Antagomir-17-5p Abolishes the Growth of Therapy-Resistant Neuroblastoma through p21 and BIM

We identified a key oncogenic pathway underlying neuroblastoma progression: specifically, MYCN, expressed at elevated level, transactivates the miRNA 17-5p-92 cluster, which inhibits p21 and BIM translation by interaction with their mRNA 3′ UTRs. Overexpression of miRNA 17-5p-92 cluster in MYCN-not-amplified neuroblastoma cells strongly augments their in vitro and in vivo tumorigenesis. In vitro or in vivo treatment with antagomir-17-5p abolishes the growth of MYCN-amplified and therapy-resistant neuroblastoma through p21 and BIM upmodulation, leading to cell cycling blockade and activation of apoptosis, respectively. In primary neuroblastoma, the majority of cases show a rise of miR-17-5p level leading to p21 downmodulation, which is particularly severe in patients with MYCN amplification and poor prognosis. Altogether, our studies demonstrate for the first time that antagomir treatment can abolish tumor growth in vivo, specifically in therapy-resistant neuroblastoma.


Introduction
MicroRNAs (miRNAs or miRs) are conserved ,22 nucleotide non-coding RNAs: they repress protein expression at posttranscriptional level [1][2][3], mainly by annealing with the 39 UTR of the target mRNA, thus interfering with its translation and/or stability [4]. MiRNAs play important roles in the regulation of basic cell functions, including proliferation, differentiation and apoptosis [5][6][7][8]. Importantly, oncogenesis has been linked to deregulated expression of miRNAs, which act as tumor suppressors or oncomirs [9,10] and may contribute to tumor invasion [11].
Neuroblastoma, accounting for 8-10% of pediatric tumors, originates from precursor cells of the peripheral nervous system. The most aggressive neuroblastomas are characterized by diverse genetic aberrations, including MYCN amplification, chromosome 1p deletion and unbalanced gain of chromosome 17q [19]. MYCN amplification occurs in 25% of the cases and correlates to both an aggressive phenotype and treatment failure [19]. The neuroblastoma progression linked to MYCN amplification, although well documented, is mediated by unknown molecular mechanisms. Some miRNAs, including miR-9, miR-125 and miR-34a control neuroblastoma cell proliferation in vitro [20,21], but their function has not been linked to neuroblastoma carrying MYCN amplification. p21 Cip1/Waf1/Sdi1 (hereafter referred to as p21), the founding member of the Cip/Kip family of cyclin-dependent kinase (CDK) inhibitors, negatively regulates cell cycle progression by inhibiting a broad range of cyclin/Cdk complexes [22]. Specifically, p21 prevents cell cycle progression from G1 to S phase by inactivating the Cdk2-cyclin E complexes, that, in turn, inhibit the tumor suppressor protein retinoblastoma (pRb) required for entering Sphase [23]. It can also act as a tumor suppressor, as demonstrated by the higher susceptibility of p21-deficient mice to develop spontaneous tumors [24]. In spite of this, genetic alterations of p21 are rare in human tumor samples [25], suggesting that its oncogenic function is mostly mediated by a deregulated expression.
BIM (Bcl-2 interacting mediator of cell death) is one of the most potent pro-apoptotic BH3-only proteins: it binds to all pro-survival Bcl-2 family members with high affinity [26,27], thereby releasing Bax or Bak proteins, the critical downstream effectors of the Bcl-2dependent pathway of apoptosis [28]. BIM is a tumor suppressor gene, as demonstrated by the accelerated Myc-induced lymphomagenesis in Em-myc mice lacking BIM [29] and increased tumorigenesis of BIM-/epithelial cells [30] .
In this study, we have investigated the molecular mechanisms underlying MYCN-induced neuroblastoma progression. Our findings indicate that an enhanced MYCN level, linked or not to MYCN amplification, transactivates the miRNA 17-5p-92 cluster at transcriptional level. The upmodulation of miR-17-5p mediates the oncogenic properties of MYCN, through a direct suppression of p21 and BIM translation. Of particular interest is that treatment of MYCN-amplified neuroblastoma with antagomir-17-5p can abolish tumor growth, not only in vitro but also in vivo.

MYCN transcriptionally regulates the expression of miRNA 17-5p-92 cluster in neuroblastoma
Transcription of the miRNA 17-5p-92 cluster is induced by the oncogene c-Myc [14]. Since c-Myc and MYCN share some transcriptional target, we hypothesized that MYCN may transactivate this cluster.
To verify this hypothesis, we first analyzed the expression of the miRNA 17-5p-92 cluster in five neuroblastoma cell lines expressing MYCN at either low level (SH-EP and SK-N-AS) or high level due to MYCN overexpression (SH-SY-5Y) or amplification (LAN-5 and IMR32) ( Figure 1A). Most miRNAs encoded by the miRNA 17-5p-92 cluster were expressed at higher levels in cells overexpressing MYCN or carrying MYCN amplification, as evaluated by both Northern blot and qRT-PCR ( Figure 1B and data not shown).
Thereafter, analysis of the genomic region surrounding the miRNA 17-5p-92 cluster for putative MYCN binding sites (including the canonical E-box CACGTG and the non-canonical sequences CATGTG, CACGGG and CAAGTG) revealed the presence of five MYCN putative binding sites, four upstream and one downstream the cluster ( Figure 1C). To demonstrate the interaction between MYCN and the cluster promoter, chromatin immunoprecipitation (ChIP) experiments were performed using the Tet-21/N cell line, which expresses MYCN under a tetracycline-regulated promoter [31]. DNA immunoprecipitated with an anti-MYCN antibody from untreated or doxycyclinetreated (for 2 or 24 h) cells was PCR-amplified using five different pairs of oligonucleotide primers encompassing the five putative MYCN-binding sites ( Figure 1C). MYCN was associated with all the putative binding sites; furthermore, the signal decreased in parallel with MYCN downmodulation by doxycycline treatment. These results demonstrate the in vivo binding of MYCN with all these sites ( Figure 1D).
Finally, we evaluated the effect of MYCN on miRNA 17-5p-92 cluster expression. A ,3700 bp DNA fragment containing the four upstream MYCN binding sites was cloned at the 59 site of the luciferase gene in the reporter pGL4 vector (pGL4prom17M) ( Figure 1C). Co-transfection of this construct together with an expression vector for MYCN in SH-EP cells led to a sharp transactivation of the luciferase activity, which was not detected in the control empty vector group ( Figure 1E). Moreover, in Tet-21/ N cells downmodulation of MYCN by doxycycline treatment caused a marked decrease of all miRNAs encoded by the miRNA 17-5p-92 cluster, as evaluated by both Northern blot and qRT-PCR ( Figure 1F and data not shown).
Altogether these results indicate that, in neuroblastoma cells, MYCN transcriptionally induces the expression of the miRNA 17-5p-92 cluster by directly binding to its promoter. miRNA 17-5p-92 cluster enhances the in vitro and in vivo tumorigenesis of SK-N-AS neuroblastoma cell line In order to investigate the oncogenic role of the miRNA 17-5p-92 cluster in neuroblastoma, we generated a stable SK-N-AS transfectant expressing the cluster under a CMV promoter (SK-N-AS 17-5p cluster). These cells showed an increased expression of the cluster (data not shown), similar to that observed in MYCNamplified neuroblastoma cells ( Figure 1B).
In a series of in vitro experiments, ectopic expression of the cluster induced: (i) an increase of the proliferation rate , as compared to control cells stably transfected with an empty vector (SK-N-AS Cont) (Figure 2A, B); (ii) a decline of the percentage of cells in G1 phase and an inverse rise of the S phase population, demonstrating an accelerated cell cycle progression ( Figure 2C); (iii) an increase of the number of colonies formed in a soft-agar semisolid medium, demonstrating a rise of the tumorigenic ability ( Figure 2D).
To evaluate the effect of the cluster on in vivo tumorigenesis, we injected SK-N-AS Cont or SK-N-AS 17-5p cluster cells into nude mice. All mice (11/11) injected with SK-N-AS 17-5p cluster cells developed a tumor and died within 130 days from injection, whereas only 36% of mice (4/11) injected with SK-N-AS Cont cells showed a visible tumor ( Figure 2E).
Altogether, these data demonstrate that the miRNA 17-5p-92 cluster enhances cell proliferation and promotes tumorigenesis, both in vitro and in vivo.
miR-17-5p is responsible for the tumorigenic effect of the miRNA 17-5p-92 cluster through direct downmodulation of p21 To identify potential target genes of the miRNAs encoded by the miRNA 17-5p-92 cluster, we used two algorithms, Target Scan [32] and PicTar [33]. Both indicated that p21 is a candidate target of miR-17-5p and -20a. Since miR-17-5p and -20a share similar sequences and functions [7,14], we focused on miR-17-5p for further analysis of p21 targeting.
We first evaluated the expression of p21 in Tet-21/N cells treated or not with doxycycline for 96 h. Downmodulation of MYCN and miR-17-5p caused a strong increase of p21 expression at both mRNA and protein level ( Figures 1F and S1). Consistently, we observed a clear reduction in the level of endogenous p21 mRNA and protein in SK-N-AS 17-5p cluster cells, as well as in SK-N-AS transiently transfected with miR-17-5p, but not with miR-92 ( Figure 3A).
SK-N-AS transfected with miR-17-5p showed a higher percentage of cells in the S phase and a lower number of cells in G1 phase, similarly to SK-N-AS 17-5p cluster cells, whereas miR-92 did not affect cell cycle progression ( Figures 2C and 3C). Since regulation of p21 expression by miRNA 17-5p-92 cluster is essentially mediated by miR-17-5p, we hypothesized that the effects of the cluster on the in vitro tumorigenesis of SK-N-AS cells were also mediated by miR-17-5p. In fact, overexpression of miR-17-5p, but not of miR-92, increased the number of colonies formed by SK-N-AS in a semisolid medium, as observed for cells overexpressing the entire cluster ( Figures 2D and 3D). Altogether, these results show that miR-17-5p is the major effector of MYCNmediated in vitro tumorigenesis of SK-N-AS cells.
The role of p21 in the control of cell cycle progression and tumorigenesis of SK-N-AS cells was demonstrated by knocking down p21 with siRNA. Notably, silencing of p21 was associated with an accelerated cell cycle progression, as well as an increased ability of these cells to form colonies in a semisolid medium, as observed upon overexpression of either the miRNA 17-5p-92 cluster or miR-17-5p ( Figure 3E and data not shown). Importantly, restoration of p21 in SK-N-AS 17-5p cluster cells abolished the in vitro tumorigenic activity of these cells by blocking miRNA 17-5p-92 cluster-induced cell cycle acceleration ( Figure 3F, G). Finally, overexpression of miR-17-5p in SH-EP cells (a MYCN-notamplified neuroblastoma cell line) enhanced cell proliferation through downmodulation of p21, as observed in SK-N-AS cells, thus demonstrating that these effects were not restricted to a particular cell line (data not shown).
Altogether, these results show that downregulation of p21 mediates miR-17-5p induced tumorigenesis in neuroblastoma cell lines.

Knockdown of miR-17-5p decreases the in vitro tumorigenesis of MYCN-amplified neuroblastoma cells
To determine whether miR-17-5p mediates tumorigenesis in MYCN-amplified neuroblastoma cells, we evaluated the effect of miR-17-5p knockdown in LAN-5 cell line, which expresses miR-17-5p at elevated level ( Figure 1B). In preliminary experiments, treatment of LAN-5 with antagomir-17-5p (a chemically modified anti-miR-17-5p oligonucleotide conjugated with cholesterol [34]) efficiently downmodulated miR-17-5p expression, as compared to cells treated with PBS or control antagomir-1 (data not shown). Incubation of LAN-5 with antagomir-17-5p, but not with the control antagomir, markedly inhibited cell proliferation, decreased in vitro tumorigenesis in soft agar and blocked cell cycle progression ( Figure 4A, B and data not shown). Furthermore, inhibition of miR-17-5p was associated with an increase of p21 mRNA and protein level ( Figure 4C).
In addition, knockdown of miR-17-5p strongly promoted both early and late apoptosis ( Figure 4D). Notably, p21 overexpression did not induce apoptosis in LAN-5 (data not shown), thus implying that an additional miR-17-5p target may regulate this process. Bioinformatic analysis with Target Scan [32], Pic Tar [33] and miRanda [35] indicated the pro-apoptotic factor BIM as a putative target of miR-17-5p. Consistently, knockdown of miR-17-5p by antagomir increased BIM expression at both mRNA and protein level in LAN-5 cells ( Figure 4E). To demonstrate that this regulation occurs through a direct binding of miR-17-5p to BIM 39 UTR, we cloned a portion of the BIM 39 UTR containing the miR-17-5p putative binding site into the pGL3-Promoter vector, downstream the luciferase gene (pGL3-Prom-BIMUTR-wt). As a control, we cloned a region of BIM 39 UTR containing a mutated miR-17-5p recognition site (pGL3-Cont-BIM UTR-mut). Cotransfection of pGL3-Prom-BIMUTR-wt together with an anti miR-17-5p, but not with a scrambled anti-miR-17-5p oligonucle- otide, caused an increase of the luciferase activity in Tet-21/N cells ( Figure 4F). This increase was not observed in cells transfected with the pGL3-Cont-BIM UTR-mut construct, thus demonstrating that mutation of the miR-17-5p binding site in the BIM 39 UTR abolishes the ability of miR-17-5p to regulate its expression ( Figure 4F).
Knockdown of miR-17-5p inhibits the in vivo tumorigenic ability of MYCN-amplified neuroblastoma cells Based on the in vitro studies, we hypothesized that abolition of miR-17-5p expression may inhibit tumor growth in vivo. To address this critical question, MYCN-amplified LAN-5 cells were injected into nude mice, and tumors thereby generated were treated with antagomir-17-5p or a control antagomir for two weeks. Injection of antagomir-17-5p dramatically inhibited tumor growth: this effect , already relevant after one week of therapy was maintained through the end of the treatment, leading in 30% of cases to complete regression of the tumor mass ( Figure 5A, B). Conversely, administration of the control antagomir did not affect tumor development, as observed in PBS-treated tumors (data not shown). Tumor analysis at 24 h after the first administration of antagomir-17-5p showed a marked downmodulation of miR-17-5p, associated with a strong increase of p21 and BIM at both mRNA and protein level ( Figure 5C, D and data not shown). Consistently, TUNEL assay showed an increased apoptosis in tumors treated with antagomir-17-5p, as compared to the control group ( Figure 5C, D).
Altogether, these results demonstrate that in vivo treatment of MYCN-amplified neuroblastoma with antagomir-17-5p abolishes tumor growth by upmodulation of p21 and BIM and increased apoptosis.

Discussion
Neuroblastoma is one of the most common extra-cranial solid tumor of early childhood, accounting for .15% of cancer-related deaths in children. While the clinical diversity of neuroblastoma correlates with several genetic features, such as ploidy or allelic loss, the amplification of the MYCN gene is the best genetic marker of poor prognosis [36]. However, the mechanisms underlying MYCN-mediated neuroblastoma progression have not been identified. Our work describes a novel oncogenic pathway underlying neuroblastoma development, whereby MYCN transactivates the miRNA 17-5p-92 cluster, which in turn downmodulates the tumor suppressors p21 and BIM. Among the different miRNAs pertaining to the miRNA 17-5p-92 cluster, miR-17-5p and miR-20a, which show an almost complete homology, are the only ones targeting p21, as indicated by bioinformatic analysis and luciferase assay. Furthermore, miR-17-5p level in neuroblastoma cell lines is often more elevated than that of miR-20a. Therefore, we focused on miR-17-5p as the primary effector of MYCN-mediated tumorigenesis.
Our studies indicate that, in neuroblastoma cell lines, miR-17-5p controls cell cycle progression through p21. In diverse tumors miR-106b, structurally related to but functionally distinct from miR-17-5p [7], regulates cell cycle progression through p21 [37]. However, we observed that miR-106b is not upmodulated in MYCN-amplified neuroblastoma cells (data not shown), thus suggesting that in neuroblastoma miR-106b is not involved in p21 regulation. p21 is a tumor suppressor gene, whose expression is mainly regulated at transcriptional level by p53 [38]. Although p53 is the most frequently mutated gene in human cancers, p53 mutations have not been detected in primary neuroblastoma [39]. In addition, lack of correlation between p53 and p21 levels in MYCN-amplified neuroblastoma cell lines [40] suggests that expression of p21 may be p53-independent in neuroblastoma. Our study shows a novel p53-independent mechanism for the regulation of p21 expression. In fact, we report that miR-17-5p directly regulates p21 in both the p53 knockout SK-N-AS cell line and the SH-EP and LAN-5 cells expressing endogenous p53 [41]. Furthermore, treatment of LAN-5 with antagomir-17-5p caused an upmodulation of p21 without affecting p53 levels (data not shown). Id2 has also been proposed to mediate MYCN ability to bypass the cell cycle checkpoint imposed by Rb [42]. However, a direct binding of MYCN to Id2 promoter has not been demonstrated and Id2 expression does not seem to be associated with MYCN amplification or expression in human neuroblastoma [43].
Treatment of LAN-5 with antagomir-17-5p causes not only a block of cell cycle, but also a dramatic apoptosis. Tumor progression usually occurs through activation of different pathways, leading to increased cell proliferation and protection from apoptosis, which provides a survival advantage to cancer cells. Oncoproteins of the myc family promote apoptosis [44][45][46]. However, MYCN amplification in neuroblastoma causes resistance to chemotherapy, associated with tumor progression and poor prognosis [36]: this suggests that MYCN-induced apoptosis may be inhibited by an additional oncogenic mechanism, crucial for tumor progression. Our findings indicate that miR-17-5p is a key factor inducing protection from MYCN-primed apoptosis in neuroblastoma. Indeed, knock down of miR-17-5p is sufficient to promote massive apoptosis of MYCN-amplified LAN-5 cells. This occurs through upmodulation of the proapoptotic factor BIM, mediated by direct binding of miR-17-5p to BIM mRNA 39 UTR.
The ''combinatorial circuitry model'', predicts that a single miRNA may target multiple mRNAs. In MYCN-amplified neuroblastoma MYCN transactivates miR-17-5p, which in turn accelerates cell cycle progression by downmodulating p21 and protects cells from apoptosis by inhibiting BIM expression (Figure 7). This occurs in a p53-independent manner, thus providing a mechanistic explanation for the rarity of p53 mutations in neuroblastoma.
A second set of studies was focused on the molecular analysis of primary neuroblastoma samples, derived from a well characterized series of patients. The samples were divided in three groups. The first and second ones yielded expected results. Specifically, the first group is characterized by MYCN-amplification, poor prognosis and therapy resistance, coupled with a marked rise of miR-17-5p level and a dramatic decrease of p21. The second one features low levels of MYCN and miR-17-5p, associated with elevated p21 expression: these patients show a relatively benign clinical profile, characterized by slow disease progression and therapy response. Surprisingly, the third, large group of samples shows an elevated level of miR-17-5p in the absence of increased MYCN expression. The levels of p21 are intermediate between those of the first and the second groups, while the clinical features are relatively benign. In line with these clinical observations, we observed that, in neuroblastoma cell lines with normal MYCN level, overexpression of miRNA 17-5p-92 cluster, as well as miR-17-5p alone, is able to enhance tumorigenesis. It is apparent, therefore, that miR-17-5p does not require MYCN to exert its oncogenic activity in neuroblastoma. Notably, knock down of miR-17-5p by antagomir sharply inhibits the in vitro and in vivo tumorigenesis of MYCN-amplified LAN-5 cells, suggesting that miR-17-5p is a key oncogenic factor in both MYCN-amplified and not-amplified neuroblastoma.
The lower level of p21 observed in the first group, as compared to the third one, may be due to a direct downmodulation of p21 by MYCN. In fact, c-Myc negatively regulates the expression of p21 at transcriptional level [47]. Preliminary results indicate that MYCN binds the p21 promoter, suggesting that in MYCNamplified neuroblastoma MYCN exerts an additional suppressing activity on p21 expression at transcriptional level (data not shown). In the clinical setting, the remarkably low level of p21 in the second group is probably linked to its poor prognosis and therapy resistance.
Our antagomir studies bear a potential significance at clinical level. Despite recent advances in treatment options, aggressive neuroblastoma carrying MYCN amplification is refractory to current therapy, leading to a disease related mortality of up to 70%. Therefore, the development of new therapeutic approaches is warranted. Strategies based on MYCN repression by siRNA may prove unsatisfactory: in fact, efficient inhibition of MYCN levels is hampered by the high levels of MYCN. Conversely, antagomir-17-5p treatment may be beneficial in MYCN-amplified neuroblastoma, also in view of evidence suggesting that systemic antagomir treatment is not coupled with significant toxicity [34,48]. In conclusion, our results provide the first demonstration that antagomirs can efficiently inhibit tumor growth in vivo, thus raising the possibility that these molecules may ultimately be clinically useful in the treatment of cancer.

Cell culture and tumor samples
The SH-EP and SK-N-AS human neuroblastoma cell lines express low levels of MYCN whereas SH-SY-5Y, LAN-5 and IMR32 overexpress MYCN (LAN-5 and IMR32 as a result of gene amplification). All the cells, obtained from the American Type Culture Collection, (Manassas, VA), were grown in RPMI medium supplemented with 10% FBS (HyClone, Logan, Utah). Tumor samples were obtained from patients diagnosed with neuroblastoma after informed consent of their parents admitted to the Division of Oncology at Bambino Gesù Children's Hospital. Samples were freshly resected during surgery and immediately frozen in liquid nitrogen for subsequent total RNA extraction. Tumors were classified according to the International Neuroblastoma Pathology Classification (INCP): 6 were at stage I, 2 at stage II, 1 at stage III, and 8 at stage IV.
For proliferation assay, cells were seeded at the same density (2.5610 4 cells/ml) and counted at the indicated times. For thymidine incorporation assay, 50610 3 cells/well were plated in 96 well plates in triplicated; 24 h after seeding, each well was incubated with 1 mCi of [ 3 H] thymidine (Amersham Biosciences). After 16 h, the cells were harvested and analysed by liquid scintillation in a Microcounter (Wallac). The counts from triplicate wells were averages.
The in vitro tumorigenesis of neuroblastoma cell lines was determined by seeding the cells at low density (400-600 cells/ml) in 1.5 ml of 0.3% Agar Noble (Difco, Kansas City, Missouri) and RPMI-10% FBS and plating them on 1.5 ml of 0.6% Agar Noble and RPMI-10% FBS. Colony formation was determined after two weeks by staining with crystal violet (Fluka, St. Gallen, Switzerland) and colonies were counted visually. SK-N-AS Cont and SK-N-AS 17-5p cluster cells were obtained by transfection of the SK-N-AS cells with the empty vector or the miRNA 17-5p-92 cluster expression constructs, followed by two weeks of blasticidin selection at 0.5 mg/ml. SK-N-AS 17-5p cluster cells stably expressing the p21 expression vector or the empty pcDNA3, were obtained by two weeks of hygromicin selection at 100 mg/ml, after transfection with Lipofectamine 2000 (Invitrogen, Carlsbad, CA).

Cell cycle analysis
Cells were seeded in 6 well plates at 40% of confluence and incubated at 37uC for 24 h. Cells were then synchronized by serum depletion for 30 h and pulsed with 10 mM BrdU (Sigma) for 30 min at different times after FCS addition. After BrdU incorporation, cells were harvested and fixed in ice-cold 70% ethanol. DNA was denatured with HCl 2N/Triton 20% and labeled with an anti-BrdU antibody (BD Bioscience) for 1 h. Then, cells were resuspended in washing buffer and labeled with antimouse APC-conjugated antibody (Beckton Dickinson). Labeled cells were washed and resuspended in PBS containing 5 mg/ml propidium iodide and analysed on a FACSCanto flow cytometer (Beckton Dickinson) using the DIVA software. All the flow cytometry experiments were performed at least twice and a representative experiment is shown in each figure.

Tumorigenicity assay in nude mice
Six-week-old nude mice strain C57/BL6 were subcutaneously injected into the right flank with 25610 6 cells (SKNAS Cont, SKNAS 17-5p cluster or LAN-5). Tumor size was assessed every two days by caliper measurement. Tumor volume was calculated as follow: volume = Dxd 2 6p/6, where D and d are the longer and the shorter diameters, respectively. For survival analysis, mice were sacrificed when tumors reached the volume of 500 mm 3 .

In vitro treatment of LAN-5 with antagomir
Antagomirs were synthesized as described [34]. Sequences were 59-a s c s uaccugcacuguaagcacu s u s u s g s -Chol 39 (antagomir-17-5p), 59-u s a s cauacuucuuuacauu s c s c s a s -Chol 39 (control antagomir-1). Lower case letters represent 29-O-Methyl-modified oligonucleotides, subscript 's' represents a phosphorothioate linkage, and 'Chol' represents linked cholesterol.
LAN-5 cells were seeded in antibiotic-free media at 50-60% of confluence (6610 5  In vivo administration of antagomir-17-5p in tumors generated by LAN-5 neuroblastoma cells 25610 6 LAN5 cells were subcutaneuosly injected into the flank of 6-8-week-old athymic nude mice. After one week, when the tumors reached an average volume of ,150 mm 3 , the tumorbearing nude mice were treated with antagomir-17-5p. 100 ml of antagomir-17-5p (diluted in PBS at 2 mg/ml), or control antagomir, or PBS were injected intratumorally three times per week for two weeks. Tumor diameters were measured at regular intervals as described above.

DNA constructs
The pGL4prom17M construct was obtained by sequential cloning into the pGL4.10 vector (Promega) of 3 fragments amplified by PCR from human genomic DNA. A 1081 base pair fragment (clone A: nucleotides -3731-2900) was first amplified with the forward 17-92promAfor (59-ATAGGTACCCCGGAAT-TTCCTGAACCACAATG-39) and the revers 17-92promArev (59-GATCTCGAGGGAGTAGCCGCCACCATCTTCGGCT-39) primers. The obtained DNA was then digested with KpnI and XhoI and cloned in pGL4.10 vector. A second segment of the cluster promoter (segment B: nucleotides -2649-1425) was obtained with the primers: 17-92promBfor 59-GATCTCGAGTCCTGGTGAG-TCTGCCCGCCCCT-39 and 17-92promBrev 59-GATAGATC-TAACACCCGAGACTGCAAAGTGCCCG-39 and it was cloned downstream the fragment A by double digestion with XhoI and BglII. The last portion of the promoter (fragment C: nucleotides -1423-1) was obtained by digestion with BglII and HindIII of the pGL4prom17 construct [7] and cloned upstream of the luciferase coding sequence.

Real-time RT-PCR
For mRNA analysis, total RNA was purified with TRIZOL Reagent (Invitrogen). Reverse transcription and Real-time PCR were performed as described [7]. Expression of mature miRNAs was determined using miRNA-specific quantitative real-time PCR (qRT-PCR; Applied Biosystems, Foster City, CA). U6 snRNA was used for normalization.

Oligonucleotides and transfection experiments
For transfection experiments, cells were seeded in antibiotic-free media for 24 h and then transfected with Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen, Carlsbad, CA). Transfection of a pool of four siRNA oligonucleotides specifically targeting p21 or BIM mRNAs (Smart Pool siRNA, Dharmacon) was performed with Hiperfect (Qiagen, Hilden, Germany).
In promoter assays experiments, SHEP cells were transfected with 0.6 mg of firefly luciferase vectors (empty pGL4 or pGL4prom17M vector), in combination with 1.8 mg of pcDNA3 (Promega corporation, Madison, WI) or pIRV neo SV-MycN, together with a Renilla luciferase vector (50 ng) as internal control.
In luciferase experiments, Tet-21/N cells were transfected with 0.4 mg of firefly luciferase vectors (empty pGL3-prom, pGL3prom-p21UTR wt or mutant, pGL3-prom-BIM UTR wt or del) Immunohistochemistry Expression of p21 and BIM was analyzed by immunohistochemistry on 5-mm slices formalin-fixed paraffin-embedded sections of tumor xenografts by using monoclonal antibody to p21 (Cell Signaling, Danvers, MA) and polyclonal antibody to BIM (ProSci Incorporated, Poway, CA). Antigen was retrieved by pretreating dewaxed sections in a microwave oven at 750 W for 5 minutes in citrate buffer (pH 6) and processing them with a Super Sensitive Link-Labeled Detection System (Biogenex, Menarini, Florence, Italy). The enzymatic activity was developed using 3-amino-9-ethylcarbazole (AEC, Dako, Milan, Italy) as a chromogenic substrate. Following counterstaining with Mayer's haematoxylin, slides were mounted in aqueous mounting medium (glycergel, Dako).

Statistical analysis
Data are presented as mean and error bars indicate the standard deviation (s.d.) or the standard error (s.e.m.). The groups were compared by one-way analysis of variance (Anova, Chicago, IL) using Bonferroni's test or by the unpaired t-test with two-tailed p value. Survival data are presented as Kaplan-Meyer plots and were analysed using a log-rank (Mantel-Haenszel) method. Significance level was P,0.05.