Disrupting a Convergent Acetylation Circuit Collapses Leukemic Identity Across AML Subtypes

Background

Acute Myeloid Leukaemia (AML) is a genetically heterogeneous blood cancer characterised by recurrent mutations in genes such as FLT3, IDH1/2, or DNMT3A. Despite advances in therapy, outcomes remain poor, particularly in older patients. Current therapeutic regimen includes conventional chemotherapy alongside targeted therapies against IDH or FLT3¹. This raises an important question: instead of targeting individual mutations, could we instead target a process that every leukaemia cell depends on?

Gene expression is regulated not only by transcription factors but also by chromatin-modifying complexes that determine DNA accessibility. SAGA (Spt-Ada-Gcn5 acetyltransferase) is one such complex. It is a histone acetyltransferase complex that deposits acetyl marks on histones and helps maintain transcription of genes essential for cell identity.

Using large-scale CRISPR-dependency datasets and cell-line-based assays, the authors of this preprint ask whether genetically diverse AML cells converge on a common epigenetic dependency despite their distinct mutations.

Key findings

SAGA is essential in haematological malignancies

The study began by analysing the Cancer Dependency Map (DepMap) comprising genome-wide CRISPR-Cas9 screens across several cancer lines. By identifying protein-complex dependencies unique to haematological malignancies, the SAGA complex was identified as a top candidate in AML. In contrast, the closely related ATAC complex did not show a similar dependency in lymphoid and myeloid lineages. Leukemic cells seem to possess a unique reliance on this transcriptional regulator, SAGA.

KAT2A/B degradation suppresses genetically diverse AML

The authors then targeted the catalytic acetyltransferase subunits of the SAGA complex, KAT2A/B, using GSK983. It produced potent anti-leukemic effects across diverse genetic modifications, including KMT2A rearrangements and NPM1 mutations. The treatment reduced cell proliferation, induced apoptosis and differentiation, impaired colony formation (Figure 1; preprint Figure 2E – I), and depleted the stem-cell-like populations. Importantly, GSK983 remained effective even in model systems carrying mutations associated with therapeutic resistance. This suggested that the dependency on the SAGA complex is common across molecular subtypes.

Figure 1 Colony-forming unit (CFU) assay for AML PDX sample (E) and human cord blood-derived CD34 cells transformed with different subtypes of AML mutations (F-I). Preprint Figure 2E-I, made available under a CC-BY-NC-ND 4.0 International License.

SAGA maintains the leukemic identity via a convergent acetylation circuit

KAT2A/B degradation selectively reduced H3K9 acetylation at promoters of key oncogenes, without globally disrupting chromatin acetylation. Beyond reducing histone acetylation, loss of KAT2A/B destabilised ENL- and SEC-associated transcriptional condensates. That led to rapid collapse of the transcriptional program in the leukemic cells. These findings place SAGA at the intersection of chromatin regulation, epigenetic modulation, transcriptional condensates and leukemic cell identity.

Conclusions and future directions

This study proposes a shift in how we view AML biology and leukemic vulnerabilities. Instead of treating each molecular subtype as a distinct entity and an independent form of disease, the authors of the preprint identify a shared epigenetic dependency across genetically diverse AML subtypes. By disrupting a shared acetylation circuit, leukemic identity itself can be destabilised (Figure 2).

Figure 2 Conceptual overview of SAGA-dependent acetylation circuit maintaining the leukemic identity across genetically diverse AML subtypes. Created by the author using PowerPoint. Icons and symbols obtained from Microsoft PowerPoint. Based on findings reported in the referenced bioRxiv preprint.

GSK983 demonstrates notable preclinical efficacy in the diverse AML cell line models. However, its clinical translation is yet to be established. Future studies will determine whether KAT2A/B degraders can translate these promising preclinical findings into durable clinical responses. Whether combining them with existing targeted therapies or immunotherapies could overcome the disease burden in select patients, particularly older patients, will have to be tested. It will be interesting to see whether such common pathways exist in other genetically heterogeneous cancers such as Acute Lymphoid Leukemia (ALL).

What I like about this preprint

In my doctoral studies, the focus has been on the genetically diverse forms of B-cell Acute Lymphoid Leukemia (B-ALL). I found it intriguing that, though similarly heterogeneous, the authors have not searched for another subtype-specific vulnerability, which has been the norm in leukaemia studies in the recent past. They have instead identified a shared epigenetic vulnerability that appears to be required to unify the diverse forms of AML.

Questions for the authors

1. Besides SAGA, several chromatin complexes showed selective dependencies in AML. What made SAGA the most compelling candidate to pursue, and do you think some of them might prove therapeutically valuable as well?
2. Do you think that KAT2A/B dependency will extend beyond AML to other malignancies such as B-ALL, which is also equally diverse?
3. As transcriptional condensates play a primary role in this mechanism, do you think they could serve as a predictive biomarker for response to KAT2A/B directed therapies?

References

1. Jeurkar C, King L, Baek D, Wilde L, Keiffer G, Kasner M. Management of Acute Myeloid Leukemia: A Review. Cancers (Basel) [Internet]. 2026 Feb 1 [cited 2026 Jul 31];18(4):659. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC12939490/