Deficiency in the p110α subunit of PI3K results in diminished Tie2 expression and Tie2^-/-–like vascular defects in mice

Phosphoinositide 3-kinases (PI3Ks) are important mediators of many transmembrane signaling receptors, including tyrosine kinases and 7-transmembrane G-protein–coupled receptors. The class Ia PI3K (abbreviated as PI3K from here on) typically contains p85/p55 as a regulatory subunit and p110α, p110β, or p110δ as the catalytic subunit, and their activities are regulated by Ras and tyrosine kinases (for a review, see Fruman et al¹ ). The p110α catalytic subunit is widely expressed in a variety of cell types, including endothelial cells.^2-6  Numerous studies have demonstrated that PI3K plays important roles in regulating endothelial proliferation, migration, and survival.^7-12  Although these studies indicated that the transmembrane tyrosine kinase Flk-1/VEGF-A receptor-2/KDR plays a crucial role in regulating PI3K activity in endothelial cells,^9,13  Tie2 (tunica intima endothelial kinase 2), a tyrosine kinase receptor for angiopoietin-1 (Ang-1) and related molecules, also signals through PI3K to regulate endothelial properties.^14-19 

Targeted disruption of the gene encoding the p110α subunit (Pik3ca) of PI3K in mice resulted in embryonic lethality in homozygous mutants between embryonic day 9.5 (E9.5) and E10.5,²⁰  but it is unclear whether PI3K is important for vascular development. In this paper, we report that these mice experience significant reductions in Tie2 expression and multiple vascular defects that are highly reminiscent of those in Tie2 mutant mice.

Gene targeting for Pik3ca, which encodes p110α, has been described previously.²⁰  Stocks of heterozygous mice (Pik3ca^+/del [Pik3ca^+/-]) were maintained in both 129X1 and CD1 strains, but embryos from CD1 background were used in this study. Genotypes were determined by polymerase chain reaction (PCR) of tail DNA samples using oligonucleotide primers: 5′-CAA GCC TGG GGA GAT GGT CA-3′ and 5′-CCC GAA GAT GGT CGT GGA GG-3′ for wild-type allele (431 base pairs [bp]); 5′-CAA GCC TGG GGA GAT GGT CA-3′ and 5′-CCC GAA GAT GGT CGT GGA GG-3′ for the mutant allele (526 bp). To introduce an endothelial cell–specific reporter gene, Pik3ca^+/- mice were crossed with transgenic mice expressing LacZ under the control of Tie2 regulatory sequences²¹  (T. Sato/Jackson Laboratory, Bar Harbor, ME). X-gal and immunohistochemical (IHC) staining reactions were performed as described.^22-24  Primary antibodies included anti–Flk-1 (clone Avas 12a1) (BD Biosciences, PharMingen, San Diego, CA), anti-CD31 (clone Mec 13.3) (BD Biosciences), and anti-Tie2 (clone TEK4, raised against mouse Tie2 extracellular domain) (eBioscience, San Diego, CA).

For Western blotting, embryos proper or yolk sacs were lysed in RIPA buffer (1 × phosphate-buffered saline [PBS]; 1% NP-40; 0.5% sodium deoxycholate; 0.1% sodium dodecyl sulfate [SDS]) containing protease inhibitors (Complete; Roche Applied Science, Indianapolis, IN). Proteins from cleared lysates were subject to Western blotting analyses with the following antibodies: anti-Tie2 (clone 33, raised against human Tie2 extracellular domain) (BD Biosciences), antiphospho-Akt (Cell Signaling Technology, Beverly, MA), anti-Akt (BD Biosciences), or anti–β-actin (Santa Cruz, Santa Cruz, CA). Signals were revealed using horseradish peroxidase (HRP)–conjugated secondary antibodies and chemoluminescence detection system (Amersham Biosciences, Piscataway, NJ). Band intensities were determined using the National Institutes of Health (NIH) Image program.

For p110α knockdown experiments, MS1 endothelial cells²⁵  (American Type Culture Collection [ATCC], Manassas, VA) were cultured in Dulbecco modified Eagle medium (DMEM; Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS; HyClone, Logan, UT) and were transfected with specific siRNA directed against mouse p110α (SmartPool; Dharmacon, Lafayette, CO). The specificity of p110α SmartPool siRNA was established by the supplier. In addition, we included LacZ siRNA as a control for nonspecific effects. Transfections were performed using Lipofectamine 2000 (Invitrogen) following the manufacturer's recommendations. Seventy-two hours after transfection, cells were lysed in ice-cold RIPA buffer as described for embryonic tissues. Proteins in cleared lysates were analyzed by Western blotting. Anti-p110α was affinity purified rabbit polyclonal antibodies from Cell Signaling Technology (Beverly, MA).

In CD1 strain background, Pik3ca^-/- embryos were similar to those in 129X1, but they survived a few hours longer and were still viable at E9.5, as indicated by beating hearts. Dissection at E9.5 first revealed that yolk sacs of Pik3ca^-/- embryos were invariably pale, and blood congestion was apparent in the head (Figure 1A-D). To document vascular defects in Pik3ca^-/- mice, we visualized the vascular system by X-gal staining after crossing the Pik3ca^+/- mice with a transgenic mouse line expressing the Escherichia coli lacZ reporter gene under the control of Tie2 promoter/enhancer,²¹  which is essentially endothelial specific. In Figure 1E and F, we show that normal and mutant embryos developed dorsal aortae (red arrowheads). In Figure 1E, the anterior cardinal vein (ACV) is clearly visible (white arrowheads). In the mutant, only patches of X-gal–positive signals are seen where the ACV normally forms (Figure 1F), suggesting incomplete development of the ACV. In transverse sections of normal embryos, ACVs are seen as prominent lumens flanking the dorsal aortae (Figure 1G). In mutants, ACV lumens are significantly smaller (Figure 1H). These observations further confirm that ACV was incompletely formed in Pik3ca^-/- embryos. Defective development of the ACV may be responsible for the accumulation of blood in the head because blood return from the head to the atrium may be blocked as a result of ACV abnormality.

X-gal staining also revealed a significant reduction in endocardial branching resulting from the Pik3ca^-/- mutation (Figure 1I-J). We compared a large number of serial sections and found that endocardial lumens appeared highly complex in normal embryos but that they were much simpler in mutants (Figure 1K-L). This observation is consistent with the whole-mount staining data shown in Figure 1I and J, which demonstrate that the difference in endocardial organization is global throughout the heart. In the yolk sac, the differentiation of large vessels failed in mutant embryos (Figure 1M-N), and lumens were generally small in histologic sections (Figure 1O-P). In addition to these defects, vessels in the forebrain region were dilated (Figure 1Q-R), and the expression of the LacZ reporter was significantly reduced at the anterior end of Pik3ca^-/- embryos (Figure 1S-T).

Morphologic manifestations of Pik3ca^-/- mice were strikingly similar to those in Tie2^-/- embryos,^22,26  or embryos in which Tie2 signaling is disrupted by a targeted disruption of angiopoietin-1 or by the overexpression of Ang-2.^23,24  Therefore, we wondered whether Tie2 expression was compromised in p110α-deficient embryos. Because Tie2–LacZ reporter gene expression was similar between Pik3ca^-/- mutants and controls (except for the forebrain region), Tie2 promoter activity was probably not drastically affected by Pik3ca^-/- mutation. Consistent with this notion, we were unable to detect significant reductions of Tie2 expression by multiple repeats of semiquantitative reverse transcription–PCR (RT-PCR) (data not shown). However, from more than 40 embryos examined by anti-Tie2 immunohistochemical staining, we found that though blood vessels in wild-type and heterozygous embryos were consistently positive for staining, none of the Pik3ca^-/- embryos showed vascular-specific positive staining (Figure 2A-D), except for some ventral tissues at the posterior end. A similar observation was also made in yolk sacs (Figure 2E-H). To exclude the possibility that reduced Tie2 protein level in Pik3ca^-/- embryos might be caused by a nonspecific and global failure of gene expression in mutant endothelial cells, we performed anti–Flk-1 and anti-CD31 staining. These antibodies stained positively in normal and mutant embryos, and examples from anti–Flk-1 staining are shown in Figure 2I to L.

To further evaluate the effect of p110α deficiency on Tie2 protein levels, we analyzed yolk sacs and embryos by Western blotting. As shown in Figure 3A, Tie2 levels were significantly lower in p110α-deficient samples (42% and 31% of normal levels in the yolk sac and embryo proper, respectively). In addition, we also verified that p110α deficiency did not affect the expression of Akt but resulted in significantly reduced phosphorylation of Akt, suggesting that p110α is the major isoform of catalytic subunit mediating PI3K activity in E9.5 mouse embryos. To further establish the relationship between defective PI3K signaling and reduced Tie2 expression, we analyzed the effect of siRNA-mediated knockdown of p110α in cultured endothelial cells. In MS1 cells transfected with 80 pmol/mL of p110α-specific siRNA, p110α was expressed only at 16% of the normal level, indicating successful knockdown. At 20 pmol/mL, p110α was expressed at 34%. The levels of Tie2 protein correlated with the abundance of p110α and were present at 37% and 72% of normal levels, respectively.

These data clearly indicate that PI3K in endothelial cells is required for normal expression of Tie2, an important regulator for vascular development. PI3K might directly regulate the expression or activity of a factor required for the control of Tie2 protein level. Alternatively, PI3K might be a regulator of another signaling pathway that influences Tie2 expression. In cultured endothelial cells, many studies have shown that PI3K is a downstream effector of Tie2 signaling.^14-19  However, our observation does not contradict a role for PI3K as a Tie2 effector. Our data do not exclude the intriguing possibility that PI3K might help maintain a normal Tie2 protein level by its interaction with Tie2. The dual roles of PI3K in downstream Tie2 signaling and in maintaining Tie2 protein levels may help explain the development of Tie2^-/-–like vascular defects in Pik3ca^-/- mice. Given the severity of developmental defects and early embryonic lethality of p110α-deficient embryos, additional mechanisms are likely involved as well. For example, of 4 classes of vascular defects found in p110α-deficient mice, 2 of them (yolk sac defects and lack of ACV) are also present in Notch1^-/-/Notch4^-/- double knockout mice.²⁷  Further studies are needed to investigate the relevance between these individual signaling pathways and the observed vascular defects in PI3K-deficient mice.

We thank Nancy Ryan for preparation of the histology sections.

Dr Etienne Lelievre's current address is CNRS UMR 8526, Institut de Biologie de Lille, 1 rue Calmette, 59021 Lille Cedex, France.