Novel KAT6B::NUTM2B fusion oncogene in congenital acute myeloid leukemia and implications for menin inhibitor therapy

Congenital acute myeloid leukemia (AML) is rare and may initially present as myeloid sarcoma. The most common non-Down syndrome-related congenital leukemia is AML, with over half involving KMT2A or KAT6A fusions.^1,2 KMT2A-rearranged infant AML has a poor prognosis, while some cases with KAT6A fusions may spontaneously remit.² Studies of molecular pathogenesis have revealed a functional connection between KMT2A and KAT6A in transcriptional activation. Both KMT2A and KAT6A fusion proteins drive aberrant transcription and self-renewal in hematopoietic progenitors, notably activating HOXA genes.^3,4 Inhibitors targeting the KMT2A/KAT6A transcriptional complex, especially menin inhibitors, have shown therapeutic promise.^5,6 Here, we report the identification of a novel KAT6B::NUTM2B fusion, involving the KAT6A paralog KAT6B, in congenital AML and demonstrate its oncogenic activity and sensitivity to menin inhibition through in vitro studies in mouse hematopoietic stem and progenitor cells (mHSPC). Informed consent for clinical genomics testing was obtained from the patient’s legal guardians, and this study was approved by the Institutional Review Board at St. Jude Children’s Research Hospital (SJCRH). The findings expand the knowledge of KAT6A/B-rearranged AML and suggest menin inhibitors as a potential therapy.

The patient presented with a blueberry muffin rash at birth. A biopsy of the rash revealed myeloid sarcoma. A bone marrow sample collected at day 9 after birth showed 37% blasts, positive for MPO and NSE (Figure 1A), and flow cytometry indicated AML with monocytic differentiation (positive for CD11c, CD64, CD33, HLA-DR, CD123, and MPO; CD14 loss in most blasts; negative for CD34, CD117, CD133, NG2, CD19, and CD3) (Online Supplementary Figure S1A). No clonal abnormalities or KMT2A rearrangement were detected by karyotyping or fluorescence in situ hybridization. Bone marrow aspirate and biopsy samples of skin lesions were collected for molecular diagnosis in a CLIA-certified laboratory at SJCRH. The fresh bone marrow aspirate was subjected to DNA and RNA extraction followed by clinical genomics testing as previously described,⁷ including whole-genome sequencing and whole-transcriptome sequencing. Formalin-fixed paraffin-embedded tissue of a skin lesion biopsy was subjected to RNA extraction and whole-transcriptome sequencing only. Integrated whole-genome sequencing and whole-transcriptome sequencing analyses identified a cryptic deletion at chromosome region 10q22 resulting in an in-frame KAT6B::NUTM2B fusion, preserving the exons encoding the histone acetyltransferase domain of KAT6B and most of the coding sequence of NUTM2B (Figure 1B). This fusion was the only tumor-acquired likely pathogenic alteration identified and was also detected in the skin rash by whole-transcriptome sequencing. Transcriptomic profiling clustered the case with KAT6A- and KMT2A-rearranged AML (Figure 1C) and high expression of HOXA9 and HOXA10 was observed (Figure 1D). An initial watch-and-wait management was chosen due to possible spontaneous remission.¹ At 4 months of age, disease progressed (44% blasts in bone marrow, hemophagocytosis), prompting standard pediatric induction chemotherapy (Online Supplementary Figure S1B) and subsequent haploidentical stem cell transplant. Full donor engraftment was achieved and the bone marrow was negative for measurable residual disease by flow cytometry, a status sustained for several months after the transplant. Relapse occurred at 1 year (77.6% blasts, testicular myeloid sarcoma). Unfortunately, the patient died at 16 months despite further chemotherapy.

To our knowledge, this patient represents the first reported case in an infant and the fifth case of a myeloid malignancy harboring a KAT6B gene fusion, as well as the first reported instance of a KAT6B::NUTM2B fusion. The four previously reported cases all carried a KAT6B::CREBBP fusion, which is considered a variant of KAT6A::CREBBP (Online Supplementary Table S1). KAT6A::CREBBP is a rare but recurrent AML-defining fusion,⁸ with other KAT6A fusion partners (NCOA2, NCOA3, EP300, LEUTX, ASXL2) also reported in myeloid malignancies (Online Supplementary Table S1). Review of infant AML with KAT6A fusions shows that over 80% of patients presented with skin lesions (myeloid sarcoma). Bone marrow involvement typically showed morphological features consistent with AML French-American-British subtype M4 or M5, with erythrophagocytosis observed in up to 50% of cases. Hepatosplenomegaly was frequently present, and up to 30% of patients showed disseminated intravascular coagulation. Spontaneous remission occurred in up to 45% of congenital AML patients with KAT6A rearrangements. Overall, the current case with a novel KAT6B::NUTM2B fusion shares clinical features similar to those observed in infant AML with KAT6A fusions; however, despite remaining clinically stable with improvement of skin lesions for approximately 4 months without treatment, the patient ultimately had a poor clinical outcome.

To gain deeper insights into the novel KAT6B::NUTM2B fusion, the full-length coding sequence of the KAT6B::NUTM2B fusion as well as controls (wild-type KAT6B and NUTM2B) were cloned into MSCV-IRES-GFP (MPG) lentiviral expression vector and transduced into C57BL/6J murine bone marrow mHSPC to assess the transforming properties of K AT-6B::NUTM2B. Initially, Hoxa9 and Hoxa10 mRNA were assessed in mHSPC expressing either wild-type KAT6B (KAT6B-mHSPC) or KAT6B::NUTM2B (KN-mHSPC). Consistent with elevated HOXA9/10 expression observed in the patient’s leukemia sample, there was a significant upregulation of Hoxa9 and Hoxa10 in KN-mHSPC (Online Supplementary Figure S2B). The mHSPC transduced with KAT6B::NUTM2B or controls were plated in semi-solid methylcellulose-based media for colony-forming unit assay. Colonies were counted, harvested, and serially replated every 7 days (Figure 2A-I). The KN-mHSPC exhibited a significant and sustained increase in colony formation, with successful serial replating for 4 weeks or longer, in contrast to wild-type KAT6B or NUTM2B, or empty vector controls (Figure 2B). Considering the HOXA9/10 activation in the patient’s leukemia, KN-mHSPC were treated with SNDX-5613, a clinical-grade menin inhibitor, at three different concentrations over a 3-week period (Figure 2A-II and III). We evaluated the sensitivity of the KAT6B::NUTM2B fusion to SNDX-5613 in comparison with two common fusions in pediatric AML: RUNX1::RUNX1T1 (9a isoform⁹) serving as a negative control as it is not considered a primary target of menin inhibitors, and NUP98::KDM5A, which is associated with HOXA activation and sensitivity to menin inhibitors,¹⁰ serving as a positive control. An empty vector control was also included. By the end of week 3, colony formation of KN-mHSPC was reduced by approximately 90-95% with SNDX-5613 treatment at all three concentrations, NUP98::KDM5A-transduced cells also exhibited a dose-dependent reduction in colony formation with SNDX-5613, albeit less prominently than KN-mHSPC. In contrast, no significant change was seen in controls (empty vector or RUNX1::RUNX1T1-transduced cells) (Figure 2C). Besides the reduction of clonogenic growth, SNDX-5613 treatment resulted in a significant increase in Gr1 and CD11b positivity and a concomitant decrease in CD117 (cKIT) expression in both KAT6B::NUTM2B and NUP98::KDM5A transduced mHSPC, indicating myeloid maturation and loss of stemness (Figure 2D). Additionally, menin inhibition resulted in significantly reduced Hoxa9/10 mRNA expression (Figure 2E) and morphological changes in KN-mHSPC (Figure 2F).

Figure 1.Diagnostic workup of the patient’s bone marrow sample. (A) Morphology and immunohistochemical staining of diagnostic bone marrow. Upper panels: Wright-Giemsa staining captured at 20x and 50x magnification highlights blasts. Lower panels: immunohistochemical staining for myeloperoxidase and neuron-specific enolase shows reactivity in blasts. (B) Identification of the novel KAT6B::NUTM2B fusion in the diagnostic bone marrow sample. Upper part: schema of KAT6B (NM_012330) and NUTM2B (NM_001278495) gene structures as well as the cryptic deletion of approximately 4.8 Mb in relation to the formation of the K AT6B::NUTM2B fusion, followed by a partial sequence of the KAT6B::NUTM2B fusion transcript along with the predicted amino acid sequence. The transcript is an in-frame fusion of KAT6B exon 17 to NUTM2B exon 2 (mid-exon). Lower part: schematic diagrams of KAT6B, NUTM2B and the predicted KAT6B::NUTM2B chimeric proteins with their functional domains: N-terminal extension of MYST; plant homeodomain; histone acetyltransferase domain; acidic domain (transcriptional activation and interaction with co-activators) and a serine-methionine-rich region. (C) Global gene expression profiling. Global gene expression profiling was conducted using uniform manifold approximation and projection (UMAP) analysis as previously described (doi:10.1038/s41588-023-01640-3). The KAT6B::NUTM2B-positive case clustered with the KAT6A-rearranged and KMT2A-rearranged subtypes by UMAP. The UMAP plot was generated in the context of a previously published cohort of pediatric acute myeloid leukemia (AML) (N=887, doi:10.1038/s41588-023-01640-3) and cord blood CD34⁺ cells (N=5). (D) Normalized expression of HOXA9 and HOXA10 genes. Expression levels are shown for the KAT6B::NUTM2B-positive patient along with reference cases from selected AML subtypes, including molecular categories known to be associated with HOXA activation (KAT6A-rearranged, KMT2A-rearranged, NUP98-rear-ranged and NPM1-mutated) and those that do not have HOXA activation (RUNX1::RUNX1T1 and CBFB::MYH11). The reference cases (N=374) were available from St. Jude Cloud (https://www.stjude.cloud) and were included in a previously published pediatric AML study cohort (doi:10.1038/s41588-023-01640-3). Expression values are shown as variance-stabilized counts derived from HTSeq counts as previously described.⁷ AA: amino acids; MPO: myeloperoxidase; MYST: histone acetyltransferase domain; NEMM: N-terminal extension of MYST; NSE: neuron-specific enolase; PHD: plant homeodomain; SM: serine-methionine-rich region; VST: variance-stabilized counts.

Both KAT6A and KAT6B encode histone acetyltransferases (HAT) that are core components of the MOZ/MORF complex¹¹ and closely interact with the KMT2A (a.k.a. MLL) complex, facilitating histone acetylation and cooperatively promoting transcription of genes regulating hematopoiesis.³ While wild-type KAT6A and KAT6A fusions have been functionally characterized, the role of KAT6B in hematopoiesis has only recently been recognized.¹² Fusions involving KAT6A or KAT6B retain 5’ exons encoding the MYST (histone acetyltransferase) domain (Figure 3), thereby preserving N-terminal functional domains of KAT6A/B in the resulting chimeric proteins. Notably, nearly all previously reported fusion partners of KAT6A/B are transcriptional co-activators (such as CREBBP, EP300, NCOA2, NCOA3) or are predicted to possess a transcriptional activation domain (e.g., LEUTX), with these activation domains consistently retained as C-terminal regions in the fusion proteins. Consequently, KAT6A/B fusions typically contain one or more HAT domains capable of modifying chromatin and activating genes such as the HOXA cluster. While NUTM2B has been reported in YWHAE::NUTM2B fusion in sarcomas, the function of wild-type NUTM2B is unclear. Interestingly, a recent study reported that YWHAE::NUTM2B can interact with EP300 and CREBBP.¹³ Although the specific functional domain mediating the interaction with EP300/CREBBP was not elucidated in the study of YWHAE::NUTM2B, it raises the possibility that KAT6B::NUTM2B may similarly recruit these co-activators to promote the transcription of HOXA genes, akin to KAT6A/B::NCOA2 fusions.

Figure 2.In vitro studies with KAT6B::NUTM2B-transduced murine bone marrow hematopoietic stem and progenitor cells. (A) Schematic depiction of workflow. GFP⁺-transduced cells were sorted and plated in methylcellulose for 1 week (W1). After this initial incubation, cells were divided as follows: (i) to continue colony-forming unit (CFU) assays without (I) and with (II) SNDX-5613 treatment, and (ii) to assess the effects of SNDX-5613 treatment in liquid culture (III). (B) CFU assay. Transduced murine bone marrow hematopoietic stem and progenitor cells (mHSPC) were plated in triplicate in methylcellulose (2×10⁴ cells/35 mm dish). Colonies were counted and serially replated every 7 days. Results from one representative CFU assay are presented, with similar findings observed across multiple independent repeats. (C) Impact of SNDX-5613 treatment on colony formation in the CFU assay. Results are shown from a representative experiment (1 of 4 independent replicates) conducted as outlined in Figure 2A-II. Transduced mHSPC were plated in methylcellulose and treated with SNDX-5613 at 0.3 μM, 0.5 μM, or 1.0 μM, or with 0.1% dimeth-ylsulfoxide (DMSO) as a control. Treatment with the indicated concentrations of SNDX-5613 started after the initial 1-week incubation in semi-solid culture. The process continued for 3 weeks, with cells replated and treated in fresh methylcellulose every 7 days. Colonies were counted prior to each replating. The x-axis shows time points after 1, 2, and 3 weeks of treatment (T1, T2, T3). The y-axis shows the relative change in CFU counts compared to the DMSO control (% of DMSO control). Error bars represent the standard deviation. Statistical significance was assessed using a two-sided Welch t test. *P<0.05; **P<0.01; ***P<0.001. (D) Flow cytometry analysis of myeloid differentiation markers. The mHSPC transduced with KAT6B::NUTM2B and NUP98::KDM5A were treated with SNDX-5613 at three concentrations or DMSO (control) in liquid culture (RPMI with cytokines) as outlined in Figure 2A-III. Results from a representative experiment are depicted. Bar plots display the log₂ fold change in mean fluorescence intensity for CD11b, Gr-1 and CD117, compared to their expression in transduced mHSPC treated with DMSO control (SNDX-5613 vs. DMSO). Statistical significance was assessed using a one-sided Welch t test based on the expected direction of change for each marker, *P<0.05; **P<0.01; ***P<0.001. (E) SNDX-5613 treatment reduced Hoxa9 and Hoxa10 expression at the transcriptional level. The mRNA expression of Hoxa9 and Hoxa10 in KAT6B::NUTM2B-transduced mHSPC was significantly downregulated by SNDX-5613 treatment compared with DMSO treatment (control). Cells in liquid culture were harvested after 1 and 2 weeks of treatment (T1 and T2, see Figure 2A-III); results are shown for T2. Bars represent log₂ fold change (SNDX-5613 treated vs. DMSO) for Hoxa9 and Hoxa10. Statistical significance was assessed using a one-sided Welch t test (expected decrease). ****P<0.0001. (F) Wright-Giemsa stained cytospins of KAT6B::NUTM2B-transduced cells obtained from the CFU assay harvested after 1 week of SNDX-5613 treatment (T1) as illustrated in Figure 2A-II. Images are shown at 40X magnification. EV: empty vector; WT: wild-type.

Figure 3.Schematic representation of KAT6A or KAT6B fusion proteins. Upper part: KAT6A/B with known fusion junctions associated with different fusion partners. In all chimeric proteins, the N-terminal region of KAT6A/B is retained, including the MYST domain which mediates histone acetyltransferase activity. Lower part: C-terminal fusion partners to KAT6A and/or KAT6B reported in literature. Fusion junctions are indicated by red stars. Black boxes highlight retained transcriptional activation domains in all fusion partners. aa: amino acids; AD: activation domain; ASXH: Asx homology domain: E: exon; KIX: CREB/kinase inducible domain interacting domain; MYST: histone acetyltransferase domain; NCBD: nuclear coactivator binding domain; NEMM: N-terminal extension of MYST; PHD: plant homeodomain; SM: serine-methionine-rich region; TAD: transactivation domain; TAZ1/2: transcriptional adaptor zinc-binding domain 1/2.

Previous studies have shown that the KAT6A::NCOA2 fusion protein targets chromatin in a wild-type KMT2A-dependent manner.³ Menin plays a central role by tethering the KMT2A complex to DNA⁶ and recruiting KAT6A/B-fusion proteins.³ Menin inhibitors have demonstrated therapeutic efficacy in KMT2A-rearranged AML,¹⁴ and emerging evidence from other HOXA-activating AML, including KAT6A-rearranged³ and NUP98-rearranged AML,¹⁰ suggests a similar vulnerability. Consistent with this model, treating KN-mHSPC with the menin inhibitor SNDX-5613 impaired colony formation, promoted myeloid differentiation, and reduced Hoxa9/10 expression, supporting menin dependency in KAT6B::NUTM2B leukemogenesis. Our findings align with a previous study on KAT6A::NCOA2³ and further indicate that menin inhibition may hold therapeutic potential against KAT6A/B-rearranged AML. In this study, we report a novel KAT6B::NUTM2B fusion in a patient with congenital AML whose clinical and transcriptomic features resembled those of patients with KAT6A::CREBBP-positive infant AML, indicating a shared mechanism. The KAT6B::NUTM2B fusion drove oncogenic transformation in vitro, and KN-mHSPC were vulnerable to a menin inhibitor, supporting the therapeutic potential of menin inhibition in KAT6B-rearranged AML. Further studies utilizing primary leukemia samples from AML patients harboring KAT6A/B-fusions are required to validate these findings. Given the increasing recognition of resistance to menin inhibitors,¹⁵ alternative therapeutic strategies, such as KAT6 inhibitors or combination therapies, should be investigated. Overall, our findings provide critical new insights into the biology of KAT6B fusions and indicate a therapeutic strategy for this rare leukemia.

Footnotes

• Received December 10, 2025
• Accepted April 10, 2026

Correspondence

L. WANG - Lu.Wang2@stjude.org

Y-C. LIU - Yen-Chun.Liu@stjude.org

Disclosures

No conflicts of interest to disclose.

Contributions

in vitro

in vitro

Funding

This work was funded by the ALSAC of SJCRH. The SJCRH Flow Cytometry and Cell Sorting Shared Resource (FCCSSR) is supported in part by ALSAC and National Cancer Institute grant P30 CA021765.

Acknowledgments

We thank Paula Perez-Sanchez at SJCRH Center of Excellence for Leukemia Studies Flow Cytometry Core; and Maria Cardenas in the Department of Computation Biology for providing HOXA9/10 expression data of leukemia samples from St. Jude Cloud. We sincerely thank Dr. Yiping Fan from the Center for Applied Bioinformatics at SJRCH for his invaluable assistance with analysis of the RNA-sequencing data. We thank the staff of SJCRH Flow Cytometry and Cell Sorting Shared Resource (FCCSSR).