Epigenetic inactivation of the RASSF10 candidate tumor suppressor ...

Oncogene (2011) 30, 978–989

& 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc

ORIGINAL ARTICLE

Epigenetic inactivation of the RASSF10 candidate tumor suppressor gene is a frequent and an early event in gliomagenesis VK Hill1,4, N Underhill-Day1,4, D Krex2, K Robel2, CB Sangan3, HR Summersgill3, M Morris1, D Gentle1, AD Chalmers3, ER Maher1 and F Latif1 1

Medical and Molecular Genetics, School of Clinical and Experimental Medicine, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK; 2Klinik und Poliklinik fu¨r Neurochirurgie, Universita¨tsklinikum Carl Gustav Carus, Technische Universita¨t Dresden, Dresden, Germany and 3Department of Biology and Biochemistry, Centre for Regenerative Medicine, University of Bath, Bath, UK

We have recently described the N-terminal RAS association domain family of genes, RASSF7–10. Previously, we cloned the N-terminal RASSF10 gene and demonstrated frequent methylation of the associated 50 -CpG island in acute lymphoblastic leukemia. To characterize RASSF10 gene expression, we demonstrate that in developing Xenopus embryos, RASSF10 shows a very striking pattern in the rhombencephalon (hind brain). It is also expressed in other parts of the brain and other organs. Due to the well-defined expression pattern in the brain of Xenopus embryos, we analyzed the methylation status of the RASSF10-associated 50 -CpG island in astrocytic gliomas. RASSF10 was frequently methylated in WHO grade II–III astrocytomas and WHO grade IV primary glioblastomas (67.5%), but was unmethylated in grade I astrocytomas and in DNA from age matched control brain samples. RASSF10 gene expression both at the mRNA and protein levels could be switched back on in methylated glioma cell lines after treatment with 5-aza-20 -deoxycytidine. In secondary glioblastomas (sGBM), RASSF10 methylation was an independent prognostic factor associated with worst progression-free survival and overall survival and occurred at an early stage in their development. In cell culture experiments, overexpression of RASSF10 mediated a reduction in the colony forming ability of two RASSF10methylated glioma cell lines. Conversely, RNAi-mediated knockdown of RASSF10-stimulated anchorage-independent growth of U87 glioma cells, increased their viability and caused an increase in the cells’ proliferative ability. We generated and characterized a RASSF10-specific antibody and demonstrated for the first time that RASSF10 subcellular localization is cell-cycle dependent with RASSF10 colocalizing to centrosomes and associated microtubules during mitosis. This is the first report demonstrating that RASSF10 can act as a tumor suppressor gene and is frequently methylated in gliomas and can potentially be developed into a prognostic marker for sGBM. Correspondence: Professor F Latif, Medical and Molecular Genetics, School of Clinical and Experimental Medicine, College of Medical and Dental Sciences, Institute of Biomedical Research, University of Birmingham, Birmingham, West Midlands B15 2TT, UK. E-mail: [email protected] 4 These authors contributed equally to this work. Received 7 June 2010; revised 11 August 2010; accepted 3 September 2010; published online 18 October 2010

Oncogene (2011) 30, 978–989; doi:10.1038/onc.2010.471; published online 18 October 2010 Keywords: gliomas; RASSF10; epigenetics

Introduction The RAS association domain family of genes comprises 10 members (RASSF1-10). These are subdivided into classical (RASSF1-6) and N-terminal (RASSF7–10) RASSF genes (Richter et al., 2009; Sherwood et al., 2009). RASSF1A, the first member of this gene family, is one of the most frequently methylated genes in a wide range of tumor types, including lung, breast, kidney and brain tumors (Hesson et al., 2007a). Biologically, RASSF1A is also the best-characterized member of the group. RASSF1A is a bona fide tumor suppressor gene; RASSF1A knockout mice are prone to tumor development (Tommasi et al., 2005; van der Weyden et al., 2005), whereas RASSF1A protein is involved in microtubule stability, cell cycle progression, regulation of apoptosis, cell migration and forms a part of the mammalian Hippo tumor suppressor network (Donninger et al., 2007; Hesson et al., 2007b; Dallol et al., 2009). Some of the classical RASSF members have been shown to bind RAS proteins and hence may act as RAS effectors. RASSF7–10 have only recently been identified and little information is available regarding genetic/ epigenetic and biological properties of these latest RASSF family members. The RASSF7–10 proteins differ structurally from the classical members in having a RAS association domain/ubiquitin fold at their N terminus (C terminal in classical RASSF members) while also lacking the SARAH domain (SALVADORRASSF-HIPPO) present in RASSF1-6. These differences may have implications for protein interactions and subsequent biological function. We have demonstrated that RASSF7 is important for cell cycle progression and cell survival (Sherwood et al., 2008). There is some evidence that RASSF7 expression is upregulated in certain cancers, hence it may promote cancer development, whereas the majority of the classical RASSF genes show loss of gene expression in tumors and exhibit tumor suppressor properties. There is increasing evidence

RASSF10 and gliomagenesis VK Hill et al

979

that RASSF8 may act as a tumor suppressor gene in lung cancer; we have demonstrated that RASSF8 is essential for maintaining adherens junction function in epithelial cells and has a role in epithelial cell migration (Lock et al., 2010). Drosophila RASSF8 (dRASSF8) has been shown to interact with Drosophila ASSP (dASSP) and this interaction is conserved in mammals. The dASPP–dRASSF8 complex has been shown to regulate cell–cell adhesion (Langton et al., 2009). RASSF9 has been shown to bind to RAS proteins and is involved in vesicle trafficking (Chen et al., 1998; Rodriguez-Viciana et al., 2004). We have recently cloned and characterized the latest member of the N-terminal RASSF gene family, namely, RASSF10 and demonstrated that RASSF10 is frequently methylated in acute lymphoblastic leukemia (Hesson et al., 2009). Nothing is known about the biological properties of this latest N-terminal

RASSF protein. In the present report, we analyzed the in vivo expression of RASSF10 in Xenopus embryos and determined the methylation status of the RASSF10 gene in a large panel of astrocytic gliomas. This is the first report describing some of the biological properties of the RASSF10 protein. Results RASSF10 expression in Xenopus tadpoles To begin to establish which tissues express RASSF10, tadpoles of the model organism Xenopus laevis were used for in situ hybridization. Prominent expression of Xenopus RASSF10 was found in many areas of the brain and in several other tissues (Figures 1a and b). In the brain, RASSF10 expression was seen in the dorsal

Xenopus RASSF10 me

di

rh

df

te ey

cg

tg

ba

1

2

me

1

2

3

cg

4

ey ba

4

rh

Spinal cord

3

Hindbrain

midbrain

Forebrain

te

df

Figure 1 Xenopus RASSF10 shows prominent expression in the brain. In situ hybridization was used to analyze the expression of Xenopus RASSF10 in stage-30 tadpoles. Expression was found in many regions of the brain and also in the eye, trigeminal ganglion, branchial arches, cement gland and dorsal fin. (a) A lateral view of a stage-30 wholemount RASSF10 in situ hybridization (anterior to the left and dorsal to the top). (b) Sections of a stage-30 tadpole showing expression of RASSF10 in the forebrain, midbrain and hindbrain (dorsal to the top). Expression was not found in the spinal cord but was present in the overlying dorsal fin. The wholemount image shows the position of the sections along the anterior/posterior axis. ba, branchial arches; cg, cement gland; df, dorsal fin; di, diencephalon (forebrain); ey, eye; me, mesencephalon (midbrain); rh, rhombencephalon (hindbrain); Te, telencephalon (forebrain); tg, trigeminal ganglion. Oncogene


980 A172

H4

RASSF10

α-Tubulin UnM -

+

M

U

A172

U

D H4

D

U

D

U

U343

U87

D T17

No RT controls A172 +

H4 -

+

A172 -

+

H4 -

+

-

Merge

A172

5-Aza-2’-dC

untreated

RASSF10

GAPDH

Hoescht

5-Aza-2’-dC

D

+

untreated

RASSF10

U

-

5-Aza-2’-deoxycytidine

H4 Figure 2 Cell line methylation and expression status. (a) COBRA results are shown for glioma cell lines with undigested product next to digested products. (b) Reexpression at the mRNA level following treatment with 5-azaDC ( þ ) is shown for methylated cell lines, H4 and A172, next to untreated cell lines (). Corresponding GAPDH results are shown along with no RT controls for both RASSF10 and GAPDH. (c) Reexpression following treatment with 5-azaDC ( þ ) at the protein level is shown for methylated cell lines, A172 and H4, by western blot (top) and immunofluorescence (bottom) using an anti-RASSF10 antibody. D, digested product; M, methylated product; U, undigested product; UnM, unmethylated product.

and ventral forebrain (telencephalon and diencephalon), weaker expression in the dorsal midbrain (mesencephalon) and striking striped expression in the hindbrain (rhombencephalon). No expression was detected in the spinal cord, although expression could be seen in the overlying dorsal fin. RASSF10 was also expressed in other neural tissues, the eye and the trigeminal ganglion, a nerve that innervates the cement gland. In addition, RASSF10 had expression in the dorsal fin, cement gland and branchial arches. RASSF10 methylation and gene expression in glioma cell lines Due to the striking pattern of RASSF10 in the Xenopus brain, we analyzed the methylation status of the 50 -CpG island associated with the RASSF10 gene in a series of well-characterized human astrocytic gliomas and glioma cell lines. Using our previously designed combined Oncogene

bisulfite restriction analysis (COBRA) primers specific for the RASSF10 50 -CpG island (Hesson et al., 2009), we demonstrated that four (A172, H4, U343, T17) glioma cell lines were heavily methylated for RASSF10, whereas glioma cell line U87 was unmethylated (Figure 2a). We treated two (A172 and H4) of the methylated glioma cell lines with 5-aza-20 -deoxycytidine (5-azaDC) and showed that RASSF10 was reexpressed after treatment with the demethylating agent at the mRNA level (Figure 2b). We also generated and characterized a RASSF10-specific antibody (Supplementary Figure S3). The methylated glioma cell lines demonstrated that RASSF10 protein could also be reexpressed after treatment with 5-azaDC (Figure 2c). RASSF10 methylation in gliomas We analyzed a series of well-characterized astrocytic gliomas (grades I-IV, see Supplementary Table S2-S3 for patient characteristics) for RASSF10 methylation


981

NONCANCEROUS BRAIN TISSUE

UnM M U

D NO1

U

D

NO2

U

D

NO3

D

U

NO4

D

U

NO5

N05

MI = 2.2%

A3

MI = 1.4%

A20

MI = 44.5%

A26

MI = 82.8%

G3

MI = 59.6%

UnM M

GRADE I U

D

U

D

U

U

A5

A4

A1

D

D

U

D

A3

A6

UnM GRADE II

M U

D

A15

U

D

U

A19

D

U

A12

D

U

A11

D

A20 UnM M

GRADE III U

D

U

D

A23

A22

U

D A27

U

D

A21

U

D

A26 UnM M

GRADE IV D

U G8

U

D

G16

U

D

G17

D

U G4

U

D G3

Figure 3 Primary glioma tumor methylation. COBRA results are shown for a selection of samples for each tumor grade and noncancerous brain tissue. Undigested products are shown next to digested products. Clone sequencing results for one sample of each tumour type are shown with methylation indexes. Each line represents a single allele, white circles represent unmethylated CpGs and black circles represent methylated samples. RASSF10 gene schematic is shown with RASSF10 in dark gray, CpG island in black and region being analyzed by COBRA is shown in light gray. D, digested product; M, methylated product; MI, methylation indexes; U, undigested product; UnM, unmethylated product.

status. RASSF10 was unmethylated in all WHO grade I astrocytomas (n ¼ 10), whereas WHO grade II, III astrocytomas and grade IV primary glioblastoma multiforme (pGBM) were frequently methylated for RASSF10 (grade II, 6/10, 60%; grade III, 8/10, 80%; grade IV pGBM, 13/20, 65%), (P ¼ 0.0001 grade I vs grade II/III/IV) (Supplementary Table S1 and Figure 3). In non-neoplastic brain DNA samples, RASSF10 showed weak methylation in 1 out of 14 samples. To confirm the COBRA results and to determine the extent and pattern of methylation within the CG-rich region, we cloned and sequenced colonies from bisulfite modified gliomas and control brain DNA samples. Bisulfite sequencing confirmed our COBRA results, showing heavy methylation across the 50 -CG-rich region in gliomas, whereas control brain DNA samples were unmethylated across the same region (Figure 3 and Supplementary Figure S1). Glioblastomas arising from lower-grade gliomas (WHO grade II or WHO grade III) are referred to as secondary glioblastomas (sGBM), but more frequently they develop without evidence of a lesser malignant

precursor entity, these are referred to as pGBM. sGBM and pGBM are clinically distinct glioblastomas and there is evidence that they also show distinct genetic profiles. Hence, we determined the methylation status of the RASSF10 gene in pGBMs as well as sGBMs. RASSF10 was equally methylated in the two types of glioblastomas (pGMB, 13/20, 65%; sGBM, 9/13, 69%), Supplementary Table S1. MGMT and RASSF1A methylation and IDH1 and IDH2 mutation status in gliomas We have previously demonstrated that MGMT and RASSF1A methylation occur frequently in gliomas. In the present series of gliomas, MGMT and RASSF1A were methylated in 26 and 84%, respectively (Supplementary Figure S2, Supplementary Table S1). There was no association between methylation status of RASSF10 and RASSF1A or MGMT methylation. A recent genome-wide mutational study identified somatic mutations at codon 132 of the isocitrate dehydrogenase 1 (IDH1) gene in B12% of glioblastomas Oncogene


982 100

80

unmethylated

Survival probability (%)

Survival probability (%)

100

60 40

methylated

20 0

unmethylated

80 60 40

methylated

20

P= 0.0318

P= 0.0231 0

20 40 60 80 100 120 140 160 Days

0

20

40

60 80 100 120 140 Days UnM M

U

D

U

grade II

D

grade IV 1

U

D

U

grade II

D

grade IV 2

Figure 4 Survival of patients with sGBM. Kaplan–Meier survival curves for (a) overall survival and (b) progression-free survival in relation to RASSF10 methylation status. (c) RASSF10 methylation in sGBM and corresponding grade II astrocytomas. COBRA digestion products are shown next to undigested products for grade II and subsequent grade IV gliomas for two cases. D, digested product; M, methylated product; U, undigested product; UnM, unmethylated product.

(Parsons et al., 2008; Yan et al., 2009a), most frequently in tumors that were known to have arisen from lowergrade gliomas (sGBM). A further study demonstrated that gliomas without IDH1 mutations often had mutations in the corresponding amino-acid (R172) of the isocitrate dehydrogenase 2 (IDH2) gene. Patients with mutations of the IDH genes had a better clinical outcome than those with wild-type IDH genes. In our series of gliomas, IDH1 mutations were identified in 10 of the 20 grade II and grade III gliomas, in 4 out of the 10 sGBM and in 2 of the 20 pGBM (P ¼ 0.0119 grade II/III/sGBM vs pGBM), no IDH2 mutations were detected in any of our samples (Supplementary Table S1). There was no association between IDH1 mutations and RASSF10 or RASSF1A methylation. In sGBM, RASSF10 methylation was associated with worst overall survival and worst progressionfree survival (P ¼ 0.0318 and 0.0231, respectively) (Figures 4a and b). With Cox proportional hazard regression analysis, RASSF10 methylation and age at 1st operation were independent prognostic factors for OS (P ¼ 0.01952 and 0.02404, respectively) and PFS (P ¼ 0.02724 and 0.03223), whereas IDH1 mutations were independent prognostic factor for OS (P ¼ 0.02699) but did not quite reach significance for PFS (P ¼ 0.06723). Timing of RASSF10 methylation Matched DNA from earlier grade lesions (diffused astrocytomas, WHO grade II) was available for seven of the RASSF10-methylated WHO grade IV sGBM. Oncogene

RASSF10 methylation was observed in the corresponding earlier grade lesions in six of the seven cases (86%), whereas in one case only, the sGBM demonstrated RASSF10 methylation (Figure 4c). These results suggest that RASSF10 methylation is an early event in gliomagenesis. Of the six pairs that show methylation in both the early grade lesion and the higher grade IV samples, four pairs show a very similar pattern of methylation by COBRA both in the early grade lesion and the high grade sGBM, whereas the remaining two samples, shown in Figure 4c, show slightly different patterns of methylation; however, in both cases the overall level of undigested sample is greater in the early grade lesion compared with the higher grade sGBM. RASSF10 subcellular localization is cell-cycle dependent As yet there is no commercially available RASSF10 antibody, hence, we generated a RASSF10-specific rabbit polyclonal antibody raised against a short nonhomologous peptide sequence from the RASSF10 protein. Characterization of the antibody is described in Supplementary Figure S3. We demonstrate that RASSF10 is localized predominantly to the cytoplasm, with some nuclear staining, in several tumor cell contexts (Figure 5a). Cell lines (negative for RASSF10 methylation and hence expressing RASSF10) derived from lung (A549), glioma (U87) and breast (MCF-7) cancers all show RASSF10 protein expression with staining patterns ranging from diffused cytoplasmic to specific foci patterning or perinuclear distribution depending upon cell context.


983 RASSF10

-Tubulin

Nuclear

Merge

Interphase

RASSF10

Nuclear

Prophase

Merge

A549 Prometaphase

U87

Metaphase

Anaphase MCF-7

Telophase

Cytokinesis

A549 Figure 5 RASSF10 protein spatial and temporal distribution. (a) Endogenous RASSF10 expression and localization in cell lines derived from lung, glioma and breast tumor. Cells were fixed, stained with anti-RASSF10 antibody and nuclear stain (DAPI), and processed by immunofluorescence microscopy. In all three cell contexts, endogenous RASSF10 is distributed predominantly in the cytoplasm often with intense localization in the perinuclear region. (b) A549 cells analyzed for RASSF10 cellular distribution during mitosis. Nuclear RASSF10 was observed at early and late stages of the mitotic cycle associating with spindle poles during metaphase/ anaphase before relocating back to the cytoplasm during late telophase/cytokinesis. Images shown are representative of at least three separate experiments.

However, although nuclear exclusion of RASSF10 was a general feature during interphase, we observed RASSF10 relocation to the nucleus specifically in mitotic cells (Figure 5b). In A549 cells, RASSF10 was colocalized with a-tubulin at developing centrosomes during prophase and displayed persistent localization with centrosomally radiating microtubule bundles until late telophase. Strong nuclear staining was not observed during cytokinesis. RASSF10 associated at spindle poles particularly during metaphase and anaphase before relocating back to the cytoplasm. We observed a similar pattern in T47D breast tumor cells (Supplementary Figure S4). Role of RASSF10 in cell viability, cell proliferation and anchorage-independent growth Following on from the observed mitosis-specific redistribution of RASSF10 into the nucleus, we assessed the effect of RASSF10 depletion on cell viability and

proliferation. Two different RASSF10-siRNA oligonucleotides (see Materials and methods) were used to transiently knockdown RASSF10 in U87 glioma cells (unmethylated for RASSF10 and express endogenous RASSF10) to assess cell growth parameters in comparison to non-silencing control small interfering RNA (siRNA). RASSF10 expression levels decreased on average by 70% for the two RASSF10-siRNA oligonucleotides (Figure 6a). We also tested the effect of transient transfection of targeted siRNA on RASSF10 cellular expression and distribution and observed a significant decrease in the cytosolic staining patterns described (Figure 6b). A small level of staining persisted consistent with low-level residual expression and/or non-specific background staining. We next examined the effect of RASSF10 depletion on cell viability over the course of 7 days (Figure 6c). Equal numbers of cells were reverse transfected with either RASSF10-RNAi or NSC-RNAi and cell viability, Oncogene


984 RASSF10

RASSF10

Merge

NSC RNAi

-tubulin

Hoescht

NSC RASSF10 RASSF10 RNAi RNAi1 RNAi 3

RASSF10 RNAi1

80 60 40

RASSF10 RNAi3

RASSF10 / tubulin

100

20 0

U87 cells

NSC RNAi RASSF10 RASSF10 RNAi 1 RNAi 3

U87 40 80000

70000 ***

65000 60000

NSC RNAi

***

RASSF10 RNAi 1 55000 RASSF10 RNAi 3 50000

% EDU incorporation

Fluorescence (560/590nm)

35

***

75000

30

*** **

25 20 15 10 5

45000

0

40000 3 days

5 days Time

7 days

NSC RASSF10 RASSF10 RNAi RNAi 1 RNAi 3

Figure 6 RASSF10 role in cellular growth. (a) NSC-RNAi and RASSF10-RNAi-transfected U87 glioma cells were analyzed for RASSF10 protein expression by immunoblotting with anti-RASSF10 polyclonal antibody (upper panel). Two different RASSF10RNAi oligonucleotides were used separately. a-Tubulin expression was assessed as a loading control. Densitometric analysis of RASSF10 and a-tubulin expression bands showed a significant reduction in RASSF10/a-tubulin protein ratios in RASSF10-RNAi cells compared with NSC-RNAi cells (lower panel). (b) U87 cells transfected with NSC-RNAi or RASSF10-RNAi oligonucleotides were cultured on coverslips for 5 days, fixed and stained with anti-RASSF10 antibody and Hoechst nuclear stain (Sigma Aldrich, St Louis, MO, USA). Anti-RASSF10 antibody staining is markedly reduced in U87-RASSF10 cells compared with staining patterns in U87-NSC cells. (c) Equal numbers of U87 cells were reverse transfected with RASSF10-RNAi or NSC-RNAi oligonucleotides and analyzed for cell viability after 3, 5 and 7 days using CellTiter-Blue Cell Viability Assay (Promega, Madison, WI, USA). The trend shown is representative of three separate experiments. Both RASSF10-RNAi oligonucleotides resulted in increased U87 cell viability. ***Pp0.001. (d) Incorporation of 5-ethynyl-20 -deoxyuridine (EdU) as a measure of cell proliferation and DNA synthesis. %EdU incorporation by U87 cells transfected with RASSF10-RNAi is significantly increased compared with control NSC-RNAi. ***Pp0.001, **Pp0.01.

as measured by the cells’ ability to metabolize the chemical resazurin into a fluorescent compound, assessed after 3, 5 and 7 days. At each time point, RASSF10 knockdown caused a significant increase in fluorescence, indicating that loss of RASSF10 leads to an increase in the viable cell population compared with NSC control cells (Figure 6c; Po0.001 in all cases). This preliminary viability data suggested that RASSF10-depleted cells proliferate at an increased rate. We assayed 5-ethynyl20 -deoxyuridine incorporation as an index of DNA synthesis in dividing U87 glioma cells. RASSF10depleted cells exhibited significantly more 5-ethynyl20 -deoxyuridine incorporation compared with NSC control Oncogene

cells (Figure 6d; Po0.001 and Po0.01 for RASSF10-siRNA oligonucleotides 1 and 3, respectively). Thus RASSF10depleted cells proliferate at an enhanced rate confirming a putative role for RASSF10 in tumor cell growth. To elucidate the potential tumor suppressor function of RASSF10 in glioma cancer cells, we transfected H4 and U343 glioma cells (both cell lines are methylated for RASSF10 and do not express endogenous RASSF10) with either pCMV-Tag1 parental vector or pCMVTag1-RASSF10. Cells were selected in G418 for 3 weeks and surviving colonies were stained with crystal violet before counting. For both cell lines, RASSF10-transfected cells showed lower numbers of colonies compared with


985 U343 140

140

250

P=0.0174

P=0.0098

120

*** Mean colonies > 100m

120

100

100 % control

% control

U87

H4

80 60

80 60

40

40

20

20

Control

RASSF10

**

150 100 50 0

0

0

200

Control

NSC RNAi RASSF10 RASSF10 RNAi 1 RNAi 3

RASSF10

NSC RNAi

RASSF10 RNAi 1

RASSF10 RNAi 3

Figure 7 RASSF10 acts as a tumor suppressor. (a) U343 and H4 cells were transfected with pCMV-Tag1 parental vector or pCMVTag1-RASSF10 for 2 days, replated at several dilutions and cultured in Dulbecco0 s modied Eagle0 s medium and G418 for 3 weeks to assess the effect of RASSF10 reexpression on colony forming ability. In both cell lines, RASSF10 reexpression caused a decrease in the number of cell colonies. Mean colony counts ( þ s.e.m) from triplicate plates are displayed as percentage of parental control. P-values for U343 (highly significant) and H4 (significant) cells are shown. (b) U87 RASSF10-RNAi- and NSC-RNAi-transfected cells were cultured in soft agar for 4 weeks to assess anchorage-independent cell growth. The total number of colonies 4100 mm across were counted. Pp0.01 for RNAi 1 and Pp0.001 for RNAi 3. ***Pp0.001, Pp0.01.

those transfected with the parental vector only (Figure 7a). Growth inhibition was statistically significant over three independent experiments (U343 cells; P ¼ 0.0098, H4 cells; P ¼ 0.0174). To assess anchorageindependence growth, we used transient knockdown of RASSF10 in U87 cells followed by growth in soft agar. RASSF10-depleted cells demonstrated an enhanced ability to grow in soft agar compared with the control cells (Po0.01 for RNAi 1; Po0.001 for RNAi 3) (Figure 7b), demonstrating that RASSF10 can inhibit cell growth in an anchorage- independent manner. This is the first report demonstrating that RASSF10 can suppress growth of cancer cells in vitro. We also assessed the ability of cells to migrate using an in vitro scratch wound healing assay. We tested the effect of RASSF10 overexpression on U343 and H4 cells but observed no difference in migratory ability into the wound between RASSF10-expressing cells and control cells out to 5 days (results not shown). At this point, all wounds were closed. We infer from this that RASSF10 has no effect on cell migration when reexpressed in glioma cells.

Discussion Glioblastomas represent the most frequent and malignant type of brain tumors in adults. These are highly invasive and aggressively growing tumors and prognosis for patients is correspondingly poor. Despite current treatments comprising of surgery, radiotherapy and chemotherapy, the median survival time after diagnosis

is still in the range of just 12 months. Glioblastomas may develop rapidly after a short clinical history and without evidence of a less malignant precursor lesion (primary or de novo glioblastomas) or they may develop slowly through progression from lower grade diffused or anaplastic astrocytomas (sGBM). pGBM and sGBM represent two clinically distinct entities as well as showing differences in their genetic makeup. pGBM show frequent loss of heterozygosity on 10q, amplification of EGFR, deletions of p16 and mutations of the PTEN gene. sGBM and their lower grade precursor lesions have frequent mutations of the TP53 gene and the recently identified mutations in the IDH1 and IDH2 genes. Loss of 10q25-qter is found in both pGBM and sGBM (Hesson et al., 2008; Parsons et al., 2008; Yan et al., 2009a b). In addition, there have been several large scale, genome-wide methylation studies to define the glioblastoma epigenome and develop biomarkers that may predict patient prognosis and response to therapy (Martinez et al., 2009; Noushmehr et al., 2010; Wu et al., 2010). MGMT promoter methylation is by far the most exiting epigenetic marker in gliomas. MGMT encodes for an O-6-methylguanine methyltransferase that removes alkyl groups from the O-6 position of guanine. Many studies have shown that MGMTpromoter methylation is associated with longer survival of patients treated with alkylating agents such as temozolomide and hence is of great clinical significance (Esteller et al., 2000; Weller et al., 2010). More recently the TCGA research network identified a CpG island methylator phenotype in a distinct subset of human gliomas with specific clinical features (Noushmehr et al., 2010). Hence, identification of epigenetically inactivated Oncogene


986

genes may provide further markers that can be utilized in improving cancer prevention, detection and therapy. RASSF10 is the latest member of the N-terminal RASSF (RASSF7–10) family of genes. We recently cloned and characterized the genomic structure of the RASSF10 gene and demonstrated frequent hypermethylation of the associated 50 -CpG island in leukemia (Hesson et al., 2009). Meanwhile, Schagdarsurengin et al., 2009. have demonstrated frequent RASSF10 methylation in thyroid cancer. Thus far there are no reports on the biological properties of the RASSF10 protein product. In the present report, we determined the expression pattern of RASSF10 in Xenopus embryos. RASSF10 is expressed in several tissues in the Xenopus tadpole but showed a particularly striking pattern of expression in the brain. Interestingly, a potential Drosophila homolog (CG32150) (Sherwood et al., 2009) is expressed in the peripheral nervous system (Reeves and Posakony, 2005). Hence to begin with, we looked in brain tumors (gliomas) for epigenetic inactivation of the RASSF10 gene. In this report, we describe for the first time epigenetic inactivation of RASSF10 in human brain tumors. RASSF10 was frequently methylated in grade II–III gliomas as well as in sGBMs and pGBMs, but was unmethylated in grade I gliomas (and in DNA from non-neoplastic brain samples). This would indicate that WHO grade I pilocytic astrocytomas, which rarely undergo malignant transformation arise through a different mechanism. In sGBM, RASSF10 methylation was associated with worst PFS and OS. We demonstrated that RASSF10 methylation was an early event in the development of sGBM. Methylation of RASSF1A and MGMT was detected in 84 and 26%, respectively, of gliomas. As has been demonstrated previously (Yan et al., 2009b), IDH1 mutations were detected frequently in grade II–III gliomas and sGBM, but occurred far less frequently in pGBM. More importantly, RASSF10 methylation was an independent prognostic marker for survival in sGBM. In order to begin to understand the biological properties of the protein product of the RASSF10 gene, we generated and characterized a RASSF10-specific antibody. The endogenous protein localizes to the cytoplasm during interphase but relocates into the nucleus at the beginning of mitosis. Interestingly, RASSF10 protein has putative nuclear localization and export sequences (Supplementary Figure S3), and thus has the potential to ‘shuttle’ between cellular compartments. In the nucleus, RASSF10 colocalizes with developing spindle poles and with microtubules during prophase through late telophase before relocating back to the cytoplasm. This specific temporal and spatial distribution suggests that RASSF10 has a regulatory function within the mitotic cycle but protein interactions confirming such a role remain to be discovered. Overexpression studies in glioma cell lines that do not express endogenous RASSF10 (due to RASSF10 methylation), demonstrate that cells transfected with RASSF10 suppress colony formation ability as compared with vector-transfected cells. The reverse happens Oncogene

in soft agar experiments, wherein we depleted endogenous RASSF10 in glioma cells and compared it with control siRNA cells. The RASSF10-depleted cells demonstrate a marked ability in their anchorageindependent growth potential compared with control cells. Cell viability assays and 5-ethynyl-20 -deoxyuridine incorporation experiments further confirmed the growth inhibitory effects of RASSF10 on cell proliferation. In summary, we have demonstrated that the RASSF10 gene is frequently methylated in WHO grade II–IV astrocytic gliomas. In sGBM, RASSF10 methylation is associated with worst survival, independent of the IDH1 mutation status and occurs at an early stage in their development. We have demonstrated that RASSF10 suppresses growth and proliferation of glioma cells in vitro and is likely a cell-cycledependent protein involved in mitotic regulation. Our results implicate RASSF10 as a candidate tumor suppressor gene involved in gliomagenesis.

Materials and methods Cell line and primary tumor DNA samples A cell line panel of 5 glioma lines and 63 astrocytic gliomas (20 pGBM WHO grade IV (pGBM), 13 sGBM WHO grade IV (sGBM), 10 anaplastic astrocytomas WHO grade III, 10 diffused astrocytomas WHO grade II and 10 pilocytic astrocytomas WHO grade I; 28 males and 35 females) and 7 matched WHO grade II astrocytomas corresponding to the sGBM and 15 non-cancerous adult brain tissue samples were used (see Supplementary Table S2-S3 for patient characteristics). Tumor tissues and tissues from epileptic surgery procedures were taken intraoperatively and were snap frozen at 80 1C. All histopathological diagnosis were performed by the Institute of Neuropatholgy at the University Hospital Carl Gustav Carus, Dresden, Germany. Frozen tumor tissue was sectioned using a cryotome (Cryostat Jung CM 1800, Leica, Weztlar, Germany). To assure a tumor cell content of at least 80% for nucleic acid extraction, control slides were stained with hematoxylin and eosin, and were examined by two independent reviewers. For DNA isolation, the QIAmp DNA Mini Kit 50 (Qiagen, Hilden, Germany) was used according to the manufacturer’s instructions. Ethical guidelines were followed for sample collection. Methylation analysis All DNA samples were bisulfite modified using Qiagen EpiTect kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. RASSF10 methylation status was assessed using COBRA on the same region assessed previously (Hesson et al, 2009), with some modifications to primer sequences; forward: 50 -TTGTTTTTGTTGTTTTYGTYGTTTTAGTA GATT-30 , forward nested: 50 -GTGTGGATTTGTTAGGAA GAGAAGT-30 , reverse nested: 50 -CTATTTCTCCTAAAT CATAACCAAACTAA-30 , reverse: 50 -CRATTAAACTTAAC CAATTTACRAAAAACCTTA-30 . Touchdown PCR was used with final annealing temperature 55 1C for primary reaction and standard PCR with annealing temperature 56 1C for secondary reaction. PCR products were digested with BstUI (Fermentas, Burlington, Canada) (CGCG) overnight at 37 1C before visualization on a 2% agarose gel. Clone sequencing of PCR products was used to assess internal levels of methylation for individual samples. PCR products of two samples for each glioma grade and non-cancerous brain tissue


987 were cloned into the pGEM-T easy vector (Promega) according to the manufacturer’s instructions. Up to 12 clones were picked and subsequently sequenced, following single colony PCR. Methylation indexes were calculated as a percentage; number of CpG dinucleotides methylated/number of CpG dinucleotides sequenced 100. Methylation status of MGMT and RASSF1A was assessed using methylationspecific PCR (MSP). Previously described MSP primers were used for both MGMT (Esteller et al., 2000) and RASSF1A (Honorio et al., 2003). Expression analysis Glioma cell lines H4, A172, T17 and U343 were treated with 5 mM 5-azaDC to achieve genome demethylation. Treatment was carried out over 5 days with daily media changes and addition of fresh 5-azaDC. Cell lines were maintained in DMEM media (Sigma Aldrich) supplemented with 2 mM glutamine and 10% FCS at 37 1C, 5% CO2. Total RNA was extracted from 5-azaDC treated and non-treated cells using RNA bee (AMS biotechnology, Abingdon, UK). Superscript III (Invitrogen, Carlsbad, CA, USA) and random hexamer primers (Fermentas, Burlington, Canada) were used to synthesize complementary DNA from 1 mg total RNA. As RASSF10 is a single exon gene, all RNA samples were treated with DNase I (Fermentas) before complementary DNA synthesis to remove any DNA contamination. Expression of RASSF10 was assessed using the following primers; forward: 50 -GTTCAGCAGGAGGAGTTGCT-30 and reverse: 50 -CGT CAGCTCCAAAGTGTGCAA-30 . A touchdown PCR was used with a final annealing temperature of 56 1C, total 40 cycles. GAPDH expression was carried out concurrently using the following primers; forward: 50 -TGAAGGTCGGAGT CAACGGATTTGGT-30 , reverse: 50 -CATGTGGGCCATGA GGTCCACCAC-30 and the same PCR conditions, with the exception of cycle number (28 instead of 40). IDH1 and IDH2 mutation analysis For the amplification of a 129 bp fragment of the IDH1 gene and a 150 bp fragment of the IDH2 gene, spanning the respective sequence of the catalytic domains and codon 132 of IDH1 and codon 172 of IDH2, respectively, primers were used as previously published (Hartmann et al., 2009). IDH1-forward: 50 -CTCCTGATGAGAAGAGGGTTG-30 and IDH1-reverse: 50 -TGGAAATTTCTGGGCCATG-30 , IDH2forward: 50 -TGGAACTATCCGGAACATCC-30 and IDH2reverse: 50 -AGTCTGTGGCCTTGTACTGC-30 . PCR conditions were for IDH1 and IDH2: 20 ng of DNA and Go Taq polymerase with standard buffer conditions employed 35 cycles with denaturing at 95 1C for 50 s, annealing at 58 1C for 50 s and extension at 72 1C for 50 s in total volume of 25 ml. Bidirectional cycle sequencing of PCR-amplified fragments was performed using an ABI 3730x (Applied Biosystems, Foster City, CA, USA) semiautomated sequencer. Cell culture Cell culture medium and reagents were purchased from Invitrogen. Cell lines (A172, H4, U343, U87, T17, A549, MCF-7, T47-D) were maintained at 37 1C, 5% CO2 in Dulbecco0 s modied Eagle0 s medium (DMEM), with 10% FCS, 1% Penicillin/streptomycin, 1% glutamine and 1% sodium pyruvate added. Protein expression and antibodies Protein extracts were prepared by lysis in 50 mM Tris–HCl pH 7.5, 1 mM EDTA, 1 mM ethylene glycol tetraacetic acid (EGTA), 50 mM sodium fluoride, 5 mM sodium pyrophosphate,

1 mM sodium orthovanadate, 0.27 M sucrose, 1% triton and protease inhibitor cocktail (Roche, Basel, Switzerland). A total of 20 mg of protein extracts were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene fluoride (PVDF) membrane and probed with appropriate antibodies: anti-RASSF10 rabbit polyclonal (raised against RASSF10 peptide sequence, Eurogentec (Lie`ge, Belgium), Supplementary Figure S3), anti-Tubulin and anti-FLAG (Sigma Aldrich), anti-HA (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-rabbit-HRP (Cell Signalling Technology, Danvers, MA, USA), anti-mouse-HRP (Dako, Copenhagen, Denmark), anti-rabbit and anti-mouse AlexaFluor (594, 488) secondary IgG (Invitrogen, Molecular Probes, Carlsbad, CA, USA). RASSF10 constructs pCMV-Tag1-RASSF10 expressing the short 507 amino-acid RASSF10 protein (A6NK89) was a kind gift from Dr Reinhard Dammann. To generate pCMV-HA-RASSF10, we amplified the RASSF10 gene (GenBank: NM_001080521) from IMAGE clone 4413366 (Source Bioscience Geneservice, Nottingham, UK) with primers; 50 -ATGGACCCTTCGGAAAAGAAG-30 and 50 -CTACACAAGGGATTCGCACAT-30 containing additional overhang restriction sites for EcoR1 and BamH1, respectively. The 1524 bp RASSF10 gene product was cloned into pCMV-HA vector to generate the short 507 amino-acid RASSF10 protein (A6NK89) coupled to a hemagglutinin (HA) epitope tag at its N terminus. RASSF10 Knockdown Predesigned RASSF10-siRNA oligonucleotides directed against RASSF10 were purchased from Qiagen (SI02814048: CACCGT CGTCCTGTTCGTCCA, SI04229617: CTGAGCTCTATGCA TAGCCAA) or a nonsense siRNA sequence directed against luciferase (non-silencing control). Cells were transfected with siRNA using Fugene-HD transfection reagent (Roche) as per the manufacturer’s instructions. Immunocytochemistry Cells were cultured on acid-washed glass coverslips, fixed with 4% paraformaldehyde in phosphate buffered saline and stained with primary antibodies and appropriate secondary antibodies as specified in the text. Nuclei were stained with 40 ,6-diamidino-2-phenylindole. For immunofluoresence microscopy, cells were visualized using a Zeiss Axioplan microscope (Zeiss, Germany) equipped with camera (Photometrics Sensys, Tucson, AZ, USA) and images were captured and processed using SmartCapture2 software (Digital Scientific, Cambridge, UK). Viability assay Cells were plated into 96-well plates and transfected with siRNA against RASSF10 or luciferase (NSC). After 3, 5 and 7 days, cell viability was measured using CellTiter-Blue reagent (Promega) as per the manufacturer’s instructions. Fluorescence ratio (A560/A590) was obtained on a Victor-X3 plate reader (Perkin Elmer, Waltham, MA, USA) to give a measure of cell metabolic activity and thus proportional cell number/ viability. 5-ethynyl-20 -deoxyuridine incorporation assay The EdU Click-it system (Invitrogen, Molecular Probes) was used as per the manufacturer’s instructions to analyze proliferation as a measure of the percentage of cells going through S phase, under different experimental conditions, as specified Oncogene


988 in the text. The mean and s.e.m. of three separate experiments, performed in triplicate, are shown. At least 500 cells per experiment were counted, repeated 3 times. Soft agar growth assays For transient knockdown of RASSF10, U87 glioma cancer cells were plated at 5 104 cells per well of a 6-well plate and treated with siRNA oligo directed against luciferase or against RASSF10 using Fugene-HD (Roche). After 24 h, cells were trypsinized and plated at 5 104 cells per well in soft agar in a 6-well plate. Cells were cultured at 37 1C for 2 weeks, colonies were photographed and the average number of colonies greater than 100 mm was calculated and expressed as fold change compared with luciferase siRNA-treated controls. Colony forming assay H4 and U343 cells were transfected (Interferin, Polyplus Transfection, New York, NY, USA) with either parental pCMV-Tag1 vector or pCMV-Tag1-RASSF10 and incubated at 37 1C for 48 h. Cells were then replated in G418-DMEM media and incubated at 37 1C for 3 weeks. Colonies were stained with 0.4% crystal violet solution and counted. The mean and s.e.m. of three separate experiments, performed in triplicate, are shown.

Xenopus in situ hybridization A Xenopus laevis rassf10 pBluescript SK() plasmid, clone Xl086p24, was previously identified, obtained from the XDB website (a kind gift from the NIBB/NIG/NBRP Xenopus laevis EST project) and sequenced (Sherwood et al., 2008). This plasmid was used to generate a dig-labeled in situ probe, embryos were obtained and in situ carried out and sectioned as described (Chalmers et al., 2002). Statistical analysis Statistical analysis was performed as indicated, Po0.05 was considered significant.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements We thank Dr Reinhard Dammann for providing the RASSF10 construct. F Latif laboratory funded in part by Breast Cancer Campaign, CRUK and Sport Aiding Medical Research for Kids (SPARKS).

References Chalmers AD, Welchman D, Papalopulu N. (2002). Intrinsic differences between the superficial and deep layers of the Xenopus ectoderm control primary neuronal differentiation. Dev Cell 2: 171–182. Chen L, Johnson RC, Milgram SL. (1998). P-CIP1, a novel protein that interacts with the cytosolic domain of peptidylglycine alphaamidating monooxygenase, is associated with endosomes. J Biol Chem 273: 33524–33532. Dallol A, Hesson LB, Matallanas D, Cooper WN, O’Neill E, Maher ER et al. (2009). RAN GTPase is a RASSF1A effector involved in controlling microtubule organization. Curr Biol 19: 1227–1232. Donninger H, Vos MD, Clark GJ. (2007). The RASSF1A tumor suppressor. J Cell Sci 120: 3163–3172. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V et al. (2000). Inactivation of the DNA repair gene MGMT and the clinical response of gliomas to alkylating agents. 343: 1350–1354. Hartmann C, Meyer J, Balss J, Capper D, Mueller W, Christians A et al. (2009). Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1010 diffuse gliomas. Acta Neuropathol 118: 469–474. Hesson LB, Cooper WN, Latif F. (2007a). The role of RASSF1A methylation in cancer. Dis Markers 23: 73–87. Hesson LB, Cooper WN, Latif F. (2007b). Evaluation of the 3p21.3 tumour-suppressor gene cluster. Oncogene 26: 7283–7301. Hesson LB, Dunwell TL, Cooper WN, Catchpoole D, Brini AT, Chiaramonte R et al. (2009). The novel RASSF6 and RASSF10 candidate tumour suppressor genes are frequently epigenetically inactivated in childhood leukaemias. Mol Cancer 8: 42. Hesson LB, Krex D, Latif F. (2008). Epigenetic markers in human gliomas: prospects for therapeutic intervention. Expert Rev Neurother 8: 1475–1496. Honorio S, Agathanggelou A, Wernert N, Rothe M, Maher ER, Latif F. (2003). Frequent epigenetic inactivation of the RASSF1A tumour suppressor gene in testicular tumours and distinct methylation Oncogene

profiles of seminoma and nonseminoma testicular germ cell tumours. Oncogene 22: 461–466. Langton PF, Colombani J, Chan EH, Wepf A, Gstaiger M, Tapon N. (2009). The dASPP-dRASSF8 complex regulates cell-cell adhesion during Drosophila retinal morphogenesis. Curr Biol 19: 1969–1978. Lock FE, Underhill-Day N, Dunwell T, Matallanas D, Cooper W, Hesson L et al. (2010). The RASSF8 candidate tumor suppressor inhibits cell growth and regulates the Wnt and NF-kB signalling pathways. Oncogene 29: 4307–4316. Martinez R, Martin-Subero JI, Rohde V, Kirsch M, Alaminos M, Fernandez AF et al. (2009). A microarray-based DNA methylation study of glioblastoma multiforme. Epigenetics 4: 255–264. Noushmehr H, Weisenberger DJ, Diefes K, Diefes K, Phillips HS, Pujara K et al. (2010). Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 17: 510–522. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P et al. (2008). An integrated genomic analysis of human glioblastoma multiforme. Science 321: 1807–1812. Reeves N, Posakony JW. (2005). Genetic programs activated by proneural proteins in the developing Drosophila PNS. Dev Cell 8: 413–425. Richter AM, Pfeifer GP, Dammann RH. (2009). The RASSF proteins in cancer; from epigenetic silencing to functional characterization. Biochim Biophys Acta 1796: 114–128. Rodriguez-Viciana P, Sabatier C, McCormick F. (2004). Signaling specificity by Ras family GTPases is determined by the full spectrum of effectors they regulate. Mol Cell Biol 24: 4943–4954. Schagdarsurengin U, Richter AM, Wo¨hler C, Dammann RH. (2009). Frequent epigenetic inactivation of RASSF10 in thyroid cancer. Epigenetics 4: 571–576. Sherwood V, Manbodh R, Sheppard C, Chalmers AD. (2008). RASSF7 is a member of a new family of RAS association domain-containing proteins and is required for completing mitosis. Mol Biol Cell 19: 1772–1782.


989 Sherwood V, Recino A, Jeffries A, Ward A, Chalmers AD. (2009). The N-terminal RASSF family: a new group of Ras-association-domaincontaining proteins, with emerging links to cancer formation. Biochem J 425: 303–311. Tommasi S, Dammann R, Zhang Z, Wang Y, Liu L, Tsark WM et al. (2005). Tumor susceptibility of Rassf1a knockout mice. Cancer Res 65: 92–98. van der Weyden L, Tachibana KK, Gonzalez MA, Adams DJ, Ng BL, Petty R et al. (2005). The RASSF1A isoform of RASSF1 promotes microtubule stability and suppresses tumorigenesis. Mol Cell Biol 25: 8356–8367.

Weller M, Stupp R, Reifenberger G, Brandes AA, van den Bent MJ, Wick W et al. (2010). MGMT promoter methylation in malignant gliomas: ready for personalized medicine? Nat Rev Neurol 6: 39–51. Wu X, Rauch TA, Zhong X, Bennett WP, Latif F, Krex D et al. (2010). CpG island hypermethylation in human astrocytomas. Cancer Res 70: 2718–2727. Yan H, Parsons DW, Jin G, McLendon R, Rasheed BA, Yuan W et al. (2009a). IDH1 and IDH2 mutations in gliomas. N Engl J Med 360: 765–773. Yan H, Bigner DD, Velculescu V, Parsons DW. (2009b). Mutant metabolic enzymes are at the origin of gliomas. Cancer Res 69: 9157–9159.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene