Functional annotation of MPN clones reveals mutation-specific fitness and fate bias

Abstract Introduction Improving outcomes for patients with myeloproliferative neoplasms (MPNs) requires therapies that prolong event-free survival (EFS). MPNs' pathology is driven by the emergence, expansion, and persistence of hematopoietic clones carrying driver and co-occurring mutations that confer a competitive fitness advantage and alter lineage bias. The mutated stem and progenitor cells (HSPCs) sustain abnormal blood cell counts and drive thrombosis, disease progression, and leukemic transformation. Therefore, reducing clonal fitness—particularly in lineages responsible for pathology—is a critical therapeutic path towards prolonging EFS. While JAK2V617F is known to promote stem cell expansion and myeloid bias, its lineage-specific fitness effects due to allele burden, co-occurring mutations (present in 50% of cases), and variants of unknown significance (VUS) remain poorly defined. To address this, we developed a single-cell pipeline integrating immunophenotype and genotype to quantify mutation-specific clonal fitness across hematopoiesis using primary patient samples.MethodsWe adapted the MissionBio Tapestri platform to develop an automated, single-cell genotyping and immunophenotyping workflow optimized for primary MPN blood samples. The original MissionBio system uses 312 amplicons to cover 45 genes (Myeloid DNA panel) and 45 oligo-conjugated antibodies (Biolegend TotalSeq-D). We designed a custom DNA panel of 114 amplicons that captures 100% of mutations associated with MPNs observed in 9,700 samples from 1,243 patients at Weill Cornell Medicine. We designed an antibody panel with 29 antibodies, which identified all major immunophenotypic populations. Peripheral blood was re-proportioned using CD34 enrichment to ensure adequate representation of HSPCs and mature leukocytes in a single-tube cell suspension. Cells were labeled with barcoded antibodies and genotyped using the Tapestri instrument for single-cell emulsification. Sequenced DNA and protein libraries were analyzed using custom Python-based tools. The Mosaic library enabled single-cell variant calling and clone assignment. We trained a machine learning model on 80,000 cells to assign cell identity based on antibody expression. Genotype-immunophenotype integration enabled clone identification and lineage tracking. Fitness across each hematopoietic lineage was defined as relative enrichment compared to wild-type counterparts for each clone.ResultsFitness of JAK2V617F clones depended on zygosity, co-occurrent mutations, and differentiation trajectory. Both monoallelic and biallelic JAK2V617F clones favored myeloid over lymphoid differentiation but differed in the magnitude of fitness and lineage preference. Myeloid fate preference varied between samples with JAK2V617F clones from some patients exhibiting stronger granulocytic vs erythroid differentiation and vice versa, suggesting that additional factors shape hematopoietic fitness. For instance, in a sample with JAK2V617F and a co-occurring DNMT3A R882H mutation, we identified six main clones. When comparing the JAK2V617F monoallelic clones, fitness toward the neutrophil lineage was more pronounced in the double-mutated clones (log2FC 1.94) and even higher in the double-mutated clone with DNMT3A loss of heterozygosity (log2FC 2.64). On the other hand, erythroid differentiation was primarily increased only in the JAK2V617F biallelic clone (log2FC 1.91). Monoallelic JAK2V617F was strongly depleted and biallelic clones were virtually absent in lymphoid populations, consistent with findings from orthogonal approaches. (Choi et al, Leukemia 2024) The pipeline also enabled in vivo assessment of several VUS, some of which showed measurable lineage-specific fitness effects, suggesting pathogenicity. Validation using paired flow cytometry and digital PCR analysis as previously reported (Abu-Zeinah et al, Blood Advances 2022) showed very high concordance, confirming the accuracy of the approach.ConclusionThis patient-centered, single-cell pipeline enables in vivo quantification of mutation-specific clonal fitness and differentiation bias in human hematopoiesis. This framework offers a scalable, biologically meaningful approach to dissect clonal architecture, functionally assess VUS, and guide therapy based on clone-level fitness. Longitudinal analysis of readily available blood samples promises to inform how specific treatments reshape MPN clonal dynamics and influence MPN event risk.