Tumorigenic role of Musashi-2 in aggressive mantle cell lymphoma

SOX11 overexpression has been associated with aggressive behavior of mantle cell lymphomas (MCL). SOX11 is overexpressed in embryonic and cancer stem cells (CSC) of some tumors. Although CSC have been isolated from primary MCL, their relationship to SOX11 expression and contribution to MCL pathogenesis and clinical evolution remain unknown. Here, we observed enrichment in leukemic and hematopoietic stem cells gene signatures in SOX11+ compared to SOX11– MCL primary cases. Musashi-2 (MSI2) emerged as one of the most significant upregulated stem cell-related genes in SOX11+ MCLs. SOX11 is directly bound to the MSI2 promoter upregulating its expression in vitro. MSI2 intronic enhancers were strongly activated in SOX11+ MCL cell lines and primary cases. MSI2 upregulation was significantly associated with poor overall survival independently of other high-risk features of MCL. MSI2 knockdown decreased the expression of genes related to apoptosis and stem cell features and significantly reduced clonogenic growth, tumor cell survival and chemoresistance in MCL cells. MSI2-knockdown cells had reduced tumorigenic engraftment into mice bone marrow and spleen compared to control cells in xenotransplanted mouse models. Our results suggest that MSI2 might play a key role in sustaining stemness and tumor cell survival, representing a possible novel target for therapeutic interventions in MCL.


MCL cell lines and primary samples
Z138 cell line was used for SOX11 knock out (KO) using CRISPR-Cas9 genome editing technology (see below). SOX11 expression was rescued in Z138-SOX11KO cell line by lentiviral transduction to obtain Z138-SOX11KO SOX11+ (FLAG-SOX11 tagged protein) cell line. We silenced MSI2 expression in Z138-SOX11KO cell line generating Z138-SOX11KO sh5MSI2 cell line. Z138 and Granta-519 wild type SOX11+ MCL cell lines were stable transduced to obtain MSI2 knockdown (KD) and control (CT) in vitro models, by lentiviral transduction, in which we performed functional in vitro analyses.
MSI2 expression was rescued in Z138 sh4 and sh5MSI2 cells by lentiviral transduction to obtain Z138 sh4MSI2 and sh5MSI2 MSI2-FLAG cell lines. JVM2-SOX11+ (FLAG-SOX11 tagged protein) MCL cell line was previously obtained [1]. Transduced or wild type Z138, Granta- MCL primary cells from peripheral blood (PB) samples were isolated by Ficoll-Hypaque density gradient centrifugation (GE Healthcare). Some MCL primary cases (n=12) were used for RNA extraction and RNA-seq experiments (Table S2). Other MCL primary cases (n = 8) were used to test the activity of the stem cell marker ALDH on ex vivo experiments.

Microarray analyses
For differential expression analysis in microarray data, gene filtering was done with featureFilter in order to maintain only probes sets with Entrez Gene ID and with the highest interquartile range of the probe sets annotated by the same Entrez Gene ID. Differentially expressed genes (DEG) were obtained by linear models and empirical Bayes methods with limma package (3.42.2) RNA-seq RNA from MCL cell lines was extracted using RNeasy Plus kit (QIAGEN) following manufacturer's instructions. RNA from MCL primary cases was extracted with TRIZOL (Life Technologies). RNA quantity and quality were examined using NanoDrop (Thermo Scientific) and RNA 6000 Nano Assay on the Agilent 2100 Bioanalyzer (Agilent Technologies). RNA samples from cell lines or primary cases with RIN higher than 8 were selected to generate mRNA or total RNA libraries using the TruSeq RNA Sample Prep Kit v2 or the TruSeq Stranded Total RNA kit with Ribo-Zero Gold (Illumina), respectively, following manufacturer's specifications. RNA-seq libraries were amplified, pooled and sequenced on a HiSeq2500 to generate 75 bp paired-end, 50 bp single-end or paired-end reads (Samples information, RNA extraction, library and sequencing conditions are shown in Table S2). More than 30 million of reads were sequenced for each sample. Sequencing reads were quality checked with FastQC (version 0.11.9), depleted from rRNA reads with SortMeRNA (version 4.3.2), trimmed to remove adapters and low-quality reads with trimmomatic (version 0.40), and pseudo-aligned to reference human genome GRCh38.p13 version with kallisto (version 0.46.1) to extract gene-level counts (Ensembl release 100) Gene level counts were imported to R (version 3.6.3, https://www.r-project.org) using tximport (version 1.12.3). Differential expression was conducted using DESeq2 package (version 1.24.0), shrinking the size factor with apeglm method. Variance stabilized expression matrix were obtained for gene set enrichment analysis (GSEA) and clustering.
Lists of DEG were used for functional annotation analysis using DAVID (https://david.ncifcrf.gov/tools.jsp).
For Z138CT and Z138-SOX11KO or JVM2CT and JVM2-SOX11+, samples were sequenced per triplicated. For MSI2 silencing, 4 samples from Z138shCT, 2 from Z138sh4MSI2 and 2 from Z138sh5MSI2 were sequenced. For Ro inhibitory experiments, Z138 cell line was independently treated per triplicated with Ro 08-2750 20 µM or DMSO 0.1% for 4 hours and samples were sequenced. Previously published RNA-seq data from MCL primary cases, cell lines and normal B-cells were obtained from BLUEPRINT consortium [6,7,9].

GSEA
Microarray data and variance-stabilized expression matrices were analyzed with GSEA v4.0.3. C2 curated gene sets related to stem cells and cell death were used in GEP data. Gene sets derived from differential expression analysis in RNA-seq data (Z138-Ro vs Z138-DMSO) were generated for GSEA (Table S10).
For the analyses, data was randomized by 1000 permutations phenotype (datasets with >7 samples per phenotype) or gene_set (datasets with <7 samples per phenotype). For microarray data, probes were collapsed to Gene Symbol using Max-probe. The metric for ranking genes was Diff_of_Classes, which uses the difference of class means to calculate fold change for log scale data. Enrichment was considered at P-value <0.05 and FDR <0.2.

Generation of stable cell lines by lentiviral transduction
HEK-293T were transfected at 60-70% of confluence with lentiviral packaging and envelope plasmids psPAX2 and pCMV-VSV-G (Addgene #12260 and #8454, respectively), and the corresponding plasmid of interest (see Plasmid generation) by using Lipotransfectin (Niborlab). Viral supernatants were collected 48 h after transfection, filtrated and concentrated adding Lenti-X Concentrator (Takara) at 1/10 following manufacturer's instructions. 1.5x10 6 Granta- respectively, for 1 week. Z138sh4 and sh5MSI2 empty or MSI2-FLAG cells were selected with 500 μg/ml of G-418 (Sigma) for 1 week. For SOX11KO, Z138 GFP+ cells were sorted 72 h after transduction in FACSAria II Cell Sorter (BD) and seeded at 500 cells/well in order to obtain pools with SOX11KO. Disruption of SOX11 locus using CRISPR/Cas9 technology was verified by using the Genomic Cleavage Detection Kit (GeneArt) following manufacturer's indications and by PCR using specific primers (Table S4)

RNA immunoprecipitation
For RNA immunoprecipitation (RIP), Magna RIP RNA-binding protein immunoprecipitation kit (Millipore) was used following manufacturer's instructions. MCL growing cells were lysed for 5 minutes at 4 ºC and stored at -80 ºC for 2 hours. Magnetic beads protein A/G were washed and incubated with 3 µg of MSI2 or normal Rabbit IgG antibody (Table S5) on rotating wheel for 1 hour at 4 ºC. Then, lysates were incubated with each immune-complex (30x10 6 cells per 3 ug of antibody bound to beads) on rotating wheel overnight at 4 ºC, separating a 10% of input fraction before antibody-beads addition.
Supernatants (unbound fractions) were saved for western blot (WB) experiment. Immunocomplexes bound to lysates were washed 6 times. One sample incubated with MSI2 antibody was used for WB analysis adding 25 ul of Sample Buffer mixed with DTT and incubating at 95 ºC for 10 minutes. Remaining samples incubated with MSI2 and IgG antibodies and input fraction were used for protein digestion with Proteinase K Buffer at 55 °C for 30 min with shaking. Then, purification of RNA was done by Phenol:Chloroform:Isoamyl Alcohol (125:24:1) method by using Phase Lock Gel Heavy tubes (Quantabio). Ethanol precipitation was performed at -80 ºC overnight. After 80% ethanol wash, RNA was eluted with 12 µl of RNase-free water. NanoDrop (Thermofisher) was used for RNA quantification. RNA from input fraction (1 µg) and the whole RNA eluted from MSI2 and IgG RIP (< 1 µg) were used for cDNA generation by using Verso cDNA Synthesis Kit (Thermofisher) following manufacturer's guidelines and using a blend of oligo dT primers and random hexamers. CDK6, NOTCH1 and GUSB mRNA enrichment was analyzed by qPCR in StepOnePlus (Thermofisher) using Fast SYBR Green Master Mix (Applied Byosistems) and specific primers (Table S4)

Colony assay
For colony assay, 500 growing cells were mixed with 1 ml of Human Methylcellulose Complete Media (R&D System). and plated in a 24-well plate for triplicate. We counted the number of colonies (1 colony >50 cells) after 14 days and we obtained bright field images (88 images stitched) of the colonies after 21 days, using Cytation 5 Imaging Reader.

Surface and intracellular antigens staining for flow cytometry
Cell death analyses were performed using 5x10 4

Western Blot
For protein extraction, 3-5x10 6 growing cells were washed with PBS and lysed in RIPA Buffer (Sigma-Aldrich) with Protease and Phosphatase inhibitor cocktail (Thermo Fisher Scientific) for 20 minutes at 4 ºC. Lysates were obtained by centrifugation at 13000 rpm for 15 min, collecting the supernatant. Protein was quantified by Protein Assay (Bio Rad) in Sinergy HT spectrophotometer (λ 595 nm) and 50-70 μg of total protein mixed with Sample Buffer 5X and DTT was used for WB. Protein extracts were separated by 6-9% SDS-PAGE, transferring them to a 0.45 μm nitrocellulose membrane. Membranes were blocked with 5% of milk powder, BSA or 2.5% of PhosphoBLOCKER (Cell Biolabs) in TBS-T for 1 h and incubated overnight at 4 ºC with specific antibodies (Table S5). After incubation, membranes were washed 3 times with TBS-T and incubated 1 h with the corresponding secondary antibody (Table S5). After washing, the membrane was incubated with Pierce ECL reagent (Thermo Fisher Scientific) to detect the proteins in ImageQuant LAS4000 (Fujifilm).

ALDEFLUOR assay
For ALDEFLUOR assay (STEMCELL Technologies), 5x10 5 MCL primary cells were treated with 1 and/or 5 µM of Ro or DMSO 0.05%, for 24 h. The ALDEFLUOR reagent was incubated for 40 minutes at 37 ºC. DEAB was used as negative control. The ALDH activity was defined by the corrected mean fluorescence intensity (MFI ALDEFLUOR -MFI ALDEFLUOR+DEAB samples).

Engraftment in xenograft mice models
For the generation of MSICTLuc+ and MSI2KDLuc+ xenograft mice models, NSG mice (NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ, Janvier LABS), 20-week of age from both sexes, were intravenously inoculated into their tail veins with 10x10 6 cells of Z138shCT-Luc and Z138sh5MSI2-Luc cells (5 mice per group). Mice were intraperitoneal injected with 112 mg/kg of D-Luciferin (tebu-bio) and anesthetized with isofluorane (Esteve). Tumor dissemination and growth was captured by luciferase bioimage (LBI) twice a week for 5 weeks. Bioluminiscence was captured 10 min after the injection of D-luciferin by IVIS Lumina III In Vivo Imaging System (PerkinElmer) and total flux (photons/s) from the tumoral cells was quantified using Living Image Software.
MSICT-Luc+ and MSI2KD-Luc+ MCL xenograft mice models were euthanized at was collected with capillary action blood collection tubes (SAI Infusion Technologies).
Spleens were obtained, photographed, weighed, homogenized and filtered through 70 μm nylon cell strainers (Fisher Scientific). Femur and tibias were extracted, cleaned of all connective tissues, and flushed with PBS and EDTA using a 27-gauge needle (Becton Dickinson). Bones were grinded with a mortar and filtered along with the BM flushes by using a 70 μm nylon cell strainer. Erythrocytes from PB, spleens and BM were lysed incubating samples with ACK buffer (Quality Biological) for 10 minutes. Samples were washed with PBS and the process was repeated until pellets remained white. Cells were filtered and resuspended in PBS to determine the percentage of MCL cells measuring GFP+ cells fluorescence by flow cytometry.
The protocol was approved by the animal testing ethical committee of the University of Barcelona.

Statistics
Welch's correction was applied to compare samples with different variances.
Overall survival (OS) calculated from time of sampling was used for Kaplan-Meier curves, and log-rank test was performed to measure the association of OS with categorical variables. Maximally selected rank statistics was applied to obtain cutoffs of continuous          Table S10. Gene sets used for gene set enrichment analysis and GO analysis containing differentially expressed genes upregulated and downregulated in Z138 Ro