Nuclear phosphorylated Y142 β-catenin accumulates in astrocytomas and glioblastomas and regulates cell invasion

Abstract

Glioblastoma multiforme (GBM) is a fast growing brain tumor characterized by extensive infiltration into the surrounding tissue and one of the most aggressive cancers. GBM is the most common glioma (originating from glial-derived cells) that either evolves from a low grade astrocytoma or appears de novo. Wnt/β-catenin and Hepatocyte Growth Factor (HGF)/c-Met signaling are hyperactive in human gliomas, where they regulate cell proliferation, migration and stem cell behavior. We previously demonstrated that β-catenin is phosphorylated at Y142 by recombinant c-Met kinase and downstream of HGF signaling in neurons. Here we studied phosphoY142 (PY142) β-catenin and dephospho S/T β-catenin (a classical Wnt transducer) in glioma biopsies, GBM cell lines and biopsy-derived glioma cell cultures. We found that PY142 β-catenin mainly localizes in the nucleus and signals through transcriptional activation in GBM cells. Tissue microarray analysis confirmed strong nuclear PY142 β-catenin immunostaining in astrocytoma and GBM biopsies. By contrast, active β-catenin showed nuclear localization only in GBM samples. Western blot analysis of tumor biopsies further indicated that PY142 and active β-catenin accumulate independently, correlating with the expression of Snail/Slug (an epithelial-mesenchymal transition marker) and Cyclin-D1 (a regulator of cell cycle progression), respectively, in high grade astrocytomas and GBMs. Moreover, GBM cells stimulated with HGF showed increasing levels of PY142 β-catenin and Snail/Slug. Importantly, the expression of mutant Y142F β-catenin decreased cell detachment and invasion induced by HGF in GBM cell lines and biopsy-derived cell cultures. Our results identify PY142 β-catenin as a nuclear β-catenin signaling form that downregulates adhesion and promotes GBM cell invasion.

Abbreviations

β-cat

β-catenin

CK2

Casein kinase-2

EGF

Epidermal Growth Factor

EMT

Epithelial mesenchymal transition

ERK

Extracellular-regulated kinase

FBS

foetal bovine serum

Fox M1

Forhead box M1

Fz

Frizzled

GBM

Glioblastoma multiforme

GSK

Glycogen Synthase Kinase

HGF

Hepatocyte Growth Factor

LEF

Lymphocyte Enhancer Factor

PBS

Phosphate Buffer Saline

PFA

paraformaldehyde

PY142 β-catenin

phosphorylated Y142 β-catenin

RTK

Receptor tyrosine kinase

SDS

sodium dodecyl sulfate

TCF

T-cell factor

TMA

tissue microarray

WT

wild-type

Introduction

β-catenin is an structural component of adherent junctions and a key effector of Wnt canonical signaling.¹ In the cell-cell adhesion complex, β-catenin interacts with cadherins and α-catenin that links the complex to the actin cytoskeleton. Downregulation of cell adhesion is in part achieved by phosphorylation of β-catenin, which promotes the release of β-catenin from the adhesion complex and activates β-catenin nuclear signaling.² Several cytoplasmic tyrosine kinases and growth factor-activated receptor tyrosine kinases (RTK) phosphorylate β-catenin, thus promoting a migratory phenotype.^3-5 Dynamic regulation by tyrosine phosphatases counteracts β-catenin phosphorylation.^1,2In addition, β-catenin is a transducer of Wnt signaling that plays central roles in development, the balance of cell proliferation/differentiation and cancer.^5,6 In the absence of Wnt ligands, β-catenin assembles into a complex containing Glycogen Synthase Kinase (GSK)-3β and additional Wnt pathway components. The S/T kinase GSK-3β phosphorylates β-catenin at its N-terminal domain, which favors its proteasomal degradation. Upon Wnt binding to Frizzled (Fz) and LRP5/6 receptors, β-catenin escapes degradation, accumulates in the cytoplasm and translocates to the nucleus, where it interacts with Lymphoid Enhancer Factor (LEF)/T-cell factor (TCF) to regulate transcription of target genes.^7-9

Glioblastoma multiforme (GBM; grade IV, World Health Organization) is an aggressive brain tumor that develops de novo (primary GBM; accounting for 90% of the cases) or evolves from a previous low grade astrocytoma (secondary GBM). GBM is the most common malignant glioma, displaying uncontrolled proliferation, angiogenesis, necrosis, resistance to apoptosis and profuse infiltration into the brain parenchyma. Average survival of the patients is of 12–14 months despite treatment. RTK-activated pathways are hyperactive in GBM, including Epidermal Growth Factor (EGF) Receptor and c-Met signaling.^10-12 Hepatocyte Growth Factor (HGF) and its receptor c-Met are both overexpressed in GBM, contributing to tumor growth invasion, angiogenesis and conferring a stem-like phenotype and poor prognosis.^10,13-16 Although activating mutations of β-catenin have not been identified in GBM¹⁷, overexpression of β-catenin and other Wnt pathway components (including Fz¹⁸) together with epigenetic regulation of Wnt inhibitors results in Wnt/β-catenin activation in GBM.^19-21 Overexpression of the Forkhead box M1 (FoxM1) transcription factor represents a critical mechanism further contributing to β-catenin signaling in GBM.²² FoxM1 promotes β-catenin nuclear accumulation and together they form a complex with TCF4 required for glioma stem cell self-renewal and gliomagenesis.^22,23 Crosstalk between EGFR, c-Met and Wnt/β-catenin is well documented in cancer cells, thereby linking β-catenin signaling and cell migration induced by growth factors.^24-29 Thus, stimulation of epithelial cells with EGF or HGF through the phosphorylation of cell adhesion proteins diminishes cell adhesion while promoting epithelial-mesenchymal transition (EMT). Phosphorylation of Y142 β-catenin by c-Met affects β-catenin interaction with α-catenin³⁰ and promotes a β-catenin switch from adhesive to transcriptional functions that facilitates pro-migratory phenotypes.^31,32 Moreover, EGFR signaling involving Extracellular-regulated kinase (ERK) and Casein kinase 2 (CK2) results in α-catenin phosphorylation, β-catenin transactivation and GBM cell invasion.³³

Here we studied the role of β-catenin phosphorylated at Y142 (PY142) in cell invasion in GBM, a tumor in which total and dephospho S/T β-catenin (a classical Wnt transducer; hereon active β-catenin) have already received some attention.^20,34,35 We used GBM biopsies, cell lines and cell cultures established from tumoral tissue. Our findings identify a nuclear pool of PY142 β-catenin in GBM cells. β-catenin activity assay confirms that PY142 β-catenin signals through transcriptional regulation. Western blot and tissue microarray (TMA) analysis of astrocytoma (grade II and III) and GBM (grade IV) biopsies indicates that PY142 β-catenin and active β-catenin accumulate independently in grade III astrocytoma and GBM (grade IV) samples, correlating with Snail/Slug and Cyclin D1, respectively. GBM cells stimulated with HGF increase PY142 β-catenin and Snail/Slug levels. Interestingly, mutant Y142F β-catenin decreases GBM cell detachment and invasion in GBM cell lines and biopsy-derived primary cultures. Together, these results indicate that PY142 β-catenin signaling contributes to GBM progression by regulating cell invasion.

Results

We performed an immunocytochemical study of β-catenin forms in U251MG and U87MG GBM cell lines and primary cultures established from astrocytoma (grade II) and GBM (grade IV) biopsies. Immunostaining for total β-catenin revealed a nuclear pool in GBM cell lines, primary astrocytoma and GBM cultures, in addition to its presence at cell-cell and cell-substrate contacts (Fig. 1). Immunostaining for PY142-β-catenin revealed a readily detectable PY142-β-catenin fraction in basal culture conditions, uncovering relatively high and stable levels of this β-catenin form. PY142-β-catenin immunostaining pattern displayed a diffuse cytoplasmic and nuclear localization in U251MG cells, and a predominantly nuclear localization in U87MG cells. Similar results were obtained in primary GBM cultures, showing increased PY142 β-catenin cytoplasmic and nuclear immunoreactivity in GBM cells compared to astrocytoma grade II cells (Fig. 1). Immunostaining for dephosphorylated S/T β-catenin (active β-catenin) also revealed a nuclear localization in GBM cultures, in addition to a cytoplasmic and plasma membrane localization. In contrast, active β-catenin was markedly cytoplasmic and perinuclear in astrocytoma grade II cells (Fig. 1). Together, primary glioma cultures and GBM cell lines exhibit a prominent nuclear β-catenin pool, consistent with the presence of nuclear PY142 β-catenin in astrocytomas and GBMs, and of nuclear active β-catenin in GBM cells.

Figure 1.

Distinct nuclear β-catenin pools are found in GBM cell lines and primary astrocytoma and GBM cultures. (A) Astrocytoma grade II and GBM (grade IV) cultures established from 2 GBM biopsies or (B) U251MG and U87MG (grade IV) cell lines were immunostained for total β-catenin (N-terminal epitope; N-term), PY142 β-catenin or active β-catenin and co-stained for Hoestch. Arrows indicate immunoreactivity in nuclei. Asterisks indicate perinuclear immunostaining and negative nuclei in astrocytoma grade II cells immunostained for active β-catenin. Bar = 15 μm.

Next, we analyzed active and PY142 β-catenin forms in tissue homogenates obtained from human control brain samples, grade II and III astrocytomas and GBM (grade IV) biopsies. Active β-catenin was detected in one control and increased in grade II astrocytoma and some grade III and IV samples, where it appears to correlate with total β-catenin levels (Fig. 2). PY142 β-catenin was detected in the majority of grade III astrocytoma and grade IV GBM biopsies, whereas it was undetectable or showed low levels in grade II astrocytomas. These results indicate a stronger presence of PY142 β-catenin in high grade astrocytoma and GBM biopsies (Fig. 2). Interestingly, PY142 β-catenin levels accumulated independently of active or total β-catenin in astrocytoma grade III and GBM biopsies.

Figure 2.

PY142 β-catenin increases in grade III astrocytoma and GBM (grade IV) biopsies, correlating with Snail/Slug expression. Western blot of tissue homogenates from control brain, grade II and III astrocytoma or GBM (grade IV) biopsies were analyzed for PY142, active and total β-catenin (A and B), c-Met and/or PY1234/5 active c-Met (A and B) and possible target genes (Snail/Slug, Cyclin D1 and Cdc25a (B)). β-actin was used as a loading control (A and B). PY142 β-catenin is detectable in grade II astrocytomas and increases in grade III astrocytoma and GBM samples, where it seems to distribute differently to active β-catenin. PY1234 and total c-Met increase in grade III astrocytoma and GBM samples. Snail/Slug expression is found in grade III and IV cases displaying high levels of PY142 β-catenin, whereas Cyclin D1 and Cdc25a correlate with active β-catenin among the different samples (except for control brain and grade II astrocytomas, in which Cyclin D1 is not detected). Asterisks refer to the same samples run in Fig. 4 (showing high levels of TCF-4 and TCF-1).

To further evaluate the levels and the subcellular localization of β-catenin forms in gliomas, we performed immunohistochemical studies using tissue microarrays (TMAs) including astrocytoma (grade II and III) and GBM (grade IV) biopsies. Active β-catenin immunostaining was cytoplasmic in grade II and III astrocytomas and GBM samples (Fig. 3). Histoscore analysis indicated that active β-catenin was localized in the nucleus in ˜71% of GBM samples, whereas nuclear immunostaining was not observed in astrocytoma samples (p < 0,00001; Pearson;Table 1A). On the other hand, nuclear immunostaining for PY142 β-catenin was found in grade II and III astrocytomas and GBM biopsies (Fig. 3). Immunoreactivity for nuclear PY142 β-catenin was 37 % higher in astrocytomas compared to GBMs (Fig. 3;Table 1B). Thus, in agreement with results obtained from cell cultures, PY142 β-catenin showed a nuclear localization in astrocytomas and GBMs, while active β-catenin was nuclear only in GBM biopsies. These findings demonstrate that active and PY142 β-catenin are distinct β-catenin forms that differently accumulate in the nucleus during astrocytoma to GBM progression.

Figure 3.

Nuclear PY142 β-catenin, Snail/Slug and c-Met are present in astrocytoma and GBM biopsies. (A) Representative immunostainings for active β-catenin, PY142 β-catenin, Snail/Slug and c-Met of selected grade II, III and IV samples from the TMAs. Nuclei are counterstained by haematoxylin. Although active β-catenin cytoplasmic immunostaining is present from grade II to grade IV samples, it concentrates in the nucleus only in grade IV (arrows). PY142 β-catenin immunoreactivity is detected in the cytoplasm and nuclei from grade II to IV samples, and its levels increase in the nuclei of astrocytomas grade III cells, similarly to c-Met and Snail/Slug. (B) Statistical analysis of the Histoscore data obtained from the TMA indicates that nuclear PY142 β-catenin and nuclear Snail/Slug immunostainings increase in astrocytomas vs GBM (p= 0,01 and 0,03, respectivelly). Nuclear c-Met (center) immunoreactivity is higher in astrocytomas of grade III compared to astrocytomas of grade II (p=0, 02). Bars represent mean immunostaining levels and segments one standard deviation. P-values were obtained from a Mann-Whitney test.

Table 1.

Statistical analysis of immunohistochemical studies using the TMA. (A) Association between positivity/negativity in nuclear active β-catenin in gliomas (astrocytomas vs. GBMs). Absolulte frequency (and percentage) is shown for each glioma type. P-value from a Fisher test is shown to assess the significance of the association. (B) Differential immunostaining of c-Met, Snail, PY142 β-catenin (β-cat) and active β-catenin in astrocytomas grade II, III and GBMs (grade IV). Mean and standard deviation shown for astrocytomas (globally or separately according to histopathological classification) and GBMs. P-values from a Mann-Whitney test provided assess the differences between astrocytomas (grade II vs. III) and between astrocytomas and GBMs (P-values indicating significant differences are in bold).

(A).

	Nuclear active β-catenin
	Negative	Positive	Total
Astrocytomas	15 (100%)	0 (0%)	15 (100%)
GBMs	16 (29%)	39 (71%)	55 (100%)

p-value < 0,00001

(B).

					p-value
					Astrocytomas	Astrocytomas
		n	mean	SD	II vs III	vs GBM
c-Met cytosolic	Astrocytomas	16	108,44	11,21	0,21	0,74
	Astrocytoma II	7	145,71	35,05
	Astrocytoma III	7	120,71	29,78
	GBMs	55	110,74	34,10
c-Met nuclear	Astrocytomas	16	131,88	36,82	0,02	0,97
	Astrocytoma II	7	105,71	30.06
	Astrocytoma III	8	151,25	30.06
	GBMs	54	131,59	41.06
Snail nuclear	Astrocytomas	16	157,03	70,29	0,27	0,01
	Astrocytoma II	7	132,50	52,22
	Astrocytoma III	8	171.25	83,28
	GBMs	56	108.26	51,86
PY142 β–cat nuclear	Astrocytomas	16	209,69	75,88	0,41	0,03
	Astrocytoma II	7	200,00	56,43
	Astrocytoma III	8	210,00	95,33
	GBMs	25	152,53	82,44
PY142β–cat cytosolic	Astrocytomas	8	115,63	46,40	0,15	0,48
	Astrocytoma II	3	75,00	47,70
	Astrocytoma III	4	135,00	26,50
	GBMs	11	138,64	47,75
Active β–cat cytosolic	Astrocytomas	15	133,67	32,65	0,21	0,25
	Astrocytoma II	7	145,71	35,05
	Astrocytoma III	7	120,71	29,78
	GBMs	55	116,95	41,26

c-Met, a receptor tyrosine kinase that phosphorylates β-catenin at Y142³¹, is hyperactive in GBM^10,15,36,37 and associates with recurrence.³⁶ We investigated whether c-Met activation is related to PY142 β-catenin levels in glioma biopsies. We studied total c-Met and PY1234/5 active c-Met in tissue homogenates obtained from astrocytomas and GBMs. Western blot for PY1234/5 c-Met confirmed that active c-Met is commonly found in high grade astrocytomas and GBMs (Fig. 2). However, PY1234/5 c-Met and PY142 β-catenin levels did not consistently correlate. Immunohistochemical analysis of the TMA showed a predominantly nuclear expression of c-Met in glioma biopsies (Fig. 3A), as observed in biopsy-derived cell cultures (Supplementary Fig. 1). Statistical analysis demonstrates that nuclear c-Met increases astrocytomas of grade III compared to astrocytomas of grade II (Fig. 3B). In addition, we observed a significant correlation between nuclear c-Met and nuclear PY142 β-catenin immunostainings (Pearson; p = 0,04) in high grade as-trocytomas.

We also evaluated possible targets regulated by active β-catenin and PY142 β-catenin in glioma biopsies. Cyclin D1 and Cdc25a were assessed as β-catenin targets involved in cell cycle progression, which deregulation associates with tumorigenesis.^38,39 Snail/Slug was studied as a pro-invasive marker regulated by HGF/c-Met and Wnt signaling.^40-42 Western blot analysis shows that high levels of Cdc25a and Cyclin D1 correlated with active and total β-catenin levels in grade III and IV gliomas (Fig. 2B). However, in contrast to Cdc25a, Cyclin D1 was not detected in control brain and grade II astrocytoma samples (Fig. 2B). Interestingly, high levels of Snail/Slug are found in grade III astrocytomas and GBMs (grade IV) biopsies displaying the highest levels of PY142 β-catenin (Fig. 2B). Furthermore, immunohistochemical analysis of biopsies showed a predominantly nuclear localization for Snail/Slug in astrocytomas and GBMs (higher in astrocytomas) (Fig. 3;Table 1B). Together, results from glioma biopsies link active β-catenin levels and Cyclin D1 expression, and suggest a relationship between PY142 β-catenin and Snail/Slug in high grade astrocytomas and GBMs.

TCF transcription factors collaborate with β-catenin in Wnt signaling.⁴³ We analyzed TCF-1 and TCF-4 levels in tissue homogenates from glioma biopsies of different histopathological grade. TCF-1 was detected in grade II, III and IV samples (Fig. 4A). High levels of TCF-4 and TCF-1 were found in biopsies of grade III and IV that displayed high PY142 β-catenin levels (Fig. 4A). This observation suggests that PY142 β-catenin involves transcriptional regulation. To address this point, a TOP-Flash plasmid that reports β-catenin transcriptional activation was co-expressed together with wild-type (WT), mutant Y142F or Y654F β-catenin in U87MG cells. Whereas expression of WT or mutant Y654F β-catenin produced significant transcriptional activation at com-parable levels, expression of mu-tant Y142F β-catenin potently blocked (>80%) β-catenin transcriptional activity (Fig. 4B). This result demonstrates that PY142 β-catenin signals through transcriptional regulation in GBM cells.

Figure 4.

PY142 β-catenin signals through transcriptional regulation in GBM. (A) Western blot analysis of TCF-4 and TCF-1 in tissue homogenates from control brain, astrocytomas of grades II and III and GBM (grade IV) biopsies, together with β-catenin and β-actin (loading control). Bands of ˜60–80 kDa were detected using anti-TCF-4 antibodies especially in grade III samples. A prominent band at around 50 kDa was immunodetected by anti-TCF-1 antibodies and additional bands in the ˜30–50 kDa range. TCF1 immunoreactive bands increased in grade II, III and IV tumor samples compared to control brain. Asterisks indicate the grade III and IV samples displaying high expression of TCF factors, low total β-catenin and high PY142 β-catenin levels (shown in Fig. 2B). (B) Luciferase assay reporting β-catenin transcriptional activity. U87MG cells were co-transfected with WT, Y654F or Y142F β-catenin together with the TOP-Flash plasmid. Whereas cells expressing WT or Y654F β-catenin show similar luciferase activity values, expression of mutant Y142F β-catenin strongly reduced β-catenin transcriptional activation in GBM cells (*** p ≤ 0,001; T test).

Next, we aimed at investigating the relationship between c-Met, PY142 β-catenin and migration markers in GBM cells. U87MG cells were stimulated with the ligand of c-Met. Treatment with HGF (100 ng/ml) increased PY142 β-catenin, Snail/Slug and N-cadherin levels (Fig. 5A). Quantification of PY142 β-catenin indicates that HGF produced an increase of PY142 β-catenin peaking at about a 3-fold increase, compared to untreated cells, at 25 min of treatment (Fig. 5B). Snail/Slug levels also increased significantly after stimulation with HGF (Fig 5B). The c-Met inhibitor, SU11274 (2 μM), potenly reduced PY1234/5 c-Met (active Met) levels and GBM cell invasion (SupplementaryFig. 2). However, treating U87MG cells with HGF and SU11274 did not reduce PY142 β-catenin levels (Fig. 5A). Our results suggest that HGF signaling increases PY142 β-catenin and Snail/Slug in GBM. The lack of effect of a c-Met inhibitor on PY142 β-catenin levels may indicate that other kinases, in addition to c-Met, regulate β-catenin phosphorylation at this site downstream of HGF signaling, or it may be explained by c-Met active fragments (see Discussion).

Figure 5.

HGF signaling regulates PY142 β-catenin and Snail/Slug in GBM cells. (A) Cell lysates of U87MG cells deprived of serum and treated with HGF (100 ng/ml) and SU11274 (2μM, overnight) as indicated. HGF treatment increased PY142 β-catenin levels. N-cadherin and Snail/Slug levels also increased upon HGF treatment. β-actin was used as a loading control. (B) Quantification of PY142 β-catenin (top) and Snail/Slug (bottom) levels normalized against the levels of β-actin. Quantification results are shown as fold increase vs. control (* p ≤ 0,05, ** p ≤ 0,01, *** p ≤ 0,001; T test).

Phosphorylation of β-catenin at Y residues increases cell migration and decreases cell adhesion.¹ We studied the involvement of PY142 β-catenin in GBM cell invasion and migration in Transwell and cell spreading assays, respectively, using a mutant Y142F β-catenin. U87MG cells expressing Y142F β-catenin showed a decrease of ˜30% in the number of cells that migrated across the Transwell and invaded the Matrigel layer, compared to cells expressing WT β-catenin (used as control;Fig. 6A). The c-Met inhibitor SU11274 together with HGF reduces the migration of GBM cells (SupplementaryFig. 2). In line with this, U87MG cells expressing WT or Y142F β-catenin treated with SU11274 showed a significant reduction in the number of cells that invaded the Matrigel, compared to untreated cells expressing WT β-catenin (Fig. 6A). We next addressed whether mutant Y142F β-catenin affects cell adhesion and migration in a cell spreading assay,⁴⁴ in which round, loosely adherent (poorly spread, i.e. highly migratory) cells are counted. U87MG cells expressing DsRed alone or together with WT β-catenin showed an increase of 20% in the number of round cells upon treatment with HGF compared to untreated cells (Fig. 6B), in agreement with the role of HGF inducing cell migration. Interestingly, Y142F β-catenin-expressing cells showed a decrease of 20% in the number of round (loosely adherent/migratory) cells compared to WT β-catenin-expressing cells, indicating that mutant Y142F β-catenin increases cell adhesion. Furthermore, in contrast to WT β-catenin-expressing cells, HGF treatment did not increase the number of round cells in Y142F-expressing cells (the number of round cells in Y142F-expressing cells untreated or treated with HGF was not significantly different;Fig. 6B). These results indicate that PY142 β-catenin is required for cell detachment and migration of U87MG cells induced by HGF.

Figure 6.

PY142 β-catenin increases cell detachment and promotes GBM cell migration and invasion. (A) Transwell assay of U87MG cells expressing WT or Y142F β-catenin, untreated or following preincubation with SU11274 (2 μM, overnight). Y142F β-catenin-expressing cells showed reduced invasion compared to WT β-catenin-expressing cells. c-Met inhibition diminished invasion compared to untreated cells in a similar manner in WT and Y142F β-catenin-expressing cells. Plot represents values normalized to untreated cells expressing WT β-catenin. (B) Cell spreading assay performed in U87MG cells expressing Ds red (control), Ds red and WT or Y142F β-catenin. Cells were treated with HGF (100 ng/ml) where indicated. Plot represents the number of round cells (displaying low spreading/adhesion; high migration) normalized to control Ds red-expressing cells. HGF increased the number of round (pro-migratory) cells in Ds red and WT β-catenin-expressing cells, but not in cells expressing mutant Y142F β-catenin. Cells expressing Y142F β-catenin (untreated and treated with HGF) showed round cells numbers below those of control condition, indicating that the Y142F β-catenin mutant increases cell adhesion. (C and D) Transwell assays of C65 (C) and C18 (D) primary GBM cultures. GBM cells expressing WT or Y142F β-catenin were left untreated or treated with HGF and/or SU11274. Y142F β-catenin-expressing cells showed reduced invasion compared to WT β-catenin-expressing cells in both GBM cases. HGF significantly increased invasion of WT β-catenin-expressing cells, but not in cells expressing Y142F β-catenin. HGF together with c-Met inhibitor SU11274 reduced invasion vs. untreated cells expressing WT β-catenin. Plot represents values normalized to untreated cells expressing WT β-catenin (* p ≤ 0,05, ** p ≤ 0,01, *** p ≤ 0,001; T-test).

Invasion experiments were also performed in 2 primary GBM cultures, C65 and C18. HGF increased the invasion of GBM cells expressing WT β-catenin compared to untreated control cells in both GBM cultures. However, in cells expressing mutant Y142F β-catenin, the effect of HGF was blocked and the number of migratory cells remained below control values (Fig. 6C and 6D), suggesting that HGF signaling involves PY142 β-catenin. Treatment with HGF and the c-Met inhibitor reduced the number of migratory cells both in WT and Y142F β-catenin-expressing cells (45%-60% below control;Fig. 6C and 6D). These results demonstrate the involvement of PY142 β-catenin in GBM invasion downstream of HGF signaling.

Discussion

β-catenin is a transducer of Wnt signaling that plays key roles in physiology and disease. As a component of adhesive contacts, β-catenin controls cell adhesion and migration. High β-catenin expression was observed in gliomas, including astrocytomas and GBMs.²⁰ Furthermore, Wnt/β-catenin signaling is regulated by PLAGL2 and FoxM1 proto-oncogenes during gliomagenesis.^18,22 Here, we identified PY142 β-catenin as a nuclear β-catenin form that signals through transcriptional regulation in GBM. PY142 β-catenin is different from classical active β-catenin: PY142 β-catenin accumulates in the nucleus already in astrocytoma (grade II and III) samples and correlates with high levels of the migration marker Snail/Slug in tumor biopsies. In contrast, active β-catenin localizes to the nucleus in GBMs but not in astrocytomas, and appears to correlate with Cyclin D1 levels. PY142 β-catenin is regulated by HGF signaling in glioma cells. High invasion abilities are remarkable features of astrocytoma and GBM cells contributing to the extensive brain infiltration and to the tumor recurrence after treatment.⁴⁵ Importantly, studies using a Y142F β-catenin mutant indicate that HGF promotes cell migration and invasion through PY142 β-catenin in GBM cell lines and biopsy-derived GBM cultures.

Phosphorylation of β-catenin at Y142 is regulated in vitro by cytosolic tyrosine kinases Fer and Fyn and by RTK c-Met.^4,30,31,46 c-Met interacts with β-catenin⁴⁷ and PY142 β-catenin is increased in developing hippocampal neurons stimulated with HGF.³¹ We report here that PY142 β-catenin levels are detected in GBM biopsies and cell cultures (under basal conditions), which increase upon treatment with HGF in U87MG cells. U87MG cell line expresses HGF and c-Met^29,37,48, suggesting that autocrine or paracrine HGF/c-Met signaling could be responsible for the basal PY142 β-catenin levels found in GBM. Although active and total c-Met levels increase in grade III astrocytoma and grade IV GBM biopsies (compared to control and grade II astrocytomas) as observed for PY142 β-catenin, a correlation between c-Met and PY142 β-catenin could not be confirmed by Western blot in all the biopsies. Importantly, however, HGF decreases cell adhesion and stimulates cell migration and invasion, which are inhibited by mutant Y142F β-catenin in GBM-biopsy derived cultures. These results indicate that HGF signaling promotes GBM migration and invasion by regulating cell detachment through PY142 β-catenin.

Treatment with the c-Met inhibitor SU11274 does not affect PY142 β-catenin levels, while significantly reducing cell invasion in GBM cell lines and primary GBM cultures. These results confirm a key role of c-Met in GBM progression.^29,48-51 The lack of regulation of PY142 β-catenin by a c-Met inhibitor does not allow us to conclude if c-Met is phosphorylating this β-catenin site in GBM cells and rather suggests the involvement of other kinases. The proteolytic processing of c-Met is well described and nuclear c-Met has been linked to invasive cancers.^52-54 Proteolytic fragments of c-Met could explain the nuclear c-Met immunostaining observed in astrocytoma/GBM biopsies included in the TMA and in cell cultures. Catalytically active fragments of c-Met (unable to interact with SU11274) could regulate PY142 β-catenin in GBM, putatively explaining PY142 β-catenin basal levels and the lack of regulation by SU11274. This suggestion is supported by the correlation observed between nuclear PY142 β-catenin and nuclear c-Met in high grade astrocytomas.

Active β-catenin levels correlate with the expression of Cdc25a and Cyclin-D1, 2 well known regulators of the cell cycle and Wnt/β-catenin targets.^39,55 These findings are in agreement with a classical role of Wnt signaling in cell proliferation.⁵⁵ Snail/Slug proteins induce a pro-mesenchymal phenotype (typically working as transcriptional repressors) and control cell migration downstream of different signaling pathways.^40,56 Snail/Slug proteins have already been involved in GBM invasion.^57,58 We demonstrate the regulation of Snail/Slug, in parallel to increased PY142 β-catenin, upon stimulation of GBM cells with HGF. Statistical studies indicate that both nuclear Snail/Slug and nuclear PY142 β-catenin peak in astrocytomas of grade III. By Western blot, grade III astrocytoma and grade IV GBM biopsies show higher amounts of PY142 β-catenin than control and grade II samples. Interestingly, higher PY142 β-catenin is found in samples with the highest levels of Snail/Slug, further linking PY142 β-catenin and cell migration abilities. PY142 β-catenin may be considered a GBM biomarker conferring high invasion properties. Because different β-catenin forms are involved in GBM progression further studies on β-catenin are warranted to evaluate its possible therapeutic targeting in GBM.

Materials and Methods

Human biopsies

Tumor samples were obtained from Hospital Universitari Arnau de Vilanova (HUAV; Lleida) and Institute of Neuropathology (Hospital of Bellvitge, Barcelona) after obtaining specific informed consents from the patients. Control brain samples were obtained from autopsies (Institute of Neuropathology, Hospital of Bellvitge, Barcelona) within maximum 7 hours from death. The study was approved by the local Ethics Committee of Human Experimentation.

Reagents and antibodies

Most biochemical reagents were purchased from Sigma-Aldrich (Buchs, Switzerland) unless otherwise stated. Cell culture reagents were obtained from Thermo Fisher Scientific (Massachusetts, USA). Antibodies were obtained from: β-catenin, Cyclin D1 (DCS-6) from BD Bioscience (New Jersey, USA); PY142 β-catenin (Ab27798) and Snail/Slug (Ab6337) from Abcam (Cambridge, UK); c-Met (sc10 and sc161)and cdc25a (sc97) from Santa Cruz Biotechnology (Texas, USA); PY1234/5 c-Met (3126), β-catenin N-terminal (9581) and TCF-1 (C46C7) from Cell Signaling Technology (Massachusetts, USA); active β-catenin (8E7) from Merck-Millipore (Massachusetts, USA); TCF-4 from Upstate (California, USA), and β-actin (A5441) from Sigma-Aldrich (Buchs, Switzerland).

Tissue microarrays, immunohistochemistry and statistical analysis

Two tissue microarrays (TMA) were constructed including a total of 74 brain tissue samples, for which formalin-fixed paraffin-embedded blocks were available. The first TMA was constructed from 17 samples of astrocytoma tissue (7 grade II and 10 grade III). The second TMA was composed of 57 GBM samples. A Tissue Arrayer device (Beecher Instrument; Maryland, USA) was used to construct the TMAs. Briefly, all the samples were histologically reviewed and representative areas were marked in the corresponding paraffin-blocks. Two selected cylinders (0.6 mm in largest diameter) from 2 different areas were included in each case.

TMA blocks were sectioned at a thickness of 3 μm, dried for 1 h at 65° before pre-treatment procedure of deparaffinization, rehydration and epitope retrieval in the Pre-Treatment Module, PT-LINK (DAKO; Glostrup, Denmark) at 95°C for 20 min in 50x Tris/EDTA buffer, pH 9 or citrate buffer, pH 6.1. Before staining the sections, endogenous peroxidase was blocked. Antibodies used were rabbit polyclonal PY142 β-catenin (1:100 dilution), mouse monoclonal active-β-catenin (1:200 dilution), rabbit polyclonal c-Met (1:200 dilution; sc-161) and rabbit polyclonal Snail/Slug (1:200 dilution). The reaction was visualized with the EnVision FLEX Detection Kit (DAKO; Glostrup, Denmark) using diaminobenzidine chromogen as a substrate. Sections were counterstained with haematoxylin.

Immunohistochemical results were evaluated by following uniform pre-established criteria. Immunoexpression was graded semi-quantitatively by considering the percentage and intensity of the staining. A histological score was obtained from each sample, which ranged from 0 (no immunoreactivity) to 300 (maximum immunoreactivity). The score was obtained by applying the following formula, Histoscore = 1X (% light staining) + 2X (% moderate staining) + 3X (% strong staining). The reliability of such score for interpretation of immunohistochemical staining has been shown previously^59-61. Since each TMA included 2 different tumor cylinders from each case, immunohistochemical evaluation was done after examining both samples. Data are presented as mean (and standard deviation) or absolute frequency (and percentage) as convenient for quantitative or qualitative variables, respectively. Association of positivity with glioma histological type was assessed with a Fisher test. Differential immunostaining in the selected biomarkers with regards to histological type was assessed with a Mann-Whitney test. Relationship between biomarkers was assessed using Pearson correlation. All analyses were performed using R statistical package, setting the threshold for significance at 5 % (α = 0.05).

Tissue homogenates

Brain samples from control brain and glioma biopsies were homogenized in a glass homogenizer in 1 ml of homogenizer buffer containing 5 M urea, 50 mM Tris-HCl pH 7,4, 100 mM NaCl, 10 mM EDTA, 0.5% sodium deoxycholate, 0.5% Igepal CO-630, 2% sodium dodecyl sulfate (SDS) and Complete protease inhibitor cocktail (Roche; Basel, Switzerland) plus phosphate inhibitors (25 mM sodium fluoride, 1 mM sodium orthovanadate and 40 mM β-glycerophosphate). Homogenates were then centrifuged (15.000 g, 5 min, 4°C) and pellets were discarded. The protein concentration of the resulting supernatants was determined using the BCA protein assay reagent (Thermo Scientific; Massachusetts, USA) and bovine serum albumin as standard.

Cell culture

Tumor biopsies were obtained from HUAV (Lleida) and processed before 30 min from surgical dissection. All patients signed specific informed consents prior to collection of specimens. To establish adherent GBM cultures⁶², tissue was washed with phosphate buffer saline (PBS), cut into small pieces (2 mm²) and incubated (2 h, 37°C under shaking) in PBS containing 155 U/ml of collagenase (Worthington; Ohio, USA) and 12 μg/ml of DNase-I. Samples were filtered through a 70 μm cell strainer (BD; New Jersey, USA) and the cell suspension was washed twice with PBS. Cells were plated in DMEM media containing 10% FBS. Media was changed every 2 d.

Glioma cel lines U251MG and U87MG were obtained CLS Cell Lines Service (Eppelheim, Germany). Cells were maintained in MEM media containing 10% foetal bovine serum (FBS), penicillin/streptomycin, L-glutamine, aminoacids and sodium pyruvate at 37°C and 5% CO2 atmosphere and subcultured by trypsinization 1:3 twice a week.

Western blot

After the appropriate treatments (HGF 100 ng/ml, Preproptech, New Jersey, USA; SU11274 2 μM, Sigma-Aldrich, Buchs, Switzerland), cells were washed with PBS and lysed in Tris 62,8 mM pH 6,8 and 2% SDS. Cell lysates were resolved in 8 or 10% acrylamide-polyacrylamide gels and transferred to a PVDF membrane. Membranes were blocked with 5% bovine serum albumin or milk and incubated overnight with primary antibodies. Blots were developed using Enhanced Chemiluminiscence (ECL, Thermo Scientific; Massachusetts, USA) or Luminata Forte horseradish peroxidase substrate from Merck-Millipore (Massachusetts, USA), kits. Band intensities were quantified with respect to β-actin (used as loading control) using Image J software.

Immunofluorescence

Cells were fixed with 4% paraformaldehyde (PFA; 20 min, room temperature), washed with PBS and blocked and permeabilized in PBS containing 5% FBS, 5% horse serum, 0.2% glycine and 0.1% Triton X100 before incubation with primary antibodies. Secondary antibodies were coupled to Alexa Fluor488 or Alexa Fluor594 (Thermo Scientific; Massachusetts, USA). Coverslips were mounted on Mowiol. Micrographs were obtained using an inverted Olympus IX70 microscope (10x, 0.3 NA or 10x, 0.4 NA; Pennsylvania, USA) equipped with epifluorescence optics and a camera (Olympus OM-4 Ti). Images were acquired using DPM Manager Software.

Transfection

Glioma cell lines and primary cultures were transfected with WT, Y142F or Y654F β-catenin expression vectors (a kind gift of M. Duñach, UAB, Barcelona, Spain). Cells were plated the day before transfection at 80% of confluence. 50 μl of Optimem containing the DNAs were mixed with 50 μl of Optimem containing 1,5 μl of Lipofectamine 2000 (Thermo Scientific; Massachusetts, USA). The mix was incubated (10 min, room temperature) and added dropwise to the cells for 4 h, when media was removed and complete media added.

Transwell invasion assay

To mimic 3-dimensional extracellular environment and investigate the infiltrative potential of the cells, the external face of Transwell polycarbonate membrane inserts (8.0 μm pore size; BD Biosciences; New Jersey, USA) was coated with Matrigel (BD Biosciences; New Jersey, USA). After 30 min at 37°C the Matrigel solidified and served as an extracellular matrix for invasion analysis. After 48 h of transfection, cells were trypsinized and resuspended in 200 μl of media containing 10% FBS without or with HGF (100 ng/ml). Cells (20.000 cells/well) were placed on top of the inserts and the bottom chambers were filled with serum-containing media. For c-Met inhibition, cells were pre-incubated overnight with 2 μM SU11274. Cells were incubated for 8 h at 37°C and then fixed with 4% PFA. Non-migrating cells were removed from the upper face of the insert using cotton swabs. Cell invasion into the Matrigel-coated lower side of the membrane was detected by Hoechst staining. Experiments were performed in duplicates. Six random fields were pictured under an epifluorescence microscope and nuclei counted. Results were expressed as fold increase vs. the number of nuclei found in the Matrigel layer in control conditions. Statistical significance was calculated by T-test.

Spreading assay

Cells were co-transfected with DsRed (0,2 μg) and WT or Y142F β-catenin (0,8 μg) expression vectors. Transfected cells were incubated for 48 h and treated with HGF (100 ng/ml, 1 h) and with SU11274 (2 μM, for the last 16 h) as required. Cells were then trypsinized and seeded on plates coated with 10 μg/ml fibronectin. After 30–45 min cells were fixed with 2% PFA and washed with PBS. Phase contrast and DsRed fluorescence pictures were taken. Spread cells (displaying DsRed fluorescence) were counted. Round bright cells under phase contrast microscopy were considered not spread (displaying low adherence and high migration abilities).⁴⁴

Luciferase activity

To determine β-catenin transcriptional activation status, luciferase assay was performed following transfection of the TOP-Flash plasmid (1,4 μg) that carries a synthetic promoter containing 3 copies of the TCF-4 binding site upstream of a firefly luciferase reporter gene. U87MG cells were plated at a density of 200 cells/mm² and transfected with Lipofectamine 2000 on the day after plating. When indicated, cells were co-transfected with TOP-Flash (0,7 μg) and β-catenin (0,7 μg) DNAs. Treatments (HGF 50 ng/ml; Wnt-3a 100 ng/ml; SU11274 2 μM) were applied for 24 h following transfection. After 48 hours of transfection, cells were lysed in lysis buffer 25 mM glycylglycine, pH 7.8, 15 mM Mg₂SO4, 1% Triton X-100, 5 mM EGTA) and rocket on ice for 15 minutes. Luciferase activity in the cell lysates was determined in Luciferase Buffer (25 mM glycylglycine, 15 mM KHPO₄, pH 7,8, 15 mM Mg₂SO₄, 1% Triton X-100, 5 mM EGTA, 1 mM dithiothreitol, 2 mM ATP, 100 mM acetyl-coenzymeA and 100 mM luciferin) using a microplate fluorescent reader FLX 800 (Biotek Instruments; Vermont, USA). Luciferase activity was normalized against protein concentration in each condition.

Acknowledgments

We thank Mireia Dunach (Universitat Autonoma de Barcelona, Barcelona, Spain) for the gift of WT, Y654F and Y142F β-catenin plasmids.

Supplementary Material

Supplemental data for this article can be accessed on the publisher's website.

Funding

This work was funded by grants of Instituto de Salud Carlos III with the support of Fondo Europeo de Desarrollo Regional (FEDER) to J.H. (PI080790 and PI1301980). M.N. is a predoctoral fellow of UdL and D.B. was a predoctoral fellow of Agaur-Generalitat de Catalunya. IRBLleida Biobank is supported by Xarxa de Bancs de Tumors de Catalunya (XBTC), IRBLleida Biobank (B.0000682) and Plataforma Biobancos PT13/0010/0014.