Tracking hematopoietic stem cells and their progeny using whole-genome sequencing

Abstract

Despite decades of progress in our understanding of hematopoiesis through the study of animal models and transplantation in humans, investigating physiological human hematopoiesis directly has remained challenging. Questions on the clonal structure of the human hematopoietic stem cell (HSC) pool, such as "how many HSCs are there?" and "do all HSC clones actively produce all blood cell types in equal proportions?" remain open. These questions have inherent value for understanding normal human physiology, but also directly inform our comprehension of the process by which the system is subverted to drive diseases of the blood, in particular blood cancers and bone marrow failure syndromes. The critical link between normal and abnormal hematopoiesis is perhaps best illustrated by the recent discovery of clonal hematopoiesis in healthy people with no abnormal blood parameters. In such individuals, large clones derived from single cells are present and are dominant relative to their normal counterparts, but their presence does not necessitate abnormal blood cell production. Intriguingly, however, these individuals are also at a significantly greater risk of developing leukemias and of cardiovascular events, underscoring the importance of understanding how blood stem cell clones compete against each other.

Figure 1

Numerous possible routes to clonal hematopoiesis. (A) Model A depicts what would happen if the clone in question has expanded in size in an independent fashion (i.e., increased self-renewal) without affecting or being affected by the normal stem cell pool, thus increasing the total number of stem cells. (B) Model B depicts what would happen if the clone actively competes with the normal stem cell pool to take up a larger share of the total stem cell number without actually changing that number. (C) Model C depicts what would happen if the stem cell pool is decreasing in size, but the clone in question displays greater resilience and represents a greater proportion of the total stem cell pool not because it has expanded greatly, but rather because the other stem cells have been depleted more (i.e., the denominator shrinks). In all cases, a static measurement of the clone would show an ∼60%–70% contribution of the clone but the cellular mechanism would be very different.

Figure 2

Effect of cell types that are sequenced on studies. (A) Sequencing progenitors reveals mutations that occurred in stem cells. Mutations that occurred in stem cells accumulate over time. When a progenitor from an adult is sequenced, most of the mutations discovered occurred in stem cells. (B) The cell types that are sequenced may affect the phylogeny that is reconstructed if stem cells do not produce balanced output. In the top phylogeny, all stem cells produce both myeloid and lymphoid cells. Sequencing only myeloid progenitors samples the phylogeny evenly, such that an accurate representation of the phylogeny is reconstructed. In the bottom phylogeny, some stem cells produce disproportionately more lymphoid than myeloid progenitors. Sequencing only myeloid progenitors undersamples parts of the phylogeny that are biased toward lymphoid cells.

Figure 3

Phylogenies from (A–C) Lee-Six et al. and (D–F) Osorio et al. . (A) The phylogeny of cells is shown in gray, with branch lengths proportional to the numbers of somatic mutations (y-axis). Each tip of the phylogeny leads to a stem or progenitor cell that has been whole-genome sequenced. Information from targeted sequencing of peripheral blood granulocytes is overlain. This is shown more clearly in the inset (B), which zooms in on one portion of the tree. On top of the underlying structure of the phylogeny (gray) are placed horizontal bars. Each bar represents a mutation in the bait set for targeted sequencing. The bars are colored according to the proportion of cells in the sample that carry the mutation, indicated by the color scale. Undetectable mutations are colored gray and shown as smaller bars. Mutations are assigned to a branch based on the colonies in which they are present. (C) This schematic explains that the allele fractions of targeted mutations in peripheral blood decline down the branches because of undetected coalescences with stem cells that were not whole-genome sequenced, but are producing granulocytes. (D) The phylogeny of whole-genome sequenced clones. Branches are labeled a–m, each of which represents a mutation that defines the lineage. The presence (hematopoietic stem cells: black, multipotent progenitors: dark gray) or absence (pale gray) of each mutation in genotyped clones is shown in the panel below the phylogeny. (E) Continuation of the phylogeny for whole-genome sequenced clones, with branch lengths proportional to mutation load. (F) Allele fractions of mutations a–m in mature blood cell populations.

Figure 4

Simulations illustrate that phylogenies reflect the population size trajectory. All simulations were produced using a Fisher–Wright model of neutral drift. At the end of each simulation, 100 cells are sampled and their phylogeny reconstructed. (A) A constant population size. (B) A sudden strong bottleneck in population size at the point in time indicated by the red dashed line. (C) An increase in population size between the red dashed lines, which stabilizes thereafter.