Increased numbers of immature/low ploidy MKs are a hallmark feature of myelofibrosis (MF). MF MKs produce pathological amounts of pro-fibrotic factors including TGFβ causing a severe bone marrow (BM) fibrosis. Isolating MKs in MF is difficult as the fibrosis prevents BM aspiration. Also, the cellular hierarchy of normal megakaryopoiesis remains unclear, complicating any in-depth study of abnormal MK development in MF. Using a single-cell approach, we recently identified CD41 as a marker of a novel, rare population of MK-primed progenitors within peripheral blood (PB) immunophenotypic megakaryocyte-erythroid progenitors (MEP) of mobilized, healthy donor controls (Psaila et al, 2016). However, our analysis did not include MF patients and we did not analyze earlier hematopoietic stem/progenitor cells (HSPCs), which may also include MK-biased subsets that might be disrupted in MF. To examine the abnormalities in MF hematopoiesis more comprehensively and without bias from pre-selected surface markers/gene expression assays, we performed single-cell RNA-sequencing on ~50,000 individual CD34+ Lineage- PB cells of healthy mobilized donors (n=2, 18,500 cells) and 4 JAK2V617F+MF patients (n=4, 32,500 cells, 10X Genomics platform). The data were analyzed using Seurat (Sajita et al, 2015) to perform dimensionality reduction and unsupervised cell clustering and to identify differentially expressed genes. Individual cells from control and MF samples were grouped into distinct 'clusters' of cells with similar transcriptional profiles. Identities of the different clusters were determined by unsupervised gene ontology enrichment analyses and manual inspection of top differentially expressed genes for known lineage-specific identifiers.

We observed striking differences in the cellular architecture of HSPCs in the controls and patients with MF (Fig 1A). Increased expression of MK-associated genes (incl. VWF, ITGA2B /CD41) was observed throughout HSPC populations in MF (Fig 1B). Similarly, we observed prominent clusters of CD34+ progenitors expressing PF4, a more mature MK marker, and high levels of TGFβ only in MF samples and not in controls (Fig 1B). Analyses to determine the molecular basis for the bias towards MK lineage specification in MF are now underway.

As our single cell analysis showed increased expression of CD41 across a number of different HSPCs in MF patients, we compared the frequencies of CD41+, MK-primed populations in PB samples of 11 controls and 15 patients with JAK2V617F+ MF, both within more differentiated CD38+ MEP/CMP as well as primitive, CD38-neg HSPC. A selective expansion of MK-primed populations in MF was observed, with a >5-fold increase in both MK-primed MEP as well as MK-primed HSC/MPP populations, suggesting an expansion of the MK lineage and biased commitment from the apex of the HSPC hierarchy.

To study more differentiated MKs, we next carried out RNA-sequencing of immature (CD41+42-) and mature (CD41+42+) MKs cultured in vitro from CD34+ HSPC from controls and patients with JAK2V617F+MF and moderate-severe BM fibrosis. The mRNA signature of immature CD42- MKs was highly distinct for MF patients vs. controls, whereas mature CD42+ MKs were not significantly different. IPA analysis identified tissue fibrosis as the most highly dysregulated pathway. The expression of genes encoding extracellular matrix proteins e.g. elastin were also markedly increased in JAK2V617F+ MKs, suggesting that MKs may themselves produce extracellular matrix factors as well as stimulating their deposition by fibroblasts. Increased expression of hepatocyte growth factor (HGF) was also observed in MF MKs. Treatment of fibroblasts in vitro with HGF or PF4 was found to inhibit TGFβ-mediated collagen deposition (P's<0.01), suggesting a complex interplay of both stimulatory (e.g. TGFβ) and inhibitory (HGF and PF4) regulators of fibrosis in MF MKs.

In summary, we demonstrate that a single-cell approach enables a much finer dissection of HSPCs compared to "bulk" assays and allows proper comparison of normal and malignant cellular differentiation pathways, as well as the ability to uncover rare cell types. Our data constitute a rich and unique dataset, illustrating how transcriptomic analysis of HSPCs in MF can identify novel regulators of fibrosis with potential for therapeutic development.


Milojkovic: Novartis: Consultancy, Honoraria; Pfizer: Consultancy, Honoraria; BMS: Consultancy, Honoraria; ARIAD: Consultancy, Honoraria; Incyte: Honoraria, Speakers Bureau. Mead: BMS: Honoraria; Pfizer: Honoraria; Novartis: Honoraria, Research Funding, Speakers Bureau.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution