Estimating dormant and active hematopoietic stem cell kinetics through extensive modeling of bromodeoxyuridine label-retaining cell dynamics

Abstract

Bone marrow hematopoietic stem cells (HSCs) are responsible for both lifelong daily maintenance of all blood cells and for repair after cell loss. Until recently the cellular mechanisms by which HSCs accomplish these two very different tasks remained an open question. Biological evidence has now been found for the existence of two related mouse HSC populations. First, a dormant HSC (d-HSC) population which harbors the highest self-renewal potential of all blood cells but is only induced into active self-renewal in response to hematopoietic stress. And second, an active HSC (a-HSC) subset that by and large produces the progenitors and mature cells required for maintenance of day-to-day hematopoiesis. Here we present computational analyses further supporting the d-HSC concept through extensive modeling of experimental DNA label-retaining cell (LRC) data. Our conclusion that the presence of a slowly dividing subpopulation of HSCs is the most likely explanation (amongst the various possible causes including stochastic cellular variation) of the observed long term Bromodeoxyuridine (BrdU) retention, is confirmed by the deterministic and stochastic models presented here. Moreover, modeling both HSC BrdU uptake and dilution in three stages and careful treatment of the BrdU detection sensitivity permitted improved estimates of HSC turnover rates. This analysis predicts that d-HSCs cycle about once every 149-193 days and a-HSCs about once every 28-36 days. We further predict that, using LRC assays, a 75%-92.5% purification of d-HSCs can be achieved after 59-130 days of chase. Interestingly, the d-HSC proportion is now estimated to be around 30-45% of total HSCs - more than twice that of our previous estimate.

Figure 1. Conventional vs. dormant population HSC hierarchy.

The hierarchical organization of the hematopoietic system has long been recognized, with rarely-dividing multipotential HSCs producing rapidly dividing lineage-restricted transit-amplifying and committed progenitors which in turn will give rise to all differentiated cell types of the blood. Within the HSC population, two possible models can be envisaged. In the conventional model (left panel), the HSC population is homogeneous with respect to cell cycle entry with the entire HSC pool turning over every few weeks. In contrast, in the dormant HSC model (right panel), the hierarchical organization of the hematopoietic system includes the phenotypic HSC pool, in which two subpopulations can be defined based on their relative turn-over frequencies. An active HSC (a-HSC) population is responsible for the day-to-day maintenance of the hematopoietic system, while a second population, the dormant HSC pool (d-HSC), cycles only a few times over the life span of the mouse in a homeostatic situation (dashed arrow) but is activated and participates in replenishment of the hematopoietic system after injury (solid arrow).

Figure 2. Observed experimental data of BrdU uptake and chase.

Green: uptake; blue: chase; red x: mean observed data; vertical dotted line: time at which BrdU was removed. Each time point represents between 5 and 11 mice. The dose of BrdU administered is 180 mg i.p. per mouse at the start followed by water containing 800 micrograms per ml BrdU continuously for 10–13 days (for more details on the experimental procedure see [2]). Plotting BrdU uptake on a timescale of hours and chase on a timescale of days clearly shows a markable change in kinetic slope at 5 time points (black arrows). Definitive biological events can be attributed to the first three changes, which motivated a three-stage (second stage shaded in solid grey) parameter estimation strategy, as discussed in the main text. The change occurring at chase day 70 (fourth arrow) can be regarded as the starting point of the long label retaining tail in the graph. The observed data at chase day 177 (fifth arrow) seems like an outlier since subsequent time points return to the kinetic slope as observed prior to day 177.

Figure 3. Deterministic LRC model predictions of BrdU content.

Brown line: HSC chase; red line: d-HSC; green line: a-HSC; dashed line: HSC uptake; blue x: observed data. Left panel: stage 1 & 2 predictions (uptake and first 10 days of chase) on a timescale of hours; right panel: uptake and chase predictions on a timescale of days. (A) One-population model predictions. This model can satisfactorily describe BrdU uptake but not the long term label-retention. (B) Two-population model predictions with a BDT of 4 and 30% d-HSC proportion. The effect of a smaller d-HSC population is visible in the left panel of this plot. (C) Two-population model predictions with a BDT of 6 or 7 and 40% d-HSC proportion. This model gave the best overall goodness-of-fit. Activation of the d-HSCs can be clearly seen in the left panel where both d-HSC and a-HSC are predicted to take BrdU up at the same rate (cycling about once every 10 days). During chase d-HSCs return to a dormant state and are predicted to divide about once every 165 days, whilst a-HSCs divide once every 31 days, diluting label much faster than the d-HSCs.

Figure 4. Summary of experimental data and modeling conclusions.

Kinetics of uptake and loss of BrdU within the phenotypic HSCs (LinSca1CD34CD150CD48CD135) as determined experimentally (solid black line) overlaid with the relative proportion of d-HSCs (dormant HSC) and a-HSCs (active HSCs) amongst BrdU phenotypic HSCs estimated by our modeling (dark grey shaded curve). The populations that can be found at any time point of a LRC experiment amongst phenotypic HSCs are indicated in the light grey box. The time points of chase at which the d-HSCs represent 75% and 92.5% of the LRC phenotypic HSCs are indicated by solid black circles.

Figure 5. Stochastic LRC model predictions of BrdU content.

Mean observed data and variation are indicated in blue; shaded area represents estimated variance of the predictions from 1000 stochastic simulations, each with and initial 3750 HSCs. Left plot is one-population stochastic predictions. Right plot is two-population stochastic prediction corresponding to the deterministic model of Figure 3C with a BDT of 6 or 7 strands. It is clear that the one-population predictions and hence stochastic variation alone cannot explain the observed long-term label retention. The two-population predictions in turn encapsulates all observed averages, although there are still some unexplained variation at some time points.

Figure 6. Average theoretical BrdU percentage of a single cell for a given number of divisions during uptake and chase.

Yellow: cell has divided once during uptake; green: cell has divided twice during uptake; red: cell has divided thrice during uptake; violet: cell has divided four times during uptake, more than four divisions will also be in this group; red lines: detection thresholds. Slices of each uptake pie are cumulative. Uptake divisions follow in a clockwise direction and chase divisions follow in an anticlockwise direction. Slices intersected by red lines indicate the number of divisions for the cell to fall below the detection threshold.