Platelet-Derived Growth Factor Receptors Direct Vascular Development Independent of Vascular Smooth Muscle Cell Function^▿

Mouse lines.

The mouse lines used in these studies were PDGFRα^fl/fl (64), PDGFRβ^fl/fl (53), Tg(Tagln-cre)1Her/J (SM22α-Cre^Tg) (23), Meox2^Cre/+ (65), myocardin^Cre/+ (39), Tie2Cre^Tg (30), ROSA26 reporter LacZ (R26R) (61), Tie2GFP^Tg (59), and PDGFRα^GFP (19) lines. SM22α-Cre^Tg mice were purchased from Jackson Laboratories. Transgene levels of SM22α-Cre^Tg animals were detected by Southern blot analysis using a probe for the Cre gene. These mice were maintained by inbreeding lines that were homozygous for SM22α-Cre^Tg. Control embryos and yolk sacs were either SM22α-Cre^Tg littermate embryos bearing heterozygous floxed alleles or wild-type embryos (somite stage matched). Detection of a vaginal plug was defined as E0.5. PDGFR^MKO embryos were recovered up to E18.5, but we recovered fewer than expected after birth. Often, we found postnatal day 1 animals with spina bifida and a cleft palate. Previously, we had determined that the PDGFRα floxed allele is hypomorphic and is lethal in combination with a null allele. These data, combined with the fact that myocardin^Cre can lead to germ line deletion of floxed alleles, suggested that the lethality was not caused by the conditional deletion of the PDGF receptors but by loss of PDGFRα signaling regardless of the myocardin^cre status of the mice.

Histology and immunohistochemistry.

Samples stained for β-galactosidase were fixed in 2% formaldehyde/0.2% glutaraldehyde for 10 min, stained in 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside overnight at room temperature, and postfixed in 10% buffered formalin for 20 min. Whole embryos were stained for PECAM and αSMA according to standard procedures and cleared using benzylalcohol-benzyl benzoate (1:2) for imaging (22). Yolk sacs were fixed for 1 h in 4% paraformaldehyde (PFA) at 4°C and blocked for 30 min in 1.5% normal serum. For section analysis, samples were fixed for 1 to 3 h in 4% PFA, embedded, and sectioned at 7 to 10 μm. Immunohistochemistry was performed by incubation in primary antibody for 2 h to overnight at 4°C and visualized using Alexafluor secondary antibodies or a Vectastain Elite ABC kit and a DAB peroxidase substrate kit (Vector Laboratories, Burlingame, CA) according to the manufacturer's instructions. Antigen retrieval for PECAM and αSMA paraffin sections was performed as previously described (64). Hematoxylin and eosin (HE) staining and picrosirius red staining were performed according to standard methods.

Western blotting and immunoprecipitation.

For immunoprecipitation, whole yolk sac lysates were incubated overnight at 4°C with 1 μg of antibody and then for 1 h with protein A-Sepharose beads. After being washed, the precipitated proteins were run on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. For Western blotting, yolk sac lysate was quantified using Bradford reagent, and equal amounts of protein were run on sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes following standard protocols. The membranes were incubated with 1% bovine serum albumin/0.05% Tween 20 in Tris-buffered saline for 30 min, with primary antibody overnight at 4°C, and with secondary antibody for 1 h. The results were visualized using enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ).

Antibodies.

The primary antibodies used for immunohistochemistry and whole mount included PDGFRβ (eBioscience, San Diego, CA), 1:250; PECAM (clone MEC13.3; BD Bioscience, San Jose, CA), 1:250; anti-SMA-fluorescein isothiocyanate (clone 1A4; Sigma, St. Louis, MO), 1:500; green fluorescent protein (GFP) (abcam, Cambridge, MA), 1:250; SM22α (abcam, Cambridge, MA), 1:250; collagen 1 (abcam, Cambridge, MA), 1:250; collagen 4 (Chemicon, Temecula, CA), 1:250; fibronectin (Sigma), 1:500; and laminin (Sigma), 1:250. The secondary antibodies used included anti-rabbit, anti-rat, and anti-mouse Alexafluor 488, 543, 594, and 633 (Molecular Probes, Eugene, OR), 1:500. Antibodies used for Western blots included integrin β1 (abcam), 1:1,000; integrin β1 phospho-S785 (abcam), 1:1,000; PDGFRβ (Upstate, Charlottesville, VA), 1:1,000; PDGFRα (Upstate), 1:1,000; and cytoskeletal actin (Novus Biologicals, Littleton, CO), 1:1,000.

Real-time PCR.

RNA was isolated from yolk sacs using Trizol (Invitrogen, Carlsbad, CA) and an RNeasy kit (Qiagen, Valencia, CA). Samples were isolated and homogenized in Trizol; 20% choloroform was added, mixed well, and centrifuged for layer separation. The top layer was mixed with an equal volume of 70% ethanol, added to an RNeasy kit minispin column, and centrifuged. Washes were followed according to the RNeasy protocol and eluted with diethyl pyrocarbonate-H₂O. RNA was quantified and DNase treated. RNA (1 μg) was used to generate cDNA using PowerScript reverse transcriptase (Clontech, Mountain View, CA) and random hexamers. Gene expression was quantified using standard real-time PCR methods using Sybr green master mixture on an ABI7000 instrument (Applied Biosystems, Foster City, CA). Samples were analyzed in triplicate on a minimum of three samples per genotype. Primer sequences will be provided upon request.

MMP assays.

DQ gelatin assay (Molecular Probes) was performed on fresh frozen tissue sections according to manufacturer's instructions. Briefly, samples were sectioned and incubated for 30 min in substrate buffer and then transferred to DQ gelatin (3 μg/ml) at 37°C and imaged after incubation for 0, 15, and 30 min. Zymography was performed according to standard protocols (32) with human matrix metalloprotease 2 (MMP-2) and MMP-9 standards (Chemicon). Yolk sac lysates were quantified using Bradford reagent, and equal amounts of protein were analyzed.

Allantois assay.

E8.5 allantoides were isolated from wild-type and PDGFRα^fl/fl; PDGFRβ^fl/fl embryos and plated on gelatin-coated coverslips. The medium was supplemented with 1% fetal bovine serum (FBS), 10% FBS, or 20 ng PDGFBB (R&D Systems, Minneapolis, MN). Cultures were grown for 24 to 26 h at 37°C in 5% CO₂. Samples were fixed with 4% PFA for 10 min and stained for PECAM as described above. Adenovirus transduction was performed using 1.4 × 10⁷ PFU per 1 ml of medium at the time of plating. PDGF receptor inhibitor (AG1296 [2 μM]; Calbiochem, Gibbstown, NJ) was added to appropriate stimulation medium at the time of plating. For collagen 4 assays, wells were coated for 1 h with 0.5 μg/ml, 5 μg/ml, or 50 μg/ml collagen 4 (R&D Systems), and then medium was added to the collagen 4 at the time of allantois plating.

Image acquisition and manipulation.

Whole-mount and section samples were analyzed using Nikon SMZ1000 and Zeiss Axiovert200 (Carl Zeiss, Thornwood, NY) microscopes. Images were captured using Hamamatsu Orca-ER (Hamamatsu, Bridgewater, NJ) and Olympus DP71 (Olympus, Center Valley, PA) cameras with OpenLab 3.5.1 and DP Controller software, respectively. Fluorescent images were colored using OpenLab and then processed in Adobe Photoshop. Confocal images were captured using an LSM510Meta confocal microscope (Carl Zeiss) and processed in ImageJ and Adobe Photoshop. Color images and Western blots were processed in Adobe Photoshop for a white background. Quantification and threshold measurements were calculated through ImageJ. Graphs and statistics were generated using Prism (Graphpad). The final figures were compiled using Canvas 9.

Generation and survival of PDGF receptor smooth muscle cell knockout embryos.

Previous data from PDGFRβ and PDGFB null embryos indicated that PDGFRβ is required for formation of VSMC and that myocardial development is also affected (21, 34, 37, 60, 66). Because we wanted to investigate the cell lineages dependent on PDGFRβ signal transduction, we used loxP/Cre technology to generate animals that lacked PDGFRβ in both cardiomyocytes and VSMC. We used two Cre-expressing mouse lines to accomplish the deletion, SM22α-Cre^Tg (23) and myocardin^Cre (39) lines. SM22α is expressed as early as E8.5 in cardiomyocyte development and during VSMC terminal differentiation (36), while myocardin is expressed by E8.5 in cardiomyocytes and is one of the earliest genes expressed in VSMC precursors (68). Surprisingly, deletion of PDGFRβ using either SM22αCre^Tg (referred to as PDGFRβ^SKO) or myocardin^Cre/+ (referred to as PDGFRβ^MKO) mice did not phenocopy PDGFRβ^−/− mice. Both PDGFRβ^SKO and PDGFRβ^MKO embryos and mice were viable and fertile.

Recently, we have shown that the two PDGF receptors (PDGFRα and -β) act cooperatively in the neural-crest-derived smooth muscle of the aortic arch (53). To determine if PDGFRα was compensating for loss of PDGFRβ signaling in other VSMC populations, we generated mice that contained SM22α-Cre^Tg and conditional alleles of both PDGF receptors. Genotyping of offspring from these crosses revealed that few PDGFRα^fl/fl; PDGFRβ^fl/fl; SM22α-Cre^Tg (PDGFR^SKO) pups were recovered at birth, indicating that PDGF receptor signaling through both receptors was required for viability in an SM22α-expressing cell population. Using timed matings, we recovered the expected Mendelian ratios of embryos up to E10.5 but few PDGFR^SKO embryos past that time point. The few viable mice that were doubly homozygous for the PDGF receptor floxed alleles were SM22α-Cre^Tg/0 (hemizygous for the transgene) (data not shown). We surmised that the expression level of Cre in mice hemizygous for the transgene was not sufficient to recombine all four floxed alleles efficiently. Indeed, no PDGFR^SKO pups resulted from breeders that were SM22α-Cre^Tg/Tg. All subsequent analyses were performed using mice homozygous for SM22α-Cre^Tg. Commonly, transgenic Cre lines are not maintained as homozygotes due to potential transgene insertion effects. We have ruled out these effects, because lethality occurs only in double-homozygous PDGF receptor SM22α-Cre^Tg/Tg embryos, not single PDGF receptor SM22α-Cre^Tg/Tg embryos. Because SM22α-Cre^Tg was likely to _target deletion in VSMC similarly to myocardin^Cre, we predicted that excision of the two PDGF receptor genes in this mouse line (PDGFR^MKO) would phenocopy the embryonic lethality observed in PDGFR^SKO embryos. However, PDGFR^MKO embryos survived until birth.

Expression of SM22-Cre^Tg before VSMC differentiation in the yolk sac.

The dramatic difference in survival between PDGFR^SKO and PDGFR^MKO embryos led us to investigate the profiles of Cre activity in the two Cre mouse lines. Using ROSA26 reporter mice, we determined SM22α-Cre^Tg and myocardin^Cre expression between E8.5 and E10.5. Cre activity was not detected in any tissue at E7.5 using either of the Cre lines. By E8.5, SM22α-Cre activity was observed in many cells of the yolk sac, as well as a small number of cells in the primitive heart (Fig. 1A). In contrast, myocardin-Cre activity was detected in the cardiac crescent, and only a few β-galactosidase-positive cells were observed in myocardin^Cre yolk sacs (Fig. 1E). Because Cre expression in the yolk sac was the most obvious difference between these lines, we examined histological sections of this tissue. At E8.5, SM22α-Cre activity was present throughout the yolk sac mesothelium, but no β-galactosidase-positive cells were detected in the myocardin^Cre yolk sacs (Fig. 1A and E). The yolk sac mesothelium is a mesoderm-derived epithelium-like component of the yolk sac that rests on a thin basement membrane and is believed to be important for transport and movement of fluid from the yolk sac. At E9.5, β-galactosidase-positive cells were present both in the mesothelial layer and surrounding some vessels in the SM22α-Cre^Tg mice, while myocardin^Cre activity was present in only a few cells associated with blood vessels, presumably VSMC progenitors (Fig. 1B and F). At E10.5, SM22α-Cre^Tg and myocardin^Cre yolk sacs both possessed β-galactosidase expression in cells surrounding blood vessels, but β-galactosidase expression in myocardin^Cre was lacking in yolk sac mesoderm populations that were not vessel associated (Fig. 1C and G). These data pointed to the possibility that deletion of the PDGF receptors in the yolk sac mesothelium caused the embryonic lethality.

Because Cre expression leads to indelible β-galactosidase expression, we could also use this marker to follow mesothelial and VSMC cell lineages in PDGFR^SKO and PDGFR^MKO embryos. In both PDGFR^SKO and PDGFR^MKO yolk sacs, few perivascular, β-galactosidase-positive cells were present at E10.5 (Fig. 1D and H). β-Galactosidase-positive cells were abundant in the mesothelial layer in the PDGFR^SKO yolk sac, demonstrating that loss of PDGF receptors did not lead to failure in the formation of this cell population. In addition, examination of both PDGFR^SKO and PDGFR^MKO embryos revealed abundant β-galactosidase-positive cells in the hearts and trunk areas (data not shown), demonstrating that loss of the receptors did not result in a general reduction of mesoderm cells.

PDGFRα and PDGFRβ are coexpressed in the yolk sac mesothelium and perivascular cells.

While PDGFRα and PDGFRβ expression has been documented as early as E6.5 in the extraembryonic endoderm and ectoplacental cone, and the ligands are expressed in the chorionic ectoderm and parietal endoderm (42, 48, 58), analysis of coexpression of the receptors in the yolk sac has not been done. Therefore, we investigated the expression patterns of the receptors. Expression of PDGFRα was detected using mice that expressed a nuclear-localized GFP from the PDGFRα locus that faithfully traced PDGFRα expression (19) and a PDGFRβ-specific antibody. At E8.5, PDGFRα and PDGFRβ were expressed in the mesothelium and endoderm, as previously reported (Fig. 2A). At E9.5, PDGFRβ endoderm expression was reduced, but expression of both receptors was maintained in the mesothelium (Fig. 2B and data not shown). At E10.5, both receptors were expressed in the mesothelium, but only PDGFRβ was identified in cells surrounding endothelial vessels (Fig. 2C). Presumably, these cells were VSMC. The early and persistent coexpression of PDGF receptors in the mesothelial layer (Fig. 2D), along with lethality of PDGFR^SKO but not PDGFR^MKO, supports the possibility that PDGF receptor expression in the mesothelium is required for embryo viability.

Because multiple reports have suggested that PDGFRβ is expressed by endothelial cells and to further refine our expression analysis in the yolk sac, we examined PDGFRβ expression in Tie2GFP^Tg mice. These mice express GFP in endothelial cells (59). We found no Tie2-positive cells that expressed detectable levels of PDGFRβ between E8.0 and E10.5 (Fig. 2E and data not shown). Tie2GFP^Tg cells were present in blood islands adjacent to PDGFRβ-expressing mesothelium at E8.5. This expression profile was consistent with a previous report that demonstrated PDGFRβ expression in mesoderm precursors in the yolk sac but not in differentiated endothelial cells (55).

PDGFR^SKO results in incomplete yolk sac capillary bed reorganization.

Embryonic lethality between E9.5 and E10.5 is often a result of cardiovascular failure, but αSMA staining for cardiomyocytes and PECAM staining for endothelial cells revealed that cardiac and embryonic vascular development appeared normal in E10.5 PDGFR^SKO embryos (Fig. 3A to D). No cardiovascular defects were observed in the embryo proper. This lack of a cardiomyocyte phenotype is in agreement with gene deletion analysis of PDGF receptors using an early mesoderm-expressed Cre line, MesP1^Cre (27). By contrast, whole-mount views and histological sections of PDGFR^SKO embryos at E10.5 revealed an apparent cessation of blood vessel maturation within the yolk sac (Fig. 3E and F). While endoderm and mesothelial layers appeared normal, endothelial vessels were distended and disorganized compared to vessels in control embryos. At this time point, we also observed efficient deletion of both PDGFRα and PDGFRβ in PDGFR^SKO yolk sacs (Fig. 3G and H). Analyses of E9.5 and E10.5 mutant yolk sacs indicated no abnormalities in proliferation by bromodeoxyuridine incorporation, and at the same time points, we observed no increase in apoptotic cells, as determined by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling in situ (data not shown).

To assess yolk sac vascular development in our mutants, we performed whole-mount PECAM staining. Vasculogenesis of the yolk sac begins at E7.5, and a number of signaling proteins have been implicated in the process (1). After E8.5, yolk sac vessels undergo a dramatic remodeling event in which the primitive polygonal structure of the vasculature is converted to stable and defined vessels (15). In both control and mutant yolk sacs, we observed the typical honeycomb pattern of endothelial cells at E9.5 (Fig. 4A and B), but at E10.5, PDGFR^SKO yolk sacs retained the characteristics of an immature yolk sac and failed to reorganize into the normal hierarchical array of large and small vessels (Fig. 4G).

To establish if these yolk sac defects were caused primarily by loss of one of the receptors, we analyzed yolk sac vessels by PECAM staining in single PDGF receptor mutants. Yolk sacs from PDGFRα^−/−, PDGFRβ^−/−, PDGFRα^SKO, and PDGFRβ^SKO mutants developed normally, although the remodeled vessels in the E10.5 PDGFRβ^−/− yolk sacs were slightly more disorganized than control vessels (Fig. 4D, I, and J and data not shown). We next analyzed embryos with complete deletion of both receptors to determine if this genotype would phenocopy the PDGFR^SKO embryos. To increase the probability of obtaining double-mutant embryos, we crossed our conditional animals with Meox2^Cre mice. Meox2^Cre expresses Cre in all embryonic tissues and the extraembryonic mesoderm (65). Many of the mutant embryos were being resorbed by E9.5, consistent with the phenotype of PDGF receptor double-homozygous embryos (M. D. Tallquist, unpublished observation), but a few embryos were recovered. They exhibited a complete failure in yolk sac remodeling (PDGFR^Meox2) (Fig. 4E) that resembled the hyperfusion phenotype described previously (12, 13), providing further evidence that loss of both PDGF receptors results in yolk sac vascular abnormalities.

Loss of PDGF receptors results in an increase of endothelial gene expression.

Vascular remodeling is controlled by activities of multiple cell populations, including endothelial (15) and mural (57) cells. To examine the differentiation status of these cell populations, we analyzed yolk sac gene expression by real-time PCR. First, expression analysis demonstrated an expected decrease in PDGFRα and PDGFRβ in the SM22α-Cre^Tg mutants compared to wild-type yolk sacs (Fig. 5A). Consistent with results from Hand1 mutants (45), another yolk sac-remodeling mutant, we observed enhanced expression of endothelial receptors, such as VEGFR1 (Flt1), VEGFR2 (Flk1), and Tie2 (Fig. 5B). Levels of endothelial-cell-specific genes, VEGF, PDGFB, and VE cadherin, were not significantly different between mutant and control samples (Fig. 5B). Nonetheless, elevated levels of endothelial receptors in PDGFR^SKO yolk sacs suggested that impaired vascular development might be inducing an enhanced but nonproductive angiogenic response.

To rule out a direct requirement for PDGF receptors in endothelial cells, we generated embryos that lacked all PDGF receptor expression in endothelial cells using a Tie2Cre^Tg mouse line (PDGFR^EKO) (30). We recovered viable PDGFR^EKO mutants at E12.5 and E15.5, and no obvious defects were observed in yolk sac development (data not shown). Together, these data imply that PDGF receptor signaling is not employed by endothelial cells and that the remodeling phenotype is caused by loss of the receptors in either VSMC or the mesothelium.

The remodeling defect is not caused by loss of PDGF receptor signaling in VSMC.

We next examined the expression of VSMC genes in the PDGFR^SKO yolk sacs. Consistent with the lack of VSMC we observed by lineage tracing (Fig. 1D), VSMC gene expression was reduced. Expression of the gene encoding myocardin-related transcription factor B (MRTFB), a yolk sac smooth muscle transcription factor (70), was reduced in PDGFR^SKO yolk sacs (Fig. 5C). Similarly, expression of the smooth muscle cell cytoskeletal gene αSMA was also reduced (Fig. 5C). SM22α expression, by contrast, exhibited only a partial reduction. This result was anticipated, because mesothelial cells also express SM22α, and they did not appear to be reduced in number (Fig. 1D and data not shown).

These expression data suggested a disruption in the VSMC population. Therefore, we examined E10.5 control, PDGFR^SKO, and PDGFR^MKO yolk sacs for the presence of VSMC. While the control yolk sacs had extensive networks of vessels that contained αSMA-positive cells, in both mutant yolk sac genotypes, few αSMA-expressing cells were present next to endothelial vessels compared to control yolk sacs (Fig. 4L to N). This result suggested that PDGF receptor signaling was required for VSMC formation. To identify if a specific receptor was important for VSMC formation, we examined αSMA staining in null and conditional mutants for PDGFRα and PDGFRβ individually (Fig. 4 and data not shown). In PDGFRβ^−/− and PDGFRβ conditional deletion lines, loss of PDGFRβ led to a dramatic reduction in yolk sac VSMC (Fig. 4O to Q). However, despite the lack of VSMC in these mutants, the yolk sac vasculature was organized into vessels of different calibers. These data suggest that PDGFRβ may be the essential PDGF receptor involved in VSMC development in the yolk sac. However, the presence of normal vessel remodeling in the absence of VSMC indicated that the yolk sac phenotype we observed in PDGFR^SKO yolk sacs was not caused by a failure of VSMC association. The mesothelial cells were therefore implicated as the primary cell population responsible for this phenotype, as they were the only yolk sac cells that expressed Cre in SM22α-Cre^Tg conditional mutants that did not express Cre in myocardin^Cre conditional mutants.

PDGF receptor signaling affects ECM deposition.

Mesothelial cells have a number of proposed functions, including transport of fluids, production of growth factors, and secretion of ECM. Because the vascular defects we observed resembled retinoic acid and TGF-β signaling pathway mutants that are deficient in ECM production (3, 7, 18), we examined the distribution of ECM in wild-type, PDGFR^SKO, and PDGFR^MKO yolk sacs. Staining for fibrillar collagen with picrosirius red (Fig. 6A) demonstrated that PDGFR^SKO yolk sacs had a decreased intensity and thickness of collagen. In contrast, PDGFR^MKO yolk sacs appeared similar to wild-type samples. Immunohistochemistry for collagen 1 and collagen 4 in PDGFR^SKO yolk sacs supported the picrosirius red findings (Fig. 6B and C). The reduction in collagens was limited to the mesothelium, as collagen 4 was detected in close proximity to the endothelial cells. In contrast, fibronectin and laminin appeared relatively unperturbed at this stage (Fig. 6D and E). To determine if the reduction in collagen was at the transcriptional level, we performed real-time PCR analysis. We detected only modest changes in transcript levels of collagens and fibronectin (Fig. 6F).

Because collagen synthesis was only moderately affected, we investigated other processes that would explain a reduction in collagen protein levels. One candidate mechanism was an increase in MMP activity. MMPs are a family of proteinases that are capable of degrading a wide variety of ECM proteins. When we examined yolk sacs using an MMP in situ assay, we detected increased MMP activity in mutant mesothelium compared to controls (Fig. 6G). Similarly, gelatin zymography consistently demonstrated higher levels of activated MMP-2 in PDGFR^SKO mutant yolk sacs than in control yolk sacs (Fig. 6H). Taken together, these data suggest that PDGF receptor signaling from the mesothelium may function to direct blood vessel remodeling in part by controlling the degradation of matrix in the yolk sac.

PDGF receptor signaling controls blood vessel morphogenesis.

To further examine the role of PDGF receptors in vascular development, we used an allantois culture assay. E8.5 allantoides from wild-type embryos develop rudimentary vascular structures, but when stimulated with either 10% FBS or PDGFBB, a ligand that can activate both PDGFRα and PDGFRβ, the vascular plexus expands (Fig. 7A and C). In addition, we showed that PDGF receptor signaling in these cultures was required for stabilization of the vessels. Using a Cre-expressing adenovirus, we were able to induce recombination in allantois explants from embryos homozygous for both PDGFRα and PDGFRβ floxed alleles (PDGFR^fl/fl). When both PDGF receptors were deleted using Cre, vascular expansion was severely reduced (Fig. 7B and C). Quantification of vascular expansion demonstrated that PDGF receptor stimulation yielded cultures that were comparable to those produced by serum stimulation (Fig. 7C). Similarly, addition of a PDGF receptor-specific inhibitor (AG1296) to the cultures resulted in a lack of vascular expansion similar to that of unstimulated cultures (Fig. 7D). We could not examine the effects of the PDGF receptor inhibitor on PDGFBB-stimulated cultures, as these cultures did not adhere to the coverslip. To determine the effect of exogenous addition of matrix, we plated allantois cultures on collagen 4 and found that exogenous collagen 4 resulted in increased vascular expansion in 1% serum. Treatment with collagen 4 was even capable of bypassing the effects of PDGF inhibition on cultures stimulated with either 10% FBS or PDGFBB (Fig. 7D and data not shown).

Finally, to examine how reduced ECM could affect endothelial cell signaling, we determined if integrin function within the yolk sac was disrupted. To accomplish this, we examined the activation status of integrin β1. Integrin β1 is essential for endothelial cell function (5) and is one of the key integrins in endothelial cell morphogenesis. Phosphorylation of integrin β1 on S785 modulates cell adhesion and migration. Using a phosphospecific antibody for S785 of β1 integrin (46), we found that phosphorylation on S785 was reduced, even though levels of β1 integrin were similar in control and PDGFR^SKO yolk sacs (Fig. 7E). Taken together, these data suggest that loss of PDGF receptor signaling leads to reduced vascular remodeling that may be related to a loss of matrix integrity.