Clonal analysis of murine hematopoiesis.

A: Experimental outline for lineage tracing of unperturbed hematopoiesis. M2/HSB/Tn mice were labeled with Dox for four weeks. The DsRed+ fractions of the indicated cell populations were isolated from bone marrow approximately a hundred weeks after the removal of Dox for Tn tag library preparation and sequencing.

B: Experimental outline for lineage tracing of hematopoiesis following 5-FU treatment. Three weeks after Dox withdrawal, the M2/HSB/Tn mice received three rounds of 5-FU injections at a 3-week interval. The indicated DsRed+ cell populations were isolated from bone marrow approximately eighty weeks after the last 5-FU infusion.

C: Experimental outline for comparative analysis of Ehi HSC clonal behaviors in native hematopoiesis and after bone marrow transplantation. One-fourth of the isolated donor DsRed+ Ehi HSCs were used for Tn tag detection. The rest cells were transplanted into lethally irradiated mice. The DsRed+ fractions of the indicated cell populations were isolated monthly from peripheral blood. The recipient mice were sacrificed three months after transplantation to analyze Tn tags in bone marrow cell populations.

FACS gating strategy for cell isolation.

A: The gating strategy used to identify and isolate Ehi/lo/neg HSC, HPC-1/-2, MkP, and MyP from cKit-enriched bone marrow cells. The DsRed+ fractions of the indicated cell populations are sorted for Tn tag detection.

B: The gating strategy used to identify and isolate Gr, Mono, B cells, and T cells from cKit-depleted bone marrow cells or peripheral blood cells. The DsRed+ fractions of the indicated cell populations are sorted for Tn tag detection.

C: Frequency of the DsRed+ fractions in the indicated bone marrow cell populations from aged mice sacrificed for Tn tag analysis. The two uninduced mice were analyzed at 120 and 123 weeks of age. Each circle represents data from an individual mouse (uninduced: black, n=2; unperturbed: red, n = 12; 5-FU: blue, n = 5). Bars represent means with SEM.

Characterization of the TRACE method for transposon tag detection.

A: Schematic of the semi-quantitative TRACE method for Tn tag detection.

B: Experimental design to validate the semi-quantitative property of the TRACE method for Tn tag detection. DNA from samples without clonal expansion (freshly induced granulocytes) is added at varying dilutions for Tn tag analysis. “R2” depicts the coefficient of determination of the linear regression.

Differentiation patterns of hematopoiesis in unperturbed and 5-FU-treated mice.

A: Heatmap depicting the alignment of Tn tags from all mice examined, with each row representing a unique Tn tag and each column representing the cell types.

B: Heatmap showing the result of Louvain clustering on multi-cell type clones. Tn tags are colored by log2-transformed UMI counts, which reflect their abundance. Tn tags used in the plots are pooled from mice assayed in the unperturbed condition (n = 12) and after 5-FU treatment (n = 5).

Alignment of transposon tags.

A-B:Alignment of Tn tags across all assayed bone marrow cell populations, with each row representing a unique transposon tag and each column representing a sampled cell population. Tn tags are colored by their log2-transformed UMI counts, reflecting their abundance, and are arranged by rank. Tn tags detected in mice examined in the unperturbed (A, n = 12) and 5-FU-treated (B, n = 5) mice are shown togetheror individually.

The correlation between HSC expansion and MkP-restricted differentiation.

A: The percentage of clones in each cluster that contained the indicated cell types.

B: Violin plots showing the Tn tag UMI counts of the indicated cell types in each clone cluster. Tn tags are pooled from unperturbed mice (red circle, n = 12) and 5-FU-treated mice (blue circle, n = 5).

C: Correlations between the frequencies of the indicated multipotent hematopoietic cell populations in Lin- cKit+ Sca1+ (LSK) cells and the frequencies of the indicated clone cluster in multi-cell type clones. Each dot represents data from a single mouse (unperturbed: red, n = 12; 5-FU: blue, n = 5). “rs” depicts the Spearman correlation coefficient.

D: Correlations between the frequencies of CD48- cells in MkP and the frequencies of the indicated clone cluster in multi-cell type clones. Each dot represents data from a single mouse (unperturbed: red, n = 12; 5-FU: blue, n = 5). “rs” depicts the Spearman correlation coefficient.

Expansion of HSC and CD48- MkP after 5-FU-induced regeneration.

A: Representative flow cytometry plots showing the frequency of HSC in Lin- cKit+ Sca1+ (LSK) bone marrow cells from unperturbed and 5-FU-treated mice.

B: Representative flow cytometry plots show the frequency of CD48- cells in MkP from unperturbed and 5-FU-treated mice.

The heterogeneous differentiation states of HPC-1.

A: The Δ Bayesian information criterion (ΔBIC) demonstrates the relative difference between the best model that explains the log2 ratios of Tn tag UMI counts in downstream differentiated cell populations and upstream HPC-1 (down/up pairs: MyP/HPC-1, Gr/HPC-1, or Mono/HPC1), and Gaussian mixture models (GMM) with varying numbers of components. The results from models that assume each component has equal variance (gray) and unequal variance (black) are both shown.

B: The observed distributions of the log2 ratios of Tn tag UMI counts in the three down/up pairs (gray bar) are shown alongside the distributions predicted by the one-(black curve) or two-component (red curve) Gaussian mixture models with minimal ΔBIC. The dashed lines illustrate the mean values predicted by the one-(black) or two-component (red) Gaussian mixture models.

C: Heatmaps showing the classification results of differentiation-active and -indolent clones defined based on their log2 UMI count ratios in the three down/up pairs using a two-component Gaussian mixture model with minimal ΔBIC. Tn tags are colored by log2-transformed UMI counts.

D: The number of differentiation-active and -indolent clones defined based on their log2 UMI count ratios of the three down/up pairs in each mouse.

E: The number of clones with indicated B-cell UMI counts in differentiation-active and - indolent HPC-1 clones.

G: Violin plots showing the distribution of the log2 ratios of Tn tag UMI counts in the three down/up pairs in differentiation-active and -indolent HPC-1 clones classified by the individual down/up pairs.

Except for Figure 5D, data used in this figure is pooled from mice assayed in the unperturbed state (n = 12) and after 5-FU-induced regeneration (n = 5).

Gaussian mixture model classification of MyP by their differentiation states.

A: The Δ Bayesian information criterion (ΔBIC) demonstrating the relative difference between the best model that explains the log2 ratios of UMI counts of Tn tags between downstream cell populations (Gr, Mono) and upstream MyP, and Gaussian mixture models with varying numbers of components. The results from models that assume each component has equal variance (gray) and unequal variance (black) are both shown.

B: The observed distributions of the log2 ratios between downstream cell populations (Gr, Mono) and upstream MyP (gray bar) are shown alongside the distributions predicted by the one-(black curve) or two-component (red curve) Gaussian mixture models with minimal ΔBIC. The dashed lines illustrate the mean values predicted by the one-(black) or two-component (red) Gaussian mixture models.

C: Violin plots showing the distribution of UMI counts of Tn tags detected in MyP in differentiation-active and -indolent clones.

Data in this figure is pooled from mice assayed in the unperturbed state (n = 12) and after 5-FU-induced regeneration (n = 5).

The cellular origins for clonal expansion in mature lineage blood cells.

A: Pielou’s index measures the evenness of UMI counts of Tn tags detected in the freshly induced granulocytes (Fresh) and all assayed bone marrow cell populations. Each circle represents data from a technical repeat (Fresh, n = 12) or a single mouse (unperturbed: red, n = 12; 5-FU: blue, n = 5). Bars represent means with SEM. Two-tailed unpaired Student’s t-tests are used to compare data from two experimental groups. ns, not significant, *p < 0.05, ***p < 0.001.

B and C: Frequency of clones detected in downstream mature lineages (MkP, Gr, Mono, B cells, T cells) with different UMI counts whose Tn tags also detected in multipotent hematopoietic stem and progenitor cells (HSPC: Ehi/lo/neg HSC, HPC-1, HPC-2) (B: unperturbed, n = 12; C: 5-FU, n = 5). Bars represent means with SEM.

D-J: The results of ElasticNet illustrates the linear relationships of UMI counts of Tn tags detected in downstream cell populations (MkP: C1/C2 (D), C3-C7 (E); B cells (F); T cells (G); Gr (H); Mono (I); and HPC-1 (J)) with their upstream progenitors (HSC: Ehi/lo/neg HSC (D, J); Ehi/lo/neg HSC, HPC-1, HPC-2, MyP (E); Ehi/lo/neg HSC, HPC-1, MyP (F-I)). The x axes represent a series of models with different penalty settings. The fraction of deviance explained and the number of non-zero coefficients in the corresponding models are shown separately in the lower and upper bound. The y axes show the coefficients (ai) estimated in these models. Data in this plot is pooled from mice assayed in the unperturbed state (n = 12) and after 5-FU-induced regeneration (n = 5).

Clonal analysis of hematopoiesis after transplantation.

A: The heatmap showing donor clones detected in recipient mice after transplantation. Tn tags are colored by log2-transformed UMI counts. The transposon tags used in the plot are pooled from donor mice used for transplantation (n = 2 for each experimental group).

B and C: Survival probability of different donor clone types measured by the fraction of donor clones detected in transplant recipients in all clones in the corresponding clone type. B: donor clones from unperturbed mice; C: donor clones from 5-FU treated mice.

D: The heatmap showing the result of Louvain clustering on clones detected in the recipients’ bone marrow and blood cell populations. Tn tags are colored by log2-transformed UMI counts. The Tn tags used in the plot are pooled from recipient mice transplanted with donor Ehi HSCs from the unperturbed group (n = 6) and the 5-FU group (n = 4).

E: The average frequency of the individual recipient clone types in mice receiving donor Ehi HSCs from unperturbed (n=6) and 5-FU-treated mice (n=4). Bars represent means with SEM. Two-tailed unpaired Student’s t-tests are used to compare data from two experimental groups. *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.001.

F: The number of recipient clone types generated by the seven types of donor multi-cell type clones. Clones from the unperturbed and 5-FU-treated donor mice are pooled for the calculation.

G: Alignment of recipient clones detected in two recipients. The rows represent clones, and the columns represent recipient mice. Clones are colored according to their cluster types in the recipient mice from which they are derived.

H-K: Violin plots showing the UMI counts of the different types of donor clones in the donor Ehi HSC (H), the donor mature cell lineages (Gr, Mono, B-cells, T-cells) (I), the recipient Ehi HSC (J), and the recipient mature cell lineages (Gr, Mono, B-cells, T-cells) (K). Clones from unperturbed and 5-FU-treated donors and recipients are separately pooled for the analysis.

Alignment of transposon tags of donors.

A-B:The heatmap showing all clones (A) and multi-cell type clones (B) detected in donor mice used for transplantation. Tn tags are colored by log2-transformed UMI counts. The transposon tags used in the plot are pooled from donor mice (n = 2 for each experimental group).