Protein Markers of Ischemic Insult in Brain Endothelial Cells Identified Using 2D Gel Electrophoresis and ICAT-Based Quantitative Proteomics Arsalan S. Haqqani,*,† John Kelly,† Ewa Baumann,† Reiner F. Haseloff,‡ Ingolf E. Blasig,‡ and Danica B. Stanimirovic† Cerebrovascular Research and Genomics and Proteomics Groups, Institute for Biological Sciences, National Research Council of Canada, Ottawa, ON, Canada K1A 0R6, and Leibnizinstitute of Molecular Pharmacology, 13125 Berlin-Buch, Germany Received July 28, 2006
The blood-brain barrier (BBB) is formed by endothelial cells of cerebral microvessels sealed by tight junctions. Ischemic brain injury is known to initiate a series of biochemical and molecular processes that lead to the disruption of the BBB, development of vascular inflammation, and subsequent neurovascular remodeling including angiogenesis. Molecular effectors of these changes are multiple and are regulated in a dynamic fashion. The current study was designed to analyze changes in cellular and secreted proteins in rat brain endothelial cells (BEC) exposed to ischemic insult in vitro using two complementary quantitative proteomic approaches: two-dimensional gel electrophoresis (2DE) and isotope-coded affinity tag (ICAT)-based proteomics. We show a comprehensive qualitative and quantitative comparison between the two proteomic methods applied to the same experimental system with respect to their reproducibility, specificity, and the type of proteins identified. In total, >160 proteins showed differential expression in response to the ischemic insult, with 38 identified by 2DE and 138 by ICAT. Only 15 proteins were commonly identified. ICAT showed superior reproducibility over 2DE and was more suitable for detecting small, large, basic, hydrophobic, and secreted proteins than 2DE. However, positive identification of proteins by MS/MS was more reliably done using a 2DE-based method compared to ICAT. Changes in proteins involved in nucleic acid, protein, and carbohydrate metabolism, signal transduction, cell structure, adhesion and motility, immunity and defense, cell cycle, and apoptosis were observed. The functional significance of observed protein changes was evaluated through a multifaceted protein classification and validation process, which included literature mining and comparative evaluation of protein changes in analogous in vitro and in vivo ischemia models. The comparative analyses of protein changes between the in vitro and in vivo models demonstrated a significant correlative relationship, emphasizing the ‘translational’ value of in vitro endothelial models in neurovascular research. Keywords: ICAT • 2D gel electrophoresis • isotope labeling • quantitative proteomics • mass spectrometry • brain • ischemia • endothelial cells • blood-brain barrier
Introduction Brain endothelial cells (BEC) lining the cerebral capillaries and microvessels display tightly sealed junctions that restrict the free passage of nutrients, hormones, drugs, and cells into the brain. The unique phenotype of BEC underlies the bloodbrain barrier (BBB) function of cerebral vessels and is dependent upon dynamic interactions with other cellular and acellular elements of the neurovascular unit including astrocytes, peri* Corresponding author. Arsalan S. Haqqani, Ph.D, Institute for Biological Sciences, National Research Council, 100 Sussex Drive, Room 2051, Ottawa, Ontario, Canada, K1A 0R6. Phone: 613-991-5466. Fax: 613-952-9092. Email:
[email protected]. † Cerebrovascular Research and Genomics and Proteomics Groups. ‡ Leibnizinstitute of Molecular Pharmacology.
226
Journal of Proteome Research 2007, 6, 226-239
Published on Web 11/23/2006
cytes, and the basement membrane.1 Brain insults characterized by tissue hypoxia, including stroke, induce profound structural and functional changes in brain vessels, including the BBB breakdown, proinflammatory and prothrombotic activation, and angio- and vasculogenesis.1 Although vascular expression of various mediators during ischemic brain disease has been studied using ‘classical’ approaches of in situ hybridization or immunohistochemistry,2-6 the complexity of molecular changes in brain vessels in response to hypoxic or ischemic states necessitates the use of global molecular analyses enabled by genomics and proteomics technologies. Various in vitro models including cultured primary and immortalized BEC have been developed to study effects of hypoxia or oxygen-glucose deprivation.7 Studies in these 10.1021/pr0603811 CCC: $37.00
2007 American Chemical Society
Markers of Ischemia/Reperfusion in Brain Endothelial Cells
models indicated that BEC respond to hypoxic/ischemic insults by HIF-1R and NF-κB-mediated up-regulation of inflammatory cytokines and adhesion molecules6 and increases in antioxidative enzymes8 and proangiogenic mediators.9 Genomic techniques including serial analysis of gene expression and microarrays enabled cataloguing of the BBB transcriptome10 and BEC gene responses to stimulation with inflammatory cytokine TNFR.11 The only published application of proteomics to analyze BEC responses to hypoxia and posthypoxic reoxygenation12 identified changes in abundant cellular proteins including glycolytic enzymes, endoplasmic reticulum, and cytoskeletal proteins using two-dimensional gel electrophoresis (2DE) method. The current study was designed to analyze changes in cellular and secreted proteins in rat BEC exposed to ischemic insult in vitro using two complementary quantitative proteomic approaches, 2DE and isotope-coded affinity tag (ICAT)-based proteomics. The study provides comprehensive qualitative and quantitative comparison between the two proteomic methods applied to the same experimental system with respect to their reproducibility and the type of proteins identified. Although cultured BEC are a useful model for studying brain vascular responses to ischemic conditions, brain vascular responses in vivo are integrated through interactions of BEC with other cellular components of the neurovascular unit, blood elements, and blood flow,13 most of which are absent in cell culture models. Other confounding factors that may reduce the ‘translational’ value of BEC culture models include a partial loss of the BBB phenotype and cellular ‘adaptation’ to specific culturing conditions.14 Therefore, the additional goal of this study was to evaluate to what extent protein expression changes observed in BEC in vitro correlate with protein expression changes in brain vessels after an ischemic insult in vivo. This comparison was performed using our recently published data set of protein expression changes determined using ICAT-based proteomics in the laser-capture microdissection (LCM)-extracted brain vessels from rats subjected to a transient global cerebral ischemia.15 These analyses demonstrated utility of in vitro models in translational studies of the cerebrovascular system.
Experimental Section Cell Culture. Immortalized adult rat brain endothelial cells (iRBEC), established and characterized in the laboratory as previously described,16 were chosen for this study to enable the isolation of sufficient amounts of protein for gel-based proteomics. iRBEC is a stable cell line that recapitulates key features of primary rat brain endothelial cells, including the expression of endothelial-specific Factor VIII-related antigen and uptake of acetylated LDL, and characteristics of the BBBspecific phenotype, including high activity of γ-glutamyl transpeptidase and the expression of tight junction proteins and P-glycoprotein. Detailed evaluation of this cell line as an in vitro model of the BBB has recently been published.17 iRBEC were maintained in complete medium 199 (M199; Sigma) containing 10% fetal bovine serum (Multicell Premium, Wisent, Quebec, Canada), 0.5 g/L peptone, 4500 mg/L glucose, 1× basal medium Eagle (BME) amino acids, 1× BME vitamins, and 1× antibiotics (all obtained from Sigma) in a humidified atmosphere of 5% CO2/95% air at 37 °C. In Vitro Ischemia. iRBEC were exposed to a 4-h oxygen and glucose deprivation (OGD; pO2 < 2%) in the serum- and glucose-free media in an anaerobic chamber equipped with a
research articles humidified, temperature-controlled incubator as previously described.18 Cells were recovered to normoxia (ambient air) for 24 h in glucose, amino acids, and vitamins containing serumfree media. Respective controls were maintained in normoxia in the presence of glucose, amino acids, and vitamins containing serum-free media throughout the experiments. Cells were harvested as pellets and frozen at -80 °C for further analyses. Culture media containing secreted proteins were also harvested, filtered through a 0.22 µm filter, concentrated using Centriprep columns (Millipore, Quebec, Canada), precipitated in 10 vol of cold acetone, and frozen at -80 °C until further analyses. 2DE, Quantitative Spot Analysis, and Identification of Differentially Expressed Proteins. Both cellular and secreted fractions of iRBEC were dissolved in urea buffer (7 M urea, 2 M thiourea, 4% CHAPS, and 1% dithiothreitol) and shaken for 1 h at room temperature. Debris was removed by centrifugation for 10 min at 16 000g, and the supernatants were precipitated in 10 vol of cold acetone. Protein pellets were dissolved in isoelectric focusing (IEF) buffer [urea buffer containing 10% Biolytes 3-10 (Bio-Rad Lab., Hercules, CA) and trace amounts of Orange G] and quantified by Bio-Rad protein assay. Immobilized pH gradient (IPG) strips from Bio-Rad were rehydrated passively overnight in strip holders in 300 µL of IEF buffer containing 250 µg of protein. IEF runs were carried out as follows: 1 h at 200 V, 1 h at 500 V, 5 h ramp to 5000 V, and hold at 500 V. Before the second dimension, strips were equilibrated in 50 mM Tris, 6 M urea, 30% glycerol, 2% sodium dodecyl sulfate (SDS), and 1% dithiothreitol for 10 min, then in 50 mM Tris, 6 M urea, 30% glycerol, 2% SDS, and 4% iodoacetamide for 10 min. The second dimension separation was carried out using a 10% polyacryamide gel at constant 24 mA current per gel. Gels were stained with a sensitive fluorescent dye SYPRO Ruby (Bio-Rad) according to the manufacturer’s protocol, and gel images were collected using Fluor-S Multiimager (Bio-Rad). Protein spots were detected and their intensities quantified in each gel using PDQuest software (Bio-Rad). An average of 675 protein spots was detected per gel from cellular fractions separated on pH 4-7 IPG strips. For the secreted fractions, an average of 140 protein spots was detected per gel. All gels had a similar range of spot intensities (101.3-104.2 intensity units), and the number of spots per gel was comparable between control and ischemia groups. The spots were subsequently matched among multiple gels using PDQuest software: >90% of the protein spots in all samples could be matched among replicate gels, suggesting a high probability of detecting the same protein spot in multiple replicates. Unmatched spots were assigned to be down-regulated or up-regulated by 10-fold, depending on whether they were present in the control or ischemia gels, respectively. Matching was also done between the secreted protein gels and their respective cellular protein gels to determine contamination of cellular protein in the secreted fraction. Only 2.8% of the proteins with low mean intensities (2 or
