1

MicroRNA-146a Inhibits Glioma Development by Targeting Notch1 12

Running title: miR-146a restricts gliomagenesis 34

Jie Mei1, Robert Bachoo2, and Chun-Li Zhang1,*56

1Department of Molecular Biology, 2Department of Neurology, University of Texas 7Southwestern Medical Center, Dallas, TX 75390-9148 8

910

Keywords: miR-146a, Notch1, glioma, EGFR, PTEN, INK4A/ARF 111213

Word Count: Abstract (193), Materials and Methods (1,196), Introduction, Results and 14Discussion (2,911) 15

16171819

*To whom correspondence may be addressed. 2021

Chun-Li Zhang 22Department of Molecular Biology 23University of Texas Southwestern Medical Center 24Dallas, TX 75390-9148 25

26Tel: 214-648-1670 27Fax: 214-648-1488 28Email: chun-li.zhang@utsouthwestern.edu29

30

31

Copyright © 2011, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved.Mol. Cell. Biol. doi:10.1128/MCB.05821-11 MCB Accepts, published online ahead of print on 5 July 2011

2

ABSTRACT 32

Dysregulated EGFR signaling through either genomic amplification or dominant active 33

mutation (EGFRvIII), in combination with dual inactivation of INK4A/ARF and PTEN, is a 34

leading cause of gliomagenesis. Our global expression analysis for microRNAs 35

revealed that EGFR activation induces miR-146a expression, which is further 36

potentiated by inactivation of PTEN. Unexpectedly, over-expression of miR-146a 37

attenuates proliferation, migration and tumorigenic potential of Ink4a/Arf-/-Pten-/-EgfrvIII38

murine astrocytes. Its ectopic expression also inhibits glioma development of a human 39

glioblastoma cell line in an orthotopic xenograft model. Such inhibitory function of miR-40

146a on gliomas is largely through down-regulation of Notch1, which plays a key role in 41

neural stem cell maintenance and is a direct target of miR-146a. Accordingly, miR-146a 42

modulates neural stem cell proliferation and differentiation, and reduces the formation 43

and migration of glioma stem-like cells. Conversely, knockdown of miR-146a by 44

microRNA sponge upregulates Notch1 and promotes tumorigenesis of malignant 45

astrocytes. These findings indicate that, in response to oncogenic cues, miR-146a is 46

induced as a negative feedback mechanism to restrict tumor growth by repressing 47

Notch1. Our results provide novel insights into the signaling pathways that link neural 48

stem cells to gliomagenesis and may lead to new strategies treating brain tumors.49

50

INTRODUCTION51

Gliomas are the most frequently observed brain tumors, with glioblastoma multiforme 52

(GBM) being the most common and aggressive form in adults (35). Despite major 53

therapeutic improvements made by combining neurosurgery, chemotherapy and 54

3

radiotherapy, the prognosis and survival rate for patients with GBM is still extremely 55

poor (7). The deadly nature of GBM originates from explosive growth and invasive 56

behavior, which are fueled by dysregulation of multiple signaling pathways. EGFR 57

activation, in cooperation with loss of tumor suppressor functions, such as mutations in 58

Ink4a/Arf and Pten genes, constitutes a lesion signature for GBM (1). Such 59

dysregulated genetic pathways are sufficient to transform neural stem cells (NSCs) or 60

astrocytes into cancer stem-like cells. This gives rise to high-grade malignant gliomas 61

with a pathological phenotype resembling human GBM (5, 59). However, the 62

downstream events underlying these genetic dysregulations in gliomagenic cells have 63

not been fully elucidated. 64

65

MicroRNAs are 20-22 nucleotide non-coding RNA molecules that have emerged as key 66

players in controlling NSC self-renewal and differentiation (11, 57). Aberrant expression 67

of miRNAs, such as miR-21, miR-124 and miR-137, is linked to glioma formation (49). 68

miR-199b-5p and miR-34a impair cancer stem-like cells through negative regulation of 69

several components of the Notch pathway in brain tumors (18, 21). The Notch pathway 70

is an evolutionarily conserved signaling pathway that plays an important role in 71

neurogenesis (4, 10, 23). Upon binding to its ligand, Delta, the Notch intracellular 72

domain (NICD, the activated form of Notch) is released from the membrane by 73

presenilin/ -secretase-mediated cleavage and translocates to the nucleus. In the 74

nucleus, NICD forms a complex with Rbpj and activates the expression of several 75

transcriptional repressors, such as Hes1 and Hes5, which inhibit neurogenesis (15, 27, 76

29). Thus, activation of the Notch pathway is essential to maintain both developing and 77

4

adult NSCs (36). This property of the Notch pathway enables it to promote glioma 78

growth (30), and its inhibition by drugs could abolish glioma stem-like cells and reduce 79

tumorigenesis (17). 80

81

In this study, we show that miR-146a is specifically induced as a converging 82

downstream target of EGFR and PTEN signaling in immortalized Ink4a/Arf-/- astrocytes. 83

We further demonstrate that miR-146a acts as a native safeguarding mechanism to 84

restrict the formation of glioma stem-like cells and glioma growth by directly controlling 85

the expression of Notch1. 86

87

MATERIALS AND METHODS 88

Cell culture, MTT and anchorage-independent growth assays. We isolated primary 89

NSCs by mechanical dissociation using 1 ml pipettes from embryonic day 14.5 (E14.5) 90

mouse forebrains in growth medium consisting of DMEM/F12, 1 mM L-glutamine, N2 91

(Invitrogen), 20 ng/ml EGF and 20 ng/ml FGF2 (Peprotech). These cells were cultured 92

as free-floating neurospheres under 37°C and 5% CO2. For differentiation, we exposed 93

NSCs to DMEM/F12 media with N2 supplement, further supplemented with 5 μM 94

forskolin and 0.5% FBS (FSK), 1 μM retinoic acid and 0.5% (v/v) FBS (RA), or 0.5% (v/v) 95

FBS (-GF). This was done in 60-mm dishes or 8-well chamber slides coated with 96

laminin (5 μg/ml) and poly-L-ornithine (10 μg/ml). The Ink4a/Arf-/-, Ink4a/Arf-/-EGFRvIII97

and Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes, or human U87 glioma cells were cultured in 98

DMEM containing 10% FBS. Glioma stem-like cells from malignant astrocytes or U87 99

cells were enriched by culturing in NSC medium with growth factors. Cell growth was 100

5

determined by the CellTiter 96® nonradioactive cell proliferation assay kit (MTT assay, 101

Promega) according to the manufacturer’s instructions. For anchorage-independent 102

growth by soft agar assay, 2,000 Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes transduced with 103

either wild-type or seed-region-mutated miR-146a were plated and cultured in 12-well 104

plates, as described previously (47). Five weeks later, cell colonies in the plates were 105

stained with 0.5 ml of 0.005% crystal violet and counted under an inverted microscope. 106

We conducted each experiment in triplicate. 107

108

miRNA microarray and quantitative RT-PCR (qPCR). We extracted total RNAs from 109

Ink4a/Arf-/- (I), Ink4a/Arf-/-EGFRvIII (IE), or Ink4a/Arf-/-Pten-/-EGFRvIII (IPE) astrocytes 110

using the miReasy mini kit (Qiagen). Biological replicates were collected. RNA quality 111

was examined by Bioanalyzer (Illumina). Five μg of each RNA sample was processed 112

for labeling and hybridization to GeneChip miRNA arrays (Affymetrix) by the Microarray 113

Core Facility at UT Southwestern Medical Center. After scanning and normalization to 114

control probes, the array data was analyzed by VAMPIRE software to identify 115

statistically significant differences in gene expression between sample groups 116

(http://genome.ucsd.edu/microarray). For qPCR, 1 μg of total RNA was reverse 117

transcribed using NCode miRNA First-Strand Synthesis and qRT-PCR Kits (Invitrogen). 118

qPCR reactions were performed in a 384-well plate using an ABI 7900HT instrument. 119

The PCR program consisted of 2 min at 50°C and 10 min at 95°C, followed by 40 cycles 120

of 15s at 95°C and 60s at 58°C. Primer quality was analyzed by dissociation curves. 121

The expression of miR-146a and Notch1 was normalized to that of U6 and Hprt,122

respectively.123

6

124

Lentivirus production and transduction. A genomic fragment encompassing the miR-125

146a coding region or a seed-region-mutated version was cloned by PCR into pTomo 126

vector (provided by Dr. Inder Verma at Salk Institute). To construct the miRNA sponge, 127

we inserted the synthesized DNA fragment into the BamHI site of a pCSC-SP-PW-128

IRES/GFP lentiviral vector, and the empty vector as the control. We produced the 129

lentiviruses, determined their titer, and transduced cultured cells using previously 130

described methods (38, 48). Transduction efficiency was monitored by GFP expression. 131

All experiments were performed using stably transduced cells within 6 passages.132

133

Cell migration assay. Migration of malignant astrocytes in culture was determined by 134

the ‘scratch’ assay. For this, cells were seeded into a 6-well tissue culture dish and 135

allowed to grow to 90% confluency in complete medium with 10% FBS. Cells in 136

monolayers were scratched in a single straight line using a pipette tip (1 mm in 137

diameter). Wounded monolayers were washed three times with culture medium to 138

remove cell debris and then incubated for another 24h. Migratory distance was 139

measured under a microscope equipped with a camera. Transwell migration assay was 140

conducted essentially as previously described (43). Invasive behavior of glioma stem-141

like cells in vitro was examined by measuring migration ability of cultured neurospheres. 142

Neurospheres with similar diameter were selected and plated onto 6-well culture plates 143

coated with laminin (5 μg/ml) and poly-L-ornithine (10 μg/ml) and cultured for 48h in 144

NSC growth medium. Cell migration and spreading was quantified by measuring the 145

7

distance between the edge of the neurosphere and the periphery of radially migrating 146

cells. 147

148

Luciferase reporter assay. The 3´UTR of the Notch1 gene, which contains one 149

putative miR-146a targeting site, was amplified by PCR and inserted into the SacI and 150

PmeI sites of pMIR-REPORT vector (Ambion). To express miR-146a, a 610-bp genomic 151

fragment encompassing the coding region was cloned by PCR and inserted into the 152

HindIII and BamHI sites of pCMV6 vector. A seed-sequence-mutated version of miR-153

146a was used as a control for all of the experiments. To create this mutant (pCMV6-154

miR-146a-mt), we replaced the seed sequence TGAGAACT with GCGGCCGC through 155

site-directed mutagenesis using the QuickChange kit (Stratagene). Similarly, the binding 156

site (AGTTCT) for miR-146a within the 3’UTR of Notch 1 gene was replaced with 157

ACGCGT to generate a control for the luciferase reporter assays. HEK293 cells were 158

transiently transfected using Fugene 6 reagent (Roche) in 48-well plates. The following 159

plasmids were used: pMIR-REPORT with either pCMV6-miR-146a or its mutant, 160

pCMV6-miR-146a-mt. pCMV-LacZ was used as a control to monitor transfection 161

efficiency. Total plasmid amount (200 ng) was kept constant for each transfection with 162

empty pCMV6 vector. Forty-eight hours later, cells were lysed and assayed for 163

luciferase and -galactosidase activity using the Dual-Light system (Applied 164

Biosystems). Relative reporter activities were determined by normalizing luminescence 165

units to -galactosidase expression. 166

167

8

Western blotting and immunocytochemistry. Protein expression was examined by 168

Western blotting, according to a standard procedure. Antibodies against the following 169

proteins were used: -Actin (Sigma, 1:8000), GFAP and p-AKT-T308 (Cell Signaling 170

Technology, 1:1,000), Sox2 (Chemicon, 1:1,000), Nestin (Aves, 1:10,000), Notch1 171

(Santa Cruz, 1:1,000). To partially block processing of Notch1 protein by –secretase,172

cells were treated with 10 M DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-173

phenylglycine t-butyl ester) (Sigma) for 36 h. Blots were scanned and quantified by NIH 174

software ImageJ. To examine proliferation and cell cycle exit, cultured cells were pulse-175

labeled with BrdU (10 μM) for 2 h and then fixed and processed for immunostaining, as 176

previously described (56). Antibodies against the following proteins were used: BrdU 177

(BD pharmingen; 1:1,000), Ki67 (Novocastra; 1:500), cleaved caspase 3 (Cell Signaling 178

Technology; 1:300) and Tuj1 (Covance; 1:500).179

180

Intracranial and subcutaneous cell transplantation model for tumorigenesis.181

NOD/SCID and nude mice (nu/nu) were purchased from Harlan Laboratories. They 182

were housed in a pathogen-free facility under standard 12h light/dark cycles, controlled 183

temperature conditions and had free access to food and water. Experimental protocols 184

were approved by the Institutional Animal Care and Use Committee (IACUC) at UT 185

Southwestern Medical Center. Astrocytes (5X105 cells) or neurospheres (2X105 cells) 186

were respectively implanted into the right caudoputamen of the brain (for orthotopic 187

transplantation) or subcutaneous tissues of 5 to 6 weeks old NOD/SCID or nude mice 188

using a 25 gauge-needle. The injection sites were monitored frequently. Mice 189

transplanted with neurospheres were euthanized 3 weeks later. Tumors were finely 190

9

dissected out, imaged, and measured according to the following formula: Volume of 191

tumor (mm3) = d2×D/2, where d and D are the shortest and the longest diameters, 192

respectively. For survival analysis, the mice were euthanized when they demonstrated 193

severe health deterioration according to IACUC guidelines. 194

195

Statistical analysis. Data are expressed as mean ± SD. Except for the survival studies, 196

all data were analyzed by two-tailed Student t tests. The Kaplan-Meier survival curves 197

were determined by using Graphpad Prism v5.0 (Graphpad Software, Inc.) and the Log-198

rank test. A p value of < 0.05 was considered significant. 199

200

RESULTS201

EGFR activation and PTEN inactivation synergistically induce expression of miR-202

146a. Loss of Ink4a/Arf results in immortal growth of primary murine astrocytes. 203

Subsequent activation of EGFR signaling through expression of a constitutively active 204

EGFRvIII mutant transforms these astrocytes into tumorigenic and neurosphere-forming 205

glioma stem-like cells when cultured in defined medium for NSCs (5, 24). In vivo,206

concomitant activation of EGFR and inactivation of Ink4a/Arf and Pten produce rapid-207

onset and fully penetrant high-grade gliomas that resemble GBM in humans (60), 208

suggesting cooperative actions of these signaling pathways on cell proliferation. To 209

confirm this, we performed MTT assays and direct cell counts to assess the growth rate 210

of Ink4a/Arf-/-, Ink4a/Arf-/-EGFRvIII, or Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes. Indeed, 211

constitutive activation of EGFR confers Ink4a/Arf-/- astrocytes with a growth advantage 212

that is further enhanced by ablation of Pten (Fig. 1A and B). 213

10

214

To understand the downstream molecular events of this genetic cooperation and 215

because of the emerging role of microRNA in cancers, we performed global expression 216

analysis of microRNAs in Ink4a/Arf-/-, Ink4a/Arf-/-EGFRvIII or Ink4a/Arf-/-Pten-/-EGFRvIII217

astrocytes. We identified a total of 19 unique miRNAs whose expression is altered by 218

activation of EGFR alone or in combination with loss of Pten (Fig. 1C). Among these 219

miRNAs, miR-146a, miR-182, miR-183 and miR-199a-3p are shown to be enriched in 220

diverse cancers, whereas miR-127 and miR-140 are down-regulated in gliomas (32, 46). 221

In contrast to unchanged expression of miR-146b, miR-146a was significantly induced 222

by EGFRvIII alone or in combination with ablation of Pten (Fig. 1C). The change in 223

expression was subsequently confirmed by qPCR using independent RNA samples (Fig. 224

1D). We also examined the expression of miR-146a after acute deletion of Pten through 225

Cre-mediated recombination of floxed Pten alleles in immortalized astrocytes (Fig. 1E 226

and F). Pten loss resulted in more than a 7-fold induction of miR-146a expression. This 227

induction was further increased to 9-fold after activation of EGFR. These data indicate a 228

cooperative nature of EGFR, PTEN and INK4A/ARF signaling on cell proliferation and 229

miR-146a expression. 230

231

miR-146a inhibits proliferation and oncogenic potential of malignant astrocytes. 232

The induction of miR-146a by EGFR and/or PTEN signaling in Ink4a/Arf-/- astrocytes 233

suggests that miR-146a may act as an onco-miR by transducing oncogenic signals to 234

control cell behavior. To test this hypothesis, we performed overexpression experiments 235

to examine whether exogenous miR-146a could further enhance proliferation and 236

11

tumorigenesis of malignant astrocytes. Due to extremely low efficiency of transient 237

transfections in these astrocytes, we employed a lentivirus-mediated expression system, 238

in which the CMV promoter drives expression of both miR-146a and GFP. As a control, 239

we mutated eight nucleotides within the seed region of miR-146a (Ctrl, Fig. 2A). Such a 240

mutation presumably abolishes the interactions of miR-146a with its targets. 241

Examination of GFP expression showed that nearly 100% of the cells were transduced 242

by lentivirus. qPCR analysis demonstrated a 26-fold increase of wild-type miR-146a in 243

Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes after lentiviral transduction (Fig. 2B). Since miR-244

146a had no effect on apoptosis (Fig. 2E), we examined cell growth by MTT assays and 245

direct cell counting. Interestingly, overexpression of miR-146a inhibited proliferation of 246

Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes (Fig. 2C-D), indicating that this microRNA may 247

instead function as a tumor suppressor. Indeed, overexpression of miR-146a resulted in 248

a 25% reduction in the diameter (Fig. 2F) and a 35% decrease in the number (Fig. 2G) 249

of colonies in an anchorage-independent growth assay, which is an in vitro model for 250

cellular transformation. 251

252

High-grade gliomas exhibit aggressive behavior, which is manifested by rapid cellular 253

migration under culture conditions in vitro (12, 50). Using a standard ‘scratch’ assay 254

after the cells become confluent in culture dishes, we found that enhanced expression 255

of miR-146a significantly reduced the rate of migration of Ink4a/Arf-/-Pten-/-EGFRvIII256

astrocytes when compared to the same cells transduced with a control lentivirus (Fig. 257

2H). This observation was further confirmed by a Matrigel transwell migration assay, 258

which showed a 50% reduction of migrating malignant astrocytes upon ectopic 259

12

expression of miR-146a (Fig. 2I). Finally, we examined the in vivo function of miR-146a 260

during glioma development after cell transplantation into the right caudoputamen of 261

NOD/SCID mice. For this, we grafted an equal number (5X105) of Ink4a/Arf-/-Pten-/-262

EGFRvIII astrocytes that were transduced with lentiviruses expressing either wild-type 263

miR-146a or seed-region-mutated miR-146a (Ctrl). Mice were monitored daily for 264

morbidity. A subset of mice was also sacrificed around 3 weeks post transplantation to 265

examine tumor burden by histology (Fig. 3A). Ectopic expression of miR-146a but not 266

the Ctrl significantly reduced the tumor burden and prolonged the survival of mice that 267

were transplanted with Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes (Fig. 3B). Together, these 268

data suggest that miR-146a acts as a tumor suppressor of glioma development. 269

270

miR-146a inhibits formation of glioma stem-like cells from malignant astrocytes. 271

Accumulating evidence suggests that self-renewable stem-like cells within the bulk of 272

brain tumors are the driving force for initiation and maintenance of aggressive gliomas 273

(13). These cells share common features with NSCs, including the ability to form 274

neurospheres and the expression of stem cell markers, such as Sox2 and Nestin (5, 21). 275

In fact, neurosphere formation is routinely used to enrich glioma stem-like cells from 276

primary human brain tumors (31, 41). The finding that miR-146a inhibits tumorigenesis 277

raised that possibility that it may negatively control the behavior of glioma stem-like cells. 278

We enriched these cells from Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes by culturing them 279

under serum-free conditions in the presence of 20 ng EGF and 20 ng FGF. After 7 days 280

in culture, control-virus-transduced cells readily formed large free-floating spheres, 281

robustly expressing Nestin and Sox2 (Fig. 4A-D). Additionally, these neurospheres were 282

13

able to differentiate into GFAP-positive astrocytes and Tuj1-positive neurons in 283

response to 1% fetal bovine serum and 5 μM forskolin treatment, respectively (Fig. 4C). 284

In sharp contrast, ectopic expression of miR-146a markedly diminished the ability of 285

Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes to form neurospheres, as indicated by a 40% 286

reduction of sphere size and more than a 30% decline in sphere number (Fig. 4A). miR-287

146a also impaired the self-renewal ability of these neurospheres, demonstrated by a 288

significant reduction of secondary and tertiary sphere formation upon serial passages 289

(Fig. 4B). Furthermore, exogenous miR-146a led to a considerable decrease in the 290

expression of Nestin and Sox2. This was accompanied by a sharp increase in the level 291

of GFAP expression, indicating enhanced glial differentiation (Fig. 4D). The invasive 292

behavior of glioma stem-like cells is evidenced by rapid cellular migration from 293

neurospheres when attached to the coated plates in culture. We selected neurospheres 294

with similar diameters from either control or miR-146a virus-transduced malignant 295

astrocytes and plated them onto laminin- and polyornithine-coated chamber slides. 296

Forty-eight hours later, control-virus-transduced cells rapidly migrated from the edge of 297

the attached spheres and spread out onto the coated culture surfaces. In contrast, 298

ectopic expression of miR-146a caused a 38% reduction of migratory distance from the 299

edge of the spheres (Fig. 4E). These data indicate that miR-146a controls not only the 300

number of glioma stem-like cells but also their invasive behavior in vitro.301

302

miR-146a targets Notch1. How does miR-146a impose an inhibitory function on glioma 303

stem-like cells and tumorigenesis? Using the Targetscan bioinformatics algorithm, we 304

searched for direct targets of miR-146a with an emphasis on those genes that were 305

14

shown to play a role in cancer stem cells, tumorigenesis or NSCs. Among several 306

potential candidates, Notch1 is most prominent due to its established function in 307

maintaining NSCs (23, 27) (Fig. 5A). Notch1 also interacts with the EGFR and AKT 308

pathways to positively regulate the proliferation of glioma stem-like cells and 309

tumorigenesis (54). To confirm the regulation of Notch1 by miR-146a, we first performed 310

a luciferase reporter assay by linking the 3’UTR of Notch1 to the firefly luciferase gene. 311

Luciferase activity in COS7 cells was significantly reduced (by 34-57%) when the 312

reporter was cotransfected with increasing amount of plasmids expressing wild-type 313

miR-146a. Such reduction is specific since a control plasmid expressing seed-region-314

mutated miR-146a (Ctrl) has little effect on reporter activity. Moreover, wild type miR-315

146a did not change the activity of a luciferase reporter when the binding site for miR-316

146a in 3’UTR of Notch1 was mutated (Fig. 5B). Western blot analysis further showed 317

that ectopic miR-146a in malignant astrocytes induced a dramatic reduction of Notch1 318

protein, especially the processed intracellular NICD, which is the most predominant 319

form in these glia cells (Fig. 5C). When these cells were treated with DAPT, an inhibitor 320

for -secretase, to partially block Notch1 processing, miR-146a significantly reduced the 321

appearance of both full-length Notch1 and NICD (Fig. 5D). The inhibitory role of miR-322

146a on Notch1 is mainly through posttranscriptional control since it does not change 323

the mRNA level of Notch1 (Fig. 5D). Furthermore, phosphorylated AKT, a known 324

downstream target of Notch1 signaling (22, 58), was also markedly reduced (Fig. 5C). It 325

is known that individual microRNAs can target numerous mRNAs (6), raising the 326

possibility that Notch1 may not be the major target of miR-146a in glioma stem-like cells. 327

We examined this possibility by performing a rescue experiment. Interestingly, ectopic 328

15

expression of NICD completely reversed the inhibitory effect of miR-146a on the 329

formation of glioma stem-like cells. This was indicated by a respective 50% and 60% 330

increase in the number and size of neurospheres when compared to miR-146a-331

expressing cells (Fig. 5F). Importantly, overexpressing NICD did not change the level of 332

miR-146a, suggesting that the rescue was not due to down-regulation of miR-146a 333

expression (data not shown). 334

335

miR-146a regulates NSC proliferation and differentiation. Notch signaling plays an 336

essential role in maintaining NSCs (2, 23). Its down-regulation promotes neurogenesis 337

during development or in the adult stage (40, 55). Our finding that miR-146a targets 338

Notch1 expression suggests that miR-146a may regulate NSC behavior. Indeed, qPCR 339

analysis showed over 12-fold induction of miR-146a expression after 4 days of culturing 340

primary mouse E14.5 NSCs under differentiation conditions (Fig. 6A). Ectopic 341

expression of miR-146a through lentiviral transduction enhanced neuronal 342

differentiation, indicated by a 5-fold increase of Tuj1+ cells compared to control-virus-343

transduced NSCs under low EGF and FGF concentrations (1 ng/ml) (Fig. 6B and C). 344

Exogenous miR-146a also inhibited NSC proliferation, demonstrated by a 50% 345

reduction of either BrdU+ or Ki67+ cells (Fig. 6D and E). Such reduction of proliferation 346

was accompanied by increased cell cycle exit, which was shown by a higher ratio of 347

BrdU+Ki67- cells over the total number of BrdU+ cells after 2 hr of BrdU pulse (Fig. 6F). 348

In contrast, miR-146a had no effect on apoptosis, since an equal number of active 349

caspase 3-positive cells were observed (Fig. 6D). Together, these results indicate that 350

miR-146a potentiates neuronal differentiation of NSCs by promoting cell cycle exit. 351

16

352

miR-146a prolongs the survival of mice bearing human glioblastoma cells. We 353

next examined whether the biological role of miR-146a is evolutionarily conserved in 354

humans. Similar to murine Notch1 gene, the 3’UTR of its human ortholog also harbors a 355

potential binding site for miR-146a, albeit not a perfect match (Fig. 7A). Western blotting 356

analysis showed that ectopic miR-146a markedly reduced the level of both full-length 357

and the active form (NICD) of human Notch1 protein in U87 glioblastoma cells (Fig. 7B). 358

miR-146a also diminished the ability of U87 cells to form glioma stem-like cells. This 359

was indicated by a more than 50% decline in sphere number, a 31% reduction of 360

sphere size (Fig. 7C), and a near 40% decrease on the appearance of secondary and 361

tertiary spheres (data not shown). When U87 glioblastoma cells were transplanted into 362

the caudoputamen of NOD/SCID mice, miR-146a reduced tumor burden and 363

significantly extended the survival of tumor-bearing mice (Fig. 7D-E). Together, these 364

data suggest that miR-146a function is evolutionarily conserved from mice to human 365

and is able to modulate the behavior of glioma cells both in vitro and in vivo.366

367

Down-regulation of miR-146a function enhances tumorigenesis. miR-146a is 368

induced by oncogenic EGFR and PTEN signaling (Fig. 1). However, its overexpression 369

inhibits the behavior of glioma stem-like cells and tumorigenesis by targeting Notch1. 370

These seemingly paradoxical data suggest that miR-146a may act as a feedback 371

mechanism to restrict tumorigenesis (see below). To vigorously test this hypothesis, we 372

used an ‘miRNA sponge' (14) to knock down miR-146a function in Ink4a/Arf-/-Pten-/-373

EGFRvIII astrocytes. This sponge was designed to have an imperfect binding site near 374

17

the miR-146a seed region – with a bulge from positions 10 to 13 (Fig. 8A). This design 375

is presumed to be more effective and stable for inhibiting miR-146a function. We first 376

evaluated its efficacy by using a Notch1 3’UTR-containing luciferase reporter and found 377

that the sponge completely reversed the inhibitory effect of miR-146a on this reporter 378

(Fig. 8B). We also examined this miRNA sponge on the expression of Notch1 through 379

Western blotting. As expected, transduction of Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes with 380

sponge-expressing lentiviruses totally rescued Notch1 protein level that was inhibited by 381

ectopic miR-146a (Fig. 8C). Moreover, this sponge also enhanced the expression of 382

endogenous Notch1, most likely by releasing the inhibition of miR-146a that is induced 383

by EGFRvIII and Pten-loss (Fig. 8D). These data indicate that the use of a miRNA 384

sponge is an effective and specific way to down-regulate miR-146a function. We next 385

examined the impact of knocking down miR-146a on the behavior of glioma stem-like 386

cells and tumorigenesis. As expected if miR-146a has an inhibitory role, its down-387

regulation promoted efficient formation of glioma stem-like cells from Ink4a/Arf-/-Pten-/-388

EGFRvIII astrocytes by significantly increasing the number and the size of neurospheres 389

(Fig. 8E and F). When neurospheres (2X105 cells) from lentivirus-transduced malignant 390

astrocytes were grafted into the flanks of female nude mice, knocking-down miR-146a 391

function using a miRNA sponge caused more than an 80% increase in tumor size 28 392

days post transplantation (Fig. 8G). Therefore, these results indicate that miR-146a is 393

induced as a feedback mechanism to restrict the oncogenic potential of EGFR and 394

PTEN signaling. 395

396

DISCUSSION 397

18

Here, we reveal that constitutively active EGFRvIII cooperates with the loss of Pten to 398

synergistically induce expression of miR-146a in Ink4a/Arf-/- astrocytes. 399

Counterintuitively, upregulation of miR-146a inhibits tumor growth and the formation and 400

migration of glioma stem-like cells by both malignant murine Ink4a/Arf-/-Pten-/-EGFRvIII401

astrocytes and human glioblastoma cells. We further show for the first time that miR-402

146a directly downregulates Notch1 and potentiates differentiation of normal NSCs. 403

Knocking down miR-146a function enhances glioma stem-like cell formation and 404

exacerbates tumor burden. These data suggest that miR-146a integrates oncogenic 405

cues to restrict tumor development as a feedback mechanism (Fig. 8H). 406

407

Previously, miR-146a was shown to be induced by endotoxin (lipopolysaccharide) 408

through two consensus NF- B binding sites in the promoter region (9). This region is 409

highly homologous between human and mouse, suggesting an evolutionarily conserved 410

regulatory mechanism for controlling miR-146a expression. NF- B is constitutively 411

activated in glioma and many other cancer cells, in which it promotes survival and 412

metastatic potential of these cells as well as tumorigenesis (28, 52). One of the key 413

pathways that controls NF- B activity in gliomas is the PI3K pathway (16, 39), which is a 414

converging point of EGFR- and PTEN-dependent signal transduction. Activation of 415

receptor tyrosine kinase EGFR leads to recruitment and activation of PI-3 kinase, which 416

phosphorylates phosphoinositol lipids to generate phosphatidylinositol-3-phosphates. 417

These lipids in turn, recruit and activate AKT in the plasma membrane through PDK1. 418

The function of PI3K is antagonized by PTEN, a lipid phosphatase that 419

dephosphorylates phosphatidylinositol 3,4,5-trisphosphate. Activated AKT further 420

19

phosphorylates IKK, thus promoting NF- B activation. The synergistic induction of miR-421

146a by constitutively active EGFRvIII and inactivation of Pten in Ink4a/Arf-/- astrocytes 422

may reflect a convergence of these two signaling pathways on NF- B activity. Activated 423

AKT also promotes Notch1 expression, which subsequently modulates transcription of 424

Egfr through p53 (44, 45). These data indicate that miR-146a belongs to an integrated 425

genetic circuit consisting of the PTEN, EGFR, NF- B and Notch pathways. The results 426

from our current study reveal that miR-146a regulates the activity of this circuit by 427

targeting Notch1 expression. 428

429

Recent studies indicate that certain miRNAs (such as miR-124, miR-137, miR-128, miR-430

7) function as tumor suppressors (19, 49). These miRNAs are rarely expressed in 431

gliomas; however, their overexpression restrains proliferation and self-renewal of glioma 432

stem-like cells by promoting neural differentiation (20, 21). miR-146a is unique in that it 433

is significantly enriched in human tissues of skin (melanoma), cervical, breast, pancreas 434

and prostate cancers, compared to the same non-cancerous tissues (42, 51, 53). 435

Similarly, miR-146a was also up-regulated in human glioblastoma tissues and in both 436

human and mouse primary glioma cell lines (32). In this regard, increased expression of 437

miR-146a can be viewed as a biomarker for cancers. Unexpectedly, our study reveals 438

that, instead of promoting gliomagenesis through its upregulation, miR-146a rather 439

plays an inhibitory role in restricting the formation of glioma stem-like cells and tumor 440

burden. This result is consistent with a demonstrated role of miR-146a in other cancers, 441

such as pancreatic and breast cancers, where it inhibits cancer progression and 442

invasion (8, 26, 33). A most recent report also showed a positive correlation of miR-443

20

146a expression with the survival time of gastric cancer patients (25). Furthermore, 444

overexpression of miR-146a inhibits proliferation and survival of breast, prostate and 445

pancreatic cancer cells through down-regulation of other targets in these cells, including 446

ROCK1, EGFR, and MTA-2 (8, 33, 34). Our study adds Notch1 as a major target of 447

miR-146a in glioma cells. Supporting these cell culture models of tumorigenesis, it was 448

recently reported that mice with a deletion of miR-146a spontaneously developed 449

subcutaneous flank tumors (37). This data clearly indicates that miR-146a serves as a 450

native molecular brake for oncogenesis (3). 451

452

In summary, our current results and other emerging data indicate that miR-146a 453

constitutes an endogenous feedback system to counteract the oncogenic potential of 454

dysregulated signaling pathways, such as activation of EGFR and inactivation of Pten in 455

gliomas. By regulating multiple targets, including key neural stem cell factor Notch1, a 456

miR-146a-mediated innate regulatory mechanism provides an opportunity to devise 457

novel therapeutic strategies against aggressive and deadly brain tumors. 458

459

Acknowledgements. We thank the Microarray Core Facility, Yuhua Zou and Wenze 460

Niu for technical help, Rhonda Bassel-Duby and Eric Olson for discussions, Hiroyoshi 461

Tanda for careful reading of the manuscript, Pamela Jackson for administrative 462

assistance, and Dr. Inder Verma (Salk Institute) for providing pTomo lentiviral vector.463

C-L.Z. is a W. W. Caruth, Jr. Scholar in Biomedical Research. This work was supported 464

by the Whitehall Foundation (2009-12-05), the Welch Foundation (I-1724-01), the NIH 465

21

(1DP2OD006484 and R01NS070981) and a Startup Fund from UT Southwestern 466

Medical Center to C-L.Z. 467

468

REFERENCES 469

1. 2008. Comprehensive genomic characterization defines human glioblastoma 470

genes and core pathways. Nature 455:1061-8.471

2. Ables, J. L., N. A. Decarolis, M. A. Johnson, P. D. Rivera, Z. Gao, D. C. 472

Cooper, F. Radtke, J. Hsieh, and A. J. Eisch. 2010. Notch1 is required for 473

maintenance of the reservoir of adult hippocampal stem cells. J Neurosci 474

30:10484-92.475

3. Ameres, S. L., and R. Fukunaga. 2010. Riding in silence: a little snowboarding, 476

a lot of small RNAs. Silence 1:8.477

4. Androutsellis-Theotokis, A., R. R. Leker, F. Soldner, D. J. Hoeppner, R. 478

Ravin, S. W. Poser, M. A. Rueger, S. K. Bae, R. Kittappa, and R. D. McKay.479

2006. Notch signalling regulates stem cell numbers in vitro and in vivo. Nature 480

442:823-6.481

5. Bachoo, R. M., E. A. Maher, K. L. Ligon, N. E. Sharpless, S. S. Chan, M. J. 482

You, Y. Tang, J. DeFrances, E. Stover, R. Weissleder, D. H. Rowitch, D. N. 483

Louis, and R. A. DePinho. 2002. Epidermal growth factor receptor and Ink4a/Arf: 484

convergent mechanisms governing terminal differentiation and transformation 485

along the neural stem cell to astrocyte axis. Cancer Cell 1:269-77.486

6. Bartel, D. P. 2009. MicroRNAs: target recognition and regulatory functions. Cell 487

136:215-33.488

22

7. Behin, A., K. Hoang-Xuan, A. F. Carpentier, and J. Y. Delattre. 2003. Primary 489

brain tumours in adults. Lancet 361:323-31.490

8. Bhaumik, D., G. K. Scott, S. Schokrpur, C. K. Patil, J. Campisi, and C. C. 491

Benz. 2008. Expression of microRNA-146 suppresses NF-kappaB activity with 492

reduction of metastatic potential in breast cancer cells. Oncogene 27:5643-7.493

9. Cameron, J. E., Q. Yin, C. Fewell, M. Lacey, J. McBride, X. Wang, Z. Lin, B. C. 494

Schaefer, and E. K. Flemington. 2008. Epstein-Barr virus latent membrane 495

protein 1 induces cellular MicroRNA miR-146a, a modulator of lymphocyte 496

signaling pathways. J Virol 82:1946-58.497

10. Chambers, C. B., Y. Peng, H. Nguyen, N. Gaiano, G. Fishell, and J. S. Nye.498

2001. Spatiotemporal selectivity of response to Notch1 signals in mammalian 499

forebrain precursors. Development 128:689-702.500

11. Cheng, L. C., E. Pastrana, M. Tavazoie, and F. Doetsch. 2009. miR-124 501

regulates adult neurogenesis in the subventricular zone stem cell niche. Nat 502

Neurosci 12:399-408.503

12. Chuang, Y. Y., N. L. Tran, N. Rusk, M. Nakada, M. E. Berens, and M. Symons.504

2004. Role of synaptojanin 2 in glioma cell migration and invasion. Cancer Res 505

64:8271-5.506

13. Das, S., M. Srikanth, and J. A. Kessler. 2008. Cancer stem cells and glioma. 507

Nat Clin Pract Neurol 4:427-35.508

14. Ebert, M. S., J. R. Neilson, and P. A. Sharp. 2007. MicroRNA sponges: 509

competitive inhibitors of small RNAs in mammalian cells. Nat Methods 4:721-6.510

23

15. Ehm, O., C. Goritz, M. Covic, I. Schaffner, T. J. Schwarz, E. Karaca, B. 511

Kempkes, E. Kremmer, F. W. Pfrieger, L. Espinosa, A. Bigas, C. Giachino, V. 512

Taylor, J. Frisen, and D. C. Lie. 2010. RBPJkappa-dependent signaling is 513

essential for long-term maintenance of neural stem cells in the adult 514

hippocampus. J Neurosci 30:13794-807.515

16. Fan, X., Y. Aalto, S. G. Sanko, S. Knuutila, D. Klatzmann, and J. S. 516

Castresana. 2002. Genetic profile, PTEN mutation and therapeutic role of PTEN 517

in glioblastomas. Int J Oncol 21:1141-50. 518

17. Fan, X., L. Khaki, T. S. Zhu, M. E. Soules, C. E. Talsma, N. Gul, C. Koh, J. 519

Zhang, Y. M. Li, J. Maciaczyk, G. Nikkhah, F. Dimeco, S. Piccirillo, A. L. 520

Vescovi, and C. G. Eberhart. 2010. NOTCH pathway blockade depletes 521

CD133-positive glioblastoma cells and inhibits growth of tumor neurospheres and 522

xenografts. Stem Cells 28:5-16.523

18. Garzia, L., I. Andolfo, E. Cusanelli, N. Marino, G. Petrosino, D. De Martino, V. 524

Esposito, A. Galeone, L. Navas, S. Esposito, S. Gargiulo, S. Fattet, V. 525

Donofrio, G. Cinalli, A. Brunetti, L. D. Vecchio, P. A. Northcott, O. Delattre, 526

M. D. Taylor, A. Iolascon, and M. Zollo. 2009. MicroRNA-199b-5p impairs 527

cancer stem cells through negative regulation of HES1 in medulloblastoma. 528

PLoS One 4:e4998. 529

19. Godlewski, J., H. B. Newton, E. A. Chiocca, and S. E. Lawler. 2010. 530

MicroRNAs and glioblastoma; the stem cell connection. Cell Death Differ 17:221-531

8.532

24

20. Godlewski, J., M. O. Nowicki, A. Bronisz, S. Williams, A. Otsuki, G. Nuovo, 533

A. Raychaudhury, H. B. Newton, E. A. Chiocca, and S. Lawler. 2008. 534

Targeting of the Bmi-1 oncogene/stem cell renewal factor by microRNA-128 535

inhibits glioma proliferation and self-renewal. Cancer Res 68:9125-30.536

21. Guessous, F., Y. Zhang, A. Kofman, A. Catania, Y. Li, D. Schiff, B. Purow, 537

and R. Abounader. 2010. microRNA-34a is tumor suppressive in brain tumors 538

and glioma stem cells. Cell Cycle 9.539

22. Guo, D., J. Ye, J. Dai, L. Li, F. Chen, D. Ma, and C. Ji. 2009. Notch-1 regulates 540

Akt signaling pathway and the expression of cell cycle regulatory proteins cyclin 541

D1, CDK2 and p21 in T-ALL cell lines. Leuk Res 33:678-85.542

23. Hitoshi, S., T. Alexson, V. Tropepe, D. Donoviel, A. J. Elia, J. S. Nye, R. A. 543

Conlon, T. W. Mak, A. Bernstein, and D. van der Kooy. 2002. Notch pathway 544

molecules are essential for the maintenance, but not the generation, of 545

mammalian neural stem cells. Genes Dev 16:846-58.546

24. Holland, E. C., W. P. Hively, R. A. DePinho, and H. E. Varmus. 1998. A 547

constitutively active epidermal growth factor receptor cooperates with disruption 548

of G1 cell-cycle arrest pathways to induce glioma-like lesions in mice. Genes Dev 549

12:3675-85.550

25. Hou, Z., L. Xie, L. Yu, X. Qian, and B. Liu. 2011. MicroRNA-146a is down-551

regulated in gastric cancer and regulates cell proliferation and apoptosis. Med 552

Oncol.553

25

26. Hurst, D. R., M. D. Edmonds, G. K. Scott, C. C. Benz, K. S. Vaidya, and D. R. 554

Welch. 2009. Breast cancer metastasis suppressor 1 up-regulates miR-146, 555

which suppresses breast cancer metastasis. Cancer Res 69:1279-83.556

27. Imayoshi, I., M. Sakamoto, M. Yamaguchi, K. Mori, and R. Kageyama. 2010. 557

Essential roles of Notch signaling in maintenance of neural stem cells in 558

developing and adult brains. J Neurosci 30:3489-98.559

28. Inoue, J., J. Gohda, T. Akiyama, and K. Semba. 2007. NF-kappaB activation in 560

development and progression of cancer. Cancer Sci 98:268-74.561

29. Kageyama, R., T. Ohtsuka, H. Shimojo, and I. Imayoshi. 2008. Dynamic Notch 562

signaling in neural progenitor cells and a revised view of lateral inhibition. Nat 563

Neurosci 11:1247-51.564

30. Kanamori, M., T. Kawaguchi, J. M. Nigro, B. G. Feuerstein, M. S. Berger, L. 565

Miele, and R. O. Pieper. 2007. Contribution of Notch signaling activation to 566

human glioblastoma multiforme. J Neurosurg 106:417-27.567

31. Laks, D. R., M. Masterman-Smith, K. Visnyei, B. Angenieux, N. M. Orozco, I. 568

Foran, W. H. Yong, H. V. Vinters, L. M. Liau, J. A. Lazareff, P. S. Mischel, T. 569

F. Cloughesy, S. Horvath, and H. I. Kornblum. 2009. Neurosphere formation is 570

an independent predictor of clinical outcome in malignant glioma. Stem Cells 571

27:980-7.572

32. Lavon, I., D. Zrihan, A. Granit, O. Einstein, N. Fainstein, M. A. Cohen, B. 573

Zelikovitch, Y. Shoshan, S. Spektor, B. E. Reubinoff, Y. Felig, O. Gerlitz, T. 574

Ben-Hur, Y. Smith, and T. Siegal. 2010. Gliomas display a microRNA 575

expression profile reminiscent of neural precursor cells. Neuro Oncol 12:422-33.576

26

33. Li, Y., T. G. Vandenboom, 2nd, Z. Wang, D. Kong, S. Ali, P. A. Philip, and F. 577

H. Sarkar. 2010. miR-146a suppresses invasion of pancreatic cancer cells. 578

Cancer Res 70:1486-95.579

34. Lin, S. L., A. Chiang, D. Chang, and S. Y. Ying. 2008. Loss of mir-146a 580

function in hormone-refractory prostate cancer. RNA 14:417-24.581

35. Louis, D. N., H. Ohgaki, O. D. Wiestler, W. K. Cavenee, P. C. Burger, A. 582

Jouvet, B. W. Scheithauer, and P. Kleihues. 2007. The 2007 WHO 583

classification of tumours of the central nervous system. Acta Neuropathol 584

114:97-109.585

36. Louvi, A., and S. Artavanis-Tsakonas. 2006. Notch signalling in vertebrate 586

neural development. Nat Rev Neurosci 7:93-102.587

37. Lu, L. F., M. P. Boldin, A. Chaudhry, L. L. Lin, K. D. Taganov, T. Hanada, A. 588

Yoshimura, D. Baltimore, and A. Y. Rudensky. 2010. Function of miR-146a in 589

controlling Treg cell-mediated regulation of Th1 responses. Cell 142:914-29.590

38. Marumoto, T., A. Tashiro, D. Friedmann-Morvinski, M. Scadeng, Y. Soda, F. 591

H. Gage, and I. M. Verma. 2009. Development of a novel mouse glioma model 592

using lentiviral vectors. Nat Med 15:110-6.593

39. Mischel, P. S., and T. F. Cloughesy. 2003. Targeted molecular therapy of GBM. 594

Brain Pathol 13:52-61.595

40. Oya, S., G. Yoshikawa, K. Takai, J. I. Tanaka, S. Higashiyama, N. Saito, T. 596

Kirino, and N. Kawahara. 2009. Attenuation of Notch signaling promotes the 597

differentiation of neural progenitors into neurons in the hippocampal CA1 region 598

after ischemic injury. Neuroscience 158:683-92.599

27

41. Pellegatta, S., P. L. Poliani, D. Corno, F. Menghi, F. Ghielmetti, B. Suarez-600

Merino, V. Caldera, S. Nava, M. Ravanini, F. Facchetti, M. G. Bruzzone, and 601

G. Finocchiaro. 2006. Neurospheres enriched in cancer stem-like cells are 602

highly effective in eliciting a dendritic cell-mediated immune response against 603

malignant gliomas. Cancer Res 66:10247-52.604

42. Philippidou, D., M. Schmitt, D. Moser, C. Margue, P. V. Nazarov, A. Muller, L. 605

Vallar, D. Nashan, I. Behrmann, and S. Kreis. 2010. Signatures of microRNAs 606

and selected microRNA target genes in human melanoma. Cancer Res 70:4163-607

73.608

43. Piao, Y., L. Lu, and J. de Groot. 2009. AMPA receptors promote perivascular 609

glioma invasion via beta1 integrin-dependent adhesion to the extracellular matrix. 610

Neuro Oncol 11:260-73.611

44. Purow, B. W., T. K. Sundaresan, M. J. Burdick, B. A. Kefas, L. D. Comeau, M. 612

P. Hawkinson, Q. Su, Y. Kotliarov, J. Lee, W. Zhang, and H. A. Fine. 2008. 613

Notch-1 regulates transcription of the epidermal growth factor receptor through 614

p53. Carcinogenesis 29:918-25.615

45. Rajasekhar, V. K., A. Viale, N. D. Socci, M. Wiedmann, X. Hu, and E. C. 616

Holland. 2003. Oncogenic Ras and Akt signaling contribute to glioblastoma 617

formation by differential recruitment of existing mRNAs to polysomes. Mol Cell 618

12:889-901.619

46. Rao, S. A., V. Santosh, and K. Somasundaram. 2010. Genome-wide 620

expression profiling identifies deregulated miRNAs in malignant astrocytoma. 621

Mod Pathol. 622

28

47. Rich, J. N., C. Guo, R. E. McLendon, D. D. Bigner, X. F. Wang, and C. M. 623

Counter. 2001. A genetically tractable model of human glioma formation. Cancer 624

Res 61:3556-60.625

48. Scherr, M., L. Venturini, K. Battmer, M. Schaller-Schoenitz, D. Schaefer, I. 626

Dallmann, A. Ganser, and M. Eder. 2007. Lentivirus-mediated antagomir 627

expression for specific inhibition of miRNA function. Nucleic Acids Res 35:e149.628

49. Silber, J., D. A. Lim, C. Petritsch, A. I. Persson, A. K. Maunakea, M. Yu, S. R. 629

Vandenberg, D. G. Ginzinger, C. D. James, J. F. Costello, G. Bergers, W. A. 630

Weiss, A. Alvarez-Buylla, and J. G. Hodgson. 2008. miR-124 and miR-137 631

inhibit proliferation of glioblastoma multiforme cells and induce differentiation of 632

brain tumor stem cells. BMC Med 6:14.633

50. Soroceanu, L., T. J. Manning, Jr., and H. Sontheimer. 1999. Modulation of 634

glioma cell migration and invasion using Cl(-) and K(+) ion channel blockers. J 635

Neurosci 19:5942-54.636

51. Volinia, S., G. A. Calin, C. G. Liu, S. Ambs, A. Cimmino, F. Petrocca, R. 637

Visone, M. Iorio, C. Roldo, M. Ferracin, R. L. Prueitt, N. Yanaihara, G. Lanza, 638

A. Scarpa, A. Vecchione, M. Negrini, C. C. Harris, and C. M. Croce. 2006. A 639

microRNA expression signature of human solid tumors defines cancer gene 640

targets. Proc Natl Acad Sci U S A 103:2257-61.641

52. Wang, H., W. Zhang, H. J. Huang, W. S. Liao, and G. N. Fuller. 2004. Analysis 642

of the activation status of Akt, NFkappaB, and Stat3 in human diffuse gliomas. 643

Lab Invest 84:941-51.644

29

53. Wang, X., S. Tang, S. Y. Le, R. Lu, J. S. Rader, C. Meyers, and Z. M. Zheng.645

2008. Aberrant expression of oncogenic and tumor-suppressive microRNAs in 646

cervical cancer is required for cancer cell growth. PLoS One 3:e2557.647

54. Xu, P., M. Qiu, Z. Zhang, C. Kang, R. Jiang, Z. Jia, G. Wang, H. Jiang, and P. 648

Pu. 2010. The oncogenic roles of Notch1 in astrocytic gliomas in vitro and in vivo. 649

J Neurooncol 97:41-51.650

55. Yoon, K., and N. Gaiano. 2005. Notch signaling in the mammalian central 651

nervous system: insights from mouse mutants. Nat Neurosci 8:709-15.652

56. Zhang, C. L., Y. Zou, W. He, F. H. Gage, and R. M. Evans. 2008. A role for 653

adult TLX-positive neural stem cells in learning and behaviour. Nature 451:1004-654

7.655

57. Zhao, C., G. Sun, S. Li, M. F. Lang, S. Yang, W. Li, and Y. Shi. 2010. 656

MicroRNA let-7b regulates neural stem cell proliferation and differentiation by 657

targeting nuclear receptor TLX signaling. Proc Natl Acad Sci U S A 107:1876-81.658

58. Zhao, N., Y. Guo, M. Zhang, L. Lin, and Z. Zheng. 2010. Akt-mTOR signaling is 659

involved in Notch-1-mediated glioma cell survival and proliferation. Oncol Rep 660

23:1443-7.661

59. Zheng, H., H. Ying, H. Yan, A. C. Kimmelman, D. J. Hiller, A. J. Chen, S. R. 662

Perry, G. Tonon, G. C. Chu, Z. Ding, J. M. Stommel, K. L. Dunn, R. 663

Wiedemeyer, M. J. You, C. Brennan, Y. A. Wang, K. L. Ligon, W. H. Wong, L. 664

Chin, and R. A. DePinho. 2008. p53 and Pten control neural and glioma 665

stem/progenitor cell renewal and differentiation. Nature 455:1129-33.666

30

60. Zhu, H., J. Acquaviva, P. Ramachandran, A. Boskovitz, S. Woolfenden, R. 667

Pfannl, R. T. Bronson, J. W. Chen, R. Weissleder, D. E. Housman, and A. 668

Charest. 2009. Oncogenic EGFR signaling cooperates with loss of tumor 669

suppressor gene functions in gliomagenesis. Proc Natl Acad Sci U S A 670

106:2712-6.671

672

FIGURE LEGENDS673

FIG. 1. Induction of miR-146a expression by EGFR and PTEN signaling in immortalized 674

Ink4a/Arf-/- astrocytes. (A-B) Proliferation of Ink4a/Arf-/- (I), Ink4a/Arf-/-EGFRvIII (IE), or 675

Ink4a/Arf-/-Pten-/-EGFRvIII (IPE) astrocytes determined by MTT assay (A) or by cell 676

counting (B) (n=3). (B) Heat map representation of differentially expressed miRNAs 677

identified by microarray analysis. (C) qPCR analysis to confirm miR-146a expression 678

induced by EGFR and PTEN signaling using independent RNA samples. (D) Effects of 679

acute Pten loss on expression of miR-146a in Ink4a/Arf-/- Ptenf/f (IPf/f) or Ink4a/Arf-/- 680

Ptenf/f EGFRvIII (IPf/fE) astrocytes. These cells were transduced with adenoviruses 681

expressing either GFP or GFP-Cre. (F) Acute deletion of Pten in Ink4a/Arf-/-Ptenf/f and682

Ink4a/Arf-/-Ptenf/fEGFRvIII astrocytes. Astrocytes were transduced with adenoviruses 683

expressing GFP (Ad-GFP) or GFP-Cre (Ad-GFP-Cre). Using primers located in the 684

floxed exon 5 of the Pten gene, a 157-bp PCR product can be detected in Ad-GFP-685

transduced but not Ad-GFP-Cretransduced astrocytes. Expression of Hprt was used as 686

a loading control. Each experiment was performed in triplicate, except for microarrays 687

that were in duplicate (*p < 0.05, **p < 0.01, and ***p < 0.001). 688

689

31

FIG. 2. miR-146a inhibits cellular transformation and migration of malignant astrocytes. 690

(A) Schema of miR-146a mutant (Ctrl). The seeding sequence of miR-146a is replaced 691

with a Not1 digestion site. (B) qPCR analysis of miR-146a expression in lentivirus-692

transduced astrocytes. Experiments were performed in triplicate. (C-D) Growth curves 693

of Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes transduced with lentiviruses expressing either 694

miR-146a or its mutant (Ctrl) by MTT assay (C) or by cell counting (D) (n=3). (E) 695

Apoptotic cells were detected by an antibody recognizing cleaved caspase 3 and nuclei 696

were shown by DAPI-staining. (F-G) Ectopic expression of miR-146a restrains 697

clonogenic growth of Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes in a soft-agar assay (n=3).698

Representative images (100× magnification) are shown. (H) A scratch assay to examine 699

the effect of miR-146a on migration of malignant astrocytes (n=3). Representative 700

images (40× magnification) at 0 and 24 hours are shown. (I) Transwell migration assay 701

of Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes transduced with lentiviruses expressing either 702

miR-146a or its mutant (Ctrl) (n=3) (*p < 0.05, **p < 0.01, and ***p < 0.001). 703

704

FIG. 3. miR-146a decreases tumor burden induced by intracranial transplantation of 705

Ink4a/Arf-/-Pten-/-EGFRvIII murine astrocytes. (A) Representative histological analysis of 706

brain gliomas in NOD/SCID mice 3 weeks post transplantation. In each experiment, a 707

seed region-mutated version of miR-146a was used as a control (Ctrl). (B) Kaplan-Meier 708

survival curve of mice transplanted with Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes (n=11 and 709

8 for the control and miR-146a group, respectively; p=0.002 by Log-rank Test). 710

711

32

FIG. 4. miR-146a reduces formation of glioma stem-like cells and their migration. (A) 712

Glioma stem-like cells were established by culturing 5,000 Ink4a/Arf-/-Pten-/-EGFRvIII713

astrocytes in NSC medium. The number of spheres and their diameter were determined 714

7 days later. A representative morphology of the spheres is shown in the left panels at 715

×100 magnification (n=3). (B) Number counts of primary, secondary and tertiary 716

neurospheres established by culturing 2,500 Ink4a/Arf-/-Pten-/-EGFRvIII murine717

astrocytes in NSC medium (n=3). (C) In the absence of growth factor, neurospheres 718

(derived from Ink4a/Arf-/-Pten-/-EGFRvIII astrocytes) were differentiated into GFAP-719

positive astrocytes and Tuj1-positive neurons in response to 1% fetal bovine serum and 720

5 M forskolin (FSK) treatment, respectively. (D) Ectopic expression of miR-146a 721

induces glial marker GFAP and suppresses expression of stem cell markers Sox2 and 722

Nestin. Actin was used as a loading control for Western blotting. Relative protein level is 723

shown. (E) miR-146a inhibits spreading and migration of glioma stem-like spheres. 724

Representative photographs were taken at ×100 magnification. Distance was measured 725

from the edge of the sphere to the periphery of the migrating cells (n=3; *p < 0.05, **p <726

0.01 and ***p < 0.001). 727

728

FIG. 5. Notch1 is a direct target of miR-146a. (A) Sequence alignments between 3’UTR 729

of mouse Notch1 and miR-146a. (B) miR-146a suppresses the activity of a luciferase 730

reporter that is linked to the 3’UTR of Notch1, but not mutant 3’UTR of Notch1. After 731

normalization to -galactosidase expression, which was used as a control for 732

transfection efficiency, the data was presented as the ratio to that of empty-vector–733

transfected cells. (C) miR-146a blocks protein expression of Notch1 and its downstream 734

33

target, pAKT, in Ink4a/Arf-/-Pten-/-EGFRVIII astrocytes. Protein loading was monitored by 735

-actin. Relative protein level is indicated. (D) Blocking Notch1 processing by DAPT, a 736

-secretase inhibitor, reveals the inhibitory effect of miR-146a on full-length Notch1 737

protein in murine astrocytes. (E) miR-146a has no effect on the expression of Notch1738

transcripts, determined by qPCR. (F) NICD rescues the inhibitory effect of miR-146a on 739

the formation of glioma stem-like cells from Ink4a/Arf-/-Pten-/-EGFRVIII astrocytes (*p < 740

0.05 and ***p < 0.001). 741

742

FIG. 6. miR-146a regulates normal NSCs. (A) Induction of miR-146a after culturing 743

NSCs under differentiation conditions for 4 days. GF, growth factor bFGF and EGF; RA, 744

retinoic acid; FSK, forskolin. (B) and (C) Promoting neuronal differentiation of normal 745

NSCs by miR-146a. Neurons were stained with an antibody against Tuj1 and nuclei 746

were shown by DAPI staining. Representative photographs were taken at ×400 747

magnification. (D) - (F) miR-146a inhibits NSC proliferation by inducing cell cycle exit. 748

(D) Proliferating cells were examined by BrdU incorporation and Ki67 staining, which 749

are shown in merged panels with phase-contrast images. Apoptotic cells were stained 750

with an antibody for cleaved caspase 3 (Casp3). Representative photographs were 751

taken at ×200 magnification. (E) Quantification of proliferating NSCs (BrdU+ or Ki67+). (F) 752

miR-146a enhances cell cycle exit, which was measured by the ratio of BrdU+Ki67- cells 753

to total number of BrdU+ cells. Experiments were performed in triplicate (**p < 0.01 and 754

***p < 0.001). 755

756

34

FIG. 7. miR-146a promotes survival of mice transplanted with human glioblastoma cells. 757

(A) A recognition site for miR-146a in the 3’UTR of human Notch1 gene. (B) miR-146a 758

targets human Notch1. Protein levels were examined by Western blotting analysis of 759

lysates from human U87 glioblastoma cells. Protein loading was monitored by -actin.760

Relative protein level is indicated. (C) miR-146a inhibits the formation of glioma stem-761

like cells from U87. These cells were enriched in NSC medium. The number of spheres 762

and their diameter were measured 7 days later. (D) and (E) miR-146a promotes survival 763

of tumor-bearing mice by reducing glioma burden. (D) Histological analysis of brain 764

tumors 3 weeks post intracranial transplantation. (E) Kaplan-Meier survival curve of 765

mice transplanted with human U87 glioblastoma cells (n=6 for each group; p=0.006 by 766

Log-rank Test).767

768

FIG. 8. Knocking down miR-146a function promotes tumorigenesis. (A) Design of 769

miRNA sponge. (B) Sponge suppresses miR-146a’s function on activity of a luciferase 770

reporter that is linked to the 3’UTR of mouse Notch1 gene. The data was presented as 771

the ratio to that of miR-146a mutant–transduced cells. (C) miRNA sponge reverses miR-772

146a-mediated down-regulation of Notch1 expression. Relative protein level is indicated. 773

(D) Enhancing endogenous Notch1 expression by miRNA sponge in Ink4a/Arf-/-Pten-/-774

EGFRVIII astrocytes. (E) and (F) miRNA sponge against miR-146a promotes formation 775

of glioma stem-like cells from malignant astrocytes (n=3). (G) Knocking down miR-146a 776

function by sponge exacerbates tumor burden. The volume of subcutaneous tumors 777

was measured after microdissection (n=5; *p < 0.05, **p < 0.01, and ***p < 0.001). (H) A 778

diagram showing miR-146a-involved signaling network. 779