NSD2 Mutations in Pediatric ALL Leads to a Distinct Gene Expression Profile and Epigenetic Landscape That Is Cell Context Specific

Background: Outcomes for children with acute lymphoblastic leukemia (ALL) have dramatically improved, but survival for patients who relapse remains poor. Mutations in genes encoding epigenetic modifiers are present in the majority of patients at relapse. In particular, activating mutations in NSD2 (MMSET, WHSC1), namely the glutamate to lysine substitution at amino acid 1099 (p.E1099K), are among the most common such mutations in epigenetic regulators. NSD2 converts histone 3 lysine 36 (H3K36) into its dimethylated form (H3K36me2) which in turn leads to stereotactic inhibition of EZH2 mediated H3K27me3. We and others have established that this leads to changes in chromatin state and gene expression. However, the pathways by which this leads to a clonal advantage remains elusive.

Design/Method: We previously reported that overexpression of wild-type (WT) and p.E1099K mutant (EK) NSD2 in B-ALL cell lines led to unique cell context specific chromatin alterations and altered gene expression but did not lead to changes in proliferation or intrinsic drug resistance in vitro (Pierro et. al. Blood 2017 130:2474). We reasoned that these observations could be explained by the need for cooperating pathways that together with NSD2 EK lead to a clonal advantage. Thus we modulated expression of NSD2 using short hairpin RNAs (shRNAs) in the B-ALL cell line RS4;11 which harbors a heterozygous NSD2 EK mutation (NSD2 low). As a control, RS4;11 was also stably transduced with a non-targeting shRNA sequence (NSD2 high). Knockdown of NSD2 as well as decrease in H3K36me2 in NSD2 low lines was confirmed by Western Blot. Differences in gene regulation in NSD2 low cells were assessed by ChIPseq for CTCF, H3K9Ac, H3K27Ac, H3K36me2 and H3K27me3, and the results were correlated with RNAseq data. This data was then compared to RNAseq and ChIPseq data from REH and 697 NSD2 WT and EK overexpression cell lines in an effort to identify candidate genes or pathways preferentially regulated by the NSD2 EK mutation.

Results: NSD2 low cells displayed a distinct gene expression profile compared to NSD2 high with 301 upregulated and 573 downregulated genes (LFC 0.58, P = 0.05). When compared to gene expression data from our previously reported NSD2 overexpression cell lines, there was minimal overlap across cell lines with only 15 differentially expressed genes shared between RS4;11 NSD2 knockdown and REH EK overexpression cell lines and only 24 genes shared between RS4;11 NSD2 knockdown and 697 EK overexpression cell lines. Across all cell lines (RS4;11, REH and 697), only three genes (NSD2, SCN8 and PCNXL2) overlapped, all of which were upregulated in NSD2 high cell lines. Using less stringent criteria (LFC 0.26, P = 0.1), we observed greater overlap with 34 shared up and downregulated genes among lines. Of the shared genes, only ZNF521 which is overexpressed in NSD2 high cell lines, is known play a role in leukemogenesis. Moreover, RS4;11 pathway analysis revealed several biologically relevant pathways modulated by the NSD2 EK mutation such as Ras, integrin signaling, cholesterol/steroid biosynthesis, apoptosis and cell proliferation.

Significant differences were also observed across epigenetic marks between RS 4;11 NSD2 high and low cells. In accordance with previously published data, we observed a global decrease in the H3K36me2 mark in RS4;11 NSD2 low lines. When aligned with changes in histone marks, among genes downregulated in NSD2 low cells there was a clear correlation with acquisition of the repressive H3K27me3 mark (and a decrease in the H3K9Ac mark). However among genes upregulated in NSD2 low cells we saw paradoxical increases in the H3K36me2 mark and decreases in the H3K27me3. Furthermore, gene expression was also influenced by marks not directly regulated by NSD2, namely H3K27ac and H3K9ac, indicating that local NSD2 mediated epigenetic changes are not the sole regulator of gene expression.

Conclusion: The activating p.E1099K substitution in NSD2 leads to a distinct gene expression profile in B-ALL cell lines that is cell context dependent. Moreover, while there is significant overlap in the transcriptional profile between WT and EK overexpression, there are distinct differences possibly indicating novel properties of the pE1099K substitution beyond enzyme hyperactivation. Our findings also imply that NSD2 EK collaborates with other leukemia associated alterations that result in clonal selection.