VEGF-induced neoangiogenesis is mediated by NAADP and two-pore channel-2–dependent Ca²⁺ signaling

Significance

The formation of new blood vessels (neoangiogenesis) accompanies tissue regeneration and healing, but is also crucial for tumor growth, hence understanding how capillaries are stimulated to grow in response to local cues is essential for the much sought-after aim of controlling this process. We have elucidated a Ca²⁺ signaling pathway involving NAADP, TPCs, and lysosomal Ca²⁺ release activated in vascular endothelial cells by VEGF, the main angiogenic growth factor, and we show that the angiogenic response can be abolished, in cultured cells and in vivo, by inhibiting components of this signaling cascade. The specificity of this pathway in terms of VEGF receptor subtype, intracellular messengers, target channels and Ca²⁺ storage organelles, offers new targets for novel antiangiogenic therapeutic strategies.

Abstract

Vascular endothelial growth factor (VEGF) and its receptors VEGFR1/VEGFR2 play major roles in controlling angiogenesis, including vascularization of solid tumors. Here we describe a specific Ca²⁺ signaling pathway linked to the VEGFR2 receptor subtype, controlling the critical angiogenic responses of endothelial cells (ECs) to VEGF. Key steps of this pathway are the involvement of the potent Ca²⁺ mobilizing messenger, nicotinic acid adenine-dinucleotide phosphate (NAADP), and the specific engagement of the two-pore channel TPC2 subtype on acidic intracellular Ca²⁺ stores, resulting in Ca²⁺ release and angiogenic responses. Targeting this intracellular pathway pharmacologically using the NAADP antagonist Ned-19 or genetically using Tpcn2^−/− mice was found to inhibit angiogenic responses to VEGF in vitro and in vivo. In human umbilical vein endothelial cells (HUVECs) Ned-19 abolished VEGF-induced Ca²⁺ release, impairing phosphorylation of ERK1/2, Akt, eNOS, JNK, cell proliferation, cell migration, and capillary-like tube formation. Interestingly, Tpcn2 shRNA treatment abolished VEGF-induced Ca²⁺ release and capillary-like tube formation. Importantly, in vivo VEGF-induced vessel formation in matrigel plugs in mice was abolished by Ned-19 and, most notably, failed to occur in Tpcn2^−/− mice, but was unaffected in Tpcn1^−/− animals. These results demonstrate that a VEGFR2/NAADP/TPC2/Ca²⁺ signaling pathway is critical for VEGF-induced angiogenesis in vitro and in vivo. Given that VEGF can elicit both pro- and antiangiogenic responses depending upon the balance of signal transduction pathways activated, targeting specific VEGFR2 downstream signaling pathways could modify this balance, potentially leading to more finely tailored therapeutic strategies.

In the adult the formation of new capillaries is an uncommon occurrence mostly restricted to pathological rather than physiological conditions, the majority of blood vessels remaining quiescent once organ growth is accomplished (1). Physiological neoangiogenesis is generally restricted to body sites undergoing regeneration or restructuring (e.g., tissue lesion repair and corpus luteum formation), whereas pathological neoangiogenesis takes place in different diseases ranging from macular degeneration to atherosclerosis, and is vital for the highly noxious development of solid tumors, thus representing a promising target for therapeutic strategies (2). Vascular endothelial growth factors (VEGF), and in particular the family member VEGF-A, are major regulators of angiogenesis and regulate ECs, mainly through the stimulation of VEGF receptor-2 (VEGFR2), a receptor tyrosine kinase, to induce cell proliferation, migration, and sprouting in the early stages of angiogenesis (3, 4). Antiangiogenic agents that target VEGF signaling have become an important component of therapies in multiple cancers, but their use is limited by acquisition of resistance to their therapeutic effects (5, 6). When overall VEGF receptor (VEGFR) signaling is experimentally impaired by the use of blocking antibodies or of specific tyrosine kinase inhibitors, alternative cellular and tissue strategies nullify the success of such interventions (5, 7, 8). Resistance to anti-VEGF therapies may occur through a variety of mechanisms, including evocation of alternative compensatory factors, selection of hypoxia-resistant tumor cells, action of proangiogenic circulating cells, and increased circulating nontumor proangiogenic factors. Moreover, cross-interactions (both cellular and humoral) between ECs and other environmental cues have to be taken into account for the ultimate aim of tailoring therapeutic interventions according to the specific pattern of the angiogenic microenvironment and EC conditions (5–7). The search for novel key downstream effectors is therefore of potential significance in the perspective of angiogenesis control in cancer progression.

Autophosphorylation of VEGFR2 upon binding VEGF results in the activation of downstream signaling cascades through complex and manifold molecular interactions that transmit signals leading to angiogenic responses. Stimulation of different EC types via VEGFR2 results in increases in intracellular free calcium concentrations [Ca²⁺]_i (9, 10) and the crucial role of this signaling element in the regulation of EC functions and angiogenesis is recognized (11, 12), and thought to be largely mediated by the phospholipase Cγ (PLCγ)/inositol 1,4,5 trisphosphate (IP₃) signaling pathway (10). It has been reported that IP₃ releases Ca²⁺ from intracellular stores in ECs, increasing [Ca²⁺]_i, and is augmented by store-operated Ca²⁺ influx (13). This signaling primes the endothelium for angiogenesis through the activation of downstream effectors such as endothelial nitric oxide synthase (eNOS), protein kinases C (PKC), and mitogen-activated protein kinases (MAPKs). Indeed, it has been reported that the interplay between IP₃-dependent Ca²⁺ mobilization and store-operated Ca²⁺ entry produces Ca²⁺ signals whose inhibition impairs the angiogenic effect of VEGF (14, 15). Given the complexity of both VEGF and Ca²⁺ signaling, and the crucial finding that VEGF evokes pro- and antiangiogenic responses, it is clear that the specificity of VEGF-evoked Ca²⁺ signatures deserves further investigation.

Differences in Ca²⁺ signatures, which are key to determining specific Ca²⁺-dependent cellular responses, rely upon often complex spatiotemporal variations in [Ca²⁺]_i (16). A major determinant of these are based on functionally distinct intracellular Ca²⁺-mobilizing messengers, namely IP₃ and cyclic adenosine diphosphoribose (cADPR), which mobilize Ca²⁺ from the endoplasmic reticulum (ER) stores, and nicotinic acid adenine dinucleotide phosphate (NAADP), which triggers Ca²⁺ release from acidic organelles, such as lysosomes and endosomes (17, 18). NAADP likely targets a channel distinct from IP₃ and ryanodine receptors (RyRs), known as two-pore channels (TPCs) (19–25), and the resulting localized NAADP-evoked Ca²⁺ signals may in some cases be globalized via IP₃ and RyRs through Ca²⁺-induced Ca²⁺ release (26, 27). However, in a few cell types, direct activation of RyRs and Ca²⁺ influx channels by NAADP have also been proposed as alternative mechanisms (28, 29). It has been demonstrated that NAADP-sensitive Ca²⁺ stores are present in the endothelium, and that NAADP is capable of regulating vascular smooth muscle contractility and blood pressure by EC-dependent mechanisms (30). In addition, we have previously demonstrated that NAADP is a specific and essential intracellular mediator of ECs histamine H₁ receptors, evoking [Ca²⁺]_i release and secretion of von Willebrand factor, which requires the functional expression of TPCs (31).

In the present work, we identify a novel pathway for VEGFR2 signal transduction whereby receptor activation leads to NAADP and TPC2-dependent Ca²⁺ release from acidic Ca²⁺ stores, which in turn controls angiogenic response in vitro and in vivo. These findings demonstrate, to our knowledge for the first time, the direct relationship between NAADP-mediated Ca²⁺ release and the signaling mechanisms underlying ECs angiogenesis mediated by VEGF.

Results

Evaluation of VEGF-Induced Ca²⁺ Release via VEGFR2.

Stimulation of ECs with VEGF is known to enhance production of IP₃ and elevate [Ca²⁺]_i (9). To characterize the possible contribution of each subtype of VEGFR to Ca²⁺ signaling, primary cultures of HUVECs were stimulated with different concentrations of various VEGFR agonists (Fig. 1): VEGF-A₁₆₅ (also termed VEGF), which principally binds to VEGFR2 but may also interact with VEGFR1, and VEGF-B that activates VEGFR1 selectively. This analysis shows that Ca²⁺ mobilization is triggered only through VEGFR2 (Fig. 1 A and B), the main transducer of VEGF effects on EC differentiation, proliferation, migration, and formation of vascular tubes (10). On the basis of the concentration-response curve, 100 µg/L VEGF-A₁₆₅ was used for all subsequent experiments.

Fig. 1.

VEGF mobilizes calcium through VEGFR2, activating Ca²⁺ release from acidic stores. Live imaging in single FURA-2-AM loaded cells. Intracellular Ca²⁺ levels in HUVECs after stimulation with different concentrations (10, 50, 100, 200, 300, and 400 µg/L) of VEGF-A₁₆₅, also known as VEGF (activator of both receptors, more selective for VEGFR2) (A) and VEGF-B (VEGFR1 selective agonist) (B). (C and D) Identification of VEGF activated intracellular Ca²⁺ stores. (C₁ and D₁) Traces representing Ca²⁺ release in cells stimulated with 100 µg/L VEGF after treatment with either vehicle alone (control) or 1 µmol/L thapsigargin for 15 min (C₁) or 0.5 µmol/L bafilomycin A1 for 1 h (D₁). (C₂ and D₂) Maximum Ca²⁺ concentrations after stimulation with 100 µg/L VEGF. (E–G) Cells were pretreated with 100 µmol/L Ned-19 (selective antagonist of NAADP) for 30 min, then stimulated with either 100 µg/L VEGF (E) or 100 µmol/L histamine (F, positive control) or 2 10⁻³ U/L thrombin (G, negative control). Changes in Ca²⁺ levels are shown as representative traces (E₁, F₁, and G₁) and as maximum Ca²⁺ concentrations in bar charts (E₂, F₂, and G₂). Arrow indicates time of agonist addition. Each data point in bar charts represent mean ± SEM from three to five independent experiments, n = 41–180 cells. **P < 0.01; ***P < 0.0002.

To evaluate the involvement of different intracellular Ca²⁺ storage organelles in VEGF-induced Ca²⁺ release, we adopted a pharmacological approach using bafilomycin A1, which inhibits pH-dependent Ca²⁺ uptake into acidic stores by inhibition of the vacuolar-type H⁺-ATPase pump, and thapsigargin, which inhibits ER SERCA pumps (17). VEGF-induced Ca²⁺ release was significantly impaired not only by thapsigargin (Fig. 1C), but even more so by bafilomycin A1 (Fig. 1D), showing a major involvement of acidic Ca²⁺ stores in response to this agonist. These data are in accordance with the widely observed cross-talk between NAADP and IP₃/cADPR pathways by which Ca²⁺ release initiated from NAADP-sensitive acidic stores triggers further ER-mediated Ca²⁺ release as rationalized by the “trigger” hypothesis (or two-pool model) (17).

NAADP has been described in different cell types as an important Ca²⁺ mobilizing messenger for different agonists (32–37) and has been characterized using the selective membrane-permeant noncompetitive antagonist Ned-19, which blocks NAADP-induced Ca²⁺ release (38). Pretreatment of HUVECs for 30 min with 100 μmol/L Ned-19 inhibits VEGF-induced Ca²⁺ release (Fig. 1E and Fig. S1A), and partially blocks histamine-evoked Ca²⁺ release as previously shown (31) (Fig. 1F), but failed to block Ca²⁺ responses to thrombin (Fig. 1G), known to be independent of NAADP (31). These data demonstrate that NAADP likely mediates VEGFR2-evoked Ca²⁺ signaling in HUVECs.

VEGF Induces NAADP-Dependent Ca²⁺ Mobilization Through the Phosphorylation of VEGFR2.

To investigate whether phosphorylation of VEGFR2 is necessary to induce NAADP-dependent Ca²⁺ release in VEGF signaling, we used TSU-68 (39), which inhibits tyrosine phosphorylation of VEGFR2 in VEGF stimulated HUVECs (Fig. S2A). As shown in Fig. S2 B and C, VEGF-induced, but not histamine-induced, Ca²⁺ release, is reduced by pretreatment with TSU-68. In line with VEGFR2 phosphorylation being upstream of NAADP-mediated Ca²⁺ release, we found that the VEGF-induced phosphorylation of this receptor at Tyr1175 is not affected by Ned-19 (Fig. S2D).

Involvement of NAADP-Induced Ca²⁺ Mobilization in VEGFR2-Dependent Signaling Pathways.

VEGF signaling via VEGFR2 results in essential EC responses and angiogenesis, comprising production of nitric oxide (NO), increase in endothelial permeability, cell proliferation, cell survival and migration (3, 10), and involves a complex network of intracellular transduction pathways and various downstream targets. With the aim of identifying the involvement of NAADP in VEGF-dependent signaling events, we evaluated the activation of known protein targets such as ERK1/2 MAPK, Akt, JNK, eNOS, and p38 MAPK. It is known that in ECs VEGF-induced phosphorylation of ERK1/2 MAPK (40), Akt (41), JNK (42), and eNOS (43) requires intracellular Ca²⁺ mobilization. We found that the phosphorylation of ERK1/2 MAPK, JNK, Akt, eNOS (Fig. 2A) stimulated by VEGF is reduced by treatment with Ned-19, suggesting that this posttranslational modification of these proteins requires NAADP-dependent Ca²⁺ release. In contrast, Ned-19 failed to inhibit p38 MAPK phosphorylation (Fig. 2A). It has been proposed (44) that activation of p38 MAPK negatively regulates VEGF-induced angiogenesis by reducing phosphorylation of ERK1/2 MAPK and increasing vascular permeability but the mechanisms are unknown. Consistent with this, the inhibition of p38 MAPK by SB203580 causes a significant increase in phosphorylation of ERK1/2 MAPK (Fig. 2B) and enhances angiogenesis in vitro (Fig. 2C), confirming that VEGF activates both proangiogenic and antiangiogenic pathways.

Fig. 2.

The involvement of Ned-19 in VEGFR-2 mediated signaling. (A) The activation of downstream targets after stimulation of HUVECs with VEGF for 15 min was studied. p-ERK1/2 MAPK, p-JNK, p-Akt, p-eNOS and p-p38 MAPK were tested by Western blotting and probed with specific antibodies. (B and C) HUVECs were preincubated for 1 h with p38 inhibitor SB203580, followed by treatment with VEGF. The effect was evaluated by Western blot as phosphorylation of ERK1/2 MAPK (B) and as formation of tubes on Matrigel-coated dishes in the angiogenesis assay (C). Data are representative of at least three independent experiments. β-actin was used as a loading control in Western blots.

Ned-19 Inhibits VEGF-Dependent Cell Proliferation.

As a part of the process of angiogenic sprouting induced by VEGF, ECs undergo proliferation (45). The involvement of NAADP in this process was evaluated using the NAADP antagonist Ned-19 by measuring different indicators of cell proliferation. Cell number, assessed by flow cytometry, appeared to be increased by VEGF treatment, and Ned-19 markedly reduced this (Fig. 3A and Fig. S1B). In addition, in cells treated for 24 h with VEGF, the expression of proliferating cellular nuclear antigen α (PCNAα), a nuclear protein involved in the control of DNA replication, is reduced by Ned-19 (Fig. 3B), as was the number of viable cells assessed by the thiazolyl blue tetrazolium bromide (MTT) assay for spectrophotometric determination of cell viability (Fig. 3C). EC proliferation behavior induced by VEGF is therefore dependent upon NAADP-mediated Ca²⁺ release. Accordingly, disruption of acidic Ca²⁺ stores by bafilomycin A1 similarly results in decreased cell proliferation (Fig. S3A).

Fig. 3.

Ned-19 inhibits VEGF-dependent cell proliferation of HUVECs. Cells were treated as indicated for 24 h. (A) Number of cells evaluated by flow cytometry. (B) PCNAα immunoblotting of total cell lysates. β-actin was used as loading control. (C) Cell viability quantified as OD (570 nm) by MTT assay. Data shown in B are representative of three independent experiments. Where applicable, values are expressed as mean ± SEM from three to five independent experiments. **P < 0.01; ***P < 0.0002.

VEGF-Dependent Cell Migration Is Modulated by NAADP-Dependent Ca²⁺ Release.

EC migration, an essential angiogenic process, involves degradation of the extracellular matrix by metallopeptidases and the activation of several signaling pathways which converge on cytoskeletal remodeling, resulting in cell extension, contraction and forward progression (46). The activity of the secreted matrix metallopeptidase 9 (MMP9) is associated with conversion of pro-MMP9 to MMP9. This activation was tested by gelatin zymography in culture media. VEGF-dependent secretion and activation of this metallopeptidase was markedly reduced by Ned-19 (Fig. 4A). Western blot analysis showed that VEGF-stimulated phosphorylated focal adhesion kinase (FAK), known to control cell motility within the extracellular matrix (47), was reduced by Ned-19 (Fig. 4B), again implying NAADP involvement. Further direct demonstration of the NAADP dependence of VEGF-induced cell motility came from Boyden chamber assays (Fig. 4C) and by scratch wound healing assays (Fig. 4D), which clearly showed that the capacity for VEGF-induced migration is reduced in cells treated with Ned-19. NAADP-mediated Ca²⁺ release is thus required for VEGF-dependent migratory capabilities of HUVECs.

Fig. 4.

Ned-19 inhibits VEGF-dependent cell migration of HUVECs. (A) Cell lysates were collected and subjected to gelatin zymography to measure MMP9 activity. (B) Cell lysate tested by Western blotting with a p-FAK specific antibody. β-actin was used as loading control. (C) Cells treated as indicated were allowed to migrate for 24 h across the membrane in Boyden chambers. Cells that migrated into the filter were counted in 20 fields per well. Results are expressed as % of control cell migration. (D) Scratch assay to evaluate the cell migration capacity in the indicated experimental conditions. (Upper) Wounded monolayer at the time of manual damage. (Lower) Degree of wound healing in 24 h. Data shown in A, B, and D are representative of three independent experiments. Where applicable, values are expressed as mean ± SEM from three to five independent experiments. *P < 0.05; **P < 0.01.

The Role of NAADP in the Angiogenic Process in Vitro and in Vivo.

The formation of capillary-like tubes in vivo is regarded as representative of later, differentiative, stages of angiogenesis, and is commonly assayed to test compounds for pro- or antiangiogenic effects. Plating onto matrigel matrices stimulates EC attachment, migration and differentiation into tubular structures simulating the in vivo process (48). When plated on matrigel matrices at high densities, HUVECs form cord-like capillary structures within a few hours, and this process is enhanced by VEGF. Importantly, we found that this key step of angiogenesis was inhibited by Ned-19 pretreatment (Fig. 5A). An approximate estimate of the efficiency of this process can be inferred by the extent of cellular network formation, whereby cells first align to form linear segments, and subsequently interconnect to form closed polygonal structures (49). As shown in Fig. 5B, the number of closed polygons formed in cells stimulated with VEGF in the presence of Ned-19 is reduced by 73% (P < 0.01) compared with samples stimulated with VEGF only. This indicates the involvement of NAADP-mediated Ca²⁺ signaling at the level of capillary-like formation in vitro. Furthermore, bafilomycin A1 disruption of acidic Ca²⁺ stores also resulted in impairment of capillary-like network formation in response to VEGF (Fig. S3B).

Fig. 5.

NAADP pathway inhibition impairs of VEGF-induced vessel formation in vitro and in vivo. (A) Representative images of one of three independent experiments. HUVECs were plated in Matrigel-coated dishes and incubated in EBM-2 with vehicle alone (Ctr), in medium supplemented with VEGF or Ned-19, or in medium containing both VEGF and Ned-19. Each condition was tested in triplicate for each individual experiment. (B) Quantitative evaluation of tube formation as the number of closed polygons formed in five fields for each experimental condition. (C–E) In vivo vessel formation was assessed after s.c. injection of C57BL/6 mice with Matrigel plugs containing either vehicle or VEGF or VEGF and 50, 100, or 150 μmol/L Ned-19. After 5 d, mice were killed and vascularization was evaluated both macroscopically as shown in two representative images (C) and as Hb content expressed as absorbance (OD)/1 g matrigel plug (D). (E) Representative images of H&E stained paraffin sections of VEGF-containing plugs in presence or absence of Ned-19. n = 15 plugs for each condition. Where applicable values from three independent experiments are expressed as mean ± SEM *P < 0.05; **P < 0.01.

The role of NAADP-mediated Ca²⁺ signaling in the overall angiogenic process in vivo was then analyzed by different approaches in murine models. To assess the inhibitory effect of Ned-19 on in vivo angiogenesis, matrigel plug assays were performed. Five days after s.c. injections of plugs in C57BL/6 mice, the extent of plug vascularization under different experimental conditions was evaluated by measuring the hemoglobin (Hb) content, and by histological hematoxylin and eosin (H&E) analysis (Fig. 5 C–E). As macroscopically apparent (Fig. 5C), the plugs containing VEGF, but not those containing VEGF plus Ned-19 or vehicle only, can be seen to undergo intense vascularization. Hb content in the plugs containing VEGF plus Ned-19 was significantly lower than in the plugs containing VEGF alone (Fig. 5D), and histological analysis of plug vascularization confirmed these results (Fig. 5E). These findings indicate an essential role for NAADP-mediated Ca²⁺ release in both in vitro and in vivo VEGF-induced angiogenic processes.

Involvement of TPC2 Channel in the Response to VEGF in Vitro and in Vivo.

Because TPCs have been proposed as principal targets for NAADP in its action to mobilize Ca²⁺ from endolysosomal stores (19–25), we assessed whether Tpnc2 gene silencing mimics the effects of Ned-19 in inhibiting VEGF-induced Ca²⁺ release and in vitro angiogenesis. First we transfected HUVECs with two different anti-human Tpcn2 shRNA constructs and analyzed their responses to VEGF 48 h later. The efficiency and specificity of TPC2 knock down was evaluated by quantitative RT-PCR; the expression of Tpcn2 was found to be significantly reduced (Fig. S4), whereas Tpcn1 expression was unaffected. As shown in Fig. 6A, the specific VEGF-evoked Ca²⁺ mobilization was equally inhibited in ECs treated with either Tpcn2-shRNA. Furthermore, anti-Tpcn2 shRNA-treated HUVECs, in contrast to those transfected with scrambled shRNA, failed to generate closed polygonal structures (Fig. 6 B and C). The effects of knocking down Tpcn2 expression thus recapitulate the effects of Ned-19 on VEGF-evoked Ca²⁺ release (Fig. 1) and the formation of tubular structures (Fig. 5 A and B).

Fig. 6.

TPC2 silencing inhibits both VEGF-induced Ca²⁺ release and in vitro angiogenesis. (A) Intracellular Ca²⁺ levels in HUVECs transfected with anti-human Tpcn2 shRNA constructs and stimulated with VEGF. n = 45 cells. (B) Representative images of one of three independent experiments in which HUVECs transfected with scramble or with two different anti-human Tpcn2 shRNA constructs were plated in Matrigel-coated dishes and incubated in EBM-2 with vehicle alone or VEGF. (C) Quantitative evaluation of capillary-like tube formation as the number of closed polygons formed in five fields for each experimental condition from three independent experiments. N.D. (not detectable) indicates that TPC2 silenced cells failed to form closed polygons. Where applicable, values from three independent experiments are expressed as mean ± SEM; *P < 0.05; ***P < 0.0002.

To further test whether the NAADP putative targets TPC1/TPC2 are involved in the observed angiogenic responses to VEGF, we performed in vivo angiogenesis assays in Tpcn1^−/− (50) and Tpcn2^−/− mice (19), using the matrigel plug assay. Over 5 days, VEGF produced an intense vascularization of the plugs (assessed both by macroscopic observations and by the measure of Hb content) in wild-type (WT) and Tpcn1^−/− mice (Fig. 7 A and B) but not in Tpcn2^−/− mice (Fig. 7 A and C). This finding suggests that TPC2 is essential for VEGF-induced angiogenesis, and importantly demonstrates an isoform specific role for TPC2 in this process.

Fig. 7.

Experiments in transgenic mice lacking NAADP putative targets TPC1 and TPC2. Tpcn1^−/− mice (A and B) and Tpcn2^−/− mice (A–C) were s.c. injected with matrigel plugs containing or not VEGF. 5 d later the mice were killed and the plugs analyzed for vascularization. (A) Macroscopic analysis of the matrigel plugs from Tpcn1^−/− and Tpcn2^−/− mice. (B and C) Measure of Hb content expressed as absorbance (OD)/1 gr matrigel plug in Tpcn1^−/− mice (B) and Tpcn2^−/− mice (C) (n = 10–24 plugs for each condition). Where applicable, values from three independent experiments are expressed as mean ± SEM *P < 0.05; **P < 0.01; ***; P < 0.0002.

A schematic representation based on our experimental evidence described indicates how the control of VEGF-induced angiogenesis may occur through the mediation of the VEGFR2/NAADP/TPC2/Ca²⁺ signaling pathway (Fig. 8).

Fig. 8.

Schematic representation of VEGFR2/NAADP/TPC2/Ca²⁺ signaling pathway in the control of VEGF-induced angiogenic responses.

Discussion

The formation of new vascular capillaries, in physiological, inflammatory, or cancer processes, proceeds through a defined sequence of steps as diverse as cell proliferation, migration, differentiation and morphogenesis, all of which involve control by VEGF-linked signaling cascades. We have shown here, using in vitro and in vivo models of angiogenesis, that the basic steps through which ECs form new capillaries rely upon signaling pathways involving VEGFR2, NAADP, TPC2, and Ca²⁺ release from acidic stores. The inhibition of this signaling pathway significantly decreases the activation of the known VEGFR2 downstream targets ERK1/2 MAPK, JNK, Akt, and eNOS with the exception of p38 MAPK, and blocks angiogenesis in both in vitro and in vivo models.

In line with previous reports (51, 52) our data using HUVECs confirm that intracellular Ca²⁺ increases in response to VEGF are not mediated by VEGFR1 signaling, but selectively involve VEGFR2, the major receptor subtype mediating angiogenic responses (10). The role of each subtype of VEGFR is a crucial issue for the development of targeted strategies aimed at modulating angiogenesis. VEGFR1 and VEGFR2 are differentially expressed in normal cells (53), but their absolute and relative expression levels can considerably vary in different tumors (54), and they are known to couple to different intracellular signaling pathways although they are also subject to extensive cross-talk (10). Specific antiangiogenic pathways are also known to be activated by VEGF via specific pathways operating not only through VEGFR1 (52, 55) but also through VEGFR2. In particular, inhibition of p38 MAPK has been shown to enhance in vitro and in vivo angiogenesis induced by VEGF (44). Importantly, internalization of the VEGFR2 is necessary for activation and downstream signaling involving ERK1/2 MAPK and Akt proteins, whereas p38 MAPK activation is independently regulated and is characterized by a temporally and spatially distinct pathway (56, 57). Our Western blot experiments characterizing the effects of the NAADP antagonist Ned-19 in HUVECs show that various downstream effector kinases, but not p38 MAPK, are ultimately dependent on NAADP-evoked Ca²⁺ release. Parallel in vitro experiments show that under similar conditions Ned-19 treatment strongly reduces angiogenic responses to VEGF, including proliferation, migration, and formation of capillary-like structures (Figs. 3–5). Taken together, these data indicate that angiogenic responses to VEGF are dependent on NAADP signaling, and confirm that the p38 MAPK pathway is not involved in these angiogenic responses. Because VEGF has the potential to elicit both pro- and antiangiogenic biological responses, depending upon the balance of intracellular signaling pathways activated by this agonist, the identification of a pathway apparently restricted to its proangiogenic effects without weakening antiangiogenic pathways offers a new conceptual tool to manipulate the balance between these two antagonistic responses.

The involvement of Ca²⁺ in the responses to VEGFR2 stimulation is well recognized, although the available information can hardly be framed into a comprehensive picture, given the multiplicity of the pathways involved. The role played by specific Ca²⁺ signatures are pivotal: a wealth of data demonstrates that cellular processes are controlled by spatiotemporally complex Ca²⁺ signaling patterns which are decoded to elicit different biological responses (16, 17). In ECs, we have previously demonstrated that the secretion of von Willebrand factor is differentially regulated by different Ca²⁺ signals, dependent on the recruitment of specific Ca²⁺ mobilizing messengers. NAADP was found to mediate the release of intracellular Ca²⁺ triggered specifically through histamine H1 receptor, resulting in secretion of von Willebrand factor, whereas thrombin-mediated Ca²⁺ signaling was NAADP-independent (31). In our study of Ca²⁺ signals triggered by VEGF, we have focused on the most ancient yet most recently discovered Ca²⁺ mobilizing second messenger, NAADP, and its putative target channels, TPCs on acidic stores (58). Our results show that inhibition of the NAADP pathway abolishes intracellular Ca²⁺ increases following VEGF stimulation without interfering with the phosphorylation of Y1175 residue of VEGFR2. The biphasic Ca²⁺ response to NAADP and the dependence of the sustained phase of Ca²⁺ release on the ER are consistent with the idea that NAADP-induced Ca²⁺ signals are small and localized, but act as a trigger for larger and global intracellular Ca²⁺ mobilization through coupling to the ER system (17). Our data, obtained through pharmacological impairment of Ca²⁺ increase from different Ca²⁺ store compartments, are consistent with this model. Indeed, treatment of ECs with both thapsigargin and bafilomycin, inhibited VEGF-induced Ca²⁺ release. These data indicate that VEGFR2-mediated Ca²⁺ signaling requires the recruitment and interactions of both ER and acidic Ca²⁺ stores.

The downstream role of NAADP-mediated Ca²⁺ signaling in the angiogenic responses to VEGF was assessed in two principal ways. First, the selective membrane-permeant NAADP anatgonist, Ned-19 was used. This molecule was developed by a computational ligand-based drug discovery program (38), and has been widely tested in many cellular systems and validated as a selective inhibitor of NAADP-evoked Ca²⁺ release, and as a modulator of NAADP-sensitive TPC2 channels reconsituted in lipid bilayer studies (59). Ned-19 treatment was found to be profoundly antiangiogenic in an in vivo model of neoangiogenesis using matrigel plug assays (48). In this model, plug neovascularization stimulated by VEGF was observed in the control samples, but was inhibited by Ned-19 treatment in a concentration-dependent manner (Fig. 5).

The second approach was to ablate expression of TPCs as putative endolysomal ion channel targets for NAADP. In HUVEC cells transfected with anti-Tpcn2 shRNA, VEGF-evoked Ca²⁺ signals were much reduced, consistent with a key role for these proteins in NAADP-mediated Ca²⁺ release. Anti-Tpcn2 shRNA also specifically impaired the ability of VEGF to induce tubular structures as a hallmark of angiogenic progression (Fig. 6B). These experiments were translated into an in vivo model using the matrigel plug assay. Paralleling the HUVEC experiments, Ned-19 inhibited VEGF-induced vascularization in wild type mice, whereas Tpcn2^−/− mice treated with VEGF alone phenocopied the effects of Ned-19 in WT animals. Importantly, normal VEGF-induced vascularization was seen in Tpcn1^−/−mice, showing a clear difference in roles for TPC1 and TPC2 in angiogenic responses to VEGF. Recent work has shown that TPC1, although regulated by NAADP, has marked differences in ion conductances to TPC2 and may be largely a proton channel under certain conditions (60). This, in addition to their differential endolysosomal distribution with TPC2 being the predominant late endosomal/lysosomal isoform and with TPC1 being more broadly expressed across the endosomal system, may indicate distinct roles in cell physiology as exemplified here (58, 59).

Although it has been reported that under certain conditions TPCs are PI(3,5)P₂ regulated but NAADP-insensitive lysosomal Na⁺ channels (61), a large amount of accumulating data published by others firmly suggests a major, if not direct, role for TPCs in endolysosomal Ca²⁺ release and NAADP-mediated responses (21–25, 57–59). Although additional studies are needed to resolve this controversy, our findings reported here are supportive of the role of TPC2 in mediating NAADP-evoked Ca²⁺ release and downstream responses in ECs and angiogenesis both in vitro and in vivo.

We are aware that the inability of VEGF to induce vascularization of plugs in Tpcn2^−/− mice is not easily reconciled with the fact that in these animals a functioning circulatory system develops, whereas genetic ablation of VEGF or VEGFR2 results in dramatic defects of blood island formation and development (reviewed in ref. 62). These data might indicate differences in intra- and/or intercellular signaling between the process of angiogenesis, based on the recruitment of embryonic precursors, and that of neoangiogenesis observed in the adult.

The heterogeneity of conditions and sites for neoangiogenesis no doubt introduces a range of variables so that any simplistic model of responses to angiogenic factors turns out to be unsatisfactory. Angiogenesis is regulated by a balance of factors which, upon the switch of tumor cells to an angiogenic phenotype, leads to tumor growth and progression (63). The characterization of these factors, the regulation of their production, and mechanisms of action have long been subject to intense investigation, but the ambitious aim of finely controlling angiogenesis is not yet in sight (4). The notable challenge represented by resistance to anti-VEGF therapies point to a high level of complexity. Apparently, a more comprehensive strategy has to be devised to overturn the unfavorable balance between proangiogenic and antiangiogenic signaling, requiring that specific pathways for these responses are identified (6). We believe that a thorough knowledge of the basic signaling machineries of EC in response to VEGF is fundamental to any further integration into more accurate and detailed pictures of pathological neoangiogenesis.

In our experimental models, EC responses to VEGF were analyzed independently of other angiogenic signals; under these controlled conditions we have identified a master VEGFR2/NAADP/TPC2/Ca²⁺ signaling pathway controlling the angiogenic response of ECs to VEGF. This specific intracellular pathway appears to be obligatory because its pharmacological and genetic ablation at different points in the pathway abolishes angiogenic responses both in vitro and in vivo. Moreover, our data indicate that two different VEGF-activated pathways operate: one proangiogenic, involving acidic Ca²⁺ stores and requiring NAADP as second messenger, and the other antiangiogenic, involving the p38 MAPK pathway and activated independently of NAADP-evoked Ca²⁺ release. Our exploration of specific VEGFR2-blocking strategies could lead to finely tailored therapies capable of discriminating between different signal transduction pathways activated by VEGF, and so potentiate current treatments by overturning the balance between proangiogenic and antiangiogenic VEGF effects. Understanding how to best exploit the distinct Ca²⁺ signaling pathways in angiogenesis could contribute to identifying new targets for antiangiogenic therapeutic strategies.

Materials and Methods

Cell Culture.

Human umbilical vein endothelial cells (HUVECs) were obtained from Lonza Sales Ldt (Switzerland) and were cultured in EGM-2 Endothelial Cell Growth Medium-2 (Endothelial Basal Medium EBM-2 + EGM-2 BulletKit, Lonza), + 100 mM Penicillin/Streptomycin (Sigma). They were maintained at 37 °C in a humidified 5% (vol/vol) CO₂ incubator and used at passage number 2–6.

Biochemistry.

The following antibodies were used: Phospho-44/42 MAPK (T202/Y204: E10, Cell Signaling), Phospho-p38 MAPK (Thr180/Tyr182, Cell Signaling) and Phospho-Akt (Ser-473, Cell Signaling), Purified Mouse Anti-eNOs (Ps1177, BD Transduction Laboratories), Rabbit pAb to PCNA (Abcam), Phospho-FAK (Tyr925, Cell Signaling), β-Actin HRP-conjugated (Sigma), Stabilized Goat Anti-Mouse HRP-conjugated (Pierce), Stabilized Peroxidase-conjugated Goat Anti-Rabbit [(H + L), Thermo Scientific], p-VEGFR2 [(Tyr-1175), Santa Cruz Biotechnology].

Reagents used are: Histamine (Sigma), Thrombin (Calbiochem), VEGF-A₁₆₅ and VEGF-B (Peprotech), Ned-19 (Tocris Bioscences), Bafilomycin (Sigma), Thapsigargin (Sigma), TSU-68 (SU668, Selleckchem), SB203580 (Calbiochem). In the in vitro assays, Ned-19 treatment was initiated 30 min before VEGF stimulation. Concentrations of 25, 50, and 100 μmol/L Ned-19 were found to inhibit VEGF-induced Ca²⁺ release substantially; 100 μmol/L Ned-19 was chosen for in vitro biological assays involving longer treatments.

Western Blot.

HUVECs were first serum-starved in EBM-2 for 4 h and then were incubated with Ned-19 for 30 min or with TSU-68 for 1 h or with SB203580 for 1 h before VEGF stimulation for 15 min. The intensity of Western blot bands was quantified by Image J software (NIH) from at least three independent experiments, normalized to β-actin content, and compared with vehicle-treated controls (set as 1).

Calcium Imaging.

HUVECs cultured on 35-mm dishes were incubated in culture medium containing 3.5 µmol/L FURA-2-AM (Invitrogen) for 1 h at 37 °C, and then rinsed with HBSS (Sigma). Each dish was placed into a culture chamber at 37 °C on the stage of an inverted fluorescence microscope (Nikon TE2000E), connected to a cooled CCD camera (512B Cascade, Roper Scientific). Samples were illuminated alternately at 340 and 380 nm using a random access monochromator (Photon Technology International) and emission was detected using a 510 nm emission filter. Images were acquired (1 ratio image per s) using Metafluor software (Universal Imaging Corporation). Calibration was obtained at the end of each experiment by maximally increasing intracellular Ca²⁺-dependent FURA-2-AM fluorescence with 5 µmol/L ionomycin (ionomycin calcium salt from Strepotimyces conglobatus, Sigma) followed by recording minimal fluorescence in a Ca²⁺-free medium. [Ca²⁺]_i was calculated according to the formulas previously described (64).

Flow Cytometric Assessment of Cell Proliferation.

Cells serum-starved in EBM-2 for 4 h were treated with VEGF and Ned-19. After 24 h, cells were harvested by trypsin/EDTA (Sigma), rinsed with PBS + 1% BSA (Sigma) and incubated with 1 µg/mL propidium iodide (PI, Sigma) plus 0,1 U/L of RNase for 3 h at room temperature. Cells were then analyzed using a Coulter Epics XL flow cytometer (Beckman Coulter) and data were analyzed using FCS3 express Software (De Novo Software) to determine the cell number for each experimental condition.

MTT Test.

To perform the methylthiazolyldiphenyl-tetrazolium (MTT) test as a measurement of cell proliferation 5 mg/mL MTT (Sigma) was dissolved in PBS and filtered. HUVECs were cultured in EGM-2 to 90% confluence, detached using trypsin/EDTA, then added onto 96-well plates at a concentration of 10 × 10⁴ cells in 200 μL of EGM-2 per well and incubated in the specific experimental condition for 24 h. As a control for background absorption, cells were omitted in some of the wells. After incubation the complete medium was removed, 100 μL of serum-free medium (EBM-2) containing 0.5 mg/mL MTT solution was added to each well and the plate was incubated in a humidified 5% (vol/vol) CO₂ incubator at 37 °C for 3 h to allow MTT to be metabolized. Then, 100 μL of DMSO (Sigma) was added to each well, pipetting up and down to dissolve crystals, and optical density was read at 550 nm.

Gelatin Zymography.

The MMP-9 in the media and in the cells was quantified separately by gelatin-substrate zymography. After 24 h of treatment, equal amounts of proteins from the cell-conditioned medium were separated on 8% (vol/vol) SDS/PAGE containing 1 mg/mL (final concentration) gelatin B. Electrophoresis was performed in nonreducing conditions, and SDS was removed by incubating the gel in 2% (vol/vol) Triton X-100 at 37 °C for 30 min. The gel was then soaked in 0.05 mol/L Tris⋅HCl buffer (pH 8.0) containing 5 mmol/L CaCl₂ for about 18 h at 37 °C and stained with 1% Coomassie brilliant blue R-250. The gelatinolytic activity was identified as a clear band. Molecular weights of the bands were estimated through the use of prestained molecular-weight markers. Pictures of the gel were taken and densitometry was performed using Image J software.

Boyden Chamber Assay.

Cell ability to migrate was evaluated by the Boyden chamber assay, which makes use of a chamber composed of two medium-filled compartments separated by a micro porous membrane (8 µm pore size, BD Bioscences). The lower well contained medium with 5% (vol/vol) serum and the chemoattractant agent (in our study: VEGF). Cells were placed in the upper chamber in complete medium containing the drug to assay (such as Ned-19) or the vehicle alone and allowed to migrate through the pores of the membrane into the lower compartment. After 24 h the membrane between the two compartments was fixed and stained, and the number of cells that had migrated to the lower side of the membrane was evaluated in a Zeiss Axioscope microscope.

Scratch Assay.

Confluent EC monolayers plated in 35-mm dishes were scraped in a straight line with a p10 pipette tip to create a “scratch.” The scratch created was of a similar size in the different experimental conditions to minimize any possible variation caused by the difference in its width. Debris was removed by a wash with PBS, which was then replaced with fresh medium in the presence or absence of VEGF and Ned-19. First images were acquired at time 0 before incubating cells at 37 °C for 24 h.

Silencing of TPC2.

Two shRNA constructs (based on the pRS vector) targeting human Tpcn2 and a control scramble sequence (noneffective 29-mer scrambled shRNA cassette, TR30012), all from OriGene, were used for cell transfection.

The targeting sequences were: ATCAGGCTGTGGTCTTCATCGAAGATGCT (TI303526) and CGTCATTGTGGCTCTTCCTGGAAACAGCA (TI303528). Transfection was carried out according to the manufacturer’s instructions for transient transfection of adherent cells (QIAGEN). Efficiency of knockdown was tested by quantitative RT-PCR 48 h after transfection.

In Vitro Matrigel Assay.

EC tube formation was evaluated by an angiogenesis in vitro assay. Briefly, 130 μL of Matrigel Basement Membrane Matrix (BD Biosciences) was added to each well of precooled 24-well tissue culture plate. Pipette tips and Matrigel solution were kept cold throughout to avoid solidification. The plate was incubated at 37 °C for 1 h to allow the matrix solution to solidify. A total of 4 × 10⁴ cells in a final volume of 500 μL of basal medium (EBM-2) were seeded onto the surface of each well containing the polymerized matrix. ECs were pretreated with the inhibitor or with vehicle alone and stimulated with the specific agonist (VEGF) for 4–24 h at 37 °C. Tube formation was inspected under an inverted microscope (Nikon Eclipse TS100) at 20× magnification and images were acquired by a digital camera (Nikon Coolpix995). The number of closed formed polygons from three independent experiments were counted in five random view-fields per well and the values averaged.

In Vivo Matrigel Plug Assay in C57BL/6 Mice.

To evaluate the ability of Ned-19 to modulate the neovascularization within matrigel plugs, matrigel (600 μL, BD Biosciences) supplemented with heparin (0.032 U/L, Schwarz Pharma S.p.A), VEGF (100 µg/L, Reliatech), TNFα (2 µg/L, R&D Systems), and Ned-19 at different doses (50, 100, and 150 µmol/L, Tocris Bioscience) was injected s.c. into the flank of C57BL/6 mice (4–5 wk old), where it rapidly formed a gel. The negative controls contained heparin alone, the positive controls heparin plus VEGF and TNFα. Within days, cells from the surrounding tissues have the opportunity to migrate into the matrigel and to form vascular structures connected to the mouse blood vessels. After 5 d, the mice were killed by CO₂ asphyxia and the angiogenic response was evaluated by macroscopic analysis at autopsy, and by measurement of the hemoglobin content in the matrigel plug. Hemoglobin was mechanically extracted from the pellets in water and measured using the Drabkin method by spectrophotometrical analysis (Sigma) at 540 nm. The values were expressed as optical density/100 mg matrigel. Histological analysis of fixed and paraffin embedded matrigel plugs was also performed using H&E stain.

Tpcn1^−/−and Tpcn2^−/− Mice.

Generation of Tpcn2^−/− mice was described elsewhere (19); these animals (Tpcn2^{Gt(YHD437)Byg}) carry an insertional mutation from a gene trap vector in Tpcn2 between exons 1 and 2. The Tpcn1^−/− line used in this study (Tpcn1^tm1Dgen) was obtained from EMMA (The European Mouse Mutant Archive). These mice have a targeted disruption of exons 4–5 that leads to absence of Tpcn1 mRNA expression (50).

The animals were housed at the University of Rome Histology Unit accredited animal facility, in individual cages in an environmentally controlled room (23 °C, 12 h light–dark cycle) and provided with food and water ad libitum. All of the procedures were approved by the Italian Ministry for Health and conducted according to the US National Institutes of Health guidelines.

In Vivo Matrigel Plug Assay in Tpcn1^−/−, Tpcn2^−/−, and WT Mice.

Matrigel (600 μL, BD Biosciences) supplemented with heparin (0.032 U/L, Schwarz Pharma S.p.A), VEGF (100 µg/L, Reliatech), TNFα (2 µg/L, R&D Systems) was injected s.c. into the flank of (4–5 wk old) WT and Tpcn1^−/− and Tpcn2^−/− mice as described above. After 5 d, the angiogenic response was evaluated by macroscopic analysis at autopsy, and by measurement of the Hb content in the matrigel plug as described before. The values were expressed as optical density/100 mg matrigel.

Statistical Analysis.

Data are presented as the mean ± SEM of results from at least three independent experiments. A Student t test was used for statistical comparison between means where applicable. *P < 0.05; **P < 0.01; ***P < 0.0002.