TP53突变型急性髓系白血病骨髓的高分辨率多组学解析

TP53- mutated ( TP53 m) AML is a distinct biological entity with adverse prognosis, characterized by prominent clonal heterogeneity; cells within a patient may exhibit wild type, monoallelic, or biallelic TP53 status, alongside co-occurring genomic alterations and transcriptomic heterogeneity. This complexity poses challenges in accurately aligning genotypes with transcriptomic and epigenomic profiles at single-cell resolution. To address this, we integrated single-cell RNA-seq (10x and long-read Oxford Nanopore, ONT) with open chromatin profiles (scATAC-seq) and whole-genome sequencing (WGS) to capture structural variants (SV), copy number alterations (CNV), and single nucleotide variants (SNV) from bone marrow samples taken from a cohort of 47 TP53 m, 7 TP53 wild-type (wt) AML patients, and 4 healthy donors.Cell type annotation for scRNA-seq (335K cells) utilized a bone marrow reference dataset, while scATAC-seq (130K cells) was annotated using the ArchR label-transfer method. Cell type proportions were concordant across modalities (median correlation, ρ = 0.76). SV and CNV profiles showed that TP53 m samples displayed frequent loss of Chr5q (72.5%), Chr17p (52.5%), and Chr7q (40%); chromothripsis was present in 16 (40%) samples. All these events were reported to occur in under 3% of TP53 wt AMLs. In contrast, TP53 m samples showed relative paucity of SNV co-mutations.We leveraged CNV profiles to map genetic gain/loss effect in scRNA and scATAC-seq compared to healthy donors, providing a clear separation of cells with abnormal CNV profiles. In TP53m AML, apart from mature lymphoid populations, >80% of cells including lymphoid progenitors (LMPP), displayed abnormal CNV status. Abnormal CNV cell fractions by sample strongly correlated with orthogonal tumor purity estimations from WGS for scRNA (ρ = 0.81) and ATAC-seq (ρ = 0.84).Single-cell SNV mapping was attempted with the standard 10x pipeline, where we detected driver hotspot mutations in only ~3% of non-lymphoid cells. To improve mutation calling, we developed a new SNV pipeline using long-read ONT chemistry on full-length cDNA. SNV calling rates rose to 15% in non-lymphoid cells. Mature lymphoid cells showed low alteration rates: SNVs (1-2%), CNVs (5%), while expressing lineage-specific markers, confirming cell-type assignment. Long-read TP53 transcript sequencing also enabled phased variant genotyping. Combined with WGS SV and CNV profiles, we observed that TP53 loss-of-function dosage distinctly associates with the genomic architecture: biallelic loss of TP53 was associated with extensive intrachromosomal breaks and chromothripsis, whereas monoallelic loss primarily induced numerical chromosomal changes. This multi-omic framework enables high-resolution characterization of somatic genomic alterations at the single-cell level in TP53 m AML, providing a powerful platform for mechanistic interrogation of disease biology.