Yotaro Ochi

Acute myeloid leukaemia (AML) is an aggressive blood cancer characterized by the unregulated proliferation of immature myeloblasts. Gene mutations have been shown to have a large effect on pathogenesis, inter-tumour heterogeneity and clinical outcomes in AML^{1,2,3,4,5,6,7,8}; however, the role of epigenetic alterations in these respects has been investigated less extensively. Here we use ATAC-seq (assay for transposase-accessible chromatin with sequencing) in a cohort of 1,563 individuals with a recent diagnosis of AML (the ‘eCHROMA’ cohort) to show that AML can be classified into 16 subgroups on the basis of chromatin accessibility profiles. Multiomics analyses of gene mutations, the transcriptome, DNA methylation and histone marks show that these ATAC subgroups exhibit distinct driver mutations, differentiation states, gene expression, DNA methylation and super-enhancer profiles, and are also associated with clinical outcomes. These findings were validated in independent cohorts. Single-cell ATAC sequencing reveals that all leukaemic cells in each subgroup share a common chromatin accessibility profile, which suggests that subgroup-specific epigenomic fingerprints underlie the ATAC-based classification. Mechanistically, the subgroups have distinct gene-regulatory networks that are driven by the activities of key transcription factors in haematopoiesis, and in which subgroup-specific super-enhancers have a pivotal role. Multiomics single-cell analysis further reveals deregulated trajectories of differentiation coupled with chromatin accessibility and gene expression. Notably, ATAC subgroups have an independent prognostic effect, compared with genomic classification, and are associated with particular drug sensitivities. In summary, ATAC-based chromatin profiling, combined with multiomics data, provides insights into AML pathogenesis beyond genomics and constitutes a valuable resource for AML research.

Cancer genome analysis relies on standard human reference genomes, but detecting somatic alterations in highly repetitive or individual-specific regions remains challenging. We developed the Personalized Reference genome-based Cancer Genome Analysis Pipeline (PRCGAP, https://github.com/yos-sk/PRCGAP), to our knowledge, the first comprehensive pipeline integrating haplotype-resolved analyses of somatic point mutations, structural variants, copy number, and DNA methylation on personalized diploid reference genomes, with each variant phased and annotated by genomic features. We applied PRCGAP to eight tumor-normal cell line pairs and three newly collected pediatric B-cell acute lymphoblastic leukemia (B-ALL) clinical samples. PRCGAP recovered most standard reference-based variants (86.7–96.4% across variant types), validated by orthogonal short-read sequencing, while additionally detecting variants that standard references missed (23.4% of SNVs and 32.6% of SVs). Notably, PRCGAP uncovered variants in centromeric and telomeric regions, and somatic single-nucleotide variants were enriched in centromere dip regions, the putative kinetochore formation sites. Using PRCGAP outputs, we identified L1 retrotransposition source sites absent from standard references and showed that a B-ALL IGH:: DUX4 fusion arose from a DUX4 pseudogene within an internal D4Z4 repeat unit, rather than the canonical full-length DUX4 gene. Within a single haplotype-resolved framework, PRCGAP opens regions of the cancer genome that are inaccessible to standard references.

Acute myeloid leukemia (AML) is an aggressive hematologic cancer characterized by proliferation of immature myeloblasts. It shows profound molecular heterogeneity, which has been primarily studied through genetic abnormalities, providing the basis for disease classification, prognostication, and therapeutic choice. However, genetic factors alone may not fully explain AML pathogenesis and diversity, while leaving the role of abnormal epigenome, particularly chromatin state, largely unexplored in a large cohort of patients. Here we show that AML is classified into 16 subgroups with distinct chromatin accessibility profiles based on ATAC-seq in 1,563 AML cases, including novel AML subgroups not previously recognized in conventional genomic classifications. By integrating multi-omics analyses of genome, transcriptome, and major histone marks, we show that these epigenetic subgroups exhibit unique features in clinical presentation, gene mutations, differentiation states, gene expression, and super-enhancer profiles, which are validated across independent cohorts. Single-cell sequencing demonstrates the presence of subgroup-specific ATAC signatures that are shared by all leukemic cells, confirming the definitive role of the epigenome in the ATAC-based classification. Mechanistically, each subgroup is associated with a distinct gene regulatory network centered on key transcription factors, where subgroup-specific super-enhancers play a pivotal role. These ATAC subgroups also have prognostic significance independent of genomic classification, and help reveal unexpected drug sensitivities. In summary, ATAC-based chromatin profiling in this large sample set, combined with multi-omics data, provides new insights into AML pathogenesis beyond genomic profiling and also serves as an invaluable resource for AML research.

TP53 mutations represent one of the strongest adverse prognostic factors in myelodysplastic neoplasms (MDS). While multi-hit TP53 ( TP53 ^multiHit ) alterations uniformly lead to very poor outcomes, the prognostic relevance of monoallelic TP53 ( TP53 ^mono ) mutations remains controversial. TP53 variants can cause loss-of-function, dominant-negative, or gain-of-function effects. We hypothesized that functional heterogeneity among TP53 variants contributes to the variable clinical behavior observed in monoallelic TP53 -mutated MDS. Therefore, we analyzed pretreatment samples from 4,505 patients with MDS from two independent cohorts (IWG, n=3,173; J-MDS, n=1,332), including 271 patients with TP53 ^mono and 499 with TP53 ^multiHit . Functional annotation of TP53 variants was performed using a previously published phenotype score (PS) derived from saturation mutagenesis screens, capturing dominant-negative and loss-of-function effects. Median overall survival (OS) differed significantly by TP53 allelic state ( TP53 wild-type ( TP53 ^wt ) 42.4 months; TP53 ^mono 22.9 months; TP53 ^multiHit 9.2 months; p < 0.001). Within the TP53 ^mono subgroup, functional annotation identified marked heterogeneity. Patients with high PS (≥7) showed significantly inferior OS compared with those with low PS (median OS: 13.8 vs. 39.2 months; HR 1.68, 95% CI 1.16-2.42; p = 0.006), particularly for IPSS-R and IPSS-M low-risk cases. Combining PS and variant allele frequency (VAF) further improved risk stratification. TP53 ^mono patients with PS ≥7 and VAF ≥22% had outcomes comparable to TP53 ^multiHit (median OS: 8.8, p = 0.2), whereas those with PS TP53 ^wt (median OS: 49.7, p = 0.9). Overall, functional annotation of TP53 variants refines prognostication in TP53 ^mono -mutated MDS and may enhance individualized risk assessment.

Centromeres are among the most diverse and dynamically evolving regions of the human genome and are commonly affected in various human cancers. However, organized into highly repetitive α-satellite higher-order repeats (HORs), human centromere sequences have long resisted detailed genomic analysis. Although the development of long-read sequencing platforms has enabled the analysis of complete centromere sequences, their application to a large set of samples is still largely limited, preventing our understanding of centromere variation and haplotype structures across large human populations and the structural basis of centromere-involving translocations in cancer. Here we show that rare k-mers present in centromeric regions can serve as effective markers for dissecting the complexity of centromere structure, particularly that of active α-satellite HOR arrays (aHOR arrays), across human populations and for understanding centromere-involving abnormalities in cancer. Based on rare k-mer-based clustering, centromere aHOR arrays are clustered into discrete haplogroups (aHOR-HGs) with distinct structural features. These k-mers were also used to develop a framework that enables the inference of haplogroups in a given sample based on short-read whole genome sequencing (WGS) data (ascairn). By applying ascairn to large-scale human population datasets ( n > 3,300), we revealed the diversity of aHOR-HGs and their geographic histories across populations. The rare k-mer-based approach was also applied to investigate the structure of 1p/19q co-deletion, a highly recurrent centromere-involving translocation in IDH -mutated oligodendrogliomas. Analyzing short-read WGS data from 142 cases with 1p/19q co-deletion using rare k-mers, we showed that breakpoints of 1p/19q co-deletion were mapped to aHOR arrays in chromosomes 1 ( D1Z7 ) and 19 ( D19Z3 ), which was validated by long-read sequencing of two 1p/19q co-deletion-positive cases. Notably, the translocation preferentially involved haplogroups composed of haplotypes containing larger regions susceptible to rearrangement. These results highlight the role of rare k-mers in dissecting the complexity of centromere sequences and their evolutionary history as well as understanding centromere-involving abnormalities associated with human diseases.

Venetoclax plus azacitidine (VEN + AZA) is widely used in acute myeloid leukemia (AML). This study explored the role of static and dynamic profiles of mutational clonal burden to predict outcomes by analyzing marrow samples from 228 VEN + AZA treated AML cases at “Pre-treatment” (n = 228), “Best-response” (n = 105), and “Relapse” (n = 27) phases using targeted-capture sequencing. In a multivariate model, older age, prior AZA, TP53 mutation with variant allele frequency ≥0.10, and RAS-pathway mutations predicted shorter overall survival (OS), while BCORL1 mutation predicted longer OS. Notably, myelodysplasia-related gene mutations, which constitute adverse factors in ELN 2022, predicted favorable survival. Achieving composite complete remission (CRc) significantly predicted longer OS (P < 0.001) but showed residual mutations in 76.2% of the cases. Among CRc cases, relapse-free survival was stratified by molecular clearance of mutations other than DNMT3A, ASXL1, and TET2 (P = 0.04). In addition, 37% of relapsed cases showed a change of major clones, with 40% having potential targets of molecular-targeting treatment. This study revealed the novel prognostic role of myelodysplasia-related gene mutations and established the importance of molecular response assessment in CRc phase.

External signals from the thymic microenvironment and the activities of lineage-specific transcription factors (TFs) instruct T-cell versus innate lymphoid cell (ILC) fates. However, mechanistic insights into how factors such as Notch1–Delta-like-4 (Dll4) signaling and E-protein TFs collaborate to establish T-cell identity remain rudimentary. Using multiple in vivo approaches and single-cell multiome analysis, we identified a feedback amplifier circuit that specifies fetal and adult T-cell fates. In early T progenitors (ETPs) in the fetal thymus, Notch signaling minimally lowered E-protein antagonistId2levels, and highId2abundance favored the differentiation of ETPs into ILCs. Conversely, in the adult thymus, Notch signaling markedly decreasedId2abundance in ETPs, substantially elevating E-protein DNA binding and in turn promoting the activation of a T-cell lineage-specific gene expression program linked with V(D)J gene recombination and T-cell receptor signaling. Our findings indicate that, in the fetal versus the adult thymus, a simple feedback amplifier circuit dictated by Notch-mediated signals andId2abundance enforces T-cell identity and suppresses ILC development.