Targeting p35/Cdk5 Signalling via CIP-Peptide Promotes Angiogenesis in Hypoxia

Cyclin-dependent kinase-5 (Cdk5) is over-expressed in both neurons and microvessels in hypoxic regions of stroke tissue and has a significant pathological role following hyper-phosphorylation leading to calpain-induced cell death. Here, we have identified a critical role of Cdk5 in cytoskeleton/focal dynamics, wherein its activator, p35, redistributes along actin microfilaments of spreading cells co-localising with p(Tyr15)Cdk5, talin/integrin beta-1 at the lamellipodia in polarising cells. Cdk5 inhibition (roscovitine) resulted in actin-cytoskeleton disorganisation, prevention of protein co-localization and inhibition of movement. Cells expressing Cdk5 (D144N) kinase mutant, were unable to spread, migrate and form tube-like structures or sprouts, while Cdk5 wild-type over-expression showed enhanced motility and angiogenesis in vitro, which was maintained during hypoxia. Gene microarray studies demonstrated myocyte enhancer factor (MEF2C) as a substrate for Cdk5-mediated angiogenesis in vitro. MEF2C showed nuclear co-immunoprecipitation with Cdk5 and almost complete inhibition of differentiation and sprout formation following siRNA knock-down. In hypoxia, insertion of Cdk5/p25-inhibitory peptide (CIP) vector preserved and enhanced in vitro angiogenesis. These results demonstrate the existence of critical and complementary signalling pathways through Cdk5 and p35, and through which coordination is a required factor for successful angiogenesis in sustained hypoxic condition.


Introduction
The importance of angiogenesis in relation to neuronal replenishment and survival after stroke has been clearly demonstrated. In this respect, revascularization and associated reperfusion are vital determinants of tissue survival and patient recovery after stroke and therefore a major potential target for successful therapies [1]. Angiogenesis "per se" is a tightly regulated multi-step process, by which new blood vessels are formed from endothelial cell (EC) sprouts, emanating from preexisting vessels and is mediated by the combination of a complex range of angiogenic and anti-angiogenic factors, driving the recruitment, migration, proliferation and differentiation of ECs [2]. Effective angiogenesis is strictly dependent on the proper spatial and temporal control of focal adhesions and activation of cytoskeletal organizers that regulate cell mechanical forces and the cell plasticity during migration, sprouting and cell differentiation within developing neo-capillary structures [3]. Beyond the typical canonical signalling molecules already characterized, a set of cyclindependent kinases (Cdks), has been shown to contribute significantly to the angiogenic response [4]. Among them, the Cdk5 is emerging as a potential important regulator of vasculogenesis [5].
Cdk5 is a proline-directed serine/threonine kinase highly expressed in the central nervous system. Initially due to the prevalence of its activity and activators (p35, p25 and p39) in neuronal cells, Cdk5 was considered as a neuron-specific kinase, involved in neurite outgrowth, neural migration, synaptic activity, neural cytoskeletal organization and neural survival [6]. Over the past decade, Cdk5 has been isolated in various non-neuronal tissues and cells [7], including ECs, and has been demonstrated to be essential for important cellular activities, including angiogenesis of Human Umbilical Vein ECs [5], neuronal cell adhesion and migration [7]; whereas its aberrant activation has been associated directly with neuronal cell death [8].
Increased expression of Cdk5 and of both p35 and p25 activators has been reported in human stroke [9,10] and after focal middle cerebral artery occlusion [8]. After brain ischemia, the rise in neuronal death is crucially associated by the Cdk5 hyper-activation mediated by p25 [8,9]. The excitotoxic signal transduction cascade induced by hypoxia leads to excessive activation of calpains, a group of calcium-activated cytosolic proteases, that cleave p35 to more stable p25 peptide, with the following Cdk5 hyperactivation and neuronal apoptosis; whereas modulation of p35 amount and its binding with Cdk5 confers cytoprotection [11,12]. Additionally, in our previous work, we have shown over-expression of Cdk5 together with p35/p25 in apoptotic brain ECs in hypoxic regions of stroked tissue [10]. We have shown the up-regulation and/or nuclear translocation of Cdk5 and p35 in neurons and ECs subjected to oxygen-glucose deficiency and their association with cellular damage as response to hypoxic conditions therefore, it is hypothesized that the Cdk5/p35 pathway may be a key pathophysiological determinant of ECs survival and ultimately outcome after infarction [13,10].
Currently, research focused on novel therapies directed to enhance angiogenesis, neurogenesis and improve neurologic function has highlighted the pharmacological inhibition of Cdk5/p25 signalling as a novel potential intervention point for cell protection and improved tissue remodelling after stroke [13,12]. However, since the use of some inhibitors would appear to affect also the physiological function of Cdk5 (i.e. Roscovitine [14]) and/or the regulation of synaptic plasticity (i.e. calpain inhibitors [15]), their introduction could lead to serious secondary side effects and thereby compromise any therapeutic efficacy [13]. Thus, identification and characterization of novel selective inhibitors of Cdk5/p25 signaling might produce a more viable potential therapeutic target [12] with particular emphasis to re-vascularization of the brain tissue [13].
Recently, the neuronal cyto-protective potential of a natural small peptide (CIP-peptide) was demonstrated after neurotoxic stress. CIP is a derived-p35 cleavage peptide, which selectively targets Cdk5/p25 activity without affecting Cdk5/p35 signalling [16], suggesting that CIP may represent a novel selective vascular protector/stimulator in hypoxic conditions such as those encountered after stroke.
In this study, we provide crucial evidence for a critical role of Cdk5 operating through distinct signalling mechanisms in the co-ordinated and sequential stimulation of cell spreading (actin fibres, talin and integrin-β-1) and tube differentiation and sprouting (MEF2C). Furthermore, we show that specific manipulation of Cdk5 signalling via administration to brain ECs of the CIP-peptide, allowed enhancement of in vitro cell migration with production of mature sprouts and tubes. Therefore, p35/Cdk5 manipulation may represent a novel strategy for EC protection in response to ischaemic conditions.

Cell culture and treatment with Roscovitine
Immortalized (hCMEC/D3) human brain microvascular EC cells (hBMEC; [17]), were kindly donated by Prof. Babette Weksler (Division of Haematology and Medical Oncology, Weill Medical College of Cornell University, New York). Cells were cultured on type I rat tail collagen (BD Bioscence Pharma, UK) coated plates, using microvascular EC medium-2 (EBM-2) from Clonetics (Lonza, Germany), supplemented with growth factors as recommended by the manufacturer. The concentration range (2.5-100µM) of (R)-roscovitine (Calbiochem, UK) were chosen based on other previously published studies demonstrating effectiveness [5]. 24h of incubation with roscovitine (50 µM) was found to be optimal for the study. The effects of (R)-roscovitine on cell proliferation (72h) were analysed using the CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS) (Promega Corporation, USA). Nuclear membrane damage and activation of cell apoptosis were evaluated by nuclear inclusion of propidium iodide counterstained with Hoechst 33258 (Sigma-Aldrich, Germany), and Quantikine® kit (R&D, USA) for active caspase-3 and DuoSet IC® kit (R&D, USA) for active p53, respectively. Cell treatment with staurosporine (Sigma-Aldrich), (0.5-1 µM) was used as positive control for apoptosis and cell damage analysis. Appropriate amounts of DMSO, the roscovitine vehicle, were used to ensure that the effects were roscovitinespecific.

CIP cloning and stable transfections
A modified EGFP-IRES vector containing CIP sequence [16], (pLV-CIP plasmids), was a kind gift from Prof. Sashi Kesavapany, National University of Singapore. CIP peptide is comprised of 126 residues of p35 protein (accession number: AAH26347.1) spanning from aa 154 to aa 279. CIP-PCR products from pLV-CIP plasmid, obtained by PCR (forward primer, TTA GGA TCC TTG CCT GGG TGA GTT TC and reverse primer, GTC GGA TCC TCA TGG GTC GGC ATT TAT C) and digested by BamHI (Invitrogen, UK) were then inserted by ligation into pcDNA3-GFP expression vector. After positive cloning in DH5α, purified pcDNA3-CIP-GFP vectors were stable introduced by transfection into hBMECs using TransIT-2020 transfection reagent (Mirus Bio LLC, Medison, USA). EGM-2 medium containing 500 µg/ml Geneticin (Invitrogene Life Science, UK), was used as selective media. Positive hBMECs clones were detected by GFP fluorescent signals and by the level of CIP mRNA by RT-PCR. 18S rRNA (accession number: NM_022551.2) was used as housekeeping gene (forward primer, TAG AGG GAC AAG TGG CGT TC and reverse primer, TGT ACA AAG GGC AGG GACT T).

In vitro capillary tubule-like structure formation on Matrige TM
The preparation of ECs and Matrigel TM was performed as described previously [18]. 6x10 3 cells were added to the Matrige TM reduced in growth factors (Becton Dickinson, UK; prepared according to the supplier's instructions), in complete EBM-2 media with or without scalar amount of tested compound. The cells were incubated for 24-72 h at 37°C. DMSO (vehicle) alone, at the same concentration as in the test samples, was also added in a group of controls in all experiments. The number of closed tubes was counted with a Nikon inverted microscope. All experiments were performed in triplicate.

Spheroid based in vitro angiogenesis assay
The preparation of EC spheroids was performed as described previously [19,20]. Briefly, cells were harvested from sub-confluent monolayer cultures by trypsinization and 6x10 5 cells were suspended in EGM-2 mixed with DMEM (Lonza, Germany) 10% FBS and 0.25% (w/v) carboxymethylcellulose (Sigma, UK). Cells were then seeded into non-adherent roundbottom 96-well plates to assemble into a single spheroid within 24 h at 37 °C, 5% CO 2 . Basic fibroblast growth factor (human recombinant FGF-2, (BD Bioscience, UK)) was used as positive control and added at a final concentration of 25 ng/ml. After 24h, gels were photographed and sprouting was assessed quantitatively.

Colorimetric adhesion assay
The adhesion assay was performed in a 96-well pre-coated with collagen type-I and blocked with 2% BSA (w/v PBS) for 120 min at 37°C. Then the plates were heat denatured at 85°C for 10 min, to block non-specific bindings. As negative control, well were only coated with PBS. 1x10 4 cells were seeded in each well and after 30 min from the seeding the non-adherent cells removed by washing with PBS. A value of 100% attachment was estimated by adding the cells to uncoated and fixed (methanol) plastic, without washing. After washing, 50ml of substrate solution (3.75mM p-nitrophenyl N-acetyl-β-dglucosaminide, Sigma-Aldrich, 0.05M Sodium citrate pH 5.0, 0.25% Triton-X-100) was added to the wells and the plates incubated overnight at 37°C. After stopping the reaction (50mM glycine, pH10.4.5 mM EDTA) the optical density of each well was determined at 490nm using a microplate reader (Multiskan Ascent, Thermo Life Sciences, UK). DMSO (vehicle) alone, in the same concentration as was present in the test samples was also added in a secondary group of controls in all experiments. Experiments were done in triplicate.

Spreading assay
Cells were placed in a 96-well plate and incubated for 1h. The plate was pre-treated as described for the colorimetric adhesion assay. The experiment was stopped by fixing the cells with cold methanol for 5 min at room temperature. Cells were then dried and stained with 1% methylene blue in water. Approximately 100 cells were examined in multiple microscopic fields. Multiple fields were imaged from each condition. Cells were classified as spread when flat and the cytoplasm visible around the entire circumference of the nucleus and un-spread, when round and without visible cytoplasm. The number of cells was then expressed as a % of total cells. Experiments were done in triplicate.

In vitro angiogenesis assay with continuous live cell imaging
In different experiments, hBMECs, Cdk5-wt, Cdk5-DN, and/or CIP transfectants, (1-2x10 4 /ml), were added to a 24-well plate coated with collagen type I, in complete EBM-2 medium and maintained at 37 °C and 5% CO 2 for 24 h to permit the formation of a confluent monolayer. Monolayer scratch wound with or without scalar amounts of the tested compound roscovitine (2-50 µM) was then induced by scoring the confluent cell monolayer with a sterile pipette tip to leave a scratch of approximately 0.4-0.5 mm in width. In vitro cell differentiation in tube like structures was performed seeding the cells (100,000 cells/ml) in growth media containing or excluding the roscovitine (50 µM) over a preformed gelled layer of Matrigel TM . In vitro adhesion, spreading and elongation assays were performed seeding separately non-confluent cells on to a collagen pre-coated 24-well tissue plate. Cells were maintained at 37°C and 5% CO 2 . The cell migration from the scratched monolayer and formation of tube like structures were monitored continuously in real time for 24h and 72h, respectively and cell adhesion and subsequent spreading for 2 h, using the automated cell culture Cell-IQ® Continuous Live Cell Imaging (Chip-Man Technologies Ltd, Finland). Cells were monitored via the ImagenTM program, and data analysed by AnalyserTM software. Each experiment was performed in triplicate. Hypoxia (1% O 2 , 5% CO 2 and N 2 ) experiments were performed using reduced EBM-2 (0.1% foetal bovine serum) media, with a pre-treatment of 24h. DMSO (vehicle) alone, in the same concentration as was present in the test samples was also added in a secondary group of controls in all experiments.

Western Blotting
Medium was removed quickly and cells were washed twice with cold-ice PBS buffer. Then, protein extraction was performed using ice-cold RIPA buffer in the presence of a protease and phosphatase inhibitor cocktail. After 30 min of incubation in ice, total cell lysate was centrifuged at 12,000 g x 20 min at 4°C. Bredford protein assay (Bio-rad, UK) was used to quantify protein amount in cell lysate. Protein absorbance was determined at 580 nm using a microplate reader. 30 µg of protein was loaded and subsequently separated on 10% SDS-PAGE gel. Protein samples were transferred onto nitrocellulose membranes (Whatman International Ltd.), stained with amidoblack solution and photographed to verify the equal loading and the quality of transfer. The membranes were subsequently incubated 1h with blocking solution (milk 1% in PBS-Tween buffer pH 7.4). Incubations with the primary antibodies against human Cdk5, MEF2C, Hsp70 (Abcam, UK), pTyr (15 )Cdk5, p35, integrin beta-1 (Santa Cruz Biotechnology, UK), Cdk2 and pThr (160 )Cdk2 (Calbiochem, UK), talin (Upstate, UK), and anti-GAPDH (Santa Cruz Biotechnology, UK) were performed with conditions as recommended by the manufacturers, overnight at 4°C. The membranes were then washed three time with PBS-Tween buffer before to be incubated with appropriate horseradish peroxidase conjugated secondary antibodies (Dako, UK) at room temperature. Protein bands were visualised using a chemiluminescence detection kit (Invitrogene Life Science) and protein contents were expressed as the ratio to GAPDH protein levels (Bio-Rad quantity one software). Analysis were performed in triplicate.

Confocal and fluorescence microscopical analysis.
hCMEC/D3 cells were added to a glass cover slip coated with collagen type-I, in a 24-well plate in complete EBM-2 (2.5% fetal bovine serum) medium. After fixing with 4% paraformaldehyde (w/v in PBS) cell were washed with PBS and permeabilised with 0.1% Triton X-100 (v/v in PBS). After blocking with 1% BSA (w/v in PBS) incubations with primary antibodies against selected proteins were performed with conditions as recommended by the manufacturers, overnight at 4°C. Secondary antibodies Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 488 goat anti-mouse IgG, Alexa Fluor 568 anti-rabbit and Alexa Fluor 460 donkey anti goat (Molecular Probes/Invitrogen) were incubated with conditions as recommended by the manufacturer. Rhodamine-phalloidin (Molecular Probes/Invitrogen) was used to stain filamentous actin (F-actin). Nuclei were stained with Hoechst 33258 solution (Sigma, UK). Samples without primary antibodies were used as negative controls. Confocal microscopy was performed with a Leica TCS SPE1000 system and the accompanying Leica Application Suite Advanced Fluorescence (LAS AF) software (Leica Microsystems UK Ltd, Milton Keynes, Bucks, UK). Images were acquired at 1024x1024 pixel resolution. Cells were pre-treated with roscovitine (50 µM) for 24h or the appropriate vehicle (DMSO) and then seeded on the plates.
Signal of autofluorescence was tested, including a negative control antibody (IgG) staining, as well. Experiments were performed in triplicate.

Cdk5 kinase activity
Cell lysates for immunoprecipitates and kinase assay were obtained as described for western blotting, in contrast, a modified RIPA lacking ionic detergent was used. Cdk5 was immunoprecipitated using protein A-sepharose CL-4B and incubated in vitro with Histone H1 (Sigma, UK) and [γ-32 P] ATP (Amersham Radiochemicals, UK). Before to start the immunoprecipitaiton, cell lysate was pre-cleared suing 100 µl of sepharose-A slurry per ml of cell lysate. The reaction mixture with Histone H1 was then resolved by SDS-PAGE followed by autoradiography. Active Cdk5 would result in the appearance on autoradiography film of a signal at the size of phosphorylated histone H1. A purified active complex Cdk5/p35 expressed in Sf21 cells (Millipore, Billerica, USA) served as a positive control. 2ng of the active complex was used in the assay. Immunoprecipitate using rabbit anti-goat antibody (Dako, UK) served as negative controls.

MEF2C siRNAs
Knockdown of MEF2C expression was performed using Silencer® Select Pre-Designed & Validated siRNA (Applied Biosystems, Life technologies, USA). SiRNA delivery was performed using Lipofectamine 2000 (Invitrogene, UK). After three days of treatment, in selected media (EBM-2 with reduced FBS and without antibiotics), knockdown efficiency was tested via protein assay (Western Blotting) and cells used for in vitro angiogenesis analysis. Silencer® control scrambled siRNAs was used as a negative control. Assay conditions were performed as recommended by the manufacturers.

Affymetrix Microarray
Total RNA was isolated using Trizol Reagent, and purified using the RNeasy Micro Kit (Qiagen, UK). The assay was performed according to the manufacturer's instructions. Concentration and purity was determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies) and RNA quality was assessed using Agilent RNA 6000 Nano Kit and 2100 Bioanalyzer platform (Agilent, USA). The expression pattern of untranfected cells and several clones from cells transfected with the empty vector or a dominant negative form of human Cdk5 were compared by microarray analysis. Microarrays and cDNA synthesis were performed following the GeneChip® WT cDNA Synthesis and amplification Kit (Affymetrix) and following the Affymetrix GeneChip® Whole Transcript (WT) Sense Target standard protocol [21,22]. The array data were summarized and normalized with the RMA algorithm using the Affymetrix Expression Console software. Statistical analysis was performed using Spotfire DecisionSite software (Integromics).

Calpain activity
The fluorometric assay was performed according to the manufacturer's instructions (Abcam, UK). Cytosolic proteins Cdk5/p35 Signalling in Angiogenesis PLOS ONE | www.plosone.org were specifically extracted without contaminations of cell membrane and lysosome proteases and preventing autoactivation of calpain during the extraction procedure according to the manufacturer's instructions. Calpain activity was quantified using a fluorescence microplate reader BMG Labtech, USA. The fluorometric assay included a positive control of active calpain I, as well. Data were expressed as relative fluorescence units. Experiments were performed in triplicate.

Statistical analysis
All data were expressed as mean ±SD (standard deviation). P values were calculated using the ANOVA one-way test using Student's t-test for unpaired sample led by MS and followed by the Wilcoxon 2-sample test. Regression analysis was carried out according to standard methods provided by MS Excel and to SPSS version 12. Results were considered to be statistically significant at values of P <0.05.

R-roscovitine is a selective inhibitor of Cdk5 activity in hBMECs
(R)-roscovitine is a well-recognized anti-neoplastic drug, carrying out both apoptotic [23] and anti-proliferative [24] actions with highly selective inhibitory potential on Cdk5 [5], and Cdk2 activities [25]. To ensure selective inhibition in our cellular model, we evaluated roscovitine effects on Cdk5 activity and on the protein levels of Cdk5, Cdk2 and activated pTyr (15) pCdk5 and pThr (160 )Cdk2. Cdk5 activity was estimated by measurement of histone H1 phosphorylation, one of its targets. The effect of scalar amounts of roscovitine on cell viability, cell apoptosis and proliferation were also determined. Roscovitine at the concentration of 50 µM inhibited completely Cdk5 activity ( Figure S1) and reduced significantly Cdk5 and pCdk5 protein contents ( Figure S1), without affecting Cdk2 or pThr (160 )Cdk2 ( Figure S1). Roscovitine did not affect cell viability, proliferation or apoptosis ( Figure S2). Cytotoxic effects were seen at concentrations of 70 µM and greater.

Cdk5 overexpression supports in vitro hBMEC angiogenesis during hypoxia
Recent observations suggest that apoptosis of neurons after brain ischemia is strongly associated with Cdk5 hyperactivation mediated by its activator p25 [9], however, normal signaling through the Cdk5/p35 complex promotes neuronal survival, whilst inhibition of normal Cdk5 activity may inhibit Hif-1α accumulation and concomitantly induce neuronal cell death [26]. Firstly, we optimized an in vitro model of low oxygen tension mimicking hypoxia during stroke, wherein hBMEC were exposed to 24h of low oxygen levels (1%). Hypoxia conditions were defined on the evidence that in human hypoxic brain tissue (i.e. after subarachnoid haemorrhage) the partial pressure of brain tissue oxygen (PtiO 2 ) decreased dramatically from the normal values of 40 mmHg [27] to <10 mmHg [28]. Considering the conversion of % oxygen to units of mm Hg, that assumes 100% oxygen equal to 760 mm Hg, our system was set at 1% of O 2 delivery, as previously described [10], to create a severe hypoxic environment [29].
In our model, the efficiency of hypoxia ( Figure S3) was evidenced by the increased nuclear inclusion of propidium iodide ( Figure S3), increased protein expression of heat shock protein Hsp70 (Figures S3B and S3F) and activation of calpain activity ( Figure S3E). We found that hypoxia significantly reduced in vitro angiogenesis in hBMEC, reducing cell migration, tubule formation and/or cell sprouting. This was associated with decreased p35 protein content ( Figure S3) and increased p25/p35 ratio ( Figure S3), with no evident changes in Cdk5 expression ( Figure S3). To understand the physiological importance of Cdk5/p35 signalling, Cdk5 activity was then deregulated using stable transfections of either Cdk5 kinase inactive mutant -D144N, (Cdk5-DN) or Cdk5 wild-type ( Figure  1), and by pharmacological inhibition with roscovitine ( Figure  2). The effects of Cdk5 inhibition on temporal and spatial cellular adaptations were then analysed by functional in vitro angiogenesis assays and monitored in real time using IQ Live Cell Imaging.
In normoxia, inhibition of Cdk5 kinase activity with roscovitine, resulted in a significant loss or inability to form and maintain potentially viable capillary structures (Figure 2), as shown by the inhibitory effect on in vitro cell adhesion ( Figure  S4), cell distribution between early and late spreading ( Figure  2A), cell motility ( Figure 2B), to culminate with the loss of EC tubule branching networks ( Figure 2G) and cell sprouts ( Figure  2L). Similarly, cells expressing the dominant Cdk5-kinase mutant D144N (Figure 1) were unable to spread, migrate and form new capillary structures or sprouts (Figure 1). Cells overexpressing Cdk5 wild-type ( Figure 1) showed enhanced angiogenesis in vitro, characterized by increased cell motility, tubule and sprout formation. Noteworthy, in Cdk5-wt the increased angiogenic activity was accompanied by abnormal tubule and sprouts organization (smaller and thinner tubes and sprouts) ( Figure 1C), suggesting that enhancing Cdk5 activity alone may not be a complete solution to allow protection and functional enhancement of in vitro angiogenesis. During hypoxia, inhibition of Cdk5 signalling was associated with a further inhibition of tube-formation beyond that produced by Cdk5-DN transfectants during normoxia (Figure 1), whereas cell migration, sprouting and tubule formation, were almost maintained at control-normoxic levels in the Cdk5-wt (Figure 1), thus Cdk5 overexpression preserved angiogenesis during in vitro hypoxia. pTyr (15) Cdk5 and p35 mobilization are essential for in vitro brain endothelial cell migration through association with integrin beta-1 and talin proteins Cytoskeleton dynamic and focal complex organizations are crucial determinants for successful angiogenesis. We hypothesized that Cdk5 inhibition may affect the cell cytoarchitecture and the observed inhibitory effects on angiogenesis in vitro may be the direct consequence of Cdk5 misplacement from actin and focal complexes. We also examined the involvement of the Cdk5 substrate, p35 in this cytoskeletal/focal organization.
Cdk5/p35 Signalling in Angiogenesis PLOS ONE | www.plosone.org We found a marked and specific distribution of Cdk5, pCdk5 and p35 with F-actin in all phases of cell motility (Figure 3), having a precise time-dependent subcellular localization during spreading phases. Analysis of p35 intracellular localization determined at different time/stages of cell spreading, showed us that p35, emanating from the nucleus ( Figure 3C), co-localized with actin stress fibres extending down the fibre from the nucleus as the cells spread, to culminate in filament tip expression in late spreading and moving cells (Figure 3C), suggesting a precise spatial-temporal mechanism linking p35 to stress fibre formation and/or contractility. Of particular note p35 ( Figure 4A), and pCdk5 ( Figure 4B), were clustered with talin and integin beta1 into lamellipodia during cell spreading ( Figure 4C and 4D) and at the leading edge of elongatedmoving cells, suggesting an involvement in the determination of cell polarity.
Inhibition of Cdk5 activity with roscovitine resulted in a total disorganisation of the actin-cytoskeleton and focal protein localization ( Figure 5). With roscovitine, pCdk5 ( Figure 5A) and p35 ( Figure 5B) disassociated from actin filaments and stress fibres, to form, with actin, cytoplasmic clumping and bundles. Thus, p35 and Cdk5 activation seem essential for the mechanisms of actin polymerization. Inhibition of Cdk5 in Cdk5-DN mutant or roscovitine treated cells markedly increased talin fragmentation ( Figure 5) and reduced the formation of focal talin tips ( Figure 5C), preventing integrin beta-1/talin bundling ( Figure 5D). Therefore, we hypothesised Cdk5/p35 to have a multipart role where Cdk5 and p35 contribute to the mechanisms inducing matrix/focal cytoskeletal stability and on the other side, in those governing actin polymerization and depolymerization, as required for cell traction and progression of cell movement. In this scenario, p35 could be the/a stimulus for F-actin fibre polymerization and polarity leading to the formation of actin/focal plaque complexes. Inhibition of Cdk5 activity or reduced p35 content, as seen during hypoxia ( Figure  S3) may reduce cell tension, and consequently affects spreading architecture and in vitro migration.

Cdk5 and MEF2C promote proper spatial organization of brain endothelial cells
The chaotic in vitro organization of EC tubule branches observed in Cdk5-wt transfectants, prompted us to speculate about the interplay of Cdk5 with other pathways involved in the promotion of EC differentiation. Using an affymetrix microarray ( Figure S5) we analysed a cluster of genes involved in angiogenesis and the potential impact of Cdk5 deregulation on their expression. By Western blot analysis and coimmunoprecipitation assays, we identified the myocyte enhancer factor MEF2C ( Figure S5), a transcription factor involved in cell differentiation and angiogenesis, as a putative  Figure S5).
Moreover, blocking Cdk5 activity in Cdk5-DN, notably reduced MEF2C protein levels and the opposite effect was seen in Cdk5-wt ( Figure S5). In control cells, MEF2C knock down by siRNAs (Figure 6), reduced in vitro tubule formation ( Figure 6C) and cell sprouting ( Figure 6E), but had only a minor, non-significant effect on cell migration ( Figure S6). In view of that, we did not find any effect of MEF2C knock down on actin fibre formation and/or on the co-localization of p35 with talin ( Figure S6C) in moving cells. This could suggest that in hBMECs, MEF2C may be mainly required at the stage of differentiation. Accordingly, we observed that in Cdk5-wt ( Figure 6C), MEF2C knockdown diminished the chaotic organization of capillary tubule like structure, demonstrating MEF2C as a Cdk5 molecular intermediate required to control the proper time and spatial capillary network in hBMECs. This interaction would likely take place at the nuclear level -since both proteins were observed to co-localise in the nuclei ( Figure  S5E).

Modulation of Cdk5/p25 signalling rescued in vitro angiogenesis in acute hypoxia
The neuro-cytoprotective effect of CIP peptide has been amply demonstrated during neurotoxic stress, wherein the protection is conferred by the inhibition of Cdk5/p25 complex formation [16]. Thus, we asked whether the introduction of this peptide into brain ECs may prevent the detrimental effects induced by hypoxia, possibly reducing the cleavage of p35 into the more stable p25 peptide.
We found that CIP expression ( Figure S7) was able to rescue cell migration ( Figure 7A) and the in vitro angiogenic potential during hypoxia, increasing the capillarization ( Figure  7B) and the formation of new sprouts ( Figure 7C). At the cytoskeletal level the organization of stress fibres with p35 clustering, which was de-regulated during hypoxia in control cells (Figure 8Ai), was markedly stabilized (Figure 8Aii). This was accompanied by an increased p35/p25 protein ratio (Figure 8, C and D), indicating that CIP potentially can interfere with/inhibit p35 cleavage induced by hypoxia ( Figure S3C).
Notably, Cdk5 localization with actin fibres was only marginally affected by hypoxia (Figure 8Bi) or in CIP transfectants ( Figure  8 Bii). CIP vector insertion also demonstrated a positive in vitro pro-angiogenic effect during normoxic conditions, promoting cell migration ( Figure S8) and increasing talin content, specifically with production of a greater number of talin tips containing p35 protein ( Figure S9).

Discussion
Cdk5 hyper-activation mediated by the cleaved p35 fragment, p25, and initiated via calpains is crucially associated with neuronal cytoskeletal derangement, neuronal apoptosis and cell death during ischemic brain insult [9], while activation of Cdk5 through the parent p35 is associated with cytoprotection [11]. Cdk5 and p35/p25 are concomitantly overexpressed in ECs in hypoxic regions of stroke tissue [10], suggesting that the balance of signaling between these pathways may help to define cellular fate in relation to angiogenesis or cell protection after stroke [13]. Our study demonstrates that Cdk5 activation through its major substrate p35 provides a key trigger for initiation and maintenance of in vitro brain EC angiogenesis and that this interaction and signalling cascade is crucial for ensuring correct cytoskeletal organisation/dynamics, involving integrin beta-1 and talin, and allowing cell spreading and subsequent cell migration. Operating via the novel binding partner, nuclear transcription factor MEF2C, angiogenesis in vitro proceeds with cell differentiation and tube-like-structure formation. Our results also show that introducing a natural inhibitor of p25/Cdk5 complex formation, the CIP peptide [16], we were able to alter the balance of p25/p35-Cdk5 signalling in favour of p35 and concomitantly and significantly increase in vitro brain EC angiogenesis during normoxia and most importantly, promoting cell survival and capillarisation in hypoxic conditions. Several studies in animal and in vitro models have described important functions of the myogenic transcriptional effector MEF2C, a crucial pro-survival protein, during vascular development, vasculogenesis and angiogenesis [30,31]. Targeted deletion of the MEF2C gene in mice, results in embryonic lethality with significant cardiovascular defects and severe vascular abnormalities, wherein endothelial cells retain the capability to differentiate but are unable to organize into a functional vascular network [32]. Here, we found a significant association between MEF2C protein levels and degree of in vitro angiogenesis using both Cdk5-wt and DN-kinase mutant transfectants. An increase in MEF2C expression in Cdk5-wt resulted in abnormal and chaotic vascular network production, higher levels of cellular migration and longer but thinner spheroid-sprout lumens. In contrast, the reduced amount of MEF2C seen in DN-kinase mutants was coupled with significantly reduced and disorganised EC spatial organisation. Knockdown of MEF2C using small-interfering RNAs in Cdk5wt, re-established the proper cellular network, whilst reducing abnormal cell migration and sprouting. In non-transfected cells MEF2C knockdown blocked almost completely the formation of neo-capillaries, suggesting that MEF2C could be an important Cdk5-signalling intermediate in brain ECs, and a key requirement for functional in vitro angiogenesis. A relationship between Cdk5 and MEF2C in angiogenesis has not been reported previously. However, Cdk5/p25 complexes have been shown to phosphorylate other members of MEF2 family (MEF2A and MEF2D), inhibiting their pro-survival activity, through a caspase-3 dependent degradation, in response to neurotoxic stress [33,34], without show any targeting on MEF2C [33]. Notably, restoration of the active form of MEF2 has been proved to be protective following focal ischemia [35] and increasing MEF2 phosphorylation levels (ser59) has been recently described in rat brain ischemic stroke models after inhibition of calpain/Cdk5/p25 signalling, by sildenafil, used to antagonise chemical hypoxia induced by malonate [12].
Here, using a protein pull down assay, we were able to coimmunoprecipitate the two proteins, but whether this association may be due to a regulatory mechanism on MEF2C stability mediated by Cdk5 phosphorylation is still to be defined. Further work should investigate whether Cdk5 is acting as a showing a more diffuse cytoplasmic distribution, while Cdk5 (Bi) maintained its co-localization with actin fibre tips (arrows in merge, upper right panel) (ii). Compared to the controls CIP insertion protected p35 (Aii) and Cdk5 (Bii) intracellular localization with actin filaments and fibres from the effects of hypoxia, this was associated with increased p35 protein contents (C) and increased p35/p25 protein ratio (D). Calculated results are reported in graph D; optical density (OD) of p35/p25 protein ratio in control untransfected cells and in CIP transfectants determined in normoxic and hypoxic conditions. Typical images from n=2 independent experiments. Immunofluorescence confocal analysis; bars, 20 µm. Original magnification, x 100. Each experiment was performed in triplicate. doi: 10.1371/journal.pone.0075538.g008 Cdk5/p35 Signalling in Angiogenesis direct regulator of MEF2C gene expression, perhaps via the E2F1-pRb axis [36,37].
Published work has demonstrated non-nuclear functions linked to Cdk5 in the regulation of cytoskeleton dynamic and motility [5], showing Cdk5 as a key modulator of talin stability and therefore essential to cell migration [38]. Thus, we investigated the interplay of Cdk5 and p35 with classical molecular players of the control of mechanical forces and motility. We found a specific time/spatial interaction of active (phosphorylated) p-Cdk5 and p35 with integrin beta-1 and talin, directed to the formation of microvesicles in late-spreading and elongated/moving cells. Accordingly with previous observations [5,39,40], p35, Cdk5 and pCdk5 localized at the focal tips of actin stress fibres. Inhibition of this interaction, by roscovitine or in DN-mutants blocked in our cellular model, cell protrusion and stable adhesions. Notably, we found that p35 localization alongside actin fibres during spreading is time-dependent and culminates with the formation of focal clusters with Cdk5, integrin beta-1 and talin. On this basis, we hypothesised that p35 may be a putative adaptor implicated in the generation of cellular tension. The association of p35 with talin and integrin beta-1 may result in the formation of a molecular bridge among cytoplasmic domains of integrins and actin fibres [38]. p35mediated activation of Cdk5 at that sites, in conjunction with talin, may be required for activation/phosphorylation of integrins and to the resulting actin polarised organization [40]. Hence, we suggest that the formation of p35 complexes with pCdk5 at focal adhesion points may relate to Cdk5 activation at focalregions and be involved in integrins kinase/activation.
In our study, hypoxia markedly impaired in vitro angiogenesis, leading to a noticeable alteration in stress fibre cyto-structure with delocalization of p35 from these fibres. This was reflected in a marked inhibition of cell spreading and migration. These changes were associated with decreased p35 expression and a parallel increase in calpain expression and in the cleavage product p25. Notably, modulating Cdk5/p35 activation with the novel natural inhibitor of p25/Cdk5 complex, the CIP-peptide, we were able to reduce p35 cleavage to p25 and ameliorate angiogenesis during in vitro hypoxia. CIP increased recovery of cell spreading and formation of stable neo-capillaries, potentially increasing p35 localization with actin fibres and therefore the stabilization of the actin cytoskeleton. On this base, we propose that p35/cdk5 signalling elicits a specific protective role during in vitro angiogenesis and together with the observed protective action of the CIP peptide, may potentially aid in tissue remodelling after ischaemic stroke.
In conclusion, our results show that the p35/Cdk5 is a pathway that is critically involved in controlling cell dynamics of brain EC. Activation of p35/Cdk5 signalling at the expense of p25-Cdk5 plays a protective role in hypoxic conditions, and positively contributes to preserve cell motility and the proper spatial and temporal control of cytoskeletal dynamics, which is essential for successful angiogenesis.