Cell lines
SMB21, SMB55 and SMB56 cell lines were previously derived from spontaneous MB tumors from three individual Ptch+/- mice and show uniform GCNP lineage marker expression as well as aberrant SHH activation [65]. SMB21 cells with loss of Sufu and Gli2 amplification have been previously described as SMO-inhibition resistant cell models [64, 65]. All SMB cells were cultured as neurospheres in ultra-low attachment culture flasks (Corning) with Dulbecco´s Modified Eagle´s Medium/Ham´s F-12 50/50 Mix medium (DMEM/F12) (Corning) supplemented with 2% B27 and vitamin A (Thermo Fisher Scientific), 1% GlutaMax (Thermo Fisher Scientific) and 1% penicillin–streptomycin (Thermo Fisher Scientific). Human pediatric MB cell line DAOY was cultured in Roswell Memorial Institute Medium (RPMI) (Thermo Fisher Scientific) supplemented with 10% fetal calf serum (Thermo Fisher Scientific) and 50µg/ml gentamycin (Thermo Fisher Scientific). For all cell lines, the optimal cell density was determined in order to achieve optimal growing conditions. 200,000 cells per ml was the optimal seeding density for SMB cells and 5,000 cells per cm2 for DAOY cells. All cell lines were kept at 37°C humidity-controlled incubator with 5% CO2 and regularly tested for mycoplasma contamination.
Patient-derived xenografts organoids generation and maintenance
Patient-derived xenografts organoids (PDXOs) have been generated and maintained, as previously described [30]. Briefly, PDXOs were maintained in patient-derived organoids medium (PDOs medium) containing 1:1 Neurobasal (Gibco, 21,103,049):DMEM/F12 (Gibco, 11,320,074), 50X B27 supplement (Gibco, 17,504,044), 100X GlutaMax (Gibco, 35,050,038), 100X N2 supplement (Gibco, 17,502,001), 20 ng/ml FGF2 (Peprotech, 100-18B), 20 ng/ml EGF (Peprotech, 100–47), penicillin (100 U/ml)/streptomycin (100 μg/ml) (Gibco, 15,140,122), and Heparin 2.5 μg/ml (Sigma Aldrich, H3149-10KU). PDXOs were cultured in 6-cm/10-cm plates (Sarstedt, 82.1194.500, 82.1472.001) in suspension in PDOs medium on an orbital shaker (70 rpm) placed in a 37°C, 5% CO2 incubator. Twice per week a complete medium change was performed. All PDXOs cultures were regularly tested and confirmed free of Mycoplasma.
Animals
Math1-cre [41], Math1CreERT2 [39], SmoM2-YFPFl/Fl [40] mice were obtained from Ulrich Schüller, University Hospital Hamburg, Germany. Dnmt1Fl/Fl [24] mice were obtained from Rudolph Jänisch, Whitehead Institute for Biomedical Research, Cambridge, USA. Genotyping was performed by PCR using genomic DNA from ear samples. All mice were maintained on a 12-h dark/light cycle and animals of both sexes were used for all experiments.
CRISPR-Cas9 knockout dependency screens
Cas9-expressing SMB21 cells were screened with the guide-only Brie library, which delivers 78,637 different gRNAs targeting 19,674 murine genes [8], and DAOY cells were screened with a corresponding human library, all-in-one Brunello, which provides 76,441 gRNAs targeting 19,114 human genes [8]. Both cell lines were transduced by spinfection with a predefined viral volume, achieving a MOI (multiplicity of infection) of approximately 0.3. 24 h post transduction, cells were split into technical triplicates with cell numbers estimated to reach a 500 × library coverage, meaning that each gRNA would be present on average in 500 cells. Following puromycin selection for 5 days, cells were propagated in culture for 21 days. To ensure proper transduction rate during the screening, an in-line assay was performed in parallel. At the last day of the screen, genomic DNA was extracted from the remaining cells using the QIAamp DNA Blood Maxi Kit (QIAGEN) from which the integrated sgRNA sequences were PCR-amplified and subjected to Next Generation Sequencing at the Broad Institute at MIT (Cambridge, USA).
CRISPR-Cas9 chemogenetic screens
Prior to the screen, SMB21 cells were transduced with the lentiviral vector lentiCas9-Blast (Addgene #52,962) and selected with blasticidin, in order to generate SMB21-Cas9 expressing cells, as validated via immunoblots. Similar to the dependency screen, cells were transduced by spinfection with a predetermined volume of the Brie library (#73,633, Addgene) with a MOI of approximately 0.3. Next day, cells were selected with puromycin for 5 days and in-line assay was conducted in parallel to ensure transduction efficiency. 7 days post transduction, cells were split into either DMSO or 5-azacytidine drug arm in duplicates at a 500 × library coverage. Applied 5-azacytidine concentration had been previously optimized, for it to have a cytostatic effect. Following 2 weeks of DMSO control and drug treatment, genomic DNA was isolated from the surviving cells using the QIAamp DNA Blood Maxi Kit (QIAGEN) from which the sgRNA sequences were PCR-amplified and sequenced with next generation sequencing at the Broad Institute (Cambridge, USA).
CRISPR-Cas9 dependency and drug screens analysis
To account for gene-independent cell responses to CRISPR-Cas9 targeting, we used CRISPRcleanR, which corrects sgRNA log2 fold changes (LFC) in an unsupervised manner [22]. For the downstream comparative analysis of the two screens, murine gene names were converted to homologous human gene names. For direct comparability of screens from human and murine descent, dependency scores were generated by scaling the corrected LFCs on the basis of pan-species non-essential and pan-essential control genes. Furthermore, we followed two distinct statistical approaches, in order to identify cell-specific essential genes. We used MAGeCK RRA algorithm to identify negatively selected genes [34], as well as supervised BAGEL2 algorithm, which calculates the likelihood that one gene belongs to essential or non-essential class [27]. Shared genes from both methods at FDR < 5% are considered essentials for each screen. Comparative gene enrichment analysis of both screens was performed using clusterProfiler R package. For the drug screen, we compared the sgRNA distribution of 5-azacytidine drug arm to either DMSO control arm or reference plasmid using MAGeCK MLE algorithm, which calculates β-scores indicative of the degree of selection per gene [33]. Screen results were visualized using MAGeCKFlute R package.
DNA methylation profiling
SMB21 and SMB55 cells were seeded at their optimal density and treated the next day with 3µM and 5µM 5-azacytidine, respectively, and DMSO control for 24 h. DNA was extracted from cells using QIAamp DNA Blood Mini Kit (QIAGEN). Methylation profiling was performed using Infinium™ Mouse Methylation BeadChip according to the manufacturer´s instructions in the Microarray Unit in DKFZ (Heidelberg, Germany). Differential methylation analysis was performed using the SeSAMe R package [67]. To define significantly hypo- and hypermethylated probes, we set a threshold of |β value|≥ 0.1 and padjusted value < 0.05. Gene enrichment analysis using gene ontology terms was conducted in ShinyGO 0.77.
RNA sequencing
SMB21 cells were seeded at their optimal density and treated the next day with 3 µM 5-azacytidine for 2 and 48 h, as well as with DMSO control for 48 h. RNA was isolated form cells using the RNeasy Plus Mini Kit (Qiagen). For RNA sequencing, mRNA fraction was enriched using polyA capture from 200 ng of total RNA using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB). Next, mRNA libraries were prepared using the NEB Next Ultra II Directional RNA Library Prep Kit for Illumina (NEB) according to the manufacturer’s instructions. The libraries were sequenced as paired-end 50 bp reads on an Illumina NovaSeq6000 (Illumina) with a sequencing depth of approximately 25 million clusters per sample. RNA raw data QC and processing was performed using megSAP (version 0.2–135-gd002274) combined with ngs-bits package (version 2019_11-42-gflb98e63). Reads were aligned using STAR v2.7.3a. Further details can be found under https://nf-core.re/rnaseq. Differential gene expression analysis was performed using DESeq2 R package [37]. To define significantly differentiated expressed genes (DEG), we set a threshold of |log2fold change|≥ 0.58 and padjusted value < 0.05. All heatmaps were visualized on Morpheus (Broad Institute). For Gene Set enrichment analysis (GSEA), we used the GSEAPreranked tool (Broad Institute), for which genes were ranked based on their log2fold change values, according to which gene sets with a false discovery rate (FDR) of less than 25% were considered significantly enriched in our analysis.
Growth rate inhibition assays
For the acute cytotoxic assays, cells were seeded in 50µl per well at their optimal cell density in 96-well plates. Serial dilutions of the tested drugs were generated and 50µl of concentrated drug was added into the cells 24 h post seeding, corresponding to final concentrations ranging from 0.01nM to 50µM. DMSO-treated wells were used to normalize data. After 72 h, cell viability was measured using CellTiter 96 Aqueous One Solution (Promega) on plate reader GloMax (Promega). To account for differences in division times among different cell lines, drug response was calculated as normalized growth rate inhibition (GR), assessing the potency and efficacy of the tested drugs, using GRmetrics R package [16]. The compounds tested, sonidegib, vismodegib and 5-azacytidine, were purchased from Selleckchem and diluted according to manufacturer´s instructions. GR values range from -1 to 1, with -1 to 0 indicating cell death, equal to 0 cytostatic drug effect and 0 to 1 partial growth inhibition. GR50 values refer to the concentration at which cell count is half of the control.
Cell proliferation assays
For the 8-day drug cell proliferation assays, cells were seeded at their optimal cell density in triplicate wells of 6-well plates and treated with corresponding monotherapies or drug combinations. Cell numbers were determined on day 4 and day 8 post seeding. DMSO-treated cells were used as a control.
Synergy assays
For the cytotoxic synergy assays between 2 drugs, cells were seeded in 50µl per well at their optimal cell density in 96-well plates. Next day, cells were treated with 4 different concentrations of single drugs and 16 concentration combinations of both drugs. DMSO-treated wells were used to normalize data. Following 72 h treatment, cell viability was measured using CellTiter 96 Aqueous One Solution on GloMax. Synergistic scores of the drug combinations were calculated based on zero interaction potency model (ZIP) using synergyfinder R package [66].
PDXOs treatment assay
For the drug treatment experiment, PDXOs (SHH MB—MED1712 and G3 MB—HT0pGF1 [30]) were cultured for 7 days in PDOs medium added with the following drugs: 10µM 5-azacytidine, 1µM sonidegib and 10µM 5-azacytidine + 1µM sonidegib. DMSO-treated PDXOs were used as a control. PDXOs were kept in Ibidi uncoated 96-well black µ-plates (Ibidi, 89,621) placed in a 37°C, 5% CO2 incubator. A complete change medium was performed every 48h for all drug conditions.
Genetic validation in vitro and in vivo
In order to generate knockdowns of Smo and Dnmt1, one sgRNA per gene was cloned into lentiCRISPRv2 puro vector (#98,290, Addgene) according to the manufacturer´s protocol. sgRNA sequence for sgSmo is 5´-CACCGGAACTCCAATCGCTACCCTG-3´: and for sgDnmt1: 5´-CACCGACCTCGGGCCAATCAATCAG-3´. Lentivirus was produced in HEK293FT cells and SMB21, SMB55 and SMB56 cells were transduced by spinfection. Following puromycin selection for 3 days, transduced cells were seeded in 96-well plates and viability was determined after 3 and 7 days post seeding, using CellTiter Blue. Viability was normalized to parental cells.
Dnmt1Fl/Fl mice [24] were crossed with Math1-cre::Dnmt1Fl/+ mice, in order to generate mice of genetic background Math1-cre::Dnmt1Fl/Fl and Math1-cre::Dnmt1Fl/+. These animals were sacrificed at p5 and p21, while Math1-cre mice were used a wildtype control. To investigate Dnmt1 ablation during SHH-MB development, Math1-cre::Dnmt1Fl/+ mice were crossed with Dnmt1Fl/Fl::SmoM2-YFPFl/Fl mice, resulting in Math1-cre::Dnmt1Fl/Fl::SmoM2-YFPFl/+ and Math1-cre::Dnmt1Fl/+::SmoM2-YFPFl/+ mice. Animals were either sacrificed at p5 or monitored for clinical symptoms and sacrificed when manifesting neurological symptoms. Math1-cre::SmoM2-YFPFl/+ mice were used as a tumor control group. All animal experimental procedures were approved by the regional council of Tuebingen and conducted according to animal welfare regulations (N10-21G license).
Western blotting
Proteins lysates were extracted from cells and murine tumor tissue using Pierce RIPA buffer (Thermo Fisher Scientific) with phosphatase inhibitors cocktail (1:100) (#5870, Cell Signaling) and sonicated, in order to ensure DNA shearing. 40µg of protein lysates were mixed with 4 × LDS sample buffer and incubated for 10 min at 70°C. Samples were separated electrophoretically in 4–12% NuPage precast gels (Thermo Fisher Scientific) and blotted on nitrocellulose membranes. After blocking the membranes with 5% nonfat dry milk-TBST buffer for 1 h at RT, they were probed with primary antibodies overnight and next day, they were incubated with goat anti-rabbit (1:5000, ab97051, abcam) or goat anti-mouse (1:5000, ab97023, abcam) horse radish peroxidase (HRP)-conjugated secondary antibodies. Proteins were visualized using SuperSignal West Dura solution (Thermo Fisher Scientific) and images were acquired in ChemiDoc imaging machine (BioRad). The following primary antibodies were used at indicated dilutions: GLI1 (1:1000, #2534, Cell Signaling), β-tubulin (1:1000, #86,298, Cell Signaling) and PCNA (1:1000, #2586, Cell Signaling). GLI1 protein bands were quantified using ImageJ by normalizing to corresponding β-tubulin levels of each sample.
Mouse treatment study
Math1-creERT2 mice were first mated with SmoM2-YFPFl/Fl mice, in order to generate mice of genetic background Math1-creERT2::SmoM2Fl/+. For the induction of Cre activity, pups were injected intraperitoneally (i.p.) at postnatal day 5 (P5) with 1mg tamoxifen (T5648-1G, Sigma) dissolved in corn oil (Sigma-Aldrich). Starting at P50, mice were randomized into vehicle control and drug treatment groups. Mice were treated 5 days a week for 3 weeks consecutively with either vehicle control (i.p.), 2.5mg/kg 5-azacytidine monotherapy (i.p.), 20mg/kg sonidegib monotherapy (oral gavage) or combination of both drugs. 5-azacytidine (Selleckchem) was dissolved in 5% DMSO and 30% PEG, sonidegib (Selleckchem) was dissolved in 2% DMSO and 98% corn oil, while vehicle control was diluted in 5% DMSO and 30% PEG300 (Selleckchem). Mice were monitored for clinical symptoms according to a stringent scoring sheet and when they exhibited neurological symptoms, they were sacrificed by transcardiac perfusion. All animal experimental procedures were approved by the regional council of Tuebingen and conducted according to animal welfare regulations (N10-21G license).
Histological and Immunohistochemical analysis
For histological analysis of mouse experiments, dissected brains were cut in the midline, snap-frozen in liquid nitrogen, fixed in Tissue-Tek medium (O.C.T, Sakura Finetek) and half-brains were sectioned sagittally at 8 µm thickness using a cryotome (LEICA CM 3050 S). For Hematoxylin and Eosin (H&E) staining, sections were fixed with acetone (− 20 °C) and 80% methanol (4°C), washed with PBS and stained with 0.1% hematoxylin (SIGMA-ALDRICH) for 10 min. Following counterstaining with 1% eosin (Care Roth) for 2 min, slides were passed through a graded series of ethanol. Luxol fast blue (LFB) staining was performed in the Institute for Pathology and Neuropathology (Tübingen, Germany). Briefly, sections were fixed in 4.5% formalin for 5 min, incubated o/n in LFB staining solution (~ 55°C) and countersrained with 0.1% cresyl violet.
Immunohistochemistry of all murine brains was performed by fixing sections either in 4% PFA (RT) or acetone (-20°C) and 80% methanol (4°C), depending on the antibody used. Slides were blocked with 10% BSA in PBS-Tween 0.3% for 1 h and incubated with primary antibodies diluted in 2% BSA in PBS-Tween 0.06% o/n at 4°C. The primary antibodies used are Ki67 (1:100, ab16667, abcam), Pax6 (1:400, ab19045, abcam), NeuN (1:400, #24,307, Cell Signaling), Cleaved Caspase 3 (1:100, #9664, Cell Signaling), Dnmt1 (1:200, ab19905, abcam), Dnmt1 (1:100, #5032, Cell Signaling), Dnmt1 (1:200, #MA5-32,547, Thermo Fisher), Dnmt3a (1:100, #3598, Cell Signaling) and Dnmt3b (1:200, #ab2851, abcam). Next day, sections were incubated with horse anti-rabbit (H + L, BA-1100, Vector Laboratories) or goat anti-mouse (H + L, BA-9200, Vector Laboratories) IgG secondary biotinylated antibodies diluted at 1:400 in 2% BSA in PBS-Tween 0.06% for 1 h at RT, followed by a 30-min incubation with avidin/biotin-based peroxidase solution (VECTASTAIN Elite ABC, Vector Laboratories) and staining with NovaRed Substrate-HRP solution (Vector Laboratories) for 1–5 min. Finally, all sections were counterstained with Hematoxylin for 45 s and dehydrated with graded ethanol.
Images were acquired using bright-field microscopy (Zeiss, Axioplan 2) and analyzed on Adobe Photoshop CS5.1 and Gimp (2.10.34). In order to calculate the percentage of antibody-positive cells, we counted the total number of cells in the region of interest (ROI) and the number of cells stained positive for the marker. ImageJ was used for quantification.
To evaluate relative tumor area of Math1-creERT2::SmoM2Fl/+ mice treated with vehicle control or corresponding drugs, we manually outlined the entire cerebellum and evident tumor region in a total of 10 H&E-stained sections per brain in three mice per treatment group using the freehand selection tool in ImageJ. Percentage of tumor to total cerebellum area was calculated in µm2.
PDXOs were fixed in 4% paraformaldehyde in PBS at 4°C overnight, cryoprotected in 30% sucrose in distilled H2O at 4°C overnight and embedded in Frozen Section Compound (Leica, 3801480). Frozen PDOs were kept at -20°C until processing. PDXOs cryosections at 20 μm were prepared with a cryostat (Thermo Scientific HM525 NX) on glass slides (Thermofisher Scientific, J1800AMNZ). Slides were stored at -20°C until immunohistology. For immunofluorescence staining of PDXOs, cryosections were treated with a permeabilization solution (PBS supplemented with 3% BSA, Seqens/H2B, 033IDB1000-70; 0.3% Triton™ X100, Sigma-Aldrich, T8787; 5% goat serum, Gibco, 16,210,064) for 1h at room temperature. Primary antibody for Ki67 (Rabbit polyclonal anti-Ki67, 1:500, Abcam, ab15580) was incubated overnight at 4°C in antibody solution (PBS supplemented with 3% BSA, Seqens/H2B, 033IDB1000-70; 0.1% Triton™ X100, Sigma-Aldrich, T8787; 1% goat serum, gibco, 16,210,064) and secondary antibody (Alexa Fluor 546 goat anti-rabbit IgG, 1:500, Thermofisher Scientific, A11035) for 1h at room temperature. Nuclei were counterstained with DAPI 10 mM (Abcam, ab228549). Sections and coverslips (Thermofisher Scientific, 15,747,592) were mounted with permanent mounting medium (Histo-Line laboratories, PMT030).
PDXOs immunohistology images were acquired by either confocal imaging by Leica TCS Sp8 (20X objectives) and Leica Application Suite X software (version 3.5.7.23225) or by Nikon TI2 equipped with spinning disc X-light V2 (10X objective) with NIS Element software (version 5.21.03). For all quantifications of immunohistology, samples being compared were processed in parallel and images were acquired using the same settings and lasers power. For Ki67 quantification, a total number of 9–20 images per sample was used and cells positive for the determined markers were manually quantified using the cell counter function in ImageJ. A specific area of ROI was defined and used across all images, avoiding edges or bad regions of the images. A total number of 600–950 DAPI + cells were counted inside the ROI, equally splitting them across considered images. These DAPI + cells were then checked for the positivity for the marker of interest. Data are presented as mean ± s.e.m. of the percentage of Ki67 + cells/DAPI; each dot represents a ROI/image.
Statistical analysis
All statistical analyses were performed in GraphPad Prism 9 or R (v4.0.5). To compare three or more groups, we used two-way ANOVA with Tukey´s test for multiple comparisons. For the statistical analysis of all cell quantifications in mouse experiments, Fisher’s exact test was used. For the survival analysis of Kaplan–Meier curves, we used the Log-rank test. Differences are considered significant at P < 0.05 unless otherwise specified. Venn diagrams were generated using Venn Diagram R package and statistical analysis of intersections is derived from SuperExactTest R package. For PDOX experiments, the Shapiro–Wilk test was used to validate the assumption of normality. Statistical significance was then determined using Kruskal–Wallis test with Dunn’s post hoc test for data with non-normal distribution.
Add Comment