Synaptojanin-2 Binding Protein Stabilizes the Notch Ligands DLL1 and DLL4 and Inhibits Sprouting AngiogenesisNovelty and Significance
Rationale: The formation of novel blood vessels is initiated by vascular endothelial growth factor. Subsequently, DLL4-Notch signaling controls the selection of tip cells, which guide new sprouts, and trailing stalk cells. Notch signaling in stalk cells is induced by DLL4 on the tip cells. Moreover, DLL4 and DLL1 are expressed in the stalk cell plexus to maintain Notch signaling. Notch loss-of-function causes formation of a hyperdense vascular network with disturbed blood flow.
Objective: This study was aimed at identifying novel modifiers of Notch signaling that interact with the intracellular domains of DLL1 and DLL4.
Methods and Results: Synaptojanin-2 binding protein (SYNJ2BP, also known as ARIP2) interacted with the PDZ binding motif of DLL1 and DLL4, but not with the Notch ligand Jagged-1. SYNJ2BP was preferentially expressed in stalk cells, enhanced DLL1 and DLL4 protein stability, and promoted Notch signaling in endothelial cells. SYNJ2BP induced expression of the Notch target genes HEY1, lunatic fringe (LFNG), and ephrin-B2, reduced phosphorylation of ERK1/2, and decreased expression of the angiogenic factor vascular endothelial growth factor (VEGF)-C. It inhibited the expression of genes enriched in tip cells, such as angiopoietin-2, ESM1, and Apelin, and impaired tip cell formation. SYNJ2BP inhibited endothelial cell migration, proliferation, and VEGF-induced angiogenesis. This could be rescued by blockade of Notch signaling or application of angiopoietin-2. SYNJ2BP-silenced human endothelial cells formed a functional vascular network in immunocompromised mice with significantly increased vascular density.
Conclusions: These data identify SYNJ2BP as a novel inhibitor of tip cell formation, executing its functions predominately by promoting Delta-Notch signaling.
Angiogenesis, the formation of novel blood vessels from preexisting vessels, is essential for embryonic development as well as for tissue growth and wound healing in the adult.1 Excessive angiogenesis is frequently observed in solid tumors and in some hematological malignancies. For instance, increased blood vessel densities have been detected in the bone marrow of patients with leukemia or myeloma. Consistently, angiogenesis inhibitors show some clinical efficiency.2 Uncontrolled angiogenesis also underlies choroidal neovascularization, which leads to vision threatening (wet macular degeneration)3 and might be causative for the formation of vascular malformations.4 Antiangiogenic drugs have been approved for treatment of several tumor entities and wet macular degeneration. Although blocking vascular endothelial growth factor (VEGF) in the eye is highly efficient,3 most malignant tumors become resistant to anti-VEGF therapy.1 Thus, targeting of additional angiogenic factors will be necessary for future therapy.
Editorial see p 1186
In This Issue, see p 1181
Angiogenic sprouting is initiated in response to stimulatory cues like VEGF, which destabilize cellular contacts, promote degradation of the extracellular matrix, and stimulate quiescent endothelial cells to become migratory and invasive.5 It is essential that only few endothelial cells respond to the growth stimuli and subsequently guide the formation of new capillary sprouts as tip cells. This selection process between tip cells and the trailing stalk cells, which are less motile but form the lumen of the new vessel, is predominantly regulated by Notch signaling. Tip cells are characterized by extensive filopodia, which express high levels of VEGF receptor-2 and VEGF receptor-3 and act as sensors of the VEGF gradient.6 VEGF induces the expression of the Notch ligand DLL4 in tip cells. The transmembrane DLL4 protein binds to Notch1 receptors in the adjacent stalk cells, leading to cleavage and release of the intracellular Notch domain (NICD). NICD translocates to the nucleus and acts as a transcriptional regulator controlling the expression of multiple cell type–specific genes. The induction of common target genes of the HES or HEY transcription factor families is generally observed in many different cell types and is considered an indicator of Notch activity.7 HEY and HES proteins repress the expression of VEGF receptors, thus rendering stalk cells and endothelial cells of mature vessels less responsive to VEGF.8 Consistently, inhibition of Notch by genetic or pharmacological approaches leads to higher tip cell numbers with increased angiogenic sprouting and branching, whereas Notch gain-of-function inhibits vessel branching.9–11 Interestingly, Notch signals are also required in vessels of the adult to maintain the quiescent endothelial phenotype. Long-term deletion of DLL4-Notch signaling in the adult may consequently result in the formation of vascular tumors12,13 or vascular malformations.4
It is important to note that tip cells and stalk cells change their positions and thus change their cellular functions during angiogenesis.14 This appears to be guided by oscillatory gene expression loops of Notch, BMP, and probably additional signaling pathways. These oscillations determine time periods in which cells are responsive for tip or stalk cell–promoting factors.15,16 Second, the expression of Notch ligands in tip and stalk cells is not a simple black-and-white situation. Although tip cells express high amounts of DLL4, this protein was also detected in stalk cells, capillaries, and arterioles, although at lower levels.9,17 Moreover, the related Notch ligand DLL1 is expressed in the stalk cells and is absent from tip cells.17 Consistently, Notch activity was observed throughout the stalk cell plexus in Notch reporter mice.9,17 This indicates that Notch receptors on stalk cells are activated not only by ligand-expressing tip cells but also by ligands on adjacent stalk cells. This is supported by genetic loss-of-function experiments of either Notch1 or DLL4, which demonstrated tip cell formation and excessive branching within the stalk cell plexus8–11 and subsequent defects in vascular remodeling.18
The present study was aimed at identifying novel regulators of Delta-Notch signaling during angiogenesis. Among the most promising candidate molecules, a yeast 2-hybrid screen identified Synaptojanin-2 binding protein (SYNJ2BP, also known as ARIP2 or OMP25) as a novel intracellular interaction partner of DLL1 and DLL4. SYNJ2BP is a small protein (145 amino acids) containing a single PDZ domain. PDZ domain proteins often act as scaffolds and modify the function, stabilization, or localization of transmembrane proteins. SYNJ2BP regulates the localization of Synaptojanin-2 and the endocytosis of activin type II receptors.19,20 This PDZ protein is widely expressed in mouse and rat tissues20,21 and may play a role in cancer progression.22 We hypothesized that SYNJ2BP may act as an important regulator in the control of Delta-Notch signaling during angiogenesis and, consequently, we pursued a systematic study of SYNJ2BP function in the regulation of angiogenic endothelial cell functions.
An expanded Methods section is available in the Online Data Supplement.
Endothelial Migration, Proliferation, Adhesion, Tube Formation, and Sprouting
Analyses of human umbilical vein endothelial cell (HUVEC) migration, proliferation, and spheroid sprouting in collagen beds of siRNA-transfected or viral-transduced HUVECs were performed as described.23
Cycloheximide Assay/DLL1 and DLL4 Half-Life Determination
To assess protein stability, human umbilical artery endothelial cells (HUAECs) were transduced with green fluorescent protein (GFP) or SYNJ2BP adenovirus. After 48 hours, the growth medium was replaced with medium containing 0.1 mmol/L cycloheximide to halt protein biosynthesis. Cells were treated for 0 to 8 hours and then harvested for Western blotting.
In Vivo Spheroid-Based Angiogenesis Assay
The in vivo spheroid-based angiogenesis assay was performed as described24,25 with spheroids from pooled HUVECs injected subcutaneously into the flank of 6-week-old to 8-week-old female CB17 severe combined immunodeficiency mice (Charles River). Plugs were removed after 28 days and fixed in 4% formaldehyde before embedding in paraffin. Sections (5 µm) were stained with antihuman CD34 (QBEND10; Menarini Diagostics) and anti-α-smooth muscle actin-Cy3 (Sigma-Aldrich).
Quantitative Reverse-Transcription Polymerase Chain Reaction
Total RNA was isolated using the RNeasy Kit (Qiagen), and 2 µg RNA was transcribed into cDNA using the SuperScript II reverse transcriptase and random hexamer primers (Invitrogen). Quantitative polymerase chain reaction was performed with the POWER SYBR Green Master Mix on an ABI StepOnePlus cycler. All polymerase chain reaction primer sequences are listed in Online Table I.
Results are expressed as means±SD. Comparisons between groups were analyzed by t test (2-sided). Degradation rates were analyzed with a 2-way ANOVA. P<0.05 were considered significant.
DLL4 Physically Interacts With SYNJ2BP, MUPP1, and MAGI Proteins
Yeast 2-hybrid screening using the intracellular domain of DLL4 as bait and an embryonic cDNA library as prey identified SYNJ2BP, MUPP1, MAGI2, and MAGI3 as novel putative binding proteins of DLL4. The same interacting proteins were also retrieved in a screen using the intracellular domain of DLL1 as bait (Figure 1A). MAGI1 was also isolated but showed characteristics of a typical false-positive clone in subsequent analyses in yeast. This is relevant insofar as MAGI1 and MAGI2 (Acvrinp1) have previously been reported to bind DLL1.26,27 All identified proteins contain PDZ domains and thus should bind to the carboxyterminal PDZ binding motif (amino acids IATEV) of DLL1 and DLL4. In silico analyses using a statistical model that predicts PDZ domain–peptide interactions28 supported this hypothesis and predicted a high probability for interactions of MAGI2, MAGI3, MUPP1, and SYNJ2BP with DLL1 and DLL4 (Figure 1B). This algorithm predicted much lower interaction probabilities for putative binding of the Notch ligand Jagged-1 (JAG1), which harbors a different PDZ binding motif (amino acids MEYIV; Figure 1B). Yeast 2-hybrid experiments showed no binding of JAG1 to SYNJ2BP (Figure 1A). The PDZ domain/PDZ binding motif interactions were verified in yeast cells in which the intracellular domains of DLL1 and DLL4, but not those lacking the terminal 4 amino acids ATEV, bound to MAGI2, MAGI3, MUPP1, and SYNJ2BP (Figure 1A). The interaction of SYNJ2BP with DLL1 and DLL4 was also observed in transiently transfected human HEK293 cells. Full-length human proteins were tagged with protein A or luciferase and the interactions between SYNJ2BP and Delta ligands were detected by bioluminescence after coimmunoprecipitation (LUMIER assay29; Figure 1C). Moreover, DLL4 and SYNJ2BP could be coprecipitated in HUVECs when overexpressed (Figure 1D). Based on this, we decided to analyze the impact of SYNJ2BP on Delta-Notch signaling in endothelial cells.
SYNJ2BP Partially Overlaps With DLL4 Expression
Primary human endothelial cells of venous (HUVEC) and arterial origin (HUAEC) expressed similar amounts of SYNJ2BP (Online Figure I). SYNJ2BP protein expression was observed in intracellular vesicles and also at the cell membrane, and partially colocalized with DLL4, when overexpressed (Figure 2A). Endogenous SYNJ2BP protein could also be detected in murine primary lung endothelial cells (Online Figure IIA and IIB). In mouse embryos, SYNJ2BP protein was ubiquitously expressed with especially strong expression in the neural tube and in blood vessels of venous and arterial origin (Figure 2B). SYNJ2BP expression was much more abundant compared with DLL4, the expression of which is more restricted to the arterial lineage. In line with this, SYNJ2BP and DLL4 only partially colocalized in the postnatal retinal vasculature of mice. This vascular bed developed after birth in a stereotypical manner. SYNJ2BP was expressed in arterial and venous beds with highest levels in stalk cells (Figure 2C). It is well-known that Delta ligand expression is not only restricted to tip cells. DLL4 is the only Notch ligand in tip cells, but together with DLL1 it may also be present in stalk cells at lower levels, where it overlaps with SYNJ2BP expression (Figure 2D). DLL1 is expressed in stalk cells, arteries, and veins of the developing retina9,17; this also overlaps with SYNJ2BP expression.
Finally we determined quantitative expression of SYNJ2BP and DLL4 in retinal endothelial cells undergoing hypoxia- induced angiogenesis. Newborn mice at postnatal day 7 were exposed to hyperoxia (75% oxygen) for 5 days and then returned to room air. This procedure causes vasoregression of retinal vessels, followed by excessive outgrowth of immature vessels. Thus, this experiment mimics retinopathy of prematurity. The retinae were harvested at day 17, that is, during the angiogenic period.30 Quantitative polymerase chain reaction was performed with normalization to CD31 and von Willebrand factor mRNA, which are rather exclusively expressed in endothelial cells. This revealed that expressions of SYNJ2BP, DLL4, and the Notch target gene HEY1 were significantly decreased in this immature vasculature (Figure 2E). This is consistent with the findings that decreased Notch activity causes excessive angiogenic sprouting of immature vessels and that SYNJ2BP may be involved in the control of angiogenesis.
SYNJ2BP Inhibits Blood Vessel Formation In Vivo
Because of the lack of a suitable knockout mouse model (because the BayGenomics gene trap ES cell line RRB066 did not produce a null allele), we used an established xenografting assay to determine the role of SYNJ2BP in vivo. This assay allows the generation of a human vasculature in mice by subcutaneous implantation of genetically modified HUVEC spheroids.4,23–25 HUVECs were lentivirally infected with shRNA targeting SYNJ2BP, causing a stable knockdown of SYNJ2BP expression to ≈30% (Online Figure IIC). HUVEC spheroids were transplanted in a growth factor–rich Matrigel and fibrin matrix into severe combined immunodeficiency mice. The grafted human endothelial cells formed blood vessels, which anastomozed with the mouse vasculature. This enabled perfusion of the human vessels and recruitment of mural cells (Figure 3). The plugs were resected 4 weeks after grafting and were analyzed by immunohistochemistry. The perfusion rate and coverage with α-smooth muscle actin–positive mural cells were not altered between the 2 groups (Figure 3A–D). However, the microvascular density was significantly higher in the plugs grafted with SYNJ2BP-silenced HUVECs compared with such expressing control shRNA (Figure 3A, 3B, and 3E). Thus, SYNJ2BP acted as an antiangiogenic molecule in vivo.
Second, we analyzed how forced expression of SYNJ2BP affects the outgrowth of capillary sprouts from segments of the mouse aorta. Aortic rings were adenovirally infected with SYNJ2BP or GFP cDNA and embedded in Matrigel. VEGF induced the outgrowth of capillary-like structures under control conditions, and these cells expressed the endothelial cell marker CD31. However, enforced SYNJ2BP expression abolished angiogenesis from aortic rings (Figure 3F and 3G). This again indicates that SYNJ2BP acts as an antiangiogenic protein.
SYNJ2BP Expression Inhibits Sprouting Angiogenesis
We have previously shown that DLL4-Notch loss-of-function enhances sprouting angiogenesis, whereas gain-of-function prevents growth factor–induced sprouting, proliferation, and migration of endothelial cells.23 Forced SYNJ2BP expression (Online Figure IID and IIE) phenocopied the functions of Notch and blocked VEGF-induced or FGF2-induced sprouting (Figure 4A and 4B). Silencing of SYNJ2BP by siRNA duplexes caused increased angiogenesis even under basal conditions (Figure 4C and 4D). This increase of the cumulative sprout length because of increased length of individual sprouts and because of higher sprout numbers (Online Figure III). The observed SYNJ2BP functions are in line with the known roles of DLL4-Notch signaling as a negative regulator of tip cell formation and endothelial cell proliferation.
SYNJ2BP Restricts Endothelial Proliferation and Migration
Angiogenesis is dependent on several cellular processes, including endothelial cell migration, proliferation, and adhesion. Migration of HUVECs during closure of a gap within an endothelial monolayer was significantly reduced by SYNJ2BP overexpression and enhanced after silencing of SYNJ2BP expression (Figure 5A and 5B; Online Figure IVA and IVB). This was not because of altered endothelial cell adhesion because SYNJ2BP had no effects on adhesion of HUVECs to the extracellular matrix components collagen or fibronectin (Online Figure IVC).
The phosphorylation of AKT/protein kinase B involved in cell survival signaling was not altered by SYNJ2BP (Online Figure IVD), and the apoptosis rate was not changed. However, proliferation of HUVECs was impaired by SYNJ2BP expression and enhanced after siRNA-mediated silencing of SYNJ2BP expression (Figure 5C and 5D). Consistently, SYNJ2BP had a negative effect on MAPK signaling as indicated by decreased phosphorylation of ERK proteins (Figure 5E). This again was reminiscent of established NOTCH1 functions in endothelial cells.1,8,9,23
SYNJ2BP Promotes Notch Signaling
The experiments described so far indicate that SYNJ2BP executes antiangiogenic functions in vitro and in vivo. The protein interactions with DLL4 and activin type II receptors imply that these signaling pathways are responsible for the observed functions. We analyzed whether SYNJ2BP gain-of-function or loss-of-function (Online Figure II) in HUVECs could influence Notch and activin signaling. SYNJ2BP expression attenuated activin-mediated phosphorylation of SMAD2/3, supporting the notion that SYNJ2BP inhibits activin signaling.19,31 However, activin was described as an inhibitor of sprouting angiogenesis,32 and our data support this (Figure 6A and 6B). Therefore, inhibition of activin signaling cannot explain the antiangiogenic roles of SYNJ2BP.
Forced lentiviral SYNJ2BP expression led to increased transcription of the primary Notch target genes HEY1 and LFNG, indicating increased Notch activity (Figure 6C). This could be verified in a Western blot showing increased levels of activated Notch (NICD) protein after forced expression of SYNJ2BP (Figure 6D). The glycosyltranferase LFNG modifies Notch receptors and thereby enhances DLL4-Notch signaling during angiogenesis.33 Furthermore, expression of ephrin-B2 was induced. Consistently, lentiviral silencing of SYNJ2BP with 2 independent shRNA constructs (Online Figure IIC) decreased Notch activity and ephrin-B2 expression (Figure 6C). Ephrin-B2 is crucial for VEGF receptor trafficking, arterial fate differentiation, and tip/stalk cell selection during sprouting angiogenesis.34,35 To test whether SYNJ2BP regulates HEY1, LFNG, and ephrin-B2 directly through Notch signaling, HUVECs were transduced with SYNJ2BP-expressing lentivirus and the γ-secretase inhibitor N-[(3,5-difluorophenyl)acetyl]-l-alanyl-2-phenyl]glycine-1,1-dimethylethylester (DAPT) was added after 24 hours to block Notch receptor cleavage. This normalized gene expression levels of LFNG and HEY1 strongly reduced ephrin-B2 mRNA levels (Figure 6E). Furthermore, we observed that constitutively active constructs of NOTCH1 (ca-NICD) or its transducer RBPJκ (ca-RBPSUH-VP16) induced ephrin-B2 expression, whereas blockade of canonical Notch signaling with dominant-negative Mastermind-like 1 reduced ephrin-B2 mRNA levels (Figure 6F). These results correspond to previous observations that ephrin-B2 is a direct Notch target gene in endothelial cells.36 Taken together, SYNJ2BP expression in endothelial cells induced Notch signaling, which is known to promote the stalk cell phenotype.1,8,9
Next, we determined the mRNA levels of several other known factors regulating endothelial guidance and tip/stalk cell formation. VEGF-A, SLIT1, SLIT2, ROBO1, ROBO4, Semaphorin-3A, Neuropilin-1, and Neuropilin-2 expression levels were not altered by SYNJ2BP. However, mRNA amounts of the proangiogenic factor VEGF-C and the tip cell–enriched genes37 angiopoietin-2 (ANGPT2), ESM1, and Apelin were reduced by SYNJ2BP overexpression and increased after SYNJ2BP silencing (Figure 6G). The expressions of ESM1 and Apelin were dependent on Notch signaling, whereas ANGPT2 and VEGF-C were not (Figure 6H). Thus, SYNJ2BP regulates several crucial factors guiding angiogenesis. The majority occurs in a Notch-dependent manner, whereas ANGPT2 and VEGF-C are at least not directly affected by Notch signaling.
SYNJ2BP Promotes the Stalk Cell Phenotype
Notch signaling is intimately linked to the selection of tip and stalk cells, and ANGPT2 expression is highly enriched in tip cells.37,38 To analyze tip cell behavior, we mixed HUVECs labeled with GFP or a red fluorescent dye in a 1:1 ratio before spheroid formation and embedded them in a collagen matrix. The probability of detecting a red or green endothelial cell at the growing tip was ≈50%. However, if SYNJ2BP expression was increased by viral transduction 48 hours before, then the SYNJ2BP-expressing HUVECs acquired the tip position significantly less frequently, whereas SYNJ2BP-silenced cells were found at the tip position at much higher rates (Figure 7A and 7B). Thus, SYNJ2BP expression not only inhibits expression of tip cell–enriched genes like ANGPT2, Apelin, and ESM1 (Figure 6G) but also limits the probability of acquiring the tip cell position. This fits well with the observed expression pattern of SYNJ2BP protein, which was predominantly found in stalk cells of the murine retina (Figure 2C and 2D).
Antiangiogenic Functions of SYNJ2BP Are Predominantly Mediated by Notch and ANGPT2 Signaling
To test the functional significance of these deregulated genes, we applied recombinant VEGF-C, ANGPT2, or Notch inhibitors to endothelial cells overexpressing SYNJ2BP. Recombinant VEGF-C and VEGF-A could not rescue sprouting angiogenesis (Figure 4A and Figure 7C). However, the application of recombinant ANGPT2 in combination with VEGF-A was sufficient to restore sprouting capacity (Figure 7D and 7E).
Finally, SYNJ2BP-expressing HUVEC spheroids were treated with the Notch inhibitor DAPT or the solvent DMSO as control. SYNJ2BP strongly inhibited VEGF-induced sprouting angiogenesis, but the γ-secretase inhibitor DAPT fully restored the angiogenic potential (Figure 7F and 7G). The reciprocal relationship between SYNJ2BP and Notch activity could also be shown in a complementary experiment in which the expression of active NOTCH1 (NICD) normalized endothelial sprouting behavior after SYNJ2BP silencing (Figure 7H and 7I). Moreover, the effects of SYNJ2BP on endothelial cell proliferation could also be rescued by dominant-negative Mastermind-like 1 protein (Figure 7J). This indicates that ANGPT2 and Notch signaling are 2 crucial, independent signaling cascades downstream of SYNJ2BP.
SYNJ2BP Stabilizes the Notch Ligands DLL1 and DLL4
The question arose how the physical binding of SYNJ2BP to DLL1 and DLL4 could enhance Notch signaling in endothelial cells. SYNJ2BP enhances the endocytosis of activin type II receptors.19 Because Notch ligand endocytosis is essential for Notch activation,7 we investigated whether SYNJ2BP expression influences the endocytosis rate of DLL1 (analysis of DLL4 failed because of a lack of suitable antibodies) during normal culture conditions. An immortalized HUVEC line was stably infected with lentivirus expressing DLL1. The internalization rate of membrane surface proteins was determined with a biotinylation/debiotinylation assay39 after SYNJ2BP or GFP expression. The internalization rate of DLL1 proteins was not significantly altered (Online Figure VA). However, we detected that the amount of membrane-bound DLL1 protein was higher in SYNJ2BP-expressing cells (Online Figure VB).
This intrigued us, so we tried to determine DLL1 and DLL4 protein expression on SYNJ2BP expression. Immunostaining of HUVECs revealed that SYNJ2BP expression strongly enhanced the cellular amount of DLL4 protein. DLL4 was localized in intracellular vesicles as well as at the cell membrane (Figure 8A), as expected. Western blotting confirmed increased DLL1 and DLL4 protein expression (Figure 8B; Online Figure VIA). However, this was not because of increased mRNA transcription (Figure 8C). The DLL4 gene is transcribed in oscillatory-like waves on VEGF treatment,15,16 and this was not altered on SYNJ2BP expression (Figure 8D). These results indicated that SYNJ2BP may promote the stability of DLL1 and DLL4 proteins. The inhibition of protein synthesis by cycloheximide revealed that forced SYNJ2BP expression increased the half-life of DLL4 protein from ≈80 to 180 minutes in HUAEC monolayers (Figure 8E and 8F). SYNJ2BP also increased the half-life of DLL1 protein from 130 to 200 minutes in HUAECs (Figure 8G; Online Figure VIB). Similarly, SYNJ2BP increased the DLL1 half-life from 130 to 190 minutes in HUVECs (Online Figure VIC). This results in an accumulation of Notch ligands, which should be capable of stimulating Notch receptors in trans.
To analyze this, HUVECs were first cocultured with C2C12 myoblasts. It is well-known that the induction of Notch signaling in C2C12 cells inhibits the differentiation of these progenitor cells into muscle cells.40,41 When HUVECs were cocultured with C2C12 cells, they could not prevent serum-induced differentiation into myocytes. However, SYNJ2BP-expressing HUVECs—exhibiting high DLL4 and DLL1 levels—sufficiently inhibited differentiation of neighboring C2C12 cells (Online Figure VC). Second, we mixed HUVECs silenced for either SYNJ2BP or DLL4 with primary mouse lung endothelial cells and measured Notch target gene expression in the murine cells using species-specific primer pairs. Similar to silencing of DLL4, the reduced amounts of SYNJ2BP in the human cells also led to decreased Notch activity in the adjacent murine cells (Figure 8H). Taken together, these data indicate that SYNJ2BP promotes Delta protein stabilization in endothelial cells, and this induces Notch signaling in endothelial and nonendothelial cells.
SYNJ2BP most likely interferes with the degradation of DLL1 and DLL4. The mode of Notch ligand proteolysis is not fully resolved yet, and degradation through the proteasome or the lysosome is discussed.42–44 We found no alterations in DLL4 ubiquitinylation on SYNJ2BP expression (Online Figure VD) and, consistently, inhibition of the proteasome did not change the half-life of DLL4 (Figure 8I). However, blockade of lysosome activity prolonged the half-life of DLL4 to approximately the same extent as did SYNJ2BP expression under control conditions. Importantly, SYNJ2BP expression had no additive effect to the lysosome inhibitor (Figure 8J). This could also be observed for DLL1 (Online Figure VID). Therefore, we conclude that SYNJ2BP expression interferes with lysosomal degradation of DLL1 and DLL4; this prolongs its half-life, leading to protein accumulation, in part also at the cell membrane, and increased Notch activity in trans. Notch activity subsequently inhibits angiogenic sprouting and promotes the stalk cell phenotype.
In summary, we suggest that (Online Figure VII) VEGF signaling induces acquisition of the tip cell phenotype, followed by strong transcription of DLL4. The DLL4 protein activates Notch1 in adjacent cells, which adopt the stalk cell phenotype.8–11 The poor sensitivity of stalk cells toward VEGF eliminates the strongest inducer of DLL4 expression, causing a decline of DLL4 protein expression. However, SYNJ2BP expression in stalk cells stabilizes the DLL1 and DLL4 protein expression, supporting Delta-Notch signaling to occur also within the stalk cell plexus. Such continuous Notch signaling is necessary to prevent the conversion into the tip cell phenotype,9 to allow further vessel remodeling,18 and even to prevent angiogenesis in mature vessels.12,13
Angiogenesis is coordinated by the fine-tuned interplay of pro- and antiangiogenic pathways. VEGF/VEGF receptor, Delta/Notch, Angiopoietin/Tie, Semaphorins/Neuropilins/Plexins, Slit/Robo, and Ephrin/Eph are among the key signaling pathways that guide angiogenic sprouting, but additional factors are needed to fine-tune this complex process.6 The Notch cascade acts as a central coordinator of vessel outgrowth, critically controlling the selection of sprouting tip cells versus remodeling stalk cells by decreasing VEGF sensitivity of stalk cells. This is of particular interest for antiangiogenic therapy, because Notch inhibition by blocking antibodies or γ-secretase inhibitors leads to excessive angiogenic sprouting and branching with a poorly functional vasculature.10,11,45,46
The complexity of Notch signaling is mechanistically still poorly understood. Although Delta-Notch interactions in trans activate the Notch receptors, binding in cis can be antagonistic.47 Likewise, it has recently been shown that the glycosyltransferase fringe can modify Notch receptors to enhance DLL4-mediated signaling in endothelial cells. Conversely, the Notch ligand JAG1 may antagonize this DLL4-Notch signaling during sprouting angiogenesis.33 Additionally, several other modifications of Notch signaling can be seen. Notch ligands need to undergo endocytosis to become fully competent to activate Notch receptors.
This study was aimed at identifying novel modifying factors of Notch signaling through a yeast 2-hybrid screen. This approach and subsequent validation experiments identified the membrane-associated guanylate kinase homologue proteins MAGI2 and MAGI3, the multiple PDZ domain protein MUPP1, as well as the PDZ domain protein SYNJ2BP (ARIP2) as novel interaction partners of DLL1 and DLL4. MAGI1 has previously been reported to bind the PDZ peptide of DLL1.26 All of the isolated interaction partners were PDZ domain proteins. Thus, it can be assumed that there are multiple ways PDZ proteins bind or compete for binding to Notch ligands. It can be seen that Notch ligands behave differently depending on the actual intracellular binding partner.
The biochemical and angiogenesis-regulating functions of SYNJ2BP in endothelial cells were subsequently studied in more detail. The physical interaction with Delta ligands appeared to stimulate Notch signaling. SYNJ2BP enhanced the expression of DLL4 protein as shown by Western blotting and immunostaining. This was independent of gene transcription. SYNJ2BP bound to DLL4, and this interaction stabilized the Notch ligand and extended its half-life from 80 to 180 minutes, probably by inhibition of lysosomal degradation. Similarly, the cell surface DLL1 protein levels were increased as DLL1 was stabilized by SYNJ2BP, although to a lesser extent compared with DLL4. This allowed SYNJ2BP-expressing endothelial cells to stimulate neighboring cells carrying Notch receptors. In this regard, Notch signaling was enhanced not only in endothelial cells but also in C2C12 myoblasts that were cocultured with SYNJ2BP-expressing HUVECs. Thus, the recruitment of SYNJ2BP appears as a novel mechanism to stabilize Delta ligands and to allow increased activation of Notch receptor-carrying cells.
Functionally, SYNJ2BP phenocopied Delta-Notch signaling in endothelial cells. Overexpression of this PDZ protein inhibited endothelial proliferation, migration, and sprouting angiogenesis. The data suggest that SYNJ2BP expression promotes the stalk cell phenotype6 with high Notch activity and LFNG expression. Fringe glycosyltransferases can modify Notch receptors to sensitize them to Delta ligands in endothelial cells.33 Consistently, expression of the tip cell molecule ANGPT237 was strongly reduced and the tip cell promoting VEGF factors were less active when added to SYNJ2BP-expressing cells.
The present study furthermore demonstrated that SYNJ2BP acts as an antiangiogenic molecule in vivo. To unravel antiangiogenic functions of SYNJ2BP, we implanted SYNJ2BP-silenced human endothelial cells in a growth factor–rich matrix subcutaneously in immunocompromised mice. This xenotransplantation technique allows the formation of a human vasculature in mice, which anastomoses with the mouse vasculature, becomes perfused, and matures by recruitment of host mural cells. The assay has previously been validated as a powerful in vivo system to study effects of vascular growth factors and inhibitors24 and also to genetically modify the angiogenic program.4,23 Silencing of SYNJ2BP expression led to a significant increase in blood vessel density, indicating that SYNJ2BP acts as a negative regulator of angiogenesis, preventing uncontrolled angiogenesis in a Notch-dependent manner. In vivo DLL4 blockade leads to increased microvessel density combined with lesser vessel perfusion in grafted tumors.11 In this study, we showed that SYNJ2BP is capable of modifying Notch signaling in such a way that it phenocopies most of the effects seen after DLL4-Notch manipulation. Because SYNJ2BP can also influence activin signaling, and because SYNJ2BP and DLL4 expression patterns only partially overlap,19,20 it is obvious that its inhibition cannot completely resemble every single aspect of a DLL4-Notch loss. In summary, the present study identified SYNJ2BP as a novel binding protein of the Notch ligands DLL1 and DLL4, which led to stabilization and accumulation of both Delta-like proteins and activation of Notch signaling. Consistently, SYNJ2BP restrained growth factor–induced angiogenesis in vitro and in vivo.
We thank the members of the Fischer laboratory for fruitful discussions. We are grateful to Dr Achim Gossler, Hannover, Germany, for providing DLL1 antibodies.
Sources of Funding
This work was supported by grants from the SFB/Transregio 23 (to A.F. and H.G.A.) and the Chica and Heinz Schaller Foundation (to A.F).
In August 2013, the average time from submission to first decision for all original research papers submitted to Circulation Research was 12.8 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.113.301686/-/DC1.
- Nonstandard Abbreviations and Acronyms
- green fluorescent protein
- human umbilical artery endothelial cell
- human umbilical vein endothelial cell
- intracellular Notch domain
- synaptojanin-2 binding protein
- Received April 29, 2013.
- Accepted September 11, 2013.
- © 2013 American Heart Association, Inc.
- Wüstehube J,
- Bartol A,
- Liebler SS,
- Brütsch R,
- Zhu Y,
- Felbor U,
- Sure U,
- Augustin HG,
- Fischer A
- Lobov IB,
- Cheung E,
- Wudali R,
- Cao J,
- Halasz G,
- Wei Y,
- Economides A,
- Lin HC,
- Papadopoulos N,
- Yancopoulos GD,
- Wiegand SJ
- Matsuzaki T,
- Hanai S,
- Kishi H,
- Liu Z,
- Bao Y,
- Kikuchi A,
- Tsuchida K,
- Sugino H
- Nemoto Y,
- De Camilli P
- Brütsch R,
- Liebler SS,
- Wüstehube J,
- Bartol A,
- Herberich SE,
- Adam MG,
- Telzerow A,
- Augustin HG,
- Fischer A
- Mizuhara E,
- Nakatani T,
- Minaki Y,
- Sakamoto Y,
- Ono Y,
- Takai Y
- Barrios-Rodiles M,
- Brown KR,
- Ozdamar B,
- et al
- Krneta J,
- Kroll J,
- Alves F,
- Prahst C,
- Sananbenesi F,
- Dullin C,
- Kimmina S,
- Phillips DJ,
- Augustin HG
- del Toro R,
- Prahst C,
- Mathivet T,
- Siegfried G,
- Kaminker JS,
- Larrivee B,
- Breant C,
- Duarte A,
- Takakura N,
- Fukamizu A,
- Penninger J,
- Eichmann A
- Felcht M,
- Koenen W,
- Weiss C,
- Weina K,
- Geraud C,
- Faulhaber J
- Dahlqvist C,
- Blokzijl A,
- Chapman G,
- Falk A,
- Dannaeus K,
- Ibâñez CF,
- Lendahl U
- Nofziger D,
- Miyamoto A,
- Lyons KM,
- Weinmaster G
- Chapman G,
- Sparrow DB,
- Kremmer E,
- Dunwoodie SL
- Dyczynska E,
- Sun D,
- Yi H,
- Sehara-Fujisawa A,
- Blobel CP,
- Zolkiewska A
- Koutelou E,
- Sato S,
- Tomomori-Sato C,
- Florens L,
- Swanson SK,
- Washburn MP,
- Kokkinaki M,
- Conaway RC,
- Conaway JW,
- Moschonas NK
- Sainson RC,
- Aoto J,
- Nakatsu MN,
- Holderfield M,
- Conn E,
- Koller E,
- Hughes CC
Novelty and Significance
What Is Known?
The outgrowth of a new blood vessel is guided by a group of leading tip cells, which are followed by stalk cells.
Vascular endothelial growth factor (VEGF) induces tip cell formation and expression of the DLL4 protein.
DLL4 signals to adjacent cells and induces Notch signaling, which promotes the stalk cell phenotype. The related DLL1 protein signals to Notch receptors within the stalk cell plexus.
What New Information Does This Article Contribute?
Synaptojanin-2 binding protein (SYNJ2BP) was identified as a novel binding protein to DLL1 and DLL4 that prolongs their half-life and enforces Notch signaling.
SYNJ2BP is predominantly expressed in the stalk cell plexus.
SYNJ2BP inhibits tip cell formation and vessel sprouting.
Sprouting angiogenesis requires the coordinated response of endothelial cells toward growth factors like VEGF. Delta-Notch signaling has emerged as a major regulator for selection of the tip cells that are responsive to VEGF. In contrast, Notch activity decreases sensitivity of the trailing stalk cells to VEGF. This coordination restricts vessel branching and allows formation of a functional vessel network. This study identified a novel regulator of Notch signaling during angiogenesis. SYNJ2BP expression in endothelial cells promotes strong Notch signaling between neighboring cells. This prevents tip cell formation and further vessel branching. Our results introduce a novel mechanism of sprouting angiogenesis regulation via controlling the Notch ligand half-life.