Genomic analyses identify recurrent MEF2D fusions in acute lymphoblastic leukaemia

Chromosomal rearrangements are initiating events in acute lymphoblastic leukaemia (ALL). Here using RNA sequencing of 560 ALL cases, we identify rearrangements between MEF2D (myocyte enhancer factor 2D) and five genes (BCL9, CSF1R, DAZAP1, HNRNPUL1 and SS18) in 22 B progenitor ALL (B-ALL) cases with a distinct gene expression profile, the most common of which is MEF2D-BCL9. Examination of an extended cohort of 1,164 B-ALL cases identified 30 cases with MEF2D rearrangements, which include an additional fusion partner, FOXJ2; thus, MEF2D-rearranged cases comprise 5.3% of cases lacking recurring alterations. MEF2D-rearranged ALL is characterized by a distinct immunophenotype, DNA copy number alterations at the rearrangement sites, older diagnosis age and poor outcome. The rearrangements result in enhanced MEF2D transcriptional activity, lymphoid transformation, activation of HDAC9 expression and sensitive to histone deacetylase inhibitor treatment. Thus, MEF2D-rearranged ALL represents a distinct form of high-risk leukaemia, for which new therapeutic approaches should be considered.

The samples used for gene expression analysis are divided into 3 groups: (1) samples with MEF2D-BCL9 fusion, (2) other MEF2D fusions (MEF2D-others) (3) or without MEF2D rearrangement (Others). The gene expression levels for MEF2D (a) and BCL9 (b) are shown in rlog. One-way ANOVA test was performed and followed by Tukey's multiple comparisons test. ***, P ≤ 0.001; ****, P ≤ 0.0001. c, Bigwig files are displayed in UCSC Genome Browser with 2 B-ALL samples without MEF2D rearrangement and 4 samples with different MEF2D-BCL9 fusion isoforms showing lack of BCL9 expression in non-rearranged B-ALL samples, and elevated expression distal to the BCL9 rearrangement break points. RNAseq library for "Control-1" is unstranded polyA-selected mRNA library while "Control-2" is stranded total RNA library. Translocation break points from exon 9 (E9) or 10 (E10) for each MEF2D-BCL9 fusion isoforms are indicated by red arrows.

Supplementary Figure 3. Identification of MEF2D fusion protein expression by immunoblotting.
M/B-1 and M/B-3, two MEF2D-BCL9 isoforms; M/H, MEF2D-HNRNPUL1; M/S, MEF2D-SS18. Molecular weight of markers is indicated on the left side in kilodalton (kDa). Sample names are aligned above the corresponding lanes, with expected protein weights in kDa provided in parenthesis. Primary antibodies are shown on the right side. The bands corresponding to the fusion proteins are indicated by red arrows. Samples with indistinguishable target bands are not marked.

Supplementary Figure 4. Fluorescence in situ hybridization confirms MEF2D rearrangements.
For the top three rows of images, sequential FISH was performed first with probes for the 5' (red) and 3' (green) regions of MEF2D to demonstrate disruption of the locus. The second hybridization used a probe to the 3' region of BCL9 (green, right panel) to confirm rearrangement of the two loci. White arrows in the right hand panels indicate the fusions. The images in the last row show lack of IGH to BCL9 fusion.

Supplementary Figure 5. DNA copy number alterations at MEF2D and partner loci.
Single nucleotide polymorphism (SNP) 6.0 microarray data is shown for representative samples as log 2 ratio copy number data aligned to hg18 in the UCSC genome browser (track may be accessed at http://genome.ucsc.edu/cgibin/hgTracks?hgS_doOtherUser=submit&hgS_otherUserName=cmulligh&hgS_otherUserSessi onName=MEF2D_fusions_SNP_3_18Jan16_hg18). Copy number alterations (CNA) are shown as probe level data (vertical lines) above (copy gain) or below (copy loss) the x axis. The regions of gain and loss are marked by red and blue lines respectively. The breakpoints at BCL9 and MEF2D are shown across the panels by dotted vertical lines. a, data spanning from BCL9 to MEF2D region. Samples with MEF2D-BCL9 fusions were observed with CNA breakpoints at BCL9, but not for samples with other MEF2D rearrangements; all MEF2Drearranged cases were observed with CNA breakpoints at MEF2D. b, data zoomed in for MEF2D.

Supplementary Figure 6. Genomic rearrangements of MEF2D-BCL9.
Two cases with MEF2D-BCL9 fusion have been whole-genome sequenced and analyzed by Complete Genomics (Complete Genomics, Inc., CA). a, Sample SJCOGALL010915 with translocation at 3' of exon 6 (E6) on MEF2D and 5' of exon 9 (E9) on BCL9, and b, SJCOGALL010884 with translocation at 3' of exon 5 (E5) on MEF2D and 5' of exon 10 (E10) on BCL9. Chromosomal break points of the MEF2D-BCL9 rearrangements were identified in intron regions and consistent with the fusions identified by RNAseq data (Fig. 1). The assembled contigs are presented and the sequences from MEF2D, BCL9 and random insertions are shown in red, blue and dark colors, respectively. The contigs are mapped to human reference genome and shown in UCSC genome browser, with MEF2D and BCL9 segments indicated by red and blue bars.

Supplementary Figure 7. Ras pathway mutations in different ALL cohorts.
Ras pathway mutations (NRAS, KRAS, NF1 and PTPN11) are presented by subgroups (see "Final Group", Supplementary Data 1). Only the groups with more than 10 cases are shown. "Ras" in the legend represents the four Ras pathway genes. Distribution of Ras mutations across different ALL subgroups are significantly different (Pearson's Chi-squared test, P value = 8.82e-06).

Supplementary Figure 8. Heatmap of top 1000 MAD genes in RNAseq samples.
Unsupervised hierarchical clustering of 1000 genes ranked highest median absolute variation (MAD) across the 219 samples showed clustering of 19 MEF2D-rearranged cases, with the exception of the case harboring MEF2D-CSF1R, which clustered with Ph-like ALL samples. Genes with top 800 and 500 MAD value were also tested by unsupervised clustering and also observed with similar clustering patterns, and the 19 MEF2D-rearranged samples were always clustered as a tight subgroup (data not shown). D30C, Depth ≥30x coverage; stranded-PE100, total stranded paired-end 100bp RNAseq; unstranded-PE100, unstranded mRNA paired-end 100bp RNAseq; unstranded-PE75, unstranded mRNA paired end 75bp RNAseq. Ex vivo panobinostat sensitivity test in different subtypes of B-ALL.

Supplementary Figure 18. Expression of proteins by retroviruses by immunoblotting.
Protein weight (kDa) is put in parenthesis at the top of each lane except MIG. Target protein is marked by "*" on top of specific band.  P=positive at an intensity typical for most cases of ALL; this is usually higher than normal cells for CD58 and CD9, and comparable to normal for CD19 and CD45 N=Negative; includes cases with only focal (<10%) positive cells H=High; at the level of normal hematogones L=Low For CD10, the number refers to the log decade of positivity. Most cases of B ALL would score as 3 or 3-4