Selective Enhancer Dependencies in MYC-Intact and MYC-Rearranged Germinal Center B-cell Diffuse Large B-cell Lymphoma

Abstract

Aberrant MYC activity defines the most aggressive GCB-DLBCLs. We characterized a mechanism of MYC transcriptional activation via a native enhancer that is active in MYC-intact GCB-DLBCL, establishing fitness-sustaining cis- and trans-regulatory circuitry in GCB-DLBCL models that lack MYC enhancer-hijacking rearrangement. See related commentary by Mulet-Lazaro and Delwel, p. 149.

Figure 1.

Overview of CRISPRi screening in lymphoma cell lines. A, Design strategy for sgRNA library and CRISPRi screen (NFR CRISPRi) targeting candidate enhancers in the MYC locus and RP regions, TF genes, and controls. B, Comparison of CRISPRi screening results in the MYC locus with the tiling sgRNA library [the y-axis shows the log₂ fold change (FC) averaged over a 20-sgRNA sliding window] and NFR-focused sgRNA library (the y-axis shows the log₂ fold change of individual sgRNAs) in MCL (Granta-519, SP-49) and GCB-DLBCL (Karpas-422) cell lines. Essential 5′ elements in MCL cell lines (NFR 148 and NFR 150) correspond to previously described Notch/RBPJ regulated MYC enhancers (21). C, Schematic diagram of chromosomal regions targeted in NFR-focused CRISPRi screens, including MYC locus regions and cell line–specific RP regions. The color code used for each region corresponds to that used for individual NFRs in (D). See Supplementary Table S2 for hg19 coordinates of each region. D, Summary of fitness effects for all targeted candidate enhancers (NFRs) in eight lymphoma NFR CRISPRi screens. Each data point represents one NFR, with log₂ fold change (Y-axis) and FDR P value rank (X-axis) calculated by the MAGeCK RRA algorithm from all sgRNAs targeting that interval. Target NFRs are ranked along the X-axis by lowest negative selection FDR P value (left to right, intervals with log₂ fold change <0) or lowest positive selection FDR P value (right to left, intervals with log₂ fold change >0). Intervals are color-coded by genomic region as in (C). See Supplementary Table S2 for NFR and sgRNA genomic position and Supplementary Table S3 for details of MAGeCK analysis.

Figure 2.

MYC activation via intrachromosomal enhancer-hijacking rearrangements. A, H3K27ac ChIP-seq and CRISPRi screening results for the MYC locus and fusion partner locus in JeKo-1 cells. Input chromatin coverage shows regions of genomic copy gain. Black arrows at bottom indicate chromosomal fusions, with breakpoints marked by vertical dotted lines. The position of the most essential enhancer identified in this screen (NFR 108) is highlighted. Note that the second panel is shown in reverse genomic orientation. B, H3K27ac ChIP-seq and CRISPRi screening results for the MYC locus and fusion partner loci in OCI-Ly1 cells. Input chromatin coverage shows regions of genomic copy gain. Chromosomal breakpoints, fusions, and strongly essential enhancers are indicated as in (A). Note that all panels except the first are shown in reverse genomic orientation. C,MYC transcript levels (qRT-PCR) after CRISPRi repression of selected enhancers in JeKo-1 cells (error bars show the SD of n = 5 technical replicates). D,MYC transcript levels (qRT-PCR) after CRISPRi repression of indicated targets in OCI-Ly1 cells. Cells were transduced with one tagBFP reporter–expressing and one mCherry reporter–expressing sgRNA vector and then flow sorted to obtain dual-transduced populations prior to dox induction of KRAB-dCas9 (two-tailed Student t test of n = 3 technical replicates for dual-enhancer–targeted samples vs. other samples; error bars are 95% CI). E, Detail of sgRNA CRISPRi scores in Jeko-1 and ChIP-seq data in JeKo-1 (H3K27ac) and GM12878 B-lymphoblastoid cell line (CTCF and EBF1) at NFR 108. F, Depletion of EBF1 and RBPJ promoter-targeting sgRNAs in eight lymphoma CRISPRi screens. Circles are sized proportionally to −log₂(FDR), with neg|FDR used for log₂FC <0 and pos|FDR used for log₂FC >0. Filled circles represent neg|FDR <0.01. G,MYC transcript levels (qRT-PCR) after CRISPRi knockdown of indicated target genes in selected cell lines (for JeKo-1, error bars show the SD of n = 2 biological replicates, for SP-49, error bars show the SD of n = 5 technical replicates). *, P < 0.05; **, P < 0.01; ***, P < 0.001. FC, fold change; neg ctrl, negative control; pos, positive.

Figure 3.

Activation of MYC by ternary complex–bound modules of the BCL6 super-enhancer. A, Schematic diagram of the normal BCL6 locus and parsimonious models for structural rearrangements involving the MYC and BCL6 locus in three DLBCL cell lines based on optical mapping and 4C-seq data. Red lines indicate intra- or inter-chromosomal fusions. See Supplementary Fig. S3A–S3C for details. B, CRISPRi sgRNA depletion or enrichment (log₂ fold change) for BCL6 distal super-enhancer regions in the MYC::BCL6-LCR+ cell lines SU-DHL-4 and WSU-DLCL2. Also shown is ChIP-seq signal for H3K27ac and ternary complex TFs MEF2B and OCT2 in SU-DHL-4 (scale: fragments per million mapped fragments). BCL6 locus super-enhancer regions SE1 and SE3 are shown as previously defined (15); region SE1 is equivalent to BCL6-LCR. C, Plot of NFR-level sgRNA depletion or enrichment (MAGeCK RRA log₂ fold change, y-axis) for targeted intervals in BCL6 distal super-enhancers in six GCB-DLBCL cell lines. Circles are sized proportionally to −log₂(FDR), with neg|FDR used for log₂FC <0 and pos|FDR used for log₂FC >0. Filled circles represent neg|FDR <0.01. The three MYC::BCL6-LCR+ cell lines are indicated in the legend with a red box. D, Change in viable cellularity for dox-inducible KRAB-dCas9–expressing MYC::BCL6-LCR+ DLBCL cell lines expressing the two indicated sgRNA lentivectors. Cells were transduced with both vectors, sorted for dual tagBFP- and tagRFP-expressing cells, and then grown with or without dox for 7 days. The Y-axis shows the ratio of CellTiter-Glo signal for dox treated/untreated cells (two-tailed Student t test of n = 3 biological replicates per condition for dual-enhancer–targeting sgRNA-transduced vs. dual-control sgRNA-transduced; error bars are 95% CI). E, Differential expression of MYC and BCL6 (qRT-PCR) for populations of MYC::BCL6-LCR+ DLBCL cell lines expressing dox-inducible KRAB-dCas9 and two sgRNA vectors as in (D) (two-tailed Student t test of n = 3 technical replicates per condition for dual-enhancer–targeting sgRNA-transduced vs. dual-control sgRNA-transduced, color-coded by measured transcript per legend; error bars are 95% CI). F, Differential expression of MYC and BCL6 (qRT-PCR) in MYC::BCL6-LCR+ DLBCL cell lines expressing dox-inducible KRAB-dCas9 after transduction with indicated sgRNAs targeting MEF2B, POU2F2, or POU2AF1 (error bars are the SD of n = 5 technical replicates per sgRNA; bars are color-coded by measured transcript per legend). ***, P < 0.001. FC, fold change; neg ctrl, negative control; pos, positive.

Figure 4.

MYC-intact GCB-DLBCL is dependent on a 3′ developmental MYC enhancer. A, H3K27ac ChIP-seq signal and CRISPRi sgRNA depletion or enrichment (log₂ fold change) from MYC-intact GCB-DLBCL cell lines. Yellow highlight shows the location of GME-1. B, Comparison of H3K27ac ChIP-seq data and CRISPRi sgRNA depletion in GCB-DLBCL cell lines for genomic regions containing GME-1 and the BCL6-LCR elements NFR 31 and NFR 41. Also shown at top are H3K27ac ChIP-seq data from normal human centroblasts and naïve B cells (87) and GCB-DLBCL patient biopsies [“HGB” datasets were previously published (15); “MHR” datasets are from the current work]. Note lack of GME-1 acetylation in DLBCL biopsies and cell lines with MYC rearrangement (MYC-R, indicated at right). C, Change in viable cellularity for MYC-intact DLBCL cell lines or MYC-rearranged OCI-Ly1 (dox-inducible KRAB-dCas9–expressing) after transduction with indicated sgRNAs targeting MYC or BCL6 locus NFRs. Each replicate was divided into media with or without dox, and the ratio of CellTiter-Glo signal (induced/uninduced) was determined at day 6 after induction (two-tailed Welch’s t test of pooled biological replicates, n = 3 per sgRNA, for two NFR 195-targeting sgRNAs vs. two control sgRNAs; error bars are 95% confidence interval). D, Change in MYC transcript levels (qRT-PCR) for MYC-intact DLBCL cell lines or MYC-rearranged OCI-Ly1 (dox-inducible KRAB-dCas9 expressing) after transduction with indicated sgRNAs targeting MYC or BCL6 locus NFRs (error bars are the SD for n = 3–5 technical replicates per sgRNA). E, H3K27ac ChIP-seq signal from sorted primary human B lineage subpopulations (Blueprint consortium) showing acetylation specific to GC B cells (GCB) at the GME-1 enhancer. Also abbreviated are naïve B cells, CD38-negative (naïve B, 38-), memory B cells – class-switched [“Mem B (cs)”], and plasma cells (“PC”). Conservation (Phastcons) and CRISPRi signal shown for comparison. “Homology analysis” shows position of intervals evaluated for conservation in other mammalian genomes. F, ATAC-seq data from sorted primary murine B lineage subpopulations (Immgen consortium), centered on sequences syntenic to human NFR 195. Note accessibility specific to GC B cell centroblasts “GCB-(CB)” and centrocytes “GCB-(CC).” Intervals with substantial homology between hg38 and mm10 are indicated at bottom. Note that subregions of human NFR 195 (“a” and “b”) are conserved in mouse, whereas human NFR 196 (“c” and “d”) lacks an identifiable homolog in mouse. G, Left: Pseudobulk ATAC-seq profiles for human tonsil B-cell populations defined via integrative single-cell RNA-seq (scRNA-seq) and ATAC-seq (36) at genomic regions containing essential MYC (NFR 195-196) and BCL6 (NFR 31 and NFR 41) locus enhancers identified in GCB-DLBCL cell lines. Populations were defined per the original publication, with population names followed by asterisks consisting of two or more similar populations that were merged (see “single-cell ATAC-seq and RNA-seq analysis” in “Methods” for details of populations merged). GC DZ and LZ population names are abbreviated: “DZ nonproliferative,” “DZ–LZ transition,” “LZ–DZ reentry commitment,” “LZ proliferative,” and “LZ–DZ transition.” The arrow at left indicates the proposed sequence of these populations in the GC cycle as per the original publication. Normalized Y-axis signal range (Signac CoveragePlot) is labeled at top for each genomic region and is the same for all tracks. Right: Relative expression of MYC and BCL6 transcripts from scRNA-seq data in the same populations as assigned in the original publication (harmonized between scATAC-seq, multiome, and scRNA-seq). **, P < 0.01; ***, P < 0.001; ns, not significant, P ≥ 0.05. expr., expression; FC, fold change; neg ctrl, negative control.

Figure 5.

The GME-1 MYC enhancer shows selective transactivation in MYC-intact cell lines and is directly regulated by the OCT2 ternary complex. A, Detail of the GME-1 enhancer with ATAC-seq, H3K27 ChIP-seq, and ChIP-seq for OCT2, OCA-B, MEF2B, and p300 in the indicated cell lines. OCT2 BioChIP in BJAB cells is from Hodson and colleagues (39) and represents binding of a transgenically expressed biotinylated OCT2 protein, whereas all other ChIP-seq datasets profile endogenous proteins. Inset at bottom shows further detail of the Phastcons conserved element track for sub-region 195B (see Supplementary Fig. S5E for further detail), with the positions of the conserved OCT2 motif and the genomic duplication identified in the SU-DHL-5 cell line indicated. B, Differential expression of MYC and BCL6 (qRT-PCR) in GME-1–dependent DLBCL cell lines expressing dox-inducible KRAB-dCas9 after transduction with indicated sgRNAs targeting MEF2B, POU2F2, and POU2AF1 (error bars are the SD of five technical replicates per sgRNA; bars are color-coded by measured transcript per legend). C, Transcriptional reporter assays conducted in MYC-intact and MYC::BCL6-LCR+ DLBCL cell lines with conserved subregions of essential MYC and BCL6 NFRs cloned into the STARR-seq_Ori luciferase vector. See Supplementary Fig. S5E–S5H for details of the indicated regions. The signal ratio and statistical comparison for MYC enhancer 195B vs. each BCL6 enhancer is indicated for each cell line (two-tailed Student t test of biological replicates (n = 3); error bars are 95% confidence interval). D, Transcriptional reporter assays in MYC-intact cell lines for constructs bearing the reference NFR 195B enhancer region or the same region with deletion of the OCT2 motif (“195B_MYC_OCT2Del”) indicated in (A) [two-tailed Student t test of biological replicates (n = 3); error bars are 95% CI]. E, Transcriptional reporter assays conducted in HT cells expressing inducible KRAB-dCas9 and transduced with the indicated sgRNAs. Following puromycin selection, cells were electroporated with the indicated constructs and induced with dox for 48 hours prior to Dual-Glo luciferase assay [two-tailed Student t test of biological replicates (n = 3); error bars are 95% CI]. F, Transcriptional reporter assays in MYC-intact cell lines for constructs bearing the reference NFR 195B enhancer region or a clone of the same region from the SU-DHL-5 cell line bearing a partial tandem duplication (“195B_MYC_Dup”) as indicated in (A) [two-tailed Student t test of biological replicates (n = 3); error bars are 95% CI]. **, P < 0.01; ***, P < 0.001; ns, not significant; P ≥ 0.05. FC, fold change; neg ctrl, negative control.

Figure 6.

The GME-1 MYC enhancer is recurrently amplified and shows topological interactions with MYC in DLBCL biopsies lacking MYC rearrangements. A, Positions of focal genomic gains of the GME-1 MYC enhancer in primary DLBCL samples. Light red bars mark the positions of isolated genomic copy gains involving the GME-1 enhancer detected in DLBCL WGS datasets that lacked evidence for a long-distance MYC rearrangement. See Supplementary Fig. S6A for details of breakpoint and copy-number analysis for these events. B, Contact matrices for Hi-C performed on GCB-DLBCL biopsies, showing topological interactions within the MYC locus for two MYC-intact lymphomas (top) and between the MYC and BCL6 loci in two MYC::BCL6-LCR+ lymphomas (bottom). Corresponding H3K27ac ChIP-seq data from the Karpas-422 cell line and normal GC B cells (GCB; Blueprint consortium) are shown at top. Color-coded markers indicate the location of the MYC and PVT1 promoters (MYC-P and PVT1-P) and GME-1 and BCL6-LCR enhancers on the X and Y axes. Black ellipses highlight increased interactions between the GME-1 enhancer and the MYC and PVT1 promoters. Black arrows indicate chromosomal fusions identified by Hi-C Breakfinder that link the MYC gene to the BCL6-LCR. Note that the fusion in AGRR_002 is telomeric to the BCL6-LCR but in reverse orientation [chr8 (+) strand fused to chr3 (−) strand]. C, Virtual 4C contact profiles (red) for topological interactions with the MYC promoter in the four lymphoma Hi-C datasets shown in (B). The Y-axis scale shows normalized interactions (Knight–Ruiz matrix balancing) with the MYC promoter for each genomic bin. Significant looping interactions detected by hiCDetectLoops (P < 0.05) are indicated by arcs at the bottom of each profile. NFR CRISPRi screen data for Karpas-422 and SU-DHL-5 are shown at top for comparison (black, sgRNA log₂ fold change is shown on the Y axis). GME-1 and other enhancers detected as essential in Karpas-422 are highlighted—note correlation with loops detected in MYC-intact biopsies.

Figure 7.

The PVT1 promoter is a cis-repressor of MYC activation by native enhancers and a subset of rearrangements. A, Genomic coverage plots of WGS data from three DZSig+ DLBCL biopsies (13) that show deletion of the PVT1 promoter (black arrow). The position of the GME-1 enhancer is labeled (red arrow and shading) to highlight copy gain of this element in biopsies 03-1110 and 01-20774. The position of a focal copy gain of NFR-202 in 01-20774 is also indicated. See Supplementary Fig. S7A for details of all SVs involving the MYC TAD in these datasets. B, Depletion or enrichment (log₂ fold change) for sgRNAs targeting the MYC and PVT1 promoter regions in MYC locus tiling (20 sgRNA sliding window) and NFR-focused CRISPRi screens (log₂ fold change of individual sgRNAs). Cell line names are color-coded as follows: blue, MYC-intact GCB-DLBCL; red, MYC::BCL6-LCR+ GCB-DLBCL with the breakpoint downstream of the PVT1 promoter; green, MYC-intact MCL; and black, DLBCL and MCL cell lines with MYC rearrangement breakpoint between the MYC gene and the PVT1 promoter. C, Volcano plots of MAGeCK analysis on NFR CRISPRi screens (cropped to highlight enriched regions), showing enrichment of sgRNAs targeting the PVT1 promoter (NFR 156). Cell line names are color-coded by subtype and rearrangement as in (B). D, Relative PVT1 and MYC transcript levels (qRT-PCR) in GME-1-dependent cell lines (dox-inducible KRAB-dCas9 expressing) after transduction with indicated sgRNAs targeting the MYC or PVT1 promoters (one-tailed Welch t test of mean values for four PVT1-P-targeting sgRNAs vs. two control sgRNAs; error bars show the SD of n = 5 technical replicates). E, Summary of MYC activation mechanisms identified in MYC-rearranged and MYC-intact mature BCL cell lines. *, P < 0.05; **, P < 0.01. CLL, chronic lymphocytic leukemia; FC, fold change. Neg. ctrl, negative control.