Files in this Data Supplement:
Fig. S1. Targeted inactivation of the integrin α5 gene. (A) Map and expected Southern blot bands of the previously published complete integrin α5-KO (Yang et al., 1993). (B) Targeting construct and genotyping strategy of the conditional α5 integrin allele. B, BamHI; S, SacI; X, XbaI; K, KpnI. Correct targeting vector integration and Flp and Cre recombinase-mediated deletion steps were detected using the indicated SacI or BamHI genomic digestions followed by standard Southern blotting using the indicated probes: HT6, 5′ external; HT7, 5′ internal; HT3-2, 3′ external. Deletions by Flp and Cre recombinases and the predicted fragments from the deleted alleles are indicated. (C) Southern blot validation of Tie2-Cre-mediated α5 integrin conditional knockout in genomic DNA from lymphocytes digested by SacI or BamH1 using 5′ external probe HT6. Band sizes and genotypes are as indicated in A,B. (D) PCR genotyping of α5 conditional knockout mice on tail biopsies. Genotypes and the primer pairs (5′-3′) with expected band sizes are as follows. For α5 genotyping (see also B): HT030 (gcaggattttactctgtgggc) and HT031 (tcctctggcgtccggccaa) amplify a 694 bp α5 wild-type and a 821 bp α5 floxed specific band; primers HT030 and HT032 (gaggttcttccactgcctccta) produce a 501 bp α5-excised band; primers HT032 and HT038 (accgcttcctcgtgctttac) amplify a 977 bp α5-ko-neo band. For αv genotyping: primer triplet GACGCCTTCAACCTGGAC, CTGGATGCTGAGTGTCAGGT and ATATTGCTGAAGAGCTTGGCG detects a 860 bp αv-neo-KO band and a 291 bp αv wild-type product. Primer set CACAAATCAAGGATGACCAAACTGAG and ggtgactcaatGtgtgaccttcagcT amplifies a 550 bp αv wild-type, a 700 bp αv-floxed and 150 bp αv-excised bands. A 544 bp Tie2-Cre-specific product was obtained using primer pair CCCTGTGCTCAGACAGAAATGAGA and CGCATAACCAGTGAAACAGCATTGC. R26R-lacZ was according Jackson strain information and Immorto mice primer set gagtttagtttgtcagtgtatc and tttgtaaccattataagctgcaa, which gives a 376 bp SV40 large T antigen product. PCRs were performed in a 30 µl volume according to TaKaRa Taq, for 5 minutes at 95°C followed by 35 cycles of 30 seconds at 94°C, 1.5 minutes at 60°C and 1 minute at 72°C. Asterisk indicates non-specific PCR products.
Fig. S2. Characterization of α5-deficient endothelial cell lines derived from adult lung or brain tissue. (A) Immunoblot of total α5 and αv integrins in immortalized endothelial cell lines derived from murine lung (mLEC) or brain (mBEC) showing depletion and reconstitution by human α5 integrin by probing against the conserved cytoplasmic domain. Treatment of mLEC samples with PNGaseF results in lower molecular weight α5 bands due to cleavage of high mannose and complex-type N-linked oligosaccharides of glycoproteins. (B) FACS analysis of α5 and αv integrin surface expression levels on several of the isolated endothelial cell lines, indicating slight upregulation of αv integrin surface expression levels upon loss of α5 integrin. ICAM2 levels are shown for the mBEC cell lines. (C) Adhesion of mLEC endothelial cells to fibronectin for 15 minutes. α5-KO endothelial cells show a slight and concentration-dependent decrease in adhesion, which is reversed on re-expression of α5.
Fig. S3. Immunohistochemistry for α5 and αv integrin localization in lung endothelial cells plated on fibronectin. (A-F) Control cells express α5 (A) and vinculin (B) but little αv (C) in focal contacts. By contrast, note the loss of α5 integrin expression (B) and the less fibrillar and more peripheral organization of focal adhesions in the α5-KO cells (D,F) when compared with control cells (A). Note relocalization of αv integrin (D) into vinculin-positive (F) focal adhesions in α5-KO endothelial cells as compared with the diffuse αv distribution in the control cells (C,E). Scale bars: 30 µm.
Fig. S4. Efficient Tie2-Cre-mediated depletion of both α5 and αv integrins in E10.5 endothelial cells. FACS analysis of freshly isolated cells from E10.5 embryos gated on immune cells (PECAM1+, VE-Cad−), endothelial cells (PECAM1+, VE-Cad+) and control 'other' cells (PECAM1−) as described in Fig. S2. Cells were analyzed for integrin expression levels of α4 (upper), α5 (middle) and αv (bottom). Color coding of the FACS profile per genotype is indicated in the legend and as follows: black shaded, isotype control antibody; green shaded, integrin signals from α5; αv conditional double-hemizygous (cdHemi); blue line, α5-cKO; αv-cHemi; purple line, α5-cHemi; αv-cKO; red line, α5/αv-cdKO in endothelial cells. Note the similar α4 expression levels between the various genotypes in the upper panels, and the near complete loss of α5 or αv from endothelial and hematopoietic cells for the matched genotypes, but not from control 'other' cells.
Fig. S5. Apparently normal vasculature in E12.5 α5/αv-cdKO embryos. (A-L) Whole-mount PECAM1/α-SMA (smooth muscle actin) staining showing no obvious vascular defects in E12.5 embryos. (A,C,E,G,H,I) α5/αv-cHemi control embryos. (B,D,F,H,J,L) α5/αv-cdKO embryos. (A-D) Double stains for endothelial (PECAM1) and smooth muscle (α-SMA) cells on embryo heads (cranial plexus) showing no obvious defects. (E,F) Capillaries in head stained for endothelial cells (PECAM1). (G,H) PECAM1 staining showing cell-cell junctions in arteries of control and cdKO embryos. (I,J) Intersomitic blood vessel staining by PECAM1. (K,L) Endothelial staining of capillaries in the eyelid of embryos, showing no obvious differences.
Fig. S6. Apparently normal vasculature and lymphatics in rare 13-week-old α5/αv-cdKO mice. (A-F) Whole-mount stain of ear skin. Stained for endothelium (PECAM1) and lymphatics (LYVE1) (A,B) and endothelium (PECAM1) and smooth muscle cells (α-SMA) showing normal vasculature. (E,F) Showing lymphatics stained for PECAM1. (G,H) Whole-mount stain for endothelial cells (PECAM1) in cHemi and cdKO mouse trachea. Note that cHemi trachea was stretched more during fixation, resulting in fewer cartilage rings/picture frame. (I,J) Apparently normal cell polarity in aortic endothelial cells as visualized by VE-cadherin stain (I,J) and localization of centromere (pericentrin) magenta dots (I′,J′) compared with the nuclei.
Fig. S7. Patent ductus arteriosus in α5-cKO mice. Castings of the great arteries of adult 6- to 20-week-old mice with Batson�s No. 17 Plastic Replica and Corrosion Kit (Polysciences). (A) α5-cKO mouse showing a small patent ductus arteriosus connecting the pulmonary arteries (PA) and descending aorta (Dao). (B) Normal aortic arch of an α5flox/+; Tie-Cre control littermate. (C,D) Patent ductus arteriosus in α5-cKO mice. (E) α5/αv-cdKO adult mouse with a patent ductus arteriosus. Note also the apparent normal aortic arch with great arteries of the α5/αv-cdKO. For abbreviations see legend of Fig. 3. Scale bars: 2.5 mm.
Fig. S8. Cardiac cushion formation is normal in α5/αv-cdKO. Whole-mount lacZ reporter staining in E11.5 embryos shows no obvious defects in endothelial cardiac cushion formation. Tie2-Cre; R26lacZ-marked endothelial cells migrate into the cardiac cushion jelly and differentiate into mesenchymal cells endothelial-mesenchymal transition (EMT). (A-F) Transverse/cross-sections (head-to-tail order of sectioning). (A,B) Showing comparable cardiac cushions in the outflow tract (OFT); (C-F) showing comparable atrioventricular cushions. (G-L) Coronal/frontal sections in dorsal-to-ventral sectioning order. (G,H) Comparable formation of the 3rd, 4th and 6th branchial arch arteries; (I-L) comparable formation of inferior atrioventricular cushion tissue (IAC) and superior atrioventricular cushion tissue (SAC). BC, bulbus cordis; RA, right atrium; LA, left atrium; VV, venous valve. Scale bars: 200 µm.
Fig. S9. Characterization of α5/αv-dHemi, α5-KO/αv-Hemi and α5/αv-dKO endothelial cell lines derived from E13.5 embryos. (A) FACS analysis of α4, α5, αv, β1 and β3 integrin surface expression showing loss of expected integrins and expression of α4 integrin. (B) Adhesion to varying concentrations of fibronectin. Note that the embryonic α5/αv-dKO lines show very poor adhesion.
Fig. S10. Assembly of fibronectin into DOC-insoluble matrix by mLEC and mBEC cells. (A) Immunoblots of 1% DOC-insoluble fibronectin matrix assembled by adult murine lung endothelial cell lines plated for the indicated times on gelatin-coated plates. Blots show representative results from clonal control cell lines (α5flox/+; Tie-Cre) and clonal α5-KO with matched human α5 integrin reconstituted cell lines. Note the reduced insoluble fibronectin in α5-KO endothelial cells at the 1-day time point and the increased assembly of endogenous fibronectin in the matched α5-reconstituted cell lines. After 48-96 hours of culture, assembled matrix levels appear more similar between the various genotypes. Also note that, although the total secreted amount of fibronectin (media, 24 hours) seems less in knockout and the matching α5-reconstituted cell lines, the endogenous fibronectin level does not seem to affect the α5-dependent fibronectin assembly in α5-reconstituted cell lines. In addition, experiments in which 10 µg/ml of exogenous fibronectin had been added gave similar fibronectin assembly results (not shown). (B) One percent DOC-insoluble endogenous fibronectin assembly by doubly α5/αv-floxed control cells, in vitro depleted α5/αv-dKO mLECs and fibronectin-KO endothelial cells as negative controls. Note that the adult in vitro depleted α5/αv-KO cells do not assemble fibronectin and neither do the fibronectin-KO endothelial cells.