Advanced glycation of the Arg-Gly-Asp (RGD) tripeptide motif modulates retinal microvascular endothelial cell dysfunction

Preparation of AGE-modified fibronectin

Unless otherwise stated, all reagents were purchased from the Sigma Chemical Co. Ltd., (Poole, Dorset, UK). AGE-modified FN (AGE-FN) was prepared as previously described [19,22]. Briefly, four-well chamber slides were coated with FN at 2 μg/cm² and incubated with MGO at 10, 50, or 100 μM or with PBS (137 mM NaCl, 2.7 mM KCl, 100 mM Na₂HPO₄, 2 mM KH₂PO₄) alone for one week at 37 °C. Following incubation, slides were extensively washed with PBS to remove MGO and unbound adducts. The levels of the MGO-derived AGE adducts argpyrimidine and CEL were assayed by HPLC and competitive AGE-ELISA, respectively, and we have already reported this [19].

Retinal microvascular endothelial cells attachment and spreading

Retinal microvascular endothelial cells (RMECs) were isolated from bovine retina by established protocols [23]. Briefly, bovine eyes were transported from a local abbatoir on ice and the retinas removed and washed free of RPE in Dulbecco's minimal essential medium (MEM; Invitrogen Life Technologies, Paisley, UK). The neural retina was then homogenized in MEM and filtered through an 87 µm filter. The trapped microvessels were digested at 37 °C for approximately 20–30 min in PBS containing 200 µg/ml pronase, 200 µg/ml DNAase, and 50 µg/ml collagenase. The filtrate was microscopically examined to determine the end point for maximum retrival of endothelial cells. Vessel fragments were then trapped in a 53 µm filter and resuspended in Dulbecco’s modified Eagles medium (DMEM) containing antibiotics (0.2 mg streptomycin sulfate, 0.12 mg benzyl penicillin and 0.2 mg kanamycin), fungizone (2.5 µg/ml) and 15% fetal calf serum (FCS; Invitrogen Life Technologies). This mixture was seeded into 25 cm² Falcon flasks and maintained at 37 °C in a mixture of 5% CO₂ and air. Endothelial cell growth was supported by DMEM containing 7.5% human platelet-poor plasma-derived serum, 5 µg/ml insulin and antibiotics. All subsequent experiments were performed on confluent monolayers of cells derived from passages 1 to 4 and plated out onto multiwell dishes. RMEC attachment was visualized in vitro by phase-contrast microscopy and quantified using manual cell counts. RMECs at a cell density of 0.9×10⁴ were seeded onto 3 cm² Petri dishes (Nunc Plasticware, Rochester, NY) that had been coated with FN or AGE-FN. After 3 or 6 h, the cells were washed twice with fresh PBS to dislodge nonadherent cells after which the adherent cells were fixed in 4% paraformaldehyde (PFA) for 1 h at room temperature. Fixation was followed by washing in PBS. The number of adherent cells was counted in six separate fields of view with the aid of an eyepiece graticule.

For assessment of RMEC spreading, 0.9×10⁴ cells were allowed to attach and spread on FN or AGE-FN for 6 h at 37 °C. Following incubation, cells were visualized and designated as either “spread” or “not spread.” Spread cells were defined as those that achieved normal cell morphology similar to that found in sparse culture, while not spread cells were morphologically typified by a rounded up, phase-bright appearance. The numbers of spread and not spread cells were counted in six separate fields of view and the percentage of spread cells determined.

To assess the impact of adding “exogenous” RGD peptide, we added the Arg-Gly-Asp-Ser (RGDS) peptide and a scrambled control peptide (Ser-Asp-Gly-Arg-Gly (SDGRG)) to substrates. SDGRG does not stimulate integrins and is often used as an appropriate control peptide for RGD [24]. The peptides were added to the FN or AGE-FN substrates on 3 cm² Petri dishes at concentrations of 1, 3, or 6 mM. The peptide was allowed to dry onto the substrate after which the cells were added as we have described.

Integrin signaling

Cells were exposed to FN or AGE-FN as outlined. Following the treatment period, attached cells were then lysed on the culture dishes using lysis buffer that consisted of double distilled water containing 10 mM Tris Ph 7.2, 158 mM NaCl, 1 mM Na₃VO₄, 1 mM EDTA, 0.1% SDS, 1% sodium deoxycholate, 1% Triton X-100, as well as protease inhibitors (1X Complete^TM mini protease inhibitor cocktail; Boehringer Mannheim, Mannheim, Germany). Protein was extracted on ice and quantification of total protein (µg/µl) determined using a bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL).

An aliquot of each extracted protein sample was diluted in Laemmli buffer (containing 4% SDS; 20% glycerol; 10% 2-mercaptoethanol; 0.004% bromphenol blue in 0.125 M Tris HCl). Protein from each sample and 10 µl ProSieve^® molecular weight marker (Cambrex Bioscience Inc., Rockland, ME) were loaded on a denaturing 4%–20% Tris-HEPES-SDS gel (Pierce) and separated by molecular weight using sodium dodecyl sulfate PAGE (SDS–PAGE). Following SDS–PAGE, proteins were transferred onto a nitrocellulose membrane (Pall Life Sciences, Portsmouth, UK). After transfer, all washes were performed with PBS containing 0.1% Tween-20. The blocking buffer and antibody diluents contained 0.1% Tween-20 and 5% Blotto™ non-fat dry milk (Santa Cruz Biotechnology, Santa Cruz, CA). After blocking, the membrane was incubated overnight in 1:500 dilution of polyclonal primary antibody (anti-phospho-FAK (tyrosine³⁹⁷; Upstate Biotechnology). After washing at room temperature for 4×10 min, the membrane was incubated for 1 h in 1:1,000 dilution of horseradish peroxidase (HRP) labeled secondary antibody (Santa Cruz Biotechnology). HRP was detected with SuperSignal^® West Pico Chemiluminescent Substrate (Pierce) using a UVP^® AutoChemi gel analysis system equipped with Labworks^™ 4.0 image acquisition and analysis software (UVP. Inc., Upland, CA). To calculate relative expression, we reprobed membranes with a primary monoclonal antibody to β-actin. Densitometry software (Labworks^™ 4.0, UVP. Inc., Upland) was used to calculate expression of each protein sample relative to the density of β-actin.

Confocal microscopy for visualization of integrin α5β1

α5β1 integrin was detected using an immunofluorescence protocol. RMECs growing on FN or AGE-FN within four-well chamber slides were fixed in 4% PFA and subsequently permeabilized with a PBS solution containing 0.1% Triton X. To reduce nonspecific binding, we incubated cells with 5% normal goat serum (NGS) overnight at 4 °C. The RMECs were then incubated with a monoclonal anti-integrin α5β1 antibody (clone HA5; Chemicon, Chandlers Ford, Hampshire, UK). This antibody has been previously used to localize this integrin combination in several cell types [25]. The antibody was diluted in PBS containing 5% NGS overnight at 4 °C after which they were washed with PBS and blocked for a further hour before incubation with a 1:1,000 dilution of Alexa-488 fluorophore labeled secondary antibody (anti-mouse Alexa-488; Invitrogen Life Technologies). After extensive washing, the cells were then incubated with 5 µg/ml propidium iodide (PI) for 20 min at room temperature to stain cell nuclei. Cells were washed extensively in PBS, mounted in Vectashield (Vector Laboratories, Peterborough, UK) and examined with a Bio-Rad Microradiance confocal scanning laser microscope (CSLM) equipped with a 25 mV Argon and 1 mV green helium-neon laser. Negative controls, in which primary antibody was omitted, ensured that the staining achieved was specific for α5β1 integrin.

RMEC mitochondrial permeability and caspase-3 activity

Mitochondrial changes were evaluated in pre-apoptotic RMECs growing on FN or differentially modified AGE-FN (using 10, 50, and 100 uM MGO). This change was evaluated using the cationic fluorescent probe 5,5′,6,6’-tetrachloro-1,1’,3,3′-tetraethylbenzimidazolylcarbocyanie (JC-1) that assesses voltage changes in membrane potential and thus mitochondrial membrane permeability. Healthy mitochondria have the ability to concentrate the green monomeric JC-1 species (emission peak 530 nm) to yield red fluorescent J aggregates (emission peak 590 nm). Dysfunctional mitochondria, however, are unable to concentrate the monomer, and there is a net decrease in the red:green fluorescence intensity ratio. Changes in the membrane transmembrane potential RMECs using JC-1 was quantified using flow cytometry as previously described [16]. Briefly, RMECs used for these studies were cultured for 3 h on FN-coated 75 cm² flasks that had been modified using 10, 50, and 100 uM MGO. Following the treatment period, JC-1 dye at a final concentration of 5 ug/ml was added to the DMEM for 30 min at 37 °C in the dark. Adherent cells were then detached via brief exposure to trypsin and versene, pelleted by light centrifugation, and resuspended in PBS at a population density of 1×10⁶/ml. Once the cells were in suspension, the JC-1 cells were immediately sorted, to exclude cell debris or doublets, using a FACSCaliber flow cytometer (Becton Dickinson, Oxford, UK) equipped with a 488 nm argon laser. Each sample was examined and gated in a forward scatter versus a side scatter dot plot. The population of RMECs showing mitochondrial permeability was determined by dividing the mean fluorescence value obtained from red fluorescence channel by the mean fluorescence value obtained from the green fluorescence channel. This allowed calculation of the red:green fluorescence intensity ratio.

As a complementary apoptosis assay, caspase-3 activity was examined using an Enzcheck^® Caspase-3 Assay kit (Molecular Probes, Invotrogen Life Technologies) in accordance with the manufacturer’s instructions. RMECs were cultured for 3 h on FN- or AGE-FN-coated 75 cm² flasks (with or without RGD peptide coating) after which the cells were washed in PBS. Adherent cells were then detached via brief exposure to trypsin and versene, pelleted, and subsequently re-suspended in PBS at a density of 1×10⁶/ml. Next, 50 µl of 1X cell lysis buffer was added to each sample for 30 min, and the samples were centrifuged at 2.5 xg to pellet cellular debris. Afterwards, 40 µl of the supernatant from each sample was transferred into individual wells of a 96 well plate (Nunc), and 50 µl of the 1X cell lysis buffer was used as a negative control to determine substrate fluorescence. Next, 40 µl of 2X substrate working solution was added to each sample followed by incubation at room temperature for 30 min. The fluorescence of each sample (excitation=342 nm; emission=441 nm) was measured on a spectrofluorimeter (Tecan Safire, Tecan, Männedorf, Switzerland).

Superoxide assay

Released superoxide (O₂^-) was measured from RMECs growing on native FN or AGE-FN with an enhanced chemiluminescence detection using lucigenin as a substrate [26]. RMECs growing on FN or AGE-FN were trypsinized. Next, 0.4×10⁴ cells from each group were added to polycarbonate luminometer tubes that contained Tyrodes-HEPES buffer. The buffer comprised the following: 140 mmol/l NaCl, 6.0 mmol/l HEPES, 2.0 mmol/l Na₂HPO₄, 2.0 mmol/l MgSO₄, and 5.6 mmol/l dextrose (pH 7.40). Ten µl lucigenin was added and each sample dark adapted at room temperature for 10 min after which they were transferred to the tube housing of a Sirius luminometer (Berthold Detection Systems, Bleichstr, Germany). Luminescence was recorded in both control and AGE-FN-exposed cells in the both presence and absence of β-nicotinamide adenine dinucleotide 3′-phosphate (NADPH) and polyethylene-glycol-superoxide dismutase (PEG-SOD). Peak emission of luminescence was used to construct bar charts.

In a parallel assay, dihydroethidium (DHE) was used to assess O₂^- in AGE-exposed RMECs. Cytosolic DHE exhibits blue fluorescence; however, once this probe is oxidized by O₂^- to 2-hyroethidium (2-HE) and ethidium it fluoresces red [27]. 2-HE is a specific O₂^--derived product and is completely inhibited by PEG-SOD. RMECs were plated onto FN-coated and AGE-FN-coated coverslips in six-well dishes (Nunc) and cultured for 24 h in 20% growth medium. After 24 h, the medium was gently aspirated, washed thrice with PBS and replaced with growth medium containing 2% serum for further 24 h incubation. For DHE staining, cells were incubated for 30 min in either Krebs-HEPES buffer, which contained 99 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO₄, 1.0 mM KH₂PO₄, 1.9 mM CaCl_2, 25 mM NaHCO_3, 11.1 mM glucose, and 20 mM HEPES pH 7.3, or 100 units/ml PEG-SOD that was reconstituted in Krebs-HEPES buffer. After this initial incubation, the relevant buffer was aspirated from the cells and quickly replaced with 100 µl Krebs-HEPES buffer (unstimulated) or stimulated with 100 µl 10 mM NADPH reconstituted in Krebs-HEPES. Immediately after addition of NADPH, cells were incubated in 20 µM DHE for 10 min and subsequently subjected to confocal scanning laser microscopy. The settings remained constant throughout the course of the assay. When microscopy was completed, fluorescence intensity for each sample was analyzed using Lucia G/F (Version 47) image analysis software (Nikon, Kingston upon Thames UK Ltd).

Quantitative RT–PCR

RMECs propagated on FN or AGE-FN were harvested and subjected to total RNA extraction using the RNeasy mini kit (Qiagen, West Sussex, UK) as per the manufacturer’s instructions. The amount of RNA was quantified using a Nano drop, ND-1000 spectrophotometer (NanodropTechnologies, Wilmington, DE) by measuring absorbance at 260/280 nm. cDNA was prepared from 1 μg equivalent of the total RNA using the Sensiscript Reverse Transcription (RT) kit (Qiagen) as per the manufacturer’s instructions.

cDNA polymerase chain reaction (PCR) primers were designed using Informax’s sequence analysis and primer design software, Vector NTI (Invitrogen™ Life Technologies) from bovine DNA and RNA sequences obtained from The National Center for Biotechnology Information (NCBI, Bethesda, MD; Table 1).

Quantitative RT–PCR was performed using a LightCycler rapid thermal cycler system (Roche Diagnostics Ltd, Lewes, UK). Reactions were performed in a 10 μl volume containing 1 μl cDNA template, 0.5 μM sense and anti-sense primers, deoxyribonucleotide triphosphates (dNTPs), Taq DNA polymerase, buffer and SYBR Green master mix (Qiagen). A typical protocol included an initial denaturation step at 95 °C for 15 min followed by amplification of the template for 36 cycles with 95 °C denaturation for 15 s, 53 °C annealing for 15 s, and 72 °C extension for 10 s. The annealing temperature was optimized for each primer set, and the extension period depended on the length of the expected product (approximately 5 s/100 base pairs, minimum 10 s). Labeled product was detected at the end of the 72 °C extension period. To confirm amplification specificity, we subjected the PCR products from each primer to a melting curve analysis and for new primer pairs, subsequent electrophoresis on a 2% agarose gel. A Lightcycler quantification report was used to obtain average crossover values. Data was expressed in terms of ΔΔ^ct (relative gene expression crossover value in comparison to housekeeping mRNA) values in comparison to the control group.

Statistical analysis

Data were expressed as the mean values ±standard error of the mean (SEM). Statistical differences in the mean were assessed using one-way ANOVA followed by the Tukey-Kramer post-hoc test for multiple comparisons unless stated specifically. All statistical analyses were performed using SPSS 14.0, (SPSS, Chicago, IL) or GraphPad InStat 3.0, (GraphPad Software, San Diego, CA). Data was considered significant at 95% (p<0.05).

RMEC attachment and spreading

RMECs seeded onto AGE-FN, which had been increasingly modified with 10, 50, and 100 µM MGO, for 3 and 6 h showed a significant step-wise reduction in attachment capacity in comparison to cells grown on native FN (p<0.001; data not shown). Using this data as a foundation, RMECs were again seeded onto AGE-FN (modified with 100 µM MGO) that had been treated with RGDS peptides or control SDGRG peptides (Figure 1) to determine if replacing the RGD motif could replenish AGE-derived loss of attachment signals. The cells demonstrated a significant reduction in attachment capacity on AGE-FN when compared to cells cultured on native FN, while supplementation with 3 mM and 6 mM RGDS peptide significantly enhanced cell attachment (p<0.01–0.001; Figure 1). The scrambled peptide had no notable effect (Figure 1). A similar response was observed for RMEC spreading responses (Figure 1).

Since integrins are central to endothelial cell matrix interactions, downstream signaling responses were evaluated. RMEC signaling subsequent to attachment on FN and AGE-FN substrates were evaluated by western analysis. The levels of phospho-FAK (tyrosine³⁹⁷) were reduced when RMECs were grown on AGE-FN when compared to FN control (Figure 2A). This was especially evident at 45 min and 60 min (Figure 2A).

FAK signaling requires clustering of integrins on the plasma membrane. Investigation of integrin α5β1 immunoreactivity showed that RMECs on FN had a relatively uniform immunofluorescence immediately adjacent to the basal surface (Figure 2B). By contrast, RMECs on AGE-FN exhibited a more intense immunofluorescence for α5β1 at the basal plasma membrane and also in perinuclear organelles (compare Figure 2B,C)

RMEC mitochondrial function and caspase-3 activity

Appropriate integrin-substrate interactions are known to evoke pro-survival pathways, therefore mitochondrial permeability, as an early indicator of apoptosis, was assessed. Inner mitochondrial membrane potential was evaluated by flow cytometry using JC-1, which exhibits dual emission properties depending on mitochondrial membrane potential. As an entire population, cells grown on FN showed little mitochondrial permeability, as indicated by a constitutive red:green fluorescence intensity ratio that was significantly shifted by exposure to AGE-FN (p<0.05; Figure 3). It was apparent that AGE-exposed cells showed a degree of heterogeneity suggesting that, at this time point, some cells were more prone to mitochondrial changes than others (Figure 3).

Caspase-3 activity in RMECs was analyzed by spectrofluorometry. This approach demonstrated that cells exposed to AGE-FN had significantly increased activity of this pro-apoptotic enzyme. Scrambled peptide plated onto the AGE-modified substrate had no influence, but RGDS peptide significantly, albeit incompletely, reduced caspase-3 activity in RMECs growing on AGE-FN (*p<0.05; Figure 4).

Superoxide production and mRNA expression changes in RMECs

O₂^- generation is not only linked to many AGE-mediated cell responses [28] but is also central to endothelial cell apoptosis [29]. Following on from the attachment studies it was observed that a subset of RMEC remained attached and spread onto the AGE-modified substrate. This “surviving population” of RMECs was investigated for in situ O₂^- production using the lucigenin-enhanced chemiluminescence assay. AGE-exposed cells demonstrated a significant increase in O₂^- when compared to controls (p<0.01; Figure 5A,B). RMECs from both substrate groups were stimulated with NADPH. This resulted in marked O₂^- production, although the AGE-exposed cells showed significantly more than RMECs growing on native FN (Figure 5C,D; **p<0.01).

As a complement to the lucigenin approach for measuring O₂^- production, an assay based on DHE fluorescence was used. RMECs again showed an AGE-mediated increase in O₂^- (Figure 6). RMECs growing on native FN demonstrated basal levels of O₂^-, but this was significantly increased when the cells were treated with NADPH (p<0.001). O₂^- was attenuated with co-incubation with PEG-SOD (p<0.001; Figure 6). RMECs growing on AGE-FN produced greater levels of O₂^- when compared to FN controls, and this response was significantly enhanced by NADPH exposure (p<0.01). As observed with RMECs growing on FN, PEG-SOD treatment significantly reduced NADPH-mediated O₂^- production (p<0.01; Figure 6).

Previous studies have shown that AGE-modification of the BM can increase expression of the component proteins at both the protein and mRNA level, and this may contribute significantly to the BM thickening lesion [30]. In the current in vitro system it was anticipated that endothelial cells respond to an inadequate substrate by increasing synthesis of matrix proteins and enzymes that re-model the BM. Using this as a rationale, we assessed mRNA expression of the BM protein collagen IV, matrix metalloproteinases (MMP), and their regulators. Quantitative RT–PCR demonstrated that RMEC exposure to AGE-FN induced significant mRNA upregulation of collagen IV when compared to cells growing on native-FN (p<0.05; Figure 7). The proteases MMP-1, MMP-2, and MMP-14 and also the protease inhibitors tissue inhibitor of metalloproteinases-1 (TIMP-1) and TIMP-2 were also upregulated in RMECs growing on AGE-FN when compared to native FN (p<0.05; Figure 7).