast, anti-BEX2 siRNA-transfected Hs683 cells no longer demonstrated these vasculogenic processes (Figure 4C). Figure 4D confirms that the siRNA against BEX2 actually decreased its expression compared with wt and CT siRNA-transfected Hs683 cells, although this decrease of BEX2 expression was not as strong as what was obtained with the previous transfection (Figure 2C).

Figure 4.

Wild-type (A) and CT siRNA-transfected (B) Hs683 cells developed vasculogenic capillary networks when cultured on Matrigel, whereas Hs683 cells (C) transfected with anti-BEX2 siRNA did not. (D) Confirmation of decrease in BEX2 mRNA in anti-BEX2 siRNA-transfected (aBEX2 siRNA) compared with wild-type nontransfected (wt) and CT siRNA-transfected (CT siRNA) Hs683 cells on days 5 and 7 after transfection, assessed by RT-PCR.

Decreasing BEX2 Expression in Hs683 Cells Impairs Their Migration Characteristics

Reducing BEX2 expression in Hs683 cells seemed to have no impact on anchorage-dependent or -independent growth, cell cycle kinetics or proliferation rate, cell motility when cultured on plastic or Matrigel, susceptibility to apoptosis, or increased sensitivity to two cytotoxic drugs investigated (TMZ and carmustine). These data are summarized for the sake of clarity in Table 1.

Table 1.

Summary of Experiments Undertaken to Investigate BEX2-Mediated Biological Effects in Human Hs683 GBM Cells.

Cell Biology Feature Evaluated	Techniques Used	References Fully Detailing Technique Used	Comments to Data Obtained in the Current Study
Cell population kinetics
Anchorage-dependent global cell growth	MTT colorimetric assay	[39,40]	Decreasing BEX2 expression in Hs683 cells did not modify their anchorage-dependent growth.
Anchorage-independent cell growth	Soft-agar clonogenic assay	[34]	Decreasing BEX2 expression in Hs683 cells did not modify their anchorage-independent growth.
Cell proliferation (cell cycle kinetics)	Flow cytometry analysis of propidium iodide-stained cells	[32]	Decreasing BEX2 expression in Hs683 cells did not modify cell cycle kinetics or their proliferation rate.
Cell death (apoptosis)	Flow cytometry analysis of TUNNEL-propidium iodide-stained cells	[32]	Decreasing BEX2 expression in Hs683 cells did not induce apoptosis.
Cell migration Motility	Computer-assisted phase-contrast microscopy (quantitative videomicroscopy)	[10,11,35]	Decreasing BEX2 expression in Hs683 cells did not modify their motility levels when cultured on plastic or Matrigel.
Sensitivity to anticancer drugs
Anchorage-dependent cell growth	MTT colorimetric assay	[15]	Decreasing BEX2 expression in Hs683 cells did not modify their sensitivity to the cytotoxic effects of TMZ or BCNU.

However, striking effects relating to cell migration including adhesion and invasion were observed when decreasing BEX2 expression. Indeed, decreasing BEX2 expression significantly (P < .001) increased the levels of adhesion of Hs683 cells cultured in Matrigel or in fibronectin (Figure 5A). In addition, although quantitative videomicroscopy revealed that impairing BEX2 expression did not modify Hs683 two-dimensional cell motility when cultured on plastic or Matrigel (Table 1), decreasing BEX2 expression significantly impaired their ability to invade (three-dimensional) Matrigel-coated membranes in Boyden chamber assays (Figure 5B).

Figure 5.

(A) Characterization of the adhesive properties of wild type (wt), control siRNA- (CT siRNA), or anti-BEX2 siRNA-transfected (aBEX2 siRNA) Hs683 cells on fibronectin (gray bars) or on Matrigel (black bars) supports. Values are the mean ± SEM; ***P < .001 compared with CT siRNA. (B) Characterization of Hs683 cell invasiveness in Boyden chambers coated with Matrigel of wild type (wt), CT siRNA-, or anti-BEX2 siRNA-transfected cells after 24 (gray bars) or 48 hours (black bars). Values are the mean ± SEM; ***P < .001 compared with CT siRNA.

Decreasing BEX2 Expression in Hs683 Cells Modifies the mRNA Levels of Several Genes Implicated in Cell Migration

Previous studies and present data show that 1) Gal-1 exerts potent promigratory effects [10–12], 2) impairing Gal-1 expression reduces BEX2 expression (Figure 2B), and 3) BEX2 seems to be implicated in migration (Figure 5). For these reasons, we carried out microarray analysis to compare gene expression in wt, anti-BEX2 siRNA-transfected, and control siRNA-transfected Hs683 cells. Of the 26 genes whose expression was significantly upregulated or downregulated (i.e., ≥1.6- or ≤0.6-fold, respectively) 8 are known to exert a role in cell migration (MAP2, ITGB6, VIL1, MMP7, PLXNC1, CCL2, NT5E, and SWAP70; Table 2).

Table 2.

Migration-Related Genes Whose mRNA Expression Was Modified in Hs683 GBM Cells with Reduced BEX2 Expression.

Official Gene Symbol	Gene Name	Ratio 1: wt/aBEX2 siRNA	Ratio 2: CT siRNA/aBEX2 siRNA	References to Cell Migration
MAP2	Microtubule-associated protein 2	0.4	0.5	[41,42]
ITGB6	Integrin beta 6	0.4	0.5	[43,44]
VIL1	Villin 1	0.4	0.5	[45,46]
MMP7	Matrix metalloproteinase 7 (matrilysin, uterine)	0.6	0.5	[47]
PLXNC1	Plexin C1	0.3	0.6	[48]
CCL2	Chemokine (C-C motif) ligand 2	0.6	0.6	[49–51]
NT5E	5′-Nucleotidase, ecto (CD73)	1.6	1.7	[52,53]
SWAP70	SWAP-70 protein	1.6	2.1	[54,55]

Ratios of gene expression in 1) wild type (wt)/anti-BEX2 siRNA-transfected Hs683 cells and 2) CT siRNA/anti-BEX2 siRNA-transfected Hs683 cells.

Figure 6 illustrates the data obtained at the protein level with respect to four of the eight genes listed in Table 2: microtubule-associated protein 2 (MAP2), plexin C1, switch-associated protein 70 (SWAP70), and integrin β₆. Transiently decreasing BEX2 expression in Hs683 cells markedly increased MAP2 expression, slightly increased plexin C1 expression, and abolished SWAP70 expression (Figure 6A). Decreasing BEX2 expression also markedly increased integrin β₆ expression (Figure 6B). Moreover, IF analysis revealed a diffuse staining pattern for integrin β₆ in Hs683 cells transfected with control CTsiRNA (Figure 6, Ca and Cb), whereas marked integrin β₆ clustering at the level of lamelipodia became apparent when decreasing BEX2 expression (Figure 6, Cc and Cd).

Figure 6.

(A) Western blot analysis of MAP2, plexin C1, and SWAP70 expression in wild type (wt), CT siRNA-transfected (CT siRNA), anti-BEX2 siRNA-transfected (aBEX2 siRNA), and anti-Gal-1 siRNA-transfected (aGal-1 siRNA) Hs683 cells on day 5 after transfection. (B) Western blot analysis of integrin β₆ expression in wt, CT siRNA, aBEX2 siRNA-transfected, and aGal-1 siRNA-transfected Hs683 cells on day 5 after transfection. (C) Immunofluorescence analysis (with bright field controls) of integrin β₆ expression in CT siRNA- (Ca and Cb) and aBEX2 siRNA-transfected (Cc and Cd) Hs683 cells on day 5 after transfection.

Moreover, transiently decreasing Gal-1 expression in Hs683 cells also increased MAP2 and plexin C1 (Figure 6A). These data strongly suggest that BEX2 could exert its modulatory influences on MAP2 and plexin C1 expression under Gal-1 control, whereas it could exert such an influence on integrin β₆ and SWAP70 without Gal-1 control (Figure 6A).

Discussion

Galectin 1 has been shown to be a key player in astroglioma and oligodendroglioma cell migration [10–12] and oligodendroglioma neoangiogenesis [16] and chemoresistance [15]. In a previous study, we observed that the gene whose expression is the most decreased in response to siRNA-targeted against Gal-1 in Hs683 oligodendroglioma cells is BEX2 [16]. It would also be interesting to analyze the levels of expression of BEX2 in oligodendroglioma cell types that are not expressing Gal-1 and to investigate whether exogenous addition of Gal-1 will stimulate BEX2 expression. However, we could not find any oligodendroglioma (and even astroglioma) cell lines that do not express Gal-1.

In this study, BEX2 expression has been found to be lower in glioblastomas than in normal brain tissues and low-grade oligodendrogliomas (Figure 1). This could relate to the fact that epigenetic processes may silence BEX2 expression in glioblastomas [21]. When Foltz et al. [21] treated astroglioma cells with trichostatin A, a histone deacetylase inhibitor or with 5-aza-2′-deoxycytidine, a DNA methytransferase inhibitor, to identify epigenetically silenced genes, they identified BEX1 and BEX2 to be silenced in all tumor specimens because of extensive promoter hypermethylation. Viral-mediated reexpression of either BEX1 or BEX2 led to increased sensitivity to chemotherapy-induced apoptosis and showed potent tumor suppressor effects in vitro and in a xenograft mouse model in vivo [21]. In sharp contrast to the data reported by Foltz et al., Naderi et al. [22] could not detect any correlation in their series of breast cancers between BEX2 methylation and expression. The BEX2 promoter is not methylated in Hs683 oligodendroglioma cells (data not shown). Naderi et al. [22] also showed that BEX2 was overexpressed in a subset of primary breast cancers and that this expression was necessary for the antiapoptotic function of the nerve growth factor in C2-treated cells. These authors suggest that the nerve growth factor/BEX2/nuclear factor-κβ pathway could be involved in regulating apoptosis in breast cancer cells and in modulating the response to tamoxifen in primary breast cancers [22]. In the current study, decreasing BEX2 expression in Hs683 oligodendroglioma cells did not increase their sensitivity to TMZ or carmustine or lead to modifications in cell proliferation, cell death (apoptosis), or anchorage-dependent or -independent growth (Table 1). In addition, in contrast to the study of Foltz et al. [21], which relied on the analysis of astroglial cells, transiently decreasing BEX2 expression in Hs683 oligodendroglioma cells induced modest but nevertheless significant increases in the survival of Hs683 orthotopic xenograft-bearing mice. It thus seems that BEX2 biological functions and activity could markedly differ between astroglioma and oligodendroglioma cells, a feature which we already emphasized in the introduction with respect to the gene products PDGF [5], EGFR [7], vimentin [25], integrin β₄ [24], ATAD 3B [26], and Notch2 [27].

Decreasing BEX2 expression in Hs683 oligodendroglioma cells impairs angiogenesis inHs683 xenografts when these cells are orthotopically grafted in vivo into the brains of immunocompromised mice (Figure 3). It is worthwhile to note that the type of staining we performed with CD34 does not allow for differentiation between tumor neoangiogenesis and normal vessels trapped within tumor tissues. The use of antibodies directed against specific markers of tumor endothelial cells, such as CD105 [56] or Dkk-3 [57], would be useful to determine if BEX2 is indeed involved in tumor neoangiogenesis. However, although these markers have been proven effective to determine neoangiogenesis through immunohistochemistry in human tissue [56,57], we failed to be able to use these markers in the present study to make this determination in mouse tissue (data not shown).

A previous study [16] and the current one revealed that BEX2 expression is controlled, at least partly, by Gal-1. Galectin 1 has been shown to exert its proangiogenic effects, at least partly, through the endoplasmic reticulum transmembrane kinase/ribonuclease IRE-1α-mediated control of the expression of oxygen-related protein 150, which, in turn, controls VEGF expression [16]. Once more, the patterns of VEGF expression [58] and vasculature profiles [59] differ between astrogliomas and oligodendrogliomas. Transiently decreasing BEX2 expression in Hs683 cells did not modify VEGF expression and decreasing IRE-1α expression in these Hs683 cells did not modify BEX2 expression (data not shown). This suggests that Gal-1-mediated control of BEX2 is independent of the activation of IRE-1α and that the BEX2-related effects on angiogenesis and on vasculogenic mimicry are therefore IRE-1α-independent. The signaling pathway by which BEX2 exerts its effects on angiogenesis remains to be elucidated.

Decreasing BEX2 expression in Hs683 oligodendroglioma cells markedly impacts cell migration features in vitro on Matrigel or on fibronectin supports with respect to both cell adhesion and cell invasion (Figure 5). Maniotis et al. [60] suggested that formation of vasculogenic mimicry channels is related to the invasive ability of tumor cells. The fact that decreasing BEX2 expression impairs the invasive properties of Hs683 glioma cells (Figure 5B) might then partly explain the effects observed on vasculogenic mimicry in vitro (Figure 4) and angiogenesis in vivo (Figure 3). The effects observed on cell adhesion and invasion could be partly explained by the fact that decreasing BEX2 expression induces the up or down-regulation of a number of genes involved in migration (Table 2) including notably MAP2, plexin C1, integrin β₆, and SWAP70.

MAP2, a member of the microtubule-associated proteins that can bind both microtubules and F-actin [61], has been shown to play a role in cell migration [41,42]. Interestingly, we have shown previously that the knockdown of Gal-1 expression in U87 astroglioma cells also decreased their motility and induced an increase in MAP2 expression [12].

Plexin C1 is a member of the plexin family of semaphorin receptors. Semaphorin binding to plexin receptors can regulate actin cytoskeleton rearrangement through the regulation of the activity of small GTPases of the Rho and Rac families [62]. Galectin 1 controls, at least partly, glioma cell migration through RhoA-mediated modifications of the actin cytoskeleton [11]. Semaphorins have also been shown to regulate adhesive and migratory properties of malignant cells [63], at least through modulation of integrin-mediated adhesion and migration [48]. Decreasing BEX2 in Hs683 oligodendroglioma cells induces a marked increase in integrin β₆ expression (Figure 6B) and its clustering in lamelipodia (Figure 6C). The expression of integrin β₆ has been shown to increase migration and the invasive behavior of oral squamous cell carcinomas [43]. Our observation that the localization of integrin β₆ changed toward clustering in lamelipodia (Figure 6C) when BEX2 expression decreases could possibly explain the resulting phenotype of increased adhesion (Figure 5A) and reduced invasion (Figure 5B). We recently showed that depletion of Gal-1 through both stable knockdown and transient-targeted siRNA treatment against Gal-1 induces an intracellular accumulation of integrin β₁ coincident with a diminution of integrin β₁ at points of cellular adhesion at the cell membrane in Hs683 oligodendroglioma and U87 astroglioma cells [64]. Galectin 1 depletion does not alter the gene expression level of integrin β₁ but effectuates the perinuclear accumulation of protein kinase C epsilon and the intermediate filament vimentin, both of which have been shown to mediate integrin recycling in motile cells [64].

Finally, it was observed that transiently decreasing BEX2 expression in Hs683 cells strongly reduced the expression of SWAP70 (Figure 6), a form of Rac-GEF that transduces signals from tyrosine kinase receptors to Rac [65]. SWAP70 has been shown to be associated with a subset of actin filaments generated in lamelipodia and the membrane ruffles of motile cells [54]. Moreover, SWAP70 has also been shown to modulate cell adhesion to fibronectin, homotopic cell aggregation, and migration [55].

Moreover, we have shown that the decrease of Gal-1 expression also impairs the expression of MAP2 and plexin C1. These results tend to confirm that Gal-1 could control the expression of BEX2, which, in turn, modulates the expression of genes involved in cell migration.

In conclusion, the present study has revealed Gal-1-mediated control of BEX2 expression and the involvement of BEX2 in oligodendroglioma neoangiogenesis and oligodendroglioma cell migration. It is also suggested that these effects may be partly explained by modulation of the expression of a number of genes such as MAP2, plexin C1, integrin β₆, and SWAP70 that have already been shown to play a role in cancer cell migration.

Abbreviations

BEX

brain-expressed X-linked gene

GBM

glioblastoma

MAP2

microtubule-associated protein 2

SWAP70

switch-associated protein 70

TMZ

temozolomide