AUTHORS
Imaz Khalid, Joachim Wolff, PhD2, Kunzangla Bhutia3, Claudia Bruedigam, PhD4, Florian Perner, MD5, Wencke Walter, PhD6, Torsten Haferlach, MD, PhD7, Tina M. Schnoeder8, Florian H. Heidel IV, MD9
AFFILIATIONS
1 Hannover Medical School, Hannover, Niedersachsen, Germany
2 Hannover Medical School, Hannover, Niedersachsen, Germany
3 Hannover Medical School, Hannover, Niedersachsen, Germany
4 Monash University, Melbourne, Australia
5 Hannover Medical School, Hannover, Niedersachsen, Germany
6 MLL Munich Leukemia Laboratory, Munich, Germany
7 MLL Munich Leukemia Laboratory, Munich, Germany
8 Hannover Medical School, Hannover, Niedersachsen, Germany
9 Hannover Medical School, Hannover, Niedersachsen, Germany, Leibniz Institute on Aging, Jena, Germany
ABSTRACT
Reciprocal chromosomal rearrangements define biologically distinct subclasses of AML. The rare t(8;16)(p11;p13) translocation fuses KAT6A (MOZ) with CREBBP (CBP), accounts for 0.3% of AMLs and confers a poor prognosis. Distinctively among AML-fusions, it unites 2 histone acetyl-transferases from separate enzyme families (MYST, p300/CBP) within a single chimeric protein. The breakpoint characteristically retains the MYST-type catalytic HAT domain of KAT6A together with the bromodomain and HAT domains of CREBBP.
To define the genetic landscape of KAT6A::CREBBP (hereafter KC) AML, we analyzed the largest cohort to date (n=50): a complex karyotype (CK) was present in 31/50 patients (62%) and TP53 was the most frequent co-mutation (14/50; 28%), followed by TET2, ASXL1, and DNMT3A-frequencies far exceeding those in de novo and fusion driven AML (CK 10–12%; TP53 5–10%). This enrichment of KC and TP53/DTA co-mutations aligns KC-AML closely with secondary or adverse-risk AML.
To functionally assess for self-renewal capacity of KC-AMLs, we generated retroviral models for KC and 2 comparators: the related KAT6A::TIF2 (MT2) fusion and the canonical MLL::AF9 (MA9) (all in MSCV-IRES-GFP). In CFU-assays, KC-transduced murine HSPCs acquired serial replating capacity, albeit less robustly than MT2- or MA9-transformed cells. To assess disease propagation in vivo, equal numbers (5×104) of KC, MT2, or MA9 cells were transplanted into sublethally irradiated C57BL/6 mice. Here, KC recipients developed AML with comparable penetrance and latency (median OS: KC 29, MT2 29, MA9 26 days). KC-AML was characterized by leukocytosis, splenomegaly, and co-expression of monocytic markers. Consistently, a newly established KC-driven PDX-model showed stable engraftment in NXG recipients and serial transplantability with a median OS of 39 days.
Flow cytometric analysis of murine HSPC subsets was performed to define leukemia-initiating cells (LICs) in KC-AML. KC- and MA9-AMLs were dominated by granulocyte-macrophage progenitor (GMP)-like progenitors (Lin⁻Kit⁺Sca-1⁻CD34hiCD16/32hi; 61.6% and 95.6%, respectively). To assess for functional LICs in this compartment, we sorted GMP-like cells from KC-, MT2-, and MA9-AMLs and transplanted them at limiting dilution (5×105, 5×104, 5×103) into sublethally irradiated recipients. Here, KC-AML showed an LIC-frequency comparable to MA9 (1/33,099 vs 1/47,690).
To test whether KC-AML retains dependency on KAT6A catalytic activity, KAT6A was genetically and pharmacologically inactivated. The selective KAT6A/B inhibitor PF-9363 preferentially reduced KC-AML viability (at 500nM: 8.4%±3.2%) over MT2 (25.6%±8.0%) and MA9 (46.0%±15.9%). Mechanistically, PF-9363 suppressed H3K14 and H3K23 acetylation at nanomolar concentrations (50nM) in KC and MT2 but at micromolar doses (5µM) in MA9, indicating KAT6A-dependent maintenance of these marks in KAT6A-fusion leukemias.
Transcriptomic profiling of murine KC, MT2, and MA9 AMLs was performed to compare the transcriptional consequences of distinct epigenetically driven fusion-oncogenes. Although all 3 fusions converge on acetyltransferase/chromatin dysregulation, their transcriptomes appeared distinct. KC-AML was characterized by upregulation of stemness-associated genes (Mn1, Lin28b) and telomerase components (Terc, Tert). Consistently, the telomerase inhibitor imetelstat (IM) significantly reduced viability of KC-AML (at 5µM: 22.3% ± 3.06%), nominating IM as a therapeutic strategy for this poor-prognosis subtype, which responds poorly to conventional chemotherapy and hypomethylating agents.
As TP53 was the most frequent co-mutation in human KC-AML, we tested whether its loss may compromise these dependencies, generating CRISPR-engineered Trp53-knockout (p53-null) KC models. Both agents retained significant activity in the p53-null context and revealed suppressed proliferation (PF-9363, DMSO vs 500nM: p<0.0001; IM, DMSO vs 2.5µM: p<0.0001). Their combination resulted in synergistic activity (PF-9363/comb: p<0.0001; IM/comb: p<0.0552), an approach currently tested in PDX models.
Collectively, our integrated genetic and pharmacologic analysis defines t(8;16)/KC as a poor-prognosis, secondary AML-like disease marked by CK and TP53/DTA co-mutations. Despite uniting 2 histone acetyltransferases in one chimera, it remains dependent on KAT6A catalytic activity and a distinct stemness/telomerase transcriptional program, exposing KAT6A and telomerase inhibition as a rational therapeutic strategy.
DISCLOSURES
Imaz Khalid: No relevant conflicts of interest to declare.