Secreted Frizzled-related proteins inhibit motility and promote growth of human malignant glioma cells


Cellular resistance to multiple proapoptotic stimuli and invasion of surrounding brain tissue by migrating tumor cells are main obstacles to an effective therapy for human malignant glioma. Here, we report that the Wnt family of embryonic differentiation genes modulate growth of malignant glioma cells in vitro and in vivo and inhibit cellular migration in vitro. sFRPs (soluble Frizzled-related proteins) are soluble proteins that bind to Wnt and interfere with Wnt signaling. We find that sFRP-1 and sFRP-2 are produced by the majority of longterm and ex vivo malignant glioma cell lines. Glioma cells that ectopically express sFRPs exhibit increased clonogenicity and enhanced resistance to serum starvation. In contrast, sFRPs do not modulate glioma cell susceptibility to apoptosis induced by the cytotoxic cytokines, CD95 (Fas/APO-1) ligand (CD95L) or Apo2 ligand/tumor necrosis factor-related apoptosis-inducing ligand (Apo2L/TRAIL), or various cytotoxic drugs. sFRP-2 strongly promotes the growth of intracranial glioma xenografts in nude mice. In contrast, enhanced expression of sFRPs inhibits the motility of glioma cells in vitro. sFRP-mediated effects on glioma cells are accompanied by decreased expression and activity of matrix metalloproteinase-2 (MMP-2) and decreased tyrosine phosphorylation of β-catenin. Thus, sFRPs promote survival under non-supportive conditions and inhibit the migration of glioma cells. We suggest that the regulation of these cellular processes involves expression of MMP-2 and tyrosine phosphorylation of β-catenin. These data support a function for Wnt signaling and its modulation by sFRPs in the biology of human gliomas.


Glioblastoma multiforme, the most frequent intrinsic malignant brain tumor, is largely resistant to established treatment modalities. Even after macroscopic resection, involved-field radiotherapy and adjuvant chemotherapy, recurrence of malignant glioma occurs regularly within few months. Single glioma cells show a tendency to invade the normal brain tissue as far as several centimetres from the macroscopic tumor border. At these sites, the glioma cells escape cytoreductive surgery and even involved-field radiotherapy with a safety margin of no more than 2 cm around the presurgical tumor volume. Similarly, invading cells are protected from many chemotherapeutic drugs because the blood-brain barrier in the infiltration zone may be intact, thereby inhibiting the access of cytotoxic drugs (reviewed in Roth and Weller, 1999).

Several mechanisms underlie the migration and invasion of malignant tumor cells. The degradation of the extracellular matrix can be accomplished by tumor-secreted matrix-degrading proteinases such as matrix metalloproteinases (MMP), e.g., MMP-2 in glioma cells (Apodaca et al., 1990; Yamamoto et al., 1996; Amberger et al., 1998). Cellular locomotion is essentially influenced by the cell adhesion proteins that mediate cell-cell contacts as well as adhesion to other biological surfaces. For example, integrins and CD44 contribute to the attachment of cells to substrates by binding to extracellular ligands such as vitronectin and hyaluronic acid (Friedlander et al., 1996; Merzak et al., 1994). E-cadherin mediates cell-cell adhesion by the zipper-like interaction of E-cadherin proteins expressed on adjacent cell surfaces. Thus, E-cadherin is responsible for the integrity of functional epithelia (Christofori et al., 1999). Intracellular β- and γ-catenin bind directly to the membrane-bound cadherins. α-catenin and other proteins link the cadherin-catenin complex to the actin filament network. Moreover, it has been speculated that the regulation of cell adhesion and migration is somehow linked to the regulation of proliferation and cell cycle control (Wong et al., 1998).

Here, we present the first evidence to suggest that Wnt signaling is involved simultaneously in the regulation of migration and proliferation of malignant glioma cells. The Drosophila segment polarity gene, wingless (wg), and its vertebrate homologs, the Wnt genes, encode secreted signaling molecules that regulate cell proliferation and differentiation during embryonic development (Dierick and Bejsovec, 1999). Wnt proteins are believed to induce an intracellular signalling cascade upon binding to the Frizzled receptors. However, this signaling pathway can be modulated by a recently identified group of soluble proteins, the soluble Frizzled-related proteins (sFRPs; secreted apoptosis-related proteins, SARPs) (Wang et al., 1997; Leyns et al., 1997; Rattner et al., 1997; Finch et al., 1997; Melkonyan et al., 1997). These proteins interfere with Wnt signaling presumably by binding to Wnt proteins, or, alternatively, directly by binding to Frizzled receptors (Bafico et al., 1999). Importantly, sFRP-1 has been shown to act as biphasic modulator of Wnt signaling, promoting Wnt-induced effects at low concentrations and counteracting them at higher concentrations (Uren et al., 2000). The interaction between Wnt proteins and Frizzled receptors triggers a multistep cascade of hitherto unknown events that lead to the inhibition of the serine-threonine kinase glycogen synthase kinase 3β (GSK-3β), resulting in decreased phosphorylation of adenomatous polyposis coli protein (APC) and β-catenin. Due to dephosphorylation or other as yet ill-defined events, β-catenin becomes less prone to degradation by the proteasome pathway. Instead, free β-catenin translocates to the nucleus where it serves as a transactivator for the transcription factor T-cell factor (TCF), leading to the transcription of several target genes. On the other hand, a change in the rheostat of complex-bound, cytoplasmic, and free, uncomplexed β-catenin also affects its role as a structural protein in the cell adhesion complexes (Müller et al., 1999).

Importantly, Wnt signaling as well as β-catenin and APC are involved in the development and progression of malignant tumors (Miller et al., 1999). Mutations in the β-catenin and APC genes are essential steps in the carcinogenesis of colon carcinoma, medulloblastoma, pilomatricoma, thyroid carcinoma, hepatocellular carcinoma and other malignancies (Tetsu and McCormick, 1999; Hamilton et al., 1995; Chan et al., 1999; Zurawel et al., 1998; Garcia et al., 1999; Ogawa et al., 1999). For example, Aoki et al. (1999) demonstrated that covalent linkage of TCF to β-catenin induces oncogenic transformation of embryonic fibroblasts. In light of these oncogenic properties of Wnt signaling, we were interested in the role of the Wnt pathway in malignant glioma.

Here, we report that interfering with the Wnt signaling cascade by ectopic expression of the soluble Wnt modulators, sFRPs, promotes growth and inhibits motility of malignant glioma cells in a complex, context-dependent manner.


Malignant glioma cells express sFRPs

To examine whether malignant glioma cells express sFRPs, we performed immunoblot analysis for sFRP-1 and sFRP-2 in 12 long-term human malignant glioma cell lines grown to 90% subconfluency. Since sFRPs are secreted proteins, cell culture supernatants of the cell lines were harvested. After concentration of supernatants, identical protein amounts were separated by SDS–PAGE and subjected to immunoblot analysis (Figure 1a). sFRP-1 protein was detected in eight of 12 cell lines. The strongest expression became apparent in U251MG and LN-308 cells. sFRP-2 was detected in six of 12 cell lines. Again, U251MG cells exhibited the highest level of expression. sFRP expression was also investigated in polyclonal primary glioma cell cultures that were freshly prepared from surgical glioma specimens (less than five passages, see Materials and methods). sFRP-1 expression was detected in four of four, and sFRP-2 expression in three of four, primary cell lines (Figure 1a). Since sFRP expression has been reported to depend on the phase of cell growth (Melkonyan et al., 1997), we compared sFRP levels in the supernatants of exponentially growing and highly confluent glioma cells (Figure 1b). sFRP expression was strongly up-regulated upon confluency in most of the cell lines with endogenous sFRP production.

Figure 1

Expression of sFRPs by malignant glioma cells. (a) Supernatants of long-term glioma cell lines or ex vivo primary glioma cell cultures (45/4, 54/6, 61/4, 67/4) were analysed for the expression of endogenous sFRP-1 and sFRP-2 proteins by immunoblot analysis. sFRP-1 and sFRP-2 proteins have a size of approximately 35 kDa (Melkonyan et al., 1997). Supernatants from U373MG sFRP transfectants were used as positive controls (see also Figure 2a). (b) sFRP expression is regulated by cellular confluency. The cells were allowed to grow exponentially or were grown to dense confluency. After harvesting and concentrating supernatants, identical protein amounts were used for immunoblot analysis (10 μg/lane)

sFRPs promote clonogenicity of glioma cells and survival under conditions of serum starvation

To elucidate the function of sFRPs in glioma cells, we produced stable sFRP-1 and sFRP-2 transfectants of the U87MG, LN-229 and U373MG cell lines. These cell lines were chosen because they had little endogenous sFRP expression and, except for LN-229, were previously found to be tumorigenic in nude mice (Roth et al., 1999; unpublished results). As controls, cells of the same lines were transfected with the empty control vector pcDNA3 (neo) only. sFRP expression of stably transfected glioma cells was confirmed by immunoblot analysis as shown in Figure 2a. Expression of the transgene was not influenced by cell density, and the morphology of transfected cells was not different from neo control cells or parental cells (data not shown). Further, there were no significant differences in doubling times between sFRP transfectants and neo control cells (Figure 2b). However, colony formation was significantly enhanced in glioma cells with ectopic sFRP expression (Figure 2c). In contrast to the increased colony number of sFRP transfectants, we did not observe substantial differences in colony size between sFRP transfectants and control cells.

Figure 2

Enhanced expression of sFRPs promotes growth of glioma cells and confers resistance to serum starvation. (a) Ectopic expression of sFRP-1 and sFRP-2 in U87MG, LN-229 and U373MG human malignant glioma cells. Supernatants of transfected cells were subjected to immunoblot analysis as described above. (b) Doubling times of human malignant glioma cells expressing the empty control vector pcDNA3, the sFRP-1 transgene or the sFRP-2 transgene. The cells were cultured in complete medium. Cell numbers were counted at regular intervals, and the doubling times of the cells during logarithmic growth were calculated. The differences were not statistically significant as assessed by t-test (P>0.05, transfectants compared with neo control cells). (c) For clonogenicity assays, 100 U87MG glioma cells/well were seeded in 6-well plates and allowed to proliferate for 9 days in full DMEM medium. Thereafter, cells were stained with crystal violet, and colonies comprising more than 20 cells were counted. This experiment was performed three times with similar results. The data shown in the figure are means of colony numbers in three wells (±s.d., ast;P<0.05, t-test). (d) Increased resistance to serum starvation by ectopic expression of sFRPs. 24 h after seeding in 96-well plates, full medium was replaced by serum-free medium. Density of viable U87MG cells was measured daily by crystal violet assays. The growth advantage of sFRP transfectants was confirmed in several independent experiments. Data are expressed as means of triplicates and s.d. Neo control cells exhibited a significantly decreased cell density compared with s-FRP transfectants (P<0.05, t-test). The curves derived from s-FRP-1- and s-FRP-2-expressing cells were not significantly different at any time point. (e) Decreased cell death during serum starvation by ectopic expression of sFRPs. After 4 and 6 days the percentages of dead, that is, detached, cells were determined by flow cytometry

It has been reported that serum starvation-induced G1 arrest of breast epithelial cells was associated with induction of sFRP-1, suggesting a role of sFRPs in the regulation of the cell cycle (Zhou et al., 1998). Therefore, we carried out cell cycle analysis by flow cytometry, but did not find significant differences between the neo and sFRP-expressing cell lines under normal growth conditions. Similarly, serum deprivation for 24–96 h induced S phase reduction and G0/1 accumulation in all cell lines but no different regulation of cell cycle in control cells compared with cells expressing the sFRP transgenes. These data were interesting given the strong impact of confluency, which is associated with slowed cell cycle progression, on sFRP expression (Figure 1b). Next, we examined the resistance of the different cell lines to serum withdrawal (Figure 2d). Glioma cells expressing sFRP-1 or sFRP-2 were more resistant to serum deprivation than neo control cells in that more viable sFRP transfectants could be detected after 1–4 days of serum starvation. Additionally, we assessed tumor cell death after serum withdrawal by flow cytometry. After 4 and 6 days the percentages of dead, that is, detached, cells were significantly higher in control cells than in sFRP-transfected cells (Figure 2e). All results demonstrated in 2c–e were obtained with U87MG glioma cells. Similar results were obtained with U373MG and LN-229 cells (data not shown).

sFRPs have been reported to modify the susceptibility of MCF7 breast carcinoma cells to apoptosis induced by tumor necrosis factor-α, ceramide or adriamycin (Melkonyan et al., 1997). Therefore, we asked whether glioma cells engineered to express ectopically sFRPs exhibited an altered susceptibility to the induction of apoptosis. Table 1 shows that sFRPs did not modulate glioma cell sensitivity to the cytotoxic cytokines, CD95L and Apo2L/TRAIL, or cytotoxic drugs with different modes of action, lomustine (CCNU), teniposide or cisplatin.

Table 1 No modulation of susceptibility to cytotoxic cytokine- and drug-induced apoptosis

sFRP-2 promotes glioma cell growth in vivo

Next, we investigated sFRP-mediated modulation of glioma growth in vivo. sFRP transgene-expressing U87MG and U373MG cells and, as a control, U87MG neo and U373 neo cells, were implanted intracerebrally into athymic mice. The animals were sacrificed 25 days after inoculation. Figure 3a,b shows that sFRP-2 expressing glioma xenografts were significantly larger than xenografts consisting of control cells. Of note, sFRP-1 transfectants produced only marginally larger tumors than control cells, and this difference did not reach statistical significance. LN-229 cells were not tumorigenic in CD1TM athymic mice. Histologically, U87MG and U373MG xenografts were densely cellular and in general not diffusely infiltrating (Figure 3c,d). Only few cells infiltrated the adjacent brain tissue at the tumor border. Compared with U87MG cells, U373MG tumors were often surrounded by small nodules of tumor cells. Necrosis or blood vessels were not seen in xenografts of either cell line. There were no morphological differences at the tumor margins between xenografts formed by U87MG or U373MG neo and sFRP cells. By means of routine immunohistochemical methods, we did not observe differences with regard to the few scattered tumor cells at the xenograft borders (Figure 3c,d). sFRP-1 and sFRP-2 expression of stably transfected, intracerebrally growing U87MG gliomas was confirmed by immunohistochemistry (Figure 3e,f). Sections derived from neo control tumors showed no specific labeling (Figure 3g,h). sFRP-1 and sFRP-2 antibodies were not cross-immunoreactive (not shown).

Figure 3

sFRP-2 promotes glioma growth in vivo. (a,b) U87MG or U373MG glioma cells were inoculated intracranially into athymic mice. Each group (neo, sFRP-1, sFRP-2) consisted of four animals. The animals were sacrificed 25 days after implantation. The volumes of tumors expressing sFRP-2 were substantially larger than the tumors of the other two groups, and only mice bearing sFRP-2 tumors showed neurologic symptoms at day 25. In contrast, the mean sFRP-1 tumor volume was only slightly increased compared with neo tumors, and the difference did not achieve statistical significance (P>0.05, t-test). (c,d) Intracerebrally inoculated U87MG glioma cells form densely cellular xenografts in nude mice (magnification: bar=500 μm in c, 125 μm in d). (e–h) Ectopic expression of sFRP-1 and sFRP-2 in human U87MG glioma xenografts was examined by immunohistochemistry. Glioma xenografts overexpressing sFRP-1 (e) and sFRP-2 (f) stained positive for the transgene products, whereas, in U87MG neo tumors, sFRP-1 (g) or sFRP-2 (H) were not detected (bar=400 μm in e–h)

sFRPs inhibit glioma cell migration and down-regulate MMP-2

β-catenin, one of the main targets of Wnt signaling, is crucial for intercellular adhesion and cellular locomotion. Therefore, we asked whether interfering with the Wnt signaling pathway by ectopic expression of sFRPs would result in a modulation of motility in glioma cells. Figure 4a shows that sFRP-1- and sFRP-2-transfected glioma cells exhibited decreased migratory activity compared with control cells. We hypothesized that the decreased motility of sFRP transfectants might be caused by changes in integrin receptor expression. Of the integrins, αvβ3 integrin has been identified to play a particular role in gliomas (Paulus and Tonn, 1994; Platten et al., 2000). Flow cytometry did not disclose significant differences of αvβ3 integrin levels at the cell surface between neo and sFRP-transfected cells (Figure 4b).

Figure 4

Suppression of glioma cell migration by sFRPs: association with decreased MMP-2 activity. (a) 3×104 U87MG glioma cells per well were applied to a 48-well micro chemotaxis chamber and allowed to migrate through the membrane. After 24 h the migrated cells on the bottom side of the membrane were fixed, stained and counted by microscope using a microgrid. This experiment was repeated twice with similar results. Data are expressed as means of cell numbers in three wells (±s.d., *P<0.05, t-test). (b) No regulation of αvβ3 integrin expression by sFRPs. αvβ3 integrin expression in the transfected cell lines was analysed by flow cytometry. SFI values are depicted in the graph. (c) The expression of MMP-2 was determined by immunoblot analysis after harvesting and concentrating supernatants of the different glioma cell lines. MMP-2 migrates at 72 kDa. (d) Decreased activity of MMP-2 in sFRP-producing glioma cells. U87MG glioma cells were grown until they reached 80% confluency. Equal amounts of protein (50 μg) from the conditioned medium were subjected to gel electrophoresis. Subsequently, the extent of MMP-2-induced digestion of gelatin was visualized by staining the gel with Coomassie Brilliant Blue. MMP-2 in the supernatants of sFRPs-producing cells was less active than MMP-2 of control cells

Since matrix metalloproteinases (MMPs), especially MMP-2, play an important role in invasion and migration of glioma cells, we examined the expression and activity of MMP-2 in glioma cells. The expression levels of released MMP-2 were significantly reduced in sFRP transfectants compared with control cells (Figure 4c). To corroborate this finding, we directly assessed the activity of MMP-2 by zymography (Figure 4d). In this assay, MMP-2-mediated digestion of gelatin in a SDS gel leads to quantitative loss of Coomassie staining at the locus of migration of the gelatinolytic enzyme. In accordance with the immunoblot findings, sFRP transfectants exhibited decreased MMP-2 activity. This finding was more prominent in sFRP-2 transfectants than in sFRP-1 expressing cells. Of note, the lower of the two signals reflects the gelatinolytic activity of the activated form of MMP-2 (Deryugina et al., 1997). A concomitant decrease in the activity of MMP-9 was not observed in sFRP transfectants (data not shown).

Ectopic expression of sFRPs results in decreased phosphorylation of β-catenin

To investigate whether proteins of the cell-adhesion complex, such as β-catenin, are involved in the sFRP-mediated effects on glioma growth and migration, we carried out immunoblot analyses of catenins and cadherins. The immunoblot analysis of α-catenin, E-cadherin and N-cadherin in the different cell lines did not reveal any difference attributable to sFRP expression (data not shown). Activation of Wnt signaling leads to the accumulation of free β-catenin in the cytosol and, thereby, enables β-catenin to translocate into the nucleus (Willert and Nusse, 1998; Dierick and Bejsovec, 1999). Therefore, we examined the expression levels of β-catenin in whole cell lysates as well as cytoplasmic and nuclear fractions of U87MG cells. Figure 5a shows that the constitutive level of cellular, cytoplasmic and nuclear β-catenin was unchanged in cells with enhanced sFRP expression. It has been demonstrated, however, that the phosphorylation state of β-catenin rather than its overall expression level is decisive for β-catenin function. Therefore, we performed immunoprecipitation studies using an anti-phosphotyrosine antibody to examine the levels of tyrosine-phosphorylated β-catenin. In fact, phosphorylation at tyrosine residues was substantially weaker in sFRP-producing glioma cells than in the control cells (Figure 5b), confirming that sFRPs interfere with an established down-stream target of Wnt signaling, that is, β-catenin. Since serine/threonine phosphorylation of β-catenin is a frequently described downstream effect of Wnt activity, we also carried out immunoprecipitation studies to detect changes in the serine/threonine phosphorylation of β-catenin. However, ectopic sFRP expression was not associated with a change in the phosphorylation of β-catenin at serine residues (Figure 5c).

Figure 5

Decreased tyrosine phosphorylation of β-catenin in glioma cells expressing sFRPs. (a) β-catenin levels were examined in whole cell lysates and, separately, in the cytoplasmic and nuclear cell fractions of U87MG glioma cells. No substantial differences of expression levels could be observed in the different cell lines. (b) Ectopic expression of sFRPs resulted in decreased phosphorylation of β-catenin at tyrosine residues. Control staining with a β-catenin antibody did not reveal differences in β-catenin levels. (c) Serine phosphorylation of β-catenin was examined as in b. sFRPs did not alter phosphorylation of β-catenin at serine residues


Several gene families that play a role during embryogenesis and differentiation are reactivated during the process of tumorigenesis, including Wnt proteins and their receptors, Frizzled. sFRPs are soluble proteins that bind to Wnt and modulate Wnt signaling (Rattner et al., 1997; Bafico et al., 1999; Uren et al., 2000). The biological role of Wnt signaling in human malignant gliomas has not been examined in detail. Here, we report that the majority of long-term as well as freshly prepared ex vivo glioma cells produce sFRPs (Figure 1a). The extent of sFRP expression correlates with cell density in vitro in that exponentially proliferating cells produce less sFRPs than confluent cell cultures (Figure 1b). To examine sFRP-induced effects on Wnt signaling in glioma cells in vitro, we stably expressed sFRP transgenes in human malignant glioma cell lines with (U373MG) or without (U87MG, LN-229) endogenous sFRP expression. Contrary to recent reports which have suggested an association of serum deprivation-induced G1 arrest and induction of sFRPs (Zhou et al., 1998), ectopic expression of sFRP-1 or sFRP-2 had no effect on cell cycle distribution (data not shown). Thus, although sFRP expression depended on confluency which is associated with accumulation in G0/1 in glioma cells (Hueber et al., 1998), cell cycle regulation was unaffected by ectopic expression of sFRP, suggesting that sFRP do not mediate the confluency-induced G0/1 arrest.

Activation of the Wnt pathway may result in accumulation of the free pool of cytoplasmic β-catenin, translocation into the nucleus and, consecutively, increased transactivation of the transcription factor TCF by virtue of the transactivator function of β-catenin. On the other hand, β-catenin belongs to structural proteins like the other catenins and the cadherins. By constituting the cell-adhesion complex, these proteins are responsible for cell-cell adhesion and more complex processes such as cellular motility (Liu et al., 1997; Müller et al., 1999). Therefore, we investigated whether interfering with the Wnt pathway by ectopic expression of the soluble sFRPs would lead to a modulation of cellular motility. In chemotaxis chamber assays, migration of sFRP transfectants was significantly impaired compared with control cells (Figure 4a). Since human glioma xenografts in nude mice hardly invade brain tissue, we were not able to demonstrate the inhibitory effect of sFRPs on migration or invasion in vivo. A thorough histological assessment did not disclose any differences in morphology or angiogenesis of xenografts (Figure 3).

Diverse proteins are supposed to play key roles in glioma cell motility, e.g. integrins, MMPs and tissue inhibitors of metalloproteinases (TIMPs) (Giese and Westphal, 1996). Thus, we examined the potential involvement of integrins in sFRP-mediated inhibition of migration. Integrin αvβ3 has a particular role in glioma cell migration (Paulus and Tonn, 1994; Platten et al., 2000). Flow cytometry failed to disclose a regulation of αvβ3 integrin expression in the process of sFRP-mediated inhibition of migration. Next, we investigated a modulation of MMP-2 expression or activity by sFRPs. MMP-2 has an important role in invasion and motility of glioma cells (Apodaca et al., 1990; Yamamoto et al., 1996). Interestingly, immunoblot analysis as well as zymography revealed a reduction in MMP-2 protein levels and MMP-2 activity in sFRP transfectants (Figure 4c, d), providing an explanation for the impaired motility of glioma cells expressing sFRP-1 and sFRP-2.

To further characterize the sFRP-induced inhibition of cellular migration, we examined the expression of proteins which constitute cell adhesion complexes. We observed that the protein levels of α-catenin, β-catenin, N-cadherin and E-cadherin were unchanged in the sFRP-transfected cell lines. Further, since Wnt signaling may affect the rheostat of free and complex-bound β-catenin which might be influenced by sFRPs (Lin et al., 1997; Bafico et al., 1999; Dennis et al., 1999), we were interested in the subcellular localization of β-catenin. However, the levels of cytosolic and nuclear β-catenin were not different in sFRP transfectants and control cells (Figure 5a). Of note, the cytosolic fraction detected by this method comprised free, uncomplexed β-catenin, as well as β-catenin complexed with adhesion proteins such as E-cadherin. However, altered β-catenin functioning is not necessarily accompanied by a change in the overall expression level of β-catenin. Several reports suggest that phosphorylation may regulate β-catenin function. For example, epidermal growth factor (EGF)-induced scattering of carcinoma cells does not alter the expression level of β-catenin, but is associated with phosphorylation of β-catenin at tyrosine residues (Shibamato et al., 1994; Hazan and Norton, 1998). Phosphorylation of β-catenin causes dissociation of the cadherin-catenin complex, resulting in mobilization of β-catenin from its complex-bound sites to the free cytoplasmic pool (Jawhari et al., 1999). Moreover, Müller et al. (1999) showed that cellular migration is associated with tyrosine phosphorylation of β-catenin. Therefore, we examined whether (de)phosphorylation of β-catenin might be involved in the inhibition of glioma cell motility. We expected that sFRP-induced inhibition of migration should be associated with decreased phosphorylation of β-catenin. In fact, tyrosine phosphorylation of β-catenin was reduced in sFRP-expressing cells compared with neo control cells (Figure 5b). In contrast, the extent of serine phosphorylation of β-catenin was unchanged, suggesting that Wnt signaling in human glioma cells may employ mechanisms of functional β-catenin targeting other than serine phosphorylation. Thus, tyrosine phosphorylation might be responsible for sFRP-induced effects on glioma cell motility and growth regulation.

The finding of a sFRP-induced decrease of phosphorylated β-catenin and, given the constant overall level of β-catenin, the respective increase of the dephosphorylated form offers an explanation for the growth-promoting effects of sFRP expression found in this study, too. Glioma cells engineered to express ectopically sFRPs showed a prominent increase in clonogenicity and an enhanced survival under conditions of serum withdrawal (Figure 2c–e). These findings are consistent with a survival advantage of sFRP-expressing cells in a non-supportive microenvironment that was modelled in vitro by serum starvation and isolation of cells in clonogenicity assays. In line with this hypothesis, intracerebral glioma xenografts formed by sFRP-2-expressing cells exhibited a prominent growth advantage in the brains of nude mice in vivo (Figure 3). In contrast to breast carcinoma cells (Melkonyan et al., 1997), sFRP-producing glioma cells exhibit an unaltered susceptibility to distinct forms of apoptosis, e.g., exposure to the death ligands, CD95L or Apo2L/TRAIL, or cytotoxic drugs (Table 1). Thus, we suggest that sFRPs initiate specific subcellular mechanisms that negatively regulate some, but not all, cell death pathways. Importantly, enhanced expression of, or mutations in the β-catenin gene are involved in tumorigenesis, e.g., in sporadic medulloblastoma (Zurawel et al., 1998). Moreover, the Wnt pathway has been linked to the development or progression of malignancies, e.g. through the oncogenic actions of increased transactivation of TCF by β-catenin (Aoki et al., 1999; Miller et al., 1999). We suppose that the growth-promoting activity of sFRPs observed in this study is based on an enhanced signaling function of the free pool of β-catenin which becomes stabilized by dephosphorylation. Moreover, sFRPs do not simply act as inhibitors of Wnt signalling. sFRP-1 has shown to be a biphasic modulator of Wnt-induced effects, thereby promoting stabilization of β-catenin at low concentrations (Uren et al., 2000).

The exact mechanism of sFRP-mediated dephosphorylation of β-catenin remains to be elucidated. Several studies indicate an important role for the balanced action of protein kinases and protein phosphatases in regulating the different functions of β-catenin. Besides the serine-threonine kinase glycogen synthase kinase 3β (GSK-3β), the tyrosine kinase activity of the EGF receptor or the hepatocyte growth factor/scatter factor (HGF/SF) may be involved in the regulation of β-catenin phosphorylation (Shibamoto et al., 1994). This is even more intriguing since a molecular cross-talk between Wnt and EGF signalling is well known (Dierick and Bejsovec, 1999). Besides the kinases, several protein tyrosine phosphatases have been linked to β-catenin function (Hoschuetzky et al., 1994; Brady-Kalnay et al., 1995; Kypta et al., 1996). Future work will reveal which of the kinases or phosphatases is crucial for maintaining β-catenin in a dephosphorylated state in the case of sFRP-mediated signaling. Another unresolved question is the mechanism by which sFRPs mediate decreased expression of MMP-2. Importantly, another matrix metalloproteinase, MMP-7, was shown to be a target of β-catenin transactivation in intestinal tumors (Crawford et al., 1999). Since MMP-2 is a secreted extracellular matrix-degrading proteinase similar to MMP-7, it may also be regulated by β-catenin.

In this study, we have for the first time presented evidence that interfering with Wnt signaling results in a modulation of growth and motility of malignant glioma cells. Our data support a role for the Wnt family and its natural modulators, sFRPs, in the pathophysiology of malignant brain tumors. Taken together, sFRPs inhibit migration and promote proliferation under non-supportive conditions. The underlying mechanisms may involve sFRP-induced actions to maintain β-catenin in a dephosphorylated state, thereby promoting its signaling functions. We speculate that inhibition of MMP-2 activity as well as growth promotion may result from released β-catenin-mediated signaling. By transactivating TCF or other unknown transcription factors, β-catenin may influence the activity of diverse target genes that are involved in the regulation of proliferation, cellular resistance and motility. In general, one might assume that the ability to survive for longer time in a poor microenvironment would lead rather to enhanced migration because of permitting anchorage-independent cell survival during the process of migration. Since we observed decreased migration upon enhanced expression of sFRPs, we suggest that promotion of survival under non-supportive conditions and inhibition of cellular motility are attributable to two distinct sFRP-mediated molecular pathways. It is tempting to speculate that these two pathways branch off after a common point of origin. In the light of the finding that β-catenin regulates not only diverse target genes that may influence cellular proliferation, but also the MMP-7 promotor (Crawford et al., 1999), the involvement of β-catenin in regulating MMP-2, too, remains an intriguing possibility for future investigations.

Materials and methods


Lomustine (CCNU) was obtained from Medac (Hamburg, Germany). Teniposide (VM26) was provided by Sandoz Pharma AG (Basel, Switzerland). Cisplatin and cycloheximide were from Sigma (Deisenhofen, Germany). Soluble CD95L was obtained from murine CD95L-transfected N2A neuroblastoma cells (Roth et al., 1997). Purified human Apo2L/TRAIL was kindly provided by Dr A Ashkenazi (Genentech, CA, USA). The glioma cell lines were kindly provided by Dr N de Tribolet (Lausanne, Switzerland) and have been characterized in previous studies (Weller et al., 1998). Antibodies to human and mouse sFRP proteins were prepared as described (Melkonyan et al., 1997). Further antibodies were obtained from: Alexis, San Diego, CA, USA (α-catenin, β-catenin), Santa Cruz, Santa Cruz, CA, USA (E-cadherin), R&D Systems, Minneapolis, MN, USA (N-cadherin) and Oncogene Research Products, Cambridge, MA, USA (MMP-2).

Cell culture, transfections and cytotoxicity assays

The glioma cells were maintained as described (Roth et al., 1998). Doubling times were determined during logarithmic growth in 24-well plates. The cells (103) were seeded, and viable cell counts were obtained daily for 7 days by trypan blue exclusion. For the determination of clonogenicity, 100 cells per well were seeded into 6-well plates in DMEM. At day 9, the colonies were counted by an inverted microscope after staining with crystal violet. Acute cytotoxic cell death assays were performed as described (Roth et al., 1998). Glioma cells stably expressing human sFRP-1 or mouse sFRP-2 were obtained by lipofection with SuperFect (Qiagen, Hilden, Germany) using the hSARP2-pcDNA3 and the mSARP1-pcDNA3 plasmids (Melkonyan et al., 1997). As a control, cells were transfected with the pcDNA3 neo control plasmid only. The cells were selected with G418 (500 μg/ml), starting 48 h after transfection. All experiments were carried out with pooled transfectants to avoid cloning or selection artifacts.

Preparation of primary glioma cell cultures

Human brain tumors were obtained from patients with glioblastoma who underwent surgery for tumor resection. After tumor removal, the tissues were placed immediately in petri dishes, minced mechanically and digested enzymatically using collagenase (1 h, 37°C). Subsequently, the dissociated cells were filtered through 100 μm cell strainers to remove any tissue debris. After centrifugation and lysis of erythrocytes by treatment with hypotonic water, the glioma cells were washed and resuspended in full medium (DMEM). Conditioned medium was harvested after no more than five passages.

Animal studies and immunohistochemistry

All animal work was carried out in accordance with the NIH guidelines ‘Guide for the Care and Use of Laboratory Animals’. Athymic mice (CD1TM, Charles River, Sulzfeld, Germany) were anesthetized by intraperitoneal injection of 7% chloral hydrate before all procedures. Intracranial implantation of glioma cells as well as generation and staining of cryostat sections were performed as previously described (Roth et al., 1999). For the assessment of tumor volume, cryostat sections were obtained at regular intervals, routinely stained with H&E and subjected to an analysis of tumor volumes using MCID software (Imaging Research Inc., Ontario, Canada). Immunohistochemistry was performed as outlined previously (Roth et al., 2000). Polyclonal antibodies to sFRP-1 and sFRP-2 were diluted in TRIS-buffered saline (TBS) and applied as primary antibodies to the sections overnight at a concentration of 1 : 100. Biotinylated anti-rabbit secondary antibody (Dakopatts, Hamburg, Germany), diluted 1 : 400 in TBS, was incubated for 30 min before the application of avidin-biotin-peroxidase complex (Dakopatts), diluted 1 : 200 in TBS, for 30 min. Labeled antigen was visualized with standard diaminobenzidine techniques.

Flow cytometry

Cell cycle analysis, using a Becton Dickinson FACScalibur cytometer, was carried out as previously described (Roth et al., 2000).

Detection of αvβ3 integrin expression

For flow cytometric analysis of αvβ3 integrin expression, glioma cells were detached from the culture dishes, harvested into ice-cold complete medium containing 10% FCS, centrifuged and resuspended in flow cytometry buffer (1% BSA/PBS/0.01% sodium azide). Subsequently, phycoerythrin (PE)-labeled anti-CD51/61[integrin αvb3] mouse monoclonal antibody (1 μg) (Chemicon, Hofheim, Germany) or PE-labeled mouse IgG (1 μg) as an isotype control (Sigma) were added per sample (106 cells). After incubation and washing, samples were resuspended in 300 μl PBS containing 1% formaldehyde and stored light-protected at 4°C prior to analysis by a Becton Dickinson FACScalibur cytometer. The specific fluorescence index (SFI) was calculated as the ratio of the mean fluorescence values obtained with the specific anti-αvβ3 antibody and the isotype control antibody (Weller et al., 1995). A SFI of 1.0 indicates that there is no difference in binding of αvβ3 integrin antibody compared with an isotype control antibody.

Immunoblot analysis

Immunoblots were performed as outlined previously (Roth et al., 2000). In the case of soluble proteins (sFRPs), 5×106 glioma cells were cultured in serum-free DMEM medium for 24 h. The supernatants were harvested and subsequently concentrated by centrifugal filter devices (Millipore, Eschborn, Germany). The supernatants of freshly isolated ex vivo glioma cells were prepared accordingly. Twenty μg protein (cellular lysates) or 10 μg protein (supernatants) per lane was separated on polyacrylamide gels (sFRPs: 15%; MMP-2:10%; α-catenin, β-catenin, E-cadherin, N-cadherin: 8%) and blotted onto nitrocellulose by standard procedures. The membranes were washed, incubated with primary antibody (sFRP-1:1 : 10 000; sFRP-2: 1 : 2 500; MMP-2: 1 μg/ml; α-catenin, β-catenin, E-cadherin: 2 μg/ml; N-cadherin: 10 μg/ml), washed and incubated with secondary antibody (protein A or anti-mouse IgG, Amersham, Braunschweig, Germany). Enhanced chemiluminescence (ECL) reagents (Amersham) were used for detection. For the separation of nuclear and cytosolic fractions, cellular pellets were resuspended in hypotonic buffer (5 mM TRIS-HCl pH 7.4, 5 mM EDTA, 50 mM NaF, 2 μg/ml aprotinin, 100 μg/ml phenylmethylsulfonylfluoride) and incubated on ice for 20 min. After centrifugation, the supernatants (cytosolic fraction) were harvested, and the proteins were precipitated with acetone by routine methods. The pellets (nuclear fraction) were resuspended in 0.1% SDS lysis buffer (25 mM HEPES, pH 7.3, 125 mM NaCl, 10 mM EDTA, 0.1% SDS, 0.5% deoxycholate, 1% Triton X-100, 50 mM NaF, 2 μg/ml aprotinin, 100 μg/ml phenylmethylsulfonylfluoride), incubated on ice and subjected to ultracentrifugation for 10 min. Subsequently, supernatants were harvested and stored at −20°C.


Cells were cultured to confluency and pretreated with sodium orthovanadate (1 mM) for 90 min before lysis. After washing twice with PBS, cells were lysed in ice-cold lysis buffer (50 mM HEPES, ph 7.3, 150 mM NaCl, 10% glycerol, 1.5 mM Mg Cl2, 1% Triton X-100, 1 mM EGTA, pH 8, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 25 mM NaF, 1 mM NaVO4, 1 mM phenylmethylsulfonylfluoride, 10 mM sodium pyrophosphate) and precleared by centrifugation at 3000 r.p.m. for 10 min at 4°C. 5 mg of protein lysate were adjusted to 1 ml with lysis buffer and incubated with 5 μl of β-catenin antibody (Alexis) for 1 h at 4°C. Protein A/G-agarose was added for 1 h. Precipitates were washed three times with HNTG buffer (20 mM HEPES, pH 7.3, 150 mM NaCl, 0.1% Triton X-100, 10% glycerol), and beads were resuspended in SDS sample buffer with HNTG buffer added 1 : 1 (v/v). For subsequent immunoblot analysis, proteins separated by SDS–PAGE were transferred to PVDF membranes (Amersham) and incubated with phosphotyrosine antibody (clone 4G10, Upstate Biotechnology, Lake Placid, NY, USA) or phosphoserine antibody (Alexis). Proteins were visualized with the ECL system (Amersham). Before reprobing with β-catenin antibody, blots were stripped by incubation for 30 min in 62.5 mM TRIS-HCl, pH 6.8, 2% SDS and 0.1% β-mercaptoethanol at 50°C.

Migration assay

The migration of malignant glioma cells through 8 μm pores was assessed using a 48 well micro chemotaxis chamber (Neuro Probe Inc., Bethesda, MD, USA). NIH-3T3-conditioned medium (30 μl) in the wells of the bottom chamber served as the chemoattractant. The filter membrane was placed between top and bottom chamber and equilibrated for 30 min at 37°C. The cells (3×104 per well) were applied to the upper wells and allowed to migrate through the membrane at 37°C in humidified air with 5% CO2. After 24 h, the membrane was removed, and the non-migrated cells were scraped off with a wiper blade. Migrated cells on the bottom side of the membrane were fixed in methanol and stained in thiazine/eosine using DiffQuick (Dade Behring AG, Düdingen, Switzerland). Cells migrated through the membrane pores were counted using a microgrid (Wick et al., 1998).


Analysis of MMP-2 activity was performed with SDS-polyacrylamide gels impregnated with 0.1% gelatin and 10% polyacrylamide. This assay is based on the digestion of gelatin and subsequent loss of Coomassie staining at the locus of migration of gelatinolytic enzymes. Cells were grown in full DMEM medium until 80% confluency, washed and maintained in serum-free medium, and the supernatant was collected after 24 h. Four parts of medium containing equal amounts of protein (50 μg) from the conditioned medium were mixed with one part of Laemmli sample buffer prior to electrophoresis. Gels were run at constant current, washed twice for 30 min in 50 mM TRIS-HCl, pH 7.5, plus 2.5% Triton X-100, and incubated overnight at 37°C in 50 mM TRIS-HCl, pH 7.6, 10 mM CaCl2, 150 mM NaCl, 0.05% NaNO3. Gels were stained with Coomassie Brilliant Blue R-250 and destained.


  1. Amberger VR, Hensel T, Ogata N and Schwab ME. . 1998 Cancer Res. 58: 149–158.

  2. Aoki M, Hecht A, Kruse U, Kemler R and Vogt PK. . 1999 Proc. Natl. Acad. Sci. USA 96: 139–144.

  3. Apodaca G, Rutka JT, Bouhana K, Berens ME, Giblin JR, Rosenblum ML, McKerrow JH and Banda MJ. . 1990 Cancer Res. 50: 2322–2329.

  4. Bafico A, Gazit A, Pramila T, Finch PW, Yaniv A and Aaronson SA. . 1999 J. Biol. Chem. 274: 16180–16187.

  5. Brady-Kalnay SM, Rimm DL and Tonks NK. . 1995 J. Cell Biol. 130: 977–986.

  6. Chan EF, Gat U, McNiff JM and Fuchs E. . 1999 Nat. Genet. 21: 410–413.

  7. Christofori G and Semb H. . 1999 Trends. Biochem. Sci. 24: 73–76.

  8. Crawford HC, Fingleton BM, Rudolph OL, Goss KJ, Rubinfeld B, Polakis P and Matrisian LM. . 1999 Oncogene 18: 2883–2891.

  9. Dennis S, Aikawa M, Szeto W, d’Amore PA and Papkoff J. . 1999 J. Cell Sci. 112: 3815–3820.

  10. Deryugina EI, Bourdon MA, Luo GX, Reisfeld RA and Strongin A. . 1997 J. Cell Sci. 110: 2473–2482.

  11. Dierick H and Bejsovec A. . 1999 Curr. Top. Dev. Biol. 43: 153–190.

  12. Finch PW, He X, Kelley MJ, Uren A, Schaudies RP, Popescu NC, Rudikoff S, Aaronson SA, Varmus HE and Rubin JS. . 1997 Proc. Natl. Acad. Sci. USA 94: 6770–6775.

  13. Friedlander DR, Zagzag D, Shiff B, Cohen H, Allen JC, Kelly PJ and Grumet M. . 1996 Cancer Res. 56: 1939–1947.

  14. Garcia RG, Tallini G, Herrero A, D’Aquila TG, Carcangiu ML and Rimm DL. . 1999 Cancer Res. 59: 1811–1815.

  15. Giese A and Westphal M. . 1996 Neurosurgery 39: 235–250.

  16. Hamilton SR, Liu B, Parsons RE, Papadopoulos N, Jen J, Powell SM, Krush AJ, Berk T, Cohen Z and Tetu B. . 1995 N. Engl. J. Med. 332: 839–847.

  17. Hazan RB and Norton L. . 1998 J. Biol. Chem. 273: 9078–9084.

  18. Hoschuetzky H, Aberle H and Kemler R. . 1994 J. Cell Biol. 127: 1375–1380.

  19. Hueber A, Durka S and Weller M. . 1998 FEBS Lett. 432: 155–157.

  20. Jawhari AU, Farthing MJ and Pignatelli M. . 1999 J. Pathol. 187: 155–157.

  21. Kypta RM, Su H and Reichardt LF. . 1996 J. Cell Biol. 134: 1519–1529.

  22. Leyns L, Bouwmeester T, Kim SH, Piccolo S and De Robertis EM. . 1997 Cell 88: 747–756.

  23. Lin K, Wang S, Julius MA, Kitajewski J, Moos MJ and Luyten FP. . 1997 Proc. Natl. Acad. Sci. USA 94: 11196–11200.

  24. Liu D, el Hariry I, Karayiannakis AJ, Wilding J, Chinery R, Kmiot W, McCrea PD, Gullick WJ and Pignatelli M. . 1997 Lab. Invest. 77: 557–563.

  25. Melkonyan HS, Chang WC, Shapiro JP, Mahadevappa M, Fitzpatrick PA, Kiefer MC, Tomei LD and Umansky SR. . 1997 Proc. Natl. Acad. Sci. USA 94: 13636–13641.

  26. Merzak A, Koocheckpour S and Pilkington GJ. . 1994 Cancer Res. 54: 3988–3992.

  27. Miller JR, Hocking AM, Brown JD and Moon RT. . 1999 Oncogene 18: 7860–7872.

  28. Müller T, Choidas A, Reichmann E and Ullrich A. . 1999 J. Biol. Chem. 274: 10173–10183.

  29. Ogawa K, Yamada Y, Kishibe K, Ishizaki K and Tokusashi Y. . 1999 Cancer Res. 59: 1830–1833.

  30. Paulus W and Tonn JC. . 1994 J. Neurosurg. 80: 515–519.

  31. Platten M, Wick W, Wild-Bode C, Aulwurm S, Dichgans J and Weller M. . 2000 Biochem. Biophys. Res. Commun. 268: 607–611.

  32. Rattner A, Hsieh JC, Smallwood PM, Gilbert DJ, Copeland NG, Jenkins NA and Nathans J. . 1997 Proc. Natl. Acad. Sci. USA 94: 2859–2863.

  33. Roth W, Fontana A, Trepel M, Reed JC, Dichgans J and Weller M. . 1997 Cancer Immunol. Immunother. 44: 55–63.

  34. Roth W, Wagenknecht B, Grimmel C, Dichgans J and Weller M. . 1998 Br. J. Cancer 77: 404–411.

  35. Roth W, Isenmann S, Naumann U, Kügler S, Bähr M, Dichgans J, Ashkenazi A and Weller M. . 1999 Biochem. Biophys. Res. Commun. 265: 479–483.

  36. Roth W and Weller M. . 1999 Cell. Mol. Life Sci. 56: 481–506.

  37. Roth W, Grimmel C, Rieger L, Strik H, Takayama S, Krajewski S, Dichgans J, Reed JC and Weller M. . 2000 Brain Pathol. 10: 223–234.

  38. Shibamoto S, Hayakawa M, Takeuchi K, Hori T, Oku N, Miyazawa K, Kitamura N, Takeichi M and Ito F. . 1994 Cell Adhes. Commun. 1: 295–305.

  39. Tetsu O and McCormick F. . 1999 Nature 398: 422–426.

  40. Uren A, Reichsman F, Anest V, Taylor WG, Muraiso K, Bottaro DP, Cumberledge S and Rubin JS. . 2000 J. Biol. Chem. 275: 4374–4382.

  41. Wang S, Krinks M, Lin K, Luyten FP and Moos M. . 1997 Cell 88: 757–766.

  42. Weller M, Malipiero U, Rensing-Ehl A, Barr PJ and Fontana A. . 1995 Cancer Res. 55: 2936–2944.

  43. Weller M, Rieger J, Grimmel C, Van Meir EG, De Tribolet N, Krajewski S, Reed JC, von Deimling A and Dichgans J. . 1998 Int. J. Cancer 79: 640–644.

  44. Wick W, Wagner S, Kerkau S, Dichgans J, Tonn JC and Weller M. . 1998 FEBS Lett. 440: 419–424.

  45. Willert K and Nusse R. . 1998 Curr. Opin. Genet. Dev. 8: 95–102.

  46. Wong MH, Rubinfeld B and Gordon JI. . 1998 J. Cell Biol. 141: 765–777.

  47. Yamamoto M, Mohanam S, Sawaya R, Fuller GN, Seiki M, Sato H, Gokaslan ZL, Liotta LA, Nicolson GL and Rao JS. . 1996 Cancer Res. 56: 384–392.

  48. Zhou Z, Wang J, Han X, Zhou J and Linder S. . 1998 Int. J. Cancer 78: 95–99.

  49. Zurawel RH, Chiappa SA, Allen C and Raffel C. . 1998 Cancer Res. 58: 896–899.

Download references


This work was supported by a grant of the Fortüne-Programm of the University of Tübingen to W Roth. The authors thank Dr W Wick for helpful discussions.

Author information

Correspondence to Michael Weller.

Rights and permissions

Reprints and Permissions

About this article


  • sFRP
  • Wnt signaling
  • migration
  • MMP-2
  • β-catenin
  • malignant glioma

Further reading