VEGF-A isoform-specific regulation of calcium ion flux, transcriptional activation and endothelial cell migration

Vascular endothelial growth factor A (VEGF-A) regulates many aspects of vascular physiology such as cell migration, proliferation, tubulogenesis and cell-cell interactions. Numerous isoforms of VEGF-A exist but their physiological significance is unclear. Here we evaluated two different VEGF-A isoforms and discovered differential regulation of cytosolic calcium ion flux, transcription factor localisation and endothelial cell response. Analysis of VEGF-A isoform-specific stimulation of VEGFR2-dependent signal transduction revealed differential capabilities for isoform activation of multiple signal transduction pathways. VEGF-A165 treatment promoted increased phospholipase Cγ1 phosphorylation, which was proportional to the subsequent rise in cytosolic calcium ions, in comparison to cells treated with VEGF-A121. A major consequence of this VEGF-A isoform-specific calcium ion flux in endothelial cells is differential dephosphorylation and subsequent nuclear translocation of the transcription factor NFATc2. Using reverse genetics, we discovered that NFATc2 is functionally required for VEGF-A-stimulated endothelial cell migration but not tubulogenesis. This work presents a new mechanism for understanding how VEGF-A isoforms program complex cellular outputs by converting signal transduction pathways into transcription factor redistribution to the nucleus, as well as defining a novel role for NFATc2 in regulating the endothelial cell response.


INTRODUCTION
There are 58 human receptor tyrosine kinases (RTKs) which are sub-classified into 20 families. These Type I membrane proteins regulate animal development, health and disease states (Lemmon and Schlessinger, 2010). Upon ligand-binding, RTK monomers undergo dimerisation followed by trans-autophosphorylation of cytoplasmic tyrosine residues, enabling the recruitment and phosphorylation of a vast array of signal transduction enzymes and adaptor proteins. RTKs are key targets for new therapeutics but successful drug design is complicated by the increasing number of ligands discovered to bind to each RTK. The physiological relevance for the expression of numerous ligands is unclear; many studies have based their conclusions on studying the effects of a single ligand for a specific RTK. This complexity is exemplified by the human vascular endothelial growth factor (VEGF) family, which comprises 5 family members [VEGF-A, VEGF-B, VEGF-C, VEGF-D and placental growth factor (PlGF)]. Together this family of growth factors regulate angiogenesis and lymphangiogenesis through differentially binding to an array of Class V RTKs (VEGFR1-3) and co-receptors such as neuropilins i.e. NRP1 and NRP2 (Koch et al., 2011). The VEGFA gene is located on chromosome 6p21.3 (Vincenti et al., 1996); transcription of this gene leads to the formation of a pre-mRNA transcript with a coding region that contains 8 exons and 7 introns. Alternative splicing of the VEGFA mRNA transcript gives rise to at least 7 pro-angiogenic isoforms, which all bind to both VEGFR1 and VEGFR2 (Robinson and Stringer, 2001). However, it is also believed that, the pre-mRNA splicing machinery can also generate anti-angiogenic isoforms via alternate splice site selection events (Harper and Bates, 2008). These events termed proximal splice site selection (PSS) and distal splice site selection (DSS), determine the terminal amino acid sequence (exon 8) switching between the pro-angiogenic sequence CDKPRR (exon 8a) or the anti-angiogenic sequence SLTRKD (exon 8b) (Harper and Bates, 2008). This raises the question as to the functional relevance of the different VEGF-A isoforms; most studies have focused solely on the VEGF-A 165 isoform, which is secreted by both vascular and non-vascular cells.
VEGF-A is a crucial regulator of angiogenesis, modulating diverse endothelial responses such as cell proliferation, migration, tubulogenesis, vascular permeability and leukocyte recruitment. VEGFA gene dosage is critical for normal development as heterozygous VEGFA (+/2) knockout mice embryos are not viable and die between E11 and E12 due to a deformed vascular network (Carmeliet et al., 1996;Ferrara et al., 1996). VEGFR1 and VEGFR2 can both bind different VEGF-A isoforms but it is unclear as to how the different RTK-ligand complexes regulate endothelial and vascular function. Nonetheless, both VEGFR1 and VEGFR2 encode gene products that are essential for correct vascular development and animal function (Fong et al., 1995;Shalaby et al., 1995).
VEGF-A binding to VEGFR2 triggers receptor dimerisation, linked to the activation of its tyrosine kinase domain, which triggers sustained downstream signal transduction integrated with receptor ubiquitination, trafficking and proteolysis Horowitz and Seerapu, 2012;Koch and Claesson-Welsh, 2012;Nakayama and Berger, 2013). A key aspect of VEGF-Astimulated endothelial cell signal transduction is the elevated transcription of 100-200 target genes, which regulate a variety of cellular responses (Rivera et al., 2011;Schweighofer et al., 2009). Various studies have shown that VEGF-A isoforms differentially promote VEGFR2-dependent signal transduction and cellular outcomes (Kawamura et al., 2008a;Kawamura et al., 2008b;Zhang et al., 2000). However, the mechanism(s) which link VEGF-A isoform-specific signal transduction to nuclear gene transcription and endothelial responses are ill-defined.
To address the individual role of each VEGF-A splice isoform in regulating vascular function, we evaluated VEGF-A 121 and VEGF-A 165 for their ability to regulate signal transduction events linked to physiological responses. Here, we show that these two VEGF-A isoforms produce different intracellular signalling outcomes which impact on a transcriptional 'switch' allowing for isoform-specific regulation of endothelial cell migration. Thus, VEGF-A isoforms could act as temporal and spatial cues that program endothelial responses essential for building unique vascular networks.

RESULTS
VEGF-A isoforms cause differential VEGFR2 activation and signal transduction VEGF-A-stimulation promotes VEGFR2 dimerisation and transautophosphorylation of several key tyrosine residues within the cytoplasmic domain (Koch and Claesson-Welsh, 2012) which stimulates downstream signal transduction pathways (Fig. 1A). Recruitment of factors and enzymes that bind activated VEGFR2 stimulates intracellular signalling events which modulate an array of endothelial cell responses in order to promote angiogenesis and regulate vascular development (Fig. 1A). Various studies have shown that VEGF-A isoforms promote differential VEGFR2 activation and downstream signal transduction (Kawamura et al., 2008b;Pan et al., 2007a). Although, VEGF-A-stimulated VEGFR2-dependent signalling is well understood, it is still unclear how VEGF-A isoform-specific signal transduction is converted into nuclear gene transcription to differentially regulate endothelial cell responses. In order to further investigate this phenomenon, we first compared the ability of two VEGF-A isoforms (VEGF-A 165 and VEGF-A 121 ) to regulate signal transduction events via the VEGFR2/VEGF-A signalling axis. Primary human umbilical vein endothelial cells (HUVECs) were titrated with 0.025, 0.25 and 1.25 nM of either VEGF-A 165 or VEGF-A 121 for 5 or 15 min prior to processing and immunoblot analysis of VEGFR2 activation and downstream signalling pathways (Fig. 1B). Quantification of the relative changes in phosphorylation status of VEGFR2-pY1175 in response to a dosedependent titration of VEGF-A 165 (Fig. 1C) or VEGF-A 121 (Fig. 1D) revealed that peak activation occurred within 5 min of ligand treatment. However, VEGF-A 121 -stimulated VEGFR2-Y1175 phosphorylation (Fig. 1D) was significantly reduced versus VEGF-A 165 treatment (Fig. 1C) at 0.025 and 0.25 nM. However, upon stimulation with saturating levels of VEGF-A (1.25 nM), the peak level of VEGFR2 activation achieved in response to VEGF-A 165 (Fig. 1C) was comparable to that induced by VEGF-A 121 (Fig. 1D). Interestingly, at this VEGF-A concentration, activated VEGFR2 appeared to be dephosphorylated at an increased rate upon treatment with VEGF-A 121 (Fig. 1D) compared to cells treated with VEGF-A 165 (Fig. 1C).
Angiogenesis is a tightly regulated process and many endothelial cell responses constantly need fine-tuning for homeostasis and maintenance of vascular physiology (Carmeliet, 2005). VEGF-A stimulation regulates several endothelial cell responses through the activation of downstream signalling events including those that affect endothelial cell permeability (Akt-eNOS pathway), cell migration (p38 MAPK pathway) and cell proliferation (ERK1/2 pathway). Stimulation with either VEGF-A 165 or VEGF-A 121 promoted phosphorylation of endothelial nitric oxide synthase (eNOS), p38 MAPK and ERK1/2 (Fig. 1B) enzymes. Quantification of eNOS-pS1177 levels revealed that stimulation with VEGF-A 121 (Fig. 1F) resulted in a generally lower level of activation compared to VEGF-A 165 stimulation (Fig. 1E). However, peak levels of eNOS activation were comparable between the two isoforms ( Fig. 1E,F). Additionally both VEGF-A isoforms caused a similar increase in p38 MAPK-pT180/pY182 levels at 1.25 nM ligand stimulation, with peak levels detected after 15 min (Fig. 1G,H). In contrast VEGF-A 165 promoted ,2fold increase in ERK1/2-T202/Y204 phosphorylation compared to VEGF-A 121 at the same concentration (1.25 nM); peak levels were still detected 15 min post stimulation (Fig. 1I,J). These findings suggest that VEGF-A isoforms have the capability to differentially stimulate multiple signal transduction pathways in endothelial cells.
We then carried out an analysis of cytosolic calcium ion levels in VEGF-A isoform-stimulated endothelial cells using a cellpermeable Ca 2+ -sensitive fluorescent probe (Fura-2 AM). Upon titration of either VEGF-A 165 (Fig. 3A) or VEGF-A121 (Fig. 3B) we observed different patterns of cytosolic calcium ion flux. The lowest concentration (0.025 nM) of VEGF-A 165 ligand caused a slow and sustained rise in cytosolic calcium ions (Fig. 3A); this was not observed upon treatment with 0.025 nM VEGF-A 121 . A similar trend was observed at 0.25 nM ligand concentration of both isoforms (Fig. 3A,B). Interestingly, the highest concentration (1.25 nM) of ligand elicited similar rises in cytosolic calcium ions with both VEGF-A 165 (Fig. 3A) and VEGF-A121 (Fig. 3B). Quantification of peak magnitude in ligand-stimulated rise in cytosolic calcium ion levels revealed that both VEGF-A 165 and VEGF-A 121 acted in a concentration-dependent manner (Fig. 3C). VEGF-A 165 stimulation generally promoted an increased level in intracellular Ca 2+ versus treatment with VEGF-A 121 (Fig. 3C). However, at saturating levels of VEGF-A (1.25 nM) the peak magnitude of cytosolic calcium ion rise was comparable upon stimulation with either VEGF-A 165 or VEGF-A 121 (Fig. 3C). In contrast, the time taken to reach this peak magnitude was ,2-fold longer in VEGF-A 121 -stimulated cells in comparison to VEGF-A 165 treatment  These findings suggest that the cytosolic calcium ion flux caused by the VEGFR2-PLCc1 axis is VEGF-A isoform-dependent ( Fig. 3F).
Cytosolic calcium ion fluxes have also been linked to promoting trafficking of another endothelial receptor tyrosine kinase (VEGFR1) from the Golgi apparatus to the plasma membrane Mittar et al., 2009). Due to the different abilities of the two VEGF-A isoforms in promoting the release of intracellular calcium ion stores (Fig. 3A-E), we hypothesised about effects on VEGFR1 translocation to the plasma membrane. To test this idea, we stimulated endothelial cells with either VEGF-A 165 or VEGF-A 121 for 0, 15, 30 or 60 min prior to assessing plasma membrane VEGFR1 using cell surface biotinylation (supplementary material Fig. S1A). Immunoblot analysis of mature and soluble VEGFR1 expression revealed that in non-stimulated endothelial cells, both forms of VEGFR1 are predominately located within internal, biotin probe-inaccessible compartments and not at the plasma membrane (supplementary material Fig. S1A). However, upon stimulation with VEGF-A 165 there was a significant increase (,2.5-fold) in both mature (supplementary material Fig. S1B) and soluble (supplementary material Fig. S1C) VEGFR1 (sVEGFR1/sFlt1) at the cell surface. Contrastingly, VEGF-A 121 treatment failed to promote a significant increase in cell surface levels of either mature or sVEGFR1 (supplementary material Fig. S1A,B).
VEGF-A-stimulated endothelial cell responses, we used reverse genetics combined with ligand-stimulated cellular assays (Fig. 6). We used siRNA duplexes to knockdown NFATc2 levels in endothelial cells and compared to scrambled siRNA duplextreated or non-transfected controls (Fig. 6). Treatment with NFATc2-specific siRNA duplexes caused ,90% knockdown in endothelial NFATc2 levels (supplementary material Fig. S2A,B), but did not affect expression of other endothelial proteins such as VEGFR2, VEGFR1 and ERK1/2 (supplementary material Fig.  S2A). To determine the effect of NFATc2 knockdown on endothelial cell migration and tubulogenesis, NFATc2-depleted or control endothelial cells were stimulated with 0.25 nM VEGF-A 165 or VEGF-A 121 , before fixation, staining and processing for light microscopy (Fig. 6A). As previously reported elsewhere, VEGF-A 165 has a ,2-3-fold higher efficacy for promoting endothelial cell migration (Fig. 6B,C). Interestingly, quantification revealed that depletion of NFATc2 caused ,2-3fold increase in basal cell migration in non-stimulated endothelial cells (Fig. 6B), during comparison of all values to non-transfected non-stimulated endothelial cells. In such an analysis, there appears to be relatively little change in VEGF-A isoformstimulated cell migration upon scrambled siRNA treatment or NFATc2 knockdown (Fig. 6B). However, when each experiment was compared to their own non-stimulated control which had been also subjected to scrambled or NFATc2 siRNA, a different pattern emerged (Fig. 6C). There was now a substantial 3-4-fold decrease in both VEGF-A 165 and VEGF-A 121 -stimulated endothelial cell migration in NFATc2 depleted endothelial cells (Fig. 6C).
An important aspect of the VEGF-A-stimulated endothelial cell response is the capacity to build hollow tubes (tubulogenesis) which can be monitored using an organotypic assay. Using this technique, we asked whether NFATc2 was required for VEGF-Astimulated tubulogenesis (Fig. 6D). As previously reported elsewhere, VEGF-A 165 has a ,2-3-fold higher efficacy for promoting endothelial cell tubulogenesis, in comparison to VEGF-A 121 (Fig. 6E,F). Intriguingly, NFATc2 did not significantly affect VEGF-A stimulated endothelial cell tubulogenesis (Fig. 6D). Quantification revealed that NFATc2 knockdown did not affect endothelial tubule length (Fig. 6E) nor branch point complexity (Fig. 6F).

DISCUSSION
In this study, we show that different VEGF-A isoforms have differing capabilities to modulate endothelial cell migration by regulating the activation and nuclear localisation of a key transcription factor (Fig. 7). In our proposed model, two VEGF-A isoforms with similar binding affinities differentially program VEGFR2 activation and downstream signal transduction which controls a transcriptional 'switch' that regulates endothelial cell migration. This 'switch' comprises the calciumdependent activation of the calmodulin-calcineurin pathway which targets endothelial NFATc2, a key transcription factor (Fig. 7).
A key feature of endothelial cell migration is the VEGF-A isoform-specific programming of signal transduction and this cellular response. VEGF-A 165 significantly promoted increased VEGFR2 phosphorylation at residue Y1175 in comparison to VEGF-A 121 . The generation of this unique pY1175 binding site enables recruitment of PLCc1 via its SH2 domains and subsequent phosphorylation on residue Y783. VEGF-A isoformspecific stimulation of PLCc1-pY783 levels correlated with increased cytosolic calcium ion flux (Fig. 7). Calmodulin is a key target of elevated cytosolic calcium ion levels and upon calcium ion binding to one of its four EF hands, undergoes a conformational change that promotes interaction with new cellular targets, including the protein phosphatase calcineurin (Berchtold and Villalobo, 2014). Calmodulin binding to calcineurin promotes enzymatic activation; one such target is NFATc2, resulting in its subsequent rapid dephosphorylation and nuclear targeting (Luo et al., 1996;Okamura et al., 2000). Both VEGF-A 165 and VEGF-A 121 stimulation promotes cytosolic calcium ion flux, NFATc2 dephosphorylation and nuclear translocation but to widely differing extents (Fig. 7). Depletion of endothelial NFATc2 levels significantly reduced VEGF-Astimulated endothelial cell migration suggesting that this factor plays a role in controlling gene expression linked to cell migration processes e.g. focal adhesion, stress fibre formation and actin polymerisation (Fig. 7). Although, our results show that VEGF-A isoforms promote NFATc2 translocation into the nucleus, a lacking area of investigation is to determine if VEGF-A-stimulated NFATc2 dephosphorylation and nuclear translocation leads to the transcription of well-known NFATc2 regulated genes. However, it is highly likely that this is the case, due to the effect that depletion of NFATc2 has on VEGF-Astimulated endothelial cell migration.
the regulation of VEGF-A isoform-specific endothelial cell migration (Fig. 7). However, in contrast to work elsewhere (Bala et al., 2012), we did not find that NFATc2 was essential for VEGF-A-stimulated endothelial cell tubulogenesis. One possible explanation lies within the choice of assay used to monitor endothelial cell tubulogenesis. In our study we used a 7 day human fibroblast-endothelial cell co-culture assay to monitor the formation of PECAM-1 positive endothelial tubules, where Bala and colleagues performed a Matrigel co-culture assay (Bala et al., 2012); one drawback of this assay is that does not give an accurate index of endothelial cell tube formation (Donovan et al., 2001). Another reason could be the method used to look at NFATc2 involvement in endothelial cell function. In our study we used RNAi to directly deplete 90% of endothelial NFATc2. Bala and colleagues were studying the effects of a Kaposi's Sarcoma Herpesvirus protein (K15) on stimulating NFATc2dependent RCAN1 expression (Bala et al., 2012). Therefore, it is possible that VEGF-A and K15-stimulated NFATc2 activation play different roles in determining endothelial cell fate. The fact that NFATc2 does not appear to be required for endothelial tube formation but is required for endothelial cell motility is surprising, as the latter is required for the former. This is a serious limitation to our current study and further work is required, the use of more robust 3-D angiogenesis assays such as the fibrin bead and 'ex vivo' mouse aortic ring assay would help strengthen the conclusions made here. Additionally, the monitoring of endothelial cell motility in real-time would further strengthen the role of NFATc2 in regulating VEGF-A isoform-specific cell migration. Likewise, analysis of other VEGF-A-stimulated endothelial responses (e.g. cell proliferation and permeability) for NFATc2 dependency would prove informative.
Interestingly, we found that basal NFATc2 activity inhibits the rate of cell migration in non-stimulated endothelial cells. One possible explanation could lie within the fact that NFATc2 naturally functions in an auto-inhibitory manner by promoting the expression of regulator of calcineurin 1 (RCAN1). RCAN1 forms a complex with calcineurin to directly inhibit its phosphatase activity, thus preventing calcineurin-dependent NFATc2 dephosphorylation and nuclear translocation (Holmes et al., 2010;Li et al., 2011). However, calcineurin also activates receptor for activated protein kinase C1 (RACK1), which has been implicated in regulating cell migration through binding to an array of signalling proteins. Endothelial cells constitutively express RCAN1 at a basal level (Holmes et al., 2010;Méndez-Barbero et al., 2013). Thus, depleting NFATc2 could reduce basal levels of RCAN1, subsequently leading to increased endothelial cell migration through attenuated calcineurin-stimulated RACK1 signalling. Calcineurin inhibitors have also been shown to inhibit cell migration (Espinosa et al., 2009;Siamakpour-Reihani et al., 2011), thus strengthening this proposed mechanism. RCAN1 is one of the most highly expressed gene products in response to VEGF-A-stimulation (Rivera et al., 2011;Schweighofer et al., 2009). Therefore, it is likely to have an essential role in regulating VEGF-A mediated endothelial responses. Various studies have shown that RCAN1 is essential for cellular migration (Espinosa et al., 2009;Holmes et al., 2010;Iizuka et al., 2004;Ryeom et al., 2008). Hence, VEGF-A isoform-specific elevation in RCAN1 expression could account for the reduction in VEGF-A-stimulated endothelial cell migration in NFATc2-depleted endothelial cells. Additionally, VEGF-A isoform-specific NFATc2dependent upregulation of RCAN1 could also account for the differences in their abilities to stimulate endothelial cell migration.
One question left unanswered is what is the mechanism(s) behind how these two VEGF-A isoforms with similar binding affinities, promote such diverse activation of VEGFR2 and downstream signalling enzymes. One possible answer is through the differential binding and recruitment of co-receptors (e.g. NRP-1, NRP-2 and HSPG). Neuropilin 1 (NRP-1) is a VEGF-A isoform-specific co-receptor . VEGF-A 165 binds to the co-receptor neuropilin 1 (NRP1) to form a VEGFR2/ NRP1/VEGF-A 165 signalling complex (Fantin et al., 2013;Herzog et al., 2011;Raimondi and Ruhrberg, 2013;Zachary et al., 2009). Formation of this trimeric complex has been shown to increase VEGF-A 165 -stimulated VEGFR2 activation, downstream signalling and endothelial responses (Allain et al., 2012;Kawamura et al., 2008a;Koch, 2012). However, VEGF-A 121 binding simultaneously to VEGFR2 and NRP1 has been contradicted (Pan et al., 2007b). Thus VEGF-A isoform-specific recruitment of NRP-1 could account for the differences in signalling between the 2 isoforms.
VEGF-A-stimulated NFATc2-dependent gene expression is said to occur in co-operation with the transcription factor early growth response 1 (EGR-1) (Schweighofer et al., 2007). VEGF-A stimulates EGR-1 gene expression through the MEK1-ERK1/2 pathway (Schweighofer et al., 2007). As we show that VEGF-A 165 and VEGF-A 121 have differential effects on ERK1/2 activation, one future study would be to investigate if these 2 isoforms differentially activated EGR-1 as a means of programing isoform-specific endothelial cell responses.
VEGFR1 binds VEGF-A with a much higher affinity than VEGFR2 (de Vries et al., 1992), yet its involvement in VEGF-A stimulated angiogenesis is not well understood. One view is that both membrane-bound and sVEGFR1 proteins act primarily as VEGF-A 'traps', thus limiting ligand availability for the major pro-angiogenic receptor, VEGFR2 (Rahimi, 2006). However, VEGF-A-regulated VEGFR1-linked signal transduction has been implicated in certain aspects of endothelial cell physiology (Koch and Claesson-Welsh, 2012). In nonstimulated endothelial cells, VEGFR1 is primarily inaccessible to VEGF-A, as it is located within an internal compartment resembling the Golgi apparatus (Mittar et al., 2009). VEGF-A 165 stimulation promotes VEGFR1 translocation to the cell surface via a cytosolic calcium ion-dependent mechanism where it can bind exogenous VEGF-A. In this study, VEGF-A 165 promoted significant trafficking of VEGFR1 to the plasma membrane whereas VEGF-A 121 was largely ineffective in this context; again this could be explained by signal transduction effects on cytosolic calcium ion levels. The role of VEGF-A isoforms in differential VEGFR1 trafficking could further modulate the endothelial cell response to this ligand, as not only would this affect VEGF-A bioavailability for VEGFR2, but regulate VEGFR1-specific signal transduction in response to VEGF-A, VEGF-B and PlGF isoforms and thus regulate vascular physiology.
The physiological process of angiogenesis can be destabilised in a wide variety of major disease states ranging from atherosclerosis, rheumatoid arthritis, pathogenic infection to cancer. Current anti-angiogenic therapies which try to restrict pathological angiogenesis, by sequestering endogenous VEGF-A or via inhibiting RTK activity, are not as successful as first hoped. This is partially due to tumour cell-acquired resistance through the activation of alternative angiogenic pathways, in addition to other assorted mechanisms (Vasudev and Reynolds, 2014). Therefore, further investigation into the mechanisms which regulate angiogenesis is required. This will hopefully lead to improved patient outcomes in various disease states, through the design of better therapeutics. Our study now provides a novel mechanism to explain how different VEGF-A isoforms act on a common receptor tyrosine kinase (VEGFR2) to differentially determine a specific endothelial cell outcome (i.e. cell migration; Fig. 7). Additionally, this work shines new light on the physiological importance of NFATc2 in regulating VEGF-Astimulated endothelial cell migration. A substantial future challenge will be to determine the biological significance of each VEGF-A isoform in healthy and diseased states.

Cell culture
Human umbilical vein endothelial cells (HUVECs) were isolated as previously described (Howell et al., 2004). Human umbilical cords were obtained after written informed consent from volunteers. Tissue retrieval was governed by local ethics approval from the Leeds Hospitals NHS Trust. HUVECs were only used between passage P1 and P5. Cells were seeded into 6-well plates and cultured (for at least 24 h) in ECGM until ,80% confluent, washed three times with PBS and then starved in MCDB131 + 0.2% (w/v) BSA for 2-3 h. HUVECs were then stimulated with 0, 0.025, 0.25 or 1.25 nM of VEGF-A 165 or VEGF-A 121 prior to processing.
Cell lysis and processing for immunoblotting Cells were washed with ice-cold PBS, prior to lysis in 2% (w/v) SDS in TBS containing 1 mM PMSF and protease inhibitor cocktail (Sigma-Aldrich) and stored at 220˚C. Protein concentration was then determined using the bicinchoninic acid (BCA) assay (ThermoFisher). Protein lysate was diluted in an equal amount of 26 SDS-sample buffer and subjected to SDS-PAGE prior to analysis via immunoblotting.
Cytosolic calcium ion flux assay 2.5610 4 HUVECs were seeded per well of a 96-well plate and cultured for 48 h until 100% confluent. Cells were washed twice with 100 ml SBS buffer (130 mM NaCl, 5 mM KCl, 1.2 mM MgCl 2 , 8 mM glucose, 10 mM HEPES, 1.5 mM CaCl 2 , pH 7.4) and loaded with 50 ml Fura-2 AM/SBS (2 mM Fura-2-AM, 0.01% Pluronic F-127; Life Technologies, Paisley, UK) for 60 min at 37˚C. Cells were then washed twice with 100 ml SBS and left at room temperature for 30 min to allow complete deesterification of Fura-2-AM. Desired concentration of VEGF-A isoform was made up as a 56stock in a compound plate. Cytosolic calcium ion rise was monitored by measuring the ratio of 510 nm emission achieved from excitation at 340 nm vs. 380 nm using a FlexStation Benchtop Microplate Reader (Molecular Devices, Sunnyvale, USA). VEGF-A was added automatically after 32 s and Ca 2+ levels were measured every 5 s for a total of 900 s. Change in cytosolic Ca 2+ fluctuations as a function of time were plotted and quantified by calculating the peak magnitude, time take to reach peak magnitude and the area under the curve using OriginPro 8.6 (OriginLab, USA).

Immunofluorescence analysis of NFATc2 localisation
HUVECs cultured in 96-well plates were serum starved for 3 h before being stimulated with 0.25 nM VEGF-A 165 or VEGF-A 121 . Media was aspirated and cells incubated with fixative (Sigma-Aldrich) for 5 min at 37˚C before permeabilisation in 0.2% (v/v) Triton X-100 in PBS. Cells were blocked in 5% (w/v) BSA prior to incubation with rabbit-anti-NFATc2 in 1% (w/v) BSA overnight at 4˚C. Primary antibody was aspirated and cells washed, before incubation with 4 mg/ml donkey antirabbit Alexaflour488-conjugated secondary antibody (Invitrogen, Amsterdam, Netherlands), 2 mg/ml 4,6-diamidino-2-phenylidole (DAPI) in 1% (w/v) BSA in PBS and incubated for 2-3 h at room temperature. Images were acquired using an EVOS-fl inverted digital microscope (Life Technologies). 3 random fields were captured per sample. Relative nuclear co-localisation was quantified using Image J as previously described elsewhere (Bruns et al., 2010;Jopling et al., 2011).

Lipid-based transfection of siRNA duplexes
HUVECs were reversed transfected using siRNA duplexes and Lipofectamine RNAiMAX (Life Technologies). Per well of a 6-well plate, 15 ml of 2 mM siRNA duplex was added to 481 ml of serum/ antibiotic-free OptiMEM (Life Technologies) and allowed to react at room temperature for 5 min. 4 ml of Lipofectamine was then added to the siRNA duplex/OptiMEM mixture, inverted briefly and incubated at room temperature for 20 min. 2.5610 5 HUVECs in 1 ml of OptiMEM were seeded, followed by immediate dropwise addition of the siRNA/ lipofectamine mixture. Cells were incubated at room temperature for 30 min before being returned to the incubator. After 6 h total incubation, transfection media was aspirated and replaced with ECGM. Cells were allowed to recover for 72 h prior to analysis of endothelial responses.
Cell migration assay 3610 4 HUVECs were seeded in MCDB131 media containing 0.2% (w/v) BSA onto a 8 mm pore size Transwell filter [pre-coated with 0.1% (v/v) pig skin gelatin (PSG)] inserted into a 24-well plate (BD Biosciences, Oxford, UK). MCDB131+0.2% (w/v) BSA containing60.25 nM VEGF-A isoform was added to the lower chambers to stimulate cell migration. Cells were allowed to migrate for 24 h prior to fixation and staining with 0.2% (w/v) crystal violet in 20% (v/v) methanol. Prior to fixation endothelial cells were visualised by phase contrast microscopy to ensure the number of adherent cells were comparable between Transwells. This ensures that differences seen in the number of migrating cells, was indeed caused by a defect in cell motility and not due to defective cell adhesion. Non-migrated cells were removed from the upper chamber with a moist cotton bud. Transwell filters were imaged and the number of migratory cells counted as previously described (Fearnley et al., 2014b).

Tubulogenesis assay
Primary human foreskin fibroblasts (Promocell) were cultured in 48-well plates until confluent in complete DMEM containing 10% FCS, 1% nonessential amino acids and 1% sodium pyruvate (Life Technologies). 6500 HUVECs were then seeded onto the fibroblast monolayer in a 1:1 mixture of complete DMEM containing 10% (v/v) FCS, non-essential amino acids, sodium pyruvate and ECGM and left to adhere for 24 h. Culture media was then removed and replaced with fresh ECGM 6VEGF-A as desired; media was replaced every 2-3 days for 7 days. Co-cultures were then fixed and blocked prior to overnight incubation with 1 mg/ml mouse anti-human PECAM-1 (CD31; Santa Cruz Biotechnology) at room temperature. Co-cultures were washed three times with PBS before incubation with 1 mg/ml donkey anti-mouse Alexaflour594-conjugate (Invitrogen) for 2-3 h at room temperature. Endothelial tubules were then visualised by immunofluorescence microscopy using an EVOS-fl inverted digital microscope (Life Technologies). Three random fields were imaged per well. Both the number of branch points and the total tubule length was then quantified from each photographic field using the open source software AngioQuant (www.cs.tut.fi/sgn/csb/angioquant) and values averaged. A more detailed methodology is available elsewhere (Fearnley et al., 2014b).

Statistical analysis
Statistical analysis was performed using one-way analysis of variance (ANOVA) followed by Tukey's post-hoc test or two-way ANOVA followed by Bonferroni multiple comparison test using GraphPad Prism software (La Jolla, CA, USA). Significant differences between control and test groups are denoted with p values less than 0.05 (*), 0.01 (**), 0.001 (***) and 0.0001 (****) indicated. Graphical error bars denote 6SEM (standard error of mean).