The study was approved by the Institutional Review Board (IRB) of Shanghai Children’s Hospital (SCH), Shanghai Jiao Tong University, in accordance with the principles of Declaration of Helsinki. Written informed consent to participate in this study was provided by the participants’ legal guardian or next of kin. Patients’ identities and privacy were protected and anonymized in the study. In addition, the Medical Experimental Animal Administrative Committee in Shanghai approved all animal experimental protocols.
Patients and specimens
Patients were included based on the criteria for high-risk NB reported by the COG [3]: 1) MYCN amplification despite any other conditions; 2) INSS = 4 and age ≥ 1.5 years; and 3) INSS = 4 and the existence of genomic abnormalities. Finally, 14 high-risk NB samples obtained between January 2015 and December 2019 were selected for clinical validation. These tissue samples were excised in the operating room, immediately snap-frozen in liquid nitrogen and then stored at − 80 °C.
Cell culture
The NB cell lines BE(2)-C, SK-N-BE2, IMR-32, SH-SY5Y and SK-N-SH were purchased from the Chinese Academy of Sciences Cell Bank. BE(2)-C, SK-N-BE2 and SH-SY5Y cells were cultured in DMEM/F12 (Gibco, #11330–032) supplemented with 10% fetal bovine serum (FBS; Sigma, #F2442) and a 1× penicillin streptomycin (P/S) solution (BasalMedia, #S110JV). IMR-32 and SK-N-SH cells were cultured in Eagle’s minimum essential medium (MEM) (BasalMedia, #L510KJ) supplemented with 10% FBS and 1 × P/S. The NB cell lines are generally categorized into 3 cell types based on morphological and biochemical properties of cells: N (neuroblastic) cells, S (substrate-adherent) cells and I (intermediate) cells. N-type cells exhibit neuronal marker enzyme activities. S-type cells contain tyrosine activity that is suggestive of stem cell capabilities. I-type cells differentially express both N-type and S-type characteristics in the transdifferentiation process between N-type and S-type cells [20]. SK-N-BE2, IMR-32 and SK-N-SH are S-type cells. SH-SY5Y are N-type cells and BE(2)-C are I-type cells [20]. Each cell line type exhibits different growth features and short tandem repeat (STR) results [20, 21]. Normal control RPE-1 fibroblasts and MCF-10 epithelial cells were cultured in DMEM/F12 supplemented with 10% FBS and 1xP/S. Virus-packaging HEK293T cells were cultured in Dulbecco’s modified Eagle medium/high glucose (DMEM) (BasalMedia, #L110KJ) supplemented with 10% FBS and 1 × P/S. STR sequencing was performed in NB cell lines by BIOWING Biotech Co., Ltd. (Shanghai, China) (Supplemental Table 1).
Virus packaging and transfection
Lentiviral shRNA or sgRNA plasmids were constructed by inserting target oligonucleotides into pLKO.1 (Addgene, #10878) or lentiGuide-Puro (Addgene, #52963) plasmids, respectively. Plasmid DNA was extracted using a DNA extraction kit (Vazyme, DC112–01). The lentivirus was packaged by transfecting the plasmids with packaging vectors (psPAX and pMD2.G) and PEI MAX solution (Polysciences, #24765) into HEK293T cells. Afterward, the virus supernatant was collected, filtered with a 0.45 μm strainer, concentrated with PEG6000 (Sigma, #81253), resolved in PBS and then aliquoted for subsequent transfection. Cells were infected with viruses and selected for 72 hours with puromycin (0.5–2 μg/mL, Yeasen, 60210ES25), blasticidin (5–20 μg/mL, YEASEN, 60218ES60), or G418 (100–250 μg/mL, Yeasen, 60220ES03). The target oligonucleotides are listed in Supplemental Table 2.
Immunoblotting
Cell samples were lysed in RIPA buffer (Thermo Fisher Scientific, #89900) and protein concentrations were quantified using the Pierce BCA kit (Thermo Fisher Scientific, #23225). Denatured proteins (10–20 μg/lane) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes. Afterward, the membranes were blocked with 5% fat-free milk (BD Biosciences, #232100) in Tris buffered saline containing Tween 20 (TBST) and incubated with rabbit anti-human MYCN (1:1000; Cell Signaling Technology, #84406), mouse-anti-human Flag (1:1000; Sigma, #F1804), or rabbit-anti-human β-tubulin (1:5000; Abcam, #ab6046) primary antibodies overnight. The membranes were then incubated with HRP-conjugated goat anti-rabbit or mouse IgG (0.2 μg/ml; Pierce, #31460 or #31430). A luminescent image analyzer (Fujifilm, LAS-4000) was used to visualize the bands after an incubation with enhanced chemiluminescence reagents (Millipore, WBKLS0500).
Quantitative reverse transcription PCR (Q-RT-PCR)
Approximately 25 mg tissue samples or 2.5 × 105 cells were lysed in TRIzol reagent (Thermo Fisher Scientific, #TR118) for total RNA extraction. RNA was reverse transcribed to cDNA with a High-Capacity RNA-to-cDNA kit (Thermo Fisher Scientific, #4387406). Quantitative PCR was performed using an Applied Biosystems QuantStudio™ 5 Real-Time PCR System (Thermo Fisher Scientific, #A34322) with SYBR Green Master buffer (ROX) (Thermo Fisher Scientific, #A25742). GAPDH was used as an internal control, and mRNA levels were calculated with the 2^delta Ct approach. The primer sequences are summarized in Supplemental Table 3.
Cell viability assay
Cells were plated in 96-well plates in triplicate at a density of 750 cells/well (small cells except for SK-N-SH and RPE-1 cells) or 375 cells/well (large cells) in 100 μL of culture medium. Then the CellTiter-Glo® luminescent cell viability assay (Promega, #G7573) was performed to assess cell viability at days 0, 2, 4 and 6 according to the manufacturer’s protocol.
Clone formation assay
A total of 375 cells/well (small cells) or 187 cells/well (large cells) were seeded in 12-well plates. Fresh medium was added every 5 days in the first 7 days. Half of the medium was removed, and the same amount of fresh medium was added every 4 days afterward. Generally, colonies were visible after 8–21 days, and the cells were washed with PBS, fixed with a 10% neutralized formaldehyde solution, and then stained with 0.5% crystal violet (Sigma, #C6158-100G) containing 25% methanol.
SMAD9 constructs
The sequence of the SMAD9 coding sequence (CDS) was screened to design a spot mutation without a change in amino acid sequence in the shSMAD9 target site. Wild-type (WT) and mutated (MUT) SMAD9 CDSs were cloned into the G418-resistant vector plenti-EF1α-MCS-IRES-Neo-WPRE (EcoRI and BamHI) by homologous recombination using the ClonExpress kit (Vazyme, #C112). The empty vector (EV), the rebuilt SMAD9-WT and SMAD9-MUT plasmids were applied for virus packaging in HEK293T cells and NB stably transfected cells (STCs) were established with G418 selection. Thereafter, STCs were transfected with shSMAD9 and selected with puromycin.
NB patient-derived Xenograft (PDX) and cells (PDCs)
Two primary NB samples from high-risk patients (SCH-NB010# and SCH-NB016#) were orthotopically xenografted into NSG mice in the first passage. Then, xenografts in the second, third and fourth passages in vivo were subcutaneously transplanted (passaged) into NSG mice. PDCs were extracted in the fourth passage and cultured in DMEM/F12 supplemented with 10% FBS. According to previous reports [22, 23], the formation of neuron-shaped cells indicated good cell line generation. After the formation of large numbers of neuron-like cells, NB PDCs were digested and transfected with shSMAD9 virus for further analysis.
Tetracycline-inducible SMAD9 knockdown (Tet-on)
The shSMAD9 sequences were cloned into the Tet-on puromycin-resistant plasmid (Addgene, #21915). Empty vector (EV) and Tet-on-shSMAD9 plasmids were applied for virus packaging in HEK293T cells and for NB STC establishment with puromycin selection. Doxycycline (Dox) was used in Tet-on STCs. Therefore, STCs were expanded for validation in vitro (1 μg/mL Dox to induce SMAD9 knockdown) and tumor growth in vivo (2 mg/mL Dox with 2% sucrose).
Animal experiments
BALB/c nude female mice (4–6 weeks) were purchased from the Experimental Animal Center of the Chinese Academy of Sciences (Shanghai, China). For subcutaneous cell line xenografts, 5 × 106 NB Tet-on STCs were subcutaneously transplanted in the dorsal flanks of mice on each side. When the tumors reached the volume of approximately 100 mm3, the mice were randomly divided into two groups and provided water containing 2 mg/mL Dox with 2% sucrose or 2% sucrose as a control. Tumor volumes were measured every 3 days with the formula 1/2 (long axis * short axis^2). Mice with tumors > 1500 mm3 were euthanized.
Histological analysis
Histological analyses, including hematoxylin-eosin (HE) and immunohistochemical (IHC) staining, were performed by Servicebio Biotechnology Company (Shanghai, China). IHC staining was performed using primary antibodies against Ki-67 (Servicebio, #GB121499) and cleaved caspase-3 (Cell Signaling Technology, #9661). The stained cells were counted as the percentage of total positive cells in the five random fields of view using the IHC profiler plugin in the ImageJ software (v1.52p, USA).
ChIP-seq and ChIP-qPCR analysis
The Flag-tagged SMAD9 CDS was reconstructed in the plasmid pCDH-CMV-MCS-EF1-Puro using the ClonExpress kit (Vazyme, #C112). Empty vector (EV) and Flag-tagged SMAD9 CDS plasmids were applied for virus packaging in HEK293T cells and NB STC establishment with puromycin selection. Thereafter, NB STCs (1 × 107) were digested with micrococcal nuclease (New England Biolabs, #M0247S) and sonicated for ChIP experiments. Briefly, 10 μg of chromatin were immunoprecipitated with a Flag antibody (Cell Signaling Technology, #14793S), or H3K27Ac antibody (AM39133, Active Motif) and 50 μL Pierce ChIP-grade Protein A/G Magnetic Beads (Thermo Scientific, #26162). Immunoprecipitated DNA and INPUT DNA were then purified and sequenced by Romics (Shanghai, China). ChIP-seq data were first mapped against the human genome build hg19 using BowTie2 [24]. Then, a model-based analysis of ChIP-seq data (MACS2: v2.2.7.1) was used to identify peak regions of ChIP-seq enrichment [25]. Peaks were annotated by performing a HOMER analysis [26]. DeepTools was used to generate BigWig files for visualization [27]. ChIP-seq data from biological experiments and public datasets were subjected to the same pipeline.
For the ChIP-qPCR analysis, 5–10 × 106 cells were harvested and 5–10 μg of chromatin were immunoprecipitated with a Flag antibody and 50 μL of Pierce™ ChIP-grade Protein A/G Magnetic Beads. qPCR was performed using SYBR Green buffer (Thermo Fisher Scientific, #A25742). The primers were designed according to the peak sequence of Flag on the promoter of MYCN (Supplemental Table 4).
CRISPR interference (CRISPRi) assay
Human MYCN-amplified BE(2)-C, SK-N-BE2 and IMR-32 NB cells were transduced with lenti-dCas9-KRAB-blast (Addgene, #89567) to generate dCas9-expressing STCs, and then lentiviral sgRNA was transduced into dCas9 expressing cells. Forty-eight hours after infection with the lentiviral sgRNA, positive cells were selected with puromycin for 3–5 days. Then, the screened cells were harvested for Q-RT-PCR, immunoblotting, cell viability and colony formation assays. The sgRNA target oligonucleotides were designed using online software (https://www.benchling.com/) according to the promoters or enhancers of target genes (Supplemental Table 5).
Luciferase reporter assay
A luciferase reporter assay was performed as described in a previous study [28]. The MYCN promoter or SMAD9 enhancer sequences were cloned into pGL3-luc vectors according to the peak sequences of SMAD9-Flag to MYCN or MYCN to SMAD9 based on ChIP-seq data. The cloning PCR primers are presented in (Supplementary Table 6).
For MYCN knockdown, we constructed BE(2)-C Tet-on shMYCN cells by infecting wild-type BE(2)-C cells with the pLKO-Tet-On-shMYCN virus followed by continuous puromycin selection. Then, BE(2)-C Tet-on shMYCN cells were transfected with pGL3-luc based reporter plasmids and Renilla pGL3-Rluc control plasmid (ratio, 10:1). After 24 hours, transfection media were replaced with media containing or lacking doxycycline (1 μg/ml), and cells were harvested 48 h later. Procedures similar to those described above were used for BE(2)-C Tet-on shSMAD9 cells. Samples were assayed with a dual luciferase assay system (Promega, E1910). Firefly luciferase activity was normalized to Renilla luciferase activity.
Public data acquisition
The NB tissue datasets (GSE12460, GSE13136, GSE16476, GSE49710 and GSE79910), NB cell line datasets (GSE28019, GSE80397, GSE121529 and GSE124451), NB xenograft dataset (GSE90121), normal adrenal gland datasets (GSE3526, GSE7307 and GSE8514), normal neural crest cell line dataset (GSE14340) and ChIP-seq dataset (GSE94822) were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). The NB tissue RNA-seq dataset EGAS00001001308 (abbreviated as EGAS) and microarray dataset TARGET were obtained from the R2 database (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi). The noncancerous cell line datasets included epithelium-derived MCF-10A cells (GSE17785), epidermal keratinocytes (GSE7216) and endometrial cells (GSE16906) from the R2 database. Data on expression and dependency in cancer cell lines were downloaded from the DepMap database (https://depmap.org/portal/; version 22Q1). ChIP-seq data for SEs were downloaded from the super-enhancer database (SEdb; http://www.licpathway.net/sedb/) as described in the previous reports [29].
Analysis of RNA sequencing data
For the RNA sequencing analysis of our own samples, trim galore was used to automatically detect and trim adaptors. Reads were mapped to the hg19 reference genome using HISAT2. Read counts were generated using HTSeq (version 0.11.1) and fragments per kilobase million (FPKM) values were calculated from the number of reads that mapped to each gene sequence, and the gene length and sequencing depth were considered.
Identification of differentially expressed genes (DEGs) and construction of a Venn diagram
The DEGs identified among NB tissues vs. normal adrenal glands, NB cell lines vs. neural crest cells and shScramble vs. shSMAD9 MYCN-amplified cells were analyzed using the false discovery rate (FDR) moderated limma test (package “DESeq2” in R). The cutoff was set for DEG selection based on the criterion |log2 (fold change, FC)| > 1 with P < 0.05 in public datasets and |log2FC| > 0.4 with P < 0.05 in shSMAD9 RNA-seq data. A Venn diagram was constructed to visualize the overlapping genes using an available online tool (http://bioinformatics.psb.ugent.be/webtools/Venn/).
Gene Set Enrichment Analysis (GSEA)
GSEA was performed with GSEA 4.1.0 software according to the online instructions (http://www.broadinstitute.org/gsea/index.jsp). All gene sets were downloaded from the official GSEA website, which includes 50 hallmark gene sets (v7.4) and MYCN-associated gene sets: WEI_MYCN_TARGETS_WITH_E_BOX, NMYC_1, KIM_MYCN_AMPLIFICATION_TARGETS_UP and LASTOWSKA_COAMPLIFIED_WITH_MYCN. In addition, we extracted the top 500 genes (ranked by P value) after MYCN knockdown (KD) from GSE80397 and GSE121529 and built the gene sets MYCN_TARGETS_GSE80397 and MYCN_TARGETS_GSE121529. An FDR q-value < 0.25 was considered statistically significant.
Protein-protein interaction (PPI) network, Cytoscape and Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses
The STRING (http://string-db.org/) database was applied to determine the PPI network of 784 common downregulated genes with an interaction score of 0.4, after which Cytoscape combined with the CytoHubba plugin was used to visualize the PPI network and hub genes. GO/KEGG enrichment analyses were performed using the “clusterProfiler” and “org.Hs.eg.db” packages in R, and the top items were visualized.
Statistical analyses
GraphPad Prism v9.2.0 or R v3.6.3 software was applied for statistical analysis. Comparisons between two groups were performed using unpaired two-tailed Student’s t test, and comparisons among more than two groups were performed using one-way ANOVA. Tumorigenicity in vivo was compared between the two groups using two-way ANOVA. The log-rank (Mantel Cox) test was used to analyze the Kaplan-Meier survival curves. P < 0.05 was considered statistically significant.