DRR regulates AKT activation to drive brain cancer invasion

ORIGINAL ARTICLE

DRR regulates AKT activation to drive brain cancer invasionA Dudley1, M Sater1, PU Le1, G Trinh1, MS Sadr1, J Bergeron1, GF Deleavey2, B Bedell1, MJ Damha2 and K Petrecca1

Glioblastoma (GBM) is the most common and invasive adult brain cancer. The rapid invasion of cancer cells into the normalbrain is a major cause of treatment failure, yet the mechanisms that regulate this process are poorly understood. We have identifieda novel mechanism of brain cancer invasion. We show that downregulated in renal cell carcinoma (DRR), which is newly expressedin invasive gliomas, recruits AKT to focal adhesions. This DRR- induced pathological relocalization of AKT bypasses commonlyaltered upstream signaling events and leads to AKT activation and invasion. We also developed an oligonucleotide therapeuticthat reduces DRR expression and prevents glioma invasion in an in vivo preclinical model of the disease. Our findings identify DRRas a novel GBM target and show that oligonucleotides targeting DRR is a novel therapeutic approach for the treatment ofDRR-positive GBMs.

Oncogene advance online publication, 21 October 2013; doi:10.1038/onc.2013.436

Keywords: brain cancer; glioblastoma; invasion; AKT; DRR

INTRODUCTIONGlioblastoma (GBM) is the most common adult brain cancer. Themedian survival with current therapy is 14.6 months.1,2 Cancer cellinvasion into the normal brain is a major contributor to this rapidtreatment failure, yet the mechanisms governing this phenomenonremain poorly understood.

The epidermal growth factor receptor (EGFR)/phosphatidylino-sitol 3-kinase (PI3K)/AKT pathway is a major driver of GBMinvasion.3 EGFR activates the lipid kinase PI3K, which thenphosphorylates PI(4,5)P2 converting it to PI(3,4,5)P3. PIP3

(phosphatidylinositol (3,4,5)-triphosphate) in turn binds thepleckstrin homology domain of AKT thereby recruiting it to thecell membrane. Once there, AKT is activated by phosphorylation atThr308 and Ser473 by 3-Phosphoinositide-dependent proteinkinase 1 and mammalian target of rapamycin complex 2,respectively.4–8 Activated AKT translocates from the membraneto the cytosol and the nucleus, where it drives downstreampathways affecting cell proliferation, survival, metabolism andinvasion.9–11 This cascade can be terminated by phosphatase andtensin homolog (PTEN), which interrupts PIP3 signaling bydephosphorylating PIP3 to PIP2.3,8,10

Oncogenic alterations of EGFR, PIK3CA and PTEN have beenidentified in GBM.12 Combined, this pathway is deregulated inover 80% of these tumors, and elevated phosphorylated AKT(pAKT) levels are found in 85% of GBMs.12–14 EGFR is thus anattractive therapeutic target, yet clinical trials testing EGFRinhibitors in GBM have yielded disappointing results.15–17

Importantly, AKT can also be activated independently ofreceptor tyrosine kinase (RTK) or PI3K activity. The viral oncogenev-akt is created by the addition of a myristoylation signal to theamino terminus of AKT; this confers constitutive activation to AKT,allowing it to associate with the cell membrane without the needfor upstream RTK or PI3K involvement.18,19 Similarly, a mutation inthe pleckstrin homology domain of AKT leads to the association ofAKT with the cell membrane and constitutive activation in breast,colorectal and ovarian cancers.20 Thus, events promoting AKT

localization to the cell membrane can be sufficient for itsactivation.

Downregulated in renal cell carcinoma (DRR) is a recentlyidentified driver of glioma invasion.21 It has been reported tofunction as a putative tumor suppressor protein.22–24 Normallyexpressed in the neurons, DRR has been shown to bundle actinand regulate neurite outgrowth.25 In the context of gliomas, it isexpressed in both low- and high-grade gliomas but not in normalglia.21 DRR localizes along stress fibers, at focal adhesions (FAs)and within the nucleus, and its expression promotes FAturnover.21 Despite the evidence that DRR participates in gliomainvasion, the underlying mechanism is unknown.

As the EFGR/PI3K-PTEN/AKT pathway is a driver of GBM invasionthat is altered in the large majority of GBMs, and DRR isoverexpressed in invasive gliomas, we tested whether DRR isinvolved in this pathway. We found that DRR expression leads toelevated AKT activation by recruiting AKT to FAs in a mannerdependent on SRC-family kinases (SFKs) and cell adhesion. Wealso found that reduction of DRR expression using antisenseoligonucleotides (AOs) prevents glioma invasion in an in vivoxenograft mouse model. Therefore, our study reveals that DRR is anovel GBM target and that AOs targeting DRR is a promising newtherapeutic approach to prevent brain cancer invasion.

RESULTSDRR induces AKT phosphorylation and this event is independentof EGFR signalingThe EGFR/PI3K-PTEN/AKT signaling pathway is altered in over 80%of GBMs, and pAKT is elevated in over 85% of GBMs.12–14 In thisstudy, we used the prototypical human GBM-derived U251 cells(CTL) as a model to create a stable DRR-overexpressing (DRRov)cell line. We previously characterized these cell lines and foundthat upon DRR overexpression cells become more invasive andthe rate of FA turnover is increased.21 To determine whether DRRaffects AKT activation, we assessed pAKT levels in both CTL and

1Department of Neurology and Neurosurgery, Montreal Neurological Institute and Hospital, McGill University, Montreal, Quebec, Canada and 2Department of Chemistry, McGillUniversity, Montreal, Quebec, Canada. Correspondence: Dr K Petrecca, Department of Neurology and Neurosurgery, Montreal Neurological Institute and Hospital, McGillUniversity, 3801 University Ave, Suite 109c, Montreal H3A 2B4, Quebec, Canada.E-mail: [email protected] 28 March 2013; revised 2 August 2013; accepted 12 August 2013

Oncogene (2013), 1–9& 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13

www.nature.com/onc

http://dx.doi.org/10.1038/onc.2013.436

mailto:[email protected]

http://www.nature.com/onc

DRRov cells. Gain of DRR expression was associated withsignificantly elevated AKT phosphorylation at both the Thr308and Ser473 sites compared with CTL cells, whereas total AKTexpression was similar in both cell lines (Figure 1a). pAKTimmunolabeling also revealed significantly elevated pAKT expres-sion in DRRov cells compared with CTL cells (Figure 1b). As EGFRsignaling is altered in 45–55% of GBMs and is a well-characterizedactivator of the PI3K/AKT pathway under normal and pathophy-siological conditions,12–14 we investigated EGFR proteinexpression and phosphorylated EGFR levels in the context ofDRR expression. We found that DRRov cells express significantlyhigher levels of EGFR and activated EGFR, as indicated byphosphotyrosine (Figure 1c) and phosphorylated EGFR (Tyr 845)(Supplementary Figure 1a), as well as cell surface EGFR comparedwith CTL cells (Figure 1d). We analyzed EGFR mRNA levels byquantitative real-time PCR in DRRov and CTL cells and did notobserve a significant difference between cell lines, suggestingpost-transcriptional regulation (Supplementary Figure 1b).

We used the EGFR kinase inhibitors AG1478 26 and gefitinib27 todetermine whether this increased EGFR/phosphorylated EGFRexpression was responsible for the DRR-induced AKTphosphorylation. We treated CTL cells with EGF and observedan increase in phosphotyrosine and pAKT levels, which wereinhibited after treatment with EGF in the presence of AG1478. We

also observed a decrease in phosphotyrosine levels in DRRov cellsafter AG1478 treatment, whereas the levels of pAKT did notchange (Figure 1e). Gefitinib, while effectively blocking EGF-induced EGFR phosphorylation in both cell lines, also did notdecrease pAKT levels in DRRov cells (Supplementary Figure 1c).Consistent with this observation, we found that EGFR inhibitiondid not significantly reduce DRRov cell invasion in a three-dimensional invasion assay (Figures 1f, g). Thus, althoughEGFR/phosphorylated EGFR expression is elevated in DRRov cells,it is not involved in DRR-induced AKT phosphorylation or cellinvasion.

DRR-induced AKT phosphorylation is SFK- and integrin-dependentActivating inputs that can lead to AKT phosphorylation includeRAS/RAF/MEK/ERK kinases, Rho-GTPases, integrin-linked kinase(ILK), SFKs, FA kinase (FAK) and PI3K.14,28–32 We tested each ofthese inputs in order to determine how DRR regulates AKTactivation.

The MEK inhibitor U0126 is a non-ATP competitive inhibitor thattargets MEK1 and MEK2 and effectively inhibits downstream ERKphosphorylation33 (see Supplementary Figure 2a). We found thatMEK inhibition did not reduce pAKT levels in DRRov cells(Figure 2a).

EGFR

tubulin

0h 24h 48h 72h

EGFR

(EGFR) 1.0 1.1 0.1 0.1

EGFAG1478

- +++- -

EGF - + - +

0

100

200

24h 48h 72h

tubulin

DRR

0

1

2

3

4

CTL

- +++- -

pTyr

pAKT

AKT

tubulin

18

60

60

60

55

175

170

55

17560

60

55

(S473) 7.6 1.0

DRRovunt

DRRov+AG1478 In

vasi

on d

ista

nce

(μm

)

DRRov untreated

DRRov AG1478

Rel

ativ

e T

otal

Cel

l Flu

ores

cenc

e(n

orm

aliz

ed to

con

trol

) **

CTLDRRov

pAKTT308

pAKTS473

AKT

DRRov CTLpTyr

CTLDRRov

DRRov CTL

DRRov

Figure 1. Phospho-AKT is elevated in DRR-overexpressing cells and DRR-induced AKT phosphorylation is independent of EGFR signaling.(a) Western blot showing the expression of DRR, Threonine 308 and Serine 473 AKT phosphorylation, total AKT, in DRRov and CTL cells.Tubulin is shown as a loading control. pAKT Ser473 band densitometry indicated. n¼ 3, DRRov pAKT vs CTL pAKT Po0.05 (t-test).(b) Quantification of pAKT (Ser473) fluorescence intensity in DRRov and CTL cells. Represented as mean±s.e.m with total fluorescenceintensity normalized to CTL cells. n¼ 9, **Po0.0001 (t-test). (c) Western blot showing elevated EGFR and phosphotyrosine (pTyr) levels inDRRov cells compared with CTL cells. EGFR band densitometry indicated. (d) Surface membrane EGFR expression in DRRov and CTL cells.Scale bar¼ 20 mm. (e) Western blot showing that EGFR inhibition with AG1478 does not reduce pAKT (Ser473) expression levels in DRRov cells.(f ) Three dimension (3D) invasion assay of DRRov cells untreated (unt) or treated with AG1478. (g) Quantification of invasion. Represented asmean±s.e.m. n¼ 3, P40.05 (ANOVA). EGFR inhibition does not significantly reduce DRR-induced cell invasion.

DRR activates AKT to drive brain cancer invasionA Dudley et al

2

Oncogene (2013) 1 – 9 & 2013 Macmillan Publishers Limited

Rho-family GTPases are involved in AKT regulation and also actdownstream of AKT to regulate cytoskeletal dynamics and cellmovement.29,34–36 Application of the Rho A, B and C inhibitor,exoenzyme C3 transferase, to DRRov cells predictably alteredcytoskeletal architecture and cell morphology by reducing stressfiber levels and collapsing cell structure (see SupplementaryFigure 2b). However, this treatment did not affect pAKT levels(Figure 2b).

ILK operates between integrins and RTKs and is an importantactivator of AKT.31,37 ILK expression was abolished in DRRov cellsusing an ILK-targeted small interfering RNA oligonucleotide andwas unaltered after transfection with a scrambled sequence(Figure 2c). However, silencing ILK expression did not decreasepAKT levels (Figure 2c).

SFKs are well-characterized effectors of integrin signaling andare key regulators of FA dynamics.38 Application of the SFKinhibitor PP2 (5 mM) to DRRov cells reduced pAKT levels, whereasthe inactive analog PP3 had no significant effect on pAKT levels

(Figure 2d). We next seeded DRRov cells on the integrin ligandfibronectin in order to generate physiologically relevant FAs. Inthis culture environment, we treated DRRov cells with the SFKinhibitors PP2 (10 mM) (Figure 2e) and SU6656 (5 mM)(Supplementary Figure 3a) and observed reduced pAKT levels;suggesting that SFKs are regulating DRR- induced AKT activationin a cell-matrix adhesion-dependent manner.

Activation of FAK is an early event following adhesion, isdependent upon SFK activity and leads to AKT phosphoryla-tion.38,39 To test FAK involvement in AKT phosphorylation, weblocked FAK activity by treating DRRov cells with the FAK inhibitorPF-228 for 1 h before seeding cells on fibronectin. This treatmenteffectively prevented the phosphorylation of FAK at Y397, yetpAKT levels were not reduced (Figure 2f), suggesting that DRR-induced AKT phosphorylation occurs independently of FAKactivity.

To determine whether PI3K is involved, we treated DRRov cellswith the PI3K inhibitors LY294002 (5, 10 and 20 mM) and

tubulin

EGF - + ++ + - + ++ +5 10 5 10

+ +20 20

CTL

dmso dmso

tubulin

30min 1.5hr

tubulin

0 0 0 0

125

60

55

175

60

60

55

60

60

125

125

55

(pAKT) 1.0 1.0 0.3

(pAKT) 1.0 0.31.0

pAKT

AKT

pFAK

FAK

tubulin

60

60

125

125

55

CTL CTLDRRovDRRov

30min 4hr

0.1 0.2 0

LY294002(μM)

AKT

pAKT

pTyr pAKT

AKT

pFAK

FAK

DRRovDRRov

FN+LY

FNunt

unt

FN+LY

FN

FN+PP2

FNunt

FN+PP2

FNunt

FN+PP2

FNunt

FN+PP2

FNunt

pFAK

pAKT

FN+PF

FNunt

DRRov

100nM

ILK

120nM 100nM 120nM

48hr 72hr

tubulin

pAKT

CTLDRRov

tubulin

pAKT

pAKT

tubulin

AKT

EGF - + ++ + - + ++ +5 10 5 10 dmso0 0 0 0dmso

60

55

60

55

51

60

55

60

60

55

(pAKT) 1.0 0.3 0.7

pAKT

tubulin

scr

PP3PP2

unt

siscr

siscr

siscr

siunt

unt

DRRov

U0126(μM)

unt

C3 unt

C3

CTLDRRov

Figure 2. DRR-induced AKT phosphorylation is SFK and adhesion-dependent. western blots showing (a) pAKT (Ser473) levels in DRRov andCTL cells untreated or treated with MEK inhibitor U0126 or with dimethyl sulfoxide (DMSO) in the absence or presence of EGF. (b) pAKT(Ser473) levels in response to the Rho inhibitor, C3 transferase, in DRRov and CTL cells. (c) pAKT (Ser473) levels in DRRov cells transfected for48 or 72 h with small interfering RNA (siRNA) (si) against ILK or control scrambled siRNA (scr). (d) pAKT (Ser473) and total AKT levels inresponse to the SFK inhibitor PP2 and its inactive analog PP3 in DRRov cells. pAKT band densitometry indicated. n¼ 3, PP2 vs unt Po0.05(t-test). PP3 vs unt P40.05 (t-test). (e) pAKT (Ser473) and AKT levels in response to PP2 treatment in DRRov and CTL cells seeded on fibronectin(FN) for 30min or 4 h. pAKT band densitometry indicated. n¼ 3, pAKT DRRov FNþ PP2 vs DRRov unt or DRRov FN Po0.05 (analysis ofvariance, ANOVA, Bonferroni). (f ) pFAK and pAKT (Ser473) levels after FAK inhibition with PF-228 (PF). (g) pAKT (Ser473) levels in response toPI3K inhibition with LY294002 in the absence or presence of EGF. pTyr. (h) pAKT (Ser473) levels in response to LY294002 (LY) in DRRov cellsseeded on fibronectin for 30min or 1.5 h. pAKT band densitometry indicated. n¼ 3, pAKT DRRov FNþ LY vs DRRov unt or DRRov FN Po0.05(ANOVA, Bonferroni). For all western blots, tubulin is shown as a loading control.


3

& 2013 Macmillan Publishers Limited Oncogene (2013) 1 – 9

wortmannin (100 nM). Although AKT phosphorylation was effec-tively inhibited in CTL cells after treatment with both inhibitors,neither LY294002 nor wortmannin prevented AKT phosphoryla-tion in DRRov cells (Figure 2g, Supplementary Figure 3b).Following our observation that SFK inhibition of AKT phosphor-ylation was adhesion-dependent, we examined whether PI3Kinhibition would be effective in reducing pAKT levels in anadhesion-dependent context. By repeating the experiments onfibronectin, we found that even low concentrations (5mM) ofLY294002 could decrease AKT phosphorylation (Figure 2h). Thesefindings reveal the importance of integrin-mediated adhesion inDRR-induced AKT phosphorylation.

Integrins interact with the extracellular-matrix through the RGD(arginine–glycine–aspartic acid) amino-acid sequence.40 We useda modified RGD peptide that competes with the RGD domain ofthe extracellular-matrix for integrin binding to directly test theadhesion dependence we have observed. We treated cells insuspension with the modified RGD peptide for 30 min beforeseeding cells on fibronectin. This treatment gave rise to a non-adherent population of cells. In this non-integrin binding cellpopulation there was a loss of AKT phosphorylation (Figures 3a, b).Similarly, AKT was not phosphorylated in DRRov cells that weremaintained in suspension without RGD treatment (Figures 3a, b).

DRR recruits AKT to FAsWe have previously showed that within the cytoplasm, DRR islocalized at FAs and along stress fibers.21 Our results showing thatDRR regulates AKT phosphorylation in an adhesion- and SFK-dependent manner suggest that DRR may be recruiting AKT toFAs. We tested this by immunolabeling AKT and pAKT in DRRovand CTL cells seeded on fibronectin. For each time point studied(30 m to 6 h), both AKT and pAKT shared an overlappingexpression pattern with vinculin in DRRov cells (Figure 3c). Incontrast, AKT and pAKT were expressed diffusely in the cytoplasmof CTL cells (Figure 3c).

As both DRR and SFKs regulate FA dynamics, we next testedwhether SFKs are involved in DRR-induced AKT recruitment to FAs.DRRov cells seeded on fibronectin and treated with 10 mM PP2developed FAs containing vinculin, yet pAKT was not observed atthese FAs (Figure 3d). To further explore the possibility that DRRrecruits AKT to FAs, we utilized a mutant form of DRR, DRRDPEPE,which does not localize to FAs.21 We found that when this form ofDRR is expressed, we did not observe colocalization betweenpAKT and FAs (Figure 3e), and pAKT levels were not elevated(Figure 3f).

SFK and PI3K inhibitors prevent DRR-induced invasionHaving identified a mechanistic link between adhesion, DRR, SFK/PI3K activity and AKT, we next tested whether this signaling axisdirectly affected the invasive properties of DRRov cells. To thisend, we tested whether SFK or PI3K inhibition could preventDRRov cells from invading in a three-dimensional invasion assay.We found that treatment of DRRov spheroids with either 10 mM

PP2 or LY294002 significantly reduced the invasiveness of thesecells at 48 and 96 h (Figures 4a, b), whereas CTL spheroids showedsignificant reductions only at 96 h (Figures 4a, c). We also observedreduced invasion of DRRov spheroids in the presence of LY294002and 5-Fluorouracil, an inhibitor of cell division41 (SupplementaryFigure 4). This observation suggests that the SFK/PI3K/AKTpathway is implicated in DRR-mediated invasion.

Antisense oligonucleotide-mediated DRR ablation preventsinvasion in a GBM mouse modelWe have shown that DRR regulates AKT phosphorylation to driveinvasion, and that treatment with SFK and PI3K inhibitors cansignificantly reduce this invasion. Our next step was to test

whether reduction of DRR expression could prevent brain cancerinvasion in vivo. Direct application of AOs to brain cancers couldbe an effective treatment modality.42 We designed AOs targetingDRR after having previously shown that malignant glioma cellsexpressing negligible levels of DRR displayed decreased invasiveproperties in vivo compared with those overexpressing DRR.21

GBM stem cell lines cultured in the absence of serum closelyreproduce the pathological features of human GBMs in in vivointracranial models.43 First, we assessed DRR levels in a humanGBM stem cell line (hGSC) and found that it was expressed(Figure 5a). Upon treatment with DRR-targeted AOs, we were ableto efficiently knock down DRR levels and observed a simultaneousdecrease in pAKT compared with scramble AOs (Figure 5a). Thesefindings indicate that DRR regulates AKT signaling in a gliomastem cell model and confirm our data from the DRRov cells.

We extended these findings to an in vivo model. hGSCs treatedwith Cy5-conjugated control AOs and AOs targeting DRR wereimplanted in the striatum of immuno-compromised CD1 mice. Thestem cell marker Sox-2 was used to identify all implanted cells,whereas Cy5 labeling was used to identify tumor cells that hadbeen treated with AOs. Whereas cancer cells treated with Cy5-conjugated control AOs readily invaded the peritumoral brain wellbeyond the tumor site (Figure 5b upper panels, c), cancer cellstreated with anti-DRR AOs did not, but instead remained withinthe well circumscribed tumor mass (Figure 5b lower panels, c).These findings show that targeting DRR in vivo is an efficientapproach to reduce brain cancer invasion and suggest that AOsare an effective method for reducing DRR levels.

DISCUSSIONOur studies identify a novel mechanism for AKT activation andinvasion in brain cancer. We have also developed a promisinganti-invasive therapeutic to prevent brain cancer invasion. Weshow that DRR, which is newly expressed in invasive gliomas,21

recruits AKT to FAs in a SFK- and adhesion-dependent manner. Wealso show that the inhibition of SFKs, and to a lesser extent PI3K,prevents AKT phosphorylation. Finally, we show that the reductionof DRR expression using AOs targeting DRR prevents invasion in axenograft mouse model of the disease.

Mutations in AKT that target it to the cell surface, which bypassupstream signaling events (such as RTK/PI3K activation), can besufficient for full AKT phosphorylation,18 and these alterationshave been described in cancers.20 Our findings support a model inwhich DRR pathologically relocalizes AKT to FAs in a SFK-dependent manner. AKT is phosphorylated at FAs and thiselevated level of activated AKT remains detectable for at least6 h, suggesting that under these conditions AKT is hyper-active.Furthermore, we observed that PI3K inhibition only reduces AKTphosphorylation upon integrin activation, suggesting thatdiffusely localized PI3K does not have a role in AKTphosphorylation in the presence of DRR. These results support anovel mode of AKT activation following its recruitment to integrin-based FAs.

AKT is activated in the large majority of GBMs, and emergingevidence suggests that AKT has a role in GBM invasion.44 As AKT isnot frequently mutated in GBM, its activation is presumed to beunder the control of upstream RTKs. Our findings that AKTactivation is not under the control of EGFR in DRR-positive gliomacells may provide an explanation for the negative results of EGFRinhibitors in clinical trials.45,46 In contrast, our studies point to DRRas an important molecular player in GBM invasion, making DRR avaluable target for therapeutics as well as a potentially usefulbiomarker for EGFR-independent brain cancer invasion.

Our results highlight the importance of SFKs and integrins astargets to prevent GBM invasion. Dasatinib is a potent SRC andSFK inhibitor that has been shown to inhibit migration andinvasion of GBM cells.47,48 These data correlate well with our


4


pFAK

tubulin

DRRoV

FAK

AKT

vinculin vinculin

DRRov CTL

pAKT6h

vinculin pAKTpAKT merge

merge

vinculin merge

DRRov DRRov+PP2

vinculin

tubulin

AKT

pAKT

DRRov

DRRΔPEPE

DRRΔPEPE

****

125

125

60

60

55

60

60

55

AKT30min

pAKT30min

pAKT

merge merge

pAKT

Rel

ativ

e de

nsity

(A

.U.) 1.5

1.0

0.5

0.0

DRRov

FN

RGD non

-adh

susp

ensio

n

FN No RGD

in su

spen

sion

FN+RGD

non-

adh

Figure 3. DRR recruits AKT to FAs. (a) Western blot showing that pAKT (Ser473) levels are reduced in the presence of the integrin blocker RGD.Lane 1-cells grown on FN, lane 2-cells in the non-adherent (non-adh) fraction after RGD treatment, lane 3-cells in suspension without RGDtreatment. (b) Quantification of pAKT (Ser473) band densitometry corrected for loading control. Represented as mean±s.e.m with valuesmeasured relative to the FN condition. n¼ 3, **Po0.05 (ANOVA, Bonferroni). (c) Co-immunolabeling of AKT (green) or pAKT (green) with theFA marker, vinculin (red) in DRRov and CTL cells grown on FN. (d) Co-immunolabeling of pAKT (green) and vinculin (red) in DRRov cellstreated with PP2. (e) Co-immunolabeling of pAKT (green) and vinculin (red) in DRRDPEPE showing that pAKT is no longer localized at FAs.All pAKT immunolabeling is specific for Ser473. Scale bars¼ 20mm. (f ) Western blot showing that pAKT (Ser473) is not elevated in cellsexpressing DRRDPEPE.


5


findings and suggest that SFK inhibitors may be useful inpreventing DRR-induced brain cancer invasion. A phase I/II trialinvolving patients with newly diagnosed GBM is currently testingdasatinib combined with radiotherapy and concomitanttemozolomide, followed by adjuvant dasatinib plustemozolomide.49,50 The integrin inhibitor Cilengitide is alsobeing tested as a first-line therapy in GBM.51 In these trialsprogression-free survival and overall survival are primary endpoints. Given our findings, it would be relevant to assess theextent of tumor invasion on surveillance imaging as anotherimportant biomarker of efficacy.

AOs delivered to the resection cavity of GBM patients viaimplanted catheters, thereby bypassing the need for systemicdelivery is currently in Phase II/III clinical trials. Clinical trials testingAOs targeting transforming growth factor b2 have shownencouraging results.42 We have developed an AO that targetsDRR expression and prevents brain cancer invasion in a xenograftmodel. As GBM invasion into normal brain is a local event in over85% of patients,52 delivery of AOs targeting invasion directly intothe resection cavity provides treatment to the most relevant areawith minimized toxicity. Upon tumor resection, catheters can beplaced directly into the resection cavity providing a simplemechanism for routine AO delivery. Given that this therapeuticmodality is currently being investigated, our findings suggest thatanti-DRR AOs could be a promising and practical therapy to treatDRR-positive GBM patients, with a reduction in invasion as atreatment endpoint.

MATERIALS AND METHODSAntibodies and reagentsAnti-phospho-AKT (Ser473), anti-phospho-AKT (Thr308), anti-AKT, anti-phospho-p44/42 MAPK (Thr202/Tyr204), Anti-p44/42 MAPK, anti-ILK1, anti-phospho-EGFR (Tyr 845), Signal Silence ILK1 and scramble siRNA II kit (Cell

Signaling Technology, Cat. nos. 6528 & 6201) (Boston, MA, USA). Anti-EGFR(Santa-Cruz Technology, Dallas, TX, USA). Anti-phosphotyrosine, anti-FAK(Millipore, Billerica, MA, USA). Anti-phospho-FAK (Tyr397) (Invitrogen,Burlington, ON, Canada). Anti-a-tubulin, anti-a-vinculin fibronectin frombovine plasma, 5-fluorouracil (Sigma, St Louis, MO, USA). AG1478, U0126,LY294002, wortmannin, PP2, SU6656 PF-228, GRGDSP peptide(Calbiochem, Billerica, MA, USA). Gefitinib (Cayman Chemical, Ann Arbor,MI, USA). C3 transferase (Cytoskeleton Inc., Denver, CO, USA). Recombinanthuman EGF (Invitrogen). PureCol Bovine Collagen Solution Type 1(Advanced BioMatrix, San Diego, CA, USA). SYBR Green PCR Master Mix(Roche, Ann Arbor, MI, USA).

Cell cultureDRRov cells: AmphoPack 293 cells were first transfected with the DRRplasmid pLib (Clontech, Mountain View, CA, USA) using Lipofectaminereagent according to the manufacturer’s protocol. Forty-eight and seventy-two hours post transfection, retrovirus was collected and added to thehuman GBM-derived U251 cell line. Stable cell lines were selected withG418 media. U251 brain cancer cells and DRR-overexpressing cells werecultured as previously described.21 Human GBM stem cells (hGSCs)were kindly provided by Dr Samuel Weiss (University of Calgary). hGSCswere isolated as previously performed53 and expanded in neurospherecultures. Spheres were cultured in complete Neurocult-NS-A proliferationmedium (Neurocult basal medium containing: Neurocult-NS-Adifferentiation supplement at a concentration of 1/10 dilution, 20 ng/mlrh EGF, 20 ng/ml rh bFGF and 2 mg/ml Heparin) from Stemcell Technologies(Vancouver, BC, Canada). When spheres appear large enough for passaging(o300mm in diameter), they were collected in tubes and spun at1200 r.p.m. for 3 min. To dissociate the spheres, 700 ul of Accumax(Millipore) was added to the cell pellet and incubated for 5 min at 37 1C,they were then washed with phosphate-buffered saline, centrifuged andresuspended in complete Neurocult-NS-A proliferation medium andseeded at a concentration of 200 000 cells/flask.

Pharmacological inhibition assaysCells were treated overnight with 1 mM AG1478 or 5 mM gefitinib beforebeing treated with 50 ng/ml EGF either alone or in the presence of 1 mM

0h 48h

DRRov

96h 0h 48h

CTL

96h

Untreated

LY294002

PP2

500

400

300

Inva

sion

dis

tanc

e (μ

m)

Inva

sion

dis

tanc

e (μ

m)

200

100

0

400DRRov untDRRov PP2DRRov LY294002

CTL untCTL PP2CTL LY294002300

200

100

0

Time (h)48 96

Time (h)48 96

**** ***********

******

Figure 4. SFK and PI3K inhibitors prevent DRR-induced invasion. (a) 3D invasion assays of DRRov and CTL spheroids untreated or treated with10mM PP2 or LY294002. Invasion front is demarcated by a dashed circle. (b) Quantification of 3D invasion of DRRov spheroids with indicatedtreatment. Represented as mean±s.e.m. n¼ 3, **Po0.01, ****Po0.0001 (ANOVA, Bonferroni). (c) Quantification of 3D invasion of CTLspheroids with indicated treatment. Represented as mean±s.e.m. n¼ 3, ***Po0.001, ****Po0.0001 (ANOVA, Bonferroni).


6


AG1478 or 5 mM gefitinib for 10 min at 37 1C; with 5–20 mM U0126for 2 h at 37 1C, then stimulated with EGF alone or in the presence ofU0126 for 10 min at 37 1C before lysis; with 2 mg/ml C3 transferase for 5 hat 37 1C before lysis; with 5 or 10 mM PP2 (or PP3) or 5 mM SU6656 for 5 h at37 1C before lysis; with 200 nM PF-228 for 1 h at 37 1C before lysis; with 5–20 mM LY294002 or 100 nM wortmannin for 2 h or 1 h, respectively, andthen stimulated with EGF alone or in the presence of LY294002 or

wortmannin for 10 min at 37 1C before lysis. For the experiments in whichfibronectin was used, 24-well plates were coated with 50 mg/mlfibronectin overnight at 4 1C. Fibronectin was then removed and theplate was allowed to air dry at room temperature for B1 h beforethe addition of cells. All inhibitors were dissolved in dimethylsulfoxide therefore an equivalent volume of dimethyl sulfoxide wasused as a control.

Sox-2 Cy5H&E

AO

merge

VV

V

scr

tubulin

pAKT

DRR

AOØ scr

cell

num

ber

05

1015202530

AOscr

VV

V

scr

AO

18

60

55

Figure 5. Antisense oligonucleotide-mediated DRR ablation prevents invasion in a mouse model. (a) Western blot showing DRR and pAKT(Ser473) expression levels in human GBM stem cells untransfected (Ø), transfected with control non-targeting AO (scr) or with DRR targetingAO (AO). Decreased DRR expression level using AOs in hGSC cells correlates with decreased pAKT expression level. Tubulin is shown as aloading control. (b) Mouse brain sections stained with hematoxylin and eosin (H&E), human Sox-2 to identify all implanted cancer cells (red),Cy5 fluorescence revealing cells treated with AOs (green). Dashed line demarcates tumor core. Box in inset indicates tumor location.Arrowheads indicate cells that have escaped the implanted tumor mass and invaded the normal peritumoral brain. (c) Quantification ofinvasion. Represented as mean±s.e.m. n¼ 3, scr vs AO Po0.05 (t-test).


7


RGD assayPlates were coated with fibronectin. Cells were treated with 500 mM RGD insuspension on a rotating platform at room temperature for 30 min. A totalof 200 000 cells/well were plated onto fibronectin for 30 min. Cells that didnot adhere were removed and lysed.

ImmunocytochemistryCells were grown to B70% confluency on glass coverslips (with or withoutfibronectin) in supplemented Dulbecco’s Modified Eagle Medium. Cellswere fixed with 3% paraformaldehyde, then permeabilized with 0.5%Triton X-100 before being incubated with indicated primary and secondaryantibodies. Images were captured with the LSM 700 confocal microscopeusing a � 60 oil immersion objective. Where indicated, fluorescence wasquantified using ImageJ (http://rsbweb.nih.gov).

Three-dimension invasion assayInvasion assays were performed as previously described.21 Briefly, 25 000cells/drop were plated onto the lid of a 10 cm Petri dish and hung to allowspheroid formation. On the third day, spheroids were transferred to 2%agar. On the sixth day, spheroids were implanted into a collagen type Imatrix and appropriate media was added (regular Dulbecco’s ModifiedEagle Medium±1 mM 5-Fluorouracil or supplemented with 10mM PP2/LY294002/AG1478±1 mM 5-Fluorouracil). Spheroids were imaged over 24 hintervals with a Zeiss light microscope � 5 objective. Four spheroids/condition from three independent experiments were used to quantifyinvasion. Invading area was measured by calculating the extreme diameterat four different angles and by subtracting the extreme diameter of thespheroids at time zero.

Antisense oligonucleotidesAOs were designed to target DRR mRNA (NM_007177) in the open readingframe at position 576–595. Scrambled AOs are non-targeting negativecontrol sequences. AOs were synthesized on an ABI 3400 DNA synthesizerfrom Applied Biosystems (Burlington, ON, Canada). AOs were purified byreverse phase HPLC using a Waters 1525 HPLC and a Varian Pursuit 5 semi-preparative reverse phase C18 column. AOs were characterized by Liquidchromatography–mass spectrometry.

Animal workAll animal procedures were approved by the Institution’s Animal CareCommittee and performed according to the guidelines of the CanadianCouncil of Animal Care. Female CD1 athymic nude mice (Charles River,Canada) were anesthetized at 6 weeks of age using intraperitonealinjection containing Ketamine, Xylazine and Acepromazine. The mice wereplaced on a stereotaxic apparatus and a midline scalp incision was made.A burr-hole (3–5 mm) was created 2.2 mm lateral to the bregma using ahigh-powered drill. The injection needle containing 100 000 cellspretreated with the antisense (20 nM; for 72 h) was then lowered into theburr-hole to a depth of 3.0 mm to allow tumor implantation at the centerof the caudate nucleus. Animals (n¼ 18) were euthanized 3 weeks postimplantation, and brains were harvested following phosphate-bufferedsaline/formalin animal perfusion.

Formalin fixation, tissue processingHarvested brain specimens were placed in 10% neutral-buffered formalinfor 72 h at room temperature immediately following the killing of theanimal. Following formalin fixation, specimens were processed andparaffin embedded, and 5 mm tissue sections were prepared and mountedon a poly-L-lysine-coated glass slides.

Antigen retrieval and immunolabelingSamples were baked in a standard laboratory oven at 60 1C for 1 h. Antigenretrieval was done using citrate buffer (pH 6.0) and pressure cooking for10 min. The slides were blocked for 40 min, incubated for 1 h with anti-human Sox-2 primary antibody (R&D) and 20 min with secondary antibody.Imaging was performed with a LSM 700 confocal microscope using a � 20objective. Three mice/condition from three independent experiments wereused to quantify invasion. The number of cells that invaded beyond100mm of the epicenter of the tumor were counted.

Statistical analysisEach experiment was performed at least three times (n¼ 3) fromindependent biological replicates. Data were analyzed with the statisticalsoftware GraphPad Prism 5 and has been presented as mean±standarderror of mean (s.e.m). Po0.05 was considered statistically significant.

CONFLICT OF INTERESTThe authors declare no conflict of interest.

ACKNOWLEDGEMENTSThis work was supported in part by grants from Industry Canada, Centre of Excellencefor Commercialization (KP), Canadian Institutes for Health Research—Proof-of-Principle Grant (KP and MJD) and Canadian Institutes for Health Research—Operating Grant (MJD). AD was supported by a McGill University Faculty of MedicineScholarship. GT was supported by a McGill University Faculty of Medicine Scholarship.GFD was supported by a Vanier Canada Graduate Scholarship.

REFERENCES1 Villavicencio AT, Burneikiene S, Romanelli P, Fariselli L, McNeely L, Lipani JD et al.

Survival following stereotactic radiosurgery for newly diagnosed and recurrentglioblastoma multiforme: a multicenter experience. Neurosurg Rev 2009; 32:417–424.

2 Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Taphoorn MJ et al.Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma.N Engl J Med 2005; 352: 987–996.

3 Fan QW, Weiss WA. Targeting the RTK-PI3K-mTOR axis in malignant glioma:overcoming resistance. Curr Top Microbiol Immunol 2010; 347: 279–296.

4 Sarbassov DD, Guertin DA, Ali SM, Sabatini DM. Phosphorylation and regulation ofAkt/PKB by the rictor-mTOR complex. Science 2005; 307: 1098–1101.

5 Martelli AM, Faenza I, Billi AM, Manzoli L, Evangelisti C, Fala F et al. Intranuclear3’-phosphoinositide metabolism and Akt signaling: new mechanisms for tumor-igenesis and protection against apoptosis? Cell Signal 2006; 18: 1101–1107.

6 Dillon RL, White DE, Muller WJ. The phosphatidyl inositol 3-kinase signalingnetwork: implications for human breast cancer. Oncogene 2007; 26: 1338–1345.

7 Liu P, Cheng H, Roberts TM, Zhao JJ. Targeting the phosphoinositide 3-kinasepathway in cancer. Nat Rev Drug Discov 2009; 8: 627–644.

8 Hers I, Vincent EE, Tavare JM. Akt signalling in health and disease. Cell Signal 2011;23: 1515–1527.

9 Song G, Ouyang G, Bao S. The activation of Akt/PKB signaling pathway and cellsurvival. J Cell Mol Med 2005; 9: 59–71.

10 Manning BD, Cantley LC. AKT/PKB signaling: navigating downstream. Cell 2007;129: 1261–1274.

11 Chin YR, Toker A. Function of Akt/PKB signaling to cell motility, invasion and thetumor stroma in cancer. Cell Signal 2009; 21: 470–476.

12 Network T. Comprehensive genomic characterization defines human glio-blastoma genes and core pathways. Nature 2008; 455: 1061–1068.

13 Akhavan D, Cloughesy TF, Mischel PS. mTOR signaling in glioblastoma: lessonslearned from bench to bedside. Neuro Oncol 2010; 12: 882–889.

14 Wang H, Wang H, Zhang W, Huang HJ, Liao WS, Fuller GN. Analysis of the acti-vation status of Akt, NFkappaB, and Stat3 in human diffuse gliomas. Lab Invest2004; 84: 941–951.

15 Brandes AA, Franceschi E, Tosoni A, Hegi ME, Stupp R. Epidermal growth factorreceptor inhibitors in neuro-oncology: hopes and disappointments. Clin CancerRes 2008; 14: 957–960.

16 van den Bent MJ, Dubbink HJ, Sanson M, van der Lee-Haarloo CR, Hegi M, JeukenJW et al. MGMT promoter methylation is prognostic but not predictive for out-come to adjuvant PCV chemotherapy in anaplastic oligodendroglial tumors: areport from EORTC Brain Tumor Group Study 26951. J Clin Oncol 2009; 27:5881–5886.

17 Brown PD, Krishnan S, Sarkaria JN, Wu W, Jaeckle KA, Uhm JH et al. Phase I/II trialof erlotinib and temozolomide with radiation therapy in the treatment of newlydiagnosed glioblastoma multiforme: North Central Cancer Treatment GroupStudy N0177. J Clin Oncol 2008; 26: 5603–5609.

18 Andjelkovic M, Alessi DR, Meier R, Fernandez A, Lamb NJ, Frech M et al. Role oftranslocation in the activation and function of protein kinase B. J Biol Chem 1997;272: 31515–31524.

19 Ahmed NN, Franke TF, Bellacosa A, Datta K, Gonzalez-Portal ME, Taguchi T et al.The proteins encoded by c-akt and v-akt differ in post-translational modification,subcellular localization and oncogenic potential. Oncogene 1993; 8: 1957–1963.


8


http://rsbweb.nih.gov

20 Carpten JD, Faber AL, Horn C, Donoho GP, Briggs SL, Robbins CM et al. A trans-forming mutation in the pleckstrin homology domain of AKT1 in cancer. Nature2007; 448: 439–444.

21 Le PU, Angers-Loustau A, de Oliveira RM, Ajlan A, Brassard CL, Dudley A et al. DRRdrives brain cancer invasion by regulating cytoskeletal-focal adhesion dynamics.Oncogene 2010; 29: 4636–4647.

22 Asano Y, Kishida S, Mu P, Sakamoto K, Murohara T, Kadomatsu K. DRR1 isexpressed in the developing nervous system and downregulated during neuro-blastoma carcinogenesis. Biochem Biophys Res Commun 2010; 394: 829–835.

23 Liu Q, Zhao XY, Bai RZ, Liang SF, Nie CL, Yuan Z et al. Induction of tumor inhibitionand apoptosis by a candidate tumor suppressor gene DRR1 on 3p21.1. Oncol Rep2009; 22: 1069–1075.

24 Wang L, Darling J, Zhang JS, Liu W, Qian J, Bostwick D et al. Loss of expression ofthe DRR 1 gene at chromosomal segment 3p21.1 in renal cell carcinoma. GenesChromosomes Cancer 2000;; 27: 1–10.

25 Schmidt MV, Schulke JP, Liebl C, Stiess M, Avrabos C, Bock J et al. Tumor sup-pressor down-regulated in renal cell carcinoma 1 (DRR1) is a stress-induced actinbundling factor that modulates synaptic efficacy and cognition. Proc Natl Acad SciUSA. 2011; 108: 17213–17218.

26 Gan HK, Walker F, Burgess AW, Rigopoulos A, Scott AM, Johns TG. The epidermalgrowth factor receptor (EGFR) tyrosine kinase inhibitor AG1478 increases theformation of inactive untethered EGFR dimers. Implications for combinationtherapy with monoclonal antibody 806. J Biol Chem 2007; 282: 2840–2850.

27 Anderson NG, Ahmad T, Chan K, Dobson R, Bundred NJ. ZD1839 (Iressa), a novelepidermal growth factor receptor (EGFR) tyrosine kinase inhibitor, potently inhi-bits the growth of EGFR-positive cancer cell lines with or without erbB2 over-expression. Int J Cancer 2001; 94: 774–782.

28 Guarino M. Src signaling in cancer invasion. J Cell Physiol 2010; 223: 14–26.29 Higuchi M, Masuyama N, Fukui Y, Suzuki A, Gotoh Y. Akt mediates Rac/Cdc42-

regulated cell motility in growth factor-stimulated cells and in invasive PTENknockout cells. Curr Biol 2001; 11: 1958–1962.

30 Courtney KD, Corcoran RB, Engelman JA. The PI3K pathway as drug target inhuman cancer. J Clin Oncol 2010; 28: 1075–1083.

31 Hannigan G, Troussard AA, Dedhar S. Integrin-linked kinase: a cancer therapeutictarget unique among its ILK. Nat Rev Cancer 2005; 5: 51–63.

32 Rexer BN, Ghosh R, Arteaga CL. Inhibition of PI3K and MEK: it is all about com-binations and biomarkers. Clin Cancer Res 2009; 15: 4518–4520.

33 Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley DA, Feeser WS et al.Identification of a novel inhibitor of mitogen-activated protein kinase. J Biol Chem1998; 273: 18623–18632.

34 Wolfrum S, Dendorfer A, Rikitake Y, Stalker TJ, Gong Y, Scalia R et al. Inhibition ofRho-kinase leads to rapid activation of phosphatidylinositol 3-kinase/proteinkinase Akt and cardiovascular protection. Arterioscler Thromb Vasc Biol 2004; 24:1842–1847.

35 Genot EM, Arrieumerlou C, Ku G, Burgering BM, Weiss A, Kramer IM. The T-cellreceptor regulates Akt (protein kinase B) via a pathway involving Rac1 andphosphatidylinositide 3-kinase. Mol Cell Biol 2000; 20: 5469–5478.

36 Del Re DP, Miyamoto S, Brown JH. Focal adhesion kinase as a RhoA-activablesignaling scaffold mediating Akt activation and cardiomyocyte protection. J BiolChem 2008; 283: 35622–35629.

37 Legate KR, Montanez E, Kudlacek O, Fassler R. ILK, PINCH and parvin: the tIPP ofintegrin signalling. Nat Rev Mol Cell Biol 2006; 7: 20–31.

38 Parsons SJ, Parsons JT. Src family kinases, key regulators of signal transduction.Oncogene 2004; 23: 7906–7909.

39 Wang S, Basson MD. Protein kinase B/AKT and focal adhesion kinase: two closesignaling partners in cancer. Anticancer Agents Med Chem 2011; 11: 993–1002.

40 Guo W, Giancotti FG. Integrin signalling during tumor progression. Nat Rev MolCell Biol 2004; 5: 816–826.

41 Maret D, Gruzglin E, Sadr MS, Siu V, Shan W, Koch AW et al. Surface expression ofprecursor N-cadherin promotes tumor cell invasion. Neoplasia 2010; 12:1066–1080.

42 Bogdahn U, Hau P, Stockhammer G, Venkataramana NK, Mahapatra AK, Suri Aet al. Targeted therapy for high-grade glioma with the TGF-beta2 inhibitor tra-bedersen: results of a randomized and controlled phase IIb study. Neuro Oncol2011; 13: 132–142.

43 Lee J, Kotliarova S, Kotliarov Y, Li A, Su Q, Donin NM et al. Tumor stem cellsderived from glioblastomas cultured in bFGF and EGF more closely mirror thephenotype and genotype of primary tumors than do serum-cultured cell lines.Cancer Cell 2006; 9: 391–403.

44 Molina JR, Hayashi Y, Stephens C, Georgescu MM. Invasive glioblastoma cellsacquire stemness and increased Akt activation. Neoplasia 2010; 12: 453–463.

45 Barkovich KJ, Hariono S, Garske AL, Zhang J, Blair JA, Fan QW et al. Kinetics ofinhibitor cycling underlie therapeutic disparities between EGFR-driven lung andbrain cancers. Cancer Discov 2012; 2: 450–457.

46 Vivanco I, Robins HI, Rohle D, Campos C, Grommes C, Nghiemphu PL et al. Dif-ferential sensitivity of glioma- versus lung cancer-specific EGFR mutations to EGFRkinase inhibitors. Cancer Discov 2012; 2: 458–471.

47 Lu KV, Zhu S, Cvrljevic A, Huang TT, Sarkaria S, Ahkavan D et al. Fyn and SRC areeffectors of oncogenic epidermal growth factor receptor signaling in glio-blastoma patients. Cancer Res 2009; 69: 6889–6898.

48 Milano V, Piao Y, LaFortune T, de Groot J. Dasatinib-induced autophagy isenhanced in combination with temozolomide in glioma. Mol Cancer Ther 2009; 8:394–406.

49 Franceschi E, Stupp R, van den Bent MJ, van Herpen C, Laigle Donadey F,Gorlia T et al. EORTC 26083 phase I/II trial of dasatinib in combination withCCNU in patients with recurrent glioblastoma. Neuro Oncol 2012; 14:1503–1510.

50 Lu-Emerson C, Norden AD, Drappatz J, Quant EC, Beroukhim R, Ciampa AS et al.Retrospective study of dasatinib for recurrent glioblastoma after bevacizumabfailure. J Neurooncol 2011; 104: 287–291.

51 Stupp R, Hegi ME, Neyns B, Goldbrunner R, Schlegel U, Clement PM et al. Phase I/IIa study of cilengitide and temozolomide with concomitant radiotherapy fol-lowed by cilengitide and temozolomide maintenance therapy in patients withnewly diagnosed glioblastoma. J Clin Oncol 2010; 28: 2712–2718.

52 Petrecca K, Guiot MC, Panet-Raymond V, Souhami L. Failure pattern followingcomplete resection plus radiotherapy and temozolomide is at the resectionmargin in patients with glioblastoma. J Neurooncol 2012; 111: 19–23.

53 Kelly JJ, Stechishin O, Chojnacki A, Lun X, Sun B, Senger DL et al. Proliferation ofhuman glioblastoma stem cells occurs independently of exogenous mitogens.Stem Cells 2009; 27: 1722–1733.

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


9


http://www.nature.com/onc

DRR regulates AKT activation to drive brain cancer invasion

Documents

Transcript of DRR regulates AKT activation to drive brain cancer invasion