Impaired angiogenesis and altered Notch signaling in mice overexpressing endothelial Egfl7

Abstract

Epidermal growth factor-like domain 7 (Egfl7) is important for regulating tubulogenesis in zebrafish, but its role in mammals remains unresolved. We show here that endothelial overexpression of Egfl7 in transgenic mice leads to partial lethality, hemorrhaging, and altered cardiac morphogenesis. These defects are accompanied by abnormal vascular patterning and remodeling in both the embryonic and postnatal vasculature. Egfl7 overexpression in the neonatal retina results in a hyperangiogenic response, and EGFL7 knockdown in human primary endothelial cells suppresses endothelial cell proliferation, sprouting, and migration. These phenotypes are reminiscent of Notch inhibition. In addition, our results show that EGFL7 and endothelial-specific NOTCH physically interact in vivo and strongly suggest that Egfl7 antagonizes Notch in both the postnatal retina and in primary endothelial cells. Specifically, Egfl7 inhibits Notch reporter activity and down-regulates the level of Notch target genes when overexpressed. In conclusion, we have uncovered a critical role for Egfl7 in vascular development and have shown that some of these functions are mediated through modulation of Notch signaling.

Figure 1

Egfl7 mRNA and protein expression in Tie2-Egfl7 transgenic embryos. (A) Schematic organization of Tie2-Egfl7 transgene construct used to generate mice that overexpress Egfl7. (B) Average Egfl7, CD31, and miR126 expression in whole embryos at E12.5, as measured by quantitative RT-PCR. The levels of Egfl7 were normalized to endothelial cell number using CD31 expression. Values are represented as fold difference relative to wild-type: WT, n = 6; TG, n = 4. *P < .01. (C) Detection of EGFL7 and β-tubulin protein in cytosolic fraction isolated from wild-type and transgenic embryos at E12.5. Quantification of protein levels is shown in the graph, and values are made relative to wild-type expression level. White bar represents wild-type; and black bars, transgenic. A vertical line has been inserted to indicate a repositioned gel lane.

Figure 2

Overexpressing Egfl7 results in partial embryonic lethality and hemorrhaging. (A) The expected and observed number of embryos and pups obtained from Tie2-Egfl7 hemizygous intercrosses at E9.5, E10.5, E12.5, and P5. *P = .004. (B) Gross phenotype of wild-type (i-ii) and Tie2-Egfl7 transgenic embryos (iii-iv) at E10.5 and E12.5. Arrows indicate hemorrhaging in Tie2-Egfl7 mice at E12.5 (iv). Original magnification ×20 (E10.5), ×12 (E12.5). Images were captured on Discovery V20 Stereomicroscope (Carl Zeiss).

Figure 3

Embryonic Egfl7 overexpression causes defects in the developing heart and in head and yolk sac vasculature. (A) CD31 immunostaining of whole-mount wild-type (i) and transgenic (v) embryos at E10.5. White and red lines represent the atrium (A) and ventricle (V), respectively. Original magnification ×30. CD31 immunostaining of parasagittal sections of wild-type (ii-iv) and Tie2-Egfl7 transgenic embryos (vi-viii). High-magnification images of black-boxed areas in subpanels ii and vi (iii,vii). High-magnification images of boxed areas in subpanels iii and vii (iv,viii). Original magnification ×1.25 (ii,vi), ×4 (iii,vii), and ×20 (iv,viii). Black asterisks represent reduction in traberculation; and red asterisks, branchial arch. (B) CD31 immunostaining of whole-mount embryos at E10.5. Tie2-Egfl7 transgenic embryos (iii-iv) showed abnormal endothelial aggregates (arrowheads), a decrease in the number of major cranial vessels (red dots), and increased branching (white dots) compared with wild-type littermates (i-ii). Original magnification ×60. (C) CD31 immunostaining of cross sections from wild-type (i-iv) and Tie2-Egfl7 transgenic embryos (v-viii) at E10.5. High-magnification images of black-boxed areas in subpanels i and v (ii,vi). High-magnification images of the boxed areas in subpanels iii and vii (iv,viii). Arrows indicates carotid artery (CA); and arrowheads, hemorrhaging from CA. Original magnification ×4 (i,iii,v,vii) and ×10 (ii,iv,vi,viii). (D) CD31 immunostaining of E12.5 wild-type (i-ii) and transgenic (iii-iv) yolk sacs. Arrowheads indicate vitelline vessels; and arrow, knot-like structure in vein. Quantification of vitelline vessel diameter in wild-type (□, n = 2) and transgenic (■, n = 4) yolk sacs. (v) *P < .03. Original magnification ×25.5 (i,iii) and × 41 (ii,iv). Whole-mount images were captured on Discovery V20 Stereomicroscope (Carl Zeiss) and immunohistochemistry images on the Axioplan 2 Imaging Upright Microscope (Carl Zeiss).

Figure 4

Tie2-Egfl7 transgenic mice exhibit defects in vascular cell stratification at branch points of the dorsal aorta. Immunofluorescence staining of the dorsal aorta at E12.5. CD31⁺ cells are shown in green (Alexa Flour 488), and SMA⁺ cells are shown in red (Cy3). Images were captured on the TCS AOBS SP2 microscope (Leica). (A-B) Wild-type and transgenic dorsal aorta, respectively. Original magnification 20×/0.7NA, water objective. (C-D) High-magnification image of boxed areas in panels A and B. Original magnification 63×/1.2NA, water objective. (E) Summary of phenotypes observed in Egfl7 transgenic embryos.

Figure 5

Overexpressing Egfl7 in the postnatal retina results in arterial and venous defects and an increase in vascular coverage. (A) Quantitative RT-PCR analysis of Egfl7 and miR126 expression in P5 retinas isolated from wild-type (□, n = 4) and transgenic (■, n = 6) mice. Egfl7 expression is normalized to endothelial cell number using CD31 expression. *P < .03. (B) Wild-type (i-iii) and transgenic (iv-vi) retinal vasculature at P5, stained with fluorescein isothiocyanate–labeled BS-1 lectin. Red dots represent branching at the distal ends of the vessel; white dots, branching along the vessel length; arrowheads, tortuous veins; and arrows, venous knot-like structures. Original magnification 10×/0.4NA, water objective (i,iv), 20×/0.7NA, water objective (ii,v), and 63×/1.2NA, (iii,vi). (C) Vascular coverage in wild-type (i) and transgenic (ii) retinas at P5. Boxed regions correspond to areas where coverage was measured. (iii) Average vascular coverage at the peripheral and central plexus for wild-type (□, n = 8) and transgenic (■, n = 14) retinas. *P < .05. (iv) CD31 mRNA expression in wild-type (□, n = 4) and transgenic (■, n = 6) retinas. Filopodia in wild-type (v) and transgenic (vi) retinas. (vii) Average filopodia number per 100-μM vessel length for wild-type (□, n = 5) and transgenic (■, n = 9) retinas. *P < .05. Values for vascular coverage and filopodia number are represented as fold difference relative to wild-type. Original magnification 20× (i,ii) and 40×/1.3NA, water objective (v,vi). (D) Summary of phenotypes observed in Egfl7 transgenic retinal vasculature at P5. Images of the retinal vasculature were captured on the TCS AOBS SP2 microscope (Leica), and filopodia was imaged using the LSM 5Live DuoScan microscope (Carl Zeiss).

Figure 6

EGFL7 functions as an antagonist of Notch in HUVECs. (A) Human EGFL7 expression in lentivirally generated HUVEC lines, as measured by quantitative RT-PCR. Empty lentiviruses pCCL GFP and pLKO served as controls for EGFL7 overexpression and knockdown, respectively. Data are represented as fold induction relative to the lentiviral controls. (B) Proliferation assay of HUVEC lines grown in complete medium for 4 days. (C) Capillary sprouting assay. Control (GFP and pLKO), EGFL7 overexpressing (EGFL7), and knockdown (pLKO 61) HUVEC lines were coated on a cytodex bead, beads embedded in fibrin gels, and visualized on day 7. (D) HUVEC monolayer wounding assay at 0 and 24 hours. Dotted lines highlight the edges of the monolayer. (E) Quantitation of HUVEC migration in monolayer wounding assay, represented as percentage of area filled at 8 hours. (F) Transactivation of Notch/CSL-luciferase reporter in EGFL7 overexpressing and knockdown HUVEC lines. Data represented as relative luciferase units (RLU). (G) Notch4-dependent HUVEC morphogenesis assay. Control HUVECs (pCCL-GFP or pLKO) or EGFL7 overexpressing (EGFL7) or knockdown (pLKO61) cells were grown as monolayer on a fibrin gel. GFP- or Notch4/GFP-expressing HUVECs were overlaid on top of the monolayer. At day 7, cocultures were visualized and the number of GFP⁺ cells undergoing morphogenesis, as seen by the extending of processes into the surrounding matrix per field, was determined. Experiments were performed in triplicate, and error bars represent SD.

Figure 7

EGFL7 interacts with Notch receptors and regulates Notch target gene expression in vivo. (A) Alignment of the DSL domain of Jagged1, Serrate, Delta, and Lag-2 with the putative DSL domain in EGFL7. Red letters represent the consensus sequence. (B) Yeast-2-hybrid assay (left panel): EGFL7 interacts with NOTCH4 and DLL4. Full-length EGFL7, DLL4, or the extracellular domain of NOTCH4 were fused to either the DNA-binding domain or the transcriptional activation domain of GAL4, and protein-protein interactions were monitored by the ability of the transformed yeast to grow on defined medium, and expression of α-galactosidase. Yeast-3-hybrid assay (right panels): EGFL7 abolishes NOTCH4-DLL4 interaction. The Egfl7 ORF was cloned downstream of a methionine repressible promoter (Met25) and transformed into a yeast strain expressing Notch4-GAL activating domain and DLL4-GAL4 DNA-binding domain fusions. Expression of α-galactosidase was then assayed on X-gal selection plates with or without methionine. (Ci-ii) Coimmunoprecipitation assays with protein extracts prepared from HEK293 cells transfected with plasmids encoding MYC/His-tagged-Egfl7 and Notch4 ECD or Notch1 ECD. (i) A NOTCH4 antibody was used to immunoprecipitate NOTCH4 ECD, and protein complexes were probed for NOTCH4 and EGFL7 by Western blot. (ii) A MYC antibody was used to immunoprecipitate EGFL7, and protein complexes were probed for NOTCH1 and EGFL7 by Western blot. (iii) Coimmunoprecipitation assay with protein extracts prepared from HUVECs infected with an adenovirus encoding MYC-tagged-Egfl7. A MYC antibody was used to immunoprecipitate EGFL7, and protein complexes were probed for NOTCH1 and EGFL7 by Western blot. (iv) Coimmunoprecipitation assays using protein lysates prepared from E12.5 embryos. An antibody against NOTCH4 was used to immunoprecipitate NOTCH4, and protein complexes were probed for NOTCH4 and EGFL7 by Western blot. Vertical lines have been inserted to indicate a repositioned gel lane. (D) Notch target gene expression in wild-type (□, n = 4) and Tie2-Egfl7 transgenic (■, n = 6) retinas. Gene expression was measured by quantitative RT-PCR and normalized to endothelial cell number using CD31 expression. (E) Notch target gene expression in wild-type (white bars, n = 6) and Tie2-Egfl7 transgenic embryos (black bars, n = 4). Data are represented as fold change compared with wild-type. *P < .05.