Quantitative lineage tracing strategies to resolve multipotency in tissue-specific stem cells

Abstract

Lineage tracing has become the method of choice to study the fate and dynamics of stem cells (SCs) during development, homeostasis, and regeneration. However, transgenic and knock-in Cre drivers used to perform lineage tracing experiments are often dynamically, temporally, and heterogeneously expressed, leading to the initial labeling of different cell types and thereby complicating their interpretation. Here, we developed two methods: the first one based on statistical analysis of multicolor lineage tracing, allowing the definition of multipotency potential to be achieved with high confidence, and the second one based on lineage tracing at saturation to assess the fate of all SCs within a given lineage and the "flux" of cells between different lineages. Our analysis clearly shows that, whereas the prostate develops from multipotent SCs, only unipotent SCs mediate mammary gland (MG) development and adult tissue remodeling. These methods offer a rigorous framework to assess the lineage relationship and SC fate in different organs and tissues.

Keywords: mammary gland; multipotency; progenitor; prostate; stem cell; unipotency.

Figure 1.

Colabeling of BCs and LCs by Lgr5 and Lgr6CreER in the MG. (A–F) Confocal imaging of immunostaining of K14 and Tomato 1 wk after TAM administration to Lgr5-EGFP-IRES-CreER^T2/Rosa-tdTomato pubertal (15 mg) (A–C) or adult virgin (1.5 mg) (D,E) mice. (A,D) Isolated BCs. (B,E) Isolated LCs. (C,F) Doublets of BCs and LCs. In adult mice, doublets could be observed only at high doses of TAM (15 mg). (G) Distribution of basal K14⁺ and luminal K8⁺ cells among Tomato⁺ cells (2865 cells out of three mice). (H–M) Confocal imaging of immunostaining of K14 and Tomato cells 1 wk after TAM administration (15 mg) to Lgr6-EGFP-IRES-CreER^T2/Rosa-tdTomato pubertal (H–J) or adult virgin (K–M) mice. (H,K) Isolated BCs. (I,L) Isolated LCs. (J,M) Doublets of BCs and LCs. (N) Distribution of basal K14⁺ and luminal K8⁺ cells among Tomato⁺ cells (4683 cells out of three mice). Arrows depict BCs, and arrowheads depict LCs. (A,B,D–F,H,I,K,L) Orthogonal projections of three-dimensional (3D) stacks. (C,J,M) Single-plane images from a 3D stack. Bars, 20 µm. Histograms represent the mean. See Supplemental Table S1 for further information on the statistics.

Figure 2.

Colabeling of LCs and BCs by K19 and Sox9CreER in the MG. (A–F) Confocal imaging of immunostaining of K14 and Confetti cells 1 wk after TAM administration (15 mg) to K19CreER^T/Rosa-Confetti pubertal (A–C) or adult virgin (D–F) mice. (A,D) Isolated LC. (B,E) Isolated BCs. (C,F) Doublets of BCs and LCs. (G) Distribution of basal K14⁺ and luminal K8⁺ cells among Confetti⁺ cells (21,408 cells out of three mice). (H–M) Confocal imaging of immunostaining of K14 and Confetti cells 1 week after TAM administration (5 mg) to Sox9CreER^T2/Rosa-Confetti pubertal (H–J) or adult virgin (K–M) mice. (H,K) Isolated LCs. (I,L) Isolated BCs. (J,M) Doublets of BCs and LCs. (N) Distribution of basal K14⁺ and luminal K8⁺ cells among Confetti⁺ cells (4258 cells out of three mice). Arrows depict BCs, and arrowheads depict LCs. (A,B,D,E,H,I,K,L) Orthogonal projections of 3D stacks. (C,F,J,M) Single-plane images from a 3D stack. Bars, 20 µm. Histograms represent the mean. See Supplemental Table S2 for further information on the statistics.

Figure 3.

K14CreER^T2 targets initially and independently unipotent BCs and LCs in the MG. (A) Scheme summarizing the genetic strategy used to target Confetti expression in K14-expressing cells. (B) Scheme summarizing the protocol used to study the fate of cells targeted at puberty using K14CreER^T2/Rosa-Confetti mice. (C,D) Percentage of Confetti⁺ cells in basal K14⁺ and luminal K8⁺ cells (C) and distribution of basal K14⁺ and luminal K8⁺ cells among Confetti⁺ cells (D) 3 d after TAM injection (1.5 mg) at puberty in K14CreER^T2/Rosa-Confetti mice (15524 cells out of three mice). (E–G) Confocal imaging of immunostaining of K14 and Confetti 3 d after TAM administration (1.5 mg) at puberty in K14CreER^T2/Rosa-Confetti mice. Examples of independently and initially labeled BCs and LCs (E), UPs (F), and BPs (G) containing BCs and LCs. (H) Scheme summarizing the protocol used to study the fate of cells targeted in adulthood using K14CreER^T2/Rosa-Confetti mice. (I–L) Confocal imaging of immunostaining of K14 and Confetti at 1 wk after TAM administration (1.5 mg) to K14CreER^T2/Rosa-Confetti adult virgin mice. Examples of independently and initially labeled BCs (I) and LCs (J) as well as UPs (K) and BPs (L) containing BCs and LCs. (M,N) Percentage of Confetti⁺ cells in basal K14⁺ and luminal K8⁺ cells (M) and distribution of K14⁺ BCs and K8⁺ LCs among Confetti⁺ cells (N) 1 wk after TAM injection (1.5 mg) to K14CreER^T2/Rosa-Confetti adult virgin mice (55287 cells out of five mice). (O) Frequency of Confetti⁺ patch compositions 1 wk after TAM injection (1.5 mg) to K14CreER^T2/Rosa-Confetti adult virgin mice. (P) Fraction of UPs among all pairs 1 wk after TAM injection (1.5 mg) to K14CreER^T2/Rosa-Confetti adult virgin mice. Experimental bars (gray) represent the number of UPs divided by the total number of pairs. Model bars (black) represent the prediction of the fraction of UPs among all pairs, which depends only on the relative frequencies of the Confetti colors. (Q) Fraction of total pairs 1 wk after TAM injection (1.5 mg) to K14CreER^T2/Rosa-Confetti adult virgin mice. Experimental bars (gray) represent the total number of pairs divided by the total number of labeled BCs. Model bars (black) represent the prediction of the overall fraction of pairs, which is based on the probability of random labeling of BCs. Although a formal expression is derived in the Supplemental Material, this fraction is roughly proportional to three factors: the degree of chimerism, the specificity of the CRE, and the architecture of the tissue (i.e., how many LCs touch a BC). (R) Scheme summarizing the parameters taken into account in the model. In a hypothetical situation with only two colors, variation of any of these parameters influences the observed number of UPs. Arrows depict BCs, and arrowheads depict LCs. (E,I,J) Orthogonal projections of 3D stack. (F,G,K,L) Single-plane images from a 3D stack. Bars, 20 µm. Error bars represent mean ± SEM, except in O, where it represents one-σ binomial confidence levels for the observed frequencies of patch compositions, and in P and Q, where it represents standard deviation. See Supplemental Table S3 for further information on the statistics.

Figure 4.

K5CreER^T2 targets multipotent BCs that give rise to LCs in the prostate. (A) Scheme summarizing the genetic strategy used to induce Confetti expression in K5-expressing cells during prostate postnatal development. (B) Scheme summarizing the protocol used to study the fate of cells targeted during postnatal development using K5CreER^T2/Rosa-Confetti mice. (C) Distribution of basal K14⁺ and luminal K8⁺ cells among Confetti⁺ cells 10 d after TAM (0.1 mg) administration at birth to K5CreER^T2/Rosa-Confetti mice (1268 cells out of five mice). (D–F) Confocal imaging of immunostaining of K5 and fluorescent Confetti cells 10 d after TAM (0.1 mg) administration at birth to K5CreER^T2/Rosa-Confetti mice. Isolated BCs (D) and LCs (E) as well as UPs containing BCs and LCs (F) were observed. In D, the section depicted represents the surface of the duct, meaning that the cells expressing K5 are located at the surface of the duct and not inside. This is illustrated by the red and green lines in the orthogonal sections of the WM. (G) Frequency of Confetti⁺ patch compositions 10 d after TAM (0.1 mg) injection at birth to K5CreER^T2/Rosa-Confetti mice. (H) Fraction of UPs among all pairs 10 d after TAM (0.1 mg) administration at birth to K5CreER^T2/Rosa-Confetti mice. Experimental bars (gray) represent the number of UPs divided by the total number of pairs. Model bars (black) represent the prediction of the fraction of UPs among all pairs, which depends only on the relative frequencies of the Confetti colors. (I) Fraction of total pairs 10 d after TAM (0.1 mg) administration at birth to K5CreER^T2/Rosa-Confetti mice. Experimental bars (gray) represent the total number of pairs divided by the total number of induced BCs (approximated by the number of cohesive groups of BCs). Model bars (black) represent the prediction of the overall frequencies of pairs, which is based on the probability of random labeling of BCs, taking into account the architecture of the tissue (i.e., how many LCs touch a BC), the degree of chimerism, the CRE specificity, and the frequency of recombination of each Confetti color. (D,E) Orthogonal projections of 3D stacks. (F) Single-plane image from a 3D stack. Bars, 20 µm. Histograms represent the mean. Error bars represent standard deviation. See Supplemental Table S4 for further information on the statistics.

Figure 5.

BCs are a self-sustained unipotent lineage in the MG during puberty and adult remodeling. (A) Theoretical outcomes of tracing BCs at saturation: BCs can be either a self-sustained unipotent lineage (top panel) or a lineage comprising bipotent SCs that produce labeled LCs over time (bottom panel). (B) Scheme summarizing the genetic strategy used to target YFP expression in K14rtTA/TetO-Cre/Rosa-YFP mice. (C) Scheme summarizing the protocol used to study the fate of cells targeted during puberty using K14rtTA/TetO-Cre/Rosa-YFP mice. (D–G) Confocal imaging of immunostaining of K14 or K5 and YFP at low magnification (D), at the adult stage (E), during pregnancy (F), and during lactation (G) in K14rtTA/TetO-Cre/Rosa-YFP mice treated for 5 wk with DOX at puberty. (H,I) Percentage of YFP⁺ cells in basal K14⁺ (H) and distribution of basal K14⁺ and luminal K8⁺ cells among YFP⁺ cells (I) 3 d after induction, in adult virgin mice, during pregnancy, and during lactation in K14rtTA/TetO-Cre/Rosa-YFP mice treated for 5 wk with DOX at puberty (12,945, 11,346, 19,609, and 25,583 cells out of three mice per time point). (J–M) FACS analysis of CD24 and CD29 expression in Lin⁻ (J) and Lin⁻YFP⁺ (K–M) cells in adult virgin K14rtTA/TetO-Cre/Rosa-YFP mice treated for 5 wk with DOX at puberty. (N) Percentage of YFP⁺ cells in Lin⁻CD24⁺CD29^Hi and Lin⁻CD24⁺CD29^Lo populations 3 d after induction, at the adult stage, and during pregnancy in K14rtTA/TetO-Cre/Rosa-YFP mice treated for 5 wk with DOX at puberty. (D–G) Orthogonal projections of 3D stacks. Bars: D, 50 µm; all others, 20 µm. Error bars represent mean ± SEM. See Supplemental Table S5 for further information on the statistics.

Figure 6.

LCs are a self-sustained unipotent lineage in the MG during puberty and adult remodeling. (A) Theoretical outcomes of tracing LCs at saturation: LCs can be either a self-sustained unipotent lineage (top panel) or a lineage comprising luminal progenitors that are replaced over time by bipotent basal SCs (bottom panel). (B) Scheme summarizing the genetic strategy used to target Tomato expression in K8rtTA/TetO-Cre/Rosa-tdTomato mice. (C) Scheme summarizing the protocol used to study the fate of cells targeted during puberty using K8rtTA/TetO-Cre/Rosa-tdTomato mice. (D–F) Confocal imaging of immunostaining of K14 and E-cadherin and fluorescent Tomato cells at the end of treatment (D), at 20 wk after induction (E), and during pregnancy (F) in K8rtTA/TetO-Cre/Rosa-tdTomato mice pulsed for 5 wk with DOX at puberty. (G) FACS analysis of CD24 and CD29 expression in Lin⁻Tomato⁺ cells at the end of treatment in K8rtTA/TetO-Cre/Rosa-tdTomato mice treated for 5 wk with DOX at puberty. (H) Percentage of Tomato⁺ cells in Lin⁻CD24⁺CD29^Hi and Lin⁻CD24⁺CD29^Lo populations at the end of treatment, at 20 wk after induction, and during pregnancy in K8rtTA/TetO-Cre/Rosa-tdTomato mice pulsed for 5 wk with DOX at puberty. (D–F) Orthogonal projections of 3D stacks. Bars, 20 µm. Error bars represent mean ± SEM.

Figure 7.

BCs contain multipotent SCs contributing to luminal expansion in the prostate. (A) Scheme summarizing the genetic strategy used to target YFP expression in K14rtTA/TetO-Cre/Rosa-YFP mice. (B) Scheme summarizing the protocol used to study the fate of cells targeted during prostate postnatal development using K14rtTA/TetO-Cre/Rosa-YFP mice. (C,D) Confocal imaging of immunostaining of K14 and YFP in the tip region at the end of DOX treatment (C) and 2 wk after induction (D) in 10-d-old K14rtTA/TetO-Cre/Rosa-YFP mice pulsed for 5 d with DOX. (E) Percentage of YFP⁺ cells in basal K14⁺ and luminal K8⁺ cells at the end of treatment and 2 wk after induction in the tip region in 10-d-old K14rtTA/TetO-Cre/Rosa-YFP mice pulsed for 5 d with DOX (40,925 and 43,395 cells out of four and three mice, respectively). (F) Scheme summarizing the genetic strategy used to target Tomato expression in K8rtTA/TetO-Cre/Rosa-tdTomato mice. (G) Scheme summarizing the protocol used to study the fate of cells targeted during prostate postnatal development using K8rtTA/TetO-Cre/Rosa-tdTomato mice. (H–J) Confocal imaging of immunostaining of K14 and fluorescent Tomato cells at the end of treatment (H), 2 wk after induction (I), and 4 wk after induction (J) in 10-d-old K8rtTA/TetO-Cre/Rosa-tdTomato mice pulsed for 7 d with DOX. (K) Percentage of Tomato⁺ cells in basal K14⁺ and luminal K8⁺ cells at the end of treatment and 2 wk after induction in the tip region in 10-d-old K8rtTA/TetO-Cre/Rosa-tdTomato mice pulsed for 7 d with DOX (26,687, 46,015, and 24,949 cells out of three, three, and two mice, respectively). (C,D,H,I,J) Orthogonal projections of 3D stacks. P-values in E and K represent a Student's t-test (paired, equal variances). Bars, 20 µm. Error bars indicate mean ± SEM. See Supplemental Table S6 for further information on statistics.