Cellular and Molecular Life Sciences (2019) 76:2231–2243 https://doi.org/10.1007/s00018-019-03041-4 Cellular and Molecular Life Sciences Functional high‑throughput screening reveals miR‑323a‑5p and miR‑342‑5p as new tumor‑suppressive microRNA for neuroblastoma Aroa Soriano1 · Marc Masanas1 · Ariadna Boloix1,2 · Núria Masiá3 · Laia París‑Coderch1 · Olga Piskareva4 · Carlos Jiménez1 · Kai‑Oliver Henrich5 · Josep Roma1 · Frank Westermann5 · Raymond L. Stallings4 · Constantino Sábado6 · Josep Sánchez de Toledo1,6 · Anna Santamaria3 · Soledad Gallego1,6 · Miguel F. Segura1 Received: 1 October 2018 / Revised: 28 January 2019 / Accepted: 4 February 2019 / Published online: 15 February 2019 © The Author(s) 2019 Abstract Current therapies for most non-infectious diseases are directed at or affect functionality of the human translated genome, barely 2% of all genetic information. By contrast, the therapeutic potential of targeting the transcriptome, ~ 70% of the genome, remains largely unexplored. RNA therapeutics is an emerging field that widens the range of druggable targets and includes elements such as microRNA. Here, we sought to screen for microRNA with tumor-suppressive functions in neuro- blastoma, an aggressive pediatric tumor of the sympathetic nervous system that requires the development of new therapies. We found miR-323a-5p and miR-342-5p to be capable of reducing cell proliferation in multiple neuroblastoma cell lines in vitro and in vivo, thereby providing a proof of concept for miRNA-based therapies for neuroblastoma. Furthermore, the combined inhibition of the direct identified targets such as CCND1, CHAF1A, INCENP and BCL-XL could reveal new vulnerabilities of high-risk neuroblastoma. Keywords Pediatric cancer · Non-coding RNA · High-throughput screening · 14q32 · Epigenetics · Cancer therapy Introduction Approximately 15,000 new cases of pediatric cancers are diagnosed yearly in Europe, with around 10% corresponding to neuroblastoma (NB), an embryonal tumor of the sympa- thetic nervous system [1]. NB accounts for 15% of all can- Aroa Soriano and Marc Masanas contributed equally to this work. cer-related deaths in children, and is the embryonal tumor with the lowest 5-year relative survival [2]. Significant Electronic supplementary material The online version of this article (https​://doi.org/10.1007/s0001​8-019-03041​-4) contains improvements are only foreseen in the field of new targeted supplementary material, which is available to authorized users. * Miguel F. Segura (VHIR)-Universitat Autònoma de Barcelona (UAB), Passeig miguel.segura@vhir.org Vall d’Hebron 119, 08035 Barcelona, Spain 4 1 Molecular and Cellular Therapeutics, Royal College Group of Translational Research in Child and Adolescent of Surgeons in Ireland and National Children’s Research Cancer, Vall d’Hebron Research Institute (VHIR)- Centre Our Lady’s Children’s Hospital, Dublin, Ireland Universitat Autònoma de Barcelona (UAB), Passeig 5 Vall d’Hebron 119‑129, Collserola Building. Lab 207, Neuroblastoma Genomics Group, German Cancer 08035 Barcelona, Spain Research Center (DKFZ), Im Neuenheimer Feld 280, 2 69120 Heidelberg, Germany Institut de Ciència de Materials de Barcelona 6 (ICMAB-CSIC) and Nanomol Technologies SA, Mòdul de Pediatric Oncology and Hematology Department, Hospital Recerca B, Campus UAB, 08193 Bellaterra, Spain Universitari Vall d’Hebron-Universitat Autònoma 3 de Barcelona (UAB), Passeig Vall d’Hebron 119, Cell Cycle and Cancer Laboratory, Biomedical Research 08035 Barcelona, Spain Group in Urology, Vall d’Hebron Research Institute 13 Vol.:(0123456789) 2232 A. Soriano et al. therapies and personalized medicine programs which, at pre- results support the use of miRNA-based restoration therapies sent, are effective for a small number of patients. Therefore, as an alternative tool against NB resistant to conventional new approaches must be considered. therapies. RNA therapeutics is an emerging field that widens the range of druggable targets and includes elements such as small interference RNA (i.e., siRNA, shRNA and micro- Materials and methods RNA). MicroRNA (miRNA) are small non-coding RNA that interfere with the translation and stability of coding mRNA Cell lines in a sequence-specific manner [3]. Mounting evidence shows miRNA to be deregulated and functionally contributing to SK-N-AS, SH-SY5Y, IMR-32 and HEK293T cell lines the development and progression of different human can- were purchased from American Type Culture Collection cers, including NB [4, 5]. An overall reduction in miRNA is (ATCC, Manassas, VA, USA), CHLA-90 cell line from the observed in advanced NB, mainly due to alterations in the Children’s Oncology Group Cell Culture and Xenograft miRNA-processing machinery [6]; therefore, miRNA res- Repository (Lubbock, TX, USA). SK-N-BE(2) and LA1-5s toration represents an attractive novel therapeutic approach. acquired from Public Health England Culture Collections To identify miRNA with therapeutic potential in NB, we (Salisbury, UK). All cell lines purchased from the tissue carried out high-throughput functional screening of 2048 banks were amplified and stored in liquid nitrogen. Upon miRNA mimics. Several miRNA with potential tumor- resuscitation, cells were maintained in culture for no more suppressive functions were identified, among which miR- than 2 months. SK-N-AS, SK-N-BE(2), SH-SY5Y, IMR- 323a-5p and miR-342-5p had the highest therapeutic poten- 32, CHLA-90 and LA1-5s were cultured and maintained in tial in multiple NB cell lines in vitro and in vivo. These Iscove’s modified Dulbecco’s medium (Life Technologies, Fig. 1  Functional high-throughput miRNA screening identified sev- transfected cells and standardized using the Z score transformation eral tumor-suppressive miRNA. a Screening design. b Graph repre- method. MiRNA previously reported as tumor-suppressive miRNA in senting the effects of 2048 individual miRNA on cell proliferation. NB are indicated. c Pie chart representing the genomic distribution The percentage of growth inhibition was obtained by comparing the of miRNA that were capable of reducing cell proliferation ~ 50% (Z average of three independent miRNA mimics replicates with mock- score < − 2) 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2233 Thermo Fisher Scientific) supplemented with 10% heat-inac- in technical triplicates using Lipofectamine 2000 (Life tech- tivated fetal bovine serum (South America Premium, Bio- nologies, Thermo Fisher Scientific, Madrid, Spain, 0.2 µL west) and 1% of insulin–transferrin–selenium Supplement per well) using the Robotic Platform Caliper Sciclone (Cali- (Life Technologies, Thermo Fisher Scientific). HEK293T per Life Sciences, Perkin Elmer, Waltham, Massachusetts, were grown in Dulbecco’s modified Eagle’s medium (Life USA). At 96-h post-transfection, cells were fixed with 1% Technologies, Thermo Fisher Scientific) supplemented with glutaraldehyde (Sigma-Aldrich) and stained with 0.5% crys- 10% heat-inactivated fetal bovine serum. All media were tal violet (Sigma-Aldrich, Madrid, Spain). Crystals were dis- supplemented with 100 U/mL penicillin, 100 µg/mL strep- solved with 15% acetic acid (Fisher Scientific) and absorb- tomycin (Life Technologies, Thermo Fisher Scientific) and ance was measured at 590 nm using an Epoch Microplate 5 μg/mL plasmocin (InvivoGen). All cultures were main- Spectrophotometer (Biotek, Winooski, Vermont, USA). tained at 37 °C in a saturated atmosphere of 95% air and 5% ­CO2. All cell lines were frequently tested for mycoplasma Screening statistics contamination. Data quality control and analysis of different factors (e.g., MicroRNA functional high‑throughput library miRNA position, transfection time) was performed using screening “R” Statistical Software (Supplementary Fig. 1C). The absorbance value of each mock- or miRNA-trans- Chemoresistant SK-N-BE(2) cells were seeded in 96-well fected well was normalized to the median of all non- plates at 5000 cells/well using the MultidropV2 dis- transfected values of the corresponding replicate plate. To penser (Finstruments). Twenty-four hours later cells were evaluate the effect of each miRNA on cell proliferation, the transfected with a microRNA library consisting of 2048 absorbance value (3 replicates) versus the median of all human miRNA mimics (Dharmacon, Lafayette, CO, USA, mock values was compared. The statistical significance was ­miRIDIAN® microRNA Library—Human Mimic (19.0) assessed by Student’s t test and the p value was adjusted by CS-001030 Lot 13112, GE Healthcare, 25 nM each miRNA) the false discovery rate “FDR” method. The percentage of Fig. 2  Ectopic expression of miRNA located at 14q32 reduced cell compared to mock-transfected cells. b, d Relative expression ­(log2) of proliferation in multiple NB cell lines. a Heatmap representing effects miR-342 (b) and miR-323a (d) in MNA and non-MNA NB tumors. c, of the indicated miRNA on cell proliferation reduction in MYCN e Overall survival Kaplan–Meier plot of the indicated miRNA expres- amplified (MNA) and MYCN non-amplified (non-MNA) cell lines. sion in human NB tissues The overexpression effect of each miRNA on cell proliferation was 13 2234 A. Soriano et al. 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2235 ◂Fig. 3  MiR-323a-5p induced cell cycle arrest at G0/G1 phase. Cell with tumor-suppressive functions in NB. Individual miRNA cycle flow cytometry analysis of the indicated cell lines transfected mimics were transfected into the SK-N-BE(2) cells (Fig. 1a). with miR-control, miR-323a-5p or miR-342-5p at 25 nM for 96 h. a One representative histogram of three independent experiments. b, c Control miRNA (i.e., cel-miR-67 and cel-miR-239b) were Histograms representing the average percentage of living cell popula- used as negative controls and miR-497-5p as a positive con- tion from three independent experiments of SK-N-BE(2) (b) or SK- trol [7, 8] (Supplementary Fig. 1a,b). The overexpression of N-AS (c) cells in G0/G1, S, or G2/M phases. *p < 0.05, **p < 0.01, 52 miRNA was found to reduce ~ 50% cell proliferation (Z two-tailed Student’s t test. d, e Western blot of the indicated cell cycle regulatory proteins in SK-N-BE(2) (d) and SK-N-AS (e) transfected score < − 2, adjusted p value < 0.05) (Fig. 1b; Supplemen- with miR-control, miR-323a-5p and miR-342-5p (25 nM) at 72 h and tary Tables 1, 2). 96 h post-transfection Since it has been suggested that miRNA with similar functions cluster together [9], we examined the genomic distribution of miRNA whose overexpression produced proliferation was standardized using the Z score equation the highest reduction on cell proliferation. Interestingly, 7 Z = x−𝜇 𝜎 , where x is the value of cell proliferation after trans- of the 52 miRNA (13.5%) were located at 14q32 (Fig. 1c fection of each single miRNA, µ is the mean cell prolifera- and Supplementary Table 2), a locus with high miRNA tion of all miRNA and σ the standard deviation. density and frequently lost or silenced in different types of tumors, including NB [10]. Mouse xenograft SK-N-AS cells (4 × 10 6) were reverse transfected with Restoration of miRNA located at 14q32 reduced cell 25 nM of miR-control, miR-323a-5p or miR-342-5p in 100- viability in multiple NB cell lines mm plates (25 µL lipofectamine/dish). SK-N-BE(2) cells (4.7 × 106) were reverse transfected with 25 nM of miR- The tumor-suppressive effects of miRNA hits located at control, miR-323a-5p or miR-342-5p in T175 flasks (90 µL 14q32 (i.e., miR-380-5p, miR-665, miR-541-3p, miR- lipofectamine/flask). Thirty-six hours post-transfection, 299-3p, miR-654-5p, miR-323a-5p and miR-342-5p) 4 × 106 cells/flank of SK-N-AS and 5 × 106 cells/flank of SK- were further confirmed in an extended panel of six NB N-BE(2) were injected into the right flank of 6- to 8-week-old cell lines bearing genomic alterations associated with female NMRI-nude mice (n = 13 mice/condition of SK-N-AS resistance to standard NB therapies and poor patient out- and n = 15 mice/condition of SK-N-BE(2)) (Janvier Labs, Le come (Supplementary Table 3). The overexpression of all Genest-Saint-Isle, France) in 300 µL of PBS:Matrigel (1:1). miRNA tested reduced cell proliferation in multiple NB Tumor volume was measured every 2–3 days. At the end cell lines compared with mock- and miR-control-trans- of the experiment, the primary tumors were removed and fected cells being miR-323a-5p and miR-342-5p those weighted. Part of the tumors were fresh frozen and the rest ones with the highest therapeutic potential (Fig. 2a). fixed in 10% formalin and embedded in paraffin. Low miR‑323a‑5p expression levels correlate with poor patient outcome Statistical methods The expression levels of miR-323a-5p and miR-342-5p Unless otherwise stated, mean ± SEM values are repre- were analyzed in a large cohort of NB tissue samples and sentative of the average of three independent experiments. correlated with clinical parameters. Whereas lower expres- Statistical significance was determined by unpaired two- sion levels of miR-342-5p did not correlate with the clinical tailed Student’s t test (GraphPad Prism Software, La Jolla, parameters analyzed (Fig. 2b–c), low miR-323a-5p levels CA, USA.). * Means p < 0.05, ** means p < 0.01 and *** were found to correlate with MYCN genomic amplification means p < 0.001. (Fig. 2d) and shorter overall survival (Fig. 2e). Ectopic expression of miR‑323a‑5p and miR‑342‑5p Results halted cell cycle progression and induced apoptosis in NB Functional high‑throughput miRNA screening identified several miRNA with tumor‑suppressive To clarify whether the reduction in cell number upon miR- activity 323a-5p and miR-342-5p overexpression was due to inhibi- tion of cell proliferation and/or induction of cell death, cell High-throughput screening using the largest library of cycle progression was analyzed using flow cytometry. The miRNA mimics available was performed to identify miRNA overexpression of miR-323a-5p induced a modest increase 13 2236 A. Soriano et al. Fig. 4  MiR-323a-5p and miR-342-5p overexpression induced apop- and/or fragmented chromatin. Data represent mean ± SEM of tosis in NB cells. a, b Analysis of chromatin fragmentation/conden- three independent experiments (n = 3 per experiment). *p < 0.05, sation in SK-N-BE(2) (a) and SK-N-AS (b) transfected with 25 nM **p < 0.01, two-tailed Student’s t test. c, d Representative Western of miR-control, miR-323a-5p or miR-342-5p 96 h post-transfection. blot analysis of apoptosis-related proteins in SK-N-BE(2) (c) and SK- Images on the right show a representative field of NB cells stained N-AS (d) transfected with miR-control, miR-323a-5p or miR-342-5p with Hoechst dye. White arrowheads point to cells with condensed (25 nM) at 72 h and 96 h post-transfection 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2237 in the subG1 peak and an increment in G0/G1-phase popula- levels decreased in both cell lines when cells were trans- tion accompanied by a reduction in the S and G2/M phases. fected with miR-323a-5p (Fig. 5e). On the other hand, only Therefore, both cell cycle arrest and increased cell death CCND1 and BCL-XL protein levels were reduced in both might be contributing to the miR-323a-5p overexpression cell lines when cells were transfected with miR-342-5p, phenotype (Fig. 3a–c). Concurring with these observations, whereas no differences were found in AKT2 levels. Results a reduction in cyclins, D1, E1 and B1 was observed after for MKNK2 were not conclusive (Fig. 5f). miR-323a-5p overexpression at 72 h post-transfection and To ascertain whether the potential targets were regulated a consequent reduction in phospho-RB levels at later time directly by miR-323a-5p or miR-342-5p, we engineered points. Furthermore, upregulation of the cell cycle inhibitor luciferase-reporter vectors with the 3′UTR of genes bearing p27 was also observed (Fig. 3d, e). the putative miRNA-binding sites (Supplementary Fig. 3). MiR-342-5p-transfected cells showed no apparent dif- The reporter vectors were co-transfected with a miR-con- ferences regarding cell cycle progression compared to trol, miR-323a-5p or miR-342-5p and luciferase activity miR-control cells, but did show an accumulation of cells in was quantified 24 h after transfection. The overexpression subG1, thereby indicating cell death induction (Fig. 3a–c). of miR-323a-5p caused a reduction in luciferase activity in In this case, reduced levels of cyclins, D1, E1 and B1 the CHAF1A, INCENP, CCND1 and FADD 3′UTR reporter and phospho-RB, but not increased p27 levels, were also vectors and no differences were observed for KIF11 and observed (Fig. 3d, e), suggesting that cells committed to die CDC25A, thereby indicating that these last two genes are not were unable to progress through the cell cycle. direct targets (Fig. 5g). On the other hand, reduction in the Next, we examined the chromatin status of miRNA- luciferase activity of CCND1 and BCLX 3′UTR vectors by transfected cells by Hoechst staining. MiR-323a-5p and miR-342-5p overexpression confirmed both genes as direct miR-342-5p transfected cells showed a higher percentage miR-342-5p targets (Fig. 5h). of cells with condensed and fragmented chromatin stain- ing, one of the typical hallmarks of apoptosis (Fig. 4a, b). CCND1, CHAF1A and INCENP silencing mostly Furthermore, NB cells transfected with miR-323a-5p and reproduced miR‑323a‑5p overexpression effects miR-342-5p showed cleavage of the executor caspase-3 and one of its targets α-FODRIN, both indicative of apoptotic The contribution of each target to the miRNA overexpression cell death (Fig. 4c, d). phenotype was analyzed by siRNA-mediated gene silenc- ing. Silencing of CHAF1A, KIF11 and CCND1 showed MiR‑323a‑5p and miR‑342‑5p target cell cycle a reduction in cell proliferation similar to that induced by and survival genes miR-323a-5p, while INCENP and FADD silencing only produced a moderate effect (Fig. 6a). Western blot analysis To find the downstream mediators of miR-323a-5p and miR- 72 h post-transfection confirmed siRNA efficacy (Fig. 6a, 342-5p phenotypic effects in NB, a miRNA-target analysis lower panels). Overall, CCND1 depletion mirrored the best was performed comparing five different platforms (Fig. 5a, miR-323a-5p overexpression, not only the general effects b). For miR-323a-5p, 905 potential targets were predicted on cell proliferation, but also on the reduction in phospho- by at least four independent algorithms, whereas 1380 RB levels and p27 accumulation (Fig. 6b). However, when were predicted for miR-342-5p (Supplementary Tables 4, apoptosis induction was analyzed, CHAF1A and INCENP 5). According to the Kyoto Encyclopedia of Genes and were the direct targets that better phenocopied the effects of Genomes (KEGG) pathway analysis [11], a significant miR-323a-5p overexpression (Fig. 6c). These results suggest number of these predicted targets were associated with cell that the combination of CCND1, CHAF1A and INCENP cycle or commonly altered pathways in cancer (Fig. 5c, d, inhibition is sufficient to reproduce the therapeutic effects p value < 0.05). These genes were selected for validation of miR-323a-5p. based on their expression association with NB outcome and/or previous reported functional role in cancer (Sup- Inhibition of CCND1 and BCL‑XL partially reproduces plementary Tables 6, 7). Transient miRNA overexpression the effects of miR‑342‑5p proved to consistently reduce the mRNA levels of several miR-323a-5p (i.e., CHAF1A, KIF11, INCENP, CDC25A, The same strategy was used to characterize the contribu- CCND1, FADD and E2F2) and miR-342-5p (i.e., AKT2, tion of miR-342-5p targets. In this case, only the inhibition CCND1, MKNK2 and BCLX) predicted targets (Supple- of CCND1 showed a similar reduction in cell proliferation mentary Fig. 2a–d). To confirm the reduction also in pro- (Fig. 6d). In agreement with this observation, only CCND1 tein levels, Western blot was performed at 48 h and 72 h silencing caused a reduction in phospho-RB levels and post-transfection of miR-323a-5p or miR-342-5p. CHAF1A, accumulation of the cell cycle inhibitor p27 (Fig. 6e). How- KIF11, INCENP, CDC25A, CCND1 and FADD protein ever, apoptosis induction was not observed with silencing 13 2238 A. Soriano et al. 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2239 ◂Fig. 5  MiR-323a-5p and miR-342-5p modulated the expression of tumor cells overcome more easily the miR-323a-5p tumor- multiple cancer-related genes. a, b Venn diagram representing the suppressive effects. overlap of predicted target genes among five miRNA-target predic- tion algorithms of miR-323a-5p or miR-342-5p. c, d Representation of the functional annotation of predicted target genes of miR-323a-5p (c) or miR-342-5p (d) using KEGG pathways and Gene Ontology Discussion databases. Red bars indicate selected pathways to analyze potential miRNA targets. e, f Representative Western blot of predicted target genes in SK-N-BE(2) and SK-N-AS transfected with miR-control, Despite the emerging personalized medicine programs for miR-323a-5p (e) or miR-342-5p (f) at 48 h and 72 h post-transfection. pediatric tumors, which include molecular profiling of NB, g, h Luciferase 3′UTR reporter assays. Graph represents luciferase the small number of recurrent somatic mutations found at activity in HEK-293T cells co-transfected with 50 ng/well of the indi- diagnosis and the lack of established tumor drivers, limit the cated reporter vectors and 25 nM of miR-control, miR-323a-5p (g) or miR-342-5p (h). Data represented the average ± SEM of three inde- therapeutic opportunities for NB patients. On the other hand, pendent experiments (n = 3 per experiment). *p < 0.05, **p < 0.01, some patients relapse and develop drug resistance when ***p < 0.001 treated with anticancer drugs targeting single molecules, which might be due to the activation of alternative pathways. MiRNA alterations have been shown to participate in of either CCND1 or BCL-XL (Fig. 6f) suggesting that the the progression and outcome of NB (reviewed in [12]). observed effects of miR-342-5p on cell death might require We propose the use of miRNA as multi-target approach to the inhibition of both or additional targets. overcome treatment resistance in NB patients, given that a single miRNA can inhibit multiple targets involved in MiR‑323a‑5p and miR‑342‑5p overexpression tumorigenic processes. reduced tumor growth in vivo Multiple pharmaceutical companies have already included miRNA in their developmental drug pipelines We proceeded to validate the capacity of miR-323a-5p and (e.g., miR-34 [13, 14], miR-16 [15], miR-122 [16]). The miR-342-5p to reduce tumor growth in vivo by subcutaneous first challenge in applying this approach to the treat- injection of SK-N-BE(2) or SK-N-AS cells previously trans- ment of high-risk NB is to identify those miRNA with fected with miR-control, miR-323a-5p or miR-342-5p into the highest therapeutic potential. Our work is the largest the flank of NRMI-nude mice. At the end of the experiment, high-throughput functional screening performed to date to all groups showed a similar percentage of tumor incidence identify miRNA with tumor-suppressive functions in can- (~ 80–100%, Fig. 7a, e), which suggested that miRNA over- cer and particularly for NB. Of the 2048 miRNA tested, we expression did not interfere with tumor engraftment. Tumor identified 52 that significantly reduced cell proliferation growth was monitored every 2–3 days for 28 days for SK-N- with a Z score lower than − 2 (~ 50% of cell proliferation). BE(2) or 18 days for SK-N-AS. While in SK-N-AS, only the Concurring with previous reports, multiple tumor-suppres- overexpression of miR-342-5p was capable of delaying cell sive miRNA such as members of the let-7 miRNA tumor- growth in vivo, both miRNA were effective in SK-N-BE(2) suppressive family (e.g., let-7i-5p, let-7a-3p, let-7f-2-5p, cells, showing a final ~ twofold reduction in tumor growth at let-7g-5p or let-7e-3p [17, 18]) were also identified in our the end of the experiment (Fig. 7b, f). These effects were con- screening, thereby validating our approach to identifying firmed by the analysis of tumor weight (Fig. 7c, d, g, h-Sup- tumor-suppressive miRNA. plementary Fig. 4a, b). Notably, 7 of the 52 miRNA (13.5% of the total) shown In summary, our results confirm miR-323a-5p and miR- to reduce cell proliferation ~ 50% are encoded in the 14q32 342-5p as potential therapeutic tools against high-risk NB. Of genomic region. This region is frequently altered in cancer note, miR-323a-5p overexpression impaired tumor growth in and particularly in NB [10], which raises the possibility the MYCN-amplified cell line, but not in SK-N-AS, suggest- that this region contains tumor-suppressor genes. All 7 ing that higher doses of miR-323a-5p or being functional for miRNA identified in our screening showed the capacity to longer periods of time may be needed to achieve a therapeutic reduce cell proliferation in multiple NB cell lines in vitro, effect. In support of this hypothesis, miRNA levels were ana- with miR-323a-5p and miR-342-5p being those with the lyzed at the beginning and at the end of experiment. Both cell highest therapeutic potential. lines achieved similar levels of miRNA overexpression at the In agreement with our results, miR-323a-5p has been time of injection (Supplementary Fig. 5a, c) ruling out the reported to be downregulated in NB patients with MYCN possibility of differential transfection efficiency between cell amplification and associated with unfavorable outcome lines. However, at the end of experiment, SK-N-AS maintained [19, 20]. Moreover, miR-323a-5p overexpression has been much lower levels of miR-323a-5p in SK-N-AS but not in reported to reduce proliferation and promote apoptosis of SK-N-BE(2) (Supplementary Fig. 5b, d), thus suggesting that human cerebral glioma cells [21]. We found that ectopic this miRNA is processed faster in this cell line and therefore, expression of miR-323a-5p suppressed cell growth through 13 2240 A. Soriano et al. Fig. 6  MiRNA target-knockdown partially reproduced the anti- tion. c, f Western blot analysis of apoptosis-related proteins in SK- tumoral effects of miR-323a-5p and miR-342-5p. a, d Cell viability N-BE(2) and SK-N-AS transfected with miR-323a-5p, miR-342-5p assay of SK-N-BE(2) and SK-N-AS transfected with 25 nM of miR- or the indicated siRNA (25 nM) at 96 h post-transfection. Graph rep- 323a-5p, miR-342-5p or the indicated siRNA. Target protein knock- resents one of three independent experiments (n = 6 per experiment). down was analyzed by Western blot at 72 h post-transfection (lower Asterisk compares miR-323a-5p or miR-342-5p versus miR-Control panels). b, e Expression of some representative cell cycle regulatory and hash compares each siRNA versus control siRNA. * or #p < 0.05, proteins in SK-N-BE(2) and SK-N-AS transfected with miR-323a-5p, ** or ##p < 0.01, *** or ###p < 0.001, two-tailed Student’s t test miR-342-5p or the indicated siRNA (25 nM) at 96 h post-transfec- 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2241 A B 1500 C D 1500 miR-Control * miR-323a-5p # 15/15 1250 miR-342-5p 100 Tumour volume (mm ) 13/14 Tumour weight (mg) 3 # * Tumour incidence (%) 11/14 1000 1000 80 750 60 # * 40 500 500 # * 20 250 0 0 0 miR-Control miR-323a-5p miR-342-5p 6 9 12 15 18 21 24 27 30 miR-Control miR-323a-5p miR-342-5p miR-Control miR-323a-5p miR-342-5p Days E F 1500 G H 1500 miR-Control ** miR-323a-5p ns 15/15 1250 miR-342-5p ** Tumour volume (mm ) 100 3 Tumour weight (mg) 13/15 13/15 Tumour incidence (%) 80 1000 1000 60 750 *** 40 500 500 ** 20 250 0 0 0 miR-Control miR-323a-5p miR-342-5p 6 9 12 15 18 miR-Control miR-323a-5p miR-342-5p miR-Control miR-323a-5p miR-342-5p Days Fig. 7  MiR-323a-5p and MiR-342-5p overexpression reduced tumor times per week for ~ 3 weeks. c, g Representative images of dis- growth in vivo. a, e Tumor formation incidence in SK-N-BE(2) (a) sected tumors (bar indicates 1 cm). d, h Tumor weight at 18 days and SK-N-AS (e). b, f Tumor growth of xenografts derived from post-injection. Hash compares miR-323a-5p versus control and aster- SK-N-BE(2) (b) SK-N-AS (f) cells transfected with miR-control, isk compares miR-342-5p versus control. * or #p < 0.05, **p < 0.01, miR-323a-5p or miR-342-5p. Tumor measurements were taken three ***p < 0.001, two-tailed Student’s t test the direct modulation of several cell cycle-associated Although the best known functions of CCND1 are genes such as CCND1, CHAF1A, INCENP and FADD related to cell cycle control, CCND1 has CDK-independ- and caused a G1 cell cycle arrest followed by induction ent functions. For example, CCND1 regulates cell differ- of apoptosis. The silencing of the miR-323a-5p direct entiation by binding to several transcription factors such targets CHAF1A, INCENP and CCND1 were those that as the estrogen receptor α (ERα) and the androgen recep- more closely recapitulate the phenotypic effects of miR- tor (AR) (reviewed in [27]). Therefore, a direct inhibition 323a-5p overexpression. CCND1 is a well-established of CCND1 could be therapeutically more effective than oncogene frequently overexpressed and associated with CDK4/6 inhibition. poor outcome in different types of tumors including NB Another of the newly identified relevant miR-323a-5p [25]. CCND1 interacts with CDK4 and CDK6 and the target is CHAF1A, a chromatin modifier protein recently activation of this complex phosphorylates RB and other involved in maintaining the undifferentiated state of highly transcription factors which promote cell cycle progression aggressive NB [28]. Thus, in addition to halting cell cycle [26]. To date, the best approach to targeting CCND1 is progression, miR-323a-5p could also be relevant for NB through the use of CDK4/6 inhibitors such as palbociclib, therapy through repressing the expression of genes that ribociclib and abemaciclib. block the differentiation of NB cells. 13 2242 A. Soriano et al. The silencing of INCENP phenocopied the apoptosis Acknowledgements We thank members of Laboratory Animal Ser- induction of miR-323a-5p overexpression. INCENP (inner vice and Statistics and Bioinformatics Unit of Vall d’Hebron Research Institute and the staff of Biomolecular Screening and Protein Technolo- centromere protein) is a component of the chromosomal pas- gies Unit of Centre for Genomic Regulation. We thank Ms. Christine senger complex (CPC), a complex that regulates mitosis. O’Hara for text correction. INCENP is a scaffolding subunit for the CPC and activates Aurora B kinase [29]. Similar to CHAF1A, high levels of Author contributions AS and MFS designed the study. MM, AS, AB, INCENP in primary NB tumors are associated with poor LPC, CJ and MFS performed the experiments. NM and AS collabo- rated in cell cycle analysis. AS, MM, OP, RLS, FW, KOH and CS prognosis (Supplementary Table 6). In addition, silenc- analyzed the data. JST, SG provided intellectual support and expertise ing of INCENP using doxycycline-inducible shRNA led toward interpretation of results. MM and MFS wrote the original manu- to significant decreases in growth of NB xenografts and script. All authors contributed to the edition and the critical review of increases mice survival (Sun et al. Advances in Neuroblas- the manuscript. toma Research 2018). Therefore, targeting INCENP could Funding The funding was received by Ministerio de Educación, be a novel promising therapy in NB. Cultura y Deporte (Grant no. FPU16/01099), Instituto de Salud Car- Our in vivo studies suggested that the transient overex- los III (Grant no. CP11/00052), Instituto de Salud Carlos III (Grant pression of miR-323a-5p was enough to halt the proliferation no. CP16/00006), Instituto de Salud Carlos III (Grant no. PI14/00561), of MYCN-amplified NB cells but not in SK-N-AS xenograft Instituto de Salud Carlos III (Grant no. PI14/00647), Instituto de Salud Carlos III (Grant no. PI17/00564), Instituto de Salud Carlos III (Grant models, probably due to a higher rate of miRNA processing. no. RD12/0036/0016), FP7 People: Marie-Curie Actions (Grant no. At that point, and owing to technical limitations, the fact that MIRNATHERAPYNBL_PEOPLE-2013-GIC), Fundación Científica more sustained overexpression of miR-323a-5p could have a Asociación Española Contra el Cáncer (Grant no. AECC-JB-2012-01), better therapeutic response cannot be ruled out. AGAUR (Grant no. 2014 SGR 660), Asociación Acunapatata and Aso- ciación NEN. In contrast, the transient overexpression of miR-342-5p did show faster induction of cell death in vitro and a clear reduc- Open Access This article is distributed under the terms of the Crea- tion in tumor growth in vivo in both NB xenografts tested. In tive Commons Attribution 4.0 International License (http://creat​iveco​ line with our observations, miR-342-5p was reported to be mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribu- downregulated in breast cancer patients with early relapse tion, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the [22] and was able to reduce HER2-positive breast cancer cell Creative Commons license, and indicate if changes were made. growth [23] and colon cancer cells [24]. Interestingly, we found that miR-342-5p is also a direct modulator of CCND1. CCND1 silencing alone caused a dra- References matic decrease in NB cell proliferation, but was not enough to induce cell death (Fig. 6c, f). We also identified BCL-XL 1. Maris JM, Kyemba SM, Rebbeck TR, White PS, Sulman EP, as a direct target of miR-342-5p. BCL-XL is an anti-apop- Jensen SJ, Allen C, Biegel JA, Brodeur GM (1997) Molecu- totic member of the B cell lymphoma 2 (BCL-2) protein fam- lar genetic analysis of familial neuroblastoma. Eur J Cancer 33(12):1923–1928. https:​ //doi.org/10.1016/S0959-​ 8049(97)00265​ ily whose overexpression contributes to tumor progression -7 and resistance to chemotherapeutic agents [30]. Although 2. Gatta G, Ferrari A, Stiller CA, Pastore G, Bisogno G, Trama A, BCL-XL reduction alone only causes a minimal reduction Capocaccia R (2012) Embryonal cancers in Europe. Eur J Cancer in cell number and did not induce cell death (Fig. 6f), it may 48(10):1425–1433 3. Bartel DP (2009) MicroRNAs: target recognition and regulatory lower the threshold of apoptosis induction upon CCND1 functions. Cell 136(2):215–233 silencing or inhibition. Therefore, the concomitant reduction 4. Peng Y, Croce CM (2016) The role of MicroRNAs in human in CCND1 and BCL-XL would more faithfully reproduce cancer. Signal Transduct Target Ther 1:15004. https ​ : //doi. the overexpression effects of miR-342-5p. org/10.1038/sigtr​ans.2015.4 5. Stallings RL (2009) MicroRNA involvement in the pathogenesis of neuroblastoma: potential for microRNA mediated therapeutics. Conclusions Curr Pharm Des 15(4):456–462 6. Lin RJ, Lin YC, Chen J, Kuo HH, Chen YY, Diccianni MB, Lon- Our strategy is confirmed as valid for the identification of don WB, Chang CH, Yu AL (2010) microRNA signature and expression of Dicer and Drosha can predict prognosis and deline- novel tumor-suppressive miRNA such as miR-323a-5p and ate risk groups in neuroblastoma. Cancer Res 70(20):7841–7850 miR-342-5p in NB models, including but not limited to the 7. Creevey L, Ryan J, Harvey H, Bray IM, Meehan M, Khan AR, ones that are resistant to conventional therapies and reveals Stallings RL (2013) MicroRNA-497 increases apoptosis in new vulnerabilities of high-risk NB through the combined MYCN amplified neuroblastoma cells by targeting the key cell cycle regulator WEE1. Mol Cancer 12:23 inhibition of targets such as CCND1, CHAF1A, INCENP 8. Soriano A, Paris-Coderch L, Jubierre L, Martinez A, Zhou X, and BCL-XL. Piskareva O, Bray I, Vidal I, Almazan-Moga A, Molist C, Roma J, Bayascas JR, Casanovas O, Stallings RL, Sanchez de Toledo J, Gallego S, Segura MF (2016) MicroRNA-497 impairs the growth 13 Functional high-throughput screening reveals miR-323a-5p and miR-342-5p as new… 2243 of chemoresistant neuroblastoma cells by targeting cell cycle, WB, McGrady PW, Higgins DG, O’Meara A, O’Sullivan M, survival and vascular permeability genes. Oncotarget 7(8):9271– Stallings RL (2009) Widespread dysregulation of MiRNAs by 9287. https​://doi.org/10.18632​/oncot​arget​.7005 MYCN amplification and chromosomal imbalances in neuroblas- 9. Wang Y, Luo J, Zhang H, Lu J (2016) microRNAs in the same toma: association of miRNA expression with survival. PLoS One clusters evolve to coordinately regulate functionally related genes. 4(11):e7850 Mol Biol Evol 33(9):2232–2247 21. Yang HA, Wang X, Ding F, Pang Q (2015) MiRNA-323-5p pro- 10. Hoshi M, Otagiri N, Shiwaku HO, Asakawa S, Shimizu N, Kaneko motes U373 cell apoptosis by reducing IGF-1R. Med Sci Monit Y, Ohi R, Hayashi Y, Horii A (2000) Detailed deletion mapping of Int Med J Exp Clin Res 21:3880–3886 chromosome band 14q32 in human neuroblastoma defines a 1.1- 22. Perez-Rivas LG, Jerez JM, Carmona R, de Luque V, Vicioso L, Mb region of common allelic loss. Br J Cancer 82(11):1801–1807. Claros MG, Viguera E, Pajares B, Sanchez A, Ribelles N, Alba https​://doi.org/10.1054/bjoc.2000.1108 E, Lozano J (2014) A microRNA signature associated with early 11. Kanehisa M, Goto S (2000) KEGG: kyoto encyclopedia of genes recurrence in breast cancer. PLoS One 9(3):e91884. https​://doi. and genomes. Nucleic Acids Res 28(1):27–30 org/10.1371/journ​al.pone.00918​84 12. Boloix A, Paris-Coderch L, Soriano A, Roma J, Gallego S, 23. He YJ, Wu JZ, Ji MH, Ma T, Qiao EQ, Ma R, Tang JH (2013) Sanchez de Toledo J, Segura MF (2015) Novel micro RNA-based miR-342 is associated with estrogen receptor-alpha expression and therapies for the treatment of neuroblastoma. An Pediatr (Barc). response to tamoxifen in breast cancer. Exp Ther Med 5(3):813– https​://doi.org/10.1016/j.anped​i.2015.07.016 818. https​://doi.org/10.3892/etm.2013.915 13. Daige CL, Wiggins JF, Priddy L, Nelligan-Davis T, Zhao J, Brown 24. Yang H, Li Q, Niu J, Li B, Jiang D, Wan Z, Yang Q, Jiang F, Wei D (2014) Systemic delivery of a miR34a mimic as a potential P, Bai S (2016) microRNA-342-5p and miR-608 inhibit colon therapeutic for liver cancer. Mol Cancer Ther 13(10):2352–2360. cancer tumorigenesis by targeting NAA10. Oncotarget 7(3):2709– https​://doi.org/10.1158/1535-7163.mct-14-0209 2720. https​://doi.org/10.18632​/oncot​arget​.6458 14. Beg MS, Brenner AJ, Sachdev J, Borad M, Kang YK, Stou- 25. Molenaar JJ, Ebus ME, Koster J, van Sluis P, van Noesel CJ, demire J, Smith S, Bader AG, Kim S, Hong DS (2017) Phase Versteeg R, Caron HN (2008) Cyclin D1 and CDK4 activity I study of MRX34, a liposomal miR-34a mimic, administered contribute to the undifferentiated phenotype in neuroblastoma. twice weekly in patients with advanced solid tumors. Inves- Cancer Res 68(8):2599–2609. https:​ //doi.org/10.1158/0008-5472. tig New Drugs 35(2):180–188. https​://doi.org/10.1007/s1063​ can-07-5032 7-016-0407-y 26. Lukas J, Muller H, Bartkova J, Spitkovsky D, Kjerulff AA, Jansen- 15. Reid G, Kao SC, Pavlakis N, Brahmbhatt H, MacDiarmid J, Durr P, Strauss M, Bartek J (1994) DNA tumor virus oncopro- Clarke S, Boyer M, van Zandwijk N (2016) Clinical development teins and retinoblastoma gene mutations share the ability to relieve of TargomiRs, a miRNA mimic-based treatment for patients with the cell’s requirement for cyclin D1 function in G1. J Cell Biol recurrent thoracic cancer. Epigenomics 8(8):1079–1085. https​:// 125(3):625–638 doi.org/10.2217/epi-2016-0035 27. Bienvenu F, Jirawatnotai S, Elias JE, Meyer CA, Mizeracka 16. Thakral S, Ghoshal K (2015) miR-122 is a unique molecule K, Marson A, Frampton GM, Cole MF, Odom DT, Odajima J, with great potential in diagnosis, prognosis of liver disease, and Geng Y, Zagozdzon A, Jecrois M, Young RA, Liu XS, Cepko therapy both as miRNA mimic and antimir. Curr Gene Ther CL, Gygi SP, Sicinski P (2010) Transcriptional role of cyclin D1 15(2):142–150 in development revealed by a genetic-proteomic screen. Nature 17. Powers JT, Tsanov KM, Pearson DS, Roels F, Spina CS, Ebright 463(7279):374–378. https​://doi.org/10.1038/natur​e0868​4 R, Seligson M, de Soysa Y, Cahan P, Theissen J, Tu HC, Han A, 28. Barbieri E, De Preter K, Capasso M, Chen Z, Hsu DM, Tonini Kurek KC, LaPier GS, Osborne JK, Ross SJ, Cesana M, Collins GP, Lefever S, Hicks J, Versteeg R, Pession A, Speleman F, Kim JJ, Berthold F, Daley GQ (2016) Multiple mechanisms disrupt the ES, Shohet JM (2014) Histone chaperone CHAF1A inhibits dif- let-7 microRNA family in neuroblastoma. Nature 535(7611):246– ferentiation and promotes aggressive neuroblastoma. Cancer Res 251. https​://doi.org/10.1038/natur​e1863​2 74(3):765–774. https​://doi.org/10.1158/0008-5472.can-13-1315 18. Beckers A, Van Peer G, Carter DR, Mets E, Althoff K, Cheung 29. Gohard FH, St-Cyr DJ, Tyers M, Earnshaw WC (2014) Targeting BB, Schulte JH, Mestdagh P, Vandesompele J, Marshall GM, the INCENP IN-box–Aurora B interaction to inhibit CPC activ- De Preter K, Speleman F (2015) MYCN-targeting miRNAs are ity in vivo. Open Biol 4(11):140163. https​://doi.org/10.1098/ predominantly downregulated during MYCN driven neuroblas- rsob.14016​3 toma tumor formation. Oncotarget 6(7):5204–5216. https​://doi. 30. Decaudin D, Geley S, Hirsch T, Castedo M, Marchetti P, Macho org/10.18632​/oncot​arget​.2477 A, Kofler R, Kroemer G (1997) Bcl-2 and Bcl-XL antagonize the 19. Schulte JH, Marschall T, Martin M, Rosenstiel P, Mestdagh P, mitochondrial dysfunction preceding nuclear apoptosis induced Schlierf S, Thor T, Vandesompele J, Eggert A, Schreiber S, Rah- by chemotherapeutic agents. Cancer Res 57(1):62–67 mann S, Schramm A (2010) Deep sequencing reveals differential expression of microRNAs in favorable versus unfavorable neu- Publisher’s Note Springer Nature remains neutral with regard to roblastoma. Nucleic Acids Res 38(17):5919–5928. https​://doi. jurisdictional claims in published maps and institutional affiliations. org/10.1093/nar/gkq34​2 20. Bray I, Bryan K, Prenter S, Buckley PG, Foley NH, Murphy DM, Alcock L, Mestdagh P, Vandesompele J, Speleman F, London 13