Unless otherwise stated, chemicals, reagents and drugs were obtained from Sigma, Dorset, UK. Enzymes for molecular biology were from Promega, Southampton, UK.

Primary human umbilical vein ECs (HUVECs) and human aortic ECs (HAoECs) (Promocell, Heidelberg, Germany) were cultured (37°C, 5% CO₂) in EC growth medium-2 (EGM-2) consisting of EC basal medium supplemented with 2% fetal bovine serum, gentamicin/amphotericin (GA-1000), and growth supplements (Lonza, Wokingham, UK). All ECs were studied between passages 2 and 6.

HUVECs and HAoECs (4×10⁴ cells/well) were re-suspended in basal medium (1 ml) and seeded onto 24 well plates coated with Matrigel (250 µl, BD Biosciences, Oxford, UK) as described [19]; [20]. Basal medium comprised EGM-2 supplemented with ascorbic acid, heparin and GA-1000 but without serum or growth factors. Drugs, growth factors or vehicle control (0.004% ethanol v/v) were added at the start of each experiment at concentrations indicated in Figure legends.

Photomicrographs (5× magnification) of the centre of each well were obtained after incubation for 4, 8 and 24 hours [21] and TLSs were quantified by counting branch points by an investigator blinded to treatment. Data were expressed as a percentage of the number of connections in control wells.

TLS formation was assessed directly using time-lapse imaging. HUVECs (2×10⁵) in 2 ml basal medium containing 5 mM HEPES (Lonza, UK) were seeded onto SlideFlasks (Nunc, New York, USA) pre-coated with 750 µl Matrigel and incubated with vehicle or 10 ng/ml VEGF, 600 nM cortisol, or both. Images were acquired with a Leica DM IRBE microscope (10× magnification) and Q500MC image processing system (Leica Cambridge Ltd, UK) from 3 separate positions/flask every 4 minutes for 24 hours. Tubes per field of view were counted every 2 hours in reconstructed movie clips.

The endothelial nature of TLSs was assessed using immunohistochemistry. Briefly, HUVECs (10,000) were cultured in standard basal medium (250 µl; 5 hours) on 8-well permanox chamber slides (Nunc, USA) pre-coated with Matrigel (100 µl). TLSs were fixed in 10% formalin and stained using the Vectastain ABC Kit (Vector Laboratories, USA), according to the manufacturer's instructions, in combination with rat anti-mouse CD31 (1 in 50 dilution) monoclonal antibody (BD Pharmingen, USA). Slides were incubated (room temperature; 5 minutes) with 3,3′-diaminobenzidine (DAB) (Vector Laboratories, USA) with positive structures staining brown. For negative controls, the primary antibody was omitted. Slides were viewed by light microscopy (Karl Zeiss Axioskop, Carl Zeiss MicroImaging, Inc, USA) and images captured from a live-feed camera (3-CCD, JVC Professional Europe Ltd, UK) using the Microcomputer Imaging Device (MCID; InterFocus Imaging Ltd, UK).

Components of the EC cytoskeleton under basal conditions and during TLS formation were assessed using immunofluorescent staining. Briefly, HUVECs were re-suspended in EGM-2 basal medium growth medium with or without cortisol (600 nM), and seeded onto coverslips coated with phenol red-free Matrigel (30 µl). After incubation for 1, 4, 10 or 22 hours, cells were fixed, permeabilised and blocked [22], then incubated with mouse monoclonal anti-α-tubulin antibody (clone B512, Sigma; 1∶1000, 1 hour). Secondary goat anti-mouse IgG (H+L, Alexa Fluor 594, Molecular Probes-Invitrogen, Paisley, UK; 1∶500) was then added for 1 hour. Direct fluorescent labelling of filamentous-actin (F-actin) and DNA was achieved by including (1∶500) phalloidin Alexa Fluor 488 (Molecular Probes-Invitrogen) or DAPI (0.1 µg/ml), respectively, with the secondary antibody. Coverslips were mounted on slides with Mowiol-glycerol and imaged as described [22].

ECs or TLSs grown on Matrigel were recovered using MatriSperse cell recovery solution (BD Biosciences, Oxford, UK) according to manufacturer's instructions. RNA was isolated using TRIzol (Invitrogen Life Technologies, Paisley, UK) and quantified with UV spectroscopy. mRNA for GR was assessed in confluent ECs and established TLSs. Total RNA (1 µg) was reverse-transcribed (45 min, 42°C) using a Kit (Promega, Southampton, UK). cDNA templates underwent PCR amplification (35 cycles) with Taq polymerase in the presence of oligonucleotides specific for GR (5′-GTGGTTTATAGAGGGCCAAGACTTGG-3′ and 5′-GGCACAACTTCCCTTTTCTGATATACAC-3′). PCR was performed as described [23] with an annealing temperature of 62°C. Products were analyzed by electrophoresis (1.2% agarose gel w/v in TBE buffer containing 0.001% ethidium bromide v/v) and visualized under UV light. PCR product size was 354 base pairs (bp). Negative controls included either no reverse transcriptase, or no cDNA template.

The influence of glucocorticoids on the expression of angiogenic factors during TLS formation was examined using quantitative real-time PCR (QrtPCR). HUVECs (100,000/ml) were seeded on Matrigel-coated cell culture dishes in the presence of cortisol (600 nM) or vehicle (control) and incubated for 1, 4, 8 or 22 hours. RNA was recovered using Matrisperse solution, Trizol (1 ml) was added to the cell pellet and the sample was frozen (−80°C).

cDNA was synthesised from RNA using the Promega Reverse Transcription System (Promega UK, Southampton, UK). The reactions were carried out on an Eppendorf Mastercycler Gradient (Eppendorf, Germany) consisting of incubation at 42°C for 45 minutes followed by 95°C for 5 minutes and finally chilled to 4°C. Quantification of the transcript was performed using human tissue-specific TaqMan primer probe mixes purchased as ready-to-use assays (Applied Biosystems, Cheshire, UK) with the Lightcycler 480 Real Time PCR system (Roche Diagnostics Ltd, West Sussex, UK). Samples were heated to 95°C for 5 minutes for pre-incubation then underwent 50 cycles of PCR amplification (denaturation at 95°C for 10 seconds, primer annealing at 60°C for 30 seconds and elongation at 72°C for 1 second) and finally underwent cooling at 40°C for a further 5 minutes.

RNA levels were determined for each sample, run in triplicate, from standard curves generated for each primer-probe set by serial dilution of pooled cDNA from each tissue. Cyclophylin (Mm02342430_g1; Applied Biosystems, Cheshire, UK) mRNA levels were used as internal references to normalise transcript levels. Changes in mRNA levels in TLS exposed to cortisol were determined for genes involved with stimulation of angiogenesis (VEGF, VEGFR2), inhibition of angiogenesis (thrombospondin-1), cell-matrix interactions (μ₆-integrin, caveolin-1) and regulation of cell fate (delta-like-4 (Dll4), Notch).

ATP production was measured from viable cells (Cell Titer Glo assay, Promega, UK). HUVECs (1.5×10³ cells/well) were seeded onto plates for 2 hours, then incubated for 94 hours as described for BrdU. CellTiter-Glo Reagent (100 µl) was then added to each well (10 minutes, room temperature) before detection of luminescence (Wallac 1420 VICTOR² plate reader, Perkin-Elmer, Buckinghamshire, UK).

To perform BrdU incorporation assays (Calbiochem-Merck, Nottingham, UK), HUVECs (3.5×10³ cells/well) were cultured in growth medium (EGM-2 basal medium (Lonza, UK) supplemented with heparin, ascorbic acid, GA-1000, and 2% charcoal-stripped fetal bovine serum) without additional growth factors, and then incubated for 46 hours with: VEGF (25 ng/ml); VEGF plus SU5416 (1 nM-1 µM); VEGF plus cortisol (3 nM-1 µM); vehicle (0.004% ethanol v/v or 0.167% DMSO v/v); or medium alone. BrdU was added to the wells 1 hour after incubations started. After further incubation (45 hours), cells were fixed, denatured (BrdU assay kit, manufacturer's instructions), and incubated with the anti-BrdU antibody and then with horseradish peroxidase-conjugated secondary goat anti-mouse IgG. Finally, cells were incubated with tetra-methylbenzidine (TMB) and the reaction stopped with sulphuric acid (2.5 M). Optical densities (405 nm and 540 nm) were measured (Multiskan Ascent plate reader, Cheshire, UK) and the absorbance of cells without BrdU subtracted from each reading.

Boyden chambers [24] were prepared by seeding HUVECs (2×10⁵) onto the upper compartment of 8.0 µm ThinCerts (Greiner Bio-One, Gloucestershire, UK), incubating (24 hours, 37°C), and then labelling with calcein-AM (8 µM). Labelled cells were detached (Trypsin-EDTA) from the underside of ThinCerts and fluorescence measured (Fluoroskan Ascent FL plate reader, Cheshire, UK). The influence of cortisol on migration was assessed by adding VEGF (10 ng/ml), VEGF plus the tyrosine kinase inhibitor SU5416 (1 µM; which inhibits the angiogenic signalling induced by VEGF binding to its receptor), VEGF plus cortisol (600 nM), SU5416 alone or cortisol alone to the medium.

Results from triplicate wells in the same experiment were averaged and treated as single data points. Data are expressed as mean±SEM, where n indicates the number of experiments. Statistical analysis was performed by one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparison post hoc test or by repeated measures ANOVA. Differences were considered significant when p<0.05.

HUVECs cultured on plastic retained a typical cobblestone appearance (not shown). Once cultured on Matrigel, however, they formed characteristic networks of tube-like structures. These comprised cells connected by filopodia-like extensions which retained immunoreactivity for CD31 (an endothelial cell marker; Figure 1A). At higher magnification these cell-cell connections showed evidence of lumen formation. (Figure 1A).

Staining of actin filaments (with fluorescently-conjugated phalloidin) and microtubules (with α-tubulin antibodies) clearly identified cytoskeletal fibres in undifferentiated HUVECs (Figure 1B; n = 4). Actin filaments appeared as well-formed, parallel stress fibres throughout the cytoplasm while microtubules appeared to radiate from the centrosome region occupying the traditional peri-nuclear area. F-actin appeared more diffuse and homogeneous throughout cells after 1 hour of cortisol treatment, while microtubule staining appeared unaffected (Figure 1B, n = 2). Microtubules and actin filaments were both observed in TLSs formed after 4 hours on Matrigel (not shown) and remained evident after 8 (not shown), 10 and 22 hours in culture (Figure 1B).

Expression of GR was demonstrated in confluent HUVECs and HAoECs (Figure 1C), and was maintained in HUVECs 22 hours after induction of TLS formation (Figure 1C).

TLS formation by HUVECs was inhibited by cortisol (300–1200 nM; Figure 2A–B), but not by cortisone (the inert 11-keto metabolite of cortisol, 300–1200 nM; Figure 2A–B), in a concentration- and time-dependent manner. This inhibition was abolished by the GR antagonist RU38486 (Figure 2C). Inhibition of TLS formation was also obtained with the GR selective, synthetic glucocorticoid dexamethasone (600 nM; Figure 2C). Similar results were obtained in HAoECs in which inhibition of TLS formation by both cortisol and dexamethasone was evident after 5 hours (Figure 3A(i)) and maintained after 22–24 hours (Figure 3A(ii)) in culture.

VEGF stimulated TLS formation both by HAoECs (Figure 3B) and by HUVECs (Figure 4). This effect was blocked in both cell types by cortisol and dexamethasone (p<0.01) after 4 hours (data not shown) and 22–24 hours (Figures 3 & 4) in culture. PGF_2α (100 nM) also stimulated TLS formation by HUVECs (138±9% connections relative to control, p<0.01) and this effect was blocked by 600 nM cortisol (105±8% connections, p<0.05 compared with PGF_2α alone) after 5 hours.

Time-lapse imaging of HUVECs cultured on Matrigel revealed minimal cell migration or proliferation, with TLS formation consisting mainly of cell stretching to facilitate connection between adjacent cells. TLS generation began almost immediately after seeding cells on Matrigel, and peaked after approximately 8 hours (Figure 4A–4B), followed by degradation and detachment. Most cells had detached after 24 hours. Cortisol (600 nM) reduced TLS formation (Figure 4B), whereas VEGF (10 ng/ml) stimulated TLS formation and this effect was blocked by cortisol (600 nM). These observations were confirmed by comparison of the time taken to achieve maximum tube formation (Figure 4C).

The impact of exposure to cortisol on cell viability was tested by measuring ATP production. Exposure to VEGF increased ATP production by HUVECs. ATP production was inhibited in a concentration-dependent manner by SU5416 (1 nM-1 µM) but not by cortisol (3 nM–1 µM) (Figure 5A). Cell proliferation was measured using BrdU incorporation. Exposure to cortisol (3 nM-1 µM) did not alter the VEGF-induced increase in BrdU incorporation whereas SU5416 (1 nM–1 µM) produced a concentration-dependent inhibition (Figure 5B). Migration of HUVECs was assessed using a modified Boyden chamber assay. The EC₅₀ (10 ng/ml) for VEGF-stimulated HUVEC migration was established in pilot investigations (not shown). VEGF-induced migration was abolished by exposure to SU5416 whereas cortisol (600 nM) had no effect on either basal or VEGF-induced migration (Figure 5C).

QrtPCR measurements indicated gene expression varied with time during TLS development but that cortisol did not alter mRNA levels of VEGF (Figure 6A) or VEGF-R2 (Figure 6B). Similarly cortisol had no effect on Dll-4, notch-1, α₆-integrin or caveolin-1 (not shown). In contrast cortisol produced a significant (p<0.05) 2.3-fold increase in thromobospondin-1 mRNA level in TLSs after (but not 1, 4 or 22) 8 hours incubation (Figure 6C).