Central role of the proteasome in senescence and survival of human fibroblasts: induction of a senescence-like phenotype upon its inhibition and resistance to stress upon its activation.

Normal human fibroblasts undergo a limited number of divisions in culture and progressively they reach a state of irreversible growth arrest, a process termed as replicative senescence. The proteasome is the major cellular proteolytic machinery, the function of which is impaired during replicative senescence. However, the exact causes of its malfunction in these conditions are unknown. Using WI38 fibroblasts as a model for cellular senescence we have observed reduced levels of proteasomal peptidase activities coupled with increased levels of both oxidized and ubiquitinated proteins in senescent cells. We have found the catalytic subunits of the 20 S complex and subunits of the 19 S regulatory complex to be down-regulated in senescent cells. This is accompanied by a decrease in the level of both 20 S and 26 S complexes. Partial inhibition of proteasomes in young cells caused by treatment with specific inhibitors induced a senescence-like phenotype, thus demonstrating the fundamental importance of the proteasome for retaining cellular maintenance and homeostasis. Stable overexpression of beta1 and beta5 subunits in WI38 established cell lines was shown to induce elevated expression levels of beta1 subunit in beta5 transfectants and vice versa. Transfectants possess increased proteasome activities and most importantly, increased capacity to cope better with various stresses. In summary these data demonstrate the central role of the proteasome during cellular senescence and survival as well as provide insights toward a better understanding of proteasome regulation.

Normal human fibroblasts undergo a limited number of divisions in culture and progressively they reach a state of irreversible growth arrest, a process termed as replicative senescence. The proteasome is the major cellular proteolytic machinery, the function of which is impaired during replicative senescence. However, the exact causes of its malfunction in these conditions are unknown. Using WI38 fibroblasts as a model for cellular senescence we have observed reduced levels of proteasomal peptidase activities coupled with increased levels of both oxidized and ubiquitinated proteins in senescent cells. We have found the catalytic subunits of the 20 S complex and subunits of the 19 S regulatory complex to be down-regulated in senescent cells. This is accompanied by a decrease in the level of both 20 S and 26 S complexes. Partial inhibition of proteasomes in young cells caused by treatment with specific inhibitors induced a senescence-like phenotype, thus demonstrating the fundamental importance of the proteasome for retaining cellular maintenance and homeostasis. Stable overexpression of ␤ 1 and ␤ 5 subunits in WI38 established cell lines was shown to induce elevated expression levels of ␤ 1 subunit in ␤ 5 transfectants and vice versa. Transfectants possess increased proteasome activities and most importantly, increased capacity to cope better with various stresses. In summary these data demonstrate the central role of the proteasome during cellular senescence and survival as well as provide insights toward a better understanding of proteasome regulation.
Normal human fibroblasts undergo a limited number of divisions in culture and progressively they reach a state of irreversible growth arrest, a process termed as replicative senescence. Senescent fibroblasts are viable and fully functional, however, they exhibit several morphological and biochemical changes as compared with their young/proliferating counterparts (reviewed in Ref. 1). They have an irregular shape, do not line up in parallel arrays, have larger nuclei, shorter telomeres, altered gene expression, and accumulate damaged proteins. Based on the observed differences between young and senescent cells several theories have been developed to explain the aging process. It has been proposed that there is an intrinsic genetic program that regulates the phenomenon; in contrast it has been also suggested that senescence (and aging) results from the accumulation of deleterious changes over time. There is experimental evidence partially supporting each statement. For example, the telomere shortening hypothesis along with the observation that some senescence-related genes can induce senescence when overexpressed in primary proliferating cells (e.g. p21 and p16) or delay it (e.g. telomerase), indicates that senescence lies on a genetic background. However, it is also established that exposure to various cytotoxic environmental factors (such as oxidants) leads to accumulation of damage, failure of homeostasis, and senescence (reviewed in Refs. 2 and 3). Therefore more detailed and comprehensive studies are needed to unravel fully the exact cause(s) of this intriguing and inevitable biological phenomenon.
Although often neglected, protein degradation is a major intracellular function, which is not only responsible for housekeeping but also for the regulation of important cellular functions, such as homeostasis and survival. The proteasome is a non-lysosomal threonine protease. It is responsible for the degradation of many intracellular proteins, including abnormal, misfolded, denatured, or otherwise damaged proteins (reviewed in Ref. 4) as well as perfectly healthy proteins, which have to be removed for normal cellular functioning (reviewed in Ref. 5). The 20 S proteasome is a 700-kDa protease composed of 28 subunits arranged as a barrel-shaped structure of four heptameric rings. The two outside rings contain one copy each of seven different but related ␣-type subunits. Likewise, both inside rings contain one copy each of seven different but related ␤-subunits where the catalytic sites are localized. The proteasome possesses multiple endopeptidase activities including chymotrypsin-like (CT-L), 1 PGPH, and T-L (reviewed in Refs. 6 -8). In mammalian cells, the 20 S core complex can be flanked at both ends by 19 S regulatory complexes or by 11 S (PA28) complexes, giving rise to the 26 S proteasome and the PA28 -20 S proteasome complexes, respectively (reviewed in Refs. 9 and 10). The 26 S proteasome is in charge of the ATP-dependent degradation of ubiquitinated proteins (reviewed in Ref. 11).
Aging is accompanied by abnormalities in intracellular proteolysis (reviewed in Refs. 5, 12, and 13). We and others have shown that the proteolytic activities of the 20 S proteasome are remarkably reduced upon aging of several human tissues as well as in senescent primary cultures (14 -21). These findings indicate that during aging there is an impaired function of the proteasome, which in turn may result in the accumulation of oxidized proteins (reviewed in Refs. 22 and 23). Limited data exist with regard to the understanding of the molecular basis of such events. Using oligonucleotide array techniques, the expression of several proteasome subunits has been found to decline with age in: human dermal fibroblasts (24), the skeletal muscle of the C57BL/6 mouse (25), and the epididymis of the Brown Norway rat (26). Similarly, decreased proteasome quantity has been observed in senescence MRC5 human embryonic fibroblasts (17) as well as in skin fibroblasts and keratinocytes derived from aged donors (14,16,18); in contrast, unchangeable proteasome content has been reported in few other studies (19 -21). However, in all these preliminary studies only one or two representative subunits of the 20 S complex were analyzed. These fragmented data suggest that aging has profound effects on the proteasome, which in turn may influence its proteolytic activities.
In this study, we have taken a detailed molecular approach regarding the role of the proteasome in replicative senescence and survival of human fibroblasts. We first demonstrate that senescent cells exhibit reduced levels of all examined proteasomal activities that are accompanied by lower proteasome content and protein expression levels of some but not all proteasome subunits. We demonstrate that when the proteasome is inhibited partially a senescence-like phenotype is triggered. Finally, we investigate the effects of stable overexpression of ␤ 1 or ␤ 5 subunits in cellular survival to proteasome inhibitors and oxidative stress. This is the first report of enhanced cellular survival that is attributed to a better function of the proteasome because of the overexpression of one of its subunits. Accordingly, these results strongly suggest a central role of the proteasome during cellular senescence and survival in response to stress in human fibroblasts.
Cell Lines and Culture Conditions-Human embryonic fibroblasts WI38, WI38/T, IMR90, and MRC5 were obtained from the European Collection of Cell Cultures and were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (v/v), 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM glutamine, and 1% non-essential amino acids (complete medium). WI38 fibroblasts were seeded at a density of 2 ϫ 10 5 cells/75-cm 2 flask (unless otherwise noted), were subcultured at a split ratio 1:2 when cells reached confluence, until they entered senescence at about 40 cumulative population doublings. In all experimental procedures described below early passage (young; cumulative population doublings Ͻ25) and late passage (senescent; cumulative population doublings Ͼ39) WI38 cultures were used. Cells were fed ϳ16 h prior to the assay and cell number was determined in duplicates using a Coulter Z2 counter (Coulter Corp.).
Immunofluorescence Antigen Staining and Confocal Laser Scanning Microscope Analysis-For immunofluorescence labeling, cells grown on coverslips were washed with cold PBS and were subsequently fixed with 4% freshly prepared paraformaldehyde in PBS followed by cell permeabilization with 0.2% Triton X-100 in PBS; fixation with Ϫ20°C cold MetOH yielded similar results. Immunolabeling of proteasomes was carried out using antibodies against ␣ 6 and ␣ 7 subunits. The antibodies were diluted in PBS containing 0.1% Tween 20 and 3% bovine serum albumin (blocking buffer); the secondary anti-mouse IgG/ fluorescein isothiocyanate-conjugated antibody was diluted 1:300 in blocking buffer. Images of the mounted coverslips were taken using a Leica TCS SP-1 confocal laser scanning microscope. Routine procedures, applied as controls to demonstrate the specificity of the antibody used, were: (a) the usage of normal serum instead of the reactive antibody and (b) omission of the first antibody. All controls appeared free of any immunofluorescence background.
Stable Transfections-Expression vectors encoding for the fulllength ␤ 1 subunit and ␤ 5 subunit cDNAs (plasmids pREP7-hygro.␤ 1 and pBJ1-neo.␤ 5 , respectively) were the generous gifts of Drs. K. Tanaka and K. Rock (30,31). The WI38/T cells were transfected with either plasmids or empty vectors by using the electroporation method. Briefly, 10 7 cells were mixed with 50 g of DNA and were electroporated at 260 V, 960 microfarads (Gene Pulser TM , Bio-Rad). Transfected cells were split 48 h later and were maintained in complete medium containing 400 g/ml G418 (for pBJ1-neo and pBJ1-neo.␤ 5 plasmids) or 100 g/ml hygromycin-B (for pREP7-hygro and pREP7-hygro.␤ 1 plasmids). Colonies of stable transfectants were isolated after 3-4 weeks of selection and propagated into cell lines. In all cases, transfectants were tested for the appropriate protein expression levels by immunoblot analysis.
Clonogenic Assays-10 4 cells (WI38/T cell lines stably transfected with empty vectors, ␤ 1 and ␤ 5 ) were seeded in 6-well plates in duplicates. After 24 h, cells were treated with 100 nM epoxomicin, 10 M MG132, or 300 M H 2 O 2 for 2.5 h in fresh medium. Cultures were then washed thoroughly and maintained in complete medium until they formed colonies (about 10 days), which were scored by Crystal Violet staining. In parallel, the number of treated cells after the same recovery period was estimated in duplicate experiments. Empty vectors (pREP7hygro and pBJ1-neo) gave similar results in all assays (shown as mean values in Figures and Table I). Each experiment was performed at least 4 times. For immunoblot detection of carbonyl groups into proteins, 10 5 cells were seeded in 6-well plates in normal medium and 24 h later they were treated with 300 M H 2 O 2 for 30 min. Proteins were extracted 24 h after H 2 O 2 treatment.
Proteasome Inhibitors (Epoxomicin and MG132) Treatment-2 ϫ 10 5 cells were seeded in normal medium and 24 h later they were treated in triplicates with epoxomicin or MG132 (dissolved in Me 2 SO), or an equal volume of the solvent (control cultures) in the presence of normal medium. Constant treatment with epoxomicin, a more potent and specific inhibitor of the proteasome, was performed as follows: fresh medium containing 20 nM epoxomicin was added in cell cultures every 24 h continuously for 4 days (epoxomicin was replaced every day to exclude the possibility of its inactivation at 37°C). Repetitive treatment with MG132 was performed as follows: 10 M MG132 was added in cell cultures for 2 h. Cells were washed with PBS, left to recover for 22 h, and treatment was repeated 3 additional times. Cultures at this stage were named as 4 ϩ 0 days (i.e. 4 days of treatment, constantly with epoxomicin or repetitively with MG132). These cultures were re-seeded, washed thoroughly with PBS, and then maintained in normal medium for an additional period of 1 or 2 weeks. Assays were performed immediately after treatment (4 ϩ 0 days) as well as 1 (referred to as 4 ϩ 7 days) and 2 weeks (referred to as 4 ϩ 14 days) later.
Quantitative Analysis of Cellular Apoptosis-For the quantitative analysis and determination of the cytoplasmic histone-associated DNA fragments, that are indicative of on-going apoptosis, the Cell Death Detection ELISA PLUS photometric enzyme immunoassay method (Roche Diagnostics) was used. Briefly, 10 5 cells were seeded in tripli-cates in 6-well plates for 24 h and were then treated with 100 nM epoxomicin, 10 M MG132, or 300 M H 2 O 2 for 2.5 h, followed by a recovery period in complete medium for an additional 24 h. The extent of the induced apoptosis was then quantitatively measured by the ELISA PLUS death assay, according to the manufacturer's instructions. For the determination of the induced apoptosis in WI38 cells continuously treated with epoxomicin, 10 5 cells were seeded in triplicates in 6-well plates for 24 h and were then treated with 20 nM epoxomicin or the solvent (Me 2 SO, control cultures) for 4 days as described in the previous section. Triplicate sets of samples were analyzed right after treatment (4 ϩ 0 days), 1 (4 ϩ 7 days), and 2 weeks (4 ϩ 14 days) after treatment.
Cellular Proliferation-DNA Synthesis Analysis-For the determination of DNA synthesis the Cell Proliferation ELISA 5-bromo-2-deoxyuridine colorimetric immunoassay method (Roche Diagnostics) was used. Briefly, 2 ϫ 10 3 cells were seeded in 96-well plates for 24 h and were then treated with 20 nM epoxomicin or the solvent (Me 2 SO, control cultures) for 4 days as described previously. Sextuplicate sets of samples were labeled for 3 h with 5-bromo-2-deoxyuridine and DNA synthesis was analyzed right after treatment (4 ϩ 0 days), 1 (4 ϩ 7 days), and 2 weeks (4 ϩ 14 days) after treatment, according to the manufacturer's instructions.
Proteasome Peptidase Assays and Protein Determination-CT-L, PGPH, and T-L activities of the proteasome in crude extracts were assayed with the hydrolysis of fluorogenic peptides, LLVY-AMC, LLE-NA, and LSTR-MCA, respectively, for 30 min at 37°C, as described previously (16). Proteasome activity was determined as the difference between the total activity of crude extracts or fractions and the remaining activity in the presence of 20 M MG132. Assays of 26 S proteasomes were carried out in 25 mM Tris/HCl buffer, pH 7.5, containing 5 mM ATP and assays of 20 S proteasomes were performed in 50 mM Hepes buffer, pH 7.5, containing 0.02% SDS as described previously (32). Fluorescence was measured using a PerkinElmer Life Sciences 650-40 fluorescence spectrophotometer. Protein concentrations were determined using the Bradford method with bovine serum albumin as a standard.
Immunoblot Analysis-Cells were harvested at the indicated time points, lysed in non-reducing Laemmli buffer, and fractionated by SDS-PAGE (12% separating gel) according to standard procedures (33). After electrophoresis, proteins were transferred to nitrocellulose membranes for blotting with appropriate antibodies. Secondary antibodies conjugated with horseradish peroxidase and enhanced chemiluminescence were used to detect the bound primary antibodies. Immunoblot detection of carbonyl groups into proteins was performed with the OxyBlot TM protein oxidation detection kit (Intergen) according to the manufacturer's instructions. Protein loading was tested by stripping each membrane and reprobing it with ␤-actin antibody.
Preparation of Cell Extracts and Separation of Proteasome Complexes by Gel Filtration-Cells were lysed in 20 mM Tris/HCl buffer, pH 7.5, containing 5 mM ATP and 0.2% Nonidet P-40. Extracts were centrifuged at 11,500 ϫ g for 10 min at 4°C and then an equal amount of protein from early/late passage cells (0.5-1.5 mg) was fractionated by gel filtration using a Pharmacia Superose 6 FPLC column equilibrated in 20 mM Tris/HCl buffer, pH 7.5, containing 10% glycerol, 5 mM ATP, and 100 mM NaCl (34). Fractions were collected and samples were analyzed by SDS-PAGE and immunoblot analysis. The peak fractions containing 26 S and 20 S proteasomes were determined by activity assays (CT-L activity), which were carried out in duplicate.
Immunoprecipitation of Proteasomes and Two-dimensional Gel Electrophoresis-Immunoprecipitated proteasomes were prepared as follows. Cell monolayers were washed and scraped into cold PBS containing 10 mM phenylmethylsulfonyl fluoride and 10 g/ml aprotinin. For radioactive-labeled immunoprecipitation, prior to washing and scraping, cells were starved in methionine-deficient medium for 1 h, followed by pulse labeling for 4 h with 100 Ci/ml [ 35 S]methionine in methionine-deficient medium. Scraped cells were diluted directly in 20 mM Tris/HCl buffer, pH 7.5, containing 5 mM ATP, 10% glycerol, and 0.2% Nonidet P-40 containing 10 mM phenylmethylsulfonyl fluoride and 10 g/ml aprotinin (lysis buffer). Cell extracts (all equilibrated in lysis buffer) were then cleared by adding normal mouse serum and protein-A agarose beads for 3 h at 4°C. Meanwhile, protein A-agarose beads equilibrated in lysis buffer were coupled with 1-2 g of antibody against ␣ 6 subunit for 3 h at 4°C with constant rocking. The target antigen was then immunoprecipitated by adding in the pre-coupled antibody against ␣ 6 subunit-protein A the pre-cleared extracts. Binding reactions were performed overnight at 4°C with constant rocking. Immunoprecipitated protein complexes were collected, washed 4 times in 50 mM Tris/HCl buffer, pH 7.5, containing 5 mM ATP, 75 mM NaCl, 10% glycerol, and 0.2% Triton X-100 (washing buffer), and eluted from the agarose beads by boiling for 5 min in the washing buffer without NaCl. Controls used to demonstrate the specificity of the observed immunoprecipitations included the use of normal serum instead of the specific antibody. Following immunoprecipitation of proteasomes, samples were processed for one-dimensional SDS-PAGE as described above.
Two-dimensional gel electrophoresis was carried out as follows. Immunoprecipitated proteasomes (starting from 650 g of crude extracts) were diluted in sample buffer (9 M urea, 2% CHAPS, 2% Pharmalytes pH 3-10, 20 mM dithiothreitol and bromphenol blue). The first dimension was performed with Immobilines Drystrips (nonlinear, pH 3-10, length 13 cm) using the Multiphor II system (Amersham Biosciences). The Drystrip was rehydrated with the sample in a reswelling tray (Amersham Biosciences) overnight at room temperature and then focused for 50,000 volt hours (23 h). After focusing, the strips were equilibrated in 50 mM Tris/HCl buffer, pH 6.8, 6 M urea, 30% (v/v) glycerol, 1% (w/v) SDS (equilibration buffer) supplemented with 1% dithiothreitol for 15 min and in equilibration buffer containing 2.5% (w/v) iodoacetamide and traces of bromphenol blue for 15 min. The second dimension was performed on a 12% SDS-PAGE using the Protean II system (Bio-Rad). Gels were fixed, amplified, dried, and directly exposed to low energy screen of STORM 860 phosphorimager. Gels were analyzed by the "STORM 860" and "Image Master 2D Elite analysis" softwares (Amersham Biosciences).
Statistical Analysis-Statistical calculations and graphs were performed with the Microsoft Excel software. Data were analyzed by analysis of variance single factor and p value was used to determine the level of significance (p Ͻ 0.05). All values were reported as mean (the average of three independent experiments) Ϯ S.E., unless otherwise indicated.

Replicative Senescence Is Accompanied by a Decrease of Proteasome Activities and the Expression of Certain Subunits-
Initially, we aimed to reveal the changes of the overall proteasome status and function in early (young) and late (senescent) passage WI38 fibroblasts. As shown in Fig. 1A, all three major proteasomal activities (CT-L, PGPH, and T-L) were found to be significantly decreased in late passage cells as compared with their young counterparts. The activities were reduced by 2-4fold in late passage cells with the most affected being the PGPH activity.
It is well documented that oxidized proteins are degraded by the proteasome. Therefore a decrease of proteasomal activities would be expected to lead to increased levels of oxidized proteins. As shown in Fig. 1B (left panel) we observed accumulation of oxidized proteins in senescent cells consistent with the reduction in proteasome activities observed in these cells. We have also investigated the ability of senescent cells to ubiquitinate proteins for degradation by the proteasome. As shown in Fig. 1B (right panel), we observed increased amounts of ubiquitinated proteins in senescent cells.
We next investigated the loss of proteasome function in senescent fibroblasts. Therefore we examined whether there is a quantitative difference of the proteasome content between early and late passage WI38 cells. A detailed immunoblot analysis of several representative proteasomal subunits, of both the 20 S proteasome and the 19 S complex, was performed. We started by examining the expression levels of the catalytic ␤-subunits ␤ 1 , ␤ 2 , and ␤ 5 that contain the catalytic centers of PGPH, T-L, and CT-L activities of the proteasome, respectively. All three subunits were found to have decreased levels in late passage WI38 cells (Fig. 1C) compared with the early passage cells. As ␤ 1 , ␤ 2 , and ␤ 5 subunits can be replaced by ␤ 1i , ␤ 2i , and ␤ 5i , respectively, in immunoproteasomes (35) we explored the possibility of subunit substitution in senescent cells. ␤ 5i and ␤ 2i were found equally expressed in early and late passage WI38 cells, whereas ␤ 1i was not detected in either cultures (data not shown). Therefore these data suggest that the decreased levels of ␤ 1 , ␤ 2 , and ␤ 5 subunits are not compensated by ␤ 1i , ␤ 2i , and ␤ 5i subunits in senescent WI38 fibroblasts. To clarify whether there was a general decrease of the proteasomal subunits with senescence we continued the analysis by examining several other ␣and ␤-subunits of the 20 S core. A representative analysis for ␤ 7 , ␣ 4 , ␣ 6 , and ␣ 7 is shown in Fig. 1C. No significant differences in the expression levels were observed between early and late passage WI38 cells for any of these subunits.
Because the 20 S core is flanked with the 19 S regulatory complexes to form the 26 S proteasome, representative subunits of this complex were also examined. All the subunits that were analyzed (S4, S5a, S6a, S6b, S8, and S14; see Fig. 1C) were found to be down-regulated in late passage cells. In conclusion, taking together all of the data, in late passage WI38 cells there is a decrease of all three proteasome activities that is accompanied by accumulation of oxidized and ubiquitinated proteins. Moreover, senescent cells exhibit reduced expression levels of the catalytic centers of the 20 S proteasome and subunits of the 19 S regulatory complex.
To investigate further the down-regulation of specific proteasome subunits in senescent cells, next we examined the overall amount of newly synthesized proteasome in both early and late passage WI38 cells. For these experiments, immunoprecipitated radiolabeled proteasomes were analyzed by two-dimensional gel electrophoresis. Such an analysis is shown in Fig. 2, A and B. It is first worth mentioning that with the same amount of total protein, the radiolabeled starting and immu-noprecipitated material in late passage cells accounted for ϳ50 and 20% of those obtained from early passage cells, respectively. This reveals a different rate of protein synthesis between early and late passage WI38 cells, a known feature of replicative senescence (reviewed in Ref. 1), but also an even higher decrease of newly synthesized proteasome in late passage cells. Specifically, it was estimated by "Image Master 2DElite analysis" of two-dimensional gels that both 20 S and 26 S newly synthesized proteasomes were reduced approximately to less than 15% in senescent cells as compared with the young ones. Proteasome content in both early and late passage cells was further investigated by "cold" immunoprecipitation followed by immunoblot analysis. As shown in Fig. 2C, for ␣ 7 and ␤ 1 subunits, the detected amount of both subunits was decreased in the senescent cell extracts, thus indicating an overall lower amount of assembled proteasomes in late passage cells. Taking together these data show both a decreased rate of proteasome subunit synthesis as well as a decrease of assembled proteasome content in late passage cells. These observations are in agreement with the decreased proteasome activities reported in Fig. 1A.
We investigated further the proteasome complexes in early and late passage cells after separation by gel filtration. In accordance with previously reported data a decrease in both 26 S and 20 S proteasome complexes was observed in senescent cells (data not shown). Some significant differences were found in the immunoblots for the late passage fractionated cell extracts. These showed an accumulation of some low molecular weight cross-reacting material (free subunits?) for ␣-type proteasome subunits ␣ 6 and ␣ 7 in the late passage cells (Fig. 2E) that was not present in early passage cells (Fig. 2D). Although further analysis is needed, this intriguing observation could indicate that these subunits may not integrate efficiently in assembled proteasomes in senescent cells.

Partial Proteasomal Inhibition by Specific Proteasome Inhibitors Elicits a Senescence-like Phenotype in Early Passage WI38
Fibroblasts-The previous results have revealed that the proteasome is highly affected during replicative senescence. Therefore, we investigated the effect of proteasomal inhibition in early passage WI38 cells. Two proteasomal inhibitors were used, namely epoxomicin (an irreversible inhibitor) and MG132 (a reversible inhibitor), in independent experiments. It is known that total inhibition of the proteasome with specific inhibitors for long periods triggers apoptosis (36). However, there are no reports regarding the effect of these reagents in primary human fibroblasts. Therefore, in pilot experiments, we examined the time-and concentration-effect of the referred inhibitors in early passage WI38 fibroblasts. We tested different ranges of concentration of epoxomicin (1 nM to 1 M) and MG132 (100 nM to 50 M) for various periods of treatment (1 h to 4 days). In accordance with earlier reports, apoptosis was evident in high epoxomicin (1 M) and MG132 (Ͼ10 M) concentrations added constantly in medium for 24 h (data not shown). However, when cells were treated with lower concentrations we observed mainly cytostatic and to a lesser extent cytotoxic effects. Treatment with a low dose of epoxomicin or high dose of MG132 once for a short period of time (i.e. up to 2 h) resulted in a reversible growth arrest as cells recovered fully and started proliferating again soon after the withdrawal of the inhibitor from the tissue culture medium. However, when cells were left in the presence of the same concentration of the inhibitor for longer periods or repetitively (i.e. more than one treatment), an irreversible growth arrest was observed. An example of the effect of a low dose of epoxomicin (20 nM for 4 days; for conditions see "Experimental Procedures") in WI38 cells is shown in Fig. 3A (left panel). Specifically, whereas control cells (Me 2 SO treated) were increased from 2.37 ϫ 10 6 Ϯ 0.09 ϫ 10 6 (cells) right after treatment (CON/4 ϩ 0 days) to 56.65 ϫ 10 6 Ϯ 0.19 ϫ 10 6 (cells) after 2 weeks of recovery (CON/4 ϩ 14 days; ϳ24-fold increase), epoxomicin-treated cells were about half as much as the control cells right after treatment (1.27 ϫ 10 6 Ϯ 0.02 ϫ 10 6 cells at INH/4 ϩ 0 days). This difference accounts both for some cells that died by apoptosis but mainly for cells that ceased to proliferate (see below). Importantly, no further cell decrease was observed as assayed at INH/4 ϩ 7 and INH/4 ϩ 14 days (Fig. 3A, left panel, gray  bars). Instead, these cells were viable and they were unable to divide even when supplemented with fresh, inhibitor-free medium.
First we examined the characteristics of epoxomicin-treated WI38 cells. As shown in Fig. 3B (left panel), limited apoptosis is triggered during the 4 days of treatment (INH/4 ϩ 0 days); however, there is no further ongoing apoptosis 1 (INH/4 ϩ 7) and 2 (INH/4 ϩ 14) weeks after treatment, in accordance with the steady cell numbers counted in these conditions (see Fig.  3A, left panel). It is worth mentioning that WI38 cells are relatively resistant to apoptosis (enrichment factor of apoptosis was arbitrary set to 1; see comparative apoptosis analysis between WI38 and WI38/T cells in Fig. 5) and, therefore, the exhibited 1.8 times of induction of apoptosis in epoximicintreated cells (INH/4 ϩ 0 days) as compared with control cells (CON/4 ϩ 0 days) represents a small fraction of treated cells.
Because epoxomicin-treated cells were unable to divide, next we examined whether these cells show the characteristic features of senescence. As shown in Fig. 3B (right panel), treated cells right after treatment (INH/4 ϩ 0 days) exhibited significantly reduced DNA synthesis levels as compared with the control cells (CON/4 ϩ 0 days). This rate could not be restored 1 (INH/4 ϩ 7 days) and 2 weeks (INH/4 ϩ 14 days) after treatment (gray bars), in contrast to the control cells that exhibited the expected levels of DNA synthesis throughout the experiment (black bars). In addition, epoxomicin-treated cells exhibited a morphology typical of senescent cells, thus, they were enlarged, they did not line up in parallel arrays, they had larger nuclei (an example can be seen in Fig. 4C), and they were positive to the senescence biomarker ␤-galactosidase (Fig.  3C). At protein level, treated cells were also found to overexpress several biomarkers of senescence. As it is shown in Fig.  3D the epoxomicin-treated cells exhibit elevated protein levels of both forms of ApoJ (37) as well as the cyclin-dependent kinase inhibitors p21 (38) and p16 (39). Finally, the status of ]methionine, and proteasomes were immunoprecipitated using a mouse monoclonal antibody against ␣ 6 subunit as described under "Experimental Procedures." Immunoprecipitations were carried out using the same amount of extracts (650 g) from early and late passage cells; equal numbers of counts of the immunoprecipitated proteasomes were loaded on to the gels. C, immunoblot analysis of cold immunoprecipitated proteasomes from extracts of early and late passage WI38 cells. Proteasomes were immunoprecipitated from cell extracts as described above. ␣ 7 and ␤ 1 subunits were detected in the immunoprecipitates after SDS-PAGE followed by immunoblot with antibodies against those subunits. Immunoblot analysis of gel filtration fractions from early (D) and late (E) passage WI38 cells with antibodies against ␣ 6 and ␣ 7 subunits. oxidized proteins was analyzed and as expected, the treated cells had a high amount of oxidized proteins as opposed to the control cultures (Fig. 3D).
Next, we have addressed the issue whether induction of a senescence-like phenotype by epoxomicin is a feature common to other primary human fibroblasts cell lines. Therefore we have repeated epoxomicin treatment in IMR90 and MRC5 cells. As shown in Fig. 3A epoxomicin treatment resulted in the induction of irreversible growth arrest in both IMR90 (right upper panel) and MRC5 (right lower panel) cells in a manner similar to WI38 cells (left panel). Moreover, the earlier mentioned biomarkers of senescence were also tested in MRC5 and IMR90 cells and showed expression patterns similar to the ones found in WI38 cells (data not shown). Finally, a similar induction of a senescence-like phenotype was observed after delivering 10 M MG132 for 2 h for 4 continuous days in WI38 cells (data not shown), thus excluding the possibility of a specific effect of epoxomicin itself. Taking together all these data (i.e. same effect of (a) epoxomicin treatment in three different primary human fibroblast cell lines and (b) delivery of two different proteasome inhibitors in WI38 cells) we exclude the possibilities either of a cell line-specific induction of a senescence-like phenotype or of a selection of a cellular subpopulations with possible unknown phenotypic traits that could account for a different survivability. Thus partial proteasomal inhibition by specific inhibitors elicits a senescence-like phenotype in early passage fibroblasts.
We have also examined the proteasome function per se in WI38-treated cells. First, we determined the levels of the three proteolytic activities, CT-L, PGPH, and T-L, in these cells. As shown in Fig. 4A for epoxomicin-treated cells all three activities were found to be altered. As expected, both CT-L and T-L activity were immediately reduced after treatment (INH/4 ϩ 0 days) because this inhibitor targets the catalytic center of these activities. Similar low levels of CT-L activity were recorded even 2 weeks after treatment (INH/4 ϩ 14 days). Regarding T-L activity, we observed a slight increase in cultures being left to recover for 2 weeks after treatment (INH/4 ϩ 14 days). As expected, PGPH activity was not immediately affected, because epoxomicin does not primarily block this activity, but as cultures progressed into an irreversible growth arrest state it was eventually reduced to low levels. Neither the expression of biochemical markers of senescence nor the three major proteasomal activities showed any significant changes between CON/4 ϩ 0 days and CON/4 ϩ 14 days.
We next addressed the issue whether the inhibition of the proteasome in epoxomicin-treated cells could result in changes in the expression levels of proteasome subunits, by triggering a possible autoregulatory mechanism. Therefore, we examined the protein expression levels of representative subunits of the 20 S proteasome (␤ 2 and ␤ 5 catalytic subunits) as well as of the 19 S regulatory complex (S6b), as these subunits were found to be down-regulated during replicative senescence (Fig. 1C). No significant differences in the expression levels of these subunits between the control cultures (CON/4 ϩ 0 days) and the epoxomicin-treated cells (INH/4 ϩ 0, 4 ϩ 7, 4 ϩ 14 days) were observed (Fig. 4B).
Finally, to further investigate the morphology of early, late passage, and epoxomicin-treated early passage cells (INH/4 ϩ 14 days) as well as the distribution of the proteasome in these cells, we immunolocalized representative proteasome subunits. Examples for ␣ 6 and ␣ 7 subunits are shown in Fig. 4C. In accordance with previously reported data, epoxomicin-treated early passage cells exhibited identical morphology to the late passage cells. In addition, in all cases antigens were mainly found to be distributed in the nucleoplasm, whereas the nucle-olus appeared free of these proteasome subunits. A dispersed punctuate pattern was also evident in the cytoplasm. However, as the earlier presented immunoblot analysis of gel filtration fractions (see Fig. 2, D and E) indicates the possible presence of free ␣ 6 and ␣ 7 subunits in late passage cells, it is not feasible to draw additional conclusions regarding either quantitative changes or differences in intracellular distribution of assembled proteasome in these cells.
Overexpression of ␤ 1 and ␤ 5 Catalytic Subunits Leads to Increased Proteasomal Activities and to Increased Capacity of the Clones to Cope Better with Stress-As ␤ 1 and ␤ 5 subunits were found to be down-regulated in late passage cells and, in addition, they constitute the catalytic centers of PGPH and CT-L activities, respectively, we decided to overexpress these two subunits in mammalian cells to check whether the proteasome can be "activated." We have decided not to employ primary WI38 fibroblasts for these experiments for several reasons. Retroviral infections in primary cells is the regular practice to test whether the studied gene immortalizes, extends lifespan, or inhibits cellular proliferation. Thus far, only infections of few nuclear oncogenes (immortalize), telomerase (extends lifespan), and some onco-suppressors (inhibit proliferation) have given these phenotypes (40). We have no reason to believe that overexpression of a single proteasome subunit will immortalize, extend lifespan, or inhibit cellular proliferation in a similar manner to oncogenes, telomerase, or onco-suppressors, respectively. Instead, we aimed at addressing the question whether overexpression of a proteasome subunit results in proteasome activation and, in turn, in increased capacity of transfected cells to cope with stress. This particular issue cannot be addressed either, by using primary cells because by the time that is needed to select stable clones, infectants have reduced proliferative capacity as they are near senescence and therefore any follow up survival assay is not feasible. In contrast established cell lines, by not suffering from limited proliferative capacity, offer a considerable advantage for performing survival assays. Therefore we have used such an established WI38 cell line, namely WI38/T cells. First we have compared the sensitivity of the two cell lines. As shown in Fig.  5 (top panel, CON) WI38/T cells are ϳ5 times more apoptosis prone than WI38 cells (because of the introduction of SV40 T Ag, reviewed in Ref. 41), as it was assayed in normal growing conditions by an ELISA cell death detection method that measures apoptosis. Treatment of both cell lines with 100 nM epoxomicin, 10 M MG132, or 300 M H 2 O 2 for 2.5 h resulted in massive cell death in WI38/T cells in contrast to WI38 cells that were, in these conditions, relatively resistant (Fig. 5, lower  panel). Furthermore, the number of WI38/T cells that survived treatment and proliferated as assayed by clonogenic assays (for conditions see "Experimental Procedures") were significantly lower than the corresponding number of WI38-treated cells (data not shown).
Having established that WI38/T cells are more responsive to a variety of cytotoxic agents than the corresponding primary cells we have transfected constructs overexpressing the ␤ 1 and ␤ 5 proteasome subunits in these cells. Several stable clones were isolated and were propagated to cell lines. Those cell lines exhibited similar growth rates and morphological characteristics with the parental cell lines even after several months of cultivation. Four representative WI38/T clones, ␤ 1 .8, ␤ 1 .13, ␤ 5 .8, and ␤ 5 .19, were chosen for further analysis. These clones exhibited a moderate overexpression of ␤ 1 and ␤ 5 subunits (see below), thus resembling the differences of expression levels observed between early and late passage cells (see Fig. 1C). As shown in Fig. 6A, for ␤ 1 .8 and ␤ 5 .19 these stable clones were found to overexpress ␤ 1 subunit 3-5-fold and ␤ 5 subunit 2-3-fold. Moreover, and in agreement with previously published similar studies, the overexpression of the ␤ 5 subunit resulted to the overexpression of ␤ 1 subunit (31); however, the inverse phenomenon was also observed, i.e. ␤ 1 transfectants exhibited elevated levels of ␤ 5 subunit.
All of the ␤ 1 and ␤ 5 stable clones tested exhibited enhanced levels of CT-L and PGPH activities as compared with vectortransfected cell lines (2.0 -2.3 fold for CT-L and 1.4 -2.3-fold for PGPH; see Fig. 6B). Transfection with either vector alone (pBJ1-neo and pREP7-hygro) had no effect on proteasome activity. These data demonstrate that increased levels of ␤ 1 and ␤ 5 subunits cause an increase in proteasome activity. Therefore we tested the survival ability of our clones when subjected to different stresses. First, WI38/T cells transfected with ␤ 1 , ␤ 5 , and each vector were exposed to high concentrations of epoxomicin (100 nM) and MG132 (10 M) for 2.5 h. Their survival capacity was then measured (by determining colony formation and number of treated cells) after a recovery period of 10 days (Fig. 7, A and B, and Table I). All of the clones exhibited higher survival rates as compared with the appropriate vector-transfected cell line. Specifically, ␤ 1 .8 and ␤ 1 .13 overexpressing cell lines exhibit 2.8-and 1.5-fold (for epoxomicin) and 6.7-and 2.8-fold (for MG132) higher proliferating ability as compared with control cells. Similarly, ␤ 5 .8 and ␤ 5 .19 overexpressing cell lines exhibited 3-(for epoxomicin) and 5-fold (for MG132) higher proliferating ability compared with control cells, respectively. When the survival rates of the stable transfectants were calculated depending on their ability to form clones (clonogenic assays) similar results were obtained: cell lines that overexpress ␤ 1 subunit exhibit 1.3-1.7 (for epoxomicin) and 1.8 -2.3fold (for MG132) higher survival rates than control cells and cell lines that overexpress ␤ 5 subunit exhibit 1.8 -2.0-fold (for epoxomicin) and 2.2-2.4-fold (for MG132) higher survival rates than control cells (Table I).
As the proteasome is in charge of proteolysis of oxidized proteins we treated our cultures once with sublethal doses of H 2 O 2 (300 M) for 2.5 h and their survival capacity was then measured as above (Fig. 7C). Again, all ␤ 1 -and ␤ 5 -transfected cell lines exhibited higher proliferating ability compared with vector-transfected cell lines ranging from 2.3-fold for ␤ 1 .8 and ␤ 1 .13 to 4.5-fold for ␤ 5 .8 and ␤ 5 .19. When clonogenic assays were performed, cell lines that overexpress ␤ 1 subunit exhibit 2.1-3.5-fold higher survival rates compared with the vectortransfected cells. Similarly, cell lines that overexpress ␤ 5 subunit exhibit 3.1-3.5-fold higher survival rates as compared with the control cells (Table I) proteins was determined in ␤ 1 and ␤ 5 clones following treatment with 300 M H 2 O 2 . As shown in Fig. 7D all tested transfectants exhibited lower amounts of oxidized proteins than vector transfectants after a recovery period from stress of 24 h. Thus we conclude that the differences in the proteolytic activities we observed in ␤ 1 and ␤ 5 overexpressing cell lines can be translated to functional differences of the proteasome because transfectants exhibit an increased capacity to cope better with various stresses.

DISCUSSION
In this article, we have shown that senescent human fibroblasts exhibit reduced proteasome activities and content that is accompanied by lower protein expression levels of the catalytic subunits of the 20 S proteasome and subunits of the 19 S regulatory complex. We have found that when the proteasome is partially inhibited in early passage cells, a senescence-like phenotype is triggered. We also provide evidence that stable overexpression of either ␤ 1 or ␤ 5 catalytic subunits results in increased proteasome activity and enhanced cellular survival to proteasome inhibitors and oxidative stress.
A number of studies have demonstrated that impairment of proteasome function is associated with cellular senescence; however, the available data are fragmented and contradictory (15)(16)(17)(18)(19)(20)(21)42). To understand the involvement of the proteasome, we have taken a detailed molecular and biochemical approach of WI38 fibroblasts undergoing replicative senescence. It is clear that all three examined activities of the proteasome are reduced in senescent cells. Accordingly, we have observed increased amounts of both oxidized and ubiquitinated proteins in these cells. Therefore we believe that the observed accumula-tion of oxidized proteins reflects both their increased rate of production as cells progress into senescence as well as the decreased ability of the proteasome to degrade them efficiently. Similarly, as the 26 S complex of the proteasome has impaired proteolytic activity in senescence, the increased levels of ubiquitinated proteins may simply reflect queuing of "labeled for degradation proteins" derived from a functional ubiquitination system. In support of this, Taylor's group (43) has shown that upon oxidative stress and aging in the rat lens, the levels of ubiquitin mRNA, high molecular weight ubiquitin aggregates, and activity of the ubiquitination enzymes, E1 and E2, increase. Another possibility might be of a functional system of ubiquitination in senescent cells, but an impaired system of protein recognition before (i.e. no ubiquitination of the appropriate proteins) or after the ubiquitination (i.e. no entrance of the targeted proteins in the proteasomal cavity). Given the importance of this cellular system, further studies are needed to elucidate the function of the ubiquitination machinery during replicative senescence.
Can the reduced proteasomal activities of senescent fibroblasts be related to decreased proteasome content in these cells? Several lines of evidence support this hypothesis. The radiolabeled two-dimensional gel electrophoresis analysis indicates a decreased rate of synthesis of proteasome subunits. The cold immunoprecipitation/immunoblot analysis proposes the presence of less assembled proteasome in senescent cells as compared with young ones. The gel filtration data indicate the existence of reduced 26 S and 20 S proteasome complexes in these cells. In agreement with these findings, the immunoblot analysis has shown that the protein expression levels of the catalytic subunits of the 20 S complex and ATPases of the 19 S regulatory complex to be down-regulated in senescence. Interestingly, we did not observe a change in the expression levels of other subunits, for example, some ␣-subunits of the 20 S complex that can accumulate in lower molecular weight forms in late passage cells. The work of Keller et al. (42) provides strong supporting evidence as it reports differences of both ␣and ␤-20 S proteasome subunits during the aging process of the spinal cord. In addition, Lee et al. (25), using oligonucleotide array techniques for gene expression profiling, have found that the expression of some but not all proteasome subunits decline with aging in the skeletal muscle of mice, including ␤ 2 and S6a subunits. Ly et al. (24), using a similar experimental approach, found age-related alterations in expression of several proteasome subunits in human dermal fibroblasts. This intriguing observation may suggest the possibility of a rate-limiting expression of proteasome subunit(s) in senescence.
A major finding of this study is that partial inhibition of the proteasome induces a senescence-like phenotype in early passage human fibroblasts. The treated young proliferating cells adapted these senescence-like features in a short period of time as a consequence of perturbation of their proteasomal pathway. To our knowledge this is the first report that demonstrates that cells treated constantly with low doses of epoxomicin or high but chronically limited doses of MG132 induce an irreversible growth arrest state, in contrast to the known initiation of apoptosis at higher doses (36). Epoxomicin is a highly specific and irreversible inhibitor of the proteasome. It covalently modifies four catalytic subunits of the 20 S complex, namely ␤ 5 , ␤ 5i , ␤ 2 , and ␤ 2i , resulting in an inhibition of primarily the CT-L and T-L activities and to a lesser extent the PGPH activity (44). The observed slight increase of T-L in cultures being left to recover for 2 weeks after treatment (INH/4 ϩ 14 days; Fig. 4A) could be attributed to a possible activation of other protease(s) possess-ing a T-L activity, in accordance with findings by Glas et al. (45), which they demonstrate that when proteasome function is lost, another group of proteases compensate for this loss. MG132 (Z-Leu-Leu-Leu-H) is a leupeptin analogue. It is a reversible inhibitor of the proteasome and it inhibits primarily the CT-L activity and to a lesser extent the PGPH activity (46). FIG. 6. Expression of ␤ 1 and ␤ 5 proteasome subunits and proteasome activities in stable transfected WI38/T cells. A, immunoblot analysis of ␤ 1 and ␤ 5 subunits in representative stable clones that overexpress ␤ 1 (␤ 1 .8 clone) and ␤ 5 (␤ 5 .19 clone) subunits as compared with vector-transfected cells. Equal protein loading was verified by stripping the membrane and reprobing with a ␤-actin antibody (43 kDa). Lower panels show up-regulation of ␤ 5 and ␤ 1 subunits in stable clones that overexpress ␤ 1 and ␤ 5 subunits, respectively. B, proteasome activities were measured as described under "Experimental Procedures" and were expressed as % of CT-L and PGPH activities of ␤ 1 , ␤ 5 , and vector-transfected clones. Mean value of activities in WI38/T vector-transfected cells was set to 100%. CT-L and PGPH activities are enhanced significantly (p Ͻ 0.05) in clones overexpressing either ␤ 1 or ␤ 5 subunits. Few other reagents are known to induce a senescence-like phenotype when delivered into primary proliferating mammalian cells. These include introduction by retroviral infection of the ras oncogene (47), treatment with H 2 O 2 (48), or inhibition of the phosphatidylinositol 3-kinase pathway (49). However, irreversibly growth-arrested cells because of the effect of these reagents do not exhibit all features of "normal replicative senescence." For instance, H 2 O 2 -treated cells have long telomeres (48) and ras-induced senescent cells overexpress a set of genes that do not associate with normal in vitro aging (50). Similarly, our preliminary analysis indicates that proteasome inhibitortreated cells exhibit similar levels of expression of ␤ 2 , ␤ 5 , and S6b subunits as compared with non-treated cells. We were not surprised with this finding as: (a) there is no evidence that these inhibitors affect the transcription or the translation of these proteasomal subunits and (b) not every biochemical feature of replicative senescence should also be a feature of a senescence-like phenotype being induced by various non-related reagents or experimental conditions.
We have shown enhanced function of the proteasome after stable transfections of either ␤ 1 or ␤ 5 catalytic subunits. It is worth noting that overexpression of either two subunits resulted in the overexpression of the other subunit. In accordance with our data, Gaczynska et al. (31) have also shown that ␤ 5 -transfected HeLa cells exhibit ␤ 1 elevated levels. Preliminary results from both two-dimensional gel electrophoresis analysis and subunit expression levels of ␤ 1 and ␤ 5 stable cell lines suggest activation of several other proteasomal subunits. 2 These data support the argument of a possible common regulation of these subunits and, in addition, may also imply the existence of an autoregulatory mechanism. A common transcriptional regulation of the proteasomal subunits has been reported in yeast (51)(52)(53). 26 of the 32 proteasomal subunit genes have been found to be preceded in their promoters by proteasome-associated control element. RPN4 has been found to be the factor that binds on this element and consequently is a transcriptional activator of these genes (51). However, no human homologue of RPN4 has been identified thus far. Wojcik and DeMartino (54) have recently applied the use of RNAi technology in S2 cells derived from Drosophila. Different double stranded RNAs were used for several proteasomal subunits. Each double stranded RNA greatly reduced the mRNA level of its respective targeted subunit and, moreover, most of them also significantly increased the mRNA levels of several others, but not all, non-targeted subunits. However, limited data exist regarding experimentation in human cells. Davies' work (reviewed in Ref. 4) indirectly implies that few proteasomal subunits may be regulated in the same way. Seven days of daily treatment with a ␣ 6 antisense oligonucleotide severely depressed the intracellular levels of several, but not all, protea-some subunits in both cultured liver epithelial cells and in K562 human hemetopoitic cells. In conclusion, although one could not rule out the possibility of a transcriptional co-regulation of proteasome subunits in human cells, by analogy to RPN4 in yeast, further studies are needed to unravel fully this intriguing issue.
Our stable clones exhibit elevated levels of proteasome activities that are accompanied by increased capacity to cope with various stresses. To our knowledge this is the first report that demonstrates that proteasome activation by overexpression of one of its catalytic subunits results in increased cell survival to both proteasome inhibitors and to oxidative stress. Gaczynska et al. (31) have also transfected ␤ 1 and ␤ 5 subunits in HeLa cells. Their clones have elevated levels of some but not all proteasomal activities (␤ 1 subunit stimulated PGPH activity without altering other peptidase activities, whereas ␤ 5 subunit reduced CT-L and T-L activities). The discrepancy regarding the proteasomal activities in ␤ 1 -and ␤ 5 -transfected HeLa and WI38/T cells is possibly because of the different cellular type used. There are several cases in the literature where stable overexpression of the same gene construct has different phenotypes when introduced into different cell lines. For instance, introduction of the catalytic subunit of telomerase efficiently extends the lifespan in WI38 cells but it does very poorly in MRC5 cells (55).
Accumulation of abnormal proteins is determined by their rates of formation, but of equal importance are their rates of hydrolysis and elimination. Therefore it is reasonable to expect that cells possessing elevated proteasome activity, like our WI38/T ␤ 1 and ␤ 5 clones, to confer enhanced protection to oxidants like H 2 O 2 . We have previously studied proteasome activity in fibroblasts derived from "control" donors of different ages (18 to 80 years old), as well as from healthy centenarians, because these individuals represent the best model of successful aging (reviewed in Ref. 56). Although we observed a decreased activity of the proteasome with increasing age, importantly, analysis of RNA and protein levels of several proteasome subunits, determination of CT-L and PGPH activities, and levels of oxidized proteins, has revealed that healthy centenarians have an active proteasome as compared with control donors (16). Thus we have concluded that cells possessing a functional proteasome may be equipped with an extra anti-aging arsenal. In support to our belief, it has been suggested by Davies and colleagues (4) that the ability of the proteasome to degrade oxidized proteins serves as a secondary cellular antioxidant defense system. Primary human diploid fibroblasts reach senescence because of both genetic and stochastic factors. Telomere shortening and changes in expression of genes involved in cell cycle checkpoints (e.g. p21 and p16) represent some examples of the genetic background that underlies senescence. Agents that induce failure of cellular homeostasis (like oxidants) belong to the even bigger category of those hundred stochastic factors that are capable of inducing senescence on top of the genetic background. We believe that the impaired function of the proteasome during replicative senescence is linked with both genetic and stochastic factors. The observed down-regulation of the ␤-catalytic subunits of the 20 S proteasome and subunits of the 19 S regulatory complex should be related to the genetic background that underlies senescence, at least in the case of WI38 fibroblasts. Identifying the pathways of regulation of these subunits in mammalian cells, similarly to the RPN4 pathway in yeast, will be the subject of future studies toward the genetic manipulation of the function of this multicatalytic enzyme. The decreased proteasome activity in senescent cells is likely to be due not only to the down-regulation of specific subunits but also 2 N. Chondrogianni and E. S. Gonos, unpublished data. mainly to the accumulation of oxidized and other damaged forms of proteins. Anti-aging strategies should be aimed based on the activation of the proteasomal proteolytic capacity or its maintenance. In any case it is expected that the regulation and the function of the proteasome will become a key issue in future aging research.