Introduction

Mammalian cells use intracellular ATP to fuel virtually all biochemical reactions. However, ATP can also be released in a controlled manner, regulating the functions of various cells in the circulatory system including endothelial cells and neutrophils through autocrine and paracrine signaling mechanisms that involve purinergic receptors [1]. ATP and its sequential degradation to ADP, AMP, and adenosine can activate different sets of the 19 known purinergic receptor subtypes [1, 2]. These receptors have different affinities to these nucleotides and nucleosides. The purinergic receptor family includes P1 adenosine receptors that are G protein-coupled receptors (GPCRs), ATP-gated ion channels named P2X, and GPCRs termed P2Y receptors. P2Y receptors differ in their affinities for ATP, ADP, UTP, UDP, and UDP-glucose [1]. Endothelial cells express several different purinergic receptor subsets, however, little is known about their functional roles in endothelial cell regulation [1, 3]. The P2Y2, P2Y1, P2Y11, and P2X4 receptor subtypes are the most abundantly expressed P2 receptor subtypes found in endothelial cells [3]. Among those, only the Gq/11-coupled receptor P2Y2 has been reported to be implicated in tissue regeneration. P2Y2 receptors have been shown to have a crucial function in the regeneration of epidermal tissue during wound healing [4, 5]. Nevertheless, the underlying mechanisms are unclear and it is unknown if P2Y2 receptors influence dermal blood vessel growth. It has also been suggested that P2Y1 mediates angiogenesis [6], although the agonist used in that study is not selective and can also activate other P2 receptors [7]. Two studies showed that P2Y2 can interact with VEGFR-2 leading to its activation upon stimulation with UTP [8, 9] and results using global knockout mice revealed that P2Y2 mediates collateral vessel formation after ischemia [10]. Furthermore, it was previously shown that the G protein Gq/11 is required for VEGF-induced angiogenesis [11], however, the receptor responsible for these effects remains unknown. Taken together, it is increasingly apparent that P2Y2 may be among the candidate GPCRs involved, as it couples to Gq/11 and can interfere with VEGFR-2 signaling via a mechanism involving Src [12,13,14]. Yet, experimental evidence demonstrating that P2Y2 is a key driver for VEGF-induced angiogenic sprouting has not been provided so far. In this study, we report that modulation of the expression of P2Y2 receptors on endothelial cells promotes endothelial sprouting by stimulating VEGFR-2 receptor signaling and driving endothelial cells towards a proangiogenic phenotype.

Methods

Reagents

Chemicals and antibodies used in this study are listed in the supplementary files (Supplementary Table 1). All other reagents are described in the respective section.

Cell culture

Human placentae and umbilical cords were obtained at term pregnancy during Caesarean sections with written and informed maternal consent and collection approved by a local ethical committee from Upper Austria (Ethikkommission des Landes Oberösterreich, #200). All methods were performed according to the relevant guidelines and regulations of the local ethical board. Until processing, placentae were stored in sterile bags (Websinger) in Ringer solution (Fresenius, Austria). Human umbilical vein endothelial cells (HUVEC) were either isolated as described previously [15] or purchased as a pooled donor batch from Lonza and used at passages 4–9. Mesenchymal stromal cells (MSC) were isolated from adipose tissue obtained from liposuction material as described before [16] and used at passages 2–10. Induced pluripotent stem cell-derived endothelial colony-forming cells (iPSC-ECFC) were obtained from Axolbio (United Kingdom) and used at passage 4–8. All cells were maintained in Endothelial Cell Growth Medium-2 (EGM-2) from Lonza or Promocell consisting of Endothelial Cell Basal Medium (EBM-2) supplemented with 5% fetal calf serum (FCS) and endothelial growth supplements. To determine that all effects occur specifically due to interference with P2Y2 signaling, we controlled for donor, passage, and virus infection and only included cells within one experiment if they were from the same donor, used at the same passage and were retrovirally infected at the same time. All HUVEC data shown were obtained from several independent experiments where at least two different biological donors from at least two different virus infections were used. In total, P2Y2 was overexpressed in cells derived from six donors, while cells from at least two donors were included in selected experiments such as knockdown/inhibition of P2Y2. Normal spheroid sprouting and fibrin matrix assays were performed with P2Y OE2 -HUVEC generated from all six donors.

siRNA-mediated knockdown

Transfection protocols and sequences and a pool of three different siRNAs were used as previously published [14, 17]. siRNAs were purchased from Qiagen. One-third of each siRNA was used in the transfection mix to knockdown P2Y2 in HUVEC.

Proliferation assay and measurement of extracellular ATP

5 × 103 HUVEC were seeded in 96-well plates and left to attach overnight. Then, cells were washed with 1 × PBS and incubated in EBM-2 ± VEGF at a concentration of 50 ng/ml for 48 h. BrdU assay was then performed as described by the manufacturer (Roche).

To measure ATP release, HUVEC were grown in 96-well plates to confluence. Cells were starved for 4 h in EBM-2 + 1% BSA (Sigma) before stimulation with VEGF 50 ng/ml or thrombin at 1 U/ml for 10 min. Supernatants were then transferred into opaque plates and ATP was determined luminometrically as previously described [18].

Plasmids and retroviral infection of HUVEC

EGFP and mCherry in pLV vectors and pBMN-Z were purchased from Addgene. pcDNA3-EYFP-HIS was purchased from Invitrogen (Thermo Fisher). Plasmids encoding human P2Y2-YFP and human P2Y11-YFP were previously described [19]. EGFP and mCherry were subcloned into pBMN after digestion with BamHI and SalI. EYFP-HIS was subcloned into pBMN after digestion with BamHI and EcoRI. Both P2Y2-YFP and P2Y11-YFP were digested with XhoI and blunted before being digested with HindIII. pBMN-Z was digested with NotI, and blunted and digested with HindIII to enable ligation with P2Y2-YFP or P2Y11-YFP. Successful subcloning of both constructs was confirmed by sequencing. Retroviral infection was performed as described previously [20]. Briefly, Phoenix ampho cells (a kind gift of Regina Grillari, University of Natural Resources and Life Sciences, Vienna) were cultured in DMEM (high glucose) and 10% FCS. Virus particle generation was performed by transfecting Phoenix ampho cells at 80% confluency using Lipofectamine 2000 or TurboFect (both from Thermo Fisher) according to the manufacturer’s instructions. Supernatant containing virus particles was mixed 50:50 with full medium and transferred onto 80% confluent HUVEC and incubated overnight. YFP-HUVEC, GFP-HUVEC, or mCherry-HUVEC were termed as CTRL-HUVEC, while P2Y2-YFP-HUVEC were labeled as P2Y OE2 - and P2Y11-YFP-HUVEC as P2Y OE11 -HUVEC, respectively. All cells were then expanded in new flasks and used for subsequent experiments.

Embedding of cells in fibrin matrices

Fibrin matrices were generated as described previously [20]. Briefly, fibrin gel components (TISSEEL®, Baxter) were prepared by warming frozen fibrinogen to room temperature and diluting 4 U/ml thrombin 1:10 in calcium chloride (CaCl2). Cells, fibrinogen, and thrombin were mixed and pipetted onto prepared coverslips. Gels polymerized at room temperature for 30 min before growth medium with aprotinin was added. The final concentration of fibrinogen used was 2.5 mg/ml, of thrombin 0.2 U/ml, and of aprotinin 100 KIU/ml. Fibrin scaffolds had a volume of 200 µl and contained 500 endothelial cells (250:250 in mosaic assays) and 500 MSC per µl fibrin. If spheroids were embedded in fibrin, the final concentration of fibrinogen was 2.5 mg/ml, of thrombin 1 U/ml, and of aprotinin 100 KIU/ml. For VEGFR-2 inhibition experiments, the selective antagonist Apatinib (YN968D1, MedChemExpress) was reconstituted in DMSO. To block VEGFR-2 in fibrin matrix assays, this antagonist was added to the normal growth medium (EGM-2, Lonza) at 1 µM with every medium change (i.e., every 2–3 days). For P2Y2 inhibition experiments, the selective antagonist AR-C 118925XX (Tocris) was reconstituted in DMSO. With every medium change, this antagonist was added to the growth medium (EGM-2, Promocell) at a concentration of 30 µM, which was previously shown to inhibit P2Y2 function in ECs [14].

Spheroid sprouting assay

Generation of spheroids and the sprouting assay were performed as described elsewhere [21]. Briefly, methylcellulose medium was prepared by dissolving 0.6 g of autoclaved methylcellulose (Sigma) in 50 ml EBM-2 (Lonza or Promocell). Cell number was adjusted to generate spheroids containing 1000 cells per spheroid. These cells were suspended in 80% EGM-2 and 20% methylcellulose medium and grown in hanging drops overnight. Spheroids were then harvested, centrifuged, and mixed with 80% methylcellulose medium and 20% FCS to avoid sedimentation of spheroids. Rat tail collagen I (Corning) was diluted to 2 mg/ml with water and 10 × DMEM high glucose for equilibration. Then, equal amounts of collagen solution and spheroid suspension were mixed and pipetted into pre-warmed wells of a 48-well plate. Alternatively, harvested spheroids were embedded in fibrin matrices as described above. After 1 h of incubation at 37 °C, EBM-2 with or without hVEGF-A165 (Peprotech or Evercyte) at a concentration of 50 ng/ml was added on top of collagen gels and samples were incubated overnight before quantification.

RNA isolation, transcriptome analysis, and RT-PCR

RNA was isolated from confluent cell monolayers using Trizol (Life Technologies) as described previously [22]. Consequently, isolated RNA from two biological replicates was used to produce biotinylated cRNA, and then purified, fragmented, and hybridized to GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix) as described elsewhere [22]. The resulting .CEL files were summarized in the Affymetrix Console Software (version 1.4.1.46) using the Robust Multi-array Average (RMA) normalization algorithm (3′ Expression Arrays: RMA, log2). Differential expression analysis was performed in ExAtlas, a gene expression meta-analysis tool [23]. For explorative pathway analysis, genes with a fold change of two or higher, in either direction, were analyzed through the use of IPA (QIAGEN Inc.) [24], focusing on endothelial cells. Gene Set Enrichment Analysis (GSEA) was done using the log2 ratio of classes (P2Y OE2 vs. CTRL) as the ranking metric. Gene sets were taken from the Molecular Signatures Database v5.2 [25]. Figures were generated using Morpheus [26] and ggplot2.

To perform RT-PCR, complementary DNA (cDNA) was synthesized using an EasyScript™ cDNA Synthesis Kit (abmGood). DNA concentration was measured with spectrophotometer (Nanodrop OneC, Thermo Fisher Scientific). cDNA was mixed with reaction buffer, dNTPs, selected primer (Supplementary Table 2), and Hot Start Taq DNA Polymerase (all from New England Biolabs). After an initial denaturation step of 95 °C for 30 s, samples were subjected to 30 cycles of 95 °C for 30 s, 55 °C for 60 s, and 68 °C for 60 s. Then, the procedure was terminated by a final extension step of 68 °C for 5 min. The amplified products were mixed with loading buffer (New England Biolabs), loaded into 1% agarose gels containing ethidium bromide, and subjected to electrophoresis.

Flow cytometry

Cells were detached using Accutase (Sigma), and centrifuged and fixed using 1% formaldehyde on ice for 15 min. To label P2Y2 or to detect total protein levels, cells were washed once and incubated in 1 × permeabilization buffer for 15 min on ice. Then, cells were stained with selected antibodies on ice as described previously [27]. To adequately detect the antibody signal, compensation settings were applied to reduce the background of the YFP fluorescence. Platelets were isolated from whole blood donated by human volunteers according to a standard protocol from Abcam (Isolation of human platelets from whole blood, Abcam). Platelets were fixed and subjected to the same P2Y2 staining procedure as mentioned above. All samples were measured on a Cytoflex flow cytometer (Beckman Coulter) and data were analyzed using FlowJo X 10.0.7.

Immunoprecipitation and immunoblotting

For immunoprecipitation, cells were grown to confluency in T75 tissue culture flasks. HUVEC were then starved for 4 h in EBM-2 + 1% BSA (Sigma) before stimulation with 100 ng/ml VEGF for 10 min. Cells were then immediately put on ice and lysed in 1 ml PLCLB lysis buffer (150 mM NaCl, 5% glycerol, 1% Triton X-100, 1.5 mM MgCl2, and 50 mM HEPES, pH 7.5) supplemented with 2 mM sodium orthovanadate, and 1 × protease inhibitor mix (Complete Mini, Roche). Cleared lysates were incubated with anti-human VEGFR-2 (clone D5B1, 1:200, Cell Signaling Technology) primary antibody for 2 h. Subsequently, the immunocomplexes were captured using protein G-Sepharose (Abcam). After three washing steps in PLCLB buffer, the proteins were separated in 7% polyacrylamide gels under reducing conditions. After blotting of the proteins to nitrocellulose membranes and blocking of the membranes in 5% nonfat milk, membranes were probed with monoclonal antibodies against VEGFR-2 (clone D5B1, 1:1000, Cell Signaling Technology) or phosphotyrosine (clone 4G10, 1:1000, Merck). After incubation with the secondary anti-rabbit (Acris Antibodies) or anti-mouse (Jackson Immuno Research) HRP-coupled antibodies, respectively, the signal was visualized by chemiluminescence (BioRad). Cell lysis of confluent, unstimulated HUVEC, and processing and immunoblotting of ERK 1/2, p-ERK 1/2, AKT, p-AKT, rS6, and p-rS6 were performed as described elsewhere [18].

Immunofluorescence staining

Endothelial cells were cultured in borosilicate glass chamber slides (Nunc). Cells were washed with 1 × PBS, fixed with 4% paraformaldehyde for 15 min, and permeabilized using 1 × permeabilization buffer (eBiosciene) for 15 min at room temperature. Cells were then stained with anti-human P2Y2 antibodies (1:200) overnight at 4 °C on a shaker following incubation with anti-rabbit Alexa Fluor 594 antibody (Invitrogen) for 1 h at room temperature. Images were taken on a laser scanning confocal microscope (Zeiss LSM510, Zeiss).

Quantification of endothelial networks, sprouts, and histological cross sections

Images of endothelial networks in fibrin were taken on an epifluorescence microscope (Zeiss Observer A1) and used for analysis. For network quantification, 3–4 images per sample were taken and analyzed as described previously [20, 27, 28]. Images were processed using Adobe Photoshop CS5 (Adobe Systems) and analyzed in a blinded manner. Measurement of number of tubules, junctions, and total and mean tubule length was done with Angiosys software (TCS Cellworks). To quantify sprouting parameters, z-stacks of spheroids were taken using a Leica DMI6000B microscope at ~ 10 µm intervals to a depth of ~ 200 µm, merged and quantified using ImageJ. All images were randomized and analyzed in a blinded manner.

Statistics

Statistical evaluations were performed with GraphPad Prism 5 software (GraphPad). Data from spheroid, fibrin matrix assays, and immunoprecipitations were analyzed by two-way ANOVA. siRNA knockdown efficiency was analyzed by paired Student’s t test. All other data were analyzed by Mann–Whitney test. Values were considered significant when p < 0.05.

Results

Detection of P2Y2 overexpression in HUVEC

To identify the role of P2Y2 in vascular network formation, we overexpressed P2Y2-YFP (P2Y OE2 ) [19] in human umbilical vein endothelial cells (HUVEC) and measured its expression via flow cytometry and RT-PCR. To confirm the specificity of the antibodies used, a P2Y2 staining and flow cytometric analysis of human platelets, control HUVEC, and P2Y2-overexpressing HUVEC was performed. In agreement with the previous reports, P2Y2 expression on platelets could not be detected [29], while moderate and high expression was detected on control and P2Y OE2 -HUVEC, respectively (Fig. 1a). In addition, a strong increase in P2Y2 mRNA could be detected upon P2Y2 overexpression in HUVEC (Fig. 1b). P2Y2-YFP was detected in the cytoplasm, on cell–cell borders and in filopodia of HUVEC cultured in a 3D fibrin matrix as visualized (Supplementary movie 1). Furthermore, staining of cultured endothelial cells with an anti-P2Y2 antibody co-localized with the YFP fluorescence signal further demonstrating the antibody specificity (Fig. 1c). Upon knockdown of P2Y2 (P2Y KD2 ) using previously published P2Y2 siRNA sequences and transfection protocols [14, 17], P2Y2 protein expression was only reduced in P2Y OE2 -cells but not in CTRL cells (Fig. 1d, e), indicating relatively low endogenous P2Y2 turnover rate in ECs [30].

Fig. 1
figure 1

Detection of P2Y2 overexpression in HUVEC. a Antibody specificity was determined by antibody staining of human platelets, control, and P2Y OE2 -HUVEC and detection by flow cytometry. Platelets did not show any expression of P2Y2, while moderate expression was detected on control cells which was increased in P2Y OE2 -HUVEC. b P2Y2 mRNA was strongly expressed in P2Y OE2 -HUVEC compared to controls as determined by RT-PCR. c P2Y2-overexpressing and vehicle-infected HUVEC were stained against P2Y2. A strong co-localization of the antibody signal and the P2Y2-fusion protein could be detected in P2Y OE2 cells particularly on cell borders. d siRNA-mediated knockdown of P2Y2 reduced P2Y2 protein expression only in P2Y OE2 cells. e On average, this resulted in a 10% reduction of protein levels in controls and 47% reduction in P2Y OE-2 HUVEC compared to cells transfected with control siRNA. Platelets were isolated from one donor. All other data are representative for three independent experiments conducted with three different HUVEC donors. * indicate a p value of < 0.05; scale bar 20 µm

Full size image

P2Y2 receptors influence sprouting and vasculogenesis

Modulation of purinergic P2Y receptor expression has previously been shown to greatly influence numerous cell functions including cell differentiation and proliferation [31,32,33]. To determine if and how P2Y2 expression alters sprouting and vascularization, we induced receptor overexpression and performed siRNA-mediated knockdown in HUVEC and examined the effects of these treatments in spheroid sprouting and fibrin co-culture assays. Neither overexpression nor knockdown of P2Y2 significantly influences the percentage of cells present in the gate used for flow cytometry analysis to determine protein expression, suggesting that cell viability is not affected by any of the above-named procedures (data not shown). Overexpression of P2Y2 resulted in an increase of spontaneous sprouting, however, being independent of additional VEGF. siRNA-mediated knockdown of P2Y2 reversed this effect and also led to impaired sprouting in response to VEGF treatment (Fig. 2a). Quantification of the number of sprouts and their cumulative lengths revealed a dependence of sprouting ability on the expression of P2Y2 (Fig. 2b). No significant difference could be detected in the average length of sprouts upon P2Y2 overexpression. However, knockdown of P2Y2 led to a reduction of the average length in VEGF-treated P2Y OE2 -HUVEC (Fig. 2b). These results demonstrate that P2Y2 receptors promote sprouting in endothelial cells and that silencing of P2Y2 receptor signaling impairs sprouting. Since this sprouting assay lacks vascular pericytes and only reflects the initial sprouting events (hours) but not functional vascular network formation (days), we employed a fibrin co-culture model of endothelial cells and mesenchymal stromal cells (MSC) [20, 27] to study how P2Y2 receptors influence vascular network formation. In line with the sprouting assay results, overexpression of P2Y2 in HUVEC induced the formation of a primitive vascular network embedded in a fibrin matrix. This occurred in the absence of supporting cells, an effect that was not observed with CTRL-HUVEC. P2Y2 receptor expression had no discernible effect when HUVEC were co-cultured with MSC. However, vascular network formation was inhibited when cells were cultured in the presence of the selective P2Y2 antagonist AR-C 118925XX (Fig. 2c) that led to significantly reduced numbers of junctions, tubules and total tubule lengths (Fig. 2d). These data indicate that endothelial P2Y2 receptors drive vascular network formation.

Fig. 2
figure 2

P2Y2 influences sprouting and vasculogenesis in HUVEC in vitro. a Representative images of HUVEC spheroids. P2Y2 overexpression (OE) or knockdown (KD) correlated with sprouting potential independently of VEGF 50 ng/ml addition. b Quantitative analysis of spheroids revealed that sprouting behaviour is significantly affected by P2Y2 expression in HUVEC spheroids as determined by the number of sprouts, average length, and cumulative length of sprouts per spheroid. c Representative images of HUVEC embedded in fibrin either alone or in co-culture with MSC, showing that enhanced P2Y2 expression or inhibition of P2Y2 by the selective antagonist AR-C 118925XX (30 µM) modulated vasculogenic network formation in fibrin. d Analysis of network parameters reveal that P2Y2 is essential for vascular network formation in fibrin. Graphs show data obtained from two independent experiments using two different HUVEC donors *, ** and *** indicate p values of < 0.05, 0.01, and 0.001, respectively. Scale bars indicate 100 µm

Full size image

P2Y2 overexpression increases expression of proangiogenic genes

To analyze how P2Y2 overexpression regulates the angiogenic potential of endothelial cells, we performed microarray expression analysis. Differentially expressed genes induced by P2Y2 receptor expression in HUVEC were primarily related to the development and function of the cardiovascular system (Fig. 3a). Many of these genes are directly involved in angiogenesis and vasculogenesis (Fig. 3b). Pathway enrichment analysis using two different databases revealed further evidence for the notion that P2Y2 expression regulates endothelial cell physiology (Sup. Fig. 1a, b). Analysis of genes essential in angiogenesis showed a strong increase in transcription of several proteins that are important for sprouting, including CXCR4, CD34, and ANGPT2 (Fig. 3c). In fact, upon P2Y2 overexpression, transcription of these and other relevant genes was increased by more than tenfold compared to basal expression (Fig. 3d). In addition, we confirmed the gene array data for CXCR4, CD34, and ANGPT2 using RT-PCR (Fig. 3e). Transcription factor network analysis revealed that NFKBIA, FOXO1, ERG, EPAS1, STAT3, NFE2L2, NOTCH1, KLF4, and PROX1 are upstream regulators of P2Y2 expression (Fig. 3f), while additional transcription factor targets seem to be enriched in response to P2Y2 overexpression (Sup. Fig. 1c).

Fig. 3
figure 3

Gene array analysis reveals increased expression of proangiogenic genes upon P2Y2 overexpression. a IPA-based functional annotation of genes which expression was significantly altered in P2Y OE2 -HUVEC, showing that mostly cardiovascular system functions and b within this gene set, primarily genes regulating angiogenesis and vasculogenesis were affected by P2Y2 overexpression. c Heatmap showing normalized gene expression of these angiogenesis-related genes in two separate donors. A strong increase upon P2Y2 overexpression of CXCR4, CD34, and ANGPT2 could be observed in both donors compared to controls. d Mean intensity values of angiogenesis genes demonstrating the strong differences in gene expression in P2Y2-overexpressing cells compared to control cells. CXCR4, CD34, and ANGPT2 showed on average a 10- to 100-fold increase in gene expression. e Confirmation of increased gene expression of CXCR4, CD34, and ANGPT2 by RT-PCR. f Transcription factor network analysis showing predicted transcriptional upstream regulators based on the gene expression pattern in P2Y2-overexpressing HUVEC. Data were obtained from two different HUVEC donor batches containing pools of three and four donors

Full size image

Enhanced P2Y2 expression drives endothelial cells towards a proangiogenic phenotype

To further elucidate the role of P2Y2 receptors in endothelial cells, we analyzed signaling downstream of P2Y2 and VEGFR-2 activation. P2Y2 overexpression impaired the ability of cells to proliferate in response to VEGF (Fig. 4a). Together with our gene array and functional assay data, this finding suggests that P2Y2 overexpression results in constitutively active VEGF receptor signaling. Furthermore, we found significantly increased phosphorylation of ERK 1/2 in P2Y OE2 -HUVEC compared to control cells (Fig. 4b). We then assessed whether P2Y2 levels affect the cell-autonomous ability to form sprouts and vascular networks. Therefore, mosaic spheroids or cell-containing fibrin gels were generated by mixing endothelial cells expressing normal or elevated levels of P2Y2 to determine which cell type is at the tip or participating in network formation. Control cells were retrovirally transduced with YFP or mCherry, while P2Y2 overexpression was achieved by transduction with a fusion of P2Y2 with YFP. We found that only P2Y2-overexpressing cells sprouted or formed a primitive network in a competitive mosaic spheroid or fibrin matrix assay at baseline conditions, while control cells did not, suggesting that no paracrine factors are involved in this mechanism (Fig. 4c, d). Indeed, in tip cell or vascular network-favoring conditions (i.e., when paracrine factors such as VEGF are added or secreted by co-cultured MSC), control cells sprouted or formed a vascular network resulting in equal contribution of P2Y OE2 and CTRL cells to the sprouts or vascular networks (Sup. Fig. 2a, b). To determine if exogenous stimulation of P2Y2 can influence sprouting, we treated EC spheroids with 100 µM UTP, another agonist of P2Y2. This treatment did not affect sprouting parameters of CTRL-HUVEC under any condition tested (Fig. 4e, f). Instead, we found filopodia-like structures that emerged from spheroids when P2Y OE2 -HUVEC were treated with UTP (Fig. 4e). These results suggest that increased P2Y2 expression elicits this functional response to UTP. Furthermore, we found that constitutive phosphorylation of VEGFR-2 in unstimulated P2Y OE2 -HUVEC was similar to that of VEGF-treated CTRL-HUVEC, which may explain why P2Y2-overexpressing cells did not respond to VEGF treatments in functional assays (Fig. 4g). In addition, we observed that total VEGFR-2 protein levels were decreased in P2Y OE2 -HUVEC (Fig. 4g). Based on our results from spheroid and fibrin matrix assays, we speculated that VEGF may induce the release of ATP to initiate autocrine P2Y2 receptor signaling. In contrast to thrombin as positive control [34], VEGF stimulation of HUVEC did not result in measurable accumulation of extracellular ATP (Fig. 4h). In addition to P2Y2, it has also been shown that endothelial cells abundantly express the purinergic receptor P2Y11 which responds to the same ligand as P2Y2 and couples to the same G protein [3]. However, overexpression of P2Y11 in HUVEC did not influence sprouting in a spheroid sprouting assay (Sup. Fig. 2c, d) demonstrating the specific involvement of P2Y2 in endothelial sprouting. Furthermore, we analyzed the influence of P2Y2 overexpression on the mTOR and AKT pathways, and found rS6 and AKT phosphorylation unchanged in unstimulated cells (Sup. Fig. 2e, f). Together, these data suggest that P2Y2 is required to facilitate VEGF-induced signaling and angiogenesis.

Fig. 4
figure 4

Mechanisms of P2Y2-induced vascularization. a VEGF failed to induce proliferation in cells overexpressing P2Y2 as determined by BrdU assay. b Total cell lysates of control and P2Y OE2 -HUVEC were subjected to immunoblotting with antibodies which recognize ERK 1/2 and p-ERK 1/2. Constitutive phosphorylation of ERK 1/2 could be measured in cells with enhanced P2Y2 expression. Representative images of collagen mosaic spheroid (c) and fibrin matrix assays (d) where either vehicle-infected or P2Y OE2 -HUVEC (both green) were co-embedded with vehicle-infected HUVEC (red). Mainly P2Y OE2 cells sprouted and formed a vascular network, while control cells did not, indicating an autocrine mechanism. e Representative images of UTP-stimulated spheroids, showing that no difference in sprouting upon stimulation could be observed in control cells, while appearance of filopodia-like structures emerged only from UTP-stimulated P2Y OE2 -HUVEC (arrows). f Quantification of number of sprouts, average length, and cumulative length demonstrates that UTP stimulation had no effect on these sprouting parameters in either sample. g VEGFR-2 is constitutively phosphorylated in cells overexpressing P2Y2 as determined by immunoprecipitation and blotting against phosphorylated tyrosine. A significant increase in tyrosine phosphorylation and decrease of total VEGFR-2 levels were observed upon P2Y2 overexpression. h ATP content was determined in supernatants of stimulated HUVEC. VEGF 50 ng/ml does not trigger ATP release of HUVEC after 10 min of stimulation. Thrombin 1 U/ml was used as a positive control. Proliferation assay, ATP release experiment, normal and mosaic spheroid, and fibrin matrix assay data were obtained from two independent experiments using two donors. Immunoblots and immunoprecipitations were performed at least four times with protein lysates obtained from three HUVEC donors. * and ** indicate p values of < 0.05 and 0.01, respectively. Scale bars indicate 100 µm

Full size image

Enhanced sprouting in response to P2Y2 overexpression is VEGFR-2 dependent

To confirm the involvement of VEGFR-2 in spontaneous endothelial sprouting of P2Y OE2 -HUVEC, we applied Apatinib, a selective VEGFR-2 inhibitor [35], to cells embedded in a fibrin matrix. Treatment of fibrin-embedded P2Y OE2 -HUVEC with 1 µM Apatinib visibly impaired network formation after incubation of 1 week (Fig. 5a). The inhibition of VEGFR-2 significantly reduced the number of junctions, tubules and the total network length compared to DMSO-treated controls (Fig. 5b). These data indicate that signaling via VEGFR-2 is at least in part responsible for the observed effects despite its reduced expression in P2Y OE2 -HUVEC. To elucidate if different Tie1 or Tie2 protein levels were responsible for the observed reduction of VEGFR-2 protein expression [36, 37] upon P2Y2 overexpression, we performed flow cytometry analysis. Tie1 total, but not surface, protein levels were influenced by P2Y2 overexpression in HUVEC (Fig. 5c). Interestingly, we observed that surface Tie2 levels were decreased, while total protein levels were not significantly affected by enhanced P2Y2 expression compared to controls, suggesting that Tie2 is internalized (Fig. 5d). Moreover, we did not find any differences in VEGFR-3 protein expression due to enhanced P2Y2 expression (Sup. Fig. 2g).

Fig. 5
figure 5

Role of VEGFR-2, Tie1 and Tie2 in P2Y2-overexpressing cells. a Representative images of P2Y OE2 -HUVEC embedded in fibrin matrices which were incubated with or without 1 µM Apatinib. b Analysis of vascular network parameters reveals that inhibition of VEGFR-2 significantly reduced the number of junctions, tubules, and the total network length but not the average tubule length of P2Y2-overexpressing HUVEC. c Tie1 surface and total protein levels were measured using flow cytometry. No significant difference can be observed in Tie1 surface protein expression in P2Y OE2 cells, while total protein levels were significantly decreased. d Tie2 surface and total protein levels were measured using flow cytometry. Upon P2Y2 overexpression, only surface but not total protein levels of Tie2 were significantly reduced. Graphs show data obtained from a minimum of three independent experiments using two different HUVEC donors. *, **, and *** indicate p values of < 0.05, 0.01, and 0.001, respectively. Scale bars indicate 100 µm

Full size image

P2Y2 influences sprouting and vasculogenesis in iPSC-ECFC

Endothelial cells may show differences in their phenotype depending on their organotypic origin in the human body [20]. To confirm that the influence of P2Y2 overexpression is not specific to endothelial cells from umbilical veins, we overexpressed P2Y2 in induced pluripotent stem cell-derived endothelial colony-forming cells (iPSC-ECFC). Flow cytometry analysis of endothelial surface proteins revealed that P2Y2 regulates surface expression of CD34, VEGFR-2, and Tie2 in both HUVEC and iPSC-ECFC to similar extents (Fig. 6a). Quantification of fluorescence intensity showed that surface levels of VEGFR-2 were significantly decreased, while CD34 levels were significantly increased upon P2Y2 overexpression (Fig. 6b). This is in line with our gene array and immunoblotting data (Figs. 3c–e, 4g). Surface protein levels of CD31, VE-Cadherin, CD73, and CD146 were either not influenced by P2Y2 or only changed in one cell type. In addition, we found increased sprouting and vascular network formation upon P2Y2 overexpression in the absence of an angiogenic stimulus in iPSC-ECFC (Fig. 6c–f). The increase in vascularization and sprouting upon enhanced P2Y2 expression could, therefore, be recapitulated in these cells, suggesting that the observed effects are not limited to specific endothelial cell types. Taken together, these data confirm that enhanced P2Y2 expression drives endothelial cells into a proangiogenic phenotype regardless of their origin.

Fig. 6
figure 6

P2Y2 influences surface protein levels, sprouting, and vasculogenesis in iPSC-ECFC and HUVEC in vitro. a Flow cytometry measurements showed the influence of P2Y2 overexpression on expression of selected surface markers in both HUVEC and iPSC-ECFC. b Analysis of geometric mean fluorescence (GMF) intensity reveals that surface protein levels of VEGFR-2 are decreased and of CD34 increased upon enhanced expression of P2Y2 in HUVEC. c Representative images of iPSC-derived ECFC embedded in fibrin either alone or in co-culture with MSC. Overexpression of P2Y2 in these cells results in formation of a primitive vascular network. d Analysis of vascular network parameters demonstrates a significant induction of network formation upon P2Y2 overexpression. e Images of spheroids with iPSC-ECFC showing that P2Y2 overexpression results in induction of sprouting. f Quantification of iPSC-ECFC spheroid images indicates a significant increase in sprouting potential upon P2Y2 overexpression as determined by the number of sprouts, average length, and cumulative length per spheroid. Graphs show data obtained from a minimum of two independent experiments using different two HUVEC donors and one iPSC-ECFC donor. *, **, and *** indicate p values of < 0.05, 0.01, and 0.001, respectively. Scale bars indicate 100 µm

Full size image

Discussion

Here, we report a novel role for the purinergic P2Y2 receptor in the sprouting of endothelial cells. To determine the function of the P2Y2 receptor during vascular growth, we forced overexpression of the receptor in ECs. Our results presented here show that sprouting and vascular network formation were significantly dependent on endothelial P2Y2 expression and activity. Furthermore, our data indicate that ECs acquire a proangiogenic phenotype and lose their responsiveness to VEGF when the expression of P2Y2 is enhanced. This is likely mediated by amplification of VEGFR-2 signaling and, thus, regulation of a major pathway of angiogenesis [13]. Indeed, inhibition of VEGFR-2 significantly impaired spontaneous vascular network formation induced by P2Y2 overexpression in our experiments, proving that VEGFR-2 is involved in this process. Interestingly, it was previously reported that stimulation of ECs with uridine adenosine tetraphosphate (Up4A) promotes angiogenesis in a 3D co-culture assay with ECs and pericytes as well as upregulation of several proangiogenic and purinergic genes including P2Y2, P2Y4, and P2Y6 [39]. Consequently, the Up4A-induced angiogenic activity could be blocked by a selective antagonist against P2Y6, suggesting that other purinergic receptors could be at least in part involved in driving Up4A-induced vascularization. To focus on the role of P2Y2 in vascular growth, we additionally carried out functional assays using siRNA-mediated knockdown and AR-C118925, a small molecule antagonist targeting P2Y2. It was previously demonstrated that this antagonist has at least 50-fold selectivity against all other human P2X and P2Y receptors, except for the P2X1 and P2X3 receptors, which were blocked by AR-C118925 at concentrations of about 1 μM [38]. However, P2X1 and P2X3 are barely detectable in endothelial cells, suggesting that these receptors may not interfere with the effects observed here [3]. Therefore, we chose to incorporate this highly selective antagonist in our assays to highlight the importance of P2Y2 in endothelial cells.

P2Y2 is known to be able to influence signaling of several tyrosine kinase receptors including epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR), and VEGFR-2 [8, 9, 12, 14, 40]. This receptor transactivation is mediated by Src-dependent activation of proline-rich tyrosine kinase 2 (Pyk2), a focal adhesion-related tyrosine kinase which is required for GPCR-mediated transactivation of receptor tyrosine kinases such as EGFR [12, 41]. In line with previously published reports, our data presented here demonstrate that interference with P2Y2 receptor signaling can influence VEGFR-2 activity [8, 9, 14, 40]. Thus, constitutive interactions between these two receptors may allow P2Y2 receptors to define functional cell responses to VEGF. However, the role of this transactivation mechanism has been unclear. Collectively, our observations here indicate that transactivation of VEGFR-2 by P2Y2 is essential for vascular network formation and sprouting. Notably, P2Y2 overexpression induces expression of several proangiogenic genes including CXCR4, CD34, and ANGPT2 in ECs. All three markers are reported to be enriched in tip cells [42,43,44]. This suggests that endothelial cells acquire a proangiogenic phenotype upon increased P2Y2 expression, since tip cell-enriched genes seem to be co-regulated with P2Y2.

We furthermore found increased CD34 and lower Tie2 surface protein levels, which represent another critical hallmark of angiogenic tip cells [44, 45]. Conversely, we detected decreased VEGFR-2 expression which has been observed in stalk cells only [46]. Nevertheless, the data presented here show lower expression of VEGFR-2 in P2Y OE2 -HUVEC, yet higher phosphorylation which may act compensatory for reduced protein levels. Moreover, it has been shown that angiogenesis can occur upon inhibition of Notch even in the absence of VEGFR-2 if VEGFR-3 is upregulated [47]. Interestingly, our gene array analysis revealed an increased expression of VEGFR-3 in P2Y OE2 -HUVEC which may substitute for lower VEGFR-2 expression levels to induce angiogenesis. However, we did not find significant differences in surface and total VEGFR-3 protein levels despite an observed increased expression in our gene array data. This indicates that VEGFR-3 may not be involved in the observed effects. In addition, our data here and previously published results suggest a correlation in protein expression between VEGFR-2 and Tie1 [36, 37]. One can speculate on a crosstalk of VEGF/VEGFR-2 and Ang/Tie signaling pathways upon P2Y2 overexpression. VEGFR-2 internalization, recycling, and degradation can also be regulated by expression or junctional localization of VE-Cadherin [48]. Interestingly, P2Y2 has been reported to be able to form a complex with VEGFR-2 and VE-Cadherin to regulate intracellular signaling factors such as Rac1 [9]. This complex formation could influence the junctional association of VE-Cadherin and thereby regulate VEGFR-2 levels. We have not detected any changes in the surface expression of VE-Cadherin, however, activation of P2Y2 has been shown to induce phosphorylation of VE-Cadherin without influencing its expression [9].

Modulation of purinergic receptor expression has been shown to drive a variety of tasks in cells including cell differentiation and p38 MAPK activation as a response to injury [31,32,33]. Recently, it was reported that overexpression of P2Y2 in human cardiac progenitor cells isolated from patients suffering from heart failure improved migration and proliferation and thereby the regenerative potential of these cells [49]. This is similar to our observations where forced expression of P2Y2 in human ECs drives sprouting and vascular network formation, demonstrating that this receptor may perform similar functions in different tissues. Interestingly, we found that ERK 1/2 and VEGFR-2 were constitutively phosphorylated, suggesting that P2Y2 may be continuously activated. Notably, reduced proliferation and increased phosphorylation of ERK 1/2 have been reported to be the characteristics of endothelial tip cells [46, 50].

The constitutive activation of P2Y2 could be a result of permanent stimulation by a ligand or of increased basal receptor activity. It has been shown that a basal activity of Gq/11 signaling exists in endothelial cells and that modulation of this activity by influencing G protein expression facilitates VEGFR-2 phosphorylation [11]. This is in line with our data, demonstrating that P2Y2 receptor overexpression induces VEGFR-2 activation in the absence of an exogenous ligand. Still, VEGF-induced release of P2Y2 ligands may provide additional information on the connection of VEGFR-2 and P2Y2, but we failed to detect any release of ATP in response to VEGF or any functional response to UTP stimulation unless P2Y2 was overexpressed. This observation is supported by other reports, indicating that altered P2Y receptor expression affects essential EC functions [31,32,33]. Nevertheless, we cannot rule out the possibility that VEGF stimulation causes the release of small amounts of localized ATP and that autocrine stimulation of P2Y2 receptors regulates functional HUVEC response to VEGF. Similar mechanisms have previously been reported to regulate pseudopodia protrusion of neutrophils during chemotaxis [2]. Release of ATP and autocrine signaling of P2Y2 activates various mechanisms [2, 14], however, enhanced expression of P2Y2 alone has been reported to be sufficient for activation of downstream signaling pathways [49]. This seems to be P2Y2-specific as overexpression of another ATP-activated and Gq/11-coupled receptor P2Y11, which is also reported to be abundantly expressed on ECs [3], does not induce any of the above-described effects. P2Y2 carries unique features as it has equal binding affinities to its natural ligands ATP and UTP, however, β-arrestin recruitment and ERK 1/2 activation kinetics appear to be ligand-specific [19]. The question whether differential activation of P2Y2 may be relevant in vascular growth remains and will be an interesting subject to be investigated in the future. Taken together, we conclude that P2Y2-dependent vascular network formation occurs as a result of autocrine signaling. In addition, our data indicate that P2Y2 overexpression in endothelial cells leads to loss of responsiveness to VEGF. Similar observations have been previously reported for endothelial colony-forming cells after stimulation with a non-selective P2 receptor agonist, which resulted in network formation on Matrigel at sub-optimal VEGF concentrations [6]. Nevertheless, the fact that inhibition or knockdown of P2Y2 can significantly influence endothelial sprouting indicates that P2Y2 does indeed play an essential role in angiogenesis. The blood vascular system is highly heterogeneous and fulfills different tissue-dependent tasks [51]. For instance, the majority of endothelial cells in a human body are found in microvascular structures, suggesting that human microvascular endothelial cells represent the most appealing source of ECs when studying wound healing and regeneration [20]. However, all of these endothelial cells from adult tissue represent a specialized cell type that may only reflect functions from the respective tissue [51]. Pluripotent cell-derived endothelial cells may offer a solution as they represent an unspecialized pan-endothelial cell type which can be used to study crucial functions in endothelial cell biology [52]. Here, we used iPSC-derived endothelial colony-forming cells which we believe represent an optimal source of unspecialized endothelial cells to confirm the role of P2Y2 in angiogenesis. Indeed, we were able to reproduce our results from experiments with HUVEC in these cells, thereby proving that our described function of P2Y2 may be similar throughout the entire endothelium.

Physiologically, P2Y2 receptors expressed by the vascular endothelium have a vital role in vasorelaxation which was confirmed by P2Y2 knockout mice experiments [1]. The trigger for vasodilation has been shown to be shear stress-induced ATP release from endothelial cells, resulting in autocrine activation of P2Y2 leading to release of nitric oxide [17]. To our knowledge, we are the first to provide findings which suggest that P2Y2 receptors fulfill a key function in vascular endothelial cells as crucial regulators of endothelial cell signaling during sprouting and vascular network formation in vitro. It is known that P2Y2 can transactivate VEGFR-2 via Src in a VEGF-independent manner [13], which may represent a mechanistic explanation for the effects observed in this study. Alternatively, P2Y2 has also been shown to co-localize with other endothelial cell surface molecules such as αvβ3 and αvβ5 integrins via its RGD motif to activate integrin signaling pathways or with VE-Cadherin which could lead to co-regulation of VEGFR-2 activity [9, 17]. Importantly, the data presented here demonstrate that P2Y2 is a previously unrecognized component in the complex signaling network that regulates VEGF-induced angiogenesis. The fact that overexpression of P2Y2 forces endothelial cells to acquire a tip cell phenotype implies that this receptor may be relevant as a therapeutic target in human vascular malignancies and pathologies.