Parkin Pathway Activation Mitigates Glioma Cell Proliferation and Predicts Patient Survival

Author Manuscript Published OnlineFirst on March 19, 2012; DOI: 10.1158/0008-5472.CAN-11-3060 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Parkin suppresses glioma cell proliferation

Parkin pathway activation mitigates glioma cell proliferation and predicts patient survival

Calvin W.S. Yeo1,2, Felicia S.L. Ng5, Chou-Chai6, Jeanne M.M. Tan6, Geraldene R.H. Koh3, Yuk Kien Chong1,5, Lynnette W.H. Koh3, Charlene S.F. Foong1,5, Edwin Sandanaraj5, Joanna D. Holbrook5, Beng-Ti Ang1,4,5, Ryosuke Takahashi8, Carol Tang3*, Kah-Leong Lim1,2,6,7*

1Department of Physiology, National University of Singapore, Singapore. 2Neurodegeneration Research Laboratory, 3Neuro-Oncology Research Laboratory, 4Department of Neurosurgery, National Neuroscience Institute, Singapore. 5Singapore Institute for Clinical Sciences, A*STAR, Singapore. 6Duke-NUS Graduate Medical School. 7A*STAR Neuroscience Research Partnership, Singapore. 8Department of Neurology, Graduate School of Medicine, Kyoto University, Kyoto, Japan.

Running title: Parkin suppresses glioma cell proliferation

* To whom all correspondence should be addressed:

Kah-Leong Lim, Ph.D., Department of Physiology, Yong Loo Ling School of Medicine, National University of Singapore, MD9, 2 Medical Drive, Singapore 117597. E-mail: [email protected]

Carol Tang, Ph.D., National Neuroscience Institute, 11 Jalan Tan Tock Seng, Singapore 308433. E-mail: [email protected]

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2012 American Association for Cancer Research. Author Manuscript Published OnlineFirst on March 19, 2012; DOI: 10.1158/0008-5472.CAN-11-3060 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Parkin suppresses glioma cell proliferation

ABSTRACT

Mutations in the parkin gene, which encodes a ubiquitin ligase, are a major

genetic cause of parkinsonism. Interestingly, parkin also plays a role in cancer as a

putative tumor suppressor, and the gene is frequently targeted by deletion and

inactivation in human malignant tumors. Here, we investigated a potential tumor

suppressor role for parkin in gliomas. We found that parkin expression was dramatically

reduced in glioma cells. Restoration of parkin expression promoted G1 phase cell cycle

arrest and mitigated the proliferation rate of glioma cells in vitro and in vivo. Notably,

parkin-expressing glioma cells showed a reduction in levels of cyclin D1, but not cyclin

E, and a selective downregulation of Akt serine-473 phosphorylation and VEGF receptor

levels. In accordance, cells derived from a parkin null mouse model exhibited increased

levels of cyclin D1, VEGF receptor and Akt phosphorylation and divided significantly

faster when compared with wild type cells, with suppression of these changes following

parkin re-introduction. Clinically, analysis of parkin pathway activation was predictive

for the survival outcome of glioma patients. Taken together, our study provides

mechanistic insight into the tumor suppressor function of parkin in brain tumors, and

suggests that measurement of parkin pathway activation may be used clinically as a

prognostic tool in brain tumor patients.

Parkin suppresses glioma cell proliferation

INTRODUCTION

Parkinson’s disease (PD) and cancer are two important but obviously disparate

human disorders. Yet, many of the PD-linked genes identified to date, including -

synuclein, parkin, DJ-1 and PINK1, are also associated with cancer (1). In particular, DJ-

1 was originally isolated as a putative oncogene (2). The overlapping of genes involved in

PD and cancer would imply a shared pathogenic pathway. However, the mechanism

underlying the opposite cellular fates of the two diseases remains unclear, the elucidation

of which could help conceive novel therapeutic approaches to both groups of disorders.

Mutations in the parkin gene, which encodes a ubiquitin ligase, were originally

identified as a genetic contributor of autosomal recessive parkinsonism (3). Accordingly,

much of the interest in characterizing the function of the parkin gene has been directed

towards understanding its role in neurodegeneration. However, in recent years, the role of

parkin in cancer has also gained much attention. It is now clear that parkin gene

alterations are not restricted to familial forms of PD but also occur frequently in a wide

variety of malignancies (4-7), which include glioblastoma (GBM) and lung cancer where

somatic loss-of-function parkin mutations were found (8). Unlike the oncogenic DJ-1,

parkin appears to act as a tumor suppressor. Supporting this, we and others have

demonstrated that restoration of parkin expression in a wide spectrum of parkin-deficient

cancer cells results in a marked decrease in their proliferation rate (8, 9). Further, at least

one line of parkin null mice exhibits a tendency to develop carcinoma (10).

Notably, parkin functions as a ubiquitin ligase (E3) associated with the ubiquitin-

proteasome system, and one of its several putative substrates identified to date is cyclin E

(11), a cell cycle-related (G1) cyclin whose accumulation is associated with cancer

Parkin suppresses glioma cell proliferation

development. It is therefore attractive to suggest that altered cyclin E proteolysis in

parkin-deficient cancer and neuronal cells may underlie their respective susceptibility to

undergo cell division and degeneration. Indeed, cyclin E accumulation is associated with

parkin deficiency in several cancer cell lines as well as in cultured post-mitotic neurons

(8, 11, 12), although this phenomenon is not universally observed (9, 13). Interestingly,

parkin also exhibits an apparent E3-independent function in the control of gene

transcription. Amongst the genes regulated by parkin is TP53, a well-established tumor

suppressor whose expression is repressed by functional parkin (14). This is rather

surprising, as the elevation of p53 levels in parkin-deficient cells should in theory retard

(instead of promote) cell division. Two other genes whose expression is also regulated by

parkin are follistatin (10) and CDK6 (9). Whereas the level of follistatin increases in

parkin-deficient cells, the reverse is true for CDK6 (9, 10). Both events are however

thought to promote carcinogenesis (9, 10). Here, we provide evidence that parkin

dysfunction is relevant to gliomagenesis, and that restoration of functional parkin

expression in glioma cells mitigates their growth via a mechanism that likely involves

parkin-mediated downregulation of the Akt signalling pathway. Importantly, we further

found that parkin pathway activation is predictive of survival prognosis of glioma

patients.

Parkin suppresses glioma cell proliferation

MATERIALS AND METHODS

Antibodies and reagents

The following antibodies were used: monoclonal anti-parkin clone PRK8 (Covance),

anti-phospho-Akt (S473 and T308), anti-Akt, anti-Cyclin D1 and E, anti-VEGFR2, anti-

phospho-PDK1, anti-PDK1, anti-mTOR, anti-Rictor, anti-Sin1, anti-phosphoFoxO3a,

anti-FoxO3a, anti-phosphoGSK3, anti-GSK3 (all from Cell Signaling), anti-Flag

peroxidase (Sigma), monoclonal anti--actin (Sigma), FITC-conjugated anti-mouse IgG

(Jackson ImmunoResearch Laboratories Inc.). All the cell lines used in this study were

purchased from American Type Culture Collection. Primary MEFs from wild type and

parkin null mice (exon 7) (kind gifts from Dr. Dawson T.M., Johns Hopkins Medicine)

were generated according to published protocol (15).

Analysis of parkin expression in glioma cells

Total RNA was isolated from various cancer cells described using the RNAeasy Mini Kit

(Qiagen). Subsequently, the isolated RNA was reverse transcribed using the Superscript

First-Strand Synthesis System (Invitrogen). Real-Time PCR was carried out in a Light

Cycler (Roche) using the FastStart DNA Master Plus Sybergreen I system (Roche)

according to the manufacturer’s protocol. A pair of parkin-specific primers (Forward: 5’-

GGAAGTCCAGCAGGTAGATCA; Reverse: 5’-ACCCTGGGTCAAGGTGAG) was

generated for this purpose. Concurrently, Real-Time PCR with a primer pair specific for

GAPDH (Forward: 5’-GAAGGTGAAGGTCGGAGTCAACG; Reverse: 5’-

TGCCATGGGTGGAATCATATTGG) was also included in the same run as an internal

control. Western Blot analyses were performed as previously described (9).

Parkin suppresses glioma cell proliferation

Generation of stable cell lines, proliferation assays and cell cycle analysis

U87MG cells stably expressing wild type or mutant parkin, or otherwise containing

vector alone were generated by means of a previously described procedure (9). All

positive cell lines used for the experiments described hereafter were maintained in serum-

containing DMEM supplemented with 200μg/ml Geneticin (Invitrogen) to prevent

extrusion of integrated constructs. A simple population growth assay was conducted by

seeding cells to be counted in duplicates at a concentration of 2 x 104 cells in 6-well

plates, and subsequently quantifying their number each day for a period of 5 days by

means of a haemocytometer. BrdU-based proliferation assay (Roche) was carried out

according to the manufacturer’s instructions. For soft-agar colony formation, 0.3% agar

containing 1 x 104 U87-vector or U87-Parkin stable cells was overlaid onto pre-cast 0.5%

o bottom agar and allowed to solidify before incubating at 37 C in a 5.0% CO2 incubator

for a period of 21 days. Colonies formed on the soft agar were visualized under light

microscopy before and after MTT staining. For cell cycle analysis, cells were seeded at a

concentration of 5 x 105cells in 10 ml of culture medium in 10-cm tissue culture plates

o and incubated at 37 C in a 5.0% CO2 incubator. After 24 hours, the culture medium was

aspirated and the cells were rinsed 3 times with serum-free culture medium replaced with

o 10ml of serum-free culture medium for 24 hours at 37 C in a 5.0% CO2 incubator to

synchronize them at G0 resting phase. Cells were stained using the BrdU/7-AAD flow kit

(Pharmingen) and then resuspended in 1ml of PBS containing 5mM of EDTA before

analysis by fluorescence-activated cell sorting (FACS) at 20,000 events using quadrant

plot analysis on a flow cytometer (FACSCalibur) (Becton Dickinson).

Parkin suppresses glioma cell proliferation

In vivo NOD-SCID/J intracranial tumor model

All procedures involving animals were approved by and conformed to the guidelines of

our Institutional Animal Care and Use Committee. Six week old NOD-SCID/J mice were

stereotactically implanted with U87-vector or U87-Parkin cells according to our previous

work (16). Coordinates used: anteroposterior, +2mm; mediolateral, +1.5mm;

dorsoventral, -2.5mm. Mice were sacrificed by transcardiac perfusion upon presentation

of neurological deficits.

Parkin suppresses glioma cell proliferation

RESULTS

Parkin deficiency is a characteristic of glioma cells and restoration of its expression

mitigates glioma growth both in vitro and in vivo.

In view of the association between parkin gene alteration and cancer, we were

interested to examine whether the expression of parkin is compromised in glioma cells.

For this purpose, we analyzed the levels of parkin transcripts quantitatively by means of

real-time PCR in a spectrum of glioma cells that included U87MG, U373MG, U251MG

and T98G. As controls, we also examined parkin expression in HEK293 cells, a non

tumor-derived cell line, and MCF7, a parkin-deficient breast cancer cell line (4, 9). Our

analysis revealed an apparent reduction in the level of parkin mRNA in all the glioma cell

lines examined, which decreases dramatically by an average of 16-fold relative to

HEK293 cells (Fig. S1A). Notably, this reduced level of parkin mRNA expression is

comparable to that in MCF7 cells (Fig. S1A), and correlates well with the amount of

parkin protein in extracts derived from the various glioma cell lines examined (Fig. S1B).

To examine the effects of restoring parkin expression in parkin-deficient glioma cells, we

generated clonal populations of U87MG glioma cells stably expressing either FLAG-

tagged parkin (U87-Parkin) or containing vector alone as a control (U87-Vector). We

have chosen U87MG cells for this purpose because it is a popular cell model for glioma

studies and is also one that is highly deficient in parkin expression (Fig. S1A). All the

parkin-positive clones express parkin at a significantly higher level than vector control or

parental cells (Fig. 1A), but are otherwise similar morphologically (not shown). We then

measured the proliferation rate of U87-Parkin and U87-Vector by means of a variety of

assays. A simple population growth assay revealed that the proliferation rate of parkin-

Parkin suppresses glioma cell proliferation

expressing U87MG cells is significantly reduced compared to control cells (Fig. 1A),

suggesting that ectopic parkin expression in these cells mitigates their growth. Supporting

this, the incorporation of BrDU, a thymidine analogue, in parkin-expressing U87MG

cells also occurs significantly less frequently compared to control cells (Fig. 1C). Similar

observation was made with the MTT-based proliferation assay (not shown). Since cancer

cells are anchorage-independent and have the ability to form colonies in soft agar, we

also examined whether ectopic parkin expression in U87MG cells compromise their

ability to generate colonies in soft agar. We found that the number of soft agar colonies

formed by parkin-expressing U87MG cells is dramatically reduced and the size of these

colonies also tend to be smaller compared to those generated by control U87MG cells

(Fig. 1D, not shown for U87-Parkin #19 and #25). Thus, elevated expression of parkin in

U87MG cells apparently could downregulate their rate of proliferation. This effect of

parkin is however dependent on its catalytic activity, as the stable expression of a

catalytically-null parkin mutant (i.e. T415N) in U87MG cells failed to exert any negative

influence on their proliferation rate (Fig. 1B).

Alongside our in vitro assays, we also examined the effects of parkin over-

expression on the ability of U87MG cells to generate solid tumor in vivo. An intracranial

glioma model (n = 5 for each group), which involved the injection of parkin-expressing

(clone #7) or control U87MG cells into the cerebral cortex of NOD-SCID mice, was used

for this purpose. At four weeks post-injection, we observed the appearance of

macroscopic tumors in four out of five dissected brains of mice injected with control

U87MG cells, but none in those injected with U87-Parkin cells (Fig. 1E) or vehicle alone

(not shown). Further, histological examination of brain sections derived from these mice

Parkin suppresses glioma cell proliferation

revealed that NOD-SCID mice injected with U87-vector control cells via the intracranial

route developed visibly larger tumors in the brain at the end of the experimental period

compared to those injected with U87-Parkin cells (Fig. 1E). Consistent with this, the

average weight of brains harvested from all five mice injected with parkin-expressing

U87MG cells is significantly reduced compared to those harvested from vector control

mice (Fig. 1F). Taken together, our results strongly support a negative role for parkin in

glioma cell proliferation.

As an extension of our in vivo study, we repeated the above experiment but

allowed injected mice from different groups (i.e. U87-vector, U87-Parkin #7 and #19; n =

10 for each group) to live beyond four weeks to examine their respective survival profile.

We found that whereas NOD-SCID mice harbouring glioma generated from U87-vector

control cells readily succumb to the tumor, those injected with U87-Parkin #7 exhibit

significantly better survival (Fig. S2A). Curiously, the U87-Parkin #19 group showed a

mortality curve that is between vector and U87-Parkin #7 groups, but with a profile

tending towards and not significantly different from the vector group (Fig. S2A). At post-

mortem analysis, we found that tumor derived from U87-Parkin #19 had lost much of its

parkin expression over time, which likely accounted for its survival profile, whereas

those derived from U87-Parkin #7 exhibit robust parkin expression even after prolonged

periods in vivo (Fig. S2B). Thus, parkin expression appears to correlate inversely with

cancer mortality.

Parkin suppresses glioma cell proliferation

Parkin-expressing glioma cells exhibit delayed entry into mitosis and reduced cyclin

D1 levels.

Conceivably, the reduced proliferation rate of parkin-expressing U87MG cells

may be due to alterations of the cell cycle program. To investigate this, we analyzed the

cell cycle profile of control U87-Vector and U87-Parkin cells via flow cytometry. In

asynchronous cells, we observed a tendency for parkin-expressing U87MG cells to

accumulate at the G1 phase relative to control cells (Fig. S3A), suggesting that parkin

expression may delay the entry of cells into the mitotic phase. In agreement with this, we

recorded a corresponding reduction in the percentage of parkin-expressing cells at the S

and G2/M phases (Fig. S3A). This trend is similarly observed in synchronized U87-

Parkin cells following serum release, but not in those that express parkin T415N mutant

(which tend to show a reverse trend) (Fig. 2A). We next examined if enhanced cyclin E

proteolysis may account for the cell cycle profile in U87-Parkin cells. Surprisingly, we

did not detect an apparent difference in the steady state level of cyclin E between parkin-

expressing and control U87MG cells (Fig. 2B). Instead, we found that the expression of

cyclin D1 (which, like cyclin E, is also involved in G1 to S transition), is dramatically

repressed in parkin-expressing U87MG cells (Fig. 2B). In contrast, cyclin D1 levels

remain largely unaltered in U87MG cells stably expressing the T415N mutant (Fig. 2C).

Together, our results suggest that defective cyclin D1 expression, rather than cyclin E,

possibly underlies parkin’s ability to influence the cell cycle program of glioma cells in

this case.

Parkin suppresses glioma cell proliferation

Akt Ser-473 phosphorylation and VEGFR-2 expression are significantly reduced in

parkin-expressing glioma cells.

Notably, a significant percentage of high-grade gliomas harbour mutations in

PTEN, a tumor suppressor whose loss of function is known to promote abnormal cellular

growth via over-activation of the PI3 kinase/Akt signaling pathway (17). Consistent with

this, Pten-deficient U87MG, U373MG and U251MG cells exhibit elevated levels of

phosphorylated Akt compared to other PTEN-expressing glioma cells (not shown). To

investigate whether exogenous parkin expression in U87MG cells exerts any effect on

Akt signaling, we immunoblotted lysates prepared from U87-Parkin and U87-Vector

cells with phosphorylation-specific Akt antibodies and found that the levels of Akt

phosphorylated at Ser-473 is significantly reduced in parkin-expressing U87MG cells

compared to control cells (Fig. 3A). This phenomenon is not due to changes in Akt

expression in the presence of parkin (Fig. 3A & B), and is not observed when we

repeated the experiment with U87MG cells stably expressing parkin T415N mutant (Fig.

3A), suggesting that the catalytic activity of parkin is required for its effect on Akt Ser-

473 phosphorylation. In contrast, phosphorylation of Akt in U87MG cells at another site,

Thr-308, appears to be unaffected by parkin overexpression (Fig. 3B). Consistent with

this, the levels of activated phosphoinositide-dependent kinase (PDK) 1, an upstream

effector of Akt Thr-308 phosphorylation, is not appreciably altered in wild type or mutant

parkin-expressing U87MG cells (Fig. 3C and S3B). Instead, the levels of TORC2

components (thought to be responsible for Akt Ser-473 phosphorylation) including

mTOR, Rictor and Sin1 as well as their associated substrates such as FoxO3a and GSK3

all appear to be reduced in U87-Parkin cells (Fig. 3D), suggesting that parkin-mediated

Parkin suppresses glioma cell proliferation

downregulation of Akt Ser-473 phosphorylation may be due to its ability (directly or

indirectly) to influence the expression of TORC2 complex. This reduction of TORC2-

associated components was not seen in parkin T415N mutant-expressing U87MG cells

(Fig. S3B).

To extend these findings, we also looked at the extent of Akt Ser-473

phosphorylation in the presence or absence of growth factor stimulation. Under serum-

starved condition, the basal level of Ser-473 phosphorylated Akt in parkin-expressing

U87MG cells is similarly reduced compared to control cells (Fig. 3E). Upon EGF

stimulation, a marked increase in Akt phosphorylation is observed in both control and

parkin-expressing U87MG cells, although the steady-state levels of Ser-473

phosphorylated Akt in U87-Parkin cells remains low relative to U87-Vector control cells

(Fig. 3D). Collectively, these results suggest that parkin-mediated suppression of glioma

cell growth is in part contributed by its ability to regulate growth factor-dependent or

independent PI3 kinase/Akt signaling pathway via the downregulation of Akt Ser-473

phosphorylation. In further support of this, we found that overexpression of Akt in U87-

Parkin stable cells can overcome the suppressive effects of parkin to result in a

significantly enhanced proliferation rate (Fig. 4A).

As cancer development is a multi-step process involving a wide spectrum of

factors, we were also interested to examine the global gene expression changes in glioma

cells triggered by exogenously-introduced parkin. We subjected parkin-expressing and

vector-control U87MG cells to microarray analyses using Affymetrix chips. Notably, the

expression of cyclin D1 transcript is reduced by about 2-fold in the presence of parkin

over-expression (not shown). Based on the changes in gene expression observed between

Parkin suppresses glioma cell proliferation

control and parkin-expressing U87MG cells (Fig. S4), we were able to derive a parkin

gene signature (see section below). Among these, one gene that caught our attention is

VEGFR-2 (also known as KDR), whose expression is reduced by nearly 4-fold in U87-

Parkin compared to vector control (Fig. S4). Aberrant VEGFR-2 expression is associated

with a wide variety of cancers, including gliomas. Consistent with our microarray

analyses result, anti-VEGFR-2 immunoblotting revealed a significant reduction in parkin-

expressing U87MG cells, but not in those that expresses the T415N mutant, as compared

to control cells (Fig. 4B). Thus, VEGFR-2, along with cyclin D1 and phospho-Akt

(S473), all of which are known to enhance cellular proliferation, exhibit significantly

reduced expression in glioma cells expressing exogenous parkin.

Parkin null fibroblasts exhibit enhanced proliferation rate that is accompanied by

increased levels of cyclin D1, phospho-Akt and VEGFR-2.

To further support the role of parkin as a suppressor of cellular proliferation, we

analysed the growth characteristics of mouse embryonic fibroblasts (MEFs) prepared

from wild type and parkin null mice that we have reported previously (18). As expected,

parkin null fibroblasts exhibit significantly enhanced rate of growth over time (Fig. 5A).

Moreover, parkin -/- MEFs also show elevated levels of cyclin D1, phospho-Akt (S473)

and VEGFR-2 (Fig. 5B). These observations were duplicated in MEFs derived from a

separate line of parkin null mice [generated by means of a targeted deletion of parkin

exon 3 (10)] (Fig. S3C). Importantly, restoration of parkin expression in parkin -/- MEF

results in a significant reduction in the expression of cyclin D1, phospho-Akt (S473) and

VEGFR-2, which is accompanied by a substantially slower growth rate (Fig. 5C-D).

Parkin suppresses glioma cell proliferation

Taken together, our results support an inverse association between parkin function and

cellular proliferation, and identified cyclin D1, phospho-Akt (S473) and VEGFR-2 as key

common denominators of this functional relationship between parkin and cellular growth

that is relevant to gliomagenesis.

Parkin gene signature predicts survival outcome of human glioma patients.

Given the tumor suppressing function of parkin and its apparent ability to reduce

mortality in the glioma mouse model, we sought to determine if parkin mutations or

deletions are present in clinical specimens. The PARK2 gene is located within a known

chromosomal fragile site (FRA6E) and consequently, rearrangements of PARK2 are

expected to be frequent in cancer cells (19-21). Furthermore, Clark et al. mate-paired

sequenced the genome of the U87MG cell line and found an intrachromosomal

translocation within the PARK2 gene (22). This, combined with the loss of expression we

observed in several glioma cell lines, suggest a rearrangement leading to loss of PARK2.

Veeriah et al. also found somatic mutations of PARK2 in 7/75 GBM samples and copy

number loss in 53/216 samples (8). To substantiate our hypothesis, we explored one of

the largest public glioma databases, REMBRANDT (23). We showed not only ~12% of

samples with incomplete copy number loss (<1.7) but a “mixed profile” with 1 group of

copy number identifiers located in the PARK2 gene showing slight gain whilst another

group showed slight loss (Fig. S5A). This is suggestive of wide-spread translocation of

PARK2. Importantly, the segmentation analysis (24) found a significant deletion across

the transcription start site of PARK2 (Fig. S5B). Collectively, these data provide evidence

for the presence of PARK2 mutations/deletions in the model systems presented.

Parkin suppresses glioma cell proliferation

Recent literature implicates that gene expression drives glioma disease

progression (25-27). As the presence of PARK2 rearrangements do not definitively

indicate if parkin-activated signalling occurs, we explored the following: (i) Verify that

PARK2 gene expression correlates with glioma disease progression; (ii) Derive a

differentially regulated gene signature from in vitro parkin-overexpressing and vector-

only U87MG cells, our initial model; and (iii) Interrogate the strength of gene signature

association with individual patient gene expression profiles in REMBRANDT (23) and

“Freije” (28), two well-known, public glioma databases. We show that PARK2

expression portends favourable prognosis and correlates inversely with tumor grade (Fig.

S6). As parkin pathway activation would be better represented by a collective set of genes

rather than a single gene, we adapted the Connectivity Map, a rank-based pattern-

matching approach that was recently successfully applied to determine the degree of

oncogenic pathway activation in gastric cancer (29, 30). Patients with strong resemblance

to the parkin gene signature (Fig. S4, Table S1, 50 genes, parkin-overexpressing versus

vector U87MG cells) would thus be expected to fare better. Indeed, the list of

differentially regulated genes (Fig. 6A, right panel, REMBRANDT; B, right panel,

Freije; Table S2; FDR~0; log2FC1.2) consistently stratified patient survival in both

databases (Fig. 6A, B; left panels); furthermore, a multivariate analysis indicated that the

gene signature prognosticated survival better than current clinical indicators, age and

histology (Table 1). This suggests that parkin pathway activation contributes to the

molecular heterogeneity of gliomas. Patients exhibiting strong concordance with the

parkin signature (parkin+) correlated with lower tumor grade (Fig. 6A, B; middle panels;

Table S3). Additionally, using an independent molecular classifier developed by Phillips

Parkin suppresses glioma cell proliferation

et al. (26), we observed parkin+ patients falling mainly into the Proneural subtype which

typifies patients with better prognosis. Conversely, parkin- patients correlated with higher

tumor grade (GBM) and the Mesenchymal subtype which typifies highly aggressive and

recurrent gliomas. Interestingly, the Mesenchymal subtype mainly correlated with PTEN

loss and EGFR amplification with activated Akt signalling, thus providing clinical

evidence and consistency with our in vitro U87MG-based model. We also observed

elevated AKT1, AKT2 and KDR expression in parkin- patient gene expression profiles,

similar to our in vitro findings (Fig. S7A, REMBRANDT; S7B, Freije).

To conclusively map our gene expression-based parkin activation pathway as a

consequence of varying PARK2 genomic levels, we utilized the copy number information

in REMBRANDT. We found that both PARK2 gene expression levels (Fig. S8A) and

PARK2 activation signature (Fig. S8B) are lower in this PARK2 DNA-lost group than in

the group with diploid PARK2. Taken together, we provide strong evidence that parkin

function correlates inversely with glioma mortality and that its pathway activation is

predictive of survival outcome.

Parkin suppresses glioma cell proliferation

DISCUSSION

The main finding of our current study is that loss of parkin function enhances

cyclin D1 expression and Akt-related growth-promoting signalling and concomitantly

promotes glioma cell proliferation. Importantly, our study also identified a signature

pathway of parkin that is predictive of the survival outcome of glioma patients and is

therefore of potential prognostic value.

It is becoming increasingly clear that parkin dysfunction not only plays a role in

neurodegeneration but also underlies the development of several types of cancers (4).

Notably, Cesari and colleagues have originally demonstrated that the chromosome 6q-

located, 1.4 Mb parkin gene is frequently targeted by hemizygous deletion and

inactivation in both malignant tumors and tumor-derived cell lines (4). Following this

discovery, several other groups including ours have reported parkin gene alterations and

expression variability in a wide variety of tumor types (4-7, 9), including GBMs (8).

Collectively, these studies strongly implicate a role for parkin as a tumor suppressor.

Further supporting this, we showed here that parkin level is dramatically reduced in

several glioma cell lines and that restoration of functional (but not a catalytically-null

mutant) parkin expression in otherwise parkin-deficient glioma cells mitigates their

growth both in vitro and in vivo. Our study also revealed that parkin negatively influences

the cell cycle program of U87MG cells but through the down-regulation of cyclin D1

instead of cyclin E expression. We do not know if this is a cell-specific phenomenon

although it is noteworthy to point out that the suggested role of parkin in cyclin E

degradation is currently controversial (9, 31).

Parkin suppresses glioma cell proliferation

Precisely how functional parkin regulates the expression of cyclin D1 is unclear to

us, although its apparent ability to down-regulate Akt phosphorylation may play a part

here. The Akt pathway is known to enhance G1/S cell cycle progression through the

increase of cyclin D1 level (32). Maximum activation of Akt requires both T308

phosphorylation by PDK1 and S473 phosphorylation by the TORC2 complex (33).

Interestingly, wild type parkin expression leads to a specific reduction in phospho-Akt

(S473) (but not T308) phosphorylation, suggesting that parkin acts on TORC2 instead of

PDK1. Consistent with this, we found that the levels of TORC2 components and

associated substrates are reduced in parkin-expressing glioma cells whereas the

expression of PDK1 remains unaltered. As over-activation of Akt signalling is a frequent

feature of tumors including gliomas, our results thus offer a mechanism underlying

parkin-mediated tumor suppressor function in gliomagenesis. Interestingly, the presence

of parkin also apparently influences the expression of VEGFR-2, one of the two high-

affinity tyrosine kinase receptors involved in tumor neoangiogenesis. Although

predominantly found on endothelial cells, VEGFR have also been detected on cancer

cells including gliomas, suggesting a possible autocrine effect on their growth (34, 35).

Importantly, autocrine regulation of GBM cell cycle progression and viability through

VEGF-VEGFR2 interplay involves co-activation of the PI3/Akt pathway (35), which

parkin appears to regulate. Taken together, our results suggest that parkin-mediated

suppression of glioma cell proliferation involves the regulation of VEGFR2-Akt-cyclin

D1 pathway.

Finally, we provide clinical evidence for our U87MG-based model. We implicate

frequent chromosomal rearrangements with loss of gene expression in the PARK2 locus.

Parkin suppresses glioma cell proliferation

We showed that PARK2 pathway activation correlates inversely with disease progression

and patient survival. Our findings are important for the following reasons: The parkin

gene signature defines molecular heterogeneity in gliomas that cannot be accounted for

by histology alone. This is impactful as histology-based diagnoses currently determine

the treatment regimens. Our findings indicate that the parkin signature can potentially

stratify patients and predict cohorts likely to receive treatment benefit from PI3K-Akt

inhibitor therapies. Thus, a re-examination of the 50-gene list including PARK2, could

conceivably provide useful prognostic indicators to monitor disease progression. Our

finding that a PARK2 deletion occurs at the transcription start site in all glioma patients

may indicate a driver role for PARK2 deletion in gliomagenesis, as a passenger

mutational status would otherwise be captured in only certain subgroups. Collectively,

these data strongly establish a role for parkin in gliomagenesis.

ACKNOWLEDGEMENTS

We thank Shirish Shenolikar for helpful discussions and Toh Tan Boon for

technical assistance. This work was supported by grants from Khoo’s Discovery Award

(L.K.L.), Singapore Millennium Foundation (C.Y.), BMRC 09/01/33/19/611 (C.T.) and

A*STAR Singapore Institute for Clinical Sciences (A.B.T.).

Parkin suppresses glioma cell proliferation

REFERENCE

1. West AB, Dawson VL, Dawson TM. To die or grow: Parkinson's disease and cancer. Trends Neurosci. 2005;28:348-52. 2. Nagakubo D, Taira T, Kitaura H, Ikeda M, Tamai K, Iguchi-Ariga SM, et al. DJ-1, a novel oncogene which transforms mouse NIH3T3 cells in cooperation with ras. Biochem Biophys Res Commun. 1997;231:509-13. 3. Kitada T, Asakawa S, Hattori N, Matsumine H, Yamamura Y, Minoshima S, et al. Mutations in the parkin gene cause autosomal recessive juvenile parkinsonism. Nature. 1998;392:605-8. 4. Cesari R, Martin ES, Calin GA, Pentimalli F, Bichi R, McAdams H, et al. Parkin, a gene implicated in autosomal recessive juvenile parkinsonism, is a candidate tumor suppressor gene on chromosome 6q25-q27. Proc Natl Acad Sci U S A. 2003;100:5956-61. 5. Denison SR, Wang F, Becker NA, Schule B, Kock N, Phillips LA, et al. Alterations in the common fragile site gene Parkin in ovarian and other cancers. Oncogene. 2003;22:8370-8. 6. Picchio MC, Martin ES, Cesari R, Calin GA, Yendamuri S, Kuroki T, et al. Alterations of the tumor suppressor gene Parkin in non-small cell lung cancer. Clin Cancer Res. 2004;10:2720- 4. 7. Wang F, Denison S, Lai JP, Philips LA, Montoya D, Kock N, et al. Parkin gene alterations in hepatocellular carcinoma. Genes Chromosomes Cancer. 2004;40:85-96. 8. Veeriah S, Taylor BS, Meng S, Fang F, Yilmaz E, Vivanco I, et al. Somatic mutations of the Parkinson's disease-associated gene PARK2 in glioblastoma and other human malignancies. Nat Genet. 2010;42:77-82. 9. Tay SP, Yeo CW, Chai C, Chua PJ, Tan HM, Ang AX, et al. Parkin enhances the expression of cyclin-dependent kinase 6 and negatively regulates the proliferation of breast cancer cells. J Biol Chem. 2010. 10. Fujiwara M, Marusawa H, Wang HQ, Iwai A, Ikeuchi K, Imai Y, et al. Parkin as a tumor suppressor gene for hepatocellular carcinoma. Oncogene. 2008. 11. Staropoli JF, McDermott C, Martinat C, Schulman B, Demireva E, Abeliovich A. Parkin Is a Component of an SCF-like Ubiquitin Ligase Complex and Protects Postmitotic Neurons from Kainate Excitotoxicity. Neuron. 2003;37:735-49. 12. Ikeuchi K, Marusawa H, Fujiwara M, Matsumoto Y, Endo Y, Watanabe T, et al. Attenuation of proteolysis-mediated cyclin E regulation by alternatively spliced Parkin in human colorectal cancers. Int J Cancer. 2009;125:2029-35. 13. Moore DJ. Parkin: a multifaceted ubiquitin ligase. Biochem Soc Trans. 2006;34:749-53. 14. da Costa CA, Sunyach C, Giaime E, West A, Corti O, Brice A, et al. Transcriptional repression of p53 by parkin and impairment by mutations associated with autosomal recessive juvenile Parkinson's disease. Nat Cell Biol. 2009;11:1370-5. 15. Garfield AS. Derivation of primary mouse embryonic fibroblast (PMEF) cultures. Methods Mol Biol. 2010;633:19-27. 16. Chong YK, Toh TB, Zaiden N, Poonepalli A, Leong SH, Ong CE, et al. Cryopreservation of neurospheres derived from human glioblastoma multiforme. Stem Cells. 2009;27:29-39. 17. Konopka G, Bonni A. Signaling pathways regulating gliomagenesis. Curr Mol Med. 2003;3:73-84. 18. Von Coelln R, Thomas B, Savitt JM, Lim KL, Sasaki M, Hess EJ, et al. Loss of locus coeruleus neurons and reduced startle in parkin null mice. Proc Natl Acad Sci U S A. 2004;101:10744-9. 19. Letessier A, Garrido-Urbani S, Ginestier C, Fournier G, Esterni B, Monville F, et al. Correlated break at PARK2/FRA6E and loss of AF-6/Afadin protein expression are associated with poor outcome in breast cancer. Oncogene. 2007;26:298-307.

Parkin suppresses glioma cell proliferation

20. Palumbo E, Matricardi L, Tosoni E, Bensimon A, Russo A. Replication dynamics at common fragile site FRA6E. Chromosoma. 2010;119:575-87. 21. Poulogiannis G, McIntyre RE, Dimitriadi M, Apps JR, Wilson CH, Ichimura K, et al. PARK2 deletions occur frequently in sporadic colorectal cancer and accelerate adenoma development in Apc mutant mice. Proc Natl Acad Sci U S A. 2010;107:15145-50. 22. Clark MJ, Homer N, O'Connor BD, Chen Z, Eskin A, Lee H, et al. U87MG decoded: the genomic sequence of a cytogenetically aberrant human cancer cell line. PLoS Genet. 2010;6:e1000832. 23. Madhavan S, Zenklusen JC, Kotliarov Y, Sahni H, Fine HA, Buetow K. Rembrandt: helping personalized medicine become a reality through integrative translational research. Mol Cancer Res. 2009;7:157-67. 24. Zhang Q, Ding L, Larson DE, Koboldt DC, McLellan MD, Chen K, et al. CMDS: a population-based method for identifying recurrent DNA copy number aberrations in cancer from high-resolution data. Bioinformatics. 2010;26:464-9. 25. Atlas TCG. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455:1061-8. 26. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu TD, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell. 2006;9:157-73. 27. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. 2010;17:98-110. 28. Freije WA, Castro-Vargas FE, Fang Z, Horvath S, Cloughesy T, Liau LM, et al. Gene expression profiling of gliomas strongly predicts survival. Cancer Res. 2004;64:6503-10. 29. Lamb J. The Connectivity Map: a new tool for biomedical research. Nat Rev Cancer. 2007;7:54-60. 30. Ooi CH, Ivanova T, Wu J, Lee M, Tan IB, Tao J, et al. Oncogenic pathway combinations predict clinical prognosis in gastric cancer. PLoS Genet. 2009;5:e1000676. 31. Ko HS, von Coelln R, Sriram SR, Kim SW, Chung KK, Pletnikova O, et al. Accumulation of the authentic parkin substrate aminoacyl-tRNA synthetase cofactor, p38/JTV-1, leads to catecholaminergic cell death. J Neurosci. 2005;25:7968-78. 32. Liang J, Slingerland JM. Multiple roles of the PI3K/PKB (Akt) pathway in cell cycle progression. Cell Cycle. 2003;2:339-45. 33. Hresko RC, Mueckler M. mTOR.RICTOR is the Ser473 kinase for Akt/protein kinase B in 3T3-L1 adipocytes. J Biol Chem. 2005;280:40406-16. 34. Masood R, Cai J, Zheng T, Smith DL, Hinton DR, Gill PS. Vascular endothelial growth factor (VEGF) is an autocrine growth factor for VEGF receptor-positive human tumors. Blood. 2001;98:1904-13. 35. Knizetova P, Ehrmann J, Hlobilkova A, Vancova I, Kalita O, Kolar Z, et al. Autocrine regulation of glioblastoma cell cycle progression, viability and radioresistance through the VEGF-VEGFR2 (KDR) interplay. Cell Cycle. 2008;7:2553-61.

Table 1. Results from Pathway Activation Score, Log Rank and Cox Regression

(+) represent patients with likeness to parkin signature; (-) represent patients with inverse gene expression relationship to parkin signature.

p Hazard p # of # of Hazard Ratio Cox -value Cox -value Dataset (+) (-) total (+)(-) %(+)(-) Log Rank p-value Ratio probes samples Multivariate Univariate 0.382 0.215 Rembrandt 111 298 38 51 89 29.87 7.69E-09** 0.006* (0.123 - 5.7E-08** (0.192 - 0.759) 0.375) 0.143 0.110 FreijeA 61 85 8 7 15 17.65 0.00284* 0.04* (0.02 - 0.00976* (0.022 - 0.917) 0.586)

Parkin suppresses glioma cell proliferation

FIGURE LEGEND

Figure 1. Overexpression of parkin in U87MG glioma cells mitigates their

proliferation rate both in vitro and in vivo. (A) Left, Anti-FLAG and anti-parkin

immunoblots of total lysates prepared from U87-vector control (V) and FLAG-parkin-

expressing stable clones (#7, 19 & 25). Equal loading of the different lysates was verified

by anti-actin immunoblotting. Right, Graph showing the growth rate of U87-vector and

U87-Parkin cells, as indicated. (B) As in (A) except that FLAG-parkin-expressing cells

are replaced with U87MG clones (i.e. TN5, 7 and 8) stably expressing FLAG-parkin

T415N, a catalytic null mutant. (C) Bar-graph showing the percentage of U87-vector and

U87-Parkin cells undergoing proliferation, as measured by BrdU assay. These

experiments were repeated at least three times. (*P < 0.05, **P < 0.001, Student’s t-test)

(D) Representative images showing the apparent decreased ability of parkin-expressing

U87 cells in forming colonies on soft agar compared to U87-vector. (E) Top, Digital

photo showing macroscopic tumors (arrows) that developed in the brains of NOD-SCID

mice 4 weeks after being injected with U87 vector cells via the intra-cranial route. In

contrast, such macroscopic tumors are not obvious in mice injected with U87 cells stably

expressing parkin. Bottom, Representative H&E stained brain sections showing the size

of tumor formed in mice injected with U87 vector versus those injected with U87-Parkin

cells, as indicated. Images shown above are of the same magnification. (F) Crude mass of

brains prepared from NOD-SCID mice injected with vector control or parkin-expressing

U87 cells.

Parkin suppresses glioma cell proliferation

Figure 2. Parkin overexpression in U87MG cells reduces the level of cyclin D1 and

delays entry into mitotic phase. (A) Bar graph showing the percentage of U87-vector

and U87-wild type parkin-expressing cells (left) or U87-vector and U87-mutant parkin

T415N cells (TN5, 7 & 8) (right) at different stages of the cell cycle as measured by flow

cytometry. (B) Immunoblots showing the levels of cyclin E and cyclin D1 in lysates

prepared from native U87MG cells (U87), U87-vector (Vector) and FLAG-parkin-

expressing stable clones (PK7, 19 & 25). The anti-cyclin D1 immunblots from three

independent experiments were used to derive the relative densitometric units of cyclin D1

(normalized to respective actin level), which is presented as a histogram on the right. (C)

As in (B) except PK7, 19 and 25 are substituted by TN5, 7 and 8.

Figure 3. Phosphorylation of Akt at Ser-473 is significantly repressed in parkin-

expressing U87MG cells. (A) Immunoblots showing the levels of phoshorylated-Akt

(S473) and total Akt in lysates prepared from U87-vector (Vector) and FLAG-parkin

(PK7, 19 & 25) or FLAG-parkin T415N (TN5, 7 & 9)-expressing stable clones. The anti-

Akt immunblots from three independent experiments were used to derive the relative

densitometric units of phospho-Akt (normalized to respective total Akt level), which is

presented as a histogram (right). (B) As in (A) except anti-phospho-Akt (T308) was used

instead of anti-phospho-Akt (S473). (C) As in (B) except anti-phospho-PDK1 and anti-

PDK1 were used to probe the immunoblots. (D) Immunoblots showing the expression

level of various TORC2 components (i.e. mTOR, Rictor and Sin1) and their associated

substrates (i.e. FoxO3a and GSK3β) in U87-vector and U87-Parkin cells. (E)

Immunoblots showing the levels of phoshorylated-Akt (S473) and total Akt in lysates

Parkin suppresses glioma cell proliferation

prepared from quiescent U87-vector (Vector) and FLAG-parkin (PK7, 19 & 25) in the

absence or presence of EGF treatment. The relative level of phospho-Akt/Akt was

determined by densitometry and presented as a histogram (bottom).

Figure 4. Parkin-mediated suppression of glioma cell growth can be overcome by

AKT overexpression. (A) (i) Immunoblots (top) and bar-graph (bottom) showing the

expression of AKT in various U87MG cells following lentiviral delivery of mCherry or

mCherry-AKT2. (ii) Bar graph showing the proliferation rate of mCherry and mCherry-

AKT2 expressing U87-Parkin cells relative to their respective vector control. (B)

Immunoblots showing the levels of VEGFR-2 in lysates prepared from native U87MG

cells (U87), U87-vector (Vector) and U87-FLAG-parkin (PK7, 19 & 25). The anti-

VEGFR-2 immunoblot from three independent experiments were used to derive the

relative densitometric units of VEGFR2 (normalized to respective actin loading control

level), which is presented as a histogram (bottom). (B) As in (A) except U87-FLAG-

parkin (PK7, 19 & 25)-expressing cells were replaced by U87-FLAG-parkin T415N

(TN5, 7 & 8)-expressing cells.

Figure 5. Expression of cyclin D1, phospho-AKT and VEGFR2 are upregulated in

parkin null fibroblasts. (A) Graph showing the growth rate of primary embryonic

fibroblasts derived from wild type (parkin +/+) or parkin null mice (parkin -/-). (B)

Immunoblots showing the levels of cyclin D1 (top), phospho-Akt (S473) (middle) and

VEGFR2 (bottom) in parkin +/+ and -/- MEFs. These experiments were repeated at least

three times and the relative densitometric units of cyclin D1/actin, phospho-AKT/AKT

Parkin suppresses glioma cell proliferation

and VEGFR2/actin in parkin +/+ and parkin -/- MEFs are presented as histogram

alongside their respective immunoblots. (C-D) As in (A-B) respectively except that data

for parkin +/+ MEFs is replaced by parkin null MEFs ectopically expressing FLAG-

parkin. (Inset in both graphs shows anti-parkin immunoblot).

Figure 6. Parkin gene signature is predictive of survival outcome of glioma patients.

The 50-gene signature stratified patient survival in 2 public glioma databases: (A)

REMBRANDT; and (B) Freije (left panels). Survival data are plotted as Kaplan-Meier

curves. (+), concordance of parkin-overexpression gene signature with patients’ gene

expression and consequently better survival; (-), inverse relation of parkin-overexpression

with clinical database gene expression and consequently poorer survival. Activation

scores denote ranking of gene signature likeness to clinical database gene expression, and

patient survival is linked to this ranking (middle panels). A positive score represents

patients with similarity to parkin overexpression, while a negative score denotes patients

with inverse relation, i.e. similarity to vector-only cells. The tumor grade and molecular

classification scheme were scored for individual patients and illustrated below the

activation score graphs. Heatmaps illustrate the expression patterns of genes in individual

patients (right panels). The corresponding parkin signature probe sets in the Affymetrix

platforms can be found in Table S2 and the list of patients in parkin+ and parkin- in each

dataset can be found in Table S3.

Parkin pathway activation mitigates glioma cell proliferation and predicts patient survival

Calvin W.S. Yeo, Felicia S.L. Ng, Chou Chai, et al.

Cancer Res Published OnlineFirst March 19, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-11-3060

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2012/03/19/0008-5472.CAN-11-3060.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2012/03/17/0008-5472.CAN-11-3060. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.