Tissue Myeloid Progenitors Differentiate into Pericytes through TGF-β Signaling in Developing Skin Vasculature

Abstract

Mural cells (pericytes and vascular smooth muscle cells) are essential for the regulation of vascular networks and maintenance of vascular integrity, but their origins are diverse in different tissues and not known in the organs that arise from the ectoderm, such as skin. Here, we show that tissue-localized myeloid progenitors contribute to pericyte development in embryonic skin vasculature. A series of in vivo fate-mapping experiments indicates that tissue myeloid progenitors differentiate into pericytes. Furthermore, depletion of tissue myeloid cells and their progenitors in PU.1 (also known as Spi1) mutants results in defective pericyte development. Fluorescence-activated cell sorting (FACS)-isolated myeloid cells and their progenitors from embryonic skin differentiate into pericytes in culture. At the molecular level, transforming growth factor-β (TGF-β) induces pericyte differentiation in culture. Furthermore, type 2 TGF-β receptor (Tgfbr2) mutants exhibit deficient pericyte development in skin vasculature. Combined, these data suggest that pericytes differentiate from tissue myeloid progenitors in the skin vasculature through TGF-β signaling.

Keywords: TGF-β; brain; capillary blood vessel; fate mapping; mural cell; myeloid; pericyte; skin; tissue macrophage; vascular development.

Figure 1. Distribution of pericytes and vascular smooth muscle cells in the developing skin vasculature

(A) Diagram of experimental procedure. The rostral back skin was dissected from mouse embryo and analyzed by whole-mount immunofluorescence confocal microscopy. (B–F) Whole-mount triple immunofluorescence confocal microscopy was performed with antibodies to a pericyte marker NG2 (B–F, green) and a vSMC marker αSMA (B–F, red) together with a pan-endothelial marker anti-PECAM-1 (B–F, blue). Close-up images (D and F) show the dotted boxed regions in C and E. Distal region of the vascular branching network appears to be covered by only pericytes, whereas both pericytes and vSMCs densely cover proximal medium-large diameter branched vessels (B–F). Note that open arrowheads indicate residual skeletal muscle fibers and arrowheads indicate hair follicles (B). (G and H) Whole-mount double labeling with antibodies to NG2 (green) and PECAM-1 (blue) at E12.5 (G) and E15.5 (H) reveals that a dynamic morphological change in pericytes during vascular development (arrows). Scale bars are 50 μm.

Figure 2. Contribution of hematopoietic cells to pericyte development in the embryonic skin

(A–H) Whole-mount triple immunofluorescence confocal microscopy of back skin from E15.5 Wnt1-Cre;R26R^EYFP (A and B), E15.5 Tie2-Cre;R26R^EYFP (C and D), E16.5 Cdh5-BAC-Cre^ERT2;R26R^EYFP injected with tamoxifen (Tam) at E11.5, E12.5 and E13.5 (E and F), or E15.5 Vav-iCre;R26R^EYFP (G–I) embryos was performed with antibodies to NG2 (A, C, E, G and H, red), PECAM-1 (A, C, E, G, H, and I, blue; D and F, red), a pan-neuronal marker Tuj1 (B, red), or a myeloid marker F4/80 (I, red), together with anti-EYFP (A–I, green). Cre-dependent EYFP reporter activity was detected in Tuj1⁺ peripheral nerves of Wnt1-Cre;R26R^EYFP skin (B, arrowheads), in PECAM-1⁺ endothelial cells of Tie2-Cre;R26R^EYFP and Cdh5-BAC-Cre^ERT2;R26R^EYFP skin (C–F, arrowheads), or in F4/80⁺ tissue-resident myeloid cells of Vav-iCre;R26R^EYFP skin (G and I, arrowheads). A significant population of EYFP⁺/NG2⁺ pericytes was found in Vav-iCre;R26R^EYFP skin (G and H, open arrows), whereas EYFP expression was scarcely detected in NG2⁺ pericytes in Wnt1-Cre;R26R^EYFP (A, open arrowheads), Tie2-Cre;R26R^EYFP, and Cdh5-BAC-Cre^ERT2;R26R^EYFP (C and E, open arrowheads) skin. Scale bars are 50 μm. (J) Quantification of EYFP⁺ cells in NG2⁺ pericytes in the skin vasculature. Each image used for counting has one branched vessel covered with NG2⁺ pericytes. EYFP⁺NG2⁺ pericytes were confirmed by their elongated morphology, with cytoplasmic processes wrapped around blood vessels (n=3 per genotype; Error bars represent mean ± SEM). (K) FACS analysis of YFP expression on CD45⁻PDGFRβ⁺ pericytes in back skin of E15.5 Wnt1-Cre;R26R^EYFP (top panel), E15.5 Tie2-Cre;R26R^EYFP (second panel), E13.5 Cdh5-BAC-Cre^ERT2;R26R^EYFP injected with Tam at E9.5, E10.5 and E11.5 (third panel), and E15.5 Vav-iCre;R26R^EYFP embryos (bottom panel). FACS data from one representative experiment along with the percentage of EYFP⁺ cells in CD45⁻PDGFRβ⁺ pericytes are shown.

Figure 3. Contribution of myeloid progenitors to pericyte development

(A and B) Whole-mount triple immunofluorescence confocal microscopy of back skin from E16.5 Vav-Cre^ER;R26R^EYFP embryos injected with tamoxifen (Tam) at E8.5, E9.5 and E10.5 was performed with antibodies to EYFP (green), NG2 (red), and PECAM-1 (blue). In addition to EYFP⁺ tissue-localized myeloid cells (arrowheads), a substantial population of EYFP⁺/NG2⁺ pericytes was found in Vav-Cre^ER mice;R26R^EYFP skin (open arrows). (C) Quantification of EYFP⁺/NG2⁺ pericytes in Vav-Cre^ER;R26R^EYFP skin (n=4 per genotype) and TdTomato⁺/NG2⁺ pericytes in CD11b-Cre;Ai14 skin (n=3 per genotype; Error bars represent mean ± SEM). Each image used for counting has one branched vessel covered with NG2⁺ pericytes: 14 EYFP⁺NG2⁺ pericytes in 104 NG2⁺ pericytes total were counted in Vav-Cre^ER;R26R^EYFP; 13 TdTomato⁺NG2⁺ pericytes in 136 NG2⁺ pericytes total were counted in CD11b-Cre;Ai14. EYFP⁺NG2⁺ and TdTomato⁺NG2⁺ pericytes were confirmed by their elongated morphology with cytoplasmic processes wrapped around blood vessels. (D–F) Whole-mount triple labeling with antibodies to F4/80 (D and F, red), CD11b (E, red; F, green), or NG2 (D and E, green), together with PECAM-1 (blue), reveals that F4/80⁺ (D) or CD11b⁺ (E) myeloid cells are distributed in close proximity to vessel-associated NG2⁺ pericytes (D and E, arrowheads) and a subset of F4/80⁺ myeloid cells is also positive for CD11b (F, arrowheads). (G–I) Whole-mount triple labeling of back skin from E15.5 CD11b-Cre;Ai14 embryos with antibodies to TdTomato (G–I, red), NG2 (G and H, green) or F4/80 (I, green), together with PECAM-1 (blue) shows a population of TdTomato⁺/NG2⁺ pericytes in CD11b-Cre;Ai14 skin (G and H, open arrows) in addition to TdTomato⁺/F4/80⁺ myeloid cells (I, arrowheads). Scale bars are 50 μm.

Figure 4. Differentiation of FACS-isolated myeloid progenitors into pericytes in culture

(A–E) FACS-isolated CD45⁺F4/80⁺PDGFRβ⁻ cells from E14.5–16.5 skin were clonally cultured in 96 well plates for 5 days in a 15% FBS-containing medium supplemented with 100 ng/ml M-CSF. Double immunofluorescence microscopy was performed with antibodies to NG2 (A–D, red) and F4/80 (A′–D′, green). The same fields from representative colonies in three independent experiments are shown in each panel: A and A′, colonies including only NG2⁻F4/80⁺ cells; B and B′, colonies including mixture of NG2^weakF4/80⁺ (arrowheads) and NG2⁺F4/80⁺ cells; C and C′, colonies including only NG2⁺F4/80⁺ cells; D and D′, colonies including only NG2⁺F4/80⁻ cells. (E) The proportion of cells in each colony. (F–I) Triple immunofluorescence confocal microscopy of the cultured CD45⁺F4/80⁺PDGFRβ⁻ cells with or without M-CSF was performed with antibodies to αSMA (F, red), PDGFRβ (F, green), NG2 (G and H, red) or F4/80 (G and H, green), together with a nuclear marker TO-PRO-3 (F–H, blue). M-CSF is required for F4/80⁺ macrophage survival but not for NG2⁺ pericyte differentiation (I). Scale bars are 50 μm. Error bars indicate mean ± SD. **, p<0.01, Student’s t-test. (J) qRT-PCR analysis of NG2 and F4/80 expression in the cultured CD45⁺F4/80⁺PDGFRβ⁻ cells with or without M-CSF. (K–T) qRT-PCR analysis of the expression of pericyte markers and myeloid markers in the CD45⁺F4/80⁺PDGFRβ⁻ cells (day 0 versus day 5) reveals that the 5-day culture induces pericyte marker expression, accompanied by a striking reduction of myeloid marker expression.

Figure 5. Genetic ablation of myeloid lineage causes defective pericyte development in the skin

(A–F) Whole-mount immunofluorescence confocal microscopy was performed with antibodies to F4/80 (A and B, green), αSMA (C and D, red), and PECAM-1 (C, D, blue) in E15.5 PU.1^−/− and control littermate skin (A–D). F4/80⁺ myeloid cell was not detectable in the mutants (A versus B), although vascular network formation appeared normal in the mutants, with remodeled vessels (C versus D), vSMC coverage (C versus D, red), and capillary network (C versus D, blue). The Quantification of PECAM-1⁺ vascular area and αSMA⁺ vSMC coverage is shown in E and F, respectively. (G–L) Whole-mount double immunofluorescence confocal microscopy with antibodies to NG2 (G–J, green), PDGFRβ (K and L, green) and PECAM-1 (G–L, blue) reveals that a drastic reduction in the number of NG2⁺ (G–J) and PDGFRβ⁺ (K and L) pericytes in E17.5 and E15.5 PU.1^−/−, respectively. The reduction of NG2⁺ and PDGFRβ⁺ pericytes is apparent in the vascular networks with small-diameter vessels (G versus H; G′ versus H′) as well as large-diameter vessels (I versus J; I′ versus J′; K versus L; K′ versus L′). Scale bars are 50 μm. (M and N) Quantification of NG2⁺ pericyte coverage distribution in back skin (M), forelimb and hindlimb skin (N) is shown (n=3). Error bars represent mean ± SEM *, p<0.05, Student’s t-test.

Figure 6. Contribution of myeloid progenitors to pericyte development in the embryonic midbrain, but not in the embryonic heart and liver

(A and B) Triple immunofluorescence confocal microscopy was performed with antibodies to PDGFRβ (green), NG2 (red), and PECAM-1 (blue) in E15.5 PU.1^−/− and control littermate midbrains. A significant reduction in the number of pericytes was observed in the mutants (B), compared to control littermate (A, arrowheads). (C and D) Triple labeling was performed with antibodies to β-gal (C, red) or TdTomato (D, red), together with anti-NG2 (green) and PECAM-1 (blue) in E15.5 Vav-iCre;R26R and CD11b-Cre;Ai14 midbrains. A significant number of β-gal⁺ or TdTomato⁺ pericytes (open arrows) was found in Vav-iCre;R26R or CD11b-Cre;Ai14 midbrain, respectively. (E–H) Triple labeling with antibodies to PDGFRβ (green), NG2 (red), and PECAM-1 (blue) in E15.5 PU.1^−/− and control littermate heart and liver reveals that no significant defect in pericyte coverage was observed in the mutants. Note that PDGFRβ expression was restricted to pericytes in the heart and liver, while NG2 expression was not restricted to pericytes in the heart. (I–N) Double labeling with antibodies to F4/80 (green) and PECAM-1 (red) in E15.5 PU.1^−/− and control littermate reveals that no F4/80⁺ myeloid cell was detectable in the mutant heart and liver. Close-up images (K and L) show the dotted boxed regions in I and J. (O and P) Triple labeling with antibodies to β-gal (green), PDGFRβ (red), and PECAM-1 (blue) in E15.5 WT1-Cre;R26R heart and liver reveals that co-localization of β-gal and PDGFRβ was found (open arrows), suggesting that mesothelium contributes to pericyte development. Scale bars are 50 μm.

Figure 7. TGF-β induces pericyte differentiation from myeloid progenitors

(A–I) FACS-isolated CD45⁺F4/80⁺PDGFRβ⁻ cells were cultured for 5 days in a 15% FBS-containing medium supplemented with 100 ng/ml M-CSF in combination with DMSO or a TGF-βR1 inhibitor, LY364947 (A–E). Similarly, the cells were cultured with or without 3 ng/ml TGF-β1 in a 0.2% FBS-containing medium supplemented with 100 ng/ml M-CSF (F–I). Triple immunofluorescence confocal microscopy was performed with antibodies to F4/80 (green) and NG2 (red), together with TO-PRO-3 (blue). The proportion of the NG2⁺F4/80⁻ (pericytes, red), NG2⁺F4/80⁺ (green), and NG2⁻F4/80⁺ (macrophages, blue) was quantified (E and I), under the culture conditions with a slight increase or decrease in total cell number in response to the LY or TGF-β1 treatment, respectively (D and H). One representative data from three independent experiments are shown. (J–L) Triple labeling of back skin from E15.5 Vav-iCre;Tgfbr2^flox/flox and control littermate embryos with antibodies to NG2 (green), αSMA (red) and PECAM-1 (blue) reveals a striking reduction of pericyte coverage in the mutants. Quantification of NG2⁺ pericyte coverage is shown in L (n=4). (M–O) Double labeling of back skin with antibodies to F4/80 (red) and PECAM-1 (blue) reveals no significant change in the number of F4/80⁺ myeloid cells in the mutant skin at E15.5 (n=3). Error bars represent mean ± SEM; **, p<0.01, Student’s t-test. Scale bars are 50 μm.