Distinct temporal requirements for Runx1 in hematopoietic progenitors and stem cells

Abstract

The transcription factor Runx1 is essential for the formation of yolk sac-derived erythroid/myeloid progenitors (EMPs) and hematopoietic stem cells (HSCs) from hemogenic endothelium during embryogenesis. However, long-term repopulating HSCs (LT-HSCs) persist when Runx1 is conditionally deleted in fetal liver cells, demonstrating that the requirement for Runx1 changes over time. To define more precisely when Runx1 transitions from an essential factor to a homeostatic regulator of EMPs and HSCs, and whether that transition requires fetal liver colonization, we performed conditional, timed deletions of Runx1 between E7.5 and E13.5. We determined that Runx1 loss reduces the formation or function of EMPs up through E10.5. The Runx1 requirement in HSCs ends later, as deletion up to E11.5 eliminates HSCs. At E11.5, there is an abrupt transition to Runx1 independence in at least a subset of HSCs that does not require fetal liver colonization. The transition to Runx1 independence in EMPs is not mediated by other core binding factors (Runx2 and/or Runx3); however, deleting the common non-DNA-binding β subunit (CBFβ) severely compromises LT-HSC function. Hence, the requirements for Runx1 in EMP and HSC formation are temporally distinct, and LT-HSC function is highly reliant on continued core binding factor activity.

Keywords: Embryo; Hematopoiesis; Hemogenic endothelium; Mouse; Runx1; Stem cells.

Fig. 1.

Experimental strategy for deleting Runx1. (A) Runx1 was deleted by culturing whole embryos ex vivo for 24 hours with 4-OHT. Embryos were staged prior to and at the end of the culture period. Hematopoietic tissues were washed, dissociated and plated in colony assays. Multiple individual colonies were analyzed by PCR for deletion of Runx1^f alleles. f/+ and Δ/+ represent colonies from a Runx1^f/+; Cre^ERT conceptus in which the Runx1^f allele was not deleted, or was deleted, respectively. f/f, f/Δ and Δ/Δ represent colonies from Runx1^f/f; Cre^ERT conceptuses in which no, one, or both Runx1^f alleles were deleted, respectively. (B) All deletions initiated at E9.5 onwards were performed by injecting pregnant dams with tamoxifen. Hematopoiesis was analyzed as shown in A. (C) The impact of Cre activation on progenitors following exposure to 4-OHT ex vivo in whole embryo cultures. Deletion was performed between E8.5 and E9.5. Genotypes of conceptuses are Runx1^f/+ (f/+) and Runx1^+/+ (+/+) (n=21), or Runx1^f/+; Actb-Cre^ERT (f/+; Cre) and Runx1^+/+; Actb-Cre^ERT (+/+; Cre) (n=8). (D) Progenitor numbers following tamoxifen injection of pregnant dams. Deletions were performed at E9.5 and embryos harvested at E10.5. All deletions were carried out with Actb-Cre^ERT. +/+, f/+ and f/f, n=10 in yolk sac and n=8 in AGM+U+V. +/+; Cre and f/+; Cre, n=5 in yolk sac and n=6 in AGM+U+V. In C and D, error bars indicate s.e.m. (E) Whole-mount immunofluorescence of E10.5 yolk sacs harvested from dams injected with tamoxifen at E9.5, using antibodies recognizing endogenous Runx1 and Kit (maximum of 20 3-μm z-sections). Panels on the right are examples of Kit⁺ cells (green) with Runx1⁺ nuclei (blue). Arrows indicate a Kit⁺ cell with low Runx1 nuclear staining. Nuclear fluorescence intensity of Runx1 in individual Kit⁺ cells was quantified and plotted in panel F. (F) The average corrected nuclear fluorescence of Runx1 in 100 Kit⁺ cells.

Fig. 2.

Deletion of Runx1 decreases EMPs in the yolk sac, but not CFU-Cs in the fetal liver. (A) Yolk sac EMPs, AGM+U+V CFU-Cs and fetal liver CFU-Cs [mean±s.e.m. per embryonic equivalent (ee)] following deletion of Runx1 with Actb-Cre^ERT. Embryonic ages in somite pairs (sp) and embryonic days (E) on the x-axis represent the stage of the embryo when harvested for methylcellulose colony assays, with the deletion initiated 24 hours earlier. Black asterisks indicate significant differences (P≤0.05) between colony numbers in f/f; Cre versus f/f and f/+ embryos. Orange asterisks indicate significant differences between f/f; Cre and f/+; Cre embryos. The difference between f/f; Cre and f/+; Cre at 18-25 sp (E9.0) in yolk sac is statistically significant (P=0.06). (B) Deletion frequency in Runx1^f/+; Actb-Cre^ERT yolk sac, AGM+U+V, and fetal liver EMPs/CFU-Cs represented as percentages of colonies (mean±s.e.m.) that had no deletion (f/+) or deletion (Δ/+) of the floxed allele. The average deletion efficiency in combined experiments was 64.7% (±4.04). Data from embryos in which the average deletion frequency was <30% were excluded from Fig. 2A. (C) Deletion frequency in Runx1^f/f; Actb-Cre^ERT progenitors represented as percentages of colonies (average±s.e.m.) that had deleted both (Δ/Δ), one (Δ/f) or no Runx1^f alleles (f/f). The average deletion efficiency was 61.4% (±19.1). The number of sample replicates and colonies are shown in supplementary material Tables S1, S2. All data were analyzed using Student’s unpaired t-test and are shown as mean±s.e.m.

Fig. 3.

Runx1 is required in VEC-Cre^ERT-expressing cells at all times. (A) Number of CFU-Cs (mean±s.e.m.) in fetal livers of various Runx1 genotypes with and without VEC-Cre^ERT, following injection of tamoxifen into pregnant dams at E10.0 and isolation of fetuses at E11.0. (B) Deletion frequency in fetal liver CFU-Cs by VEC-Cre^ERT represented as percentages of colonies (mean±s.e.m.) with no (f/+ or f/f), one (Δ/+ or Δ/f) or two (Δ/Δ) deleted Runx1 alleles. (C) Expression of VEC (Cdh5) mRNA in E10.5 AGM+U+V from sorted Runx1 (GFP⁺) VEC⁺ CD45⁺ hematopoietic cells relative to Runx1 (GFP⁺) VEC⁺ CD45^- endothelial cells (represented as fold change). (D) Number of yolk EMPs (mean±s.e.m.) following deletion of Runx1 with VEC-Cre^ERT. All deletions were performed by tamoxifen injection of pregnant dams. (E) Deletion frequency in Runx1^f/+; VEC-Cre^ERT yolk sac EMPs (mean±s.e.m.). (F) Deletion frequency in Runx1^f/f; VEC-Cre^ERT yolk sac EMPs. (G) Number of AGM+U+V CFU-Cs (mean±s.e.m.) following deletion of Runx1 with VEC-Cre^ERT. (H) Deletion frequency in Runx1^f/+; VEC-Cre^ERT AGM+U+V CFU-Cs represented as percentages of colonies (mean±s.e.m.) that had not deleted (f/+) or deleted (Δ/+) the floxed Runx1 allele. (I) Deletion frequencies in Runx1^f/f; VEC-Cre^ERT AGM+U+V CFU-Cs. The number of sample replicates and colonies assayed are shown in supplementary material Tables S3, S4. Student’s unpaired t-tests were performed on data shown in A,B,D-I.

Fig. 4.

Runx1 is essential for pre-HSC/HSC formation in the AGM prior to E11.5. (A) Experimental strategy for deletion in pre-HSCs/HSCs. Pregnant dams were injected with tamoxifen at E9.5 or E10.5. AGM+U+V were harvested at E11.5 and explanted in the absence of tamoxifen for 3 days prior to transplantation into irradiated hosts. Recipients were analyzed for engraftment, then donor-derived (CD45.2⁺) cells were FACS sorted 16-weeks post-transplantation and directly analyzed by PCR (splenic B cells and thymic T cells) or cultured in methylcellulose (bone marrow) and colonies were assayed by PCR. (B) Percentage of transplant recipients reconstituted with >1% donor Mac-1⁺ bone marrow cells at 16 weeks. Pregnant dams were injected with tamoxifen at E9.5 (i9.5), and AGM+U+V harvested at E11.5. The number of animals transplanted is indicated above the bars. The percentage contribution at the lowest dose at which all recipients were reconstituted [0.3 embryo equivalents (ee)] by Runx1^f/+; Actb-Cre^ERT HSCs ranged from 96.4% to 99.6% (mean=97.8%), and by Runx1^f/f; Actb-Cre^ERT HSCs from 87.6% to 99.0% (mean=94.2%). All recipients reconstituted with Mac-1⁺ cells contained donor-derived B and T cells. (C) Percentage of deleted Runx1 alleles from sorted donor-derived CD45.2⁺ cells (injections at E9.5), from transplants shown in B (Actb-Cre^ERT). Deletion data in the spleen and thymus represent intensity of PCR products from 40,000 cells. Deletion data in individual CD45.2 bone marrow colonies (BM CFU-C) were averaged from all recipients (±s.e.m.). Runx1^f/+; Actb-Cre^ERT=266 colonies and Runx1^f/f; Actb-Cre^ERT=340 colonies. (D) Percentage of transplant recipients reconstituted with >1% donor Mac-1⁺ bone marrow cells at 16 weeks when tamoxifen was injected into pregnant dams at E10.5, analyzed as in B. The percentage contribution at 0.3 embryo equivalents by Runx1^f/+; Actb-Cre^ERT HSCs ranged from 84.2% to 99.8% (mean=95.9%), and by Runx1^f/f; Actb-Cre^ERT HSCs from 24.2% to 79.6% (mean=56.5%). All recipients reconstituted with Mac-1⁺ cells contained donor-derived B and T cells. (E) Percentage of deleted Runx1 alleles from sorted donor-derived CD45.2⁺ cells for injections carried out at E10.5, analyzed as in C. The number of colonies analyzed by PCR was 73 for donor-derived (CD45.2) Runx1^f/+; Actb-Cre^ERT BM CFU-Cs and 168 for Runx1^f/f; Actb-Cre^ERT BM CFU-Cs. (F) Percentage of transplant recipients reconstituted with >1% donor Mac-1⁺ bone marrow cells at 16 weeks when tamoxifen was injected into pregnant dams at E10.5 (deletion with VEC-Cre^ERT). The number of animals transplanted is indicated above the bars. All recipients were transplanted with 0.16 ee of donor cells. The percentage contribution by Runx1^f/+; VEC-Cre^ERT HSCs ranged from 85% to 92.9% (mean=89.4%). (G) Percentage of deleted Runx1 alleles from sorted donor-derived CD45.2⁺ cells for injections carried out at E10.5, analyzed as in C and E. Eighty-six colonies were analyzed by PCR for Runx1^f/+; VEC-Cre^ERT donor-derived BM CFU-Cs. NA, no colonies analyzed (because Runx1^f/f; VEC-Cre^ERT cells did not reconstitute).

Fig. 5.

Deletion of Runx1 reduces the number of hematopoietic clusters in the dorsal aorta. (A) Whole-mount confocal microscopy of Runx1⁺ CD31⁺ endothelial cells (examples with arrows) and Runx1⁺ CD31⁺ Kit⁺ hematopoietic cluster cells (examples with asterisks) in the dorsal aorta. Image is a sum of three 2-μm z-sections. (B) Fraction of Runx1⁺ cells (average±s.e.m.) distributed between the endothelium (CD31⁺ Kit^-) and hematopoietic clusters (CD31⁺ Kit⁺) of the dorsal aorta at three stages (two embryos per stage). (C) Whole-mount immunofluorescence of E10.5 embryos following tamoxifen injection of pregnant dams at E9.5 (iE9.5). Each image is a 365-μm deep z-stack composite with the dorsal aorta outlined in yellow. (D) Average number (±s.d.) (n=2) of Runx1⁺ cells per somite pair in the dorsal aorta.

Fig. 6.

Runx1 is required for HSC maturation prior to E11.5 in the AGM+U+V. (A) Schematic for deletion of Runx1 ex vivo. E11.5 AGM+U+V (CD45.2⁺) were explanted in 4-OHT for three days, dissociated, and transplanted into irradiated CD45.1⁺ hosts with 250,000 CD45.1/CD45.2 spleen cells. Recipients were analyzed 16 weeks post-transplantation, and CD45.2⁺ donor cells (mean±s.e.m.) sorted from bone marrow and thymus of reconstituted recipients and analyzed for Runx1 deletion by PCR, as described in Fig. 4. (B) Percentage of donor-derived cells in the bone marrow 16 weeks post-transplantation following deletion with Actb-Cre^ERT. Contribution to peripheral blood was similar to bone marrow (not shown). Embryonic equivalents of 1.0 and 1.6 were transplanted. (C) PCR analysis of Runx1 alleles from Actb-Cre^ERT donor-derived colonies picked from methylcellulose cultures. Number of colonies assayed=31 for f/+; Actb-Cre^ERT and 137 for f/f; Actb-Cre^ERT. Each bar represents colonies from an independent recipient mouse. (D) Percentage of donor-derived cells (mean±s.e.m.) in the bone marrow 16 weeks post-transplantation following deletion with VEC-Cre^ERT. Contribution to peripheral blood was similar to bone marrow (not shown). Embryonic equivalent of 0.2 was transplanted. (E) PCR analysis of Runx1 alleles from VEC-Cre^ERT deleted donor-derived colonies picked from methylcellulose cultures from individual recipients. Number of colonies assayed=106 for f/+; VEC-Cre^ERT and 60 for f/f; VEC-Cre^ERT. Each bar represents colonies from an independent recipient mouse.

Fig. 7.

Runx2 and Runx3 do not compensate for Runx1 loss in fetal liver CFU-Cs. (A) CFU-Cs (±s.e.m.) per 20,000 E14.5 fetal liver cells from Cbfb^f/f and Cbfb^f/+ (n=6), Cbfb^f/+; Vav1-Cre (n=4), and Cbfb^f/f; Vav1-Cre fetuses (n=7). Significance was determined by one-way ANOVA. Asterisk indicates a significant difference compared with controls (#) determined by Dunnett’s Multiple Comparison test. (B) Representative flow cytometry profiles of LSK and other phenotypic HSC populations using SLAM markers (CD48, CD150) in E14.5 fetal livers. (C) Percentage (mean±s.e.m.) of phenotypic LT-HSCs in E14.5 fetal livers. Cbfb^f/+, Cbfb^f/f, n=5. Cbfb^f/+; Vav1-Cre, n=4. Cbfb^f/f; Vav1-Cre, n=4. (D) Competitive transplant of various numbers of Cbfb^f/+ and Cbfb^f/f; Vav1-Cre E14.5 fetal liver cells. Plotted are the percentages (mean ±s.e.m.) of CD45.2 donor-derived Mac-1⁺ cells in the peripheral blood at 3, 7 and 20 weeks post-transplant. Each dot represents an individual transplant recipient. CD45.1/CD45.2 bone marrow cells (2×10⁵) were used as competitors. (E) Noncompetitive transplant of 2×10⁶ E14.5 fetal liver cells. Contribution to Mac-1⁺ cells in peripheral blood is plotted (mean±s.e.m.). (F) Contribution of donor-derived cells to various populations in the bone marrow of recipients transplanted with 2×10⁶ E14.5 fetal liver cells (mean±s.e.m.). The contribution of Cbfb^f/f; Vav1-Cre cells to all populations was significantly lower than that of Cbfb^f/f cells, with the exception of short-term (ST)-HSCs (CD34⁺ Flt3^- LSK).