Figure 5. JAK2^V617F E-SLAM HSCs, but not progenitors, are tilted toward differentiation.

(A) The average clone size data for WT and JAK2^V617F E-SLAM HSCs are approximately exponential over the 10-d time course. At early times, the data for both cell types show that the expansion is geometrical, with individual clones expanding from one to two to four to eight cells. After several rounds of division, the average cell division rate appears to accelerate significantly, while nearer to 10 d, there is a deceleration most likely due to cells exiting cycle. The dashed line shows what exponential growth would look like with the average doubling rate of the first 4–5 rounds of division (1.06 for WT and 1.26 for JAK2^V617F), and the solid line represents the model fit to the actual data points for WT (left) and JAK2^V617F (right) clones (for details, see Model S1). Note that, in both cases, division rates must increase to accommodate the expansion measured at day 10. (B) A schematic of the model dynamics. In the WT situation, cells move through a differentiation hierarchy with HSCs at the apex. In the model, the division of an HSC leads to symmetric duplication or differentiation with equal probability (i.e., x = 50%). Cells at the first generation of the differentiating hierarchy then have a capacity to duplicate or symmetrically differentiate into cells in the next tier of the hierarchy, and the model can be tuned to allow x to vary throughout (see Model S1 for further details). (C) The cumulative size distribution of clones 10 d postplating by cell type in WT clones—i.e., the KSL data point at (4k, 40%) in the WT graph—shows that 40% of the colonies have at least 4,000 KSL cells, etc. (D) Comparison of the balanced self-renewal model (i.e., with x = 50% within the entire stem and progenitor cell compartments) with parameters inferred from a fit to the colony growth curve (A) and cell type averages at 10 d postplating, against the experimental data (points) taken from (C) and (E). The vertical lines (color coded by cell type) represent the expected range of fluctuations of the cumulative size distribution due to small number statistics, and are inferred from the average and first standard deviation of the results of the model simulation with 1,000 trials each with a cohort of 68 (WT) and 125 (JAK2^V617F) colonies, consistent with that used in experiment (for further details and model parameters, see Model S1). (E) The same data set as in 5C and 5D, but with model predictions when just 40% of the progenitor cell progeny (x = 40% for the non-HSCs tiers) remain at the same tier of the hierarchy. Note the departure of the line for the KSL population, which reflects the premature escape of cells from the top of the hierarchy. (F) The cumulative size distribution of colonies 10 d postplating by cell type in JAK2^V617F clones. (G) The balanced self-renewal model (i.e., with x = 50% within the entire non-HSC progenitor compartment) overlaid onto the data from JAK2^V617F clones with solid lines displaying the predictions of the model. The departure of the model from the observed data is visible in the total viable cells where the model predicts more viable cells in order to produce the observed number of KSL and Lin-/non-KSL cells. (H) Here the lines shown represent a model where HSC self-renewal has been set to 0 implying that every division of an HSC will result in differentiation to the next tier, but progenitor self-renewal remains intact within the rest of the non-HSC progenitor cell compartments. Note the strong overlap of the model with the data points from the JAK2^V617F clones. In panels C–H, the total size colony is represented by black, Lin-/non-KSL cells are beige, and KSL cells are blue.