The arterial wall adapts to alterations in blood flow and pressure by remodeling the cellular and extracellular architecture. Biomechanical stress of vascular smooth muscle cells (VSMCs) in the media is thought to precede this process and promote their activation and subsequent proliferation. However, molecular determinants orchestrating the transcriptional phenotype under these conditions have been insufficiently studied. We identified the transcription factor, nuclear factor of activated T cells 5 (NFAT5; or tonicity enhancer-binding protein) as a crucial regulatory element of mechanical stress responses of VSMCs. Here, the relevance of NFAT5 for arterial growth and thickening is investigated in mice upon inducible smooth muscle cell (SMC)-specific genetic ablation of Nfat5. In cultured mouse VSMCs, loss of Nfat5 inhibits the expression of gene sets involved in the control of the cell cycle and the interaction with the extracellular matrix and cytoskeletal dynamics. In vivo, SMC-specific knockout of Nfat5 did not affect the general vascular architecture and blood pressure levels under baseline conditions. However, proliferation of VSMCs and the thickening of the arterial wall were inhibited during both flow-induced collateral remodeling and hypertension-mediated arterial hypertrophy. Whereas originally described as a hypertonicity-responsive transcription factor, these findings identify NFAT5 as a novel molecular determinant of biomechanically induced phenotype changes of VSMCs and wall stress-induced arterial remodeling processes.—Arnold, C., Feldner, A., Zappe, M., Komljenovic, D., De La Torre, C., Ruzicka, P., Hecker, M., Neuhofer, W., Korff, T. Genetic ablation of NFAT5/TonEBP in smooth muscle cells impairs flow- and pressure-induced arterial remodeling in mice. FASEB J. 33, 3364–3377 (2019). www.fasebj.org
Adequate endothelial cell stimulation is a prerequisite for the adaptive remodelling of macro- and microvessels. A pivotal autocrine mechanism following endothelial cell activation is the release of angiopoietin-2 (Ang-2), which subsequently antagonizes the binding of Ang-1 to the Tie-2 receptor, thus sensitizing the endothelial cells to pro-angiogenic and/or pro-inflammatory stimuli. Based on the observation that hypertension in mice reduces the abundance of Ang-2 stored in arterial endothelial cells, this study was aimed at testing the hypothesis that an increase in wall stress (WS) or stretch—a hallmark of hypertension—is sufficient to release Ang-2 from endothelial cells. In fact, stretching of isolated perfused mouse arteries or human cultured endothelial cells rapidly elicited an increased release of Ang-2. In the cultured endothelial cells, this was preceded by a transient rise in intracellular free calcium, abrogated through calcium chelation and accompanied by a decrease in Tie-2 phosphorylation. Interestingly, Ang-1 abolished the stretch-induced release of Ang-2 from both cultured and native endothelial cells through inhibiting the stretch-dependent mobilization of intracellular calcium. Collectively, these results indicate that increased WS or stretch facilitates the release of Ang-2 from endothelial cell Weibel–Palade bodies, and that Ang-1 can block this by attenuating the stretch-mediated rise in intracellular calcium.
Contraction of vascular smooth muscle cells (VSMCs) in response to an increase in blood pressure is pivotal for maintaining blood flow and constitutes both consequence and cause of chronic hypertension. While a plethora of contraction-mediating mechanisms has been delineated, not much is known about molecular determinants regulating the responses of VSMCs to an increase in wall stress or biomechanical stretch. In this context, we reported that conditional ablation of Junb - a subunit of the transcription factor AP-1 (activator protein 1) - interferes with the expression of myosin light chain 9 and consequently impaired myogenic responses of arteries. We hypothesized that the arterial circumferential wall stress is increased under these conditions which may influence the VSMC phenotype and vessel wall architecture. In fact, Junb-deficient arteries showed structural remodeling as evidenced by a ~2-3-fold decrease in collagen type I and IV content (n=5, p<0.05). cDNA microarray analyses revealed that expression of regulator of G-protein signaling 5 (RGS5) was up-regulated in VSMCs (~6-fold, n=4, p<0.001) isolated from Junb-deficient arteries which was confirmed in vivo but not mediated by AP-1. However, elevation of wall stress by volume-induced hypertension in vivo or exposure of VSMCs to biomechanical stretch in vitro was sufficient to trigger RGS5 expression (3-fold, n=3, p<0.05). Overexpression of this regulator activates RhoA and enhanced stretch-induced responses of VSMCs such as stress fiber formation while its knockdown showed opposite effects. In line with this, RGS5-deficient arteries show an impaired pressure-mediated constriction (n=8, p<0.05) and both the hypertension-induced increase in diastolic blood pressure as well as structural remodeling of the arterial vessel wall were significantly attenuated in RGS5-deficient mice. Collectively, these findings indicate that biomechanical stretch mediates the expression of RGS5 in VSMCs to enhance stretch-dependent responses as a prerequisite for adequate adaptive vascular remodeling processes. This mechanism may establish RGS5 as a promising target for future strategies to combat, e.g. the deleterious consequences of arterial hypertension such as arterial stiffening.
The Delta-Notch pathway is a signal exchanger between adjacent cells to regulate numerous differentiation steps during embryonic development. Blood vessel formation by sprouting angiogenesis requires high expression of the Notch ligand DLL4 in the leading tip cell, while Notch receptors in the trailing stalk cells are activated by DLL4 to achieve strong Notch signaling activity. Upon ligand binding, Notch receptors are cleaved by ADAM proteases and gamma-secretase. This releases the intracellular Notch domain that acts as a transcription factor. There is evidence that also Notch ligands (DLL1, DLL4, JAG1, JAG2) are processed upon receptor binding to influence transcription in the ligand-expressing cell. Thus, the existence of bi-directional Delta-Notch signaling has been proposed. We report here that the Notch ligands DLL1 and JAG1 are processed in endothelial cells in a gamma-secretase-dependent manner and that the intracellular ligand domains accumulate in the cell nucleus. Overexpression of JAG1 intracellular domain (ICD) as well as DLL1-ICD, DLL4-ICD and NOTCH1-ICD inhibited endothelial proliferation. Whereas NOTCH1-ICD strongly repressed endothelial migration and sprouting angiogenesis, JAG1-ICD, DLL1-ICD and DLL4-ICD had no significant effects. Consistently, global gene expression patterns were only marginally affected by the processed Notch ligands. In addition to its effects as a transcription factor, NOTCH1-ICD promotes cell adhesion to the extracellular matrix in a transcription-independent manner. However, JAG1-ICD, DLL1-ICD and DLL4-ICD did not influence endothelial cell adhesion. In summary, reverse signaling of Notch ligands appears to be dispensable for angiogenesis in cellular systems.
Hypertension evokes detrimental changes in the arterial vessel wall that facilitate stiffening and thus lead to a further rise in mean blood pressure, eventually causing heart failure. The underlying pathophysiological remodelling process is elicited by an increase in wall stress (WS) and is strictly dependent on the activation of vascular smooth muscle cells (SMC). However, it remains unclear as to why these cells fail to maintain their contractile and quiescent phenotype in a hypertensive environment. In this context, we reveal that the knockdown of myocardin—a pivotal transcriptional determinant of the contractile SMC phenotype—is sufficient to induce SMC proliferation. In line with this observation, immunofluorescence analysis of the media of remodelling arteries from hypertensive mice demonstrated a significant decrease in the abundance of myocardin and an increase in SMC proliferation. Subsequent analyses of isolated perfused mouse arteries and human cultured SMCs exposed to cyclic stretch (i.e. mimicking one component of WS) suggested that this biomechanical force facilitates serine phosphorylation of myocardin. Furthermore, this biomechanical stimulus promotes rapid translocation of myocardin from the nucleus to the cytoplasm, inhibits its mRNA expression, and causes proteasomal degradation of the cytoplasmic protein. Collectively, these findings suggest that hypertension negates the activity of myocardin in SMCs on multiple levels, hence eliminating a crucial determinant of SMC quiescence. This mechanism may control the initial switch from the contractile towards the synthetic SMC phenotype during hypertension and may offer an interesting novel approach to prevent cardiovascular disease.
G protein‐coupled receptors (GPCR) are major regulators of contractility of vascular smooth muscle cells (SMCs) thereby controlling vascular tone, blood pressure and flow. GPCR signal transduction is modulated by the regulators of G protein signaling (RGS) which terminate G protein activation. Although RGS5 is known to inhibit Gα q activation in arterial SMCs, its role in arterial remodeling is unclear. To this end, we induced arteriogenesis – the flow and wall stress‐induced growth of collateral arterioles – in mice. Immunofluorescence analysis of remodeling arterioles showed increased RGS5 abundance in wild type medial SMCs while remodeling in rgs5 −/− arterioles was disturbed. Exposing cultured human arterial SMCs to cyclic stretch, mimicking elevated wall stress, also led to increased RGS5 levels. Over‐expressing RGS5 in human arterial SMCs prevented Gα q /Ca 2+ ‐mediated constriction upon sphingosine‐1‐phosphate (S1P) stimulation whereas S1P‐treated isolated rgs5 −/− arteries showed enhanced contractility. This shifted S1P signaling from Gα q towards Gα 12/13 ‐mediated RhoA kinase activation. In summary, elevated wall stress elicited RGS5 expression which blunts Ca 2+ ‐mediated vasoconstriction leading to Gα 12/13 ‐dependent RhoA kinase activation. This appears to be critical for arteriogenesis. This work was supported by the Deutsche Forschungsgemeinschaft (SFB TR23, projects B6/C5/C6).
Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Angiogenesis is coordinated by VEGF and Notch signaling. DLL4-induced Notch signaling inhibits tip cell formation and vessel branching. To ensure proper Notch signaling, receptors and ligands are clustered at adherens junctions. However, little is known about factors that control Notch activity by influencing the cellular localization of Notch ligands. Here, we show that the multiple PDZ domain protein (MPDZ) enhances Notch signaling activity. MPDZ physically interacts with the intracellular carboxyterminus of DLL1 and DLL4 and enables their interaction with the adherens junction protein Nectin-2. Inactivation of the MPDZ gene leads to impaired Notch signaling activity and increased blood vessel sprouting in cellular models and the embryonic mouse hindbrain. Tumor angiogenesis was enhanced upon endothelial-specific inactivation of MPDZ leading to an excessively branched and poorly functional vessel network resulting in tumor hypoxia. As such, we identified MPDZ as a novel modulator of Notch signaling by controlling ligand recruitment to adherens junctions. https://doi.org/10.7554/eLife.32860.001 eLife digest Blood vessels transport oxygen and nutrients to all our organs and also remove waste products. New blood vessels form – in a process called angiogenesis – when a tissue is not receiving enough oxygen. This happens during normal development and wound healing, but also during tumor growth. Cells at the tip of a branching blood vessel sense when a tissue lacks oxygen and use proteins on their cell surfaces to help new vessels to grow. During this process, the tip cells of an existing vessel relay the signal from the tissue to other cells ‘behind’ them, in the so-called stalk of the vessel. It is known that tip- and stalk cells communicate by using specific proteins at their interfaces. The tip cells activate proteins called Notch ligands, such as DLL4, while stalk cells express the Notch receptor. During a process called Notch signaling, the ligands bind to the receptor, which becomes active and helps to control angiogenesis. It also hinders excessive vessel branching and so prevents the blood vessels from becoming leaky and inefficient. However, it was not known exactly how Notch ligands interact with their receptors on neighboring cells, and Notch signaling is regulated. Here, Tetzlaff et al. sought to answer these questions by using blood vessel cells from the human umbilical cord grown in the laboratory and blood vessel cells in mice. The results showed that the proteins DLL1 and DLL4 interacted with a protein called MPDZ. This interaction stabilized the DLL proteins at the cell membrane, which increased the Notch-signaling activity. When Tetzlaff et al. experimentally reduced the amount of MPDZ in the laboratory-grown cells, the Notch signaling decreased. Furthermore, the cells with less MPDZ formed more branching structures. And when MPDZ was genetically removed in mice, the embryos had more branched blood vessels in their developing brains. Lastly, when mice without MPDZ were transplanted with tumor cells, the tumors contained more, but leakier, blood vessels and were not supplied with enough oxygen. This suggests that MPDZ is an important factor that helps to regulate angiogenesis by enhancing Notch signaling between tip and branch cells in a new blood vessel. The increased activity of the Notch limits new blood vessels from branching too much. A better understanding of how blood vessels form or become leaky may help to find ways to prevent tumors from growing. https://doi.org/10.7554/eLife.32860.002 Introduction Angiogenesis, the process of forming new blood vessels from pre-existing ones is essential for embryonic development, tissue growth, wound healing, and regeneration. However, angiogenesis also substantially contributes to the pathogenesis of several diseases, most notably tumor progression (Potente et al., 2011). Angiogenesis is stimulated by vascular endothelial growth factor (VEGF), which activates quiescent endothelial cells (EC). These cells subsequently degrade the extracellular matrix and migrate toward the VEGF gradient. A new vessel sprout is guided by the tip cell, which expresses high amounts of VEGF receptors (Siekmann et al., 2013). VEGF signaling induces the expression of the Notch Delta-like ligand 4 (DLL4) on activated ECs. This transmembrane protein activates Notch receptors on adjacent cells, which adopt the stalk cell phenotype and form the new vessel lumen. Notch signaling requires ligand binding in trans that triggers Notch receptor cleavage to release the intracellular domain (NICD). NICD translocates to the nucleus where it acts as a transcriptional regulator. Prototypical Notch target genes are the HES and HEY transcriptional repressors. These inhibit VEGF receptor expression which limits responsiveness to VEGF, tip cell formation and vessel branching (Bray, 2016; Fischer et al., 2004). Inhibition of DLL4/Notch signaling is a powerful tool to interfere with angiogenesis as it results in the formation of excessive tip cell numbers and vessel branches. This chaotic vessel network precludes proper blood perfusion leading to severe tumor hypoxia. Interestingly, inactivation of the Notch ligand Jagged1 (JAG1) results in a reduced sprouting angiogenesis, indicating that JAG1 and DLL4 have opposing roles during blood vessel formation (Benedito et al., 2009; Kangsamaksin et al., 2015; Kofler et al., 2011). It remains poorly understood which factors modulate and control Notch activity during angiogenesis. To become fully competent for Notch receptor activation, Notch ligands need to gain several posttranslational modifications, for example ubiquitinylation (D'Souza et al., 2008). In addition, Notch ligands need to be presented on the cell surface at an area, that is likely to be in contact with Notch receptors on adjacent cells (Shaya et al., 2017). Delta-like and Jagged proteins contain different PDZ binding motifs at their intracellular carboxyterminus, which enable binding to certain PDZ domain containing proteins. There are some indications that binding of Notch ligands to PDZ domain proteins, for example DLL1 to MAGI1 and SYN2BP as well as JAG1 to AF6 control their cellular localization or their protein stability (Adam et al., 2013; Ascano et al., 2003; Mizuhara et al., 2005). Interestingly, these proteins are associated with either adherens or tight junctions. Notch receptors are also localized at adherens junctions in several cell types (Batchuluun et al., 2017; Benhra et al., 2011; Hatakeyama et al., 2014; Sasaki et al., 2007). Therefore, clustering ligands and receptors at cellular junctions might increase the rate of physical binding events and subsequent Notch signaling activity. In yeast, two-hybrid screening approaches the Notch ligands DLL1 and DLL4, but not JAG1, interacted with the multiple PDZ domain protein (MPDZ) also known as MUPP1 (Adam et al., 2013; Estrach et al., 2007). Also a synthetic DLL1 peptide containing the 27 carboxyterminal amino acids interacted with PDZ proteins including MPDZ (Wright et al., 2004). However, this interaction had not yet been confirmed by independent methods and any potential functional consequences are elusive. MPDZ contains 13 PDZ domains and a single L27 domain (Ullmer et al., 1998). It lacks an intrinsic catalytic function, and it is assumed that its function is to cluster proteins at the cell membrane or at adherens and tight junctions (Adachi et al., 2009). Such clustering of proteins was shown to affect the strength of melatonin or the AMPA transmembrane receptor signaling (Guillaume et al., 2008; Krapivinsky et al., 2004). As Notch receptor expression is often enriched at cellular junctions (Batchuluun et al., 2017; Benhra et al., 2011; Hatakeyama et al., 2014; Sasaki et al., 2007), we analyzed how the protein interaction of MPDZ with the Notch ligands DLL1 and DLL4 affects Notch signaling during angiogenesis. Results MPDZ physically interacts with the Notch ligands DLL1 and DLL4 and promotes Notch signaling MPDZ has been identified in screening approaches as a putative binding partner of the Notch ligands DLL1 and DLL4 (Adam et al., 2013; Estrach et al., 2007; Wright et al., 2004). To verify this, we performed co-immunoprecipitation studies in HEK293T cells. Full-length DLL1 and DLL4 or mutants thereof lacking the carboxyterminal PDZ-binding site (amino acids IATEV) were co-expressed with MPDZ fused with an amino-terminal fluorescent Citrine. Co-immunoprecipitation revealed that MPDZ associated with DLL1 and DLL4 proteins. However, DLL1 or DLL4 lacking their PDZ-binding site did not interact with MPDZ (Figure 1A and B), indicating that the carboxyterminus of Delta-like ligands binds to MPDZ. This protein-protein interaction could also be detected in primary human umbilical venous endothelial cells (HUVEC) as well as in whole murine kidney lysates using a co-immunoprecipitation approach (Figure 1C and D, Figure 1—figure supplement 1A and B). Figure 1 with 1 supplement see all Download asset Open asset MPDZ interacts with DLL1 and DLL4. (A, B) HEK293T cells were transfected with Citrine-MPDZ together with HA-tagged DLL1, HA-tagged DLL1ΔPDZ (lacking the PDZ-binding site), Flag-tagged DLL4 or Flag-tagged DLL4ΔPDZ. Antibodies against Citrine were used to immunoprecipitate Citrine-MPDZ. HA and FLAG-tagged proteins as well as MPDZ were detected by immunoblot (IB). Scheme shows structures of the constructs used for co-immunoprecipitation. Input, 10% of the immunoprecipitate. Cit-MPDZ, Citrine-MPDZ; IP, immunoprecipitation; neg.ctrl., negative control. (C, D) MPDZ was co-expressed with either DLL1 or DLL4 in primary endothelial cells (HUVEC). DLL1 and DLL4 were pulled down by using specific antibodies. DLL1, DLL4 and MPDZ were detected by immunoblot (IB). Input, 5% of the immunoprecipitate. IP, immunoprecipitation; neg.ctrl., negative control. (E) HUVEC were either transduced with adenovirus expressing GFP (ctrl) or MPDZ (MPDZ). Expression level of Notch target genes HEY1, HEY2 and HES1 were analyzed by qPCR 48 hr after transduction. Data are presented as mean ±SD. n = 4; *, p<0.05; **, p<0.01 unpaired Student’s t-test. (F) Scheme of the co-culture Notch reporter assay. IMCD3 cells expressing the Notch ligand DLL4 were co-cultured with CHO-N1-CIT cells carrying a Notch luciferase reporter construct. The IMCD3 sender cells were modified by expression of MPDZ or an empty vector control. After 48 hours, cells were lysed and the light emission of the luciferin and the Renilla luciferase activities were measured. Signaling activity is calculated by normalizing the luciferase signal with the Renilla signal. Data are presented as mean ± SEM. n = 5; *, p<0.05 unpaired Student’s t-test. https://doi.org/10.7554/eLife.32860.003 Figure 1—source data 1 Source data of qantitative PCR analysis related to Figure 1E. https://doi.org/10.7554/eLife.32860.005 Download elife-32860-fig1-data1-v3.xlsx Since SYNJ2BP binds also to DLL1 and DLL4 via the PDZ-binding motif and induces Notch signaling, a competition between SYNJ2BP and MPDZ might be possible. However, pull-down studies showed that the absence of MPDZ did not overtly affect the binding of SYNJ2BP to DLL1 or DLL4 (Figure 1—figure supplement 1C and D). The activity of Notch signaling depends critically on the amount of active DLL1/4 molecules on the cell surface. We tested whether the MPDZ-DLL1/4 protein interaction could alter Notch signaling activity. Forced expression of MPDZ promoted Notch signaling in HUVEC as indicated by higher expression levels of the Notch target genes HEY1, HEY2 and HES1 (Figure 1E). To test if MPDZ would alter the ability of DLL4 to activate Notch receptors in trans, IMCD3 cells expressing DLL4 (sender cells) were transfected with plasmids encoding MPDZ cDNA or empty vector control. A Notch luciferase reporter CHO cell line (receiver cells) was co-cultured with the IMCD3 sender cells (Figure 1F). This showed that higher amounts of MPDZ in the DLL4 expressing sender cells resulted in increased Notch signaling activity in receiver cells (Figure 1F). Loss of MPDZ impairs endothelial Notch signaling in vitro and in mice To test if MPDZ contributes to basal Notch signaling in ECs, we silenced MPDZ expression in HUVEC using established lentiviruses expressing independent shRNAs (Feldner et al., 2017). The reduction of MPDZ expression (93 ± 3%, n = 4, p<0.001) resulted in a significant reduction of mRNA expression of the Notch target genes HEY1, HEY2 and HES1 (Figure 2A), indicating diminished Notch activity. Figure 2 Download asset Open asset MPDZ promotes Notch signaling activity. (A) HUVECs were either transduced with lentivirus expressing GFP (sh-ctrl) or with lentivirus expressing shRNA against MPDZ (sh-MPDZ). Expression level of Notch target genes HEY1, HEY2 and HES1 were analyzed by qPCR 48 hr after transduction. Data are presented as mean ±SD. n ≥ 3; *, p<0.05; **, p<0.01; ***, p<0.001 unpaired Student’s t-test. (B) Cardiac endothelial cells were isolated from Mpdzfl/fl and MpdzΔEC mice by magnetic beads bound with CD31 antibodies. Expression levels of Notch target genes Hey1 and Hey2 were analyzed by qPCR. Data are presented as mean ±SD. n = 3; *, p<0.05; ***, p<0.001 unpaired Student’s t-test. (C) HUVECs were either transduced with lentivirus expressing GFP (sh-ctrl) or with lentivirus expressing shRNA against MPDZ (sh-MPDZ). Expression levels of DLL1 and DLL4 were analyzed by immunoblotting 48 hr after transduction. β-actin served as loading control. Data are presented as mean ±SD. n ≥ 3; n.s., not significant. (D) HUVECs were either transduced with adenovirus expressing GFP (ctrl) or with adenovirus expressing MPDZ. Expression levels of DLL1 and DLL4 were analyzed by immunoblotting 48 hr after transduction. β-actin served as loading control. Data are presented as mean ±SD. n ≥ 3; n.s., not significant. (E) Lung endothelial cells were isolated from Mpdzfl/fl and MpdzΔEC mice by CD31 magnetic beads. Protein amounts of Dll1 and Dll4 were analyzed by immunoblotting. β-actin served as loading control. Data are presented as mean ±SD. n = 3; n.s., not significant. https://doi.org/10.7554/eLife.32860.006 Figure 2—source data 1 Source data of qantitative PCR analysis related to Figure 2A and B. https://doi.org/10.7554/eLife.32860.007 Download elife-32860-fig2-data1-v3.xlsx This could also be observed in cardiac CD31-positive ECs derived from EC-specific Mpdz-deficient mice (Tek-Cre;Mpdzfl/fl referred to as MpdzΔEC) (Feldner et al., 2017). qPCR analysis revealed that the reduction of Mpdz expression (69 ± 12%, n = 4, p<0,001) resulted in a diminished Notch signaling activity as the relative Hey1 and Hey2 mRNA amounts were lower compared to Cre-negative littermate controls (Figure 2B). MPDZ promotes cell surface localization of DLL1 and DLL4 Next, we aimed at elucidating the mechanism of how MPDZ alters the activity of Notch ligands. The total expression levels of DLL1 and DLL4 proteins in HUVEC lysates were not altered upon silencing of MPDZ (Figure 2C). The same was observed after adenoviral MPDZ overexpression (Figure 2D). Lung ECs derived from MpdzΔEC mice also did not show changes in total Dll1 and Dll4 protein expression levels compared to controls (Figure 2E). MPDZ is able to cluster several transmembrane proteins at tight and adherens junctions. For instance, MPDZ facilitates RhoA signaling by recruiting Syx to endothelial junctions (Ngok et al., 2012). The activation of Notch receptors requires the interaction with Notch ligands presented on the cell surface. MPDZ is expressed in ECs derived of arteries, veins and microvessels with pronounced localization at the cell membrane (Figure 3—figure supplement 1). To address whether MPDZ affects the cell surface expression of the Notch ligands DLL1 and DLL4, we first tested if the carboxyterminal PDZ-binding sites of these Notch ligands are important for cell surface presentation. Constructs in which either full length DLL1 and DLL4 or versions lacking the PDZ-binding motifs were fused to a mCherry tag were generated. Those constructs were expressed in HUVEC and cell surface expression was analyzed by flow cytometry. Only mCherry-positive HUVEC were gated and the cell surface presentation of the Notch ligands was analyzed using specific antibodies against the extracellular domains of DLL1 and DLL4, respectively. Cell surface expression levels of full length DLL1 and DLL4 were higher than those of their mutants lacking the PDZ-binding site (Figure 3A). This indicates that protein-protein interactions via the PDZ-binding site could influence the localization of Delta-like proteins. Figure 3 with 1 supplement see all Download asset Open asset MPDZ recruits DLL1 and DLL4 to Nectin-2. (A) DLL1 and DLL4 full length and such lacking the PDZ-binding site (ΔPDZ) constructs containing a mCherry tag were expressed in HUVEC. Cells were stained with antibodies against DLL1 or DLL4. DLL1 and DLL4 surface expression of mCherry positive cells was analyzed by flow cytometry. n = 3; *, p<0.05; **, p<0.01 unpaired Student’s t-test. (B) Endothelial cells were isolated from Mpdz+/+ embryos and Mpdz-/- littermates at embryonic day E11.5. Cells were purified by CD31 magnetic dynabeads and stained with anti-CD34 and anti-Dll4 antibodies for flow cytometric analysis. n = 4; *, p<0.05; unpaired Student’s t-test. (C, D) HEK293T control cells (293T sh-ctrl) as well as MPDZ-silenced HEK293T cells (293T sh-MPDZ) were transfected with Nectin-2 and HA-tagged DLL1 or Flag-tagged DLL4. For immunoprecipitation a Nectin-2 antibody was used and HA-tagged DLL1, Flag-tagged DLL4 and MPDZ were detected by western Blotting. Input, 10% of immunoprecipitate. IB, Immunoblot; IP, Immunoprecipitation; neg.ctrl., negative control. https://doi.org/10.7554/eLife.32860.008 Figure 3—source data 1 Source data of FACS analysis related to Figure 3A and B. https://doi.org/10.7554/eLife.32860.010 Download elife-32860-fig3-data1-v3.xlsx To test whether indeed MPDZ is involved in regulating DLL4 cell surface localization, we analyzed cell surface presentation on isolated mouse embryonic EC at the developmental stage E11.5 by flow cytometry. This revealed that the Dll4 cell surface expression of CD34-positive ECs was lower in Mpdz-deficient cells compared to wild type littermate controls (Figure 3B). MPDZ recruits DLL1 and DLL4 to the adherens junction protein nectin-2 MPDZ binds to the intracellular domain of the single-pass type I membrane glycoprotein Nectin-2, which is a component of adherens junctions (Adachi et al., 2009) and is expressed in the endothelium (Rehm et al., 2013; Wallez and Huber, 2008). Thus, MPDZ could be involved in recruiting DLL1 and DLL4 to Nectin-2 containing adherens junctions. To test this, we performed co-immunoprecipitation studies in HEK293 cells. This revealed that DLL1 and DLL4 could be co-immunoprecipitated with Nectin-2. However, silencing of MPDZ expression abolished the interaction of DLL1 and DLL4 with Nectin-2 (Figure 3C and D). Interestingly, loss of Mpdz in mice did not affect the overall formation of vascular tight junctions (e.g. staining patterns of Claudin-5 and Occludin were unremarkable), as well as adherens junctions (VE-Cadherin) and Nectin-2-containing junctions (Figure 4, Figure 4—figure supplement 1), similar as observed before (Feldner et al., 2017). In vitro experiments indicated that the co-localization of DLL1 and DLL4 with the adherens junction protein Nectin-2 at the cell membrane was diminished upon silencing MPDZ expression in ECs (Figure 4—figure supplement 2A,B), whereas silencing of Nectin-2 did not affect localization of DLL1 and DLL4 (Figure 4—figure supplement 2C). Taken together, the data suggest that MPDZ stabilizes the cell surface presentation of DLL1 and DLL4 through recruitment to the adherens junction protein Nectin-2. Figure 4 with 2 supplements see all Download asset Open asset Mpdz does not affect cell cell junction assembly. (A) Retinae isolated from 14 days old Mpdz-/- pups and control littermates (Mpdz+/+) were stained with Isolectin-B4 and antibodies against Claudin-5. Images were acquired with the confocal microscope LSM 700. Scale bar: 25 µm. (B, C) Retinae were isolated 9 days old Mpdz-/- pups and control littermates. Staining for endothelial cells with Isolectin-B4 and VE-Cadherin or Nectin-2. Images were acquired with the confocal microscope LSM 700. Scale bar: 25 µm. https://doi.org/10.7554/eLife.32860.011 MPDZ regulates sprouting angiogenesis in vitro and ex vivo DLL4/Notch signaling restricts sprouting angiogenesis and cells with high Notch signaling activity adopt the stalk cell phenotype in a growing vessel sprout (Blanco and Gerhardt, 2013; Eilken and Adams, 2010; Pitulescu et al., 2017). As such, the protein interaction of MPDZ with the Notch ligand DLL4 and its promotion of Notch signaling activity could potentially be important to control angiogenesis. To address this, MPDZ was silenced or over-expressed in HUVEC, which were embedded as spheroids into a collagen matrix to analyze endothelial sprout formation. HUVEC silenced for MPDZ expression had increased angiogenic potential and formed more capillary-like sprouts compared to control cells, both under basal conditions and after VEGF stimulation (Figure 5A). Oppositely, forced MPDZ expression resulted in impaired sprout formation under both conditions (Figure 5B). Furthermore, MPDZ-expressing cells, which exhibit increased Notch signaling activity, preferred the stalk cell position in the sprouting angiogenesis assay. In line with this result, MPDZ silenced cells, which showed less Notch signaling activity, adopted preferentially the tip cell position in a growing sprout (Figure 5C). Figure 5 Download asset Open asset MPDZ inhibits sprouting angiogenesis in vitro. (A) HUVEC were transduced with lentivirus-expressing shRNA against MPDZ (sh-MPDZ) or expressing GFP (sh-ctrl). Sprouting angiogenesis of collagen-embedded spheroids was analyzed 72 hr after transduction. Spheroids were cultured under basal conditions or stimulated with VEGF-A (25 ng/ml). Quantification shows length of all sprouts of each spheroid. n = 4 experiments with 10 spheroids per condition. **, p<0.01; ***, p<0.001 one-way ANOVA (Holm-Sidak method). (B) HUVEC were transduced with control (ctrl) or MPDZ-expressing adenovirus. Sprouting angiogenesis of collagen-embedded spheroids was analyzed 72 hr after transduction. Spheroids were cultured under basal conditions or stimulated with VEGF-A (25 ng/ml). Quantification shows length of all sprouts of each spheroid. n = 5 experiments with 10 spheroids per condition. *, p<0.05; **, p<0.01; One Way ANOVA (Holm-Sidak method). (C) Mixed spheroids of HUVEC transduced with lentivirus expressing shRNA against MPDZ (sh-MPDZ) or expressing mCherry (ctrl) or spheroids of HUVEC transduced with control (ctrl) or MPDZ-expressing (MPDZ) adenovirus were embedded in a collagen matrix and analyzed 72 hr after transduction. Cells at the most distal end were considered as tip cells. Tip cell numbers were analyzed under basal conditions or after stimulation with VEGF-A (25 ng/ml). n = 3 experiments with 10 spheroids per condition. *, p<0.05; **, p<0.01; ***, p<0.001 unpaired Student’s t-test. https://doi.org/10.7554/eLife.32860.014 Figure 5—source data 1 Source data of the sprouting assay and the tip-stalk-cell competition assay related to Figure 5 A, B, C. https://doi.org/10.7554/eLife.32860.015 Download elife-32860-fig5-data1-v3.xlsx The increased angiogenic potential of Mpdz-deficient cells could also be shown in the aortic ring assay, in which slices of the mouse aorta from MpdzΔEC or littermate control mice (Mpdzfl/fl) were embedded into Matrigel. Aortic ECs from MpdzΔEC mice showed a significantly higher outgrowth rate compared to control (Figure 6A). Figure 6 Download asset Open asset Loss of Mpdz leads to increased vessel branching in the embryonic mouse hindbrain. (A) Aortae were isolated from Mpdzfl/fl and MpdzΔEC mice. Aortic rings were embedded in Matrigel and EC outgrowth was analyzed 24 hr after embedding. n = 4 mice per genotype; *, p<0.05 unpaired Student’s t-test. (B) Embryos at developmental stage E12.5 were used for IsolectinB4-FITC staining (endothelial cells). Left panel shows whole hindbrains of Mpdz+/+ and Mpdz-/- embryos. Right panel shows zoom-ins. Left panel: scale bar, 500 µm; Right panel: scale bar, 100 µm. (C) Quantification of vessel branches and junctions per area. n ≥ 6 mice per genotype; **, p<0.01 unpaired Student’s t-test. https://doi.org/10.7554/eLife.32860.016 Figure 6—source data 1 Source data of the aortic ring assay and the blood vessel analysis of embryonic hindbrains related to Figure 6 A and C. https://doi.org/10.7554/eLife.32860.017 Download elife-32860-fig6-data1-v3.xlsx Loss of MPDZ promotes angiogenesis during brain development Based on the in vitro and ex vivo data, we tested whether MPDZ also affects angiogenesis during mouse development. Vascularization of the hindbrain occurs in a stereotypic manner and is ideally suited to examine sprouting angiogenesis (Fantin et al., 2013). Hindbrains were resected at developmental stage E12.5 to analyze the vasculature. This revealed a much denser vessel network in Mpdz-/- embryos compared to wild-type littermates (Figure 6B). Mpdz-/- embryos had a significant higher number of vessel junctions and branches compared to littermate controls (Figure 6C). This indicates that Mpdz is needed to limit developmental angiogenesis. Endothelial-specific loss of Mpdz alters tumor angiogenesis Angiogenesis is a hallmark of cancer and therefore we examined how genetic inactivation of Mpdz specifically in the endothelium would affect tumor growth and tumor angiogenesis. B16F10 melanoma and Lewis lung carcinoma (LLC) cells were injected subcutaneously into syngeneic MpdzΔEC and control mice. No significant differences in the tumor growth rates were observed between Mpdzfl/fl and MpdzΔEC mice (Figure 7A and B). The resected tumors were stained against the EC marker CD31 and α-SMA (smooth muscle cells, myofibroblasts) (Figure 7C). Microvessel density was significantly higher in MpdzΔEC compared to Mpdzfl/fl mice (Figure 7D), whereas vessel coverage was not altered (Figure 7D), similar as observed after blocking Dll4-induced Notch signaling in the tumor vasculature (Kangsamaksin et al., 2015). Figure 7 with 1 supplement see all Download asset Open asset Excessive tumor angiogenesis upon endothelial-specific inactivation of Mpdz. Tumor growth curve of B16F10 (A) and LLC (B) tumors subcutaneously implanted into Mpdzfl/fl and MpdzΔEC mice. n ≥ 5; Results are shown as mean ±SEM. (C) Representative images of B16F10 and LLC tumors grown in Mpdzfl/fl and MpdzΔEC mice, stained against CD31 (endothelial cells) and α-SMA (smooth muscle cells). Scale bar: 100 µm (D) Quantification of the vessel staining. Microvessel density was determined by counting the CD31-positive vessels per area. For the analysis of vessel coverage, the percentage of α-SMA-positive vessels was determined. n ≥ 5; results are shown as mean ±SD; *, p<0.05; ***, p<0.001; unpaired Student’s t-test. (E) Representative images of B16F10 and LLC tumors stained for Glut1 (hypoxia marker) grown in Mpdzfl/fl and MpdzΔEC mice. Scale bar: 100 µm (F) Quantification of the Glut1-positive area. n ≥ 4; results are shown as mean ±SEM; *, p<0.05; **, p<0.01; unpaired Student’s t-test. Figure legends – figure supplements. https://doi.org/10.7554/eLife.32860.018 Figure 7—source data 1 Source data of the microvessel density analysis and the Glut1 expression analysis related to Figure 7D and F. https://doi.org/10.7554/eLife.32860.020 Download elife-32860-fig7-data1-v3.xlsx Increased numbers of blood vessels can support tumor growth, whereas too many vessel branches disturb the functionality of the vessel network, impair proper perfusion, inhibit tumor growth and can lead to tissue hypoxia. For instance, this was observed after inhibition of endothelial Dll4/Notch signaling in tumors (Kangsamaksin et al., 2015; Noguera-Troise et al., 2006; Ridgway et al., 2006). Indeed, both B16F10 and LLC tumors grown in MpdzΔEC mice contained larger hypoxic areas compared to controls, as indicated by Glut1 expression (Figure 7E and F). To elucidate whether tumor perfusion is impaired, we injected Hoechst 33342 and FITC-labeled Lycopersicon Esculentum lectin into a tail vein and resected the tumors 5 min later. This revealed that B16F10 as well as LLC tumors were less well perfused in MpdzΔEC mice compared to control littermates (Figure 7—figure supplement 1A and B). In the melanoma model, the percentage of Lectin-positive tumor blood vessels was reduced in MpdzΔEC mice (Figure 7—figure supplement 1C), whereas in the LLC model, which contains a better struct
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.This study was aimed at identifying novel modifiers of Notch signaling that interact with the intracellular domains of DLL1 and DLL4.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.These data identify SYNJ2BP as a novel inhibitor of tip cell formation, executing its functions predominately by promoting Delta-Notch signaling.
Angiogenesis is coordinated by VEGF and Notch signaling. DLL4-induced Notch signaling inhibits tip cell formation and vessel branching. To ensure proper Notch signaling, receptors and ligands are clustered at adherens junctions. However, little is known about factors that control Notch activity by influencing the cellular localization of Notch ligands. Here, we show that the multiple PDZ domain protein (MPDZ) enhances Notch signaling activity. MPDZ physically interacts with the intracellular carboxyterminus of DLL1 and DLL4 and enables their interaction with the adherens junction protein Nectin-2. Inactivation of the MPDZ gene leads to impaired Notch signaling activity and increased blood vessel sprouting in cellular models and the embryonic mouse hindbrain. Tumor angiogenesis was enhanced upon endothelial-specific inactivation of MPDZ leading to an excessively branched and poorly functional vessel network resulting in tumor hypoxia. As such, we identified MPDZ as a novel modulator of Notch signaling by controlling ligand recruitment to adherens junctions.
