Profile of Macrophages in Human Abdominal Aortic Aneurysms: A Transcriptomic, Proteomic, and Antibody Protein Array Study Nicolas Lamblin,†,‡,§,| Philippe Ratajczak,†,‡,§ David Hot,‡,⊥,# Emilie Dubois,†,‡,§ Maggy Chwastyniak,†,‡,§ Olivia Beseme,†,‡,§ Herve´ Drobecq,‡,§,# Yves Lemoine,‡,⊥,#,∇ Mohammad Koussa,| Philippe Amouyel,†,‡,§,| and Florence Pinet*,†,‡,§,| INSERM U744, Lille, France, Institut Pasteur de Lille, Lille, France, University of Nord de France, USDL, IFR 141, Lille, France, Centre Hospitalier Re´gional et Universitaire de Lille, France, Laboratoire d’Etudes Transcriptomiques et Ge´nomiques Applique´es, Lille, France, UMR CNRS 8161, Lille, France, and University of Nord de France, USTL, Villeneuve d’Ascq, France Received March 18, 2010
Abdominal aortic aneurysms (AAA) are defined by an increased aortic diameter and characterized by impairment of the extracellular matrix, macrophages infiltration and decreased density of smooth muscle cells. Our aim is to identify the key molecules involved in the pathogenesis of AAAs. This study investigated transcriptomic and proteomic profiles of macrophages from AAA patients (>50 mm aortic diameter) (n ) 24) and peripheral arterial occlusion (PAO) patients without AAA detected (n ) 18), who both needed a surgery. An antibody protein microarray, generated by printing antibodies onto membranes against proteins selected from the transcriptomic and proteomic analysis, was performed to validate the proteins differentially expressed specifically in macrophages and plasma from the same patients. We found a restricted number of proteins differentially expressed between AAA and PAO patients: TIMP-3, ADAMTS5, and ADAMTS8 that differ significantly in plasma of AAA patients compared to PAO patients, as found in the macrophages. In contrast to plasma MMP-9, soluble glycoprotein V (sGPV) and plasmin-antiplasmin complex levels, plasma TIMP-3 levels were not correlated to AAA size but interestingly correlated to sGPV, a platelet activation marker. Combining transcriptomic and proteomic is a valid approach to identify diseases causing proteins and potential biomarkers. Keywords: abdominal aortic aneurysm • humans • macrophages • plasma • microarray • 2D-DIGE • antibody array • proteases • antiproteases
Introduction Abdominal aortic aneurysms (AAAs) are a potentially fatal disorder. They affect 2-9% of the population older than 65 years and are a frequent cause of morbidity and mortality.1 Incidence of AAA is increasing due to general aging of the population and an increase in the amount of screening programs.2 Accelerated degradation of collagen and elastin, the main components of the extracellular matrix (ECM) in the vascular wall, cause the development of AAAs in deteriorating aortic walls that dilate progressively and may eventually rupture.3 Furthermore, AAA formation and rupture are closely accompanied by inflammation and neovascularization.4,5 Other alterations associated with AAAs include marked changes in the cellular composition of the aortic wall, especially the * Corresponding author: Dr. Florence Pinet, INSERM U744-IPL, 1 rue du professeur Calmette, 59019 Lille cedex, France. Tel: (33) 3 20 87 72 15. Fax: (33) 3 20 87 78 94. E-mail:
[email protected]. † INSERM U744. ‡ Institut Pasteur de Lille. § University of Nord de France, USDL. | Centre Hospitalier Re´gional et Universitaire de Lille. ⊥ Laboratoire d’Etudes Transcriptomiques et Ge´nomiques Applique´es. # UMR CNRS 8161. ∇ University of Nord de France, USTL.
3720 Journal of Proteome Research 2010, 9, 3720–3729 Published on Web 06/01/2010
infiltration of macrophages and T-lymphocytes into the adventitia6 and a major reduction in the population of vascular smooth muscle cells (SMCs).7 The inflammatory cells have a role in AAA pathogenesis through secretion and activation of matrix metalloproteinases (MMPs), which degrade the ECM and weaken the aortic wall.8 Inflammatory infiltrates and invading neovessels are relevant sources of MMPs in the AAA wall and may substantially contribute to aneurismal wall instability.9 Macrophages have been shown to express all MMPs; however, differences in their expression were detected.10 This study investigated, for the first time, the transcriptomic and proteomic profiles of macrophages to determine the genes and proteins involved in aneurysmal disease. We compared two groups of patients in LILAS (the Lille Anevrysmal Study, described in Materials and Methods), those with AAAs and others with peripheral arterial occlusion (PAO). We choose PAO disease as control diseases as Thompson et al.11 described the separation of AAA disease from other atherosclerotic diseases like peripheral vascular disease and carotid disease. The transcriptomic analysis was performed using a microarray containing 137 genes of MMPs, ADAMs (A disintegrin and metalloproteinases), ADAMTS (A disintegrin and metallopro10.1021/pr100250s
2010 American Chemical Society
Protease and Antiprotease Expression in Human Aneurysmal Disease teinase with thrombospondin motifs), TIMPs (tissue inhibitor of metalloproteinase), and other proteases. The second approach, using the 2D-DIGE technology as previously described12 focused on proteome analysis of cytosolic proteins isolated from macrophages of the same patients. To validate our screening approach, we have constructed an antibody protein array that contained on one slide 42 different antibodies representing RNA and proteins found to be differentially modulated in macrophages from AAA and PAO patients. The last step was the analysis of plasma obtained from the same patients to select proteins involved in the aneurysmal process that are easily detectable, and then confirming the data using specific assays against the selected proteins, as hypothesized that circulating biomarkers can reflect inflammation and degeneration in the AAA wall.13
Materials and Methods Population Study. The Lille Anevrysmal Study (LILAS) was a case-control study that enrolled 42 men, either AAA (case patients) or PAO (control patients) who needed a vascular surgery or endovascular treatment at the Lille University Hospital Centre (Lille, France) between September 2003 and July 2005. Case patients were considered eligible if the AAA diameter, measured by abdominal ultrasound, exceeded 50 mm or if it had increased more than 10 mm during the past 6 months and required initial surgical treatment. Control patients were considered eligible if PAO was diagnosed in the aorta, iliac, or femoral arteries, if AAAs had been ruled out. This determination was based on symptoms and medical imaging (Doppler ultrasound, MRI angiography, and/or arteriography) without presence of AAA. All patients were men older than 55 years-old. Exclusion criteria were previous surgery for AAA or PAO or an infectious, traumatic, or connective-issue diseaseassociated cause (Marfan, Takayasu or Behcet syndromes) of the aneurysm. The ethics committee of the Lille University Hospital Center (France) approved the study, and each patient provided written informed consent. The protocol required a blood sample (80 mL) to be taken at inclusion, before the surgery. Clinical data items are presented as mean ( SD and frequencies with proportions as appropriate. For continuous variables, differences between groups were assessed with the nonparametric Mann-Whitney t test. For discrete variables, differences between groups were assessed with Fisher’s exact test. A p-value 9 was analyzed. For each patient, mRNA from 5 µg of total RNA was labeled by Cy5 or Cy3 fluorophores with the Agilent Fluorescent Linear Amplification kit as described by the supplier. Each sample was analyzed with two slides according to a dye-swap strategy to account for labeling and detection differences between Cy5 and Cy3. Labeled cDNAs obtained from patient RNA and control RNA were mixed together and hybridization buffer was added to a final concentration of 40% formamide, 2.5× Denhardt’s, 0.5% SDS, and 4× SSC (sodium saline citrate). Hybridization was performed under a coverslip in a hybridization chamber (CMT, Corning) at 42 °C for 16 h. Slides were then washed twice in 2× SSC and 0.1% SDS for 5 min at 42 °C, then once in 0.2× SSC for 1 min at room temperature. Arrays were then scanned with the Affymetrix 418 scanner and images were processed with ImaGene 6.0 software (Biodiscovery). Data analysis was performed with the statistical language R (v 2.0.1),16 more specifically with the LIMMA library (Linear Models for MicroArray data).17 A normalization protocol was applied to the local background to correct mean intensities of unflagged spots because of spatial and slide-to-slide inconsistency of local background values. The normalization protocol consisted first of a within-array global lowess normalization to correct for dye effect18 and then of between-array normalization to scale the distribution ratio values between the different slides. The microarray data are available in GEO database (GSE21803). In accordance with the MIAME (Minimum Information About a Microarray Experiment) guidelines, we note the steps involved in data processing: (1) each microarray was scanned twice (at high and low intensity) for each wavelength of Cy3 and Cy5; (2) two separate normalization steps were conducted: first, pinby-pin and second, a lowess fitness normalization. A moderated t-statistic with empirical Bayes shrinkage of the standard error was used to classify the statistically significant modulations.19 Because of multiple testing, p-values were corrected with the Benjamini and Hochberg method20 to control for the false discovery rate. Genes with an adjusted p-value
