Regulators of G-Protein Signaling 3 and 4 (RGS3, RGS4) Are Associated with Glioma Cell Motility

Abstract. Diffuse brain invasion is a major reason for poor prognosis of glioma patients. The molecular mechanisms underlying infiltration are different from those of other cancer types. To detect genes associated with glioma invasion, highly migratory clones were selected from U373MG glioma cells and from primary glioblastoma cells, and the gene expression pattern of these “fast” cells was compared with that of the original (“slow”) cells using oligonucleotide microarrays comprising 12,625 genes. A total of 28 genes were differently expressed in both primary and established cell populations, including 19 genes that were upregulated and 9 that were downregulated in fast cells. Most of these genes have not been linked to glioma invasion so far. Specifically, differentially expressed genes included those encoding extracellular matrix components (COL16A1, DPT), proteases (CATD, PRSS11), cytokines (MDK, 1L8), transport proteins (SLC1A3, ATP10B), cytoskeleton constituents (ACTA2, ACTSG, NEFL), DNA repair enzymes (WRN, ADPRTL2), and G- protein signaling components (GNA 12, RGS3, RGS4). RGS3 and RGS4, which are homologs of the Drosophila glia gene loco, were further functionally analyzed. U373MG glioma cell clones overexpressing RGS3 or RGS4 showed an increase of both adhesion and migration. These findings expand the spectrum of possible molecular pathways underlying the invasion of neoplastic astrocytes. Specifically, they suggest that RGS proteins and G-protein-mediated signal transduction are evolutionary conserved functional players.

Key Words: Astrocytoma; Glioblastoma; Glioma; G-protein; Invasion; Microchip; Migration.

INTRODUCTION

Glioblastoma is the most common and most malignant human brain tumor. Despite aggressive therapy, the prognosis has not remarkably improved in the last decades (1). A major reason for the unfavorable outcome of patients with a glioblastoma is the widespread, diffuse invasion of neoplastic astrocytes into brain tissue, which prevents complete surgical resection of the tumor (2). During the past few years it has become common belief that understanding the molecular mechanisms underlying this characteristic and eventually fatal invasion pattern will result in novel and potentially effective approaches in glioma management (3). Accordingly, research in this area has been greatly intensified and a large number of proteins have been identified that are correlated with and possibly involved in glioma invasion, including adhesion molecules, extracellular matrix components, proteases, growth factors, and many others (4, 5).

A major problem of current experimental research in glioma invasion is that experiments are usually based on the candidate approach, i.e. on the hypothesis that proteins or genes known to be involved in the migration/ invasion of other cells (normal non- neural cells in vitro, glia cells in the developing brain, and extracerebral cancer) also mediate migration of neoplastic astrocytes through the human brain in vivo. This approach unavoidably restricts the number of genes amenable to analysis and may miss the most relevant genes, i.e. those specific to glioma invasion. Due to the characteristic differentiation pattern of astrocytes, the unique cerebral extracellular matrix and the distinct transcerebral routes along specific basement membranes and white matter tracts, major aspects of glioma invasion are quite distinctive and different from other migration processes. However, those aspects might represent the most effective therapeutic targets.

Here we have selected cell populations with a highly migratory phenotype derived from an established glioblastoma cell line and from primary glioblastoma cells. The expression pattern of these “fast” cells was then compared with that of the parental (“slow”) cells by using oligonucleotide microarrays. DNA array analysis has been applied to gliomas in order to compare tumor versus normal tissue (6-8), high-grade versus low-grade gliomas (9, 10), cells in the center versus the periphery of the tumor (11), and tumor cells exposed versus not exposed to a certain extracellular matrix (12). Since microarray experiments are not hypothesis-driven, they are able to reveal novel, possibly pathogenetically relevant or even specific genes, and therefore appear promising for tackling mechanisms mediating glioma invasion. Using stringent statistical criteria, our experiments revealed 28 genes that were differentially expressed in slow versus fast cells in both primary and established glioma cell populations. Most of these genes have not been associated with glioma migration before. Two of these genes (RGS3 and RGS4), both of them being homologs of the Drosophila glia gene loco, were further functionally analyzed and shown to promote glioma cell migration.

MATERIALS AND METHODS

Cell Culture

Human U373MG glioma cells (European Collection of Cell Cultures, Salisbury, Wiltshire, UK) were grown in Dulbecco’s modified Eagle’s minimal essential medium (DMEM, Biochrom, Berlin, Germany), supplemented with 10% fetal calf serum (Biochrom), penicillin G (100 U ml^sup -1^, Biochrom) and streptomycin sulfate (100 g ml^sup -1^, Biochrom). Cells were incubated under standard conditions (37C in a 5% CO2 humidified atmosphere) and passaged using trypsinization.

Human TB288 primary glioma cells were prepared from a temporal glioblastoma specimen of a 59-year-old male patient undergoing tumor resection at the University Hospital Muenster. The tumor tissue was finely minced with 2 scalpels and incubated in DMEM with supplements as above in a 10-cm-diameter cell culture dish (Nunc, Wiesbaden, Germany), under standard conditions for 48 h. Subsequently, the medium containing cell and tissue debris was aspirated and the plate was rinsed twice with phosphate buffered saline (PBS, Biochrom). The remaining adherent cells were overlaid with medium.

Selection for Migration

Highly migratory U373MG glioma cells were selected using a modification of the technique reported by McDonough et al (13). Round coverslips of 10-mm diameter (Piano, Wetzlar, Germany) were attached to the center of each well in a 24-well cell culture plate (Nunc) using silicone (KAWO-SL 59, KAWO, Hildesheim, Germany). Immediately before seeding, each well was coated with 500 of matrigel (BD Bioscience, Erembodegem, Belgium) solution containing 100 g ml^sup -1^ in serum-free DMEM. Sedimentation cylinders with an inner diameter of 1.7 mm were then placed in the center of the coverslips of each well and 2,000 cells were seeded in a volume of 1 l DMEM. Sedimentation cylinders were removed 16 h after seeding and the cells were allowed to migrate over the border of the round coverslips. Then the coverslips containing the slower cells were removed and the faster cells in the wells of the culture plate were collected by trypsinization and used for another cycle of selection. After the first cycle of selection, the cells remaining on coverslips were also collected and designated U373slow. Parallel culture of these cells was maintained without selection pressure during selection experiments. The slower cells in the following cycles of selection were discarded. A total of 10 rounds of selection were performed and the resulting cells were designated U373fast.

A highly migratory population of TB288 glioma cells was obtained in the same way, although only 1 cycle of selection was performed to avoid possible alteration of the original features of the tumor cells during extended culture time. For the same reason the 24-well culture plate was not coated with matrigel. The resulting cells were designated TB288slow and TB288fast, respectively.

In Vitro Migration Assay

In vitro migration assays were performed on coated Permanox LabTek ChamberSlides(TM) with 8 chambers (Nunc) as described by Berens et al (14) with some modifications. Briefly, slides were incubated with 100 l per chamber of matrigel solution containing 100 g ml^sup -1^ in serum-free DMEM. After 1-hour incubation under normal culture conditions, chambers were washed 3 times with PBS and incubated with 100 l of a BSA solution (1 mg ml^sup -1^ in PBS, sterile filtered) for 30 min at room temperature. After additional washing with PBS, slides were used for migration assays. Sterile sedimentation cylinders (1.2-mm inner diameter, 7-mm outer diameter) were filled with medium and placed into the chambers. Two thousand cells were seeded in a volume of 1 l DMEM. For sedimentation and adhesion the slides were incubated for 16 h under standard culture conditions. After removing the sedimentation cylinders medium was gently aspirated and replaced by 200 l DMEM per chamber. At this time cells had formed round colonies and started migrating on the substrate. Colonies were fixed with 3.7% formaldehyde/PBS directly after removing the sedimentation cylinders (t = 0) and after intervals of 24 h for up to 72 h (U373MG) and 120 h (TB288). Cells were stained with Coomassie blue (0.05% w/v in 50% methanol, 10% acetic acid). Experiments were performed as 8-fold replicates.

Morphometry

Distance of migration was determined by using an Olympus BX50 microscope at 12.5-fold magnification, a digitizing CCD camera Olympus DP10 (Olympus, Hamburg, German\y), and Sigma Scan Pro 4 image analysis software (SPSS Science, Chicago, IL). The colonies were surrounded by using the distance and area measurement tool of the software, and the colony area was automatically determined. The data were exported to an Excel database and statistically evaluated.

Cell Adhesion Assay

After coating a 96-well microtiter plate (Nunc) with matrigel as described before, 1 10^sup 5^ cells per well were incubated for 1 h under standard conditions. Then medium and nonadherent cells were aspirated and the wells were washed 3 times with PBS. Cells were fixed with 100 l 3.7% formaldehyde/PBS for 15 min and stained with additional 100 l 0.5% toluidine blue in 3.7% formaldehyde/PBS overnight at 4C. After aspirating the dye solution, cells were washed 4 times with ddH^sub 2^O, air-dried, and lysed in 100 l 2% SDS on a shaker for 20 min. Absorption was measured at 620 n m using an ELISA reader. Experiments were performed as 5-fold replicates.

Cell Proliferation Assay (MTT Assay)

Proliferation was assessed with colorimetric [3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] (MTT) assay. After incubating the cells for 6 h to become adherent, medium was aspirated and 200 l of MTT solution (0.5 mg ml^sup -1^ in DMEM containing 10% fetal calf serum, 100 U ml^sup -1^ penicillin G, and 100 g ml^sup -1^ streptomycin sulfate) were added. After an additional 3 h, the MTT solution was discarded and 200 l of isopropanol were added to dissolve formazan crystals. Absorption was measured at 570 nm using an ELISA reader. For calculating cell number, absorbance values were compared with the standard curve for the appropriate cell line. Experiments were performed as 6-fold replicates.

Cell Viability Assay (LDH Release Assay)

After incubation of the cells for 6 h, 200 l of new medium was added to each well and cells were incubated for an additional 3 h. One hundred l supernatant of each well was transferred to a new microtiter plate and mixed with 100 l of reaction mixture according to supplier’s protocol (LDH Release Assay, Roche, Basel, Switzerland). Probes were incubated at room temperature protected from light for 30 min. Afterwards, absorbance was measured at 492 nm using an ELISA reader. For calculating cell viability, absorbance values were normalized against a no cell control and compared with the appropriate standard curve. Experiments were performed as 6- fold replicates.

Immunocytochemistry

Ten thousand TB288 glioma cells in 200 l DMEM were grown on glass slides using 1-cm-diameter cloning cylinders (Sigma, Taufkirchen, Germany). After cells had become adherent, cylinders were removed, cells were gently washed in PBS, and fixed with 3.7% formaldehyde in PBS (containing 1% Triton-X 100) for 10 min. Immunocytochemical staining was performed automatically in a Horizon TechMate apparatus (Dako, Hamburg, Germany) using the DAB-system kit according to manufacturer’s instructions. Primary antibodies used included those against GFAP, muscle actin (clone HHF35), CD31, CD34, S-100 protein, cytokeratins (clone KL1), vimentin (clone V9), and the proliferation antigen Ki-67 (clone MIB1). Cells were lightly counterstained with hematoxylin. For detection of RGS3 in U373MG cells using indirect immunofluorescence, cells grown for 24 h on matrigel-coated glass slides were fixed in methanol (-20C). RGS3 was detected using 2 g/ ml goat anti-RGS3 antibody (sc-9303, Santa Cruz Biotechnology, Santa Cruz, CA) and FITC-conjugated donkey anti-goat IgG secondary antibody (Dianova, Hamburg, Germany).

Western Blot

Cells and tissue were lysed with buffer containing 20 mM Tris, pH 7.6, 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1% Nonidet-P 40. Equal amounts of proteins as determined using Lowry assay (Bio-Rad, Hercules, CA) were separated on a 15% (RGS4) or 10% (RGS3) polyacrylamide gel with subsequent transfer onto nitrocellulose membrane (Schleicher and Schull, Dassel, Germany) using the wet blot technique. RGS3 was detected with 2 g/ml sc-9303 antibody and peroxidase-conjugated mouse anti- goat IgG secondary antibody A-9452 (Sigma) using the ECL plus detection kit (Amersham, Braunschweig, Germany). RGS4 was detected with a rabbit anti-RGS4 polyclonal anti-body (kind gift from Dr. Kirk M. Druey, NIAID, NIH, Bethesda, MD) at a dilution of 1:500 and peroxidase-conjugated mouse anti-rabbit IgG secondary antibody A- 2074 (Sigma) using the ECL plus detection kit (Amersham). To verify equal protein loading on each lane, blots were stripped and reprobed for actin. The intensity of protein bands was densitometrically detected using Gel-Pro Analyzer(TM) Software (Version 3.0, Media Cybernetics Inc., Silver Spring, MD).

Oligonucleotide Microarray Analysis

The Affymetrix Human Genome U95 Av2 Array (HG U95 Av2, Affymetrix, Santa Clara, CA) is complementary to 12,625 genes and ESTs. Affymetrix GeneChip analysis was performed according to the instructions of the Expression Analysis Technical Manual (Affymetrix). Cells were grown in 6-cm cell culture dishes with (U373MG cells) or without (TB288 cells) matrigel. Total RNA was extracted at approximately 80% confluency using the RNeasy kit (Qiagen, Hilden, Germany). Double-stranded cDNA was synthesized from total RNA using Gibco BRL Superscript Choice System (Invitrogen, Karlsruhe, Germany) and a T7-(dT)^sub 24^ primer (5′- GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)^sub 24^-3′). After phenol/chloroform extraction of double-stranded cDNA, synthesis of biotin-labeled cRNA was performed using the ENZO BioArray(TM) HighYield(TM) RNA Transcript Labeling Kit (ENZO Life Sciences, Farmingdale, NY). After purification with RNeasy kit (Qiagen), ethanol precipitation, and quantifying, the biotinylated cRNA was fragmented by alkaline treatment. Fifteen g of adjusted fragmented cRNA in a hybridization cocktail containing eukaryotic hybridization controls were hybridized to a single GeneChip Probe Array for 16 h. After washing and staining, the probe arrays were scanned in an Affymetrix GeneChip scanner. Data (.cel-files) were exported to a Cobi database and analyzed using GeneData Expressionist software (GeneData, Basel, Switzerland). Duplicate experiments were performed. Significant differences between expressed genes as well as expressed versus nonexpressed genes were examined.

Real-Time RT-PCR

To quantitate RGS3 and RGS4 transcripts, TaqMan assays were performed (TaqMan Universal PCR Master Mix and GeneAmp 5700 Sequence Detection System, Applied Biosystems, Foster City, CA) using cells that had been propagated for at least 5 serial passages after the end of selection for enhanced migration and microarray analysis. Total RNA was isolated as described above. Template cDNA was synthesized out of 2 g total RNA using the Omniscript(TM) Reverse Transcriptase Kit (Qiagen) and the T7-(dT)^sub 24^ primer, following manufacturer’s instructions. Forward and reverse primers and fluorescence tagged probe used for the RGS3 gene were 5′- TGGCACAGAAGCGCATCTT, 5′-GTTAATAAGGTCCAGGTAGAGGTCAGA and 5′-6FAM- AAGGACTCGTACCCTCGCTTTCTCCG-TAMRA, respectively. Forward and reverse primers and fluorescence tagged probe used for the RGS4 gene were 5′- CCGCTTCCTCAAGTCTCGAT, 5′-CAGGTGAGAATTAGGCACACTGA and 5′-6FAM- CTATCTTGATTTGGTCAACCCGTCCAGCT-TAMRA, respectively. Forward and reverse primers and fluorescence tagged probe used for the GAPDH housekeeping gene were 5′-ACCCACTCCTCCACCTTTGAC, 5′- CATACCAGGAAATGAGCTTGACAA and 5′-6FAM-CTGGCATTGCCCTCAACGACCA-TAMRA, respectively. Cycling conditions were 50C for 2 min, 95C for 10 min, followed by 40 cycles of 95C for 15 s and 60C for 1 min. The number of cycles necessary to produce a detectable amount of product was recorded for each sample and the ratio of gene expression was calculated as 2^sup (difference in cycle number)^, normalized against GAPDH. Experiments were performed in quadruplicate.

Plasmids and Transfection

To obtain overexpression of RGS3 and RGS4, the vector pRc-CMV (Invitrogen, Groningen, The Netherlands) containing human RGS3 cDNA and the vector pCR 3.1 (Invitrogen) containing human RGS4 cDNA (both gifts from Dr. Thomas Wieland, Mannheim, Germany) were transfected into U373MG glioma cells using Effectene Transfection Reagent according to manufacturer’s instructions (Qiagen). After 2 weeks of selection with 1,000 g/ml G418, clones were checked for RGS3 and RGS4 expression, respectively, using Western blot analysis. In each case, 3 clones with RGS3 and RGS4 overexpression were used for further functional assays.

RESULTS

Immunocytochemistry of Primary Glioma Cells To verify the astrocytic origin of TB288 primary cells derived from a glioblastoma biopsy specimen, immunocytochemistry was performed. Cells were positive for GPAP (98%) and vimentin (100%), and most also expressed S-100 protein (96%), while no staining for muscle actin, CD31, CD34, and cytokeratins was revealed. The Ki67 proliferation index was 44%. This expression pattern demonstrates that virtually all cells were neoplastic, while vascular cells or other mesenchymal cells were absent.

Establishment of Fast Cell Clones

U373fast cells (i.e. cells that had undergone 10 rounds of selection for migration) were compared with U373slow cells (i.e. cells passaged without selection) using an in vitro migration assay. After 72 h, the colony area of U373fast cells was 2.2-fold greater as compared to U373slow cells (p < 0.001, t-test, Fig. 1a). In addition, motile glioblastoma cells derived from a glioblastoma biopsy specimen (TB288fast cells) were compared with the less motile population derived from the same specimen (TB288slow cells). In migration assays, the colony area covered by TB288fast cells was 1.8- fold greater than the colony area covered by TB288slow cells (p < 0.001, t-test, Fig. 1b). These data confirm that selection for migration resulted in cell populations with a higher migratory phenotype.

No statistically significa\nt differences for proliferation and viability were revealed for either U373slow versus U373fast cells or TB288slow versus TB288fast cells (t-test). MTT assay revealed an increase after 48 h from originally seeded 1,500 cells to 5,864 628 cells (U373slow), 6,531 1,236 cells (U373fast), 2,474 537 cells (TB288slow), and 3,083 + 677 cells (TB288fast). LDH release within 48 h resulted in the following optical densities: 0.052 0.031 (U373slow), 0.025 0.028 OD (U373fast), 0.139 0.059 OD (TB288slow), and 0.142 0.098 (TB288fast).

Oligonucleotide Microarray Analysis

Of 12,625 genes (including ESTs) represented on the Affymetrix HG U95Av2 array, 4,851 were present in all four U373MG samples, and 4,319 genes were detected in all four TB288 samples. Intensity values of 5,444 genes in U373MG and 6,391 genes in TB288 were below the detection level in all samples. The other 2,330 genes in U373MG and 1,915 genes in TB288 were irregularly represented among the samples. Figure 2 shows all differentially expressed genes in U373MG (Fig. 2a) and TB288 (Fig. 2b), but only genes represented in all 4 samples were used for further investigation. A t-test (p < 0.05) identified 430 genes in U373MG and 831 in TB288, which were differentially expressed between slow and fast cells. From these 2 groups, all genes that differed in their expression level between slow and fast cells by at least 1.5-fold in mean intensity were selected, resulting in 206 genes (U373MG) and 470 genes (TB288), respectively. Table 1 (U373MG cells) and Table 2 (TB288 cells) show the 50 genes with the highest regulation. Of the 50 genes in U373MG cells, 37 showed higher expression and 13 lower expression in fast cells. Regulation scores (n-fold regulation) ranged between 2.2 and 3.9 for most genes, while 3 genes had scores higher than 10, i.e. genes encoding fibroblast activation protein, fibronectin, and the [alpha]1 chain of type XI collagen (Table 1). In TB288 cells, 15 of the 50 most differently expressed genes showed higher regulation scores and 35 showed lower scores in fast cells as compared to slow cells. Compared to U373MG cells, mean regulation scores were higher in TB288 glioma cells. For the 50 genes, the scores usually ranged between 3.2 and 6.0, while 4 genes (encoding secreted frizzled- related protein-1, cholesterol-25-hydroxylase, [alpha]6-integrin, and interleukin-6) had scores of between 9.7 and 34.4; all of them showing lower expression in TB288fast cells (Table 2).

Fig. 1. Results of monolayer migration assays performed on matrigel used to verify increased migration of human glioma cells following selection. Increase of colony areas at the end of the assay are expressed relative to the colony area after removal of sedimentation cylinders. U373MG colony area was measured at 72 h (a), TB288 colony area at 120 h after removal of sedimentation cylinders (b). Bars represent means of 8-fold replicates with standard deviations. Increase of colony areas was different between slow versus fast cells in both U373MG and TB288 cells (t-test, p < 0.001).

Fig. 2. Logarithmic scatter plot showing all expressed genes in U373MG (a) and TB288 (b) glioma cells. Diagonal pairs of lines indicate 1.5-fold (blue), 3-fold (red), and 6-fold regulation (green). The 8 genes that are differentially regulated in both U373MG and TB288 cells (Table 3) are indicated by red crosses.

The 2 lists of genes showing more than 1.5-fold regulation and a significant difference between slow and last cells, encompassing 206 genes (U373MG) and 470 genes (TB288), were then combined to reveal an intersect of genes regulated in the same direction. Altogether, 8 genes with measurable expression in both slow and fast cells were detected (Fig. 2). In addition, the microchip data provided 20 genes that were expressed only in either the fast or the slow samples in U373MG and/or TB288 cells. Of these 28 differentially expressed genes, 19 showed higher and 9 lower expression in fast cells (Table 3). Because RGS3 and RGS4 were the only genes involved in closely related pathways they were selected for additional expression and functional analysis.

Real-Time RT-PCR

To verify the microarray data by real-time quantitative RT-PCR, 2 genes were selected. Total RNA from U373MG and TB288 samples were analyzed for the expression of RGS3 and RGS4, relative to the level of GAPDH mRNA in each sample. Consistent with the micorarray data, higher levels of mRNA were measured for U373fast and TB288fast cells relative to U373slow and TB288slow cells. RGS3 mRNA levels were 1.8- fold higher in U373fast cells and 1.6-fold higher in TB288fast cells, while RGS4 mRNA levels were 3.4-fold higher in U373fast cells and 1.6-fold higher in TB288fast cells.

RGS3 Protein Analysis

To verify RGS3 expression at the protein level, Western blot analysis was performed using TB288slow/fast and U373slow/fast cells. In line with the RT-PCR data, TB288fast cells showed an increased expression of RGS3 protein as compared to TB288slow cells (Fig. 3a). Densitometry after normalization against actin revealed 2.2-fold increase of RGS3 protein in TB288fast cells as compared to TB288slow cells. U373slow and U373fast cells showed a low content of RGS3 protein, detectable only in immunofluorescence (Fig. 3b, c). Western blot analysis of U373MG cells stably transfected with RGS3 cDNA revealed strong overexpression of RGS3 protein as compared to parental U373MG cells. Clones RGS3-13, RGS3-15, and RGS3-35 were chosen for functional assays (Fig. 3d).

RGS4 Protein Analysis

Western blot analysis was also performed to verify RGS4 expression at the protein level. Corresponding to real-time RT-PCR, U373fast and TB288fast cells showed increased expression levels of RGS4 as compared to U373slow and TB288slow cells (Fig. 4a). Densitometry after normalization against actin showed 1.7-fold increase of RGS4 protein in U373fast cells and 3.0-fold increase in TB288fast cells as compared to their slow counterparts.

TABLE 1

Genes Differentially Expressed in U373fast versus U373slow Glioma Cells

TABLE 2

Genes Differentially Expressed in TB288fast versus TB288slow Glioma Cells

TABLE 3

Genes Differentially Expressed in Both U373MG and TB288 Glioma Cells Arranged by Function

Fig. 3. a: Western blot analysis using whole TB288 primary glioblastoma cell lysates (150 g protein in each lane) confirms higher RGS3 expression in TB288fast cells as compared to TB288slow cells at the protein level. b, c: RGS3 protein is expressed in the cytoplasm of U373slow cells (b) and U373fast cells (c) in the same punctate pattern (bar = 20 m) using indirect immunofluorescence, a technique that is not sensitive enough to reveal unequivocal differences in expression levels between slow and fast cells. d: Western blot analysis showing marked overexpression of recombinant human RGS3 in 3 stably transfected U373MG clones as compared to parental U373MG cells (90 g protein in each lane). Using the same amount of protein from U373fast and U373slow cells, faint RGS3 bands were revealed only after overexposing the blot for up to 1 h (not shown).

To evaluate the in vivo relevance of the data, human tissues were analyzed. RGS4 was expressed at different levels in all 6 glioblastoma biopsy specimens, whereas normal human cerebral cortex (obtained by autopsy) was virtually negative (Fig. 4b). U373MG cells stably transfected with RGS4 cDNA revealed strong overexpression of RGS4 protein as compared to parental U373MG cells as detected by Western blot analysis. Clones RGS4-1, RGS4-2 and RGS4-29 were chosen for functional assays (Fig. 4c).

Functional Assays of RGS3 and RGS4 Clones

Within 1 h, all RGS3 and RGS4 clones examined were more adherent than nontransfected U373MG cells (p < 0.001, t-test, Fig. 5a, b). In a monolayer migration assay the colony area covered by RGS3 clones was greater than the colony area covered by nontransfected U373MG glioma cells (RGS3-13: 1.36-fold, p = 0.016; RGS3-15: 1.42-fold, p < 0.001; RGS3-35: 1.55-fold, p < 0.001, Mest; Fig. 5c). Likewise, the colony area covered by RGS4 clones was greater than the colony area covered by non transfected U373MG cells (RGS4-1: 1.19-fold, p = 0.017; RGS4-2: 1.51-fold, p < 0.001; RGS4-29: 1.18-fold, p = 0.007, t-test; Fig. 5d). A direct relation between levels of overexpression and amount of increased adhesion/migration was not revealed (compare Figs. 3d, 5a and 5c as well as Figs. 4c, 5b and 5d). In contrast, no differences in proliferation were revealed for RGS3 and RGS4 clones versus nontransfected U373MG glioma cells (data not shown).

Fig. 4. Western blot analyses showing higher RGS4 expression in fast U373 and TB288 cells as compared to slow cells (a), expression of RGS4 in all 6 human glioblastoma biopsy specimens but not in normal human cerebral cortex (b), and marked overexpression of recombinant RGS4 in 3 stably transfected U373MG clones as compared to parental U373MG cells (c).

DISCUSSION

We have compared the expression pattern of highly migratory glioma cell populations to that of the original cells to reveal genes associated with motility. Since astrocytic gliomas are heterogeneous and undergo pronounced mesenchymal transdifferentiation in vitro, primary cells (TB288) as well as an established cell line (U373MG) were examined. Not unexpectedly, the expression pattern of genes in highly migratory cells differed markedly between these 2 populations. The lists shown in Tables 1 and 2, each consisting of the 50 genes with highest differential expression scores, comprised only 4 genes that occurred in both TB288 and U373MG cells. Only one of them (encoding serine protease- 11) was regulated in parallel (overexpressed in both fast cell populations), while the other 3 genes (KYNU, GAS1, NDP) showed inverse regulation (higher expression in U373fast cells but lower expression in TB288fast cells). We therefore applied stringent statistical criteria to reveal a relatively small group of 28 g\enes that were differentially expressed in both primary and established glioma cells in the same direction, including 19 upregulated and 9 downregulated genes in fast cells (Table 3).

Fig. 5. Adhesion assays (a, b) and monolayer migration assays (c, d) of U373MG clones overexpressing RGS3 (a, c) and RGS4 (b, d) showing an increase of both adhesion and migration of all clones as compared to U373MG parental cells on matrigel. In migration assays, increase of colony area at the end of assay is expressed relative to the colony area after removal of sedimentation cylinders. Cells had migrated for 72 h following removal of sedimentation cylinders. Bars represent means of 5-fold (adhesion) and 8-fold (migration) replicates with standard deviations (* p < 0.05, ** p < 0.01, *** p < 0.001; t-test).

Only one of these 28 genes (CATD) has been shown before to be functionally involved in glioma cell migration. The aspartic protease cathepsin D is overexpressed in malignant gliomas when compared to normal brain and low-grade gliomas, and application of blocking anti-cathepsin D antibodies inhibited glioma invasion in in vitro transwell migration assays (15, 16). Furthermore, phospholipase-C-gamma, a close relative of the PLCL2 gene product, has also been implicated in promoting glioma invasion (17). Other genes, shown here to be differently expressed in fast versus slow populations, have previously been found as being abnormally expressed in malignant gliomas as compared to low-grade gliomas and/ or non-neoplastic astrocytes, but their role in glioma cell migration has not been assessed. These genes encode midkine (9, 18, 19), low molecular weight neurofilament (20), the glial glutamate transporter (SLC1A3, EAAT1, GLAST) (21), and interleukin-8 (22). Furthermore, the gene for ADP-ribosyltransferase (PARP), a relative of ADPRTL2, is upregulated in glioblastoma (23). A few genes are involved in migratory processes outside the nervous system, including DAAM2, which is part of the Wnt signaling pathway controlling movement during development (24), and serine protease type 11 (PRSS11, HtrA1), which has been associated with melanoma metastasis (25). In conclusion, the vast majority of differentially expressed genes found in our study have not yet been examined with respect to glioma invasion, and for most of them it is unknown whether they are at all expressed in the normal or pathologic brain.

Some of the genes found here represent functional classes that have traditionally been linked to migration, such as extracellular matrix proteins, adhesion molecules, proteases, cytokines, and cytoskeletal constituents. Other genes are more surprising, such as 3 genes involved in G-protein coupled receptor signaling. GNA12 encodes G[alpha]^sub 12^ of the G^sub 12/13^ family of G proteins, while RGS3 and RGS4 belong to the RGS (regulators of G protein signaling) family, a group of more than 30 members that deactivate G protein signaling by accelerating GTP hydrolysis and returning the Ga subunit to its inactivated GDP-bound form (26). While the role of RGS proteins in migration has not yet been analysed, the demonstration that a Drosophila RGS-homolog (loco) interacting with G[alpha] subunits is expressed by peripheral glia and required for glial cell differentiation (27) suggests a conserved function for RGS proteins in glia biology. The spectrum of receptors coupling through G[alpha]^sub 12^ and affected by specific RGS proteins is still unclear, but available data on receptors and signaling pathways support the hypothesis that G-protein signaling may indeed be involved in glioma invasion. Lysophosphatidic acid (LPA), one of the few known ligands of G^sub 12/13^ receptors, induces microfilament rearrangement, focal adhesions, cell rounding, and increased migration through Rho-dependent pathways in a variety of cell types including astrocytes and glioma cells (28-30), although the underlying molecular mechanisms are largely unknown. LPA signaling leads to upregulation of phospholipase-C (31), and subsequent activation of Rho requires Daam1 (disheveled associated activator of morphogenesis-1) (24). Interestingly, close relatives of these 2 pathway elements were upregulated in the fast glioma cells studied here (PLCL2, DAAM2). Thus, it is tempting to speculate that signaling through G[alpha]^sub 12^ coupled receptors, possibly by LPA, plays a major role in glioma invasion.

We have therefore selected RGS3 and RGS4 as interesting G- protein signaling related candidates for expression and functional studies. RGS3 and RGS4 consist of 519 and 205 amino acids, respectively, the former including an extended N-terminal domain of about 300 amino acids. There are several splice variants with a 3.2- kb transcript of RGS3 and a 3.4-kb transcript of RGS4 predominating in the brain (32). They attenuate signaling not only via their GTPase accelerating activity towards several [alpha]-subunits of heterotrimeric G-proteins, but also directly inhibit G[beta][gamma]- mediated signals, while specific downstream signaling networks and their biological role are largely unknown (33).

By using real-time PCR and immunoblotting, we first confirmed that RGS3 and RGS4 are upregulated in fast cells at mRNA and protein levels. We then established human glioma cell clones with increased expression of recombinant human RGS3 or RGS4. All clones examined showed markedly increased adhesion and migration on matrigel, indicating that RGS3 and RGS4 upregulation in “fast” glioma cells, as revealed by microarray, reflects functional involvement rather than downstream events secondary to increased motility. Further in vivo studies are required to evaluate functional effects of RGS3 and RGS4 in situ. Taken together, our findings expand the spectrum of possible molecular pathways underlying the invasion of neoplastic astrocytes. Specifically, they suggest that RGS proteins and G- protein-mediated signal transduction are evolutionary conserved functional players in glia cell biology and migration.

ACKNOWLEDGMENTS

We thank B. Foppe, S. Engels, K. Klimmek and Dr. K. Richemann, Integrated Functional Genomics, Interdisciplinary Center for Clinical Research (IZKF), University Hospital Muenster, for support and advice during Affymetrix GeneChip hybridization, scanning and software analysis, Dr. T. Wieland. Institute of Pharmacology and Toxicology, University of Heidelberg/Mannheim, for providing materials and helpful hints, as well as B. Hillmann, B. Heuer and A. Wagner for excellent technical assistance.

REFERENCES

1. Behin A, Hoang-Xuan K, Carpentier AF, Delattre JY. Primary brain tumors in adults. Lancet 2003;361:323-31

2. Kleihues P, Cavenee WK. Pathology and genetics. Tumours of the nervous system. In: World Health Organisation Classification of Tumours, 2nd ed. Lyon: [ARC, 2000:9-29

3. Tysnes BB, Mahesparan R. Biological mechanisms of glioma invasion and potential therapeutic targets. J Neurooncol 2001;53:129- 47

4. Mikkelsen T, Bjerkvig R, Laerum OD, Rosenblum ML. Brain tumor invasion. New York: Wiley-Liss, 1998

5. Rutka JT, Tonn JC. The leading edge of glioma biology. New concepts in glioma proteases and invasion. J Neurooncol 2001;53:85- 235

6. Sehgal A, Boynton AL, Young RH et al. Application of the differential hybridization of Atlas Human expression arrays technique in the identification of differentially expressed genes in human glioblastoma multiforme tumor tissue. J Surg Oncol 1998;67:234- 41

7. Sallinen SL, Sallinen PK, Haapasalo HK, et al. Identification of differentially expressed genes in human gliomas by DNA microarray and tissue chip techniques. Cancer Res 2000;60:6617-22

8. Markert JM, Fuller CM, Gillespie GY, et al. Differential gene expression profiling in human brain tumors. Physiol Genomics 2001;5:21-33

9. Rickman DS, Bobek MP, Misek DE, et al. Distinctive molecular profiles of high-grade and low-grade gliomas based on oligonucleotide microarray analysis. Cancer Res 2001;61:6885-91

10. Hunter S, Young A, Olson J, et al. Differential expression between pilocytic and anaplastic astrocytomas: Identification of apolipoprotein D as a marker for low-grade, non-infiltrating primary CNS neoplasms. J Neuropathol Exp Neurol 2002;61:275-81

11. Mariani L, Beaudry C, McDonough WS, et al. Glioma cell motility is associated with reduced transcription of proapoptotic and proliferation genes: A cDNA microarray analysis. J Neurooncol 2001;53:161-76

12. Mariani L, Beaudry C, McDonough WS, et al. Death-associated protein 3 (Dap-3) is overexpressed in invasive glioblastoma cells in vivo and in glioma cell lines with induced molility phenotype in vitro. Clin Cancer Res 2001;7:2480-89

13. McDonough W, Tran N, Giese A, Norman SA, Berens ME. Altered gene expression in human astrocytoma cells selected for migration: I. Thromboxane synthase. J Neuropathol Exp Neurol 1998;57:449-55

14. Berens ME, Rief MD, Loo MA, Giese A. The role of extracellular matrix in human astrocytoma migration and proliferation studied in a microliter scale assay. Clin Exp Metastasis 1994;12:405-15

15. Sivaparvathi M, Sawaya R, Chintala SK, Go Y, Gokaslan ZL, Rao JS. Expression of cathepsin D during the progression of human gliomas. Neurosci Lett 1996;208:171-74

16. Tews DS, Nissen A. Expression of adhesion factors and degrading proteins in primary and secondary glioblastomas and their precursor tumors. Invasion Metastasis 1998;18:271-84

17. Khoshyomn S, Penar PL, Rossi J, Wells A, Abramson DL, Bhushan A. Inhibition of phospholipasc C-gammal activation blocks glioma cell motility and invasion of fetal rat brain aggregates. Neurosurgery 1999;44:568-77

18. Mishima K, Asai A, Kadomatsu K, et al. Increased expression of midkine during the progression of human astrocytomas. Neurosci Lett 1997;233:29-32

19. Kato S, Ishihara K, Shinozawa T, et al. Monoclonal antibody to human midkine reveals increased midkine expression in human brain tumors. J Neuropathol Exp Neurol 1999;58:430-41

20. Tlhya\ma T, Lee VM, Trojanowski JQ. Co-expression of low molecular weight neurofilament protein and glial fibrillary acidic protein in established human glioma cell lines. Am J Pathol 1993;142:883-92

21. Ye ZC, Rothstein JD, Sontheimer H. Compromised glutamate transport in human glioma cells: Reduction-mislocalization of sodium- dependent glutamate transporters and enhanced activity of cystine- glutamate exchange. J Neurosci 1999;19:10767-77

22. Desbaillets I, Diserens AC, Tribolet N, Hamou MF, Van Meir EG. Upregulation of interleukin 8 by oxygen-deprived cells in glioblastoma suggests a role in leukocyte activation, chemotaxis, and angiogenesis. J Exp Med 1997;186:1201-12

23. Wharlon SB, McNelis U, Bell HS, Whittle IR. Expression of poly(ADP-ribose) polymerase and distribution of poly(ADP- ribosyl)ation in glioblastoma and in a glioma multicellular tumour spheroid model. Neuropathol Appl Neurobiol 2000;26:528-35

24. Habas R, Kato Y, He X. Wnt/Frizzled activation of Rho regulates vertebrate gastrulation and requires a novel Formin homology protein Daaml. Cell 2001;107:843-54

25. Saldi A, De Luca A, Murini M, et al. The HtrA1 serine protease is down-regulated during human melanoma progression and represses growth of metastatic melanoma cells. Oncogene 2002;21:6684- 88

26. Neubig RR, Siderovski DP. Regulators of G-protein signalling as new central nervous system drug targets. Nat Rev Drug Discov 2002;1:187-97

27. Granderath S, Stollewerk A, Greig S, Goodman CS, O’Kane CJ, Klambt C. Loco encodes an RGS protein required for Drosophila glial differentiation. Development 1999;126:1781-91

28. Manning TJ Jr, Parker JC, Sontheimer H. Role of lysophosphatidic acid and rho in glioma cell motility. Cell Motil Cytoskelcton 2000;45:185-99

29. Moolenaar WH. Development of our current understanding of bioactive lysophospholipids. Ann NY Acad Sci 2000;905:1-10

30. Sah VP, Seasholtz TM, Sagi SA, Brown JH. The role of Rho in G protein-coupled receptor signal transduction. Annu Rev Pharmacol Toxicol 2000;40:459-89

31. Litosch I. Phosphatidic acid modulates G protein regulation of phospholipase C-betal activity in membranes. Cell Signal 2002;14:259-63

32. Costigan M, Samad TA, Allchorne A, Lanouc C, Tate S, Woolf CJ. High basal expression and injury-induced down regulation of 2 regulator of G-protein signaling transcripts, RGS3 and RGS4 in primary sensory neurons. Mol Cell Neurosci 2003;24:106-16

33. Wieland T, Mittmann C. Regulators of G-protein signalling: Multifunctional proteins with impact on signalling in the cardiovascular system. Pharmacol Therapeut 2003;97:95-115

Received August 13, 2003

Revision received October 23, 2003

Accepted November 5, 2003

LARS TATENHORST, VOLKER SENNER, PHD, SYLVIA PUTTMANN, PHD, AND WERNER PAULUS, MD

From Institute of Neuropathology, University Hospital, Muenster, Germany.

Correspondence to: Werner Paulus, MD, Institute of Neuropathology, University of Muenster, Domagkstr. 19, D-48149 Muenster, Germany. Tel: +49-251-83-56966. Fax: +49-251-83-56971. E- mail: werner. paulus@uni-muenster.de

Supported by grant Pa 328/5 from Deutsche Forschungsgemeinschaft (DFG).

Copyright American Association of Neuropathologists, Inc. Mar 2004