UBTF tandem duplications generate a nuclear export signal motif that drives an XPO1 dependency in leukemia

Abstract UBTF tandem duplications (UBTF-TDs) define a high-risk molecular subtype of acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS) characterized by aberrant HOX gene expression and poor clinical outcomes. Our previous work revealed that UBTF-TD mislocalizes to chromatin regions dysregulated in UBTF-TD AML, including the HOXA/B gene clusters. However, the molecular basis for this mislocalization remains poorly understood. To identify UBTF-TD-specific binding partners on chromatin, we performed on-chromatin proteomics (qPLEX-RIME) on K562 cells transduced with UBTF wildtype (UBTF-WT) or UBTF-TD. RIME identified that Exportin-1 (XPO1) enriched for UBTF-TD, along with previously reported potential interactors like KMT2A, and we validated these interactions using proximity labeling (TurboID). XPO1 has been shown to interact with NPM1-mutant via nuclear export signal (NES) motifs, leading to leukemogeneosis. We thus hypothesize that UBTF-TDs may encode NES motifs. We evaluated amino acid sequences from 97 UBTF-TD AML/MDS cases with the NES prediction tool LocNES (PMID: 25515756) and found that 78% of cases harbored a de novo canonical NES motif within the duplicated region, most commonly Class 1a or Class 3 motifs. The remaining cases encoded imperfect NES-like sequences that may represent a new XPO1 interacting motif. Protein interaction assays (GST pull-down and SPR) confirmed that these exon 13-derived regions from representative UBTF-TDs (TD48, TD54, TD84) showed enhanced XPO1-RanGTP binding compared to UBTF-WT. Alanine mutagenesis of the NES motifs (NES-M) abolished XPO1 binding across these constructs, including UBTF-TD84, which contains the imperfect NES. We next asked whether these NES motifs are required for leukemic transformation. Expression of our three representative UBTF-TDs in cord blood CD34⁺ cells promoted proliferation in liquid culture and increased self-renewal capacity as measured by CFU assays. In contrast, NES-M variants displayed proliferation, self-renewal, and differentiation rates similar to UBTF-WT and vector conrols. This included reduced KIT expression, increased CD11b⁺ expression, and morphologic changes. Bulk RNA-seq revealed that CD34+ cells with NES-M had reduced HOXA/B and MEIS1 expression, suggesting that the NES motif is critical for maintaining the leukemic transcriptional program. To test whether the NES motif mediates chromatin localization of UBTF-TD at HOXA/B and MEIS1 loci, we performed CUT&RUN. UBTF-TD proteins localized to these loci, whereas NES-M variants did not. NES-M also showed reduced H3K27ac enrichment at these loci, consistent with low enhancer activity. Given the functional importance of the UBTF-TD-NES/XPO1 interaction, we assessed the therapeutic potential of XPO1 inhibition using eltanexor. In vitro, eltanexor (25 nM) displaced UBTF-TD from chromatin within 24 hours and impaired proliferation, induced apoptosis, and promoted myeloid differentiation in CD34⁺ models. In a UBTF-TD patient-derived xenograft (PDX) model, six weeks of eltanexor treatment (5 or 10 mg/kg, oral) significantly reduced leukemic burden, spleen weight, white blood cell counts, and hCD45⁺ engraftment in peripheral blood, spleen, and bone marrow. Flow cytometry showed a shift from CD117⁺/CD34⁺ stem-like cells to CD11b⁺/CD33⁺ myeloid cells. Histology revealed decreased leukemic infiltration and increased myeloid maturation in the spleen and bone marrow. Single-cell RNA-seq of isolated human cells from the bone marrow demonstrated loss of stem-like transcripts (CD34 and KIT) and increased expression of myeloid markers (MPO, CTSG, PRTN3). CUT&RUN of treated PDX samples confirmed eviction of UBTF-TD from HOXA/B loci, with minimal effect on XPO1 binding, collectively suggesting that XPO1 inhibition is a candidate therapy for AML with UBTF-TD alterations. In summary, UBTF-TDs introduce de novo NES motifs that enable a gain-of-function interaction with XPO1 and promote aberrant chromatin binding to loci dysregulated in UBTF-TD AML. This interaction sustains a leukemic self-renewal program and proliferation. Disrupting this axis, genetically (NES-M) or pharmacologically (eltanexor), impairs chromatin targeting, induces differentiation, and suppresses leukemogenesis. These findings nominate XPO1 as a dependency in UBTF-TD AML and support clinical evaluation of XPO1 inhibitors in this subtype.