Somatic Mutations Reveal Lineage Relationships and Age-Related Mutagenesis in Human Hematopoiesis

Abstract

Mutation accumulation during life can contribute to hematopoietic dysfunction; however, the underlying dynamics are unknown. Somatic mutations in blood progenitors can provide insight into the rate and processes underlying this accumulation, as well as the developmental lineage tree and stem cell division numbers. Here, we catalog mutations in the genomes of human-bone-marrow-derived and umbilical-cord-blood-derived hematopoietic stem and progenitor cells (HSPCs). We find that mutations accumulate gradually during life with approximately 14 base substitutions per year. The majority of mutations were acquired after birth and could be explained by the constant activity of various endogenous mutagenic processes, which also explains the mutation load in acute myeloid leukemia (AML). Using these mutations, we construct a developmental lineage tree of human hematopoiesis, revealing a polyclonal architecture and providing evidence that developmental clones exhibit multipotency. Our approach highlights features of human native hematopoiesis and its implications for leukemogenesis.

Keywords: HSC; developmental lineage tree; human hematopoiesis; leukemia; mutational processes; somatic mutations.

Graphical abstract

Figure 1

Determining Somatic Mutations in Hematopoietic Progenitors (A) Schematic overview of experimental setup to catalog somatic mutations in single human blood progenitors. MSCs, mesenchymal stem cells; WGS, whole-genome sequencing. (B) Average number of base substitutions in HSCs and MPPs (extrapolated to the whole autosomal genome) of the donor A. Error bars indicate SD. Each data point represents a single HSC or MPP clone. The p value indicates no statistical difference (NS) between the number of base substitutions in HSCs and MPPs (two-sided t test). (C) Relative contribution of the indicated mutation types to the base substitution spectra in HSCs and MPPs. Error bars indicate SD. Each data point represents a single HSC or MPP clone. (D) Average number of indels in HSCs and MPPs (extrapolated to the whole autosomal genome) of the donor A. Error bars indicate SD. Each data point represents a single HSC or MPP clone. The p values indicate no statistical difference (NS) between the number of indels in HSCs and MPPs (two-sided t test).

Figure 2

Age-Associated Mutation Accumulation in Human Blood Progenitors (A) Correlation of the number of base substitutions accumulated per genome with age of the independent donors we assessed. Each data point represents a single clone. p value of the age effect in the linear mixed model is indicated above the plot (two-tailed t test). The sample size is 7 donors with a total of 22 clones sequenced (9 HSCs, 9 MPPs, and 4 cord blood progenitors). Linear mixed model was performed on the clones from adult bone marrow (5 donors; 18 clones). Dotted line indicates the extrapolation of this correlation to birth (age = 0). Subsequently, we confirmed this value at birth by performing WGS on umbilical cord blood samples (n = 4 clones) of 2 independent donors. (B) Annual base substitution rate estimated by the linear mixed model in (A). Error bars represent the 95% confidence interval of the slope estimate. (C) Correlation of the number of indels with age of the donors. p value of the age effect in the linear mixed model is indicated above the plot (two-tailed t test). Linear mixed model was performed on the clones from adult bone marrow. Dotted line indicates the extrapolation of this correlation to birth (age = 0). (D) Relative contribution of the indicated mutation types to the base substitution spectra for each donor. The total number of base substitutions and assessed clones per donor is indicated. (E) Relative contribution of the indicated mutation types to the total number of indels for each donor. The total number of indels and assessed clones per donor is indicated.

Figure 3

Signatures of Mutational Processes in HSPCs (A) Principal-component analysis of 96-base substitution spectra. Spectra of single base substitutions sequenced clones (colored dots, 16 colon, 13 small intestine, 10 liver, 18 HSPCs, and 4 cord blood clones) and 96 profiles of signatures. Mutations for all four umbilical-cord-blood-derived samples were pooled together, as mutation load was low in these samples. Directions of signature contributions between samples are indicated on the x and y axis; see Figure S3. (B) Total 96 mutational profiles for all mutations in the HSPC clones. (C) Absolute contribution of each mutational signature type (extrapolated to the whole autosomal genome) plotted against the age of the donors. The observed absolute contributions of the signatures in the umbilical cord blood are plotted as triangles, and the numbers are indicated in the plot marked with CB. The p values of the age effects per tissue are shown (linear mixed model; two-tailed t test, excluding the cord blood data) with extrapolations of signature accumulation to the birth drawn in dotted lines. (D) Transcriptional strand bias profile for all C > T transitions in HSPCs (pooled); ^∗p < 0.05; two-sided Poisson test. (E) Mean base substitution load per year per 10-MB genome of the indicated mutational signature types for the HSPC clones (n = 18) and the AML samples (n = 24). The total number of base substitutions and assessed samples per category is indicated. Error bars indicate SD. Each data point represents a single HSPC or AML sample.

Figure 4

Reconstruction of the Development Lineage Tree and Branch-Specific Contributions to Different Blood Lineages (A) Phylogenetic tree indicating the relatedness of the whole-genome sequenced clones. The different branches, which are defined by individual somatic base substitutions, are indicated with letters a–m. In the gray panel, the presence or absence of those base substitutions (y axes) in both the whole-genome sequenced clones and amplicon-based sequenced clones (x axes) are for HSCs (black) and MPPs (gray). The ɑ branch highlighted with red-dashed box showed asymmetrical contribution to HSCs and MPPs with a statistical significance (p < 0.05; permutation test). (B) As a continuation of the phylogenetic tree, the number of their unique mutations in the WGS clones (black for HSCs; gray for MPPs) is shown by the length of their branches. The red horizontal line indicates the time of birth estimated based on the average number of base substitutions in the umbilical cord blood cells with the 95% confidence interval as gray box. (C) Dendrogram depicting the correlations between the mature populations based on the hierarchical clustering of variant allele frequencies (VAFs) of the somatic mutations a–m.