|
|
|
|
|
|
|
||||
| Home Help Feedback Subscriptions Archive Search Table of Contents | ||||
| ||||||||||||||||||||
Files in this Data Supplement:
Fig. S1. Egfl7 targeting strategy and confirmation of gene knockout. The Egfl7 locus was targeted by insertional mutagenesis using a retroviral gene trap vector (A) and by homologous recombination (D). The precise genomic insertion site of the retroviral gene trap vector in the Egfl7 gene was determined by inverse genomic PCR. The insertion leads to silencing of endogenous Egfl7 transcription, and transcripts initiated from the inserted vector contain stop codons in all three frames, thus abolishing EGFL7 protein synthesis. Both knockout lines were backcrossed into a C57Bl/6 genetic background. Chimeric founders from the homologous recombination line were also crossed to 129svj to generate pure 129svj offspring. Disruption of Egfl7 transcription by the gene trap vector was confirmed by RT-PCR on various tissues using gene-specific primers located in exons 2 and 3 (B). Homologous recombination was confirmed by Southern blotting (E), and endothelial cell-specific transcription from the targeted locus was assayed by lacZ staining on E10.5 heterozygous mice (F). Immunofluorescence staining with an anti-EGFL7 monoclonal antibody confirmed the loss of EGFL7 in homozygous tissue from the gene trap (C) and the homologous recombination line (G). Tissues used for the immunofluorescent staining were neonatal lungs from Egfl7+/+ (C, top panels) and Egfl7−/− (C, bottom panels) P14 littermates, and uterus from pregnant Egfl7+/+ (G, top panels) or Egfl7−/− (G, bottom panels) females. The actual size represented by the width of the panel: 1.5 mm in C,G (all panels); 4 mm in F.
Fig. S2. Egfl7 is expressed in endothelial cells in the neonatal retina. In situ hybridization with a mouse Egfl7 antisense radioactive riboprobe (A,C-F) shows expression in the developing vascular plexus at P3 (A), P6 (C,D), P10 (E) and P12 (F). (B) Negative control using a P3 retinal section hybridized with the sense riboprobe. Egfl7 is expressed in the nerve fiber layer vasculature (white arrows) and other vessels in the eye, including capillary beds in the outer plexiform layer (red arrowhead), hyaloid vasculature (white arrowheads), tunica vasculosa lentis (blue arrowhead) and vessels in the choroid and sclera (yellow arrowheads). EC-specific expression is confirmed by staining retina from a P14 Egfl7+/− mouse for β-galactosidase activity (blue), collagen IV (brown) and nuclear fast red (magenta) (G). The actual size represented by the width of the panel: 3.25 mm in A-C,E,F; 1.3 mm in D; 455 μm in G.
Fig. S3. Vascular phenotype in the homologous recombination KO line. (A,B) Whole-mount CD31 IHC on the heads of Egfl7+/+ (A) and Egfl7−/− (B) embryonic littermates at E10.5. Red open arrowheads indicate major cranial vessels. (C-H) Isolectin B4 immunofluorescent staining on the P5 retinas of Egfl7+/+ (C,D) and Egfl7−/− (E,F) littermates. Panel F highlights aberrant EC aggregates at the vascular migration front. Quantitative measurement of the retinal vascular coverage (G) and retinal sizes (H) of multiple pairs of Egfl7+/+ and Egfl7−/− littermates at P5 indicates that vascular coverage is significantly reduced in the Egfl7−/− mice, whereas the collective retinal size is unchanged. Each dot in the graph represents one mouse. Red arrowheads indicate aberrant EC clusters (B,E,F). (I-L) Egfl7+/+ (I,J) and Egfl7−/− (K,L) littermates at P5 were perfused with fluorescein-labeled tomato lectin and peritonea examined by confocal microscopy. Horizontal arrows in K and L indicate the same branch, and vertical arrows in I and J indicate another branch. Large vessels and their branches remain in the same plane in Egfl7+/+ tissues (I,J), whereas those in the Egfl7−/− tissues do not (K,L). Vessels are smooth in the Egfl7+/+ tissues (I,J), but are tortuous in the Egfl7−/− tissues (K,L). (M-W) Adult Egfl7+/+ (M,N,T,U) and Egfl7−/− (O-S,V,W) littermates were perfused with fluorescein-labeled tomato lectin (green), and retinas were stained with smooth muscle actin (αSMA, red, or white in M,O). M and O show αSMA signal only, R-W show lectin signal only. Vasculatures in the Egfl7−/− retinas show a number of defects: major arteries are tortuous (O,R); vessels of similar diameters are aberrantly associated (P); vascular diameter is irregular along its length (S); anastamosing vessels are looping around and form ‘knots’ at the junctions between multiple vessels (blue arrow in Q); some primary vessels (largest arteries and veins) are not restricted by layer boundaries and are seen penetrating through the retina into the OPL (arrowhead in W), whereas their wild-type counterparts all stay in the NFL layer (small white arrows in T,U). T-W are 3D renderings of all three vascular layers using a series of optical sections; W is tilted by 30° to better visualize the primary vessel entering the OPL. Insets in N and P are sample z-sections of the vessels shown in the same panel. The actual size represented by the width of the panel: 3.6 mm in A,B; 3.4 mm in C,E; 340 μm in D,F,I-L; 3.8 mm in M,O; 157 μm in N,P; 163 μm in Q; 0.8 mm in R,S.
Fig. S4. Methods of quantification used in this manuscript. This figure contains images to explain a subset of quantification methods used for figures in the main text or in the supplementary information. Images used for quantification and presentation were acquired with a Photometrics Cool Snap fx camera/Olympus BX61 microscope (Metamorph Molecular Devices software), or a Spot RT Slider camera/Leica MZFLIII dissection scope. Confocal images were collected with a Zeiss LSM 510 two-photon laser microscope; 3D renditions were generated with Imaris software (Bitplane). For protein localization, single-frame confocal images were used. For cell counting, confocal projection images were used. (1) Embryonic coronary vascular coverage. Quantification was conducted on photographs of the ventral sides of CD31-stained hearts (A). Total area of a heart (A, labeled as ‘a’ in the middle panel) and the area covered by vasculature (A, labeled as ‘b’ in the right panel) were manually traced and calculated using Image J software. Vascular coverage (percentage)=vessel-covered area (b) divided by the total heart area (a)×100. Vascular coverage for multiple pairs of Egfl7+/+ and Egfl7−/− littermates is plotted in Fig. 2G; each dot in the graph represents one animal. P values between two genotypes were calculated by t-test using the statistic software Prism. (2) Embryonic cranial vascular coverage. Quantification was conducted on photographs of whole-mount heads stained with CD31. Numbers of major vessels spanning along the dorsal-ventral axis of the mesencephalon were counted (open arrows in Fig. 2E,F and Fig. S3A,B). Numbers of major cranial vessels from multiple pairs of Egfl7+/+ and Egfl7−/− littermates are plotted in Fig. 2G; each dot in the graph represents one animal. P values between two genotypes were calculated by t-test using Prism. Data for the insertional and homologous recombination lines are both included in Fig. 2H. (3) Neonatal retinal vascular coverage and retina sizes. Quantification was conducted on photographs of flatmount retinas stained with isolectin B4. The radius of each petal in each retina was measured individually (yellow lines in B, length of each radius expressed as a1, a2, a3, a4, a5). The size of each retina was calculated as the mean of the radii for all petals. A similar method was used to measure the radii of vessels-covered area (red lines in B, lengths expressed as b1, b2, b3, b4, b5), and the value for each retina was calculated as the mean of the radii for all petals. Vascular coverage was calculated as ratio between vascular radii and retinal radii (b/a), and value for each retina was calculated as the mean of values for all petals. Vascular coverage and retinal size from multiple pairs of Egfl7+/+ and Egfl7−/− littermates are plotted in Fig. 2O,P for the insertional mutant line, and in Fig. S3G,H for the homologous mutant line. Each dot in the graphs represents one animal. P values between two genotypes were calculated by t-test using Prism. (4) Aortic ring sprout extension velocity. Quantification was conducted on bright-field images of live aortic rings taken at different days (C and Fig. 4A,B). Individual sprouts were followed based on their position and shape. Length of each sprout was measured using the software Metamorph (colored lines in the right-hand four panels in C). The sprout extension velocity was calculated as: (‘length at the end day’−‘length at the start date’)/‘number of days elapsed’. The average lengths at different days and sprout extension velocities over different lengths of time are plotted in Fig. 4C and D, respectively. 28 sprouts from six Egfl7+/+ rings, and 41 sprouts from six Egfl7−/− rings, were counted. (5) Primary mouse EC transwell migration. Experiments were conducted as described in Materials and methods. ECs were isolated from four pairs of Egfl7+/+ and Egfl7−/− littermates. Triplicate samples were run using ECs from each mouse. Due to experiment-to-experiment variation in the number of cells migrated, we calculated the results as the ratios between Egfl7+/+ and Egfl7−/− cells from each experiment. Within each experiment, nine pair-wise comparisons were conducted between three Egfl7+/+ and three Egfl7−/− samples. The average and standard error of 36 wild-type/knockout ratios were plotted as shown in Fig. 4E. (6) Individual HUVEC migration velocity. Experiments were conducted as described in the Materials and methods section. Cells in three wells were analyzed for each treatment. Numbers of cells tracked are: n=12 (in the fibronectin-coated plate); n=11 (in the fibronectin+EGFL7-coated plate); n=6 (in the EGFL7-coated plate). Results are shown in Fig. 4F. (7) Tip and stalk cell counts in the retinas. Counting was conducted on high-magnification confocal projection images of P5 retinas from five pairs of Egfl7+/+ and Egfl7−/− littermates. For each sprout, a 40 μm line was drawn from the tip of the tip-cell body toward the stalk cells (white lines in Fig. 5B,E). The numbers of tip cell nuclei (black asterisks in Fig. 5B,E) and stalk cell nuclei (x in Fig. 5B,E) within this length limit were counted. Tip cells were recognized by the presence of filopodia and their position. Five to seven sprouts were counted per retina. Tip and stalk cell counts were plotted separately and are shown in Fig. 5G and H, respectively. Each dot in the plots represents one sprout. (8) Cell counts in the aortic rings. Counting was conducted on high-magnification confocal projection images of aortic rings from four pairs of Egfl7+/+ and Egfl7−/− littermates. For each sprout, a 50 μm line was drawn from the tip of each sprout towards the middle of each sprout (white lines in the right panels in D, and in Fig. 5J,L). The numbers of cell nuclei (red in the right bottom panel in D) within this length limit were counted. Ten to fifteen sprouts were counted per aortic ring. Results are shown in Fig. 5M. Each dot in the plot represents one sprout. We did not count tip and stalk cells separately because they are not as readily recognizable as in the retinas. (9) Mural cell coverage in the retinas. Quantification was conducted on high-magnification confocal images of P5 retinas from three pairs of Egfl7+/+ and Egfl7−/− littermates, with three images per retina. Pixel counts of NG2 signal and isloectin B4 signal were carried out in different color channels using Adobe Photoshop. The ratios between NG2 pixel count and isolectin B4 pixel count are plotted in Fig. S5P. (10) Endothelial cell proliferation in vivo. Digital images were taken of cryosections through equivalent levels of the retina stained with CD31+Ki67+DAPI (Fig. S7A,B). Ki67+ nuclei of CD31+ cells were counted and plotted against total nuclei of CD31+ cells; n=5 images per animal, 4 animals per genotype. Results are plotted in Fig. S7C. (11) Isolated mouse EC proliferation/survival. Primary mouse endothelial cells isolated from Egfl7+/+ and Egfl7−/− littermates were seeded into each well of 96-well plates, at 50,000 cells per well. n=12 wells per animal, 3 animals per genotype. Cells were allowed to attach for 1 hour, washed three times with PBS, and then cultured for 16 hours in EGM2, 20% FBS. The Alamar Blue assay was carried out following the manufacturer’s protocol (BioSource). Alamar Blue reduction signal was normalized to total cell number measured by CyQuant (Invitrogen). Results are plotted in Fig. S7F.
Fig. S5. Abnormal vascular development in the Egfl7<b>−</b>/<b>− retina is a primary defect. A number of cell types within the retina were investigated to determine whether the vascular phenotype in the neonatal retina is secondary due to defects in other cell types. All data in this figure are derived from neonatal retinas of the Egfl7+/+ and Egfl7−/− littermates from the insertional Egfl7-knockout line; genotypes are indicated. (A-D) Histological analysis of Hematoxylin and Eosin (H&E) stained sections of P2 (A,C) and P5 (B,D) retinas revealed no overt morphological difference between Egfl7+/+ and Egfl7−/− littermates. (E-H) We examined if astrocyte development and migration are affected by the loss of EGFL7 protein, as the astrocytic layer develops prior to endothelial invasion and serves as a migration scaffold for the extending vascular plexus. Shown here are flatmount P5 retinas stained with isolectin B4 (green) and astrocyte marker GFAP (red). The extent of astrocyte/glial cell coverage, as well as the spatial relationship between ECs and astrocytes, are unchanged in the Egfl7−/− retinas. (I,J) Since astrocytes guide the migration of endothelial cells by secreting VEGF, we examined sections of P8 retina stained with isolectin B4 (green), VEGF (red) and DAPI (blue); the arrowhead indicates the vascular migration front. VEGF distribution is unchanged in the knockout retina. (K) VEGF ELISA on retinal lysates from P5 and P8 animals. VEGF levels are not changed in the Egfl7−/− retinas. (L-O) Mural cells (pericytes and vascular smooth muscle cells) are essential for endothelial cell survival, vessel integrity and remodeling. We stained P5 retinas with isolectin B4 (green) and mural cell marker NG2 (red, autofluorescence of red blood cells is also seen in the red channel) and found that the spatial distribution of NG2+ mural cells is similar in Egfl7+/+ and Egfl7−/− retinas. (P) Quantification of the ratios between NG2 and isolectin signals from three pairs of P5 retinas. wt, Egfl7+/+; ko, Egfl7−/−. The extent of mural cell coverage throughout the retina is unaffected by the loss of EGFL7. (Q-T) Transmission electron micrographs of P5 retinal sections through capillaries in the NFL midway between the optic nerve and migration front. E, endothelial cells; P, pericytes. Arrows, points at which EC and pericyte cell membranes are in direct contact. The spatial relationship between ECs and mural cells is similar in Egfl7+/+ and Egfl7−/− retinas. In addition, the number of points at which endothelial cell and pericyte membranes make close contact (arrows in R,T) are similar between the two genotypes: 3.84 (average)±0.45 (s.e.m.) contact points per capillary (n=19 capillaries) in Egfl7+/+ retinas; 4.19±0.84 contact points per capillary (n=21 capillaries) in Egfl7−/− retinas (capillaries with sizes below 7 μm were analyzed). The actual size represented by the width of the panel: 457 μm in A-D; 365 μm in E,G; 110 μm in F,H; 470 μm in I,J; 180 μm in L-O; 8 μm in Q,S; 429 nm in R,T.
Fig. S6. Fibronectin assembly is unaffected by the loss of EGFL7. (A,B) Fibronectin plays a crucial role in supporting retinal EC migration. In the retina, a fibrous network rich in fibronectin is laid down by the developing astrocytic cell layer, providing a scaffold for the developing vasculature to migrate on. Staining of isolectin B4 (red) and fibronectin (FN, green) on P5 retinas from the Egfl7+/+ (A) and Egfl7−/− (B) littermates indicates that the fibronectin network is unaffected by the loss of EGFL7, as the size and length of the fibronectin fibers are comparable between the two genotypes. (C,D) In addition, fibronectin (green), VE-cadherin (VECad, red) and DAPI (blue) triple staining on primary ECs isolated from the Egfl7+/+ (C) and Egfl7−/− (D) littermates indicates that the ability of mutant ECs to assemble and maintain a fibronectin network is remarkably similar to wild-type ECs. Retinas and cells are from the insertional mutant line. The actual size represented by the width of the panel: 78 μm in A,B; 117 μm in C,D.
Fig. S7. EC proliferation and apoptosis are unaffected in the absence of EGFL7. (A-C) Ki67 (red), CD31 (green) and DAPI (blue) staining on P2 retinal sections from the Egfl7+/+ (A) and Egfl7−/− (B) littermates shows no difference in labeling index between the two genotypes. Ratios between the Ki67/CD31 double-positive cells and CD31-positive cells are plotted in C. (D,E) Activated caspase-3 (red), CD31 (green) and DAPI (blue) staining on P2 retinal sections from the Egfl7+/+ (D) and Egfl7−/− (E) littermates. Apoptosis is rarely seen in retinal ECs, although it is readily detectable in neurons (arrowheads) regardless of genotype. (F) Alamar Blue assay on primary endothelial cells isolated from the Egfl7+/+ (red bars) and Egfl7−/− (green bars) littermates. Plotted are Alamar Blue reduction signals (measuring the reductive environment of proliferating cells) normalized to total input cells after the cells are cultured for 1 and 2 days. The readouts are similar between the two genotypes. This result provides additional evidence that there is no significant change in cell viability/proliferation in the Egfl7 mutant tissues. Tissues and cells are derived from the insertional mutant line. The actual size represented by the width of the panel: 57 μm in A,B; 126 μm in D,E.
Fig. S8. FAK phosphorylation is reduced in the Egfl7<b>−</b>/<b>− stalk cells. Focal adhesion kinase (FAK) phosphorylation is a molecular event closely related to cell adhesion and migration on ECM substrates. In our hands, phosphorylation on tyrosine 861 (FAK-pY861) seemed to be the predominant form in endothelial cells within the neonatal retinas and aortic ring sprouts, whereas FAK-pY397 was detected predominantly in non-endothelial cells (data not shown). This figure shows the changes of pY861 in the Egfl7 mutant retinas. (A-H) Neonatal retinas from Egfl7+/+ (A,B,E,F) and Egfl7−/− (C,D,G,H) littermates were stained for isolectin B4 (red in B,D,F,H), DAPI (blue in B,D,F,H) and FAK phosphorylated at amino acid 861 (FAK-pY861; A,C and green in B,D) or total FAK (E,G and green in F,H). Arrowheads, non-endothelial signal; white arrow, stalk cell signal; red arrows, tip cell filopodia signal. A-D are retinal flatmount samples; E-H are retinal sections because the anti-FAK antibody worked poorly in whole-mount retinas. (I-P) Aortic rings from Egfl7+/+ (I,J,M,N) and Egfl7−/− (K,L,O,P) littermates were stained for CD31 (red in J,L,N,P), DAPI (blue in J,L,N,P) and FAK-pY861 (I,K and green in J,L) or total FAK (M,O and green in N,P). Whereas total FAK staining is similar between the two genotypes, significant reduction in FAK-pY861 occurs in the Egfl7-deficient stalk cells in the two tissues examined. Tissues are derived from the insertional mutant line. The actual size represented by the width of the panel: 42 μm in A-H; 104 μm in I-P.
| ||||||||||||||||||||
