Nkx2-2As Suppression Contributes to the Pathogenesis of Sonic

Published OnlineFirst December 11, 2017; DOI: 10.1158/0008-5472.CAN-17-1631 Cancer Tumor Biology and Immunology Research

Nkx2-2as Suppression Contributes to the Pathogenesis of Sonic Hedgehog Medulloblastoma Yimeng Zhang1, Ting Wang1, Shan Wang1, Yanlu Xiong2, Rui Zhang1, Xiang Zhang1, Jing Zhao1, An-Gang Yang3, Lei Wang1, and Lintao Jia1

Abstract

Aberrant Hedgehog signaling and excessive activation of the LATS2 targeting activity. Shh signaling impaired Nkx2-2as Gli family of transcriptional activators are key drivers of expression by upregulating the transcriptional repressor medulloblastoma (MB), the most common human pediatric FoxD1. In clinical specimens of Shh-subgroup MB, we vali- brain malignancy. MB originates mainly from cerebellar gran- dated coordinated expression of the aforementioned proteins. ule neuron progenitors (CGNP), but the mechanisms under- Notably, exogenous expression of Nkx2-2as suppressed tumor- lying CGNP transformation remain largely obscure. In this igenesis and prolonged animal survival in MB mouse models. study, we found that suppression of the noncoding RNA Our findings illuminate the role of noncoding RNAs in Hedge- Nkx2-2as promoted Sonic Hedgehog (Shh)-potentiated MB hog signaling and MB occurrence, with implications for iden- development. Nkx2-2as functioned as a competing endoge- tifying candidate therapeutic targets for MB treatment. nous RNA against miR-103 and miR-107, sequestering them Significance: These findings illuminate the role of noncod- and thereby derepressing their tumor suppressive targets BTG2 ing RNAs in Hedgehog signaling and an interplay between and LATS1 and impeding cell division and migration. We also the Hedgehog and Hippo pathways in medulloblastoma found that Nkx2-2as tethered miR-548m and abrogated its pathogenesis. Cancer Res; 78(4); 1–12. 2017 AACR.

Introduction genetic aberrancies lead to tumorigenesis, transgenic mice that mimic these mutations recapitulate the occurrence of MBs, pro- Medulloblastoma (MB) is the most common type of pediatric viding important tools for deciphering the precise mechanisms brain malignancy (1). Accumulating evidence suggests that MB underlying the pathogenesis of MBs (5, 6). originates from cerebellar granule neuron progenitors (CGNP), The Hedgehog pathway is a key regulator of ontogenesis, and which proliferate in a germinal matrix along the outside of the aberrant Hedgehog signaling has been implicated in carcinogen- cerebellum termed the external granule cell layer (1, 2). Due to the esis (7). In the absence of the Hedgehog ligands, the inhibitory limited prognostic value of previous histological groupings, transmembrane receptor Patched prevents high expression and molecular typing of clinical MBs is necessary (3, 4). Along these activity of the seven-membrane-spanning receptor Smoothened lines, a recent integrated genomic profiling study revealed that MB (SMO; refs. 7, 8). In mammalian cells, ligand engagement of consists of at least four distinct molecular subgroups with signif- Patched relieves SMO inhibition, allowing it to process and icant differences in their demographic, genetic, and symptomatic activate Gli transcription factors. Conversely, the Gli proteins are features: Wnt, Sonic Hedgehog (Shh), group 3, and group 4 (1, 3). directly bound and antagonized by the suppressor of fused Despite the current lack of a systemic understanding of how homolog (Sufu; refs. 7, 8). As the best-studied Hedgehog homolog, Shh signaling accounts for nearly 30% of human MBs, with the most frequent mutations detected in Patched, Sufu, and 1State Key Laboratory of Cancer Biology, Department of Biochemistry and SMO (9). Although a growing number of genes involved in cell Molecular Biology, Fourth Military Medical University, Xi'an, China. 2Department growth and differentiation have been identified as direct targets of Thoracic Surgery, Tangdu Hospital, Fourth Military Medical University, Xi'an, 3 of the Gli isoforms (Gli 1 to 3) during embryonic development, China. Department of Immunology, Fourth Military Medical University, Xi'an, the key mediators of MB downstream of Gli remain to be char- China. acterized (9, 10). Note: Supplementary data for this article are available at Cancer Research Long noncoding RNAs (lncRNA) are RNA transcripts longer Online (http://cancerres.aacrjournals.org/). than 200 nucleotides without a defined protein-coding capacity Y. Zhang, T. Wang, and S. Wang contributed equally to this article. (11). These previously designated "junk" RNAs have attracted a Corresponding Authors: Lintao Jia, Department of Biochemistry and Molecular great deal of attention due to their indispensable roles in regu- Biology, Fourth Military Medical University, 169 Changle West Road, Xi'an lating gene expression via multiple mechanisms (11, 12). Many 710032, China. Phone: 86-29-8477-6799; Fax: 86-29-8477-3947; E-mail: lncRNAs act in trans via a direct interaction with proteins to [email protected]; and Lei Wang, [email protected] modulate their function, whereas others act as competing endog- doi: 10.1158/0008-5472.CAN-17-1631 enous RNAs (ceRNA) by sequestering microRNAs (miRNA) that 2017 American Association for Cancer Research. would otherwise silence individual genes upon binding to the

www.aacrjournals.org OF1

Zhang et al.

30 untranslated region (30 UTR) of target messenger RNAs (mRNA; pooled cerebella were cut into pieces and then treated with ref. 13). Recent studies showed that lncRNAs are key regulators of trypsin-EDTA for 15 minutes at 37 Cin5%CO2. The digested diverse cell behaviors, including malignant transformation (14). cerebella tissue was aspirated into DMEM–F-12 containing 10% Here, we found that the downregulation of a lncRNA, Nkx2-2as, is fetal calf serum and penicillin–streptomycin for 5 minutes. critically involved in Shh-driven development of MB. Specifically, Enough cells suspension were collected after aspirating repeatedly Nkx2-2as functions as a ceRNA to sequester miR-103/miR-107 and filtrating, and then resuspended in Neurobasal medium and miR-548m, thereby maintaining expression of their tumor- (Gibco) with B27 supplement, 25 mmol/L KCl and antibiotics. 5 suppressive targets, B-cell translocation gene 2 (BTG2/Tis21/PC3) Finally, cells were plated at a density of 5 10 cells per poly-D- and large tumor suppressor kinase 1/2 (LATS1/2). In addition, we lysine-coated 60-mm Petri dish. investigated the mechanism underlying Shh-elicited impairment of Nkx2-2as expression in MB cells. Transgenic mouse MB model The spontaneous orthotopic MB model driven by Hedgehog Materials and Methods signaling was from The Jackson Laboratory (stock no.: 008831). These transgenic mice express SmoA1, a constitutively active Small RNAs and gene overexpression constructs mutant of the mouse homolog of the Drosophila Smo gene, under Complementary strands of siRNAs and single-stranded the control of the mouse neurogenic differentiation 2 (Neurod2) miRNA mimics or inhibitors were synthesized by GenePharma. promoter, resulting in transgene expression specific to cerebellar Target sequences of siRNAs and the sequences for miRNA granule cells (16). mimics are listed in Supplementary Tab. S1. For generation of lentiviral constructs of wild-type (WT) or mutant Nkx2-2as, Microarray analysis cDNA strands corresponding to the human or mouse Nkx2-2as CGNPs from neonatal mice were compared with MB cells from sequences (Transcript ID: ENST00000549659 and the Neurod2-SmoA1 mouse model using microarray analysis with ENSMUST00000136998.2, www.ensemble.org), those with the ArrayStar LncRNA Array protocol (Kangcheng Biotech). The putative miRNA-binding site mutations, or an RNA complemen- microarray data have been deposited in Gene Expression Omni- tary to the 170-638 nt of human Nkx2-2as (as a control) were bus (GEO, assigned accession no.: GSE85449). synthesized and cloned into the BamHI/EcoRI sites of pFUGW (GeneChem). The constructs were transfected into HEK293T cells Quantitative RT-PCR analysis along with packaging plasmids pMD2.G and psPAX2 (Gene- Total RNA was extracted using TRIzol reagent (Invitrogen) and Chem). After transfection for 48 hours, the supernatant was reverse-transcribed to cDNA. cDNA was amplified by quantitative collected, centrifuged, filtered, and used for infection of target PCR using the SYBR Premix Ex Taq (TaKaRa). To quantify cells. WT Nkx2-2as was also subcloned into vector pLV-IRES- miRNAs, total RNA was reverse-transcribed using the miScript mCherry (Clontech), followed by packaging in HEK293T cells Reverse Transcription Kit (Qiagen), and then amplified using using the Lenti-X HT packaging system (Clontech). The cDNA customized primers paired with the universal primer provided encoding FoxD1 was amplified from total RNA from Daoy cells in the kit. The levels of mRNAs and miRNAs were normalized to 0 using the following primers: 5 -AAATGACCCTGAGCACTGA- those of GAPDH, 18S rRNA, and U6. The primers for PCR are 0 0 0 GATGT-3 and 5 -AATTAACAATTGGAAATCCTAGCAGT-3 . The listed in Supplementary Table S2. resultant PCR fragment was cloned into vector pMD-18 T and subcloned into the lentivirus-based expression plasmid pLenti6/ Western blotting V5-DEST (Invitrogen). Virus packaging and infection were per- Cells were harvested at the indicated times, and total proteins formed as recommended by the manufacturer. were extracted for analysis. Protein concentrations were quantified using a BCA kit (Pierce). Proteins were separated on Cell culture, transfection, and infection SDS/PAGE gels, transferred onto PVDF membrane, and subjected Human MB cell lines, Daoy and D341 Med, and human to immunoblot analyses using the antibodies listed in Supple- embryonic kidney 293T (HEK293T) cells were purchased from mentary Table S3. Horseradish peroxidase–linked F(ab)2 frag- ATCC in 2015. All cells were cultured in DMEM (Sigma-Aldrich) ments of goat anti-rabbit and anti-mouse immunoglobulin supplemented with 10% fetal bovine serum (FBS; Gibco, BRL). (ZB-2305, Zhong Shan Jin Qiao, China) were used as secondary Cells were routinely grown to 80% confluence at 37 Cina antibodies. humidified atmosphere containing 5% CO2, and cells from passages 2 to 4 were used for experiments. Transfection was Luciferase reporter assay performed using Lipofectamine 2000 reagent (Invitrogen). For For miRNA target gene verification, the 30 UTRs of target mRNAs overexpression of intended RNA in Daoy cells, cells were infected were amplified and cloned into the XbaI site downstream of the with recombinant lentiviruses and were selected with 100 mg/mL firefly luciferase gene in vector pGL3-Promoter (Promega). The bleomycin (Sigma-Aldrich) for two weeks prior to use of homog- primers used for 30 UTR amplification were as follows: 50-GAAAT- enous pool of the infected cells for further assays. CATGACTTGTTTCTAATTC-30 and 50-ATGCAAGGCTGACTAGC- CAGCCAT-30 for BTG2, and 50-GAAAGACAGTTTTAGTTT- Cerebellar granule neuron precursor cell culture TATCTTGC-30 and 50-CCAGATAGCCAGATTTTCCTTTGCC-30 for Mouse cerebellar granule neuron precursor (CGNP) cells were LATS1. Mutant 30 UTRs were obtained via PCR using the Quik- prepared from cerebella dissected from neonatal mice as reported Change Site-Directed Mutagenesis Kit (Stratagene). The resulting previously (15). Briefly, cerebella from C57BL/6J mice at 4 to 5 constructs and a pRL-TK Renilla luciferase construct were cotrans- postnatal days were dissected into Dulbecco's Phosphate-Buff- fected along with miRNA mimics into HEK293 cells. To investi- ered Saline (Corning). The meninges were stripped, and the gate miRNA affinity and its effect on the expression of luciferase

OF2 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Represses Shh Medulloblastoma

from a construct containing Nkx2-2as in the 30 UTR, WT or Flow cytometry mutant lncRNA was amplified and cloned into the SgfI/NotI For apoptosis assays, control or Nkx2-2as–overexpressing MB site of psiCHECK-2 (Promega). The primers used for PCR cells were fixed with 4% formaldehyde/PBS for 15 minutes and amplification were as follows: 50-TTGCGATCGCTAAAAAAAG- permeabilized with 3% BSA/0.1% glycine/PBS for 30 minutes. GAAGAAATTCTCTG-30 and 50-TTGCGGCCGCTTTCGGCAG- The cells were then stained with Annexin V and propidium iodide CAGCGGTACCTG-30. The resultant constructs and a pGL3- (PI; Roche) for 15 minutes in the dark, and analyzed using a Control firefly luciferase construct were cotransfected with miRNA Coulter Epics XL-MCL flow cytometer (Beckman Coulter). For mimics into HEK293 cells. Cell extracts were prepared, cell-cycle assays, control or Nkx2-2as–overexpressing Daoy cells and luciferase activity was measured using the Dual Luciferase were seeded into 6-well plates. Thirty-six hours after transfection Reporter Assay System (Promega). with siRNA duplexes, cells were collected and fixed in chilled 70% ethanol at 20C for 2 hours, followed by washing with phos- Chromatin immunoprecipitation phate-buffered saline (PBS). The fixed cells were stained with Cells were cross-linked with formaldehyde and harvested for 50 mg/mL PI at room temperature for 20 minutes before analysis. chromatin immunoprecipitation (ChIP). Briefly, chromatin was To evaluate expression of cancer stem cell markers, tumor tissues fragmented by sonication, and precleared chromatin was immu- were isolated from the spontaneous MB mouse model, dissoci- noprecipitated overnight with antibodies or IgG as a negative ated into single-cell suspensions, and infected with control or control (Supplementary Table S3). The enrichment of specific Nkx2-2as recombinant lentiviruses. Forty-eight hours later, cells DNA fragments was analyzed by PCR. The primers used for were incubated with 10 mL of anti-CD133-PE, anti-CD15-PerCP- amplification of human promoter regions are as follows: Cy5.5, or isotype control immunoglobins (Supplementary 50-GCAGTAGGACAAACGGGAGGGCAG-30 and 50-TGTGTGAG- Table S3) for 30 minutes at room temperature in the dark, TAGCGATATTGTCAGCC-30 for Nkx2-2as;50-TTAGTTTCTAACTC- followed by analysis on a Coulter Epics XL-MCL flow cytometer. CAGGAGGGGT-30 and 50-AGGTAGGAACCGGTGAATGTTAAA- GAG-30 for Nkx2-2;50-GAAGGACAAGATGAAGGAAATGCT-30 Cell invasion assay and 50-AGGCTCCAGGACTTTGCAACTTCAAC-30 for CCND1,a The invasiveness of cells was measured as reported previously known Gli target gene; 50-TGCGACTGCGGCTGCCGGAGCTGC- (19). Briefly, a Transwell insert with a diameter of 8 mm (Milli- 30 and 50-TCGTGCTTAAATTGGGGGGCTTCGCATCA-30 for pore) was coated with 200 mL of Matrigel and preincubated with FoxD1; and 50-GGCAAATGCCTGACTCAGTGACC-30 and 50-TGA- DMEM. Cells were suspended in DMEM containing 1% FBS and CTCACCGTCCGGTCTCCCAGCA-30 for TUBB3, used as a nega- seeded into the upper chamber of the Transwell (2 104 cells/ tive control. insert), and 500 mL of DMEM containing 10% FBS was added to the lower chamber. After incubation at 37C for 12 hours, the cells RNA immunoprecipitation assays were fixed in methanol and stained with Crystal violet (Sigma). RNA immunoprecipitation (RIP) assays were conducted using Cells that invaded through the pores to the lower surface of the the EZ-Magna RIP RNA-Binding Protein Immunoprecipitation Kit filter were counted under a microscope. (Millipore). The MS2-based RIP assays were performed as previously described (17). Briefly, the cDNAs of WT and mutated Nkx2- Sphere formation assay 2as and Renilla luciferase (as a control) were subcloned into pSL- Tumors were isolated from Neurod2-SmoA1-transgenic mice, MS2-24 containing a 24-MS2-binding module (18). The resultant and single-cell suspensions were prepared. Cells were seeded at a construct was cotransfected with pMS2-GFP (18) into Daoy cells, density of 100 cells per well into ultralow attachment 24-well followed by immunoprecipitation using GFP or IgG as a control. plates (Corning) in 100 mL of DMEM/F12 serum-free medium. The precipitated RNA fraction was analyzed by qRT-PCR. The Cells were fed every 3 days by replacement of 50% of the medium, antibodies and primers used for RIP are shown in Supplementary and were infected with control or Nkx2-2as recombinant lenti- Tables S3 and S2, respectively. viruses coexpressing mCherry 7 days after seeding. Spheres were observed under an inverted microscope and a fluorescence Cell proliferation assay microscope. Cell proliferation was measured using the Cell Counting Kit-8 (CCK-8; 7Sea Biotech). Briefly, harvested cells were seeded into Immunofluorescent staining and images analysis 96-well plates at a density of 2,000 cells/well (n ¼ 5 for each time CGNP cells were fixed with 4% paraformaldehyde for 20 min- point) in a final volume of 100 mL. CCK-8 solution (10 mL) was utes after transfection with Nkx2.2as-targeted or control siRNAs. added to each well, and the absorbance at 450 nm was measured After permeabilization with 0.5% Triton and blocking with 3% after incubation for 2 hours at 37C to calculate the number of BSA, cells were incubated overnight at 4C with FITC-conjugated viable cells. or unconjugated primary antibodies (Supplementary Table S3). Then cells incubated with unconjugated primary antibody were Plate colony formation assay further stained with a Cy3-conjugated secondary antibody for 1 Suspensions of cells were inoculated in 6-well flat-bottomed hour in the dark at room temperature. Cells were next incubated 0 plates at a density of 500 cells per well. Cells were dispersed evenly with 4 -6-diamidino-2-phenylindole (DAPI) for 8 minutes to by mild shaking of the plates, and then incubated in complete stain nuclei. Fluorescent images were captured by a laser confocal medium until visible colonies appeared. Plates were then gently scanning microscopy. Daoy cells were stained similarly after washed and subjected to Giemsa staining. Viable colonies con- infection with Nkx2.2as-overexpressing or control lentiviruses. taining at least 50 cells were counted. All experiments were All captured images were analyzed by the Image-Pro Plus 6.0. The repeated in triplicate, and the average values are presented. IOD is the sum of the fluorescence intensity of every cell of the

www.aacrjournals.org Cancer Res; 78(4) February 15, 2018 OF3

Zhang et al.

images. The area is the pixel of the extracted cell, and the mean lentiviruses. Briefly, mice were anesthetized with ketamine and fluorescence intensity can be calculated by IOD/area. xylazine and placed in a Kopf stereotaxic apparatus. Sterile sur- gical procedures were followed to expose the injection site located Histological and immunohistochemical staining at AP 4.55 mm, ML 1.5–2 mm, DV 3.0 mm according to mouse Mice were anesthetized with 10% chloral hydrate and perfused brain stereotaxic coordinates (21). A 0.5-mm burr hole was placed through the ascending aorta with 150 mL of normal saline (NS; followed by an injection of 2 mL of viruses (3 109 TU/mL) over 2 0–4C) followed by 200 mL of 4% paraformaldehyde (0–4C). minutes. The head specimens after fixation were also scanned with The brain was removed, and the cerebellum was isolated, fixed in micro-CT (Inveon Micro-CT/PET, Siemens). Image acquisition 10% formalin, and embedded in paraffin. Five-micron-thick was performed at 80 kV and 500 mA. Bone mineral density (BMD) sections were cut and subjected to hematoxylin–eosin (H&E) was analyzed after reconstruction of three-dimensional images and immunohistochemical staining as previously described using the workstation and software (Inveon research workplace (20). Immunohistochemistry was also performed using tissue 2.2). All animal study protocols were approved by the Institu- arrays of clinical MBs (Alenabio; Cat. no.: BC17012). The array tional Animal Care and Use Committee of Fourth Military Med- samples were subject to GAB1 staining to distinguish Shh and ical University. other subgroups of MBs. The antibodies used for immunohistochemistry are listed in Supplementary Table S3. Chromogen Bioinformatics analysis development was performed with the ultraView Universal DAB An online tool (LncBase Predicted v.2, carolina.imis.athena- Detection Kit (Ventana Medical Systems). The percentage of innovation.gr/diana_tools/) was used to screen for Nkx2-2as– positive cells and staining intensity were multiplied to produce binding miRNAs. Potential targets of miRNAs were predicted a weighted score for each case. using the PicTar and TargetScan software. Transcription factor binding sites on gene regulatory regions were predicted using the In situ hybridization JASPAR online tool (http://jaspar.binf.ku.dk). Cerebellar sections of WT or MB-bearing mice and the aforementioned clinical MB arrays were used for in situ hybrid- Statistical analysis ization. Double-DIG-labeled probe recognizing a conserved Data were analyzed using the SPSS software as follows: (i) For 0 sequence in mouse and human Nkx2-2as (5 -GTCAA- experiments including qRT-PCR, colony formation, cell invasion, 0 GATCTGGTTCCAGAACCA-3 ) and control RNU6-6P snRNA and luciferase assays, statistical significance was evaluated using probe were purchased from Exiqon. Briefly, sections were the two-tailed Student t test for comparison of two groups, and dehydrated sequentially with 70%, 90%, and 100% ethanol, ANOVA followed by Fisher exact test for comparison of three or air-dried, and hybridized at 55 C with probes (40 nmol/L) more groups. (ii) For associations between gene expression diluted in ISH buffer (Exiqon). Sections were rinsed sequen- values, significance was evaluated by Pearson product–moment tially with 5 saline sodium citrate (SSC), SSC, and 0.2 SSC, correlation coefficient analysis. and incubated with blocking solution (Roche) followed by alkaline phosphatase (AP)-conjugated anti-DIG antibody (1:800) in 2% sheep serum (Jackson Immunoresearch). Slides Results were washed with PBS-T (PBS plus 0.1% Tween-20) and incu- The lncRNA Nkx2-2as is downregulated in Shh-driven MB cells bated in the dark with AP substrate buffer [NBT-BCIP tablet To explore the genes involved in tumorigenesis, we utilized a (Roche)in10mLof0.2mmol/Llevamisole(Fluka,Sigma)]. spontaneous orthotopic MB model driven by Hedgehog sig- The reaction was stopped with AP stop solution (Sigma). naling (The Jackson Laboratory; stock no.: 008831; ref. 16). Tissues were counterstained with Nuclear Fast Red and rinsed H&E staining revealed that MB-like lesions appeared in the with water. Sections were dehydrated as described above and cerebella of these mice (Fig. 1A); survival time of mice with the mounted with coverslips in Eukitt mounting medium (VWR). disease ranged from 4 to 8 months (Supplementary Fig. S1). Images were captured with a light microscope and processed Microarray analysis identified a large population of genes and with identical settings. The signal intensities were quantified noncoding RNAs differentially expressed in MB cells and using the intensity measurement tools in the Image-Pro Plus GCNPs (Fig. 1B; Supplementary Fig. S2, and GEO accession software package (Media Cybernetics). no.: GSE85449). The reliability of gene profile analysis was verified by the upregulation of Ccnd1 (1.83-fold, P ¼ 0.047) Treatment of mouse MB models and Igf2 (2.59-fold, P ¼ 0.0002) and downregulation of Pax6 Athymic nude mice (6–8 weeks old) were injected subcutane- (7.46-fold, P ¼ 0.004) and Jup (2.41-fold, P ¼ 0.008), ously with Daoy cells stably overexpressing Nkx2-2as and Renilla previously reported as direct targets of Hedgehog signaling. luciferase or luciferase alone (5 106 cells per mouse) to allow for Among the lncRNAs with the largest changes in expression, xenograft tumor development. Bioluminescence was imaged on a we identified Nkx2-2as, which was downregulated by 13.53- Xenogen IVIS Kinetic imaging system (Caliper). Identical illumi- fold in MB cells versus GCNPs (GEO accession no.: GSE85449, nation settings (2/f stop, 12.5 cm field of view, binning factor of 4, P ¼ 0.0002). The reduced expression of Nkx2-2as in MB in open filter, 1 min exposure time) were used for all images. comparison with normal cerebellar tissues was also validated Bioluminescence intensities were calculated using the Living by qRT-PCR and in situ hybridization (Fig. 1C and D). Knock- Image software (Caliper) and are expressed as photon flux down of Gli2, a transcriptional activator of the Hedgehog (p/s/cm2/sr). To evaluate the effect of Nkx2-2as overexpression pathway, significantly increased Nkx2-2as levels in the on the survival of Neurod2-SmoA1-transgenic mice, two-month- human MB cell line Daoy (Fig. 1E and F). Thus, Nkx2-2as old mice were randomly grouped and administered biweekly expression is impaired specificallyinMBsasaresultofaberrant intracerebellar injection with control or Nkx2-2as recombinant Hedgehog signaling.

OF4 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Represses Shh Medulloblastoma

Figure 1. Deregulation of Nkx2-2as in MB driven by Shh signaling. A, H&E staining for MB arising in two SmoA1-transgenic mice (Tg #1 and Tg #2; 4 months old; The Jackson Laboratory; stock # 008831). NM, normal tissue. Bar, 50 mm. B, Scatter plot of microarray analysis of differentially expressed lncRNAs in MB and CGNP cells (see also data submitted to GEO, accession # GSE85449). C, qRT-PCR assay of MB tissues from two transgenic mice and cerebellar tissues of two age-matched mice (WT). D, In situ hybridization for Nkx2-2as expression in MB tissues from transgenic mice and age-matched mouse cerebellar (WT) tissues. Parts of the images in the left panel are shown in a higher magniﬁcation (right). Bar, 100 mm. E and F, Western blotting (E) and qRT-PCR (F) assays using Daoy cells transfected with scrambled (Ctrl) or Gli1/Gli2-targeted siRNAs. Data are representative images or are expressed as the means SD of three independent experiments. , P < 0.01, compared with WT1 (C) or the control group (F).

Nkx2-2as suppresses the malignant phenotypes of MB cells Nkx2-2 is not a key mediator of the tumor-suppressive effect Next, we examined whether Nkx2-2as plays a role in Shh- of Nkx2-2as driven onset of MBs. In line with the wide expression of EGFP As an antisense transcript of the Nkx2-2 gene locus, Nkx2-2as in Daoy cells after infection with lentiviruses generated from was previously reported to facilitate the expression of the tran- an empty shuttle vector, cells infected with the Nkx2-2as scription factor Nkx2-2 in neural stem cells (Fig. 3A; ref. 25). recombinant viruses exhibited high expression of this lncRNA Accordingly, we investigated whether Nkx2-2 is involved in the (Fig. 2A). The control viruses express a complementary RNA regulation of malignant growth. qRT-PCR and immunohisto- fragment, which cannot be detected via qRT-PCR due to the chemical staining revealed a significant decrease in Nkx2-2 expres- lack of a primer binding site (Fig. 2A). Ectopic expression of sion in mouse MB compared with normal cerebellar tissues (Fig. Nkx2-2asinhumanMBcelllines,DaoyandD341MED, 3B and C). Overexpression of Nkx2-2as in MB cells caused a slight caused remarkable growth inhibition in both cell lines increase in Nkx2-2 levels (Fig. 3D and E); however, knockdown of (Fig. 2B). Consistent with these observations, Nkx2-2as over- Nkx2-2 caused a moderate but not significant increase in the expression suppressed in vitro colony formation by Daoy cells growth of Nkx2-2as–overexpressing Daoy cells (Fig. 3F and G) (Fig. 2C) and increased the ratio of cells undergoing serum and failed to rescue the invasive capacity impaired by Nkx2-2as withdrawal–induced apoptosis (Fig. 2D). Nkx2-2as also (Fig. 3H). These data suggest that Nkx2-2as plays a suppressive impaired invasion by Daoy cells, as measured in a Transwell role via mechanisms other than its previously reported regulation assay (Fig. 2E). In addition, Nkx2-2as overexpression signifi- of Nkx2-2. cantly inhibited the formation of tumor spheres and the expression of markers of brain tumor stem cells (Supplemen- Nkx2-2as upregulates BTG2 and LATS1 via a ceRNA mechanism tary Fig. S3A, B; refs. 22, 23). MBs were widely accepted to Many lncRNAs have been shown to act as ceRNAs, regulating occur due to the transformation of CGNPs (24). We next gene expression by acting as a "sponge" for miRNAs and thereby cultured CGNPs isolated from neonatal mice, and achieved derepressing the targets of the bound miRNAs (26). Given the siRNA-based knockdown of Nkx2-2as in these cells (Fig. 2F predominantly cytoplasmic localization of Nkx2-2as in the cer- and G). As a result, the decrease in Nkx2-2as levels improved ebellum (Fig. 1D), we next investigated whether it suppresses MB the growth of CGNPs (Fig. 2H). Therefore, Nkx2-2as functions development via the ceRNA mechanism. Among the candidate as a tumor suppressor in both CGNPs and MB cells. miRNAs predicted to be tethered by Nkx2-2as, miR-103a/107 and

www.aacrjournals.org Cancer Res; 78(4) February 15, 2018 OF5

Zhang et al.

Figure 2. Nkx2-2as acts as a tumor suppressor in MB. A, Daoy cells were infected with control RNA (complementary to 170-638 nt of Nkx2-2as, Ctrl)– or Nkx2-2as–expressing recombinant lentivirus, followed by fluorescence microscopy (left) and qRT-PCR assay (right). Left, Daoy cells infected with lentiviruses derived from an empty pFUGW shuttle vector that expressed EGFP. B, CCK-8 assay to assess the growth of Daoy and D341 Med cells after infection with the recombinant lentivirus described in A. C, In vitro colony formation assay of Daoy cells described in A. D, Annexin V/PI staining followed by flow cytometry assay to detect apoptosis in Daoy cells infected with the recombinant lentivirus described in A. E, Transwell invasion assays using Daoy cells described in A. Data are representative images or are expressed as the means SD of three independent experiments. F, CGNPs were isolated from neonatal mice, cultured in vitro and subjected to immunofluorescent staining for the indicated markers. G and H, Cells in F were transfected with Nkx2-2as–targeted or scrambled (Ctrl) siRNAs and were subjected to qRT-PCR (G) and CCK-8 (H) assays. Bar, 10 mm. , P < 0.05; , P < 0.001, compared with the control group.

Figure 3. Nkx2-2as suppresses MB independently of Nkx2-2. A, Schematic diagram showing the genomic locus containing human Nkx2-2 (NKX2-2-001) and Nkx2-2as (NKX2-2-AS1-001; http://asia.ensembl.org/index.html). B and C, qRT-PCR assay (B) and immunohistochemical staining (C) of MB tissues from transgenic mice and cerebellar tissues of age- and strain-matched wild-type mice (WT). Bar, 20 mm. D and E, Daoy cells were infected with control RNA- or Nkx2-2as–expressing lentiviruses, followed by qRT-PCR (D) and Western blot (E) analysis. F, Daoy cells were mock-transfected (Ctrl) or transfected with scrambled or Nkx2-2–targeted siRNAs and subjected to Western blot analysis. G and H, Daoy cells were infected with a control RNA– or Nkx2-2as–expressing lentiviruses and transfected with the indicated siRNAs, followed by CCK-8 assay to assess cell growth (G) and Transwell invasion assay (H). Data are representative images or are expressed as the means SD of three independent experiments. , P < 0.05, compared with WT1 (B) or the control group (D and H).

OF6 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Represses Shh Medulloblastoma

Figure 4. Nkx2-2as tethers miR-103a/107 and miR-548m in MB cells. A, MS2-based RIP assay with anti-GFP antibody in Daoy cells after cotransfection with MS2bp-GFP plasmid and MS2bs constructs of Nkx2-2as, or Renilla luciferase (Rluc) as a control (control vectors). , P < 0.05, compared with the MS2bs-Rluc group. B, Absolute qPCR analysis of copy numbers of the indicated RNAs in MB tissues from Neurod2-SmoA1–transgenic mice and cerebellar tissues of age- and strain-matched wild-type mice (Ctrl), and in control or Nkx2-2as–recombinant lentivirus-infected Daoy cells. C, Left, schematic diagram showing Nkx2-2as mutations in putative miRNA-binding sites. Right, MS2-based RIP assay with anti-GFP antibody in Daoy cells after cotransfection with MS2bp-GFP plasmid and MS2bs constructs of wild-type or mutant Nkx2-2as, or Renilla luciferase (Rluc) as a control (control vectors). ,#, P < 0.05, compared with the MS2bs-Rluc or MS2bs-Nkx2-2as group, respectively. D, HEK293T cells were transfected with psiCHECK2 constructs containing wild-type or mutant Nkx2-2as in the 30 UTR of the luciferase reporter gene, followed by transfection with indicated miRNA mimics and determination of the ratio of Renilla luciferase activity to ﬁreﬂy luciferase activity. , P < 0.05, compared with the control mimics group.

miR-548m were enriched by this lncRNA in an MS2-based RIP In the attempt to probe whether Nkx2-2as functions through assay designed to pull down endogenous Nkx2-2as–associated tethering the aforementioned miRNAs, we found that muta- miRNAs (Fig. 4A; Supplementary Fig. S4). As conserved miRNAs tions in the putative miR-103a/107– or miR-548m–binding in mammals (27, 28), miR-103a and miR-107 were expressed in sites abrogated the capability of Nkx2-2as to upregulate the mouse cerebellum and MB tissues and human Daoy cells at copy predicted miRNA targets BTG2, LATS1, or LATS2, respectively, numbers comparable to those of Nkx2-2as (Fig. 4B). In Daoy cells, in Daoy cells (Fig. 5E). The targeting of BTG2 and LATS1 by we also detected abundant expression of miR-548m, a miRNA miR-103a/107 and the silencing of LATS2 by miR-548m were that is absent in rodents but has recently been annotated as verified in HEK293T cells using a luciferase reporter construct playing critical functional roles in human malignancies for the 30 UTRofthepredictedtargettranscripts(Fig.5F). (Fig. 4B; ref. 29). The sequestering of these miRNAs was abolished Knockdown of Dicer, an endoribonuclease required for miRNA by mutations in the putative miRNA-binding sites of Nkx2-2as biogenesis, failed to further improve the levels of BTG2 and (Fig. 4C). In addition, expression of luciferase from a construct LATS1/2, suggesting a high efficacy of Nkx2-2as to sequestering containing WT Nkx2-2as, but not the putative miRNA-binding these miRNAs from their target RNA transcripts (Fig. 5G). The site–depleted mutant in the 30 UTR, was remarkably inhibited by duplex of a miRNA and its target RNA is incorporated into an miR-103a/107 and miR-548 mimics (Fig. 4D). These data suggest RNA-induced silencing complex (RISC) containing the catalytic that Nkx2-2as functions as a sponge to tether miR-103/107 and component Argonaute 2 (Ago2; ref. 30). Indeed, RIP assay miR-548m in the cell. revealed that Nkx2-2as overexpression in Daoy cells increased ceRNAs protect RNA transcripts from degradation or trans- the enrichment of Ago2 on this lncRNA, but significantly lational repression by competing for shared miRNAs (26). decreased its enrichment on the BTG2 and LATS1/2 transcripts Bioinformatics-based predictions suggested that both BTG2 (Fig. 5H), suggesting that Nkx2-2as and the aforementioned and LATS1 are potential targets of miR-103/107, and LATS2 mRNAs compete for association with RISC. might serve as a target for miR-548m (Supplementary Fig. S5). We next investigated whether BTG2 and LATS1/2 account Consistent with this, both BTG2 and LATS1 transcript levels for the tumor-suppressive role of Nkx2-2as in MB. Knock- were reduced in mouse MB in comparison with normal cere- down of BTG2 rescued cell growth impaired by Nkx2-2as, bella (Fig. 5A; GEO accession no.: GSE85449). Immunohisto- in accordance with its alleviation of the G1-phase cell-cycle chemical staining validated the downregulation of BTG2 and arrest induced by Nkx2-2as (Fig. 5I–K). LATS1/2 are negative LATS proteins in mouse MB tissues (Fig. 5B). Knockdown of regulators of yes-associated protein 1 (YAP1) in the Hippo Nkx2-2as in CGNPs induced a significant downregulation of pathway (31). Accordingly, silencing of these factors caused a BTG2 and LATS1 (Fig. 5C), whereas Nkx2-2as overexpression concomitant upregulation of YAP1, and increased the increased the levels of BTG2 and LATS1 in Daoy cells (Fig. 5D). growth and invasive capability of Nkx2-2as–overexpressing

www.aacrjournals.org Cancer Res; 78(4) February 15, 2018 OF7

Zhang et al.

Figure 5. Nkx2-2as upregulates BTG2 and LATS1/2 by sequestering miR-103a/107 and miR-548m. A and B, qRT-PCR assay (A) and immunohistochemical staining (B) for the indicated proteins in MB tissues from transgenic mice and cerebellar tissues from age- and strain-matched wild-type mice (Ctrl). C, Immunofluorescent staining of CGNPs transfected with Nkx2-2as–targeted or scrambled (Ctrl) siRNAs. Shown are images of three (#1–3) representative microscope areas and quantification results of the fluorescence intensity. Bar, 50 mm. D, Immunofluorescent staining of Daoy cells infected with control RNA– or Nkx2-2as–overexpressing lentiviruses, followed by quantification of the fluorescence intensity. Bar, 50 mm. E, Western blot analysis of Daoy cells after infection with control RNA- or wild-type/mutant Nkx2-2as–overexpressing lentiviruses. F, pGL3 constructs containing intact or mutant 30 UTR of the indicated genes were cotransfected with scrambled or miRNA mimics into HEK293T cells. Luciferase activity was measured 48 hours after transfection. G, Top, Daoy cells were transfected with scrambled (Ctrl) or Dicer-targeted siRNAs, followed by Western blot analysis of cell lysates. Bottom, Daoy cells were infected with control RNA (Ctrl)– or Nkx2-2as–overexpressing lentiviruses, and/or transfected with indicated siRNAs, followed by Western blot analysis of cell lysates. H, RIP assay of the enrichment of Ago2 on the indicated transcripts, relative to a control enrichment with nonspecific IgG, in Daoy cells infected with control RNA– or Nkx2-2as–overexpressing lentiviruses. I, Daoy cells were transfected with scrambled (Ctrl) or BTG2-targeted siRNAs and were subjected to Western blot analysis. J and K, Daoy cells were infected with control RNA (Ctrl)– or Nkx2-2as–overexpressing lentiviruses, and/or transfected with indicated siRNAs, followed by CCK-8 assay for cell growth (J)andflow cytometry assay to assess cell-cycle distribution (K). L, Daoy cells were transfected with scrambled (Ctrl) or LATS1/2-targeted siRNAs and were subjected to Western blot analysis. M and N, Daoy cells were infected with control RNA (Ctrl)– or Nkx2-2as–overexpressing lentiviruses, and/or transfected with indicated siRNAs, followed by CCK-8 assay to assess cell growth (M) and Transwell invasion assay (N). Data are representative images or are expressed as the means SD of three independent experiments. , P < 0.05; , P < 0.01, compared with the control group. #, P < 0.05, compared with the scrambled siRNA group.

OF8 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Represses Shh Medulloblastoma

Figure 6. Shh/Gli2 downregulates Nkx2-2as by transcriptionally activating FoxD1. A, Schematic diagram showing the predicted FoxD1-binding sites in the 30 ﬂanking region of the Nkx2-2as gene, predicted by an online tool (http://jaspar.binf.ku.dk). B and C, Daoy cells were transfected with scrambled (Ctrl) or FoxD1-targeted siRNAs, followed by Western blot (B) and qRT-PCR (C) assays. D, ChIP assay to assess occupancy of FoxD1 on the promoter regions of Nkx2-2as and Nkx2-2 (as a negative control). E, Immunohistochemical staining of MB tissues from transgenic mice and cerebellar tissues of age- and strain-matched WT mice (Ctrl). F, Daoy cells were transfected with scrambled (Ctrl) or Gli2-targeted siRNAs and then subjected to Western blot analysis. G, ChIP assay for occupancy of Gli2 on the promoter regions of the indicated genes. TUBB3 served as a negative control, and CCND1 is a known target gene of Gli2. H and I, Daoy cells were transfected with scrambled (Ctrl) or Gli2-targeted siRNAs with or without a FoxD1 overexpression construct. Cells were then subjected to Western blot (H) and qRT-PCR (I) assays. Data are representative images or are expressed as the means SD of three independent experiments. , P < 0.05; , P < 0.01, compared with the control group.

MB cells (Fig. 5L–N). These data suggest that Nkx2-2as silencing of Gli2 impaired the expression of FoxD1 (Fig. 6F), suppresses the malignant phenotypes of MB cells via BTG2 consistent with the observation that Gli2 is enriched on the FoxD1 and LATS1/2. promoter (Fig. 6G), and that upregulation of Nkx2-2as in response to Gli2 knockdown was reversed by overexpression of Shh/Gli2 impairs Nkx2-2as by transcriptionally activating FoxD1 (Fig. 6H and I). Thus, Hedgehog signaling impedes Nkx2- FoxD1 in MB cells 2as expression in MB cells by upregulating the transcriptional We next investigated how Nkx2-2as is downregulated by repressor FoxD1. Hedgehog signaling in MB cells. We failed to detect enrichment of Gli2 on the putative Nkx2-2as promoter region, defined as the The Gli2/FoxD1/Nkx2-2as axis is involved in in vivo 1,000 bp upstream of the transcription start site of Nkx2-2as, pathogenesis of Shh-subtype MB ruling out direct transcriptional activation by Gli2 (Supplemen- We next evaluated the therapeutic potential of lentivirus-deliv- tary Fig. S6). Bioinformatics-based prediction revealed several ered Nkx2-2as in a xenograft mouse model. Nkx2-2as–overex- putative binding sites for FoxD1 in the regulatory region of pressing Daoy cells exhibited significantly retarded tumor growth Nkx2-2as (Fig. 6A). Consistent with the reported repressive role in vivo in comparison with the unmodified Daoy cells (Fig. 7A). In of FoxD1 in target gene transcription (32), knockdown of FoxD1 addition, intracerebellar administration of Nkx2-2as recombi- dramatically increased Nkx2-2as expression in MB cells (Fig. 6B nant lentiviruses resulted in prolonged survival of Neurod2- and C). The occupancy of FoxD1 on the Nkx2-2as promoter was SmoA1–transgenic mice (Fig. 7B). Consistently, SmoA1-transgen- validated by ChIP assay (Fig. 6D). These findings were in accor- ic mice treated with Nkx2-2as–overexpressing lentiviruses dance with the high expression of FoxD1 in Shh-driven mouse MB showed a relatively normal shape of skulls and high BMD com- in comparison with normal cerebellum tissues (Fig. 6E; GEO pared with those administered with control viruses (Fig. 7C). We accession no.: GSE85449, 2.86-fold, P ¼ 0.01). Meanwhile, then examined whether Gli2/FoxD1 repression of Nkx2-2as and

www.aacrjournals.org Cancer Res; 78(4) February 15, 2018 OF9

Zhang et al.

Figure 7. Correlation of FoxD1/Nkx2-2as pathway with Shh-subtype MBs in vivo. A, A xenograft MB model was established by s.c. injection of parental (Ctrl) or Nkx2-2as-overexpressing Daoy cells, which were modiﬁed to overexpress luciferase (n ¼ 5). Tumors were visualized 10 days after inoculation via bioluminescence imaging. B and C, Two-month-old SmoA1-transgenic mice were randomly grouped and received biweekly intracerebellar injection with control or Nkx2-2as recombinant lentiviruses. B, A Kaplan–Meier survival curve was plotted. Mouse head specimens were scanned with micro-CT with representative images shown and mean BMD plotted in C. D, Representative images for immunohistochemical and in situ hybridization assays of clinical Shh (top) and non-Shh (bottom) MB samples. E, Statistical analysis of the correlations of the indicated molecules, based on the scores of immunohistochemical staining and in situ hybridization. F, A working model for FoxD1/Nkx2-2as–mediated regulation of MB development. The Hedgehog pathway transcription factor Gli2 switches on expression of FoxD1, which subsequently represses the transcription of Nkx2-2as. Nkx2-2as functions as a ceRNA to sequester miR-103/107 and miR-548m, thereby derepressing the tumor suppressors BTG2 and LATS1/2. Thus, Gli2/FoxD1-mediated downregulation of Nkx2-2as contributes to the pathogenesis of Shh- subtype MBs. , P < 0.05, compared with the control group.

downstream effectors is associated with clinical MBs. Shh- Nkx2-2as was originally identified as an antisense transcript of subgroup MBs were identified by GAB1 staining (Supplementary the gene encoding the homeobox protein Nkx2-2, a transcription Fig. S7; ref. 33). GAB1-high MB tissues expressed high levels of factor involved in the differentiation of the ventral horn neurons FoxD1 and modest levels of Nkx2-2as, BTG2, and LATS1/2 (36). Unlike other endogenous antisense transcripts of protein- (Fig. 7D). Analysis of the staining intensities validated the corre- coding genes, Nkx2-2as induces a modest increase, rather than a lation among FoxD1, Nkx2-2as, BTG2, and LATS1/2 specifically decrease, in Nkx2-2 mRNA levels (25). However, we found that in Shh-subgroup MBs (Fig. 7E). These data suggest the involve- Nkx2-2as plays a tumor-suppressive role independent of Nkx2-2, ment of Gli2/FoxD1-regulated Nkx2-2as, BTG2, and LATS1/2 in consistent with the previous finding that in cis regulation of the pathogenesis of Shh signaling–driven MB. Nkx2-2 is dispensable for the prodifferentiation role of Nkx2-2as in neural stem cells (25). Furthermore, Shahi and colleagues Discussion found that Nkx2-2 is directly targeted and upregulated by Gli1 in MB cells (37), which in combination with our current findings Hyperactivation of the Shh pathway is a key driver of MB, the suggests that this genomic region is involved in the pathogenesis most common malignant brain tumor of childhood (1). Despite of Shh-type MBs. While both transcription factors are critically the established role of Gli transcription factors in canonical Shh involved in Shh signaling, previous studies suggested that Gli2 but signaling, the downstream molecular mechanisms that account not Gli1 is required for Shh signaling and mediates inappropriate for the development of Shh-type MBs remain to be elucidated (7). activation of the pathway (38), which is in agreement with our In this study, we found that the deregulation of a lncRNA, Nkx2- finding that Nkx2-2as is downregulated solely by Gli2 via 2as, contributed to carcinogenesis in the context of constitutively transcriptional activation of FoxD1. To this end, we also cannot active Shh signaling in cerebellar granule cells. Nkx2-2as functions rule out a direct repression of Nkx2-2as by Gli2, e.g., through as a ceRNA to sequester miR-103/107 and miR-548m, resulting in binding to a distal silencer yet to be identified on the regulatory the derepression of their targets BTG2 and LATS1/2, which act as region of Nkx2-2as. Although these data validated the indepen- tumor suppressors in MB (34, 35). Gli2 impairs Nkx2-2as expres- dent transcription of Nkx2-2 and Nkx2-2as, further studies are sion by transcriptionally activating FoxD1 and thereby facilitating required to reveal the patterns of potential interactions between the proliferation of CGNPs (Fig. 7F). These findings, together with Nkx2-2 and Nkx2-2as. In addition, future studies should attempt the report that Nkx2-2as promotes oligodendrocytic differentia- to determine whether Nkx2-2 and Nkx2-2as are concurrently but tion of neural progenitors (25), suggest that this lncRNA is inversely regulated by Shh signaling in specific types of cells, critically involved in cerebellar development and may act as a wherein they play opposing roles in the development or carci- barrier for malignant transformation of CGNPs. nogenesis of the cerebellum.

OF10 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Represses Shh Medulloblastoma

The recent characterization of increasing numbers of ceRNAs Shh and Hippo pathways in the context of brain malignancies. highlights the ubiquitousness of this mode of fine-tuning of gene Collectively, our results provide novel insights into the mechan- expression (39). We propose here that Nkx2-2as acts as a ceRNA to isms underlying Shh-driving MB pathogenesis and could there- maintain BTG2 and LATS1 expression in normal CGNP cells by fore facilitate the characterization of candidate targets for clinical tethering miR-103/107 and miR-548m (Fig. 7F), as evidenced by MB treatment. the recapitulation and repression of MB cell phenotypes by Nkx2- 2as deficiency and overexpression, respectively. These observa- Disclosure of Potential Conflicts of Interest tions are in accordance with previous reports that BTG2 plays an No potential conflicts of interest were disclosed. essential role in counteracting the development of Shh-type MB (34, 40). Similarly, crosstalk between the Hippo pathway and Shh Authors' Contributions signaling has been documented in the pathogenesis of MB Conception and design: L. Wang, L. Jia (35, 41). The Hippo signaling network is a highly conserved Development of methodology: X. Zhang, L. Wang Acquisition of data (provided animals, acquired and managed patients, pathway that regulates multifaceted cell behaviors to control provided facilities, etc.): Y. Zhang, T. Wang, S. Wang, Y. Xiong tissue homeostasis and organ size, and deregulation of the Hippo Analysis and interpretation of data (e.g., statistical analysis, biostatistics, pathway underlies the progression of various disorders including computational analysis): Y. Zhang, R. Zhang, X. Zhang, J. Zhao malignancies (31, 42). In the mammalian Hippo network, Writing, review, and/or revision of the manuscript: L. Jia stimuli originating from cell contact or extracellular matrix cause Administrative, technical, or material support (i.e., reporting or organizing a sequential activation of mammalian sterile 20-like 1/2 data, constructing databases): A.-G. Yang Study supervision: L. Wang, L. Jia (MST1/2) and LATS1/2 kinases, which inhibit the activity and elicit proteasomal degradation of YAP1, a key transcription factor Acknowledgments of the Hippo pathway (31). The pathogenesis of Shh-subgroup This work was supported in part by grants from the National Natural MB has been linked to hyperactivity of YAP1, which is ascribed to Science Foundation of China (81472631 to L. Jia and 81602517 to L. Wang). both YAP1 gene amplification and direct YAP1 upregulation by The authors thank Dr. Yingfeng Lei in the Department of Microbiology, Shh signaling independent of new protein synthesis (41). Con- Fourth Military Medical University for kindly providing the pMS2-GFP and sistent with this, we showed here that Shh maintains YAP1 levels pSL-MS2-24 plasmids as detailed in Materials and Methods. – in mouse MB cells via noncoding RNA mediated downregulation The costs of publication of this article were defrayed in part by the payment of fi of LATS1. The present nding that LATS2 is also downregulated by page charges. This article must therefore be hereby marked advertisement in a similar mechanism involving a primate-specific miRNA, miR- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 548m, adds weight to the evolutionarily reinforced role of these regulators in MB pathogenesis (43). Thus, these data suggest a Received June 5, 2017; revised October 13, 2017; accepted December 5, 2017; candidate mechanism underlying the interplay between the published OnlineFirst December 11, 2017.

References 1. Gajjar AJ, Robinson GW. Medulloblastoma-translating discoveries from 13. Tay Y, Rinn J, Pandolﬁ PP. The multilayered complexity of ceRNA crosstalk the bench to the bedside. Nat Rev Clin Oncol 2014;11:714–22. and competition. Nature 2014;505:344–52. 2. Martirosian V, Chen TC, Lin M, Neman J. Medulloblastoma initiation and 14. Ling H, Vincent K, Pichler M, Fodde R, Berindan-Neagoe I, Slack FJ, Calin spread: where neurodevelopment, microenvironment and cancer cross GA. Junk DNA and the long non-coding RNA twist in cancer genetics. pathways. J Neurosci Res 2016;94:1511–9. Oncogene 2015;34:5003–11. 3. Massimino M, Biassoni V, Gandola L, Garre ML, Gatta G, Giangaspero F, 15. Kenney AM, Rowitch DH. Sonic hedgehog promotes G(1) cyclin expres- Poggi G, et al. Childhood medulloblastoma. Crit Rev Oncol Hematol sion and sustained cell cycle progression in mammalian neuronal pre- 2016;105:35–51. cursors. Mol Cell Biol 2000;20:9055–67. 4. Skowron P, Ramaswamy V, Taylor MD. Genetic and molecular altera- 16. Hallahan AR, Pritchard JI, Hansen S, Benson M, Stoeck J, Hatton BA, Russell tions across medulloblastoma subgroups. J Mol Med (Berl) 2015;93: TL, et al. The SmoA1 mouse model reveals that notch signaling is critical for 1075–84. the growth and survival of sonic hedgehog-induced medulloblastomas. 5. Markant SL, Wechsler-Reya RJ. Review: personalized mice: modelling the Cancer Res 2004;64:7794–800. molecular heterogeneity of medulloblastoma. Neuropathol Appl Neuro- 17. Gong C, Maquat LE. lncRNAs transactivate STAU1-mediated mRNA biol 2012;38:228–40. decaybyduplexingwith30 UTRs via Alu elements. Nature 2011; 6. Phoenix TN, Patmore DM, Boop S, Boulos N, Jacus MO, Patel YT, Roussel 470:284–8. MF, et al. Medulloblastoma genotype dictates blood brain barrier pheno- 18. Beckham CJ, Light HR, Nissan TA, Ahlquist P, Parker R, Noueiry A. type. Cancer Cell 2016;29:508–22. Interactions between brome mosaic virus RNAs and cytoplasmic proces- 7. Pak E, Segal RA. Hedgehog signal transduction: key players, oncogenic sing bodies. J Virol 2007;81:9759–68. drivers, and cancer therapy. Dev Cell 2016;38:333–44. 19. Bao W, Fu HJ, Jia LT, Zhang Y, Li W, Jin BQ, Yao LB, et al. HER2-mediated 8. Lee RT, Zhao Z, Ingham PW. Hedgehog signalling. Development upregulation of MMP-1 is involved in gastric cancer cell invasion. Arch 2016;143:367–72. Biochem Biophys 2010;499:49–55. 9. Dellovade T, Romer JT, Curran T, Rubin LL. The hedgehog pathway and 20. Wu TJ, Chang SS, Li CW, Hsu YH, Chen TC, Lee WC, Yeh CT, et al. Severe neurological disorders. Annu Rev Neurosci 2006;29:539–63. hepatitis promotes hepatocellular carcinoma recurrence via NF-kappaB 10. Rimkus TK, Carpenter RL, Qasem S, Chan M, Lo HW. Targeting the sonic pathway-mediated epithelial-mesenchymal transition after resection. Clin hedgehog signaling pathway: review of smoothened and GLI inhibitors. Cancer Res 2016;22:1800–12. Cancers (Basel) 2016;8:. 21. Paxinos G, Franklin KBJ. The Mouse Brain in Stereotactic Coordinates, 3rd 11. Quinn JJ, Chang HY. Unique features of long non-coding RNA biogenesis ed. Academic Press, 2008. and function. Nat Rev Genet 2016;17:47–62. 22. Garg N, Bakhshinyan D, Venugopal C, Mahendram S, Rosa DA, 12. Geisler S, Coller J. RNA in unexpected places: long non-coding RNA Vijayakumar T, Manoranjan B, et al. CD133þ brain tumor-initiating cells functions in diverse cellular contexts. Nat Rev Mol Cell Biol 2013;14: are dependent on STAT3 signaling to drive medulloblastoma recurrence. 699–712. Oncogene 2017;36:606–17.

www.aacrjournals.org Cancer Res; 78(4) February 15, 2018 OF11

Zhang et al.

23. Read TA, Fogarty MP, Markant SL, McLendon RE, Wei Z, Ellison DW, Febbo medulloblastoma in Patched1 heterozygous mice by inhibiting the PG, et al. Identification of CD15 as a marker for tumor-propagating cells in Cxcl3-dependent migration of cerebellar neurons. J Neurosci 2012;32: a mouse model of medulloblastoma. Cancer Cell 2009;15:135–47. 15547–64. 24. Lang PY, Nanjangud GJ, Sokolsky-Papkov M, Shaw C, Hwang D, Parker JS, 35. Rao R, Salloum R, Xin M, Lu QR. The G protein Galphas acts as a tumor Kabanov AV, et al. ATR maintains chromosomal integrity during postnatal suppressor in sonic hedgehog signaling-driven tumorigenesis. Cell Cycle cerebellar neurogenesis and is required for medulloblastoma formation. 2016;15:1325–30. Development 2016;143:4038–52. 36. Lovrics A, Gao Y, Juhasz B, Bock I, Byrne HM, Dinnyes A, Kovacs KA. 25. Tochitani S, Hayashizaki Y. Nkx2.2 antisense RNA overexpression Boolean modelling reveals new regulatory connections between transcrip- enhanced oligodendrocytic differentiation. Biochem Biophys Res Com- tion factors orchestrating the development of the ventral spinal cord. PLoS mun 2008;372:691–6. One 2014;9:e111430. 26. Karreth FA, Pandolfi PP. ceRNA cross-talk in cancer: when ce-bling rivalries 37. Shahi MH, Afzal M, Sinha S, Eberhart CG, Rey JA, Fan X, Castresana JS. go awry. Cancer Discov 2013;3:1113–21. Regulation of sonic hedgehog-GLI1 downstream target genes PTCH1, 27. Chen PS, Su JL, Cha ST, Tarn WY, Wang MY, Hsu HC, Lin MT, et al. miR-107 Cyclin D2, Plakoglobin, PAX6 and NKX2.2 and their epigenetic status in promotes tumor progression by targeting the let-7 microRNA in mice and medulloblastoma and astrocytoma. BMC Cancer 2010;10:614. humans. J Clin Invest 2011;121:3442–55. 38. Bai CB, Auerbach W, Lee JS, Stephen D, Joyner AL. Gli2, but not Gli1, is 28. Chen HY, Lin YM, Chung HC, Lang YD, Lin CJ, Huang J, Wang WC, et al. required for initial Shh signaling and ectopic activation of the Shh pathway. miR-103/107 promote metastasis of colorectal cancer by targeting the Development 2002;129:4753–61. metastasis suppressors DAPK and KLF4. Cancer Res 2012;72:3631–41. 39. Ergun S, Oztuzcu S. Oncocers: ceRNA-mediated cross-talk by sponging 29. Lwin T, Zhao X, Cheng F, Zhang X, Huang A, Shah B, Zhang Y, et al. A miRNAs in oncogenic pathways. Tumour Biol 2015;36:3129–36. microenvironment-mediated c-Myc/miR-548m/HDAC6 amplification 40. Farioli-Vecchioli S, Tanori M, Micheli L, Mancuso M, Leonardi L, Saran A, loop in non-Hodgkin B cell lymphomas. J Clin Invest 2013;123:4612–26. Ciotti MT, et al. Inhibition of medulloblastoma tumorigenesis by 30. Holoch D, Moazed D. RNA-mediated epigenetic regulation of gene expres- the antiproliferative and pro-differentiative gene PC3. FASEB J 2007;21: sion. Nat Rev Genet 2015;16:71–84. 2215–25. 31. Yu FX, Zhao B, Guan KL. Hippo pathway in organ size control, tissue 41. Fernandez LA, Northcott PA, Dalton J, Fraga C, Ellison D, Angers S, Taylor homeostasis, and cancer. Cell 2015;163:811–28. MD, et al. YAP1 is amplified and up-regulated in hedgehog-associated 32. Fetting JL, Guay JA, Karolak MJ, Iozzo RV, Adams DC, Maridas DE, Brown medulloblastomas and mediates Sonic hedgehog-driven neural precursor AC, et al. FOXD1 promotes nephron progenitor differentiation by repres- proliferation. Genes Dev 2009;23:2729–41. sing decorin in the embryonic kidney. Development 2014;141:17–27. 42. Nagashima S, Bao Y, Hata Y. The hippo pathway as drug targets in cancer 33. Kaur K, Kakkar A, Kumar A, Mallick S, Julka PK, Gupta D, Suri A, et al. therapy and regenerative medicine. Curr Drug Targets 2017;18:447–54. Integrating molecular subclassification of medulloblastomas into routine 43. Koufaris C.Human and primate-specific microRNAs in cancer: Evolution, clinical practice: a simplified approach. Brain Pathol 2016;26:334–43. and significance in comparison with more distantly-related research mod- 34. Farioli-Vecchioli S, Cina I, Ceccarelli M, Micheli L, Leonardi L, Ciotti MT, els: the great potential of evolutionary young microRNA in cancer research. De Bardi M, et al. Tis21 knock-out enhances the frequency of Bioessays 2016;38:286–94.

OF12 Cancer Res; 78(4) February 15, 2018 Cancer Research

Nkx2-2as Suppression Contributes to the Pathogenesis of Sonic Hedgehog Medulloblastoma

Yimeng Zhang, Ting Wang, Shan Wang, et al.

Cancer Res Published OnlineFirst December 11, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-1631

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2017/12/09/0008-5472.CAN-17-1631.DC1

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2018/02/01/0008-5472.CAN-17-1631. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.