Context-dependent functions of angiopoietin 2 are determined by the endothelial phosphatase VEPTP

ANGPT2–TIE2 Signaling Is Essential for Embryonic Lymphangiogenesis.

To identify the molecular basis of the differential functions of ANGPT2 in LECs and BECs, we characterized the role of ANGPT2–TIE2 signaling in lymphatic development, where ANGPT2 has a well-defined agonistic role (22–24). As expected, whole-body Tie2 deletion from conception using the Rosa26rtTA;tetOCre bitransgenic system in mice harboring a Tie2 conditional by inversion (COIN) allele (Tie2^WBΔE0) resulted in embryonic lethality between E9.5 and E10.5 (17, 18). However, embryos induced at E12.5 were found to survive until late gestation, allowing analysis of the lymphatic vasculature. At E16.5, all embryos were found alive. However, subcutaneous edema was observed in Tie2^WBΔE12.5 knockout embryos, which was never observed in Tie2 wild-type or heterozygous controls (Fig. 1A). This result was recapitulated in lymphatic-specific Tie2 knockouts generated using Prox1CreER (Tie2^LymΔE10.5 mice), indicating that the edema was lymphatic in origin (Fig. 1B).

Compared with control littermates, immunostaining revealed a paucity of PROX1⁺ lymphatic vessels in the dorsal skin of Tie2^LymΔE10.5 knockout mice at E14.5, confirming the importance of lymphatic-expressed TIE2 in lymphangiogenesis (Fig. 1C). Interestingly, unlike the phenotype observed in lymphatic-specific knockout embryos, dorsal skin lymphatics of whole-body knockouts were enlarged compared with littermate controls (Fig. S1A). This enlargement could be due to the combined deleterious effects of TIE2 deletion on BECs and LECs.

To determine the role for each ANGPT ligand in dermal lymphangiogenesis, we compared the phenotypes of either single or combined deletion of Angpt1 and Angpt2 genes with Tie2 conditional knockout mice. Unlike the marked edema of Tie2^WBΔE12.5 embryos, single Angpt1 or Angpt2 whole-body knockout embryos induced at E12.5 had no apparent edema when dissected at E16.5 (Fig. 1D and Fig. S1B). The fact that loss of the TIE2 receptor led to a more severe phenotype than loss of either ligand alone suggested compensation or cooperative roles of the two ANGPT ligands. To test this possibility, we generated compound mutants lacking both Angpt1 and Angpt2 from E12.5 onward. This simultaneous loss of ANGPT1 and ANGPT2 expression (A1;A2^WBΔE12.5) recapitulated the phenotype of Tie2^WBΔE12.5 mice (Fig. 1D and Fig. S1B) and embryos exhibited marked edema.

ANGPT2–TIE2 Signaling Is Essential for Mesenteric Lymphatic Development.

Angpt2 has a well-described role in the mesenteric and intestinal lymphatic vasculature. Because Tie2 whole-body or lymphatic knockout embryos induced at E12.5 were not viable, we tested whole-body deletion at a range of time points (22–24). Deletion at E13.5 or later resulted in viable mutant offspring. Tie2^WBΔE13.5 knockout pups exhibited chylous ascites with severely disturbed lymphatic vessel morphology, indicating defects of mesenteric lymphatic function (Fig. S2). We then examined the role of each ANGPT ligand in this TIE2-mediated developmental process. Whole-body Angpt1 knockout pups induced at E13.5 had no apparent phenotype, but chylous ascites were observed in mice lacking Angpt2 alone or both Angpt1 and Angpt2 (Fig. S2B). Deletion of Tie2 after E15.5 did not result in overt chylous ascites (Fig. S2A), although a reduced number of lymphatic valves was observed in the mesentery (Fig. S3).

To better understand the role of ANGPT2 in mesenteric lymphatic development, we utilized a cell-type–specific approach to delete Angpt2 in endothelial cells (24). In contrast to the well-developed lymphatic vessels in control mice, endothelial deletion of Angpt2 using a lymphatic-expressed Lyve1-Cre (Angpt2^ΔLyve1Cre) resulted in severely disturbed lymphatic vessel morphology with leakage of chyle, phenocopying the Tie2^WBΔE13.5 knockout (Fig. S2D). Collectively, these results demonstrate a requirement for ANGPT2–TIE2 signaling in mesenteric lymphatic development.

A Regulatory Phosphatase for TIE2, VEPTP, Is Absent from Lymphatic Endothelium.

TIE1 and VEPTP are both known to modulate activation status of TIE2 and are expressed in BECs (1–4). However, while TIE1 is expressed in LECs in vivo where it is required for lymphatic development (26, 31), VEPTP expression in LECs in vivo has not been reported, although it has been reported in cultured human dermal LECs (32, 33). Using a Veptp reporter mouse line (30), we determined the expression pattern of the phosphatase. Both β-galactosidase expression and its activity were strongly detected in BECs, but not in PROX1 or LYVE1⁺ lymphatics of the embryonic dorsal skin, neonatal mesentery, adult ear dermis, or adult ocular limbus (Fig. 2A and Fig. S4). Interestingly, ANGPT1-producing cells are closely associated with CD31⁺ blood vessels but not with NRP2⁺ lymphatic vessels in embryonic skin (Fig. 2B). In contrast, Angpt2 is expressed in NRP2⁺ dermal lymphatic vessels (Fig. 2C).

VEPTP Abrogates ANGPT2 Agonistic Activity on TIE2.

Given the striking difference in expression pattern of VEPTP in blood vs. lymphatic vasculature in mice in vivo and its negative regulatory role on TIE2, we reasoned that VEPTP might block ANGPT2 agonistic function in BECs. To test this hypothesis, we characterized the interaction of TIE2 and VEPTP in a heterologous cell model using HEK293 cells, where endogenous expression of both proteins was absent. In cotransfection experiments, VEPTP effectively reduced autophosphorylation of TIE2. Phosphorylation was restored by treatment with a small-molecule inhibitor of VEPTP, AKB-9785, or with recombinant human (rHu) ANGPT1 (Fig. 3 A and B). VEPTP also reduced the phosphorylation of a TIE2 gain-of-function mutant (R849W) identified in patients with hereditary venous malformations (16), confirming its high enzymatic activity (Fig. S5A). TIE2 and VEPTP form a stable complex when transfected in cells and they reciprocally regulate each other, as evidenced by the TIE2-dependent phosphorylation of catalytically inactive VEPTP (Fig. S5 B and C).

Receptor tyrosine kinases signal through trans- and autophosphorylation of tyrosine residues (34). To better understand the regulation of TIE2 by VEPTP and TIE1, we performed phosphoproteomic analysis of the full-length TIE2 and TIE1 receptors in cells. Overall, mass spectrometry detected peptide fragments covering all but Y1024 of the 19 intracellular tyrosine residues on TIE2 (Fig. 3C). Among the 18 tyrosine residues, 13 were phosphorylated at varying levels, including the C-terminal Y1102 and Y1108 that are known to recruit downstream signaling adaptors, such as p85 of PI3 kinase and DokR (35–37). Juxta-membrane Y816, which has been reported to recruit Shp2 and Grb14 for signaling, was also phosphorylated (35). Next, we tested how VEPTP coexpression modulates baseline TIE2 phosphorylation levels on individual tyrosine sites. We observed marked reduction of all phosphorylation in the presence of VEPTP, highlighting the broad impact of this phosphatase to overall TIE2 signal strength (Fig. 3C).

TIE1, a homolog of TIE2, is an orphan receptor tyrosine kinase with no known ligand (1, 3). TIE1 has been shown to interact with TIE2 and its importance both in lymphangiogenesis and angiogenesis has been well-characterized (26, 31, 38, 39). Recently, extracellular cleavage of TIE1 has been shown to play a role in ANGPT2-mediated TIE2 antagonism in BECs during inflammation (40). Our heterologous expression system showed that TIE1 maintained low levels of autophophosphorylation compared with TIE2 (Fig. S6A). Phosphoproteomic analysis of TIE1, either with or without TIE2, revealed that TIE1 phosphorylation is induced by TIE2 coexpression (Fig. S6B). Conversely, when a TIE2 kinase dead mutant protein was coexpressed with wild-type TIE1, the kinase-dead TIE2 became robustly phosphorylated (Fig. S6C), indicating reciprocal cross-talk between these kinases.

Ligand-induced TIE2 activation was studied using stable cell-lines expressing either TIE2-FLAG alone or together with VEPTP-GFP in HEK293 cells (Fig. S7A). Treatment with rHuANGPT1 and rHuANGPT2 increased TIE2 phosphorylation in TIE2-expressing cells in a dose-dependent fashion (Fig. 3 D and E). However, in the presence of VEPTP, only rHuANGPT1 was found to activate TIE2, while treatment with rHuANGPT2 had no effect (Fig. 3 D and E).

VEPTP Inhibition Restores ANGPT2 Agonistic Activity on TIE2.

In a variety of disease conditions, circulating ANGPT2 levels increase, leading to elevated ANGPT2:ANGPT1 ratios (2, 7, 19, 41). We wondered if VEPTP inhibition might enable ANGPT2 to become a TIE2 agonist in BECs as seen in LECs. VEPTP inhibition alone or in combination with rHuANGPT2 treatment increased TIE2 phosphorylation in TIE2/VEPTP-expressing stable cells, whereas rHuANGPT2 alone did not (Figs. 3E and 4A). To determine if ANGPT2 enhances TIE2 signaling above the effect of VEPTP inhibition alone, we tested the downstream signaling activity in response to either ANGPT2 or VEPTP inhibition alone or in combination in primary and transformed endothelial cell-lines. Human umbilical vein endothelial cells (HUVECs) and EA.Hy926 cells express abundant TIE2 and VEPTP (Fig. S7B). Treatment with rHuANGPT1, but not with rHuANGPT2, increased pAKT levels in these cell lines (Fig. 4B and Fig. S7C). In contrast, rHuANGPT2 increased pAKT abundance in HUVECs, which were also treated with VEPTP inhibitor (Fig. 4C and Fig. S7D). Importantly, the level of AKT phosphorylation was greater in cells treated with both rHuANGPT2 and VEPTP inhibitor than with VEPTP inhibitor alone. Further downstream, ANGPT1-mediated increases in pAKT signaling are reported to cause forkhead box O1 (FOXO1) phosphorylation, leading to its nuclear exclusion (40, 41). This finding was reproduced in our model, where rHuANGPT1 induced nuclear FOXO1 protein to translocate to the cytoplasm, while rHuANGPT2 did not (Fig. 4D). However, consistent with our pAKT findings, treatment with VEPTP inhibitor alone or VEPTP inhibitor in combination with rHuANGPT2 also markedly reduced nuclear accumulation of FOXO1 (Fig. 4D).

ANGPT1 reinforces vascular junctions and stabilizes blood vasculature by initiating TIE2 phosphorylation and activating downstream signaling networks, but also by physically bridging TIE2 receptors between juxtaposed cells (42, 43). Immunocytochemical analyses revealed TIE2-FLAG protein localized at cellular junctions following stimulation with either rHuANGPT1 or rHuANGPT2, but not VEPTP inhibitor, as previously reported (28, 42). This ANGPT2-induced cellular junctional localization was not affected by VEPTP inhibition (Fig. S7E). Collectively, our results have demonstrated that VEPTP inhibition in conjunction with ANGPT2 stimulation activates TIE2 signaling and drives translocation of TIE2 to the cell junctions in BECs, mimicking the actions of ANGPT1.

Animals.

The mouse lines used for our study have been previously described (7, 17, 18, 24, 30, 56–58). Whole-body timed deletion of target genes was achieved by using a bigenic Rosa26rtTA;tetOCre system, as previously described (7, 17, 18). The transgenic mouse lines were maintained on a mixed background due to the large number of transgenes required. However, littermate controls were used for all phenotype analyses. Full details of mouse analysis are described in SI Materials and Methods.

Cell Culture Experiments and Phospho-Proteomics.

The cells were cultured with standard methods and stimulated with rHuANGPT1 (R&D Systems), rHuANGPT2 (R&D Systems), and AKB-9785 (a VEPTP inhibitor) (54) for 30 min at 37 °C unless otherwise mentioned. For phosoho-mapping analysis, HEK293 cells were transfected with plasmid vectors for expressing TIE2-FLAG, TIE1-FLAG, and VEPTP-GFP, either alone or in combination. The proteins in corresponding SDS gel pieces were digested with trypsin and chymotrypsin, and digested peptides were analyzed with LC-MS/MS. Abundance of phophorlyation sites were semiquantitatively compared across the samples using spectral counting (number of peptide spectrum matching or PSM) (59). Full details of analysis are described in SI Materials and Methods.

Statistics and Reproducibility.

Results are expressed as means ± SEM. Statistical analysis was carried out using two-tailed Student’s t test or one-way ANOVA followed by Tukey–Kramer’s test or Dunnett's correction for multiple comparisons using GraphPad Prism software. A P value less than 0.05 was considered as statistically significant.

Study Approval.

All animal experiments were approved by the Animal Care Committee of Mount Sinai Hospital, University of Toronto, Toronto and the Institutional Animal Care and Use Committee of the Center for Comparative Medicine at Northwestern University, Chicago.