DOT1L inhibition is lethal for multiple myeloma due to perturbation of the endoplasmic reticulum stress pathway

The histone 3 lysine 79 (H3K79) methyltransferase (HMT) DOT1L is known to play a critical role for growth and survival of MLL-rearranged leukemia. Serendipitous observations during high-throughput drug screens indicated that the use of DOT1L inhibitors might be expandable to multiple myeloma (MM). Through pharmacologic and genetic experiments, we could validate that DOT1L is essential for growth and viability of a subset of MM cell lines, in line with a recent report from another team. In vivo activity against established MM xenografts was observed with a novel DOT1L inhibitor. In order to understand the molecular mechanism of the dependency in MM, we examined gene expression changes upon DOT1L inhibition in sensitive and insensitive cell lines and discovered that genes belonging to the endoplasmic reticulum (ER) stress pathway and protein synthesis machinery were specifically suppressed in sensitive cells. Whole-genome CRISPR screens in the presence or absence of a DOT1L inhibitor revealed that concomitant targeting of the H3K4me3 methyltransferase SETD1B increases the effect of DOT1L inhibition. Our results provide a strong basis for further investigating DOT1L and SETD1B as targets in MM.


Targeting of SETD1B by CRISPR
To knockout SETD1B, different inducible sgRNA constructs were cloned into a lentiviral vector (pNGx_ LV_g016; Cellecta) with a tetracycline-inducible U6 promoter, and the sgRNA with the most efficient knockout (sg_3) was selected with the target sequence (sense) CATGTGCAAGAAGTATGGGG. Infected pools of RPMI8226 and KMS-27 cells were established and knockout was induced by addition of 100 ng/ml doxycycline. Knockout efficiency was verified by western blot using a SETD1B antibody (Abcam, #ab113984). The immunoblot was probed with an actin antibody (Millipore, #MAB1501 clone C4) for normalization. Detection was performed with a western blotting detection kit (SuperSignal West Femto Maximum Sensitivity Substrate kit, Thermo Fisher).

RNA-seq of MM cell lines and analysis
12 cell lines (MM1-S, KMS-20, KMS-21BM, OPM-2, RPMI8226, NCI-H929, AMO-1, KMM-1, KMS-11, KMS-27, U266B1, L-363) were treated for 6 days with DMSO or 1 µM of the DOT1L inhibitor SGC0946 in biological triplicates. Total RNA was extracted using the RNeasy Plus Mini Kit (Qiagen, #74136) according to the manufacturer's instructions and submitted to RNA sequencing for quantification of transcript expression. Raw sequencing data was aligned to the human genome (hg19 assembly) using the EQP software [3]. The raw genelevel counts were normalized with the Voom procedure [4] and differential expression analysis was conducted with the Limma R/Bioconductor software [5]. Each SGC0946-treated cell line was contrasted to its respective DMSO baseline and the treated SGC0946-sensitive cell lines were also contrasted to the treated insensitive ones upon DMSO baseline subtraction. Pathway analysis of the RNA-seq data was described previously [6]. For the purpose of gene set enrichment analysis, the differentially expressed genes, obtained as described above by contrasting the SGC0946-treated sensitive to insensitive cell lines, were ranked in descending order according to the fold change signed minus log10 P-value of the Limma statistical test. Enrichment of the gene sets extracted from Metacore process network database was assessed with the Kolmogorov-Smirnov (K-S) test. The gene sets significantly over-represented in the SGC0946-treated sensitive or insensitive cell lines were selected by plotting the K-S statistics Dn over the minus log10 P-value of the K-S test.

RNA-seq of RPMI8226 Cas9 with inducible sgSETD1B and analysis
RPMI8226 Cas9 cells (with an inducible sgSETD1B expressing vector) were pretreated during 3 days with doxycycline (100 ng/ml) or not. Cells were then treated with either DMSO, 1 µM SGC0946, doxycycline (100 ng/ml), or combination of doxycycline and 1 µM SGC0946 for another 3 days, followed by total RNA extraction using the RNeasy Plus Mini Kit (Qiagen, #74136) according to the manufacturer's instructions. Illumina-compatible libraries for RNA-sequencing were generated by using the Illumina TruSeq Stranded Total RNA Sample Preparation Kit (including Ribo-Zero Gold rRNA depletion) and custom Library-Adapters ordered from IDT. 12 cycles of PCR were applied for library enrichment. Pathway analysis of the RNA-seq data was described previously [6].Gene set enrichment analyses (GSEAs) [7] were performed on pre-ranked gene lists using the gene list analysis module of the GSEA desktop java version 2.01.12.

ISMARA analysis
ISMARA analysis was performed using the online version of this tool [8]. Activity of a transcription factor in a sample defines how much expression (log2 transformed) of a promoter in this sample changes per 1 binding site of this TF in proximity of this promoter.

Bioinformatic analysis, 12 MM cell lines
For each processed sample, the raw H3K79me2 ChIP-seq sequencing reads were mapped to the human genome (hg19 assembly) with the Bowtie2 software. The sequencing coverage was calculated over genome from the BAM output of the alignment and displayed as a BigWig file in the UCSC genome browser. H3K79me2 positive DNA regions were identified from ChIP-seq data by calling peaks with MACS2.0.10. Differential gene expression was merged with ChIP-seq data in the following way: the consensus ChIP-seq peaks over the 6 sensitive cell lines were computed with the DiffBind 1.12.3 Bioconductor package (with a minimum overlap of peaks in r 4cell lines), and as a measure of ChIP-seq signal for each peak and sensitive cell line RPKM values were calculated with DiffBind. A global composite ChIPseq signal score was calculated for each peak by summing over the individual cell line ChIP-seq signals. Gene-centric differential expression data (treated SGC0946-sensitive to insensitive cell lines contrast) were merged with the peakcentric ChIP-seq signal by genomic region proximity using the annotate function of the ChIPpeakAnno 3.4.4 Bioconductor package. Metagene profiles were generated as follows: for every human protein coding gene of the ENSEMBL 73 database and per cell line, the base level coverage of ChiP-seq reads was calculated in a -2kb +3kb window centered at gene transcription start sites (TSS) and expressed as counts-per-million reads (CPM). Per cell line the base position coverage averages were calculated over genes overlapping with ChIP-seq sensitive / insensitive consensus peak sets (obtained as described above) and plotted as smoothed lines (smoothing factor 100 bases). All data were calculated with standard packages from R-3.1.3 and associated Bioconductor release.

Bioinformatic analysis, second ChIP-seq experiment with RPMI8226 and KMS-27 cells
Histone ChIP-seq aligment was performed using the AQUAS pipeline (https://github.com/kundajelab/ chipseq_pipeline). Drosophila normalization was performed by computing the reference-adjusted read per million (RPM) [9]. Each set of target genes was compared to a set of control genes, where matching was performed on the gene expression level and gene width. Where logfold changes were available, control genes were required to show absolute logFC < 0.5. Differential binding analysis was performed using DESeq2 with normalization factors. For each histone mark a set of consensus peaks occurring in at least 3 samples were used for downstream analysis. Statistical comparisons of metagene profiles were performed using t-tests.

In vitro growth curves
Cells were seeded into 24-well plates at a density of 0.5 × 10 6 / mL and a total volume of 1.5 mL and treated with the specified compound concentrations. Every 3 days, cell numbers were determined using a CASY model TT (Roche Innovatis) or a TC20 Automated Cell Counter (BioRad), the cells were then split back to the original seeding density if needed, and medium as well as compound were replaced. Based on microscopic inspection of the cultures, we had reason to believe that cell counts generated with the CASY instrument did not reliably exclude all dead cells. Therefore, percent of dead cells at the endpoint of the respective growth curves was measured by manually counting cells in a hemocytometer using Trypan blue dye exclusion. Samples counted on a TC20 were always stained with Trypan blue, allowing the instrument to exclude dead cells.

Cell proliferation assay
For proliferation assays with thapsigargin, cells were pretreated during 3 days either with DMSO or 1 µM SGC0946 and subsequently seeded on 96-well plates in triplicate at 80 000 cells per well and incubated with various concentrations of thapsigargin for 24 h followed by quantification of viable cells using CellTiter-Glo (Promega). IC50 values were determined with the XLFit Excel Add-In (ID Business Solutions).

Genetic alterations in multiple myeloma (MM) cell lines
We

Histone methyltransferase panel of biochemical assays
IC50 values for SGC0946 and Compound 11 in a panel of 12 histone methyltransferase assays, including DOT1L, were determined for as described before {Qi, 2012 #158}.

In vivo experiment in a MM1-S mouse xenograft model
The experimental procedures involving animal studies strictly adhered to the Association for Assessment and Accreditation of Laboratory Animal Care International guidelines as published in the Guide for the Care and Use of Laboratory Animals, and to Novartis Corporate Animal Welfare policies. Studies were performed at Novartis facilities. Subcutaneous tumors were induced by injecting MM1-S cells bearing a luciferase reporter (MM1-S-luc) in HBSS containing 50% BD Matrigel in the flank of NOD-SCID mice (NOD/MrkBomTac-Prkdscid). Compound 11 was formulated as solution in 20% Solutol HS15/80% Saline and administered subcutaneously. Treatment was started when the average tumor size had reached approximately 150 mm 3 (n = 8/group). Tumor size was measured twice a week with a caliper. Tumor volume was calculated using the formula (Length × Width) × π/6 and expressed in mm 3 . Data is presented as mean ± SEM. The statistical analysis was performed on delta tumor volume (ΔTVol) by comparing the treatment groups to the vehicle control group at endpoint by Kruskal-Wallis followed by Dunn's post-hoc test.

Quantitative real-time PCR
For mRNA expression analysis, total RNA was extracted from cells using an RNeasy mini kit (Qiagen, #74136) and reverse transcribed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, #4368814) according to the manufacturer's protocol. Real time PCR was then performed using Taqman

Viability assay and flow cytometry
Viability assay was performed by incubating the cells with Zombie Aqua Fixable Viability Kit (BioLegend, # 423101) 15 min at room temperature. For flow cytometer analysis, cells were incubated 20 min at 4° C with fluorochrome-conjugated mAbs specific for surface antigens and diluted in PBS containing 2% of FCS prior to analysis. The following monoclonal antibodies were used in this study: kappa APC (SouthernBiotech, #9180) and lambda PE (Southern Biotech, #9230-11). For intracellular staining, the cells were fixed and permeabilized using FIX & PERM Cell Fixation and Cell Permeabilization Kit (ThermoFisher Scientific, #GAS003)). Briefly, the cells were incubated 15 min at room temperature with Fixation medium. After a washing step, the cell were incubated with the Permeabilization medium and the recommended volume of intracellular antibodies (kappa-APC-Southern Biotech, #9180 and lambda-PE-Southern Biotech, #9230-11) for 20 min at room temperature. After washing, the cells were resuspended in PBS + 2% FCS for immediate analysis. Labeled cells were analyzed on a LSRFortessa (BD Biosciences) and the data were analyzed using FlowJo software. factors, whose change in activity (ISMARA score) upon DOT1L inhibition with SGC0946 as significantly different between sensitive and insensitive cell lines. P-values are from t-tests comparing the SGC0946-induced changes in sensitive vs insensitive cell lines, filtered for P < 0.05. Difference = (average change in insensitive cell lines) -(average change in sensitive cell lines). Table 3 Supplementary Table 4: Gene expression changes in RPMI8226 cells upon DOT1L inhibition, SETD1B knockout, or combination thereof after 3 days. See Supplementary Table 4