Genetic and Functional Characterization of NUP98-Rearranged Leukemia

NUP98 rearrangements (NUP98r) are recurrent alterations in pediatric acute myeloid leukemia (AML) and are associated with dismal outcomes (PMID:21813447). These structural variants result in fusion oncoproteins comprised of the N-terminus of NUP98 and the C-terminus of fusion partners. Among more than 30 fusion partners discovered so far, the associations between NUP98::NSD1 with myeloblastic AML and NUP98::KDM5A with acute megakaryocytic/erythroid leukemia (AMKL/AEL) have been appreciated, which often harbor distinct cooperating mutations (e.g., NUP98::NSD1 with FLT3-ITD and WT1 mutations). In contrast, other fusions, such as NUP98::RAP1GDS1 or NUP98::HOXA9, can be found also in myelodysplastic syndrome (MDS) or T-cell acute lymphoblastic leukemia (T-ALL) (PMID:32766874). However, how these fusions and cooperating mutations contribute to the specific phenotypes of NUP98r leukemias is poorly understood.

To investigate the genetic background of NUP98r leukemia, we collected RNA-seq, WGS, and WES data for 91 pediatric cases from St. Jude (n=71),AEIOP (n=3), and the TARGET study (PMID:29227476, n=17; Panel.A), which included AML (n=82), MDS (n=4), and T-ALL (n=5). NUP98::NSD1 (n=47, 51.6%) and NUP98::KDM5A (n=26, 28.6%) were the most common fusions, in line with previous observations. NUP98::RAP1GDS1 (n=6, 6.6%), an understudied NUP98 fusion, was the third most common fusion in this study. We also confirmed frequent mutations of WT1 and FLT3-ITD in NUP98::NSD1 AML and entire or partial chromosome 13 loss involving RB1 locus in NUP98::KDM5A AMKL, while we observed frequent mutations in RAS or epigenetics-related genes, CCND2/3, and RUNX1 preferentially in cases without alterations of WT1, FLT3 or RB1 loss. NUP98::RAP1GDS1 and NOTCH1 mutations were strongly enriched in T-ALL. These findings suggest that fusions and cooperating alterations both contribute to the disease phenotypes.

To study how NUP98 fusion oncoproteins (FOs) drive leukemogenesis and specific phenotypes, we applied Cleavage Under Targets and Release Using Nuclease (CUT&RUN) to determine NUP98 FO binding for a primary sample (myeloblastic with NUP98::NSD1), CHRF-288-11 cell line (megakaryocytic with NUP98::KDM5A), and HA-tagged CD34 NUP98::KDM5A model (Panel.B-top). CUT&RUN using antibodies against the N-terminus NUP98 and histone modifications (H3K4me3, H3K27ac) revealed that these models shared NUP98 FO binding sites at the HOXA/B and RUNX1 loci, together with active histone marks. In contrast, binding patterns at differentiation-related gene loci were heterogenous. NUP98::NSD1 uniquely binds to CEBPA, which encodes a master regulator of myeloid differentiation, contrary to NUP98::KDM5A binding to GFI1B encoding regulator of erythromegakaryocytic differentiation. These data indicate that NUP98 FOs directly and uniquely regulate differentiation-related genes. We also performed CUT&RUN in AML cells with RUNX1::RUNX1T1 or UBTF tandem duplications, in which the N-terminus NUP98 antibody did not show any detectable peaks, confirming the specificity of the N-terminus NUP98 signals in NUP98r AML models. Experiments using a HA antibody in HA-tagged NUP98::KDM5A CD34 models confirmed that fusion oncoproteins are binding to these target regions. However, the C-terminus NUP98 antibody also detected similar signals as the N-NUP98 antibody, suggesting that wild-type NUP98 is also recruited to NUP98 FO target genes, possibly by a heteromer formation with NUP98 FOs and other NPC components.

To understand the possible association between somatic mutations and the genomic binding of NUP98 FOs, we next investigated chromatin immunoprecipitation sequencing (ChIP-seq) data from the ENCODE study (PMID:22955991) for RB1, WT1, and STAT5 (downstream FLT3-ITD). Binding peaks of these proteins partly coincided with NUP98 peaks on representative target genes (Panel.B-bottom), suggesting that frequent mutations in NUP98r leukemia converge on the genomic interaction of NUP98 FOs.

In conclusion, genomic characterization of NUP98r leukemia showed the association among fusion partners, cooperating mutations, and disease phenotypes. Epigenetic profiling revealed that NUP98 FOs directly regulate target genes in a fusion and differentiation-specific manner and provide a basis for further study to investigate how NUP98 FOs and cooperating partners drive leukemogenesis and disease phenotypes.