HSPCs display within-family homogeneity in differentiation and proliferation despite population heterogeneity
- Institut Curie, Université PSL, Sorbonne Université, CNRS UMR168, Laboratoire Physico Chimie Curie, France
- Institut Curie, PSL Research University, CNRS, France
- Université Paris-Saclay, France
- INSERM, UMR1287, Gustave Roussy, France
- Gustave Roussy, France
- Université de Paris, France
- Hamilton Institute, Maynooth University, Ireland
Figures
Figure 1
High-throughput simultaneous division and differentiation tracking per ancestor.
A single-cell suspension was obtained by flushing femurs, tibia, and iliac crests. Cells were stained with fluorescently labeled antibodies for phenotypic identification. After this first antibody staining, the cell suspension was then split into four equal parts, each of which was stained with a distinct 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) and CellTrace Violet (CTV) combination. From each of these CTV/CFSE preparations, a single cell was index sorted into 90 wells of a 96-well plate, resulting in four distinctly CTV/CFSE-colored cells per well. In addition, for each ancestor type, a small bulk of 100 cells of each color combination were sorted into a single well as a control. After 24 or 48 hr of culture, cells were stained with fluorescently labeled antibodies for phenotypic identification and analyzed on a flow cytometer. The data from all wells and the small 100 cell bulks were combined and used to set gates for determination of generation number and phenotypic cell type. Those gates were then applied to the data from each well to obtain lineage, division, and differentiation information for each cell.
-
Figure 1—source data 1
Cell type assignment based on cell surface marker.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig1-data1-v2.xlsx
Figure 2 with 3 supplements
Despite population-level heterogeneity, individual hematopoietic stem and progenitor cell (HSPC) families are substantially homogeneous.
Plots are fractionated by each ancestor type (SLAM-HSCs, ST-HSCs, and multi-potent progenitors [MPPs]) and cocktail (with and without IL-3 + IL-6, indicated by + and -, respectively). (A) Simultaneous visualization of family membership, generation number, and cell type of offspring from initially seeded ancestors harvested at two time points (24 and 48 hr). Each row presents the offspring from a single ancestor. Columns identify the generation number of each recovered cell, with their phenotypic cell type indicated by color coding. Rows are sorted in increasing order of the difference between maximum and minimum generations in each ancestor’s family (generational range). (B) To test the differentiation pattern of our cell phenotypes after culture, sorted cell types as shown in Figure 1 were cultured in methylcellulose cultures and evaluated for colony forming units (CFUs) formation for megakaryocytic (MK), granulocytic/monocytic (GM), erythroid burst-forming (BFU-E), or mixed colony formation of all types (CFU-MK, CFU-GM, and BFU-E) with or without MK. Late MP did not produce any colonies in methylcellulose culture (C) The percentages of MK, CFU-MK, and mixed colony formation are indicated from the culture in serum-free fibrin clot. GMP did not produce any colonies as expected. (D) Percentage of each recovered cell type. (E) Distribution of the maximum generation per family, as indicated by color coding. (F) Proportions of recovered cell types for ancestors that have not yet divided. (G) For ancestors that have divided only once and for whom two offspring are recovered, percentage having symmetric and asymmetric fates with and without differentiation. (H) Percentage of families with each generational range. The 48 hr data (red bars) is compared to a mathematical model parameterized by a single coefficient, ρ, which encodes the correlation in whether cells in the same generation within a family divide or cease to divide (Materials and methods). Shown is prediction for both the maximum likelihood best-fit value of ρ (value and blue/white bars) and, as a reference, a range of other values of ρ (solid blue bars). For (D–H), error bars indicate 95% confidence intervals calculated via basic bootstrap (Materials and methods). Sample sizes and p-values (from permutation tests, see Materials and methods) from the panels can be found in Figure 2—source data 1 and 2.
-
Figure 2—source data 1
Number of families per progenitor and condition.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig2-data1-v2.xlsx
-
Figure 2—source data 2
Significance values from permutation testing procedures.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig2-data2-v2.xlsx
Figure 2—figure supplement 1
Cloning efficiency.
Figure 2—figure supplement 2
Fluorescence from antibody staining is not retained during culture.
To assess whether fluorescence from antibodies used during cell isolation could be retained during culture and affect the subsequent analysis of cell differentiation status, we analyzed sorted cells after 24 hr of culture. For none of the antibodies used during sort, any fluorescence was detected.
Figure 2—figure supplement 3
Proportions of recovered cell types for ancestors that have divided only once.
Error bars indicate 95% confidence intervals calculated via basic bootstrap (Materials and methods). Sample sizes for this panel can be found in Figure 2—source data 1.
Figure 3 with 3 supplements
Differentiation and division progress in tandem.
(A) Percentage of cells at each differentiation stage for each generation for each ancestor type. Error bars indicate basic bootstrap 95% confidence intervals (Materials and methods). Sample sizes for this panel can be found in source data. (B) The Uniform Manifold Approximation and Projection (UMAP) algorithm was applied to the phenotypic data pooled from all time points, conditions, and ancestor cell types. Each cell is projected into the UMAP coordinates and color-coded according to the log of their fluorescence intensity for c-Kit, Sca-1, CD150, Flt3, CD48, and CD16/32 at the time of analysis (see also Figure 3—figure supplement 2; and for each ancestor cell type plotted separately on the UMAP, see Figure 4—figure supplement 1). (C) Projection of traditionally gated data onto the UMAP, pooled data (left), 24 hr (middle), and 48 hr (right). (D) Projection of cell generation number data onto the UMAP for each ancestor type (see also Figure 3—figure supplement 3). Sample sizes for panel (A) can be found in Figure 3—source data 1.
-
Figure 3—source data 1
Number of cells per generation from a given progenitor.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig3-data1-v2.xlsx
Figure 3—figure supplement 1
Additional information plotted onto the differentiation of Uniform Manifold Approximation and Projection (UMAP).
Cells are color-coded by time point (top left), experiment (top right), CellTrace stain (bottom left), and culture condition (bottom right) and plotted on the UMAP from Figure 3. The experiments and CellTrace combinations were evenly distributed on the UMAP. Differences based on both the time point and culture condition recapitulate what is described in Figures 2B and 3B. C indicates cells originating from ancestors stained with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE), V with CellTrace Violet (CTV), C > V indicates staining with both 2.5 µM CFSE and 1.25 µM CTV, and V > C indicates staining with both 1.25 µM CFSE and 2.5 µM CTV.
Figure 3—figure supplement 2
Generation numbers projected onto the Uniform Manifold Approximation and Projection, fractionated by ancestor type and time point.
Figure 3—figure supplement 3
Per ancestor type fluorescence intensities during sort projected on the Uniform Manifold Approximation and Projection.
As in Figure 3D, but fractionated by ancestor type.
Figure 4 with 2 supplements
Families are highly concordant in differentiation.
(A, B) Number of cell types per family in the observed data compared with the average of 250,000 permutations of the data at 24 hr (A) and 48 hr (B). Error bars indicate 95% confidence intervals based on permutations (see Materials and methods). (C) Cells from the 15 families with the largest number of cells are color-coded by family and projected onto the Uniform Manifold Approximation and Projection in Figure 3. (D) The cumulative percentage of the maximum division of offspring from ancestor SLAM-HSCs rank-ordered by their expression level (fluorescence intensity) of CD48 (blue) or Sca-1 (red) during sort. (E) The cumulative percentage of offspring, presenting a given cell type, from ancestor cells rank-ordered by increasing cell surface marker expression. * indicates a significant deviation from the diagonal (black) as determined by Jonckheere's trend test (p-values in Figure 4—source data 1). Sample sizes for all panels can be found in Figure 4—source data 2.
-
Figure 4—source data 1
Significance values from the Jonckheere’s rend test between expression levels at sort and offspring maximum generation as shown in Figure 4D.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig4-data1-v2.xlsx
-
Figure 4—source data 2
95% confidence intervals of Spearman r and p-values of correlations.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig4-data2-v2.xlsx
-
Figure 4—source data 3
Significance values from the Jonckheere’s trend test between expression levels at sort and presence of a given cell type among offspring.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig4-data3-v2.xlsx
-
Figure 4—source data 4
Summary of significant findings as shown in Figure 4E.
- https://cdn.elifesciences.org/articles/60624/elife-60624-fig4-data4-v2.xlsx
Figure 4—figure supplement 1
fluorescence intensity correlation within families and correlation with division for additional markers and ancestor types.
(A) Correlation between fluorescence intensities of the given marker on each analyzed cell and the fluorescence intensity of the same marker on its ancestor index sort. Spearman’s r is reported for index sort expression levels of each ancestor type. (B) The cumulative percentage of the maximum division of offspring from ancestor cells rank-ordered by their expression as in Figure 4D, but for each ancestor cell type and marker. Sample sizes for all panels can be found in Figure 2—source data 1, and 95% confidence intervals for the correlation r can be found in Figure 4—source data 1.
Figure 4—figure supplement 2
The cumulative sum of the offspring presenting a given cell type from ancestor cells rank-ordered by cell surface marker expression as in Figure 4E, without normalizing the offspring count, and plotted for each ancestor cell type, time point, culture condition, and marker.
Solid lines indicate observed sums, and dashed lines indicate the diagonal for the corresponding observation.
Additional files
-
Source data 1
Complete dataset used in the article for each analyzed cell, its generation, assigned cell type, family membership, marker fluorescence intensities at the time point of analysis, and marker fluorescence intensities of its ancestor cell during sort.
- https://cdn.elifesciences.org/articles/60624/elife-60624-data1-v2.xlsx
-
Source code 1
All code used for generation of figures and statistical testing.
- https://cdn.elifesciences.org/articles/60624/elife-60624-code1-v2.zip
-
Transparent reporting form
- https://cdn.elifesciences.org/articles/60624/elife-60624-transrepform-v2.pdf
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
HSPCs display within-family homogeneity in differentiation and proliferation despite population heterogeneity
eLife 10:e60624.
https://doi.org/10.7554/eLife.60624
