β1 integrin expression on endothelial cells is required for angiogenesis but not for vasculogenesis

Authors

  • Harikrishna Tanjore,

    1. Division of Matrix Biology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Elisabeth M. Zeisberg,

    1. Division of Matrix Biology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Behzad Gerami-Naini,

    1. Division of Matrix Biology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Raghu Kalluri

    Corresponding author
    1. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, and Harvard-MIT Division of Health Sciences and Technology, Boston, Massachusetts
    • Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, RW514, Boston, MA 02215
    Search for more papers by this author

Abstract

Integrins are a family of cell adhesion receptors that are involved in cell–matrix and cell–cell communications. They facilitate cell proliferation, migration, and survival. Using the Cre-Lox system, we deleted β1 integrin on Tie2-positive (Tie2-cre β1 Int fl/fl) vascular endothelial cells. Deletion of β1 integrin on vascular endothelial cells results in embryonic lethality. Blood vessel defects are encountered in the Tie2-Cre β1 Int fl/fl embryos at embryonic age (E9.5), and embryos die before reaching E10.5. The embryos exhibit growth retardation and both histological evaluation and PECAM-1 staining of E9.5 embryos revealed defects in angiogenic sprouting and vascular branching morphogenesis. Large and medium-size vessel formation is not affected in these embryos. Angiogenic defects were observed in several regions of the embryo and yolk sacs. These results indicate that β1 integrin expression on vascular endothelial cells is crucial for embryonic angiogenesis but dispensable for vasculogenesis. Developmental Dynamics 237:75–82, 2008. © 2007 Wiley-Liss, Inc.

INTRODUCTION

Development of embryonic vasculature starts by the formation of new blood vessels in the yolk sac around day 6.5 of gestation (Risau,1997). This process is called vasculogenesis and occurs by differentiation of endothelial cells from precursor angioblasts in the blood island of the extra embryonic mesoderm (Risau and Flamme,1995). At embryonic day (E) E8.5, primitive vasculature undergoes extensive remodeling involving vessel expansion and regression, called angiogenesis (Risau,1997). Angiogenesis seems to be regulated in different ways than vasculogenesis. However, until now the exact mechanisms that regulate angiogenesis vs. vasculogenesis are not clear. Both vasculogenesis and angiogenesis mainly involve cross-talk between endothelial cells (influencing migration of endothelial cells) and other cell–cell and cell–matrix interactions.

Integrins are a family of heterodimeric transmembrane glycoproteins receptors consisting of α subunit and a β subunit that are involved in cell–cell and cell–matrix interactions. Several studies support the idea that β1 integrin play a role in angiogenesis (Mettouchi and Meneguzzi,2006). However, the study of β1 integrin in the formation and maintenance of vascular network has been confounded by the fact that systemic genetic ablation of β1 integrin leads to early embryonic lethality before day 5 of gestation (Fassler and Meyer,1995; Stephens et al.,1995). Here, we use Cre-Lox system to specifically delete β1 integrin in Tie2-positive endothelial cells. The tyrosine kinase receptor Tie2 is expressed on the first set of endothelial cells at the primitive streak stage. Its expression continues during the developmental stages of both vasculogenesis and angiogenesis and remains active in endothelial cells throughout adulthood (Sato et al.,1993; Schnurch and Risau,1993; Wong et al.,1997).

Endothelial integrins that contain a β1 subunit are α1β1, α2β1, α3β1, α4β1, α5β1, α6β1, and αvβ1 (Rupp and Little,2001; Stupack and Cheresh,2002). Therefore, endothelial deletion of β1 integrin in Tie2-Cre β1 Int fl/fl mice leads to the absence of all these integrins in endothelial cells, beginning as the first endothelial cells arise. Deletion of β1 integrin on vascular endothelial cells results in blood vessel sprouting defects at around gestation age of E9.5, and all embryos die before reaching E10.5. The formation of new vessels in the earlier stages of embryonic development, however, appears unaffected. These results suggest a crucial and specific role for endothelial β1 integrin during embryonic development.

RESULTS

Ablation of β1 Integrin in Tie2-Positive Endothelial Cells Leads to Embryonic Lethality Between E9.5 and E10.5

To delete β1 integrin specifically in vascular endothelial cells, we chose to cross β1 integrin floxed mice (β1 Int fl/fl) with Tie2-cre transgenic mice. In Tie2-cre transgenic mice, the expression of Cre recombinase is driven by an endothelial-specific Tie2 promoter/enhancer (Kisanuki et al.,2001). First, to confirm endothelial specific deletion by the Cre activity, we crossed Tie2-cre mice with R26RstopLacZ mice, in which expression of LacZ requires Cre-mediated excision of a stop cassette (Fig. 1A). Tie2Cre; R26Rstop LacZ double transgenic embryos were harvested at time points E9.5 and E11.5 and analyzed for LacZ expression. At both time points, LacZ expression was detected in most endothelial cells but not in any other cell types (Fig. 1B,C). These results indicate specific activity of Cre-recombinase in vascular endothelial cells by Tie2-cre transgenic mice.

Figure 1.

Endothelial cell deletion of β1 integrin. A: Cre-mediated excision in Tie2-cre/R26Rstop LacZ double transgenic mice. B,C: Embryos harvested at time points embryonic day (E) E9.5 and E11.5 from Tie2-cre/R26Rstop LacZ double transgenic mice and stained for LacZ expression. LacZ was detected in most endothelial cells (blue), eosin was used to counterstain (red). D: Tie2-Cre–mediated excision of exon 3 in β1 integrin floxed mice flanked by two loxP sites. E: Percentage of viable Tie2-Cre β1 Int fl/fl mutant embryos harvested at different time points from E8.5 to postnatal from crosses between β1 integrin fl/fl and Tie2-Cre β1 Int fl/wt mice. No viable Tie2-Cre β1 Int fl/fl mutant embryos were identified at E10.5 onward. F–H: Embryos analyzed at different gestational ages from E8.5 to E17.5 between β1 integrin fl/fl and Tie2-Cre β1 Int fl/wt mice. F: E8.5 embryos revealed no difference between β Int fl/wt (control) embryos and Tie2-Cre β1 Int fl/fl mutant embryos. G: E9.5 embryos appeared normal with the exception of Tie2-Cre β1 Int fl/fl embryos which were smaller in size, indicating growth retardation. H: Partially resorbed (dead) E10.5 Tie2-Cre β1 Int fl/fl mutant embryos compared with control embryos.

In β1 integrin floxed mice, exon 3 is flanked by loxP sites (Fig. 1D; Raghavan et al.,2000). Breeding between β1 Int fl/fl mice and Tie2-cre β1 Int fl/wt mice yielded healthy viable litters. Litter size ranged from 5–7 pups. Genotyping of a total of 125 mice from several different matings suggested that the mice were not born in the expected Mendelian ratio. Among 125 pups, 34% were Tie2-cre β1 Int fl/wt, 33% were Tie2-cre–negative β1 Int fl/wt, and 33% were Tie2-cre–negative β1 Int fl/fl. Viable Tie2-Cre β1 Int fl/fl mutants mice, which would be expected at a Mendelian ratio of 25% were not present (Fig. 1E; Table 1). These results suggested that deletion of the β1 integrin in Tie2-positive endothelial cells is embryonic lethal.

Table 1. Embryos at Different Gestation Ages and Adult Mice Obtained Between β1 Int fl/fl and Tie2 Cre β1 Int fl/fl Mice Mating
Gestation ageTotal no.β1 Intfl/+ Tie2 Cre+β1 Intfl/fl Tie2 Cre+β1 Intfl/+ Tie2 Cre−β1 Intfl/fl Tie2 Cre−
  • a

    Embryos were dead and resorbed.

Adult12543 (34%)041 (33%)41 (33%)
E17.52810 (35%)09 (32%)9 (32%)
E14.53111 (36%)09 (29%)11 (35%)
E11.5289 (32%)09 (32%)10 (36%)
E10.54915 (31%)4 (8%)a14 (29%)16 (32%)
E9.517643 (24%)46 (26%)45 (26%)42 (24%)
E8.582 (25%)2 (25%)2 (25%)2 (25%)

To further analyze the embryonic lethality due to loss of β1 integrin gene in vascular endothelial cells, timed matings were set up between β1 integrin fl/fl mice and Tie2-Cre β1 Int fl/wt mice. Pregnant mice were sacrificed at different time points post coitum, and embryos were analyzed at different gestation ages from E8.5 to E17.5 (Fig. 1E–H). A piece of the yolk sac was used for genotyping. Table 1 shows the percentage of viable and visible resorbed (dead) embryos obtained at different time points post coitum. From day E10.5 on no viable Tie2-Cre β1 Int fl/fl mutant embryos were identified (Fig. 1E). At E10.5, either empty deciduas were present (suggesting complete resorption of Tie2-Cre β1 Int fl/fl embryos) or few partially resorbed (dead) Tie2-Cre β1 Int fl/fl mutant embryos could still be detected (Fig. 1H). Until E9.5, Tie2-Cre β1 Int fl/fl mutant embryos were found viable in the expected Mendelian ratio of 25% (Fig. 1E).

At E9.5, overall morphological appearance of Tie2-Cre β1 Int fl/fl embryos was normal, except some Tie2-Cre β1 Int fl/fl embryos were smaller in size when compared with Tie2-Cre β1 Int fl/fl littermates (controls), indicating growth retardation (Fig. 1G). Analysis of E8.5 embryos revealed no difference between control embryos and Tie2-Cre β1 Int fl/fl mutant littermates (Fig. 1F). Our findings indicate that Tie2-cre β1 Int fl/fl embryos die between E9.5 and E10.5 (Fig. 1E; Table 1).

Loss of endothelial β1 integrin in Tie2-Cre β1 Int fl/fl embryos was confirmed by immunohistochemistry in mutant versus control littermate embryos at day E9.5 (Fig. 2A–F). We also studied the effect of the loss of β1 integrin on its natural binding partners α4 and α5 integrin, by performing immunofluorescent double labeling experiments with endothelial cell markers CD31 (PECAM-1), von Willebrand Factor (vWF), and β1, α4 and α5 integrins, respectively. Whereas both α4 and α5 integrin are expressed in endothelial cells of normal control embryos, these integrins are absent in Tie2-Cre β1 Int fl/fl mutant embryos (Fig. 2G–R).

Figure 2.

Double immunofluorescence labeling of β1 integrin in endothelial cells. A–F: Immunofluorescence labeling of embryonic day (E) 9.5 embryos using antibodies to β1 integrin (green) and endothelial cell marker CD31 (PECAM-1; red) confirmed deletion of β1 integrin Tie2-Cre β1 Int fl/fl mutant embryos. G–L: DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). Double immunofluorescence labeling for α4 integrin a natural partner for β1 integrin and von Willebrand factor (vWF) an endothelial cell marker reveals loss of α4 integrin in Tie2-Cre β1 Int fl/fl mutant embryos. Control embryos showed α4 integrin expression in endothelial cells. M–R: Double immunofluorescence labeling for α5 integrin and von Willebrand factor (vWF) also reveals loss of α5 integrin in Tie2-Cre β1 Int fl/fl mutant embryos. Control embryos showed α5 integrin expression.

Tie2-cre β1 Int fl/fl Mutant Embryos Display Hypoxia and Vascular Abnormalities

Mutant and control embryos were sectioned, hematoxylin and eosin (H&E) stained and analyzed for histological changes and morphological changes. H&E-stained sections of E9.5 revealed insignificant histological differences between Tie2-Cre β1 Int fl/fl mutant embryos and control littermates, except for pronounced growth retardation in approximately 75% of all mutant embryos (Fig. 3A–D). That 100% of mutant embryos died between embryonic days 9.5 and 10.5 on the one hand, in combination with the lack of a distinct organ defect on the other hand, led us to hypothesize that a general hypoxia may be involved in the cause of death of mutant embryos.

Figure 3.

Histological analyses of mutant embryos and hypoxia in Tie2-Cre β1 Int fl/fl mutant embryos. A–D: Hematoxylin and eosin (H&E) sections of embryonic day (E) E9.5 embryos revealed insignificant histological difference between control embryos and Tie2-Cre β1 Int fl/fl mutant embryos. A and B display sagittal view, C and D show cross-sectional view of H&E staining from both control and Tie2-Cre β1 Int fl/fl mutant E9.5 embryos. E–H: Hypoxia staining (brown) in E9.5 Tie2-Cre β1 Int fl/fl mutant embryos compared control embryos, which did not stain for hypoxia. Higher magnification of hypoxia staining in control (F) and Tie2-Cre β1 Int fl/fl mutant embryos (H).

Hypoxic zones in embryos were analyzed after intravenous injection of hypoxia probe, pimonidazole, into the pregnant mouse. Hypoxia, as indicated by brown staining, was detected throughout mutant E9.5 embryos but not in littermate control embryos, supporting our hypothesis that general hypoxia could lead to the death of mutant embryos (Fig. 3E–H).

To further delineate the cause for hypoxia and death, we performed whole-mount CD31 (PECAM) staining to study the vasculature in mutant versus control embryos and yolk sacs. Extensive vasculature with networks of large and smaller vessels and sprouting angiogenesis was apparent in control embryos at E9.5 (Fig. 4A). Whereas in the mutant embryos the large vessels seem equally developed, overall vascular branching and sprouting appear substantially reduced when compared with the control embryos (Fig. 4B). This reduced branching and abrupt stunting of vessels are especially apparent in the brain region (Fig. 4C,D) and the heart (Fig. 4E,F, and red dotted line). The dorsal aorta is developed in both mutant and littermate control embryos (Fig. 4E,F, red stars). Extensive sprouting and branching of vessels from the dorsal aorta region, however, are seen only in control embryos but not in Tie2-cre β1 Int fl/fl embryos (Fig. 4E,F, red arrows) and the yolk sac (Fig. 4G,H). Whole-mount immunohistochemistry of yolk sacs with CD31 (PECAM) reveals similar vascular defects in the Tie2-cre β1 Int fl/fl mutant embryos when compared with control littermate embryos at E9.5 (Fig. 4G,H). Yolk sacs in control littermate embryos have a network of branches sprouting from the large vessels, whereas in the Tie2-cre β1 Int fl/fl embryos many large vessels end abruptly, and in many places, vascular sprouts display a widened lumen compared with the control embryos (Fig. 4G,H, arrows).

Figure 4.

Vascular defects in embryonic day (E) E9.5 Tie2-Cre β1 Int fl/fl mutant embryos. A–F: Whole-mount stained E9.5 embryos with CD31 (PECAM-1) antibody. A: Control E9.5 embryos maintain extensive vascular network of large and small vessels along with sprouting angiogenesis. B: E9.5 Tie2-Cre β1 Int fl/fl mutant embryos display overall significant reduction of vascular branching and sprouting, whereas large vessels in mutant embryos appeared to develop equally compared with control embryos. C: The brain region in control E 9.5 embryos displays extensive vascular branching and sprouting angiogenesis. D: The brain region in E9.5 Tie2-Cre β1 Int fl/fl mutant embryos reveals reduced branching and abrupt stunting of vessels. E,F: Control embryos reveal extensive sprouting and branching from dorsal aorta (E, red star, red arrow), which is not observed in F. E9.5 Tie2-Cre β1 Int fl/fl mutant embryos (red star, red arrow). The red dotted line represents the heart region, with apparently reduced branching and stunting of vessels in mutant embryos. G: Whole-mount immunostaining of yolk sac from control embryos stained with CD31 (PECAM-1) antibody display extensive sprouting and branching from large vessels (red arrows). H: Whole-mount E9.5 Tie2-Cre β1 Int fl/fl mutant yolk sac stained with CD31 (PECAM-1) revealed reduced branching from large vessels, abrupt ending of vessels with distended appearance (red arrows). I: H&E sections of yolk sacs from E9.5 control embryos revealed frequent connections between mesodermal and epithelial layers (arrow) filled with an abundance of red blood cells showing a vitelline vessel. J: Yolk sacs from Tie2-Cre β1 Int fl/fl mutant embryos showed less frequent connections between two layers with large vascular compartments filled with fewer blood cells.

Histological analysis of yolk sacs shows a normal vessel pattern and abundant red blood cells between the mesodermal and epithelial layers in the control embryos (Fig. 4I). In the yolk sacs of Tie2-cre β1 Int fl/fl mutant embryos, the mesodermal and epithelial layers appear less frequently connected with each other, which leads to abnormally large vascular compartments (Fig. 4J). Additionally, in the vessels of Tie2-cre β1 Int fl/fl mutant yolk sacs, fewer blood cells are present compared with control yolk sacs (Fig. 4I,J). In summary, our results suggest that lack of β1 integrin in endothelial cells causes defective angiogenesis and reduced number of blood cells, which leads to embryonic death around embryonic day E9.5, when intact vasculogenesis becomes indispensable for the oxygen supply of the embryo.

DISCUSSION

Formation of new vessels/capillaries from the preexisting vessels by means of angiogenesis, requires proliferation and migration of endothelial cells (Risau,1997). Vascular remodeling during the embryonic development requires extensive communication between the endothelial cells. Integrins function by signaling between cells and also between cells and the extracellular matrix (Giancotti and Ruoslahti,1999).

Systemic deletion of β1 integrin leads to embryonic lethality at E5.5 due to a peri-implantation defect (Fassler and Meyer,1995). In this report, we demonstrate that preservation of β1 integrin in all cells, except endothelial cells, extends the lifespan of the embryos by 4 gestation days from E5.5 to E 9.5. All mutant embryos, however, die between days E9.5 and E10.5, a time point when the embryo needs a more complex vasculature to survive. Analysis of the mutant embryos reveals general hypoxia and severe defects of vascular branching. These results argue for a crucial role of β1 integrin in the function of endothelial cells in the development of an elaborate vascular system. The observed reduced number of blood cells may be due to the reduced overall angiogenesis. Alternatively, because Tie2 is expressed on some hematopoietic stem cells, and β1 integrin is important for homing of hematopoietic stem cells to the liver during embryonic development, the lower number of blood cells could be an expression of reduced liver hematopoiesis (Fassler and Meyer,1995). Further studies are needed to address this question.

Traditionally, β3-containing integrins such as αvβ3 and αvβ5 integrin have been implicated as critical for vascular endothelial cell function and activity during development (Drake et al.,1995; Hanahan and Folkman,1996). This notion was challenged, however, when systemic deletion of β3 and β5 integrin did not lead to embryonic defects, and mice developed normally in most cases without any vasculogenic or angiogenic defects. The β3/β5 double null mice also develop without vascular defects (Hodivala-Dilke et al.,1999; Huang et al.,2000). Interestingly, despite some studies that suggested a role of β1 integrin in tumor angiogenesis, β1 integrin is not considered a key integrin for endothelial activity and function, especially during embryonic development. The outcome of the present study comes as a surprise, as it establishes a critical role for β1 integrin in embryonic vascular development.

If we take other studies into consideration, which examined the function of α1, α2, α3, α4, α5, α6, and αν integrins (all natural partners of β1 integrin on endothelial cells), using systemic deletion of each of these integrins, one may speculate that among all integrins with a β1 subunit, α5β1 and α4β1 integrins are most crucial for endothelial function in development. Systemic deletion of neither α1, α2, α3, nor α6 integrin results in embryonic lethality, suggesting that α1β1, α2β1, α3β1, and α6β1 are not required for the embryonic vascular development/function or they may be compensated by other integrins (Gardner et al.,1996; Georges-Labouesse et al.,1996; Kreidberg et al.,1996; Holtkotter et al.,2002). A certain percentage of αν null mice die embryonically, their premature death is attributed to placental defects, and these mice display complex vasculature, suggesting that ανβ1 integrin may not be the key player in initiating angiogenesis (Yang et al.,1995; Bader et al.,1998). The vascular defects observed in our study, however, are comparable to vasculature defects seen in α5 integrin null mice (Yang et al.,1993). Therefore, while some of the β1 integrin and α5 integrin null phenotypes were attributed to fibroblast/mesenchymal component defects, it is possible that mice with systemic absence of α5 integrin die at E9.5 due to primary defects in the endothelial cell function along with possible mesenchymal cell defects. Further studies are required to clarify these interesting observations. In α4 integrin null mice, embryonic lethality was attributed mainly to placental defects, but coronary vessel defects are also observed (Yang et al.,1995). Collectively, our results demonstrate that β1 integrin is critical for endothelial function during embryonic development.

EXPERIMENTAL PROCEDURES

Mice

β1 Int fl/fl mice were purchased from the Jackson Laboratories (Bar Harbor, ME). Tie2-Cre mice were provided by Dr. Yanagisawa. Rosa26 reporter mice were purchased from the Jackson Laboratories. Mice were maintained at the Beth Israel Deaconess Medical Center animal facility under standard conditions. All animal studies were reviewed and approved by the animal care and use committee of Beth Israel Deaconess Medical Center.

Polymerase Chain Reaction Genotyping

Genotyping for β1 integrin was done by polymerase chain reaction (PCR) using the primers 5′-CCGCTCAAAGCAGAGTGTCAGTC-3′ and 5′-CCACAACTTTCCCAGTTAGCTCTC-3′ resulting in 280 bp (loxP) and 160 bp (wild-type) product. Tie2-Cre transgenes were detected by PCR using primers 5′-GATGCCGGTGAACGTG CAAAACAGGCTC-3′ and 5′-CGCCGTAAATCAATCGATGAGTTGCTTC-3′, which generates a 450-bp product. A piece of yolk sac is used for genotyping embryos. For postnatal mice, genotyping a piece of tail is used for genotyping.

Whole-Mount Immunohistochemistry

Briefly, the harvested E9.5 embryos and yolk sacs were fixed in 4% paraformaldehyde–phosphate-buffered saline (PBS) overnight at 4°C. The embryos were rinsed with PBS and dehydrated with series of methanol changes. Peroxidase was blocked in 5% hydrogen peroxide–methanol, and embryos were rehydrated in series of methanol changes. Nonspecific binding of antibody is blocked in PBSMT (2% instant milk, 0.2% Triton X-100 in PBS) for 1 hr at 4°C. Subsequently, the embryos were incubated overnight with anti-CD31 (PECAM-1) monoclonal antibody (clone MEC 13.3 [Pharmingen]). After several washes in PBSMT, the embryos were incubated overnight at 4°C with horseradish peroxidase–conjugated F (ab)2 donkey anti-rat IgG secondary antibody (Jackson ImmunoResearch). After several washes with PBSMT and PBT (0.2% bovine serum albumin,0.1% Triton X-100–PBS), the signals were developed in 0.3 mg of 3′3′-diaminobenzidine/ml–0.5% NiCl2–0.03% H2O2. The samples were post-fixed with 2% paraformaldehyde–0.1% glutaraldehyde–PBS and then photographed.

Whole-Mount LacZ Staining

To confirm cre-mediated excision, R26RstopLacZ mice were mated with Tie2-Cre mice. Embryos were harvested from pregnant Tie2Cre; R26Rstop LacZ double transgenic at time points of E9.5 and E11.5 stained for LacZ. Embryos were first fixed in LacZ fixative solution containing 0.2% glutaraldehyde, 5 mM ethyleneglycoltetraacetic acid (pH 7.3), 100 mM MgCl2 in 0.1 M NaPO4 (pH 7.3; Lobe et al.,1999) for 1 hr at room temperature and washed three times in LacZ wash buffer (0.1 M PBS [pH 7.4] 1 M MgCl2 1% sodium deoxycholate 2% Nonidet-P40 in 100 mM sodium phosphate [pH 7.3]). The whole-mounts were stained in X-Gal staining solution (1 mg/ml of 5-bromo-4-chloro-3-indolyl β-galactoside [X-Gal], 2 mM MgCl2, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 0.01% sodium deoxycholate, and 0.02% Nonidet-P40) at 37°C overnight. X-gal–labeled embryos were embedded in paraffin, and sectioned in several planes and observed for LacZ staining.

Immunohistochemistry and Immunofluorescence

E9.5 embryos were fixed in 4% paraformaldehyde (PFA), and embedded in paraffin. Several serial sagittal and cross-sections from different embryos were stained for H&E. For double immunofluorescence staining on E9.5 embryos, embryos were fixed in 4% PFA and washed with PBS twice and transferred into 10% sucrose in PBS for 2 hr and subsequently into 20% sucrose in PBS at 4°C overnight. Embryos were placed in cryoblock with OCT and frozen in liquid nitrogen. Ten-micrometer sections of E9.5 embryos were stained with endothelial cell marker CD31 (PECAM-1) antibody and β1 integrin (hamster anti-rat β1 integrin antibody [BD Biosciences]). Double immunofluorescent staining of embryos for α5 and α4 integrin and vWF was done using rat-anti mouse α5 integrin (Pharmingen), mouse anti human α4 integrin (Chemicon) and rabbit anti-human vWF (DAKO). Fluorescein isothiocyanate–conjugated secondary antibodies were used for CD31(PECAM-1) and vWF, and rhodamine-conjugated secondary antibodies were used for β1, α5, α4 integrins staining. Nuclear staining was done with DAPI using Vectashield mounting medium (Vector).

Hypoxia Labeling

Hypoxia marker, pimonidazole hydrochloride (Hypoxyprobe-1) and mouse monoclonal antibody were purchased from Chemicon international. Pregnant mice were injected with pimonidazole hydrochloride (60 mg/kg mice) intravenously. After 2 hr, embryos were harvested, fixed in 4%PFA and washed with PBS. Embryos were then embedded in paraffin and sectioned, and hypoxia staining was performed according to manufacturer's instructions.

Acknowledgements

We thank Margot A. Martino for the genotyping of β1 integrin and Tie2-Cre mice.

Ancillary