Src-mediated Post-translational Regulation of Endoglin Stability and Function Is Critical for Angiogenesis*

Background: Endoglin overexpression promotes angiogenesis but the mechanism of endoglin down-regulation is largely unknown. Results: Endoglin YIY motif is phosphorylated by Src and induces receptor down-regulation. Conclusion: The YIY motif is an endocytic signal for endoglin turnover. Significance: Given that endoglin is a vascular target, defining how endoglin expression is post-translationally regulated is crucial for anti-angiogenic therapies. Endoglin is a transforming growth factor β (TGF-β) co-receptor essential for angiogenesis and tumor vascularization. Endoglin modulates the crucial balance between pro- and anti-angiogenic signaling by activin receptor-like kinase (ALK) 1, 5, and TGF-β type II (TβRII) receptors. Despite its established role in physiology and disease, the mechanism of endoglin down-regulation remains unknown. Here we report that the conserved juxtamembrane cytoplasmic tyrosine motif (612YIY614) is a critical determinant of angiogenesis. Src directly phosphorylates this motif to induce endoglin internalization and degradation via the lysosome. We identified epidermal growth factor (EGF) and vascular endothelial growth factor (VEGF) as Src-activators that induce endoglin turnover following 612YIY614 phosphorylation. Interestingly, Src phosphorylation of endoglin-612YIY614 was also an important process for receptor down-regulation by TRACON105 (TRC105), an endoglin-targeting antibody currently in clinical trials. The regulation of 612YIY614 phosphorylation was critical for angiogenesis, as both the phosphomimetic and unphosphorylatable mutants impaired endothelial functions including proliferation, migration, and capillary tube formation. Collectively, these findings establish Src and pro-angiogenic mitogens as critical mediators of endoglin stability and function.

Endoglin expression is markedly elevated in actively proliferating ECs during angiogenesis and vascular remodeling (10). Many growth factors or conditions known to promote tumor progression also stimulate endoglin gene expression, including TGF-␤, hypoxia, and inflammation (12). Indeed, its robust expression serves as a highly sensitive marker of tumor vascularization, and strongly correlates with tumor growth, metastases, and poor overall prognosis (11,14,15). Endoglin is also recognized as a highly effective vascular target, as demonstrated by TRC105, the first humanized endoglin antibody used in clinical trials for treatment of advanced or metastatic tumors (16 -18). But despite recent advances, many facets of endoglin biology are still poorly understood at the molecular and cellular level. In particular, while extensive efforts have aimed at suppressing endoglin function to block tumor vascularization, the normal post-translational mechanism(s) mediating endoglin down-regulation remains virtually unexplored.
Previous endoglin structure-function studies reveal a large extracellular ligand-binding domain that binds to BMP-9/10, while TGF-␤ and other structurally-related family members including activins, can bind to this domain in a heteromeric complex with ALK1, T␤RII, BMP, or activin receptors (2,12). The extracellular domain is linked to a single transmembrane segment, followed by a short cytoplasmic domain that serve as key docking sites for a number of signaling/trafficking adaptor partners including zyxin, Tctex2␤, and ␤-arrestin2 (12, 19 -22). Many serine/threonine phosphorylation sites within this domain are associated with endoglin internalization. Throm-bin, for instance, has been shown to inhibit endoglin serine phosphorylation and induce receptor endocytosis via proteaseactivated receptor 1 and protein kinase C activation (23). Moreover, ␤-arrestin2 is thought to bind a phosphorylated threonine residue at position 650 to cause receptor internalization (19). Still, while endoglin serine/threonine phosphorylation appears to have direct roles in endocytic trafficking and downstream signaling, endoglin expression level remains unaffected (19,23).
Interestingly, in addition to the many serine/threonine phosphorylation and adaptor-binding sites interspersed throughout the C-terminal tail, there is an evolutionarily conserved peptide sequence containing two key tyrosine residues at positions 612 and 614 ( 612 YIY 614 ), located immediately distal to the transmembrane segment for which no functional role has been assigned. Given such a close proximity to the plasma membrane, we hypothesized that these tyrosine residues, if phosphorylated, could induce important local conformational changes that affect endoglin function, or serve as platform for tyrosine kinase signal transduction complexes.
Here we present evidence that the conserved tyrosine motif is a critical determinant of endoglin stability and angiogenesis, and further demonstrate that TRC105 requires this motif to efficiently down-regulate endoglin from the cell surface.
Immunoprecipitation-Cells were washed briefly and then lysed on ice with lysis buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM EDTA, 10 mM NaF, 10% (w/v) glycerol, 1% Nonidet Nonidet P-40) and supplemented with protease inhibitors (Sigma protease inhibitor mixture) and phosphatase inhibitors (Sigma phosphatase inhibitor mixture). The lysates were precleared by centrifugation and incubated with appropriate primary Abs for 2-4 h, and then with protein agarose G/A for 1-2 h at 4°C. The immunoprecipitates were collected by centrifugation; pellets were washed with lysis buffer, and stored in 2ϫ sample buffer before Western blot analyses. For soluble endoglin, endoglin was immunoprecipitated from the conditioned media with extracellular-targeting antibody (TRC105 or P3D1).
Immunofluorescence-EngϪ/Ϫ ECs or COS7 cells grown overnight on coverslips were transiently transfected with appropriate constructs using Lipofectamine 2000 (Invitrogen) as described. For ubiquitin staining, cells were treated with MG132 (10 M) for 2 h prior to fixation. For TRC105-induced endoglin internalization, ECs were pre-treated with TRC105 (200 ng/ml) for the indicated duration prior to fixation. 24 -48 h following transfection, cells were washed with PBS and then fixed with 4% paraformaldehyde. Cells were permeabilized in 0.1% Triton X-100 in PBS for 3-10 min, then blocked with 5% bovine serum albumin in PBS containing 0.05% Triton X-100 for 20 min. All primary Abs were incubated at room temperature for 1 h unless noted otherwise. Eng-WT, Eng-YE, and Eng-YF expression was detected using TRC105. EEA1, LAMP1, and ubiquitin Abs were used to detect early endosomes, lysosomes, and ubiquitin clusters, respectively. Proteasome was detected using 20 S Proteasome ␣6. ALK1, T␤RII, and ALK5 were detected using ALK1, T␤RII, and HA Abs, respectively. Following primary antibody incubation, cells were incubated with appropriate flurophore conjugated secondary antibodies (Alexa-Fluor) at room temperature for 30 min. Cells were costained with DAPI (Sigma) immediately before immunofluorescence microscopy analyses (Olympus FV1000 confocal system). Pearson correlation coefficient analysis was used with ImageJ to measure co-localization.
Transwell Migration Assays-Engϩ/ϩ and EngϪ/Ϫ ECs were transfected with appropriate endoglin constructs. 24 h following transfection, cells were seeded in the upper chamber of a transwell filter in complete growth media, coated both at the top and bottom with gelatin. Cells were allowed to migrate for 12 h toward the lower chamber containing growth media alone or growth media containing EGF (5 ϫ 10 Ϫ4 mg/ml) or VEGF (5 ϫ 10 Ϫ4 mg/ml). Cells that migrated to the bottom surface of the filter were fixed, stained, and then digitally imaged and counted.
Crystal Violet Cell Growth Assay-EngϪ/Ϫ ECs were plated at 15,000 in 12-well plates and transfected with the appropriate endoglin constructs. Following transfection, cells were fixed at different time points (4% paraformaldehyde in PBS for 15 min). Following fixation, cells were washed with 1ϫ water and stained with 0.1% crystal violet for 20 min. Cells were washed 3ϫ with water and allowed to air dry for 30 min. Cells were destained using crystal violet destaining solution (10% acetic acid, 50% methanol, 40% H 2 0) for 20 min, and the optical density was read at 590 nm in a microplate reader.
Endothelial Tube Formation-EngϪ/Ϫ ECs were transfected with the appropriate endoglin constructs. 24 h following transfection, cells were trypsinized and plated on a 24-well plate coated with 200 l of matrigel basement matrix (BD Biosciences) at 160,000 cells/well. 1 h following plating, growth medium was removed and 200 l of matrigel basement matrix was added. 30 min following the addition of the matrigel base-ment matrix, 300 l of growth medium was added. Endothelial tubes were digitally imaged and quantified by measuring the relative tube length and counting the number of branches per node. Tube length and number of branches per node for each endoglin construct were normalized to the value in Eng-WT.

RESULTS
There are 8 tyrosine residues found in the human endoglin peptide sequence, 6 of which reside in the extracellular domain and two located immediately distal to the transmembrane segment (Tyr 612 and Tyr 614 ) (Fig. 1A). Sequence alignment reveals that this intracellular tyrosine motif (WY 612 IY 614 ) of unknown function is evolutionarily conserved, and is present in another type III TGF-␤ co-receptor, T␤RIII (␤glycan) (Fig. 1B). To functionally test this motif, we generated cDNAs encoding a phosphomimetic (both Tyr residues to Glu; Eng-YE) and a phenylalanine mutant and thus not subject to phosphorylation FIGURE 1. Endoglin tyrosine motif is an endocytic signal for down-regulation. A, endoglin schematic shows the extracellular domain containing 6 tyrosine sites, a transmembrane segment, and an intracellular domain containing 2 tyrosine residues as indicated. The two intracellular tyrosine residues were mutated to glutamate (Eng-YE), phenylalanine (Eng-YF), or alanine (Eng-YA). B, peptide sequence alignment shows the conserved intracellular tyrosine motif (highlighted) across species for endoglin and with T␤RIII. C, a representative Western analysis of endoglin expression in EngϪ/Ϫ ECs (upper panel). A quantitative densitometry of relative endoglin expression normalized to Eng-WT and ␤-actin. Statistical analysis was based on four independent experiments. *, p Ͻ 0.05. D, endoglin localization upon expression of human Eng-WT, Eng-YE, Eng-YF, or Eng-YA in EngϪ/Ϫ ECs and stained for endoglin (green) using TRC105. DAPI co-staining (blue) indicates non-transfected EngϪ/Ϫ ECs. E, Western analysis of endoglin cleavage upon immunoprecipitation of sEng from conditioned media of EngϪ/Ϫ ECs expressing Eng-WT, Eng-YE, or Eng-YF.
Immunofluorescence microscopy revealed that both Eng-WT and Eng-YF had membrane and cytoplasmic distribution typically observed for TGF-␤ receptors including endoglin, whereas Eng-YE displayed an atypical punctate endocytic profile (Fig. 1D). To rule out the possibility of protein aggregation of Eng-YE due to structural perturbations, we tested the expression of endoglin that harbors dual alanine substitutions at the YIY motif (both Tyr to to Ala; Eng-YA). The Eng-YA expression pattern was quite similar Eng-WT and Eng-YF, indicating that the phosphomimetic mutation likely alters Eng-YE expression. Given that the membrane-bound endoglin can undergo ectodomain cleavage by the membrane-anchored matrix metalloproteinase-1 (MMP-14) to release soluble endoglin (sEng) (26), we compared sEng production derived from Eng-WT, Eng-YE, and Eng-YF-expressing EngϪ/Ϫ ECs to determine whether Eng-YE is capable of properly folding and trafficking to the cell surface. Here, sEng production was detected from the conditioned media of all three forms, albeit to a slightly lesser extent from Eng-YE-expressing cells (Fig. 1E). As no secreting isoforms of endoglin exist, these results indicated that Eng-YE is capable of trafficking to the cell surface for ectodomain cleavage despite a large localized subset in endosomal compartments.
To determine the subcellular characteristics of the Eng-YEcontaining endosomes, we co-stained Eng-YE with a panel of endosomal markers, including the early endosome (EEA1), late endosome (Rab-9), lysosome (LAMP1), ubiquitin, and the proteasome (26 S) (Fig. 2). Here, the Eng-YE-containing endocytic vesicles showed prominent co-localization with the early (EEA1) and late endosomes (Rab9), as well as the lysosome (LAMP1) (Fig. 2, A-C, E), whereas minimal co-localization was observed with ubiquitin or the proteasome (Fig. 2, D and E). Consistent with these results, Eng-YE expression was markedly enhanced in the presence of a lysosomal inhibitor (chloroquine) but much less so with a proteasome inhibitor (MG132) (Fig. 2F). Moreover, unlike control or upon MG132 treatment ( Fig. 2, D and G), chloroquine treatment produced appreciably enlarged Eng-YE-containing vesicles that strongly co-localized with LAMP1 upon accumulation in lysosomal compartments (Fig. 2G). Taken together, these findings indicated that Eng-YE down-regulation occurs primarily through the lysosomal degradation.
Given that endoglin can form a heteromeric complex with ALK1 and T␤RII, we next examined whether the endoglin YIY motif influences their subcellular trafficking. As expected, the typical membrane and cytoplasmic ALK1 localization was not significantly altered upon co-expression with either Eng-WT or Eng-YF, nor did ALK1 accumulate in LAMP1-containing vesicles (Fig. 3, A and B). In sharp contrast, ALK1 was recruited into endocytic vesicles by Eng-YE where they co-localized with late endosomal markers such as Rab9 (Fig. 3, C, D, F). Importantly, the global ALK1 expression was selectively reduced when coexpressed with Eng-YE but not with Eng-WT or Eng-YF, supporting the specific role of the YIY motif in regulating endoglin and ALK1 stability (Fig. 3E). Similar outcomes were observed for T␤RII where its membrane and diffuse cytoplasmic expression was unaltered upon co-expression with Eng-WT or Eng-YF, and was not targeted for lysosomal degradation (Fig. 4, A and B). However, T␤RII co-localized with Eng-YE in endocytic vesicles in late endosomal compartments (Fig. 4, C and D). Interestingly, while ALK5 is also capable of interacting with endoglin, this receptor sustained its normal cytoplasmic distribution when expressed alone (Fig. 4F), or when co-expressed with Eng-YE (Fig. 4G). Taken together, these findings indicate that the endoglin phosphomimetic motif selectively mediates ALK1 and T␤RII degradation but not ALK5.
To specifically test whether the YIY endocytic motif becomes tyrosine phosphorylated, we first immunoprecipitated for proteins that are tyrosine phosphorylated upon expression of Eng-WT, Eng-YE, or Eng-YF in endoglin-null background (Fig. 5A).
Here tyrosine phosphorylation was only detected for Eng-WT despite the extracellular tyrosine residues present in both Eng-YE and Eng-YF, indicating that the YIY motif is the major site of endoglin tyrosine phosphorylation ( Fig. 5A; lanes 1-4). To identify the major kinase responsible for endoglin tyrosine phosphorylation, we searched for YIY-containing peptide substrate motifs recognized by various protein tyrosine kinases. Previous studies have shown that c-Src and related family kinases recognize a diverse set of tyrosine motifs as substrates including the YIY sequence with a hydrophobic residue in position 2 and a basic amino acid in position 7 (27,28). Given that endoglin contains this YIY peptide sequence followed by a conserved basic residue arginine at position 7 (Fig. 1B), we tested c-Src as a potential kinase by inhibiting its catalytic activity in the presence of Eng-WT expression and found a striking reduction in Eng-WT tyrosine phosphorylation relative to control (Fig. 5A; lane 2 versus 5). Next, to determine whether Src regulates endoglin expression through tyrosine phosphorylation, Eng-WT and Eng-YF were expressed in the presence of increasing Src expression (Fig. 5B). Indeed, Eng-WT stability inversely correlated with increased Src expression and activation, whereas Eng-YF remained constant, indicating that the tyrosine motif is a key determinant of endoglin stability and turnover.
To determine whether endoglin is a Src substrate, we tested for their endogenous interaction by co-immunoprecipitation. Here, interaction between endogenous endoglin and Src was detected in ECs when immunoprecipitated with an antibody targeting the endoglin extracellular domain, but not upon endoglin knockdown or immunoprecipitation with a control IgG (Fig. 5C). We next tested for Src interaction with the endoglin mutants to determine whether Src binding specifically required the YIY motif. Similar co-immunoprecipitation studies revealed that Src interacted with Eng-WT and Eng-YF, but not Eng-YE, strongly suggesting that the bulky dual aromatic side chains of the WY 612 IY 614 motif serve as a structural recognition motif for the Src catalytic domain (Fig. 5D; third panel). More importantly, the Eng-WT/Src interaction resulted in endoglin tyrosine phosphorylation whereas the Eng-YF/Src complex did not, hence further supporting endoglin-612 YIY 614 as a novel Src phosphorylation motif and not an SH2 domainbinding site (Fig. 5D; lane 2

versus 4 of second panel).
We next screened for various Src-activating cytokines and growth factors that mediate endoglin tyrosine phosphorylation and degradation (29,30). While 5 ligands tested induced Src activation to varying degrees compared with no treatment, only VEGF and EGF caused notable endoglin turnover during a 2-h stimulation, indicating a selective Src-dependent response ( Fig.  5E; first and middle panel). To further test Src as a mediator of this process, we examined the effects of VEGF and EGF stimulation on endogenous endoglin expression in the presence or absence of Src inhibition (Fig. 5F). Consistent with our previous results, both VEGF and EGF-induced endoglin turnover was completely blocked upon Src inhibition. Taken together, the results provide novel evidence for Src-mediated endoglin down-regulation following stimulation by pro-angiogenic mitogens VEGF and EGF.
Antibody-induced internalization is a common process by which membrane receptors are degraded. Endoglin antibodies have also been previously reported to induce receptor internalization and presumed to become degraded through an unknown mechanism (31,32). As part of our ongoing investigation on the anti-angiogenic properties of TRC105, we tested the functional role of the tyrosine endocytic motif in TRC105induced endoglin internalization. To this end, Eng-WT and Eng-YF-expressing EngϪ/Ϫ ECs were incubated with TRC105 for various time intervals to allow antibody-induced endoglin internalization in live cells (0 to 6 h). Internalization of the receptor-antibody complex (TRC105-Eng) was monitored at each time point by labeling with a fluorescent secondary antibody upon cell fixation followed by permeabilization. As expected, at early time points there was minimal TRC105-induced internalization of cell surface labeled Eng-WT and Eng-YF (Fig. 6, A and B; 0 to 15 min). However, a significant level of TRC105-Eng-WT was observed in endosomes from 0.5 to 2 h relative to TRC105-Eng-YF (Fig. 6C), suggesting that the tyrosine endocytic motif plays a major role in the endocytic process. Co-staining with markers for early endosomes (EEA1) and the lysosome (LAMP1) further revealed that the receptor complex is ultimately degraded (Fig. 7A; upper and lower panels). Interestingly, there was a TRC105 concentration-dependent increase in endoglin tyrosine phosphorylation that was abrogated upon Src inhibition (Fig. 7, B and C), further supporting a general role for Src-induced tyrosine phosphorylation in endoglin degradation.
Finally, we began characterizing the cellular roles of the tyrosine motif by comparing the effects of Eng-WT, Eng-YE, and Eng-YF expression on endothelial functions. Consistent with our previous findings that indicated endoglin regulation of EC migration (33), here again endoglin inhibited cell motility by 30 -40% when comparing between Engϩ/ϩ and EngϪ/Ϫ ECs, but demonstrated a further inhibition upon rescue overexpres-sion of Eng-WT in EngϪ/Ϫ ECs (Fig. 8A). But unlike Eng-YF and Eng-WT, Eng-YE phenocopied EngϪ/Ϫ ECs in that it failed to suppress migration likely due to its constitutive turnover (Fig. 8A). In parallel studies, we examined the effects of VEGF and EGF stimulation, which normally enhance EC migration through mitogenic signaling (34). Indeed, Eng-WTexpressing ECs treated with either VEGF or EGF promoted EC migration relative to control, whereas Eng-YF expression impaired the VEGF and EGF-induced migratory response (Fig.  8, B and C). Although Src inactivation generally reduces mitogenic and migratory responses in many cell types including ECs, here Src inhibition had the greatest effect at suppressing EGF and VEGF-induced motility for Eng-WT expressing ECs but not Eng-YF (Fig. 8D), suggesting that endoglin turnover following tyrosine phosphorylation by Src is an integral process of VEGF/EGF-induced EC migration.
In addition to migration, we examined the role of the tyrosine motif in a three-dimensional capillary tube formation assay. Consistent with our previous findings, EngϪ/Ϫ ECs formed unstable capillary tube-like structures that regressed over time compared with those rescued with Eng-WT expression (Fig. 9,  A and B). However, the rescue expression with Eng-YE or Eng-YF only partially restored efficient capillary branching and stability relative to Eng-WT (Fig. 9B), suggesting that a proper balance in endoglin tyrosine phosphorylation is critical during capillary morphogenesis. Consistent with this notion, Src inhibition during capillary morphogenesis had minimal impact on the stability and branching capacity of all except Eng-WT expressing ECs (Fig. 9B). Although the similarly reduced levels of capillary formation observed in Eng-YE and Eng-YF was rather unexpected, this could be attributed to the fact that Eng-YE-expressing ECs formed branches that regressed over time, whereas Eng-YF-expressing ECs failed to efficiently sprout new branches from the initial stages (24 versus 72 h; Fig. 9C). Consistent with this notion, Eng-YF-expressing ECs not only had impaired VEGF and EGF-induced cell migration (Fig. 8), but also proliferated slower than Eng-YE expressing ECs (Fig. 9D). Overall, our results demonstrate that the phosphorylation status of the endoglin tyrosine motif and stability play critical roles in EC functions during angiogenesis.

DISCUSSION
Our investigation of the conserved endoglin tyrosine motif revealed one of the first post-translational mechanisms by which endoglin is regulated. Our study defines Src and potentially other family members as major tyrosine kinases contributing to endoglin turnover, and further identifies Src-activating pro-angiogenic mitogens (VEGF and EGF) as regulators of this process. A previous study investigated the post-translational effects of endoglin expression by tumor necrosis factor (TNF)-␣ (35). Here, TNF-␣-induced endoglin down-regulation appeared to be a much more gradual process requiring up to 24 h, whereas Src-induced degradation was far more rapid (Fig.  4D). Nevertheless, as the underlying mechanism for TNF-␣induced endoglin turnover has yet to be defined, it will be interesting to determine whether Src and the endoglin 612 YIY 614 motif are involved in this process.
Whereas little is known about endoglin down-regulation, the process by which T␤RII and ALK5 expression is regulated is more thoroughly characterized, with previous studies demonstrating their internalization in both clathrincoated pits and lipid raft endosomes (1, 36 -38). Furthermore, it is now known that ALK5 degradation requires the ubiquitin-associated salt-inducible kinase (SIK) and Smad7 (39,40). While we failed to detect endoglin ubiquitination and degradation via the proteasome, it still remains plausible that a subset is degraded via the proteasome as part of heteromeric complexes with T␤RII and/or ALK5, or through other endoglin binding partners. In contrast, results indicating a clear chloroquine-induced Eng-YE accumulation in the lysosomal compartments, as well as its restored overall expression (Fig. 2, F and G), strongly support a novel role for the tyrosine motif in mediating endoglin degradation through the lysosomal pathway. The fact that endoglin down-regulation occurred mostly in response to VEGF and EGF stimulation, and not other Srcactivating ligands, indicates that Src targeting of the endoglin tyrosine motif is context-specific and likely tightly regulated. Consistent with this notion, our data showed that insulin, not VEGF or EGF, yielded the highest sustained Src activation dur-FIGURE 7. TRC105 mediates Src-dependent endoglin tyrosine phosphorylation. A, Eng Ϫ/Ϫ MEECs transiently expressing Eng-WT were treated with TRC105 (200 ng/ml) for 2 h and stained for endoglin (green), EEA1 (red), and LAMP-1 (red). B, COS7 cells transiently expressing Eng-WT were treated with TRC105 for 15 min before immunoprecipitation with p-Tyr Ab, and immunoblotted for endoglin. Lysates were immunoblotted for Src, Eng, and ␤-actin for expression. C, COS7 cells expressing Eng-WT were treated with TRC105 (200 ng/ml) alone, or pre-treated with PP2 (15 M) for 15 min before treatment with TRC105 and PP2. Cells were immunoprecipitated for p-Tyr and immunoblotted for Eng. Cell lysates were immunoblotted for Eng and ␤-actin using appropriate antibodies. FIGURE 8. Endoglin tyrosine phosphorylation motif is a critical determinant of endothelial migration. A, quantification of transwell migration for Engϩ/ϩ, EngϪ/Ϫ, and EngϪ/Ϫ ECs expressing Eng-WT, Eng-YE, or Eng-YF. ECs were plated into transwells in growth media. *, p Ͻ 0.05; **, p Ͻ 0.04 (Eng-WT or Eng-YF compared with EngϪ/Ϫ). B, representative images of Eng-WT or Eng-YF transwells upon VEGF (500 ng/ml) or EGF (500 ng/ml) stimulation 16 h. C, quantification of transwell migration for EngϪ/Ϫ ECs expressing Eng-WT or Eng-YF in growth media, or growth media plus VEGF (500 ng/ml) or EGF (500 ng/ml) stimulation (16 h). *, p Ͻ 0.04; **, p ϭ 0.01 (compared with Eng-WT plus EGF treatment). D, quantification of transwell migration for EngϪ/Ϫ ECs expressing Eng-WT or Eng-YF upon VEGF (500 ng/ml) or EGF (500 ng/ml) stimulation (16 h) in the presence of PP2 (5 M). *, p ϭ 0.02 (compared with Eng-WT plus EGF treatment); **, p ϭ 0.03 (compared with Eng-WT plus VEGF treatment).
ing the 2 h treatment despite having little effect on endoglin stability ( Fig. 4D; second panel). It is unclear whether Src is first recruited to endoglin by specific receptor tyrosine kinases such as VEGFR2 and EGFR, or whether other undefined signaling complexes are involved.
But in addition to the potential involvement of receptor tyrosine kinases, it is possible that endoglin may also be subject to regulation by tyrosine phosphatases that inhibit endoglin turnover. Indeed, in one previous mouse study, the functional knock-out of an endothelium-expressed receptor tyrosine phosphatase, CD148 (DEP-1/PTP), proved lethal at midgestation due to impaired angiogenesis with a notable lack of endoglin protein expression throughout the vasculature, while other angiogenic markers such as ALK1, VEGFR2, and Tie2 were unaffected (37). While it will be important to identify endoglin-specific tyrosine phosphatases, our results suggest that tyrosine phosphatases such as CD148 may play a key role in promoting endoglin stability and signaling, whereas tyrosine kinases including Src, have the opposite effect.
The defects in capillary tube formation shared by both Eng-YE and Eng-YF were initially surprising since endoglin normally enhances tube formation, and therefore a less stable form (i.e. Eng-YE) would be expected to impede this process. Upon closer inspection, however, our data indicated that Eng-YE and Eng-YF mutants impair angiogenesis through distinct cellular mechanisms. Whereas Eng-YE-derived capillaries were unstable and regressed over time much like EngϪ/Ϫ ECs, the inability for Eng-YF-expressing capillaries to efficiently branch out may indicate a migration-specific defect due to impaired VEGF-or EGF-induced migratory response rather than proliferation, since Eng-YF ECs proliferated slightly faster than that of Eng-WT (Figs. 8C and 9D). Although the restored expression of endoglin wild type (Eng-WT) inhibited proliferation in EngϪ/Ϫ just as we reported previously, currently it is unclear as to why Eng-YF expression increases proliferation relative to Eng-WT over time ( Fig. 9D; 24 h versus 36 h). Given that endoglin modulates numerous signaling pathways closely associated with migration and proliferation, such as the Smads, ERK, and PI3K/Akt, it will be crucial in future studies to investigate the signaling properties governed by this novel tyrosine motif.
Defining the mechanisms by which TRC105 down-regulates endoglin is part of our ongoing investigation. Our current work demonstrates that TRC105 causes endoglin internalization and receptor trafficking toward lysosomal degradation (Fig. 6). Although the precise mechanism is unclear, the rapid, concentration dependent effects of TRC105 on Src activation and endoglin tyrosine phosphorylation suggest that, whether induced by exogenous (i.e. TRC105) or endogenous factors (i.e. VEGF/EGF), Src phosphorylation of the 612 YIY 614 motif may be a crucial event preceding endoglin degradation.
In conclusion, this work provides critical new information on the structure and function of endoglin. Our studies indicate that the evolutionarily conserved 612 YIY 614 motif serves as a direct substrate for Src tyrosine kinase and functions as a key determinant for endoglin down-regulation and angiogenesis. Along these lines, Src-activating mitogens such as VEGF and EGF may regulate endoglin expression as part of a negative feedback mechanism. Finally, our work provides clinically relevant data on the mechanisms by which TRC105 induces endoglin inhibition.