HSCs “shift and drift” during transplantation

In this issue of Blood, Radtke et al¹ demonstrate that transplanted hematopoietic stem cells (HSCs) produce mature cells earlier than expected and, consequently, stochastically drift into clonal pools.

The survival of patients who underwent hematopoietic transplantation depends on the timely recovery of blood and immune cells. However, the hematopoietic cells that are responsible for this life-saving function at different times after the transplantation remain unidentifiable. The reigning view holds that hematopoietic regeneration after transplantation occurs in 2 phases, with HSCs principally contributing to the long-lived second phase that initiates after the first 6 months.^2,3 Radtke et al present evidence in nonhuman primate models that early and rapid blood cell recovery is also driven by long-term persisting stem cells (see figure). These findings are profoundly significant for stem-cell transplantation and gene therapy fields, in which determining the source of modified blood cells is essential.

To quantitatively assess the clonal dynamics of transplanted stem cells in nonhuman primate models, researchers used both DNA barcodes and deep vector integration-site analysis. Prior studies by these researchers showed that increasing the detection sensitivity was vital in detecting the early contribution of long-term repopulating clones.^4,5 With these more sensitive methods, Radtke et al now tracked transplanted HSCs across multiple lineages for up to 4.5 years. Although many stem-cell clones rapidly divide and disappear after the first year, about 50% of the early-contributing clones are still detectable at the last-assessed time points. These results suggest that past studies in humans and primates might have failed to detect early contribution of HSCs because of the small size and large number of clones during the first phase. But could HSCs be sufficient to explain everything? With these densely mapped data, Radtke et al generated a simple stochastic model that accounted for stem-cell self-renewal and differentiation. Indeed, this purely stem-cell driven model was sufficient to estimate the number of persistent clones from just the input population. Intriguingly, the model was insufficient to fit the kinetics and, especially, the sharp decrease in clonal abundance at the intermediate stages of recovery. To fit these early kinetics of clonal loss, investigators had to create a two-phase model in which the proportion of dividing stem cells (per day) was very high in the early phases of the transplantation (100%) and much lower in the late phase (10%). To validate this heterogeneity, researchers performed single stem-cell cultures, which showed a wide range of clonal expansion, roughly correlating with the percentage of CD34⁺CD90⁺ cells, suggesting that stochastic stem-cell expansions during the early phase are sufficient to explain the clone size heterogeneity observed in the late phase. Radtke et al further challenged their predictions using CellTag, an expressed DNA barcoding system, which allowed them to query the clonal dynamics of CD34⁺ cells using single-cell transcriptomics.⁶ The single-cell nature of the assay allowed them to reliably quantify the clonal sizes while simultaneously verifying their exact molecular identity. These results confirmed that HSCs contained a collection of clonally expanded stem-cell pools, which are the cause for the observed distributions in mature progenitors. Thus, the researchers conclude that HSCs are sufficient to drive all phases of cell recovery during hematopoietic transplantation and that the fast cell divisions required to drive early regeneration result in the stochastic expansion of clonal stem-cell pools, which stabilize after the first year after transplantation.

Although such simple stochastic models may be sufficient to explain clonal dynamics, previous work in mouse HSCs has suggested that various regenerative behaviors, including lineage bias and stem-cell expansion, are partly deterministic, driven by intrinsic and heritable properties.^7-9 Importantly, in these cases, stochastic models were also sufficient to fit distributions of individual samples. Still, clones acted more similarly than expected when the same clone was measured across multiple independent parallel functional tests, suggesting that both stochastic and deterministic mechanisms contribute to HSC dynamics. Further experiments in nonhuman primates should be able to determine whether heritable, deterministic programs also influence regenerative behaviors in this model. Finally, although HSCs are clearly sufficient to drive blood formation across all phases, the relative contribution of other progenitors in a real transplantation scenario is still uncertain. Solving this will require distinctly labeling and tracing multiple stem and progenitor populations simultaneously.¹⁰ These and future findings will be of extreme importance to a growing community of clinical and translational HSC researchers who need to consider the hematopoietic populations that they need to target to achieve the most efficacious results for patients.

Conflict-of-interest disclosure: The author declares no competing financial interests.