Single Cell Genotype-Phenotype Analysis Demonstrates That Post-MPN AML Is Driven By Clonal Evolution and Loss of Transcriptional Heterogeneity within Stem Cell Compartments

Introduction

Myeloproliferative neoplasms (MPN) are the consequence of acquired mutations in hematopoietic stem and progenitor cells (HSPCs). Specific mutations and mutational burden determine a patient's risk of disease progression to lethal post-MPN AML.

Recent technological advancements linking single-cell gene expression and genotype have facilitated the deconvolution of tumor heterogeneity. However, these pipelines are limited in their ability to genotype low-expressed transcripts (e.g., JAK2) and multiple pathogenic loci. To overcome this, we developed a novel single-cell LOng-read genotyping of TRanscripts pipeline called LOTRSeq to explore cancer heterogeneity through target enrichment and long-read sequencing.

Methods

LOTRSeq genotypes 10X 3' single-cell RNA sequencing (scRNASeq) barcoded transcripts by a biotin-streptavidin probe-based pull down of 30 genes frequently mutated in MPN/AML, followed by Oxford Nanopore Technology (ONT) long-read sequencing. We performed scRNASeq of 50,439 CD34+ HSPCs of n=8 patients with JAK2^V617F-driven MPN (n=4 PV, n=1 ET, n=3 MF) and n=4 post-MPN AML (2 paired samples) followed by LOTRSeq. To assess transcriptional differences between mutated and wildtype HSPCs of MPN/post-MPN AML patients compared to healthy HSPCs, we integrated CD34+ HSPCs scRNASeq data of healthy subjects (n=6, Ainciburu et al. 2023).

Results

We utilized a pool of cDNA generated from 4 AML cell lines with 8 different 10X barcodes and ratios to validate the LOTRSeq pipeline for the deconvolution of similar barcodes and genotyping of lowly expressed genes. Target enrichment achieved up to 100X enrichment of transcripts over whole transcriptome sequencing. Read proportions assigned to deconvoluted barcodes highly correlated with cell mix proportions (R=0.93, p=0.0007). We achieved 95% sensitivity to call JAK2^V617F and 80% sensitivity to call TP53^M133K, with 84% and 78% accuracy, respectively. In MPN and post-MPN AML samples, expressed variant allele frequencies (VAF) of pseudo bulked long-reads highly correlated with genomic VAF (n=6, R=0.9, p=0.01). In addition, 60-80% of cells were genotyped, with genotyping % strongly correlating with Nanopore sequencing output (R=0.85, p=0.0002).

Single-cell analysis and clustering of CD34+ HSPCs agnostic to genotype reveals that cells cluster according to a deterministic progenitor phenotype. The HSPCs of JAK2^V617F-driven MPN largely maintain transcriptional heterogeneity similar to healthy but with expanded erythroid primed cluster (p=0.0047) that correlates with the percentage of JAK2^V617F cells (R=0.73, p=0.039). Consistent with previous publications (Nam et al. 2019, van Egeren et al. 2021), mutant and wildtype cells did not cluster apart. However, JAK2^V617F cells compared to wildtype and healthy, showed a higher probability of converging towards an erythroid cell fate.

Sequential analysis of a patient’s bone marrow with primary MF and post-MPN AML showed linear clonal evolution of a JAK2^V617F and IDH1^R132S mutant clone at leukemic transformation, leading to the expansion of a transcriptional HSC-like cluster. Another paired ET and post-MPN AML with a 2-year transformation period showed branching evolution where we identified at diagnosis 27% JAK2^V617F cells and 22% TP53^Y236C in independent clones and 8% TP53^R282W within the TP53^Y236C clone. Upon leukemic transformation, the JAK2^V617F clone was detected only in 0.6% of cells, outcompeted by a clone with biallelic loss of TP53 through missense mutations Y236C and R282W. This clonal population overlapped transcriptionally with Granulocyte-Megakaryocyte-Progenitor like cells.

Conclusion

We have developed a novel scRNASeq pipeline to concurrently genotype a panel of 30 MPN/AML driver genes and examine the transcriptional impact of genetic heterogeneity. We observe a progressive loss of transcriptional heterogeneity with disease progression associated with acquiring additional gene mutations and clonal evolution. Discrete gene mutations are associated with specific effects on transcriptional identity. This knowledge will be invaluable in understanding and monitoring the impact of mutation acquisition and clonal evolution during MPN disease progression and response to therapy.