Different behaviour of DVL1, DVL2, DVL3 in astrocytoma malignancy grades and their association to TCF1 and LEF1 upregulation

Abstract Key regulators of the Wnt signalling, DVL1, DVL2 and DVL3, in astrocytomas of different malignancy grades were investigated. Markers for DVL1,DVL2 and DVL3 were used to detect microsatellite instability (MSI) and gross deletions (LOH), while immunohistochemistry and immunoreactivity score were used to determine the signal strengths of the three DVL proteins and transcription factors of the pathway, TCF1 and LEF1. Our findings demonstrated that MSI at all three DVL loci was constantly found across tumour grades with the highest number in grade II (P = 0.008). Collectively, LOHs were more frequent in high‐grade tumours than in low grade ones. LOHs of DVL3 gene were significantly associated with grade IV tumours (P = 0.007). The results on protein expressions indicated that high‐grade tumours expressed less DVL1 protein as compared with low grade ones. A significant negative correlation was established between DVL1 expression and malignancy grades (P < 0.001). The expression of DVL2 protein was found similar across grades, while DVL3 expression significantly increased with malignancy grades (P < 0.001). The signal strengths of expressed DVL1 and DVL3 were negatively correlated (P = 0.002). However, TCF1 and LEF1 were both significantly upregulated and increasing with astrocytoma grades (P = 0.001). A positive correlation was established between DVL3 and both TCF1 (P = 0.020) and LEF1 (P = 0.006) suggesting their joint involvement in malignant progression. Our findings suggest that DVL1 and DVL2 may be involved during early stages of the disease, while DVL3 may have a role in later phases and together with TCF1 and LEF1 promotes the activation of Wnt signalling.


| INTRODUCTION
Astrocytomas are the most common and deadliest form of primary brain tumours. 1 According to the World Health Organization (WHO) classification, there are four grades of astrocytic brain tumours, considering their histology, molecular characteristics and prognosis. 2 The least aggressive are pilocytic astrocytomas corresponding to WHO grade I, while diffuse (grade II) and anaplastic astrocytomas (grade III) are malignant types with intrinsic ability to progress to higher grades of malignancy. Of these, glioblastoma multiforme (grade IV) is the most aggressive, fastest-growing and highly invasive tumour with survival times of about a year. 3 Despite recent advances in understanding the molecular basis of astrocytoma development and progression, additional research is required to develop more effective therapies.
By its biological characteristics astrocytomas are genetically and pathohistologically a very heterogeneous group of tumours. Complex mechanism of gliomagenesis is the outcome of overlapping between altered signalling pathways. A long-scale study conducted by The Cancer Genome Atlas (TCGA) revealed numerous data on specific genetic and epigenetic alterations underlying gliomas. 4 Although it is still not possible to define the exact number and chronology of changes during gliomagenesis, they found that key genes responsible for the formation and progression of astrocytic brain tumours are most frequently involved into deregulated oncogenic pathways: RTK/RAS/PI3K, TP53 and RB. 4,5 Recent studies have shown that altered signal transduction in Wnt signalling pathway is also involved in the molecular pathogenesis of glial tumours. [6][7][8][9] The Wnt signalling is an evolutionarily conserved pathway that plays very important roles during embryonic development and tumourigenesis. 10 In the absence of Wnt ligand, cytoplasmic β-catenin protein is constantly being degraded by the action of the destruction complex composed of Axin, APC and 2 phosphokinases: casein kinase 1 (CK1) and glycogen synthase kinase 3 (GSK3). CK1 and GSK3 phosphorylate β-catenin, which leads to its proteasomal degradation. Continuous elimination of β-catenin from the cell prevents β-catenin from reaching the nucleus, thereby repressing transcription of Wnt target genes. 11,12 Binding of Wnt ligand to a Frizzled (Fz) receptor and its coreceptor LRP5/6 activates Wnt/β-catenin pathway. In this stage the concentration of DVL protein in cytoplasm increases, resulting in recruitment of components of destruction complex to the cell membrane. These events lead to inhibition of β-catenin phosphorylation, its stabilization and accumulation in the cytoplasm. β-catenin transfers to the nucleus where it forms complexes with T cell factor 1 (TCF1) and lymphoid-enhancer factor 1 (LEF1) and in such a fashion activates Wnt target gene transcription. 13 DVL is the central component of Wnt signalling and key cytoplasmic regulator that rescues βcatenin from degradation. Three homologous Dishevelled genes (DVL1, DVL2 and DVL3), that show a high degree of similarity, have been found in humans. Knock-out mouse models have shown that each of the DVL proteins can act individually but also in the combination with other family members. 14 Dishevelled gene proteins function as branching points for the differential interpretation of distinct Wnt ligands. 15 They are activated by phosphorylation in response to Wnt signals; whereas the ubiquitination of DVL proteins leads to an effective inhibition of the β-catenin destruction complex. 13 It has been postulated that DVL overexpression may play a role in the progression of several cancers. [16][17][18][19][20] As a result it represents a potential target for cancer therapy.
In the present investigation, we searched for changes of all three human DVL genes and proteins and tried to define their involvement in specific astrocytoma grade. We were also interested about their effect on Wnt signalling activation and examined transcription factors TCF1 and LEF1.

| Tumour specimens
Eighty-three astrocytoma samples together with corresponding autologous blood tissue were collected with patients' consents from the Departments of Neurosurgery and Departments of Pathology University Hospital Centers "Zagreb" and "Sisters of Charity", Croatia and Brain Tumour Tissue Banks from Croatia and Canada.
All tumours were studied by certified neuropathologist and classified according to the WHO criteria. 2 The patients had no family history of brain tumours and did not undergo any cancer treatment (chemotherapy or radiotherapy) prior to surgery. The sample consisted of 17  10 mmol L −1 EDTA; pH 7.6) and centrifuged (15 min/5000 g). After overnight incubation with 2 mL SE buffer (Sodium-EDTA; 75 mmol L −1 NaCl; 25 mmol L −1 Na 2 EDTA; pH 8), 200 μL 10% SDS (Sodium-dodecyl sulphate) and 15 μL proteinase K (Sigma, Germany) (20 mg/ mL) at 37°C, salting-out method by isopropanol precipitation followed.
Marker D3S1262 was also used for multiplex PCR. It was amplified in the same reaction together with markers for genes SHGC-68373 (222 bp) 22 and Apex1 (321 bp) 23

| Microsatellite genotyping analysis
Loss of heterozygosity was confirmed by capillary electrophoresis performed on instrument 3730XL (Applied Biosystems Inc.). The 5′end of the forward primer was fluorescently labelled with a 6-FAM dye, and PCR amplification was performed with the AmpliTaq Gold master-mix (360) (Applied Biosystems Inc.). The labelled PCR products were separated using capillary electrophoresis. Alleles were sized relative to an internal size standard (500 LIZ; Thermo Fisher Scientific, Carlsbad, CA, USA). Raw data and graphical representation of LOH samples were reviewed using GeneMapper 5 and Peak Scanner (Thermo Fisher Scientific).

| Immunohistochemistry
In order to establish the levels of expression and cellular localization of DVL1, DVL2, DVL3 gene products immunohistochemistry was The level of expression of DVL1, DVL2 and DVL3 proteins in the healthy brain was determined by using cerebral cortex of human brain (Amsbio, Oxfordshire, UK). We have found that the levels of immunoreactivity of all homologues of the Dishevelled protein family in the healthy brain tissue were very low, and the signal was present only in the cytoplasm.
Negative controls underwent the same staining procedure but without incubating samples with primary antibodies. The frontal cortex of a healthy human brain, placenta, liver tissue and normal bronchial epithelia all served as positive controls. Antibody labelling was analysed by three independent and blinded observers using an Olympus BH-2 microscope and a digital scanner (NanoZoomer  We counted 200 cells in tumour hot spot area and performed semiquantitative analysis introducing immunoreactivity score (IRS) in order to determine the signal strength. Immunoreactivity score is a factor that best correlates with computational photo analysis and was calculated by multiplying the percentage of cells with a positive signal in the sample (PP score) with staining intensity (SI score). PP score was determined as follows: no immunopositivity in tumour cells = score 0; 1%-25% positive cells = score 1; >25%-50% = score 2; >50%-85% = score 3; >85% = score 4. SI score was assessed in three categories mirroring staining intensities, no staining or weak = score 1, moderate staining = score 2 and strong staining = score 3.
Immunoreactivity score in our study ranged from 0 to 12.   Figure 1A).

| Statistical analysis
For DVL2 gene analysis marker D17S960 was selected and it showed a lower rate of MSI in all tumour grades, 7.1% in grade I, 15.4% in grade II, 0% in grade III and 8.6% in grade VI. Again grade II tumours (diffuse) harboured the highest frequency of MSI. However, gross deletions of DVL2 were detected in all grades ( Figure 1B).
Thus, grade II and III each harboured 33.3% of LOHs, followed by grade IV with 21.7% and grade I with the lowest frequency of 12.5% ( Table 2). The distribution of genetic changes was not significantly associated to any specific grade (P = 0.843).
DVL3 gene was analysed with D3S1262 marker ( Figure 1C). To ascertain results on observed genetic changes, genotyping by capillary electrophoresis was additionally performed and LOHs were confirmed as shown in Figure 1D. Taking both approaches together our findings were additionally confirmed.
Besides gross deletions, DVL3 gene also showed amplifications in 8.6% of glioblastoma patients ( Figure 1C). Polymorphic status of used microsatellite markers and observed genetic changes of DVL1, DVL2 and DVL3 genes are presented in Table 2   This observation suggests that the increased DVL3 expression in glioblastoma leads to more frequent transfer of this protein into the nucleus (Figure 4).

| TCF1 and LEF1 protein expression levels increased with astrocytoma grades
Next we asked how the Dishevelleds' expression is impacting Wnt signalling activation. For this we analysed the expression levels of two transcriptional factors located at the end of the Wnt signalling cascade, TCF1 and LEF1, whose elevated expression indicates the activity of the pathway. Both factors were found to be expressed in our total sample where the levels of both TCF1 and LEF1 expression increased with astrocytoma grades. Accordingly, 61.1% of pilocytic astrocytomas showed the lowest levels of TCF1 expression, while weak or lack of TCF1 expression was confined to 50% of diffuse astrocytomas. The highest expression levels were found in anaplastic and glioblastomas cases, 77.8% and 80% respectively. The signal was present exclusively in the nuclei of tumour cells ( Figure 3D).
Statistical analysis for TCF1 protein showed that pilocytic had significantly more cells with low expression levels as compared with diffuse astrocytomas (P = 0.040) and glioblastomas (P = 0.003). Furthermore, diffuse (P = 0.047), anaplastic (P = 0.022) astrocytomas and glioblastomas (P = 0.001) had significantly higher number of cells with strong TCF1 expression as compared with pilocytic ( Figure 3D).
Similar results were obtained for LEF1 protein expression. Low or lack of LEF1 expression was found in 61.1% of pilocytic astrocytomas, while 70% of glioblastomas showed strong LEF1 expression.
Almost the same proportions of samples with moderate and strong expression were present in the group of diffuse astrocytoma (88.9%) and glioblastoma (90%), while in anaplastic cases the observed rate was lower (66.6%).
The cell count analysis confirmed the previous observations. Thus, pilocytic astrocytomas had significantly more cells with low LEF1 expression than both diffuse astrocytomas (P = 0.006) and glioblastomas (P = 0.001) ( Figure 3E). The highest number of cells with strong LEF1 expression was in glioblastomas comparing to pilocytic (P < 0.001), diffuse (P = 0.032) and anaplastic types (P = 0.003). Diffuse astrocytomas had significantly more cells with strong LEF1 expression as compared with pilocytic (P < 0.001) and anaplastic (P = 0.008), while anaplastic had significantly more cells with strong LEF1 expression than pilocytic (P = 0.008) ( Figure 3E).
The results on transcription factors suggest that Wnt activation is accompanying the progression of the disease.

| The correlations of molecular findings and clinical parameters
Immunoreactivity score values for DVL1, DVL2 and DVL3 were calculated in order to determine the potential correlations between the expression levels of the proteins.
There was no significant correlation between the expression levels of DVL1 and DVL2 proteins (P = 0.799) ( Figure 5A). However, DVL1 and DVL3 were significantly negatively correlated in our total sample (P = 0.002) ( Figure 5B). Lower tumour grades (I and II) showed significantly higher levels of DVL1 expression than high grades (III and IV), while strong DVL3 expression was significantly rare are studies about the roles of DVL1, DVL2 and DVL3 in human astrocytic brain tumours. [27][28][29] The results of the present study showed that genetic and protein changes of all three Dishevelleds have distinct roles in the process of astrocytoma formation and progression. were confined to diffuse cases suggesting that the increased genomic instability could be characteristic for astrocytomas that are able to progress to higher grades. It has been known that the increase of mutation accumulation results in an acceleration of the tumour cell evolution. 30  genes. However, the mutations that have been found are much more frequent in glioblastomas than low-grade tumours. The reported rate of amplifications of DVL3 is similar to the ones we found.
Taken together, we showed that MSI occurred constantly across all four tumour grades, with more frequent incidence in lower astrocytoma grades, suggesting its association with tumour formation.
Loss of heterozygosity was found to be often present in anaplastic astrocytoma and glioblastoma and therefore could be involved in the process of tumour progression as a background mechanism for inactivation of tumour suppressor genes. All things considered, it is clear that LOH and MSI contribute to the genomic profile of astrocytoma. 46 59 and promotes epithelial-to-mesenchymal transition. 62 The results of our study showed that both TCF1 and LEF1 showed elevated expression and this upregulation indicates that the Wnt pathway was activated. Moreover, the levels of both TCF1 and Positive correlation was established between DVL3 and both TCF1 (r s = 0.410, df = 30, P = 0.020) and LEF1 (r s = 0.475, df = 30, P = 0.006) indicating once again DVL3 link to progression.
There are several studies investigating the role of Wnt signalling in human astrocytoma. 27 Hospital Centers "Sisters of Charity" and "Zagreb".

ACKNOWLEDGEMENT
We thank Brain Tumour Foundation of Canada for donating brain tumour samples.

CONFLI CT OF INTEREST
All authors declare that they have no conflicts of interest.