Cell isolation, cultivation, and characterization
Human diploid fibroblasts from foreskin were seeded in 10-cm dishes at a density of 200,000 cells per dish and treated in parallel with the following compounds: untreated, vehicle treated (ethanol), oligomycin (8 μmol L−1 final concentration), FCCP (3 μmol L−1 final concentration), and AMP (3 μmol L−1 final concentration). Every day, medium was replaced and inhibitory compounds freshly added. This treatment leads to premature senescence within 14 days (data not shown, see also (Stockl et al., 2006, 2007; Zwerschke et al., 2003). RNA was prepared after 72 h of incubation and analyzed by RNA profiling as described earlier.
Human umbilical vein endothelial cells were isolated from human umbilical veins and cultured in Endothelial Cell Basal Medium (Lonza, Basel, Switzerland) supplemented with endothelial cell growth medium (EGM) Single Quots (Lonza), containing hEGF 0.5 mL, hydrocortisone 0.5 mL, GA-1000 0.5 mL, BBE 2.0 mL, and FBS 10.0 mL. The cells were subcultured by trypsination with trypsin-EDTA (Gibco Life Technologies, Vienna, Austria), seeded on cell culture dishes coated with 0.2% gelatine and grown at 37°C at ambient atmosphere containing 5% CO2. Cells were passaged at a ratio of 1:5 in regular intervals. At later passages, the splitting ratio was reduced to 1:3 and 1:2, respectively. Cells were passaged before reaching 70–80% confluence. Population doublings (PDL) were estimated using the following equation: n = (log10 F − log10 I)/0.301 (where n is the population doublings, F, number of cells at the end of one passage, and I, number of cells that were seeded at the beginning of one passage). After 50 PDL, the cells reached growth arrest, and the senescent phenotype was verified by staining for senescence-associated beta galactosidase, which was positive for ≥ 95% of cells.
Human primary PrSC were established and trans-differentiation induced by TGF-beta1 treatment, as described (Untergasser et al., 2005). Briefly, PrSC were derived from prostate cancer patients who have not received hormonal therapy (n = 3, age 65–72) after obtaining written informed consent. After radical prostatectomy and inspection by the pathologist, two tissue wedges showing no histological signs of malignancy were removed from the transition zone. These explants were minced into organoids of ∼ 1 mm3 and seeded on uncoated plastic material in stromal cell growth medium containing insulin, human basic fibroblast growth factor, 5% fetal calf serum, and gentamycin and amphotericin-B as antibiotics (SCGM; Lonza). Explants were maintained at 37°C in a humidified atmosphere of 5% CO2. These conditions produce a homogeneous fibroblast cell population after 7 days of culture. When cells reached 70% confluence, they were split at a 1:3 ratio using trypsin-EDTA to expand the population. In all experiments, cells of passage 2–4 were used directly from culture (not previously frozen).
Renal proximal tubular epithelial cells were cultivated as recently reported (Wieser et al., 2008). In brief, within 24 h after surgery, tissue from the renal cortex was fragmented and incubated at 37°C for 15–20 min in DMEM/Ham’s F12 (1:1) (Biochrom KG, Berlin, Germany) containing 1 mg mL−1 collagenase type IV (PAN-BioTech GmbH, Aidenbach, Germany) and 1 mg mL−1 trypsin-inhibitor (Sigma, Vienna, Austria). After being passed through a 105-μm nylon mesh, the filtrate was centrifuged, washed twice with phosphate-buffered saline (PBS), resuspended in medium, and dispensed into roux-flasks (Nunc, Wiesbaden, Germany). 24 hours thereafter medium was changed. The initial passage of confluent cells after 3–5 days was considered as PDL zero. Cells were passaged (1:2 to 1:4) at confluence, using 0.25% trypsin/0.02% EDTA, which was inactivated with 1 mg mL−1 trypsin-inhibitor. Cumulative PDL was calculated as a function of passage number and split ratio (4). Medium consisted of DMEM/Ham’s F12 (1:1) supplemented with 4 mm l-glutamine, 10 mm HEPES buffer, 5 pm tri-iodothyronine, 10 ng mL−1 recombinant human EGF, 3.5 μg mL−1 ascorbic acid, 5 μg mL−1 transferrin, 5 μg mL−1 insulin, 25 ng mL−1 prostaglandin E1, 25 ng mL−1 hydrocortisone, and 8.65 ng mL−1 sodium selenite (all from Sigma). After 24 PDL, the cells reached growth arrest, and the senescent phenotype was verified by staining for senescence-associated beta galactosidase, which was positive for ≥ 95% of cells. Cells at intermediate passage could be divided into two groups concerning their redox status, as monitored by dichlorofluorescein diacetate (DCFDA) staining. To address the significance of this distinction, RPTECs were sorted into DCFDAbright and DCFDAdim subpopulations, which were analyzed separately.
Mesenchymal stem cells were isolated from the iliac crest of systemically healthy individuals (young donor, age 5; elderly donor, age 56), which had been harvested for reconstructive bone surgery of defects within other areas of the body as described previously (Fehrer et al., 2007). Briefly, a small biopsy of substantia spongiosa osseum, which otherwise would have been discarded based on necessary bone for molding and re-contouring prior to insertion into the recipient site, was taken to further investigation under an Institutional Review Board-approved protocol after having obtained the parents’ and the respective patient’s written consent. After surgery, the bone was transferred into minimal essential medium (MEM) supplemented with 20% heat-inactivated fetal calf serum, 100 units mL−1 penicillin, and 100 μg mL−1 streptomycin (growth medium) for transportation from the operation theater to the clean room at room temperature. The biopsies were fragmented, and marrow cells were isolated from pieces (20–100 mm3) by centrifugation (400 g, 1 min). After centrifugation, the remaining pieces were treated with collagenase (2.5 mg mL−1 in MEM) for 2–3 h at 37°C, 20% O2, and 5% CO2. Thereafter, the specimen was again centrifuged (400 g, 1 min). Cells were resuspended and loaded on a Ficoll-Paque Plus® gradient and centrifuged at 2500 g for 30 min. Cells were harvested from the interphase (density < 1.075 g mL−1), washed, and collected by centrifugation (1500 g, 15 min). Cells were cultured at a density of 0.2–0.5 × 106 cells cm−2 at either 20% or 3% O2 in combination with 5% CO2 and 37°C (HeraCell240 – Heraeus, Thermo Scientific, Vienna, Austria; Thermo Electron Forma Series II, 3110). After 24 h, the nonadherent cell fraction was removed by washing twice with PBS. After the primary culture had reached approximately 30–50% confluence, cells were washed twice with PBS and subsequently treated with 0.05% trypsin/1 mm EDTA for 3–5 min at 37°C. Cells were harvested, washed in MEM, and further expanded at a density of 50 cells cm−2.
Isolation of CD8+CD28+ and CD8+CD28− T cells from peripheral blood of apparently healthy young (< 35 year, n = 6, mean age 29, range 26–35) and elderly (> 65 year, n = 10, mean age 72, range 66–87) donors was performed by preparing peripheral blood mononuclear cells (PBMCs) by Ficoll-Paque PLUS (Amersham Biosciences) density gradient centrifugation as approved by the Ethics committee of Innsbruck Medical University. CD8+ T cells were negatively selected from the obtained PBMC fraction by applying the magnetic separation protocol CD8+ T cell isolation kit II (depleting CD4, CD14, CD16, CD19, CD36, CD56, CD123, TCRγ/δ, and CD235a; Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions. Subsequently, purified CD8+ T cells were stained with an allophycocyanin (APC)-conjugated αCD28 monoclonal antibody (mAb) and split into CD8+CD28+ and CD8+CD28− T-cell populations using αAPC MicroBeads (Miltenyi Biotec) by passing the cell suspension through a positive selection column (LS; Miltenyi Biotec) mounted in a magnetic field. The CD8+CD28− T-cell fraction was then re-incubated with αAPC MicroBeads and run over a fresh LS-column to increase purity. For phenotypic analysis, purified T-cell fractions were labelled with a combination of mAbs (αTCRαβ-FITC, αCD16-PE, αCD4-PerCP, αCD8-PE-Cy7, αCD28-APC, and αCD3-APC-Cy7; all BD Biosciences, Heidelberg, Germany) and analyzed on a FACSCanto II (BD Biosciences) revealing that the described isolation protocol yields population homogeneities of > 95%.
Bioinformatics, clustering, and visualization of array data
Array raw data were normalized via CARMAweb (Rainer et al., 2006) using the gcrma algorithm (Wu et al., 2004). Hierarchical clustering of samples and genes (Euclidian distance, average linkage) was performed after filtering out genes with low variance on a subset of 20 000 genes with the MeV program package (Saeed et al., 2006) available online: http://www.tm4.org/mev/).
To evaluate the probability of observing an elevated number of under- or overexpressed gene occurrences, the distinct number of occurrences in which the gene is under- or overexpressed in a respective group of experiments was determined. The resulting gene list was ranked according to occurrences with respect to either age/senescence-related model systems or oxidative stress–derived data sets as well as a gene list of highly under- or overrepresented candidates present in both cases.
Statistical analysis to identify significant probe sets was performed as described by de Magalhaes et al. (2009a,b): For each probe set, a P-value was calculated with the cumulative binomial distribution (CBD):
providing the probability P for a probe set to be as often or more often differentially expressed than the times k, it was actually differentially expressed in n experiments. The threshold for differential expression of a probe set was defined as a fold change between samples ≥ ±1.5. Applying statistical analysis based on CBD, the probability P that any probe set is differentially expressed with senescence or oxidative stress was defined as the average of differentially expressed probe sets in the experimental group senescence or oxidative stress divided by the number of all probe sets per whole-genome analysis.
The q-value was calculated for each probe set using Storey’s false discovery rate (FDR) approach with the bootstrapping method (Storey et al., 2004). The robust parameter was used to make the q-values more accurate for small P-values (Storey, 2002). Statistical computation was carried out using the statistical framework R (R-Team, 2007) version 2.5.1. Bioconductor q-value package version 1.10.0 (Storey & Tibshirani, 2003) was used to calculate the FDR.
Differentially expressed genes were grouped into protein families associated with characterized pathways applying Pathway Explorer ((Mlecnik et al., 2005), available online: https://pathwayexplorer.genome.tugraz.at/), P-values were calculated from the complete expression value dataset (54 675 probe sets) with Fisher’s exact test, pathway data sources: KEGG pathway database (http://www.genome.jp/kegg/pathway.html), GenMapp (http://www.genmapp.org/), Biocarta (http://www.biocarta.com/genes/).
GiSAO.db (https://gisao.genome.tugraz.at) is a web-based database system for storing and retrieving data of genes involved in senescence, apoptosis, and oxidative stress. The application is based on a three-tier architecture consisting of a web interface, business logic, and a database. The web interface enables data input and presentation. It was implemented by using Struts framework (http://struts.apache.org/) with Java Server Pages (http://java.sun.com/products/jsp/). The business logic, which is responsible for data processing is an Enterprise JavaBeans 3 (http://java.sun.com/products/ejb/) application deployed on JBoss (http://www.jboss.org/jbossas/) application server. The data are stored in an Oracle database, a relational database management system.
The GiSAO database contains normalized gene expression values obtained from experiments evaluated with the aid of Affymetrix arrays. Gene expression values of each experiment can be displayed and compared with the gene expression values of two or more microarray experiments. Additionally, experimental data of follow-up experiments regarding candidate genes, such as qPCR or Western Blot analysis, are entered into the GiSAO.db. Furthermore, GiSAO.db contains two types of orthologue data: orthologue data provided and updated by Affymetrix from cross reference tables linking Saccharomyces cerevisiae, Caenorhabditis elegans, Drosophila melanogaster, Mus musculus and Homo sapiens facilitate comparative genomic analyses, as well as orthology data computed and entered manually. Besides mRNA expression profiles, also data from proteome analysis and further validation with respect to functional analyses together with information from public resources are available for distinct candidate genes. Moreover, external links lead to orthologs of HomoloGene (http://www.ncbi.nlm.nih.gov/homologene) and InParanoid (http://inparanoid.sbc.su.se/cgi-bin/index.cgi). The gene IDs are linked to their respective database, such as Entrez Gene or RefSeq. GiSAO.db also provides gene annotation (Gene Symbol, Gene Name, etc.) and GO terms (http://www.geneontology.org/). Finally, KEGG pathways (http://www.genome.jp/kegg/) can be displayed and data can be exported in various formats.
Lifespan analysis of gene disruption mutants of Sacharomyces cerevisiae
Experiments were carried out in BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) as the wild-type strain and respective null mutants, obtained from Euroscarf. Strains were grown at 28°C on SC medium containing 0.17% yeast nitrogen base (Difco), 0.5% (NH4)2SO4 and 30 mg L−1 of all amino acids (except 80 mg L−1 histidine and 200 mg L−1 leucine), 30 mg L−1 adenine, and 320 mg L−1 uracil with 2% glucose. For all experiments, yeast cells were grown at 28°C and 145 rpm. For chronological lifespan experiments, cultures were inoculated at an OD600 of 0.1, and aliquots were taken to perform survival plating at indicated time points.