Cell lines and culture conditions - OCI-Ly1, OCI-Ly18, SUDHL-6, SUDHL-4, HT, HBL-1, NUDUL-1 (DLBCL), RAJI and RAMOS (BL) cell lines were purchased from DSMZ. B593 DLBCL cell line was developed in house (27). NUDUL-1, OCI-Ly18, RAMOS cell lines were cultured in RPMI 1640 medium supplemented with 20% fetal bovine serum (FBS), 1% glutamine, 100 mg/mL penicillin/streptomycin. SUDHL-4, SUDHL-6, HT, B593 and RAJI cell lines were cultured in RPMI Glutamax supplemented with 20% FBS, 1 mM sodium pyruvate, 1% minimum essential medium non-essential amino acids (MEM NEAA) and 100 mg/mL penicillin/streptomycin. OCI-Ly1 cell line was grown in IMDM medium with 20% FBS and 100 mg/mL penicillin/streptomycin. Cell culture reagents were purchased from Gibco. All cells were grown at 37°C, 5% CO2.
Karyotype and FISH – B593, OCI-Ly18, OCI-Ly1, SUDHL-6, SUDHL-4, HT, HBL-1 and NUDUL-1 were prepared in a supplemented cell culture medium for cytogenetic analyses including RPMI, 15% FBS, 1% L-glutamine and 1% penicillin. Karyotype was performed on cell suspension following 17h unstimulated culture. Metaphase preparation slides for chromosome banding analysis were stained to reveal R-banding patterns. Karyotypes were described according to the current version of International System for Human Cytogenetic Nomenclature (28). Metaphase and interphase fluorescence hybridization (FISH) studies were performed on cytogenetic pellet using Dual Color breakapart probes (BCL2, BCL6 and c-MYC, Cytocell) to detect the presence of BCL2, BCL6 and/or MYC rearrangement. The detection of IGH::MYC rearrangement was performed using a dual color dual fusion probe (XL IGH::MYC DF, Metasystems). Slides were fixed in methanol/acetic acid 3:1 for 30 min, air dried and dehydrated through graded alcohols. Both FISH probes and target DNA were denatured simultaneously for 1 min at 73°C, and incubated for 20h at 37°C. The post-hybridization washes were performed according to the manufacturer’s instructions (Cytocell, Metasystems). A total of 100 nuclei were analyzed for each slide. FISH results on metaphases were interpreted with the karyotype for all cell lines tested.
Optical genome mapping (OGM) - OGM was performed for the following cell lines: B593, NUDUL-1, SUDHL4, SUDHL6, OCI-Ly1, OCI-Ly18 and HT.
DNA extraction and direct enzymatic labelling - Ultra-high-molecular-weight genomic DNA (UHMW gDNA) was extracted manually from 1.5 million cells using the Prep SP Frozen Cell Pellet DNA Isolation kit according to the manufacturer’s instructions (Bionano Genomics). Briefly, the samples were lysed and digested with Proteinase K, RNase A and Lysis and Binding buffer. DNA was precipitated with isopropanol and bound to a nanobind magnetic disk. Subsequently, DNA was resuspended in the elution buffer and quantified with Qubit Broad Range double-stranded DNA assay kits (ThermoFisher Scientific). UHMW gDNA was incubated for 48h at room temperature for homogenization.
DNA labelling was performed following manufacturer’s protocols (Bionano Genomics). Standard Direct Label Enzyme 1 (DLE-1) reaction was carried out using 750 ng of UHMW gDNA. DNA was labelled at a sequence-specific motif (CTTAAG) with DL-green fluorophores. Thereafter, DLE- 1 enzyme was digested using puregene proteinase K (Qiagen) and excess DL-green fluorophores were removed with an adsorption membrane in a microtiter plate. The labelled DNA were counterstained (blue backbone staining) and homogenized for 48h at room temperature before quantification using Qubit High Sensitivity double-stranded DNA assay kits (ThermoFisher Scientific).
Data collection and quality metrics - The labelled DNA molecules were loaded on a Saphyr G2.3 chip and molecules were imaged using the Saphyr instrument (Bionano GenomicsSingle linearized DNA molecules travel through nanochannels by electrophoretic migration. For each cell line, multiple cycles were run to collect 800 Gb of data per cell line and reach an average genome coverage of 180X that is sufficient for accurate genomic analysis of a cell line. Specific quality control parameters were evaluated for each cell line to achieve a valid OGM analysis, namely, a map rate ≥ 70%, a label density of 14 to 17 (labels per 100 kbp) and an appropriate labeled DNA length (N50) ≥ 230 kbp of filtered DNA (> 150 kbp) with ≥ 9 labeled sites, according to the manufacturer's instructions.
OGM variant calling and data filtering - The capture images were converted to a barcode pattern and aligned to the human genome reference (GRCh38) for annotation. Data were analyzed using Bionano Access software. We applied the two methods provided by the supplier to identify both structural variants (SV) and copy number variants (CNV): i) the Rare Variant Analysis (RVA) pipeline detects SV on the basis of the differences between aligned barcode patterns and the reference genome; this RVA pipeline enables to identify SV and CNV occurring at low frequency (about 10%) and is therefore largely recommended for hematological neoplasms, ii) The De Novo Assembly (DN) pipeline consists of a first reconstruction of the sample genome, which is subsequently compared to the reference genome; this DN pipeline is commonly used to identify germline chromosomal abnormalities but offers three advantages in a context of B-cell lymphoma cell lines: first, a high resolution, enabling detection of fine, cryptic or subtelomeric anomalies (like IGH/14q32.3 rearrangements); second, the capacity of detection of copy neutral loss of heterozygosity (cn-LOH); third, as a complementary and confirmation method together with the RVA pipeline.
RNA extraction & RT-qPCR - Total RNA was extracted from 3 million cells with NucleoSpin RNA extraction kit (Macherey-Nagel) according to manufacturer’s instructions. Reverse transcription was carried out following RT SuperscriptTM III kit (Invitrogen RT) from 1µg extracted RNA. Target genes were amplified using PCR primers (MYCforw GGCTCCTGGCAAAAGGTCA, MYCrev CTGCGTAGTTGTGCTGATGT, BCL2forw, GTGGATGACTGAGTACCTGAAC, BCL2rev GAGACAGCCAGGAGAAATCAA, BCL6forw TCTGGAGAGAAGCCCTACAA, BCL6rev CCACAGATTTCACAGGGATAGG, GAPDHforw CCACTCCTCCACCTTTGAC, GAPDHrev ACCCTGTTGCTGTAGCCA) and quantified by qPCR on a thermocycler CF384 Touch Real-Time PCR (Bio-Rad). Ct values (cycle threshold) were analyzed with CFX ManagerTM Software (Bio-Rad) and normalized to GAPDH.
RNA sequencing – Cell lines were treated with 500 nM JQ1 for 24h. Total RNA was extracted and purified according to the procedure mentioned in the RNA extraction & qPCR section. RNA paired-ended sequencing and stranded library were performed by BGI Genomics or GenomEast (IGBMC, Strasbourg, France) sequencing facilities from 500ng total RNA. Sequencing depth was at least 30 million reads per sample. Reads were quality-checked using FastQC. Reads were then aligned to H. sapiens genome build hg38 using STAR 2.7.1a (Dobin et al., 2013). Raw read counts for each gene were calculated using HTSeqCount 0.11.2 using defaults parameters (29). Read count data was normalized using RPM normalization and DESeq2 1.32. (30). Analyses were performed with R release 4.1.0. DESeq2 normalized read counts were used to identify differentially expressed genes with an adjusted p-value < 0.05 and a fold change below − 1.5 or above 1.5. Enrichment analysis were performed using GSEA software (31) and Cluster Profiler R package (32).
Western blot – Proteins were extracted from 5 million cells with 100 µL of RIPA buffer (150 mM NaCl, 50 mM Tris-HCl pH 8.0, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 1% NP-40). Lysates were incubated 30 min at 4°C under agitation. Samples were then sonicated with VibraCell sonicator (15 sec ON/OFF for 1 min per sample). Then, samples were centrifuged 20 min at 16,000g and 4°C to clear insoluble material and quantified with Bradford protein assay. 10 µg of each sample was loaded on a gradient gel NuPAGE 4–12% Bis-Tris (Invitrogen) and run in MES buffer (Invitrogen). The proteins were transferred on an activated PVDF membrane for 1h at 100V at 4°C in Tris Glycine 10% ethanol. Membranes were blocked in either 5% milk or BSA in PBS-Tween (PBST) or TBS-Tween (TBST) for 1 hour, and incubated overnight with primary antibodies under constant rotation at 4°C. The following primary antibodies were used: anti-MYC 1/1,000 TBST 5% BSA (rabbit, #A9169, Cell Signaling), anti-BCL2 1/1,000 PBST 5% milk (mouse, #sc-7382, Santa-Cruz Biotechnology), anti-BCL6 1/2,000 PBST 5% milk (mouse, #sc-7388, Santa-Cruz Biotechnology), anti-BRD4 1/1,000 PBST 5% milk (rabbit, # A301-985A, Bethyl laboratories), anti-BRD3 1/1,000 PBST 5% milk (rabbit, #A302-368A, Bethyl laboratories), anti-BRD2 1/2,000 PBST 5% milk (rabbit, #A302-583A, Bethyl laboratories), anti-αTubulin 1/5,000 PBST 5% milk (mouse, #T5168, Merck), anti-H4 1/5,000 PBST 5% milk (rabbit, #05-858, Upstate). The membranes were washed three times in PBST or TBST, incubated with the following secondary antibody: goat anti-rabbit IgG peroxidase 1/10,000 (#A9169-2ML, Merck) or polyclonal rabbit anti-mouse IgG peroxidase 1/2,000 (#P026002-2, Agilent) for 1h at room temperature, washed three times and revealed with Luminol/peroxidase mix Clarity Western ECL (Biorad). Acquisitions were realized using Vilber imaging system. Signal quantification was performed with ImageJ.
Establishment of inducible MYC knock-down cell lines – shRNA sequences either targeting MYC (shMYCA: GCAATCACCTATGAACTTGTT and shMYCB: GACGACGAGACC
TTCATCAAA) (33) or non-targeting (shControl: CCTAAGGTTAAGTCGCCCTCG) were used. Oligos were annealed and ligated by T4 ligase with the pLKO Tet-On inducible plasmid previously digested with AgeI/EcoRI (34). Lentiviral particles were produced by the ANIRA platform (ENS Lyon, France). B593, SUDHL-4 and OCI-Ly1 cell lines were transduced with lentiviral particles at multiplicity of infection (MOI) of 10, and NUDUL-1 with a MOI of 20. Viral particles were washed away after 24h. 1 µg/mL of puromycin was added at day 3 post-infection. Induction was performed with doxycycline at 200ng/mL during 72h for B593, SUDHL-4 and OCI-Ly1 or with 500 ng/mL during 72h for NUDUL-1. Knockdown efficiency was assessed by Western Blot.
Cell cycle analysis – 1 million cells were harvested and washed with PBS. Cell pellets were resuspended in 100µL PBS and fixed with 900µL of cold ethanol (70%) added drop by drop under constant agitation of the sample collection tube. After 30 min, the samples were stored at -20°C for some days. The day of the analysis, samples were slowly thawed at 4°C and then centrifuged (2,000 rpm, 10 min). Cell pellets were resuspended in 1 mL of PBS in order to complete rehydration of the fixed samples during 4h at 4°C. After rehydration, cells were collected by centrifugation (2,000 rpm, 5 min) and resuspended in 1 mL PBS, 10 µg/mL of RNAse A (Sigma Aldrich) and 10 µg/mL propidium iodide (Sigma Aldrich). Samples were incubated at room temperature in the dark during 30 min before being analyzed by flow cytometry (LSRII, BD Biosciences). Data were analyzed using ModFit LT software.
Assessment of cell proliferation and cell death – Cell viability and proliferation were determined by dual Annexin V-FITC (BD Biosciences) and propidium iodide (Sigma Aldrich) staining and by numeration of the absolute count of fluorescent beads (Precision count beads, Biolegends) and cells ratio. 0.1µg/mL Annexin V was diluted in Binding buffer solution (0.1 M HEPES pH7.4, 1.4 M NaCl, 25 mM CaCl2). Next, cells were harvested and washed with PBS and stained in 100 µL of Annexin V staining mix at room temperature for 10 min in the dark. Then, 100 µL of Binding Buffer 1X and 50 µL of Precision count beads were added. Finally, 5 µg/mL propidium iodide is added at the last minute before flow cytometry analysis (10,000 events analyzed per condition). Live cells, apoptotic cells, and fluorescent beads populations were gated in order to get the number of events for each population. Cell viability was determined by the ratio of the number of alive cells events (AnV-/IP-) and that of apoptotic cells events (AnV+/IP-, AnV+/IP+) comparing to the total number of events. Cell proliferation was determined based on absolute cell counts calculated as follow: absolute cell count (cells/µl) = (cell count/”precision count beads” counts) x ”precision count beads” concentration (beads/µl). Data were acquired using LSRII (BD Biosciences) or Attune NxT (Thermo Fisher Scientific) flow cytometer.
BET inhibitors treatments – Dose response experiments were performed on all DLBCL and BL cell lines in order to determine IC50 of 6 BET inhibitors of different selectivity. JQ1 was synthesized as described previously (35). I-BET726, I-BET762 and RVX-208 have been purchased from Selleckchem. GSK-046 has been purchased from Chemietek and ABBV-744 from Abmole. All compounds were dissolved in DMSO. Cell lines were treated for 6h to 72h at the indicated doses.
Statistics - Statistical analyses and graphical representations were performed with R (release 4.1.0) or GraphPad Prism (v10) as indicated
Mixed nonlinear statistical modeling - The general specification of models is: Mij = b0 + b1 Treatmentij + b2 Gene_Levelij + b3 Treatmentij X Gene_Levelij + uj, where Mij is the number of dead cells of data i within cell line j, beta0 is the fixed intercept term, b1 is the fixed coefficient of the covariate Gene_Level (RNA-seq corresponding to gene expression), b2 is the fixed coefficient of the covariate Treatment representing the effect of JQ1 treatment versus control DMSO, b3 represents the fixed effect associated with the Treatment by Gene_Level interaction and uj is the random effect associated with the intercept for cell line j. Nonlinear relationship between cell mortality and Gene_Level was explored by restricted cubic spline (RCS) modeling. Models were fitted via restricted maximum log likelihood (REML). All statistical analyses from models were performed after non-parametric bootstrap data resampling with replacement (2,000 replicates) with bootstrap samples taken independently within each cell line considered as stratum. The overall statistical significance of the interaction terms between Treatment covariate and RCS covariates evaluating the nonlinearity hypothesis was tested by a Wald test. A REML-based likelihood ratio test was used to test the variance of the random cell line effects. Both nonlinearity of models and random effects were retained with p < 0.001. Statistical significance of predicted JQ1/DMSO ratios of cell mortality number at various percentiles of Gene_Level (percentile 10th to percentile 100th) were performed by a Wald test. Elasticity (E) was used to compare the effect of different genes on cell mortality M. The elasticity of M with respect to Gene_Level is the proportional (%) change in M for a proportional (1%) change in Gene_Level. Data were analyzed using Stata 16.1. Estimate margin effects (predicted mortality) in models using transformed gene level data with RCS was obtained by the f_able post-estimation command.