p53 Mediates the Accelerated Onset of Senescence of Endothelial Progenitor Cells in Diabetes*

Adverse metabolic factors, including oxidized small and dense low density lipoprotein (ox-dmLDL) can contribute to the reduced number and the impaired functions of circulating endothelial progenitors (EPC) in diabetic patients. To elucidate the molecular mechanisms involved, EPC from normal donors were cultured in the presence of ox-dmLDL. Under these experimental conditions EPC undergo to senescent-like growth arrest. This effect is associated with Akt activation, p21 expression, p53 accumulation, and retinoblastoma protein dephosphorylation and with a reduced protective effect against oxidative damage. Moreover, depletion of endogenous p53 expression by small interfering RNA demonstrates that the integrity of this pathway is essential for senescence to occur. Activation of the Akt/p53/p21 signaling pathway and accelerated onset of senescence are also detectable in EPC from diabetic patients. Finally, diabetic EPC depleted of endogenous p53 do not undergo to senescence-growth arrest and acquire the ability to form tube-like structures in vitro. These observations identify the activation of the p53 signaling pathway as a crucial event that can contribute to the impaired neovascularization in diabetes.

replication and control antiproliferative programs (8). However, p21 together with the tumor suppressor proteins p53 and pRb is not only an important mediator of quiescence-like growth arrest but also of senescence (9 -11). So far a large body of evidence from animal studies indicates that p21 and p53 are key modulators of neointimal lesion development by efficiently attenuating neointimal thickening (12)(13)(14)(15). Remarkably, conditional overexpression of p21 has been reported to be associated with growth arrest and phenotypic features of senescence (16). Moreover, senescent endothelial cells have been detected in atherosclerotic, but not in nonatherosclerotic, vessels (17).
Vasculature remodeling does not rely exclusively on proliferation of resident endothelial cells but also involves circulating progenitor cells (EPC) (18,19). Recent works demonstrated that in patients with cardiovascular risk factors, including diabetes, the number of EPC that can be isolated from peripheral blood is reduced, and their function is impaired (20). Moreover, despite culturing EPC under normoglycemic conditions, functional differences between EPC recovered from diabetic patients and normal subjects exist (20). Therefore the reduced number and the impaired function of EPC in diabetes can rely on adverse metabolic stress factors, different from hyperglycemia (21). Increasing evidence indicates that native oxidized LDL by inducing reactive oxygen species (ROS) production activates a cascade of molecular events that mainly contribute to accelerate atherosclerosis in diabetic patients (22). In vivo small and dense LDL, because of their qualitative abnormalities, are more susceptible than natural occurring LDL to undergo oxidation and to activate the oxidative stress pathways.
In this study we investigated the effects of in vitro oxidized diabetic LDL on EPC fate and in particular the molecular mechanisms activated by oxidized small and dense LDL which can contribute to EPC dysfunction in diabetes. We found that in vitro oxidized small and dense diabetic LDL accelerate the onset of senescence of EPC recovered from normal subjects via a p53-mediated signaling pathway. Moreover, we demonstrate that depletion of endogenous p53 in EPC recovered from diabetic patients prevents the accelerated onset of senescence, indicating that a p53-mediated pathway contributes to the impaired EPC function in this setting.
Isolation, Characterization, and Oxidation of LDL-Blood from diabetic patients or healthy controls was processed according to Redgrave and Carlson (24). Plasma was brought to a density of 1.10 g/ml and processed as described previously (7). The density of the LDL was measured in the tubes where the highest levels of cholesterol were found. LDL was dialyzed against 0.02 M EDTA-free phosphate buffer, pH 7.4, containing 0.15 M NaCl. LDL was adjusted to 0.2 mg/ml protein concentration. CuSO 4 was added for 24 h (37°C) at a ratio of 25 M/mg LDL protein (25). Oxidation was stopped by the addition of 1 mM EDTA and 0.02 mM butylated hydroxytoluene.
Isolation and Culture of EPC from Peripheral Blood Mononuclear Cells-ox-dmLDL or dmLDL studies were performed on EPC recovered from normal donors. Isolation of CD34 ϩ cells was performed with the direct CD34 or CD133 ϩ isolation kit (MINIMACS system, Miltenyi Biotec) according to the manufacturer's instruction (26). The purity of sorted cells was assessed by FACS analysis (26). The isolated EPC were cultured for 10 or 15 days on 20 g/ml fibronectin-coated dishes in EGM endothelial growth medium containing 20% fetal calf serum and 20 ng/ml VEGF alone or in combination with 10 g/ml dmLDL or with 10 g/ml ox-dmLDL. FACS was used to analyze the differentiated phenotype (anti-CD34, anti-CD31, anti-CD146, or anti-VE-cadherin antibodies were used) (26). CD34 ϩ cells were also isolated from 9 diabetic patients (HbA1c Ͻ7%) and assayed for the presence of the senescenceassociated ␤-galactosidase (SA-␤-gal) marker (27).
Silencing of Endogenous p53 and Akt by Small Interfering RNAs (siRNA)-To obtain inactivation of p53, EPC cultured with ox-dmLDL were transiently transfected by Lipofectamine PLUS TM reagent (Invitrogen) according to the vendor's instructions with the vector pSUPER retro containing p53 siRNA or a scramble p53 siRNA (control siRNA) sequences (1.5 g) as described by Brummelkamp et al. (28). The pSUPER retro containing p53 siRNA and the scramble p53 siRNA were gently provided by Dr. S. Soddu. 60 h later whole cell extracts were prepared, separated on 10% SDS-PAGE, and immunoblotted with antibody against p53. To obtain inactivation of Akt EPC cultured as above were transiently transfected with purified duplex siRNAs for Akt and for a scramble control purchased from Qiagen (Valencia, CA). Transfection was performed using Lipofectamine according to the vendor's instructions. Cell viability was evaluated at the end of the experiment.
Flow Cytometry-EPC treated as indicated were fixed with 70% ethanol. After digestion with RNase, DNA was stained with propidium iodide and analyzed with a flow cytometer (FACScan, Becton Dickinson Immunocytometry Systems, San Jose, CA). In selected experiments NAC, SB203580, or LY294002 inhibitors were used. The percentage of cells in each phase of the cell cycle was determined by ModFit LT software (Verity Software House, Inc., Topsham, ME). FACS was used for analyzing VE-cadherin, CD146, or CD34 expression. Fluorescein isothiocyanate-conjugated anti-mouse IgG (Sigma) was used as secondary antibody.
Detection of ROS-DCF-DA (20 mM final concentration) was added to EPC in the various cultured conditions. At the times indicated the cells were subjected to flow cytometric analysis and processed as described previously (7). Tumor necrosis factor-␣ was used as positive control. A lucigenin-derived chemiluminescence assay was also performed. Briefly, EPC were resuspended in a glass tube containing phenol red-free basal medium and incubated in a 37°C water bath for 10 min. 5 M lucigenin and 250 M NADPH were then added and immediately placed inside the luminometer (Lumat LB 9501, Berthold Technologies, Pforzheim, Germany). The lucigenin-derived chemiluminescence assay was measured for a 10-min period. EPC were then stimulated with saline (0.1% BSA in phosphate-buffered saline), ox-dmLDL, or dm-LDL as indicated, and LDLCL was monitored for 30 min. In a few experiments 30 mM/liter NAC, 30 M/liter DPI, or 10 M/liter LY294002 was used.
Western Blot Analysis-EPC were cultured in the presence of dmLDL or ox-dmLDL in combination with VEGF as indicated. Cells were lysed (in lysis buffer including a protease inhibitor mixture (Sigma), 200 mM/liter sodium vanadate, and 20 M/liter sodium fluoride), and the protein concentration was obtained as described previously (7,29). 50 g of proteins were separated on SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes. In selected experiments 10 M/liter SB203580, 10 M/liter LY294002, or 30 mM/liter NAC was used. To evaluate the p21 half-life EPC were treated with 10 g/ml cycloheximide for the indicated intervals. 50 g of whole cell lysates were prepared at each time point and assayed for the expression of p21 and ␤-actin by Western blotting. Reaction with antibodies and detection with an enhanced chemiluminescence detection system (Amersham Biosciences) were performed as described previously (7).
For dot blot experiments a sheet of nitrocellulose was clamped between the gasket and the 96-well sample template. 200 g of LDL sample was allowed to filter through the membrane. After the antigen was immobilized, the nitrocellulose was incubated in 4% albumin blocking solution. Nitrocellulose was subsequently incubated with rabbit IgG anti-AGE at room temperature for 1.5 h and then with goat IgG antirabbit IgG labeled with alkaline phosphatase. Dots were visualized with the procedure contained in the Immun-Blot Assay kit (Bio-Rad).
In Vitro Angiogenesis-EPC cultured with dmLDL or ox-dmLDL were prelabeled with a permanent green fluorescent dye (PKH2 from Sigma). In vitro angiogenesis was studied on Matrigel-coated surface as described by Montesano and Orci (30). VEGF was added to the medium or incorporated into the Matrigel. To evaluate cell viability, trypan blue exclusion staining was used (31). In vitro angiogenesis was expressed as the percentage Ϯ S.D. of the tube-like structure area to the total Matrigel area.
Northern Blot Analysis-Northern blot analysis was performed accordingly to standard methods (32). The filter was hybridized to 32 Prandom priming-labeled cDNA probes corresponding to cyclin D1, p21, and ␤-actin.
EPC Senescence-Senescence was evaluated on EPC recovered from normal subjects cultured for 15 days with VEGF plus dmLDL or with ox-dmLDL or on EPC recovered from diabetic patients after 10 days of culture in EGM endothelial growth medium. Acidic ␤-gal activity was used for these assays (27). Briefly, EPC were washed in phosphate-buffered saline, fixed for 3 min at room temperature in 2% paraformaldehyde, washed, and incubated for 24 h at 37°C with fresh SA-␤-gal stain solution: 1 mg/ml 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal), 5 mM/liter potassium ferrocyanide, 5 mM/liter ferricyanide, 150 mM/liter NaCl, 2 mM/liter MgCl 2 , 0.01% sodium deoxycholate, and 0.02% Nonidet P-40. SA-␤-gal-positive cells were also quantified on cells cultured in the presence of NAC or LY294002 as indicated. The number of blue cells was counted manually from a total of 200 cells.
Statistical Analysis-All in vitro results are representative of at least three independent experiments performed in triplicate. Densitometric analysis using a Bio-Rad GS 250 molecular imager was used to calculate the differences in the -fold induction of protein activation or expression (* and § p Ͻ 0.05, statistically significant between experimental and control values). Significance of differences between experimental and control values was calculated using analysis of variance with Newman-Keuls multicomparison test.

LDL Preparation and
Characterization-In all diabetic subjects the peak of cholesterol was in the LDL subfractions with a density ranging from 1.037 to 1.044 g/ml. In healthy subjects the highest level of cholesterol was found in the LDL subfractions of density ranging from 1.028 to 1.035 g/ml. In diabetic samples LDL were constantly smaller and denser than normal LDL, and this effect was confirmed by the density gradient ultracentrifugation profile (data not shown) and by a lower cholesterol: apoB ratio found in these subjects (1.6 versus 1.3).
Our previous findings that unlike 8% glycated LDL, 10 and 15% glycated LDL inhibit cell cycle progression (29) indicate that the level of glycation is crucial to elicit damaging signals in primary endothelial cells. To rule out the possibility that the biological effects exerted by oxidized diabetic LDL could rely on LDL glycation, AGE immunoreactivity on LDL was evaluated. To this end a dot blot assay using an anti-AGE antibody was performed. As shown in Fig. 1A no significant difference in AGE immunoreactivity was detected in LDL recovered from normal subject (HbA1c ϭ 5%) and diabetic patients in good metabolic control (HbA1c ϭ 7%). By contrast, AGE immunoreactivity increased in LDL recovered from patients with HbA1c above 8%. Thus, only LDL recovered from diabetic patients with HbA1c Ͻ7% were subjected to oxidation. Before in vitro oxidation lipid peroxidation of natural occurring diabetic LDL used in our study was also assayed by thiobarbituric acid reactive substances (TBARS), assay and capillary electrophoresis and compared with that of the native LDL (nLDL). Consistent with our previous results (7) we failed to detect differences in TBARS and conjugates dienes between dmLDL or nLDL (data not shown).
ox-dmLDL Increases p21 Expression and Induces Growth Arrest-Cell cycle events induced by in vitro ox-dmLDL on EPC were first evaluated. The results reported in Table 1 demonstrate that the fraction of cells in S and G 2 /M decreases when cells are cultured with ox-dmLDL. This effect was not observed in EPC cultured in the presence of dmLDL not subjected to oxidation. Indeed, cells in S phase were reduced to 13% compared with cells cultured with VEGF alone or in combination with dmLDL (38 and 36%, respectively). ox-dmLDL were used at a dose of 10 g/ml because higher doses such as 50 g/ml induced an apoptotic phenotype in 10% of the cells. To dissect the molecular events leading to growth arrest, cell cycle-related proteins were analyzed. The dose-response curve shown in Fig. 1B demonstrates that treatment with increasing concentrations of ox-dmLDL is associated with an increase of p21 expression. Moreover, we failed to detect a significant difference between 10 and 50 g/ml ox-dm-LDL. On the contrary, the level of p21 is not affected by the same concentrations of dmLDL. Consistent with the increased level of p21 expression also a decreased level of cyclin D1 expression occurs (Fig. 1C). A sustained increasing level of p21 can be mediated by various mechanisms, including regulation of protein stability (8). As shown in Fig. 1D this possibility is ruled out by the finding that the p21 half-life in EPC cultured in the presence of dmLDL or ox-dmLDL is similar. In addition (Fig. 1E) the level of p21 mRNA, but not that of cyclin D1, is significantly increased in EPC cultured in the presence of ox-dmLDL, and pretreatment with actinomycin D abrogates the effect of ox-dmLDL on both p21 mRNA and protein expression. These findings suggest that mechanisms other than protein accumulation account for our results.
It is known that oxidized LDL exert their effects by interacting with membrane receptor structure different from the canonical LDL receptor (33). The possibility that ox-dmLDL affects EPC by interacting with receptors other than that of dmLDL is sustained by competition experiments (Fig. 1F). Indeed, addition of an excess of dmLDL or of LDL recovered from normal subjects (nLDL) does not change the level of p21 in EPC. On the contrary the induction of p21 expression is abrogated by blocking the receptor LOX-1 but not by blocking RAGE, indicating that p21 expression is mediated by LDL oxidation rather that by LDL glycation.
p21 Expression and Growth Arrest Are Dependent on Accumulation of p53 and the Presence of Unphosphorylated Rb Protein-p21 is a well known target of p53 (34,35). Thus, nuclear accumulation of p53 was first evaluated in EPC cultured in the presence of ox-dmLDL or dmLDL. Moreover, to exclude further the possibility that glycation of LDL can account for our results in selected experiments CML-BSA was used. As shown in Fig.  2A p53 accumulates in the nucleus in EPC cultured with ox-dmLDL but not with dmLDL or CML, strongly supporting the role of LDL oxidation in this event (Fig. 2B). Consistently phosphorylated Rb protein can be only detected in EPC cultured with dmLDL. To confirm the involvement of p53 in mediating this signaling pathway EPC were depleted of endogenous p53 by siRNA. As shown in Fig. 2D ox-dmLDL fails to induce both accumulation of p53 and p21 expression in EPC transfected with p53 siRNA (Fig. 2, C  and D). Moreover, depletion of endogenous p53 also prevents the effect of ox-dmLDL on Rb protein (Fig. 2E) and of ox-dmLDL-induced growth arrest (Table 2).
It has been reported recently that in macrophage from mice lacking p21 an increased expression of the Ink4 family member p16 can be detected (36). However, the finding (Fig. 2D) that ox-dmLDL treatment does not change the level of p16 both in EPC depleted or not depleted of endogenous p53 indicates that different mechanisms might account for our results.
Akt Is Constitutively Activated in EPC Cultured with ox-dmLDL but Not with dmLDL-The results reported in Fig. 3A demonstrate that Akt, which is known to regulate the p53/p21 signaling pathway (17), is constitutively activated in EPC cultured with ox-dmLDL. On the contrary when EPC were transiently stimulated with ox-dmLDL Akt activation is undetectable (Fig. 3B). The role of Akt is sustained by the observations that pharmacological inhibition of Akt, by LY294002, besides abrogating Akt activity also prevents p53 accumulation and p21 FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 expression as well as cell cycle events in response to ox-dmLDL (Table  3). Moreover, the role of Akt in this signaling pathway is further demonstrated by the results obtained with the p38 mitogen-activated protein kinase inhibitor SB203580 (Fig. 3B and Table 3) or with the specific Akt inhibitor 1L-6-hydroxymethyl-ciro-inositol-2-(R)-2-O-methyl-3-O-octadecylcarbonate (data not shown).

p53-mediated EPC Senescent Growth Arrest in Diabetes
In Caenorhabditis elegans a reduction-of-function mutation in the phosphatidylinositol 3-kinase/Akt pathway leads to extension of the life span, and this effect is inhibited by mutations of antioxidant genes (37). Radical scavenger genes have a protective effect against oxidative dam-age in human cells (38), and defective antioxidant enzyme production has been reported in diabetic patients (39). Thus the level of MnSOD and of catalase in EPC cultured in the presence of dmLDL or ox-dmLDL was evaluated. We found that, unlike the expression of catalase, that of MnSOD is reduced in EPC cultured with ox-dmLDL (Fig. 3D). Consistent with the effect exerted by ox-dmLDL on MnSOD, ox-dmLDL-cultured EPC show an increased ROS production (Fig. 3E). The role of Akt in the signaling pathway leading to ROS production in EPC-cultured with ox-dmLDL was also evaluated upon depletion of endogenous Akt (Fig. 3F). The lucigenin-enhanced chemiluminescence assay shown in EPC cultured with 20 ng/ml VEGF, 10 g/ml VEGF ϩ dmLDL, or 10 or 50 g/ml VEGF ϩ ox-dm-LDL were washed and fixed with ethanol and processed by FACS. 0.5 M/liter doxorubicin was used as a positive control for apoptosis.

Cell cycle phases
Percentage of cells a VEGF ox-dmLDL 10 g/ml ox-dmLDL 50 g/ml dmLDL 50 g/ml Doxorubicin  After, the nitrocellulose was incubated in 4% albumin blocking solution subsequently incubated with rabbit IgG anti-AGE antibody. Goat IgG anti-rabbit IgG labeled with alkaline phosphatase was used. Dots are visualized with the procedure contained in the Immun-Blot Assay kit. B, EPC cultured for 15 days with VEGF and with ox-dmLDL or dmLDL as indicated were lysed and processed for Western blot analysis. Cell extracts were subjected to 15% SDS-PAGE and transferred electrophoretically to nitrocellulose filters. The filter was immunoblotted with anti-p21 antiserum and reprobed with an anti-␤-actin antibody. C, EPC were cultured and processed as above and evaluated for cyclin D1 expression using an anti-cyclin D1 antibody. EPC treated for 10 min with VEGF were used as control (VEGF). D, whole cell lysates from cycloheximide-treated EPC (cultured in the presence of dmLDL or ox-dmLDL) were prepared at each time point and processed for Western blot analysis. The filter was immunoblotted with anti-p21 antiserum and reprobed with anti-␤-actin antibody. E, EPC, cultured as above, were untreated or treated with ox-dm-LDL alone or in combination with 10 g/ml actinomycin D (Act D) and evaluated for p21 expression. The filter was reprobed with an anti-␤-actin antibody. CML-BSA-treated endothelial cells were used as positive control (ϩ) (right panel). Northern blot analysis was performed on EPC treated as above (left panel). p21, cyclin D1, or ␤-actin cDNA probes were used. F, EPC cultured with VEGF plus ox-dmLDL alone or in the presence of a 200-fold excess of dmLDL or nLDL (n) and of blocking anti-RAGE or a blocking anti-LOX-1 antibodies were lysed and subjected to SDS-PAGE. The filter was immunoblotted with an anti-p21 antibody and reprobed with an anti-␤-actin antibody. Three different experiments were performed with similar results. AGE, p21, and cyclin D1 expression was also quantified by densitometric analysis, as described under "Experimental Procedures." * and §, p Ͻ 0.05.

p53-mediated EPC Senescent Growth Arrest in Diabetes
Fig. 3G demonstrates that in EPC transfected with scramble siRNA ox-dmLDL is able to induce ROS production and that this effect is inhibited by pretreatment with NAC and DPI. Moreover, consistent with the effect exerted by LY294002, depletion of endogenous Akt makes EPC unresponsive to ox-dmLDL-induced ROS production. To evaluate whether the p53/p21-mediated growth arrest relies on oxidative stress, the effect of ROS scavenger, NAC, was also examined. As assessed in Fig. 3C both the level of p53 and p21 as well as the growth arrest (Table 3) are prevented by the addition of NAC. These observations together with the finding that in EPC depleted of endogenous p53 ox-dmLDL fails to reduce MnSOD expression (Fig. 3H) sustain the possibility that, as in C. elegans, in human EPC the protective effects against oxidative damage can rely on these signaling events. ox-dmLDL Impair EPC Maturation-To determine whether the activation of the p53/p21 signaling pathway could also impair EPC mat-uration, flow cytometric analysis was performed. As shown in Fig. 4A the expression of VE-cadherin and CD146 (S-endol) and the loss of CD34 expression demonstrate the ability of EPC cultured in the presence of VEGF and dmLDL to undergo terminal maturation. On the contrary no VE-cadherin and CD146 are detected in EPC cultured with ox-dmLDL that still retain the expression of CD34, indicating that ox-dmLDL affect the ability of EPC to undergo terminal differentiation (Fig. 4, d-f). Similarly these cultured cells did not express CD31 (data not shown). Consistent with their undifferentiated phenotype EPC cultured in the presence of ox-dmLDL show an impaired ability to form tube-like structure in a Matrigel assay (Fig. 4C), p53-mediated Senescence Occurs in EPC Cultured with ox-dmLDL and in EPC from Diabetic Patients-Apart from DNA damage, certain undefined stress-causing signals also induce senescence denoted as accelerated senescence (40). p53, p21, and pRb are the major regulators FIGURE 2. p53 is involved in ox-dmLDL-mediated p21 expression. A, EPC cultured for 15 days with VEGF and dmLDL, ox-dmLDL, or CML were lysed, and total cell extracts were subjected to SDS-PAGE. The filter was immunoblotted with an anti-p53 antibody and reprobed with an anti-tubulin antibody. B, EPC cultured with VEGF and dmLDL or ox-dmLDL were lysed and processed for Western blot analysis. The filters were immunoblotted with an anti-p53, an anti-pRb, or an anti-tubulin antibody. C, EPC cultured with ox-dmLDL were transfected with p53 siRNA or with the scrambled sequence (control siRNA) and lysed. Total cell extracts were subjected to SDS-PAGE. The filter was immunoblotted with an anti-p53 antibody and reprobed with an anti-tubulin antibody. MCF7 stimulated with 25 M/liter) copper were used as positive control (ϩ). D, EPC cultured with VEGF and dmLDL or ox-dmLDL and transfected as above were lysed. Total cell extracts were subjected to SDS-PAGE, and the filters were immunoblotted with anti-p21, anti-p53, and anti-p16 antibodies. Tubulin was used as loading control. MCF7 stimulated with 25 M/liter copper and CML-treated endothelial cells were used as positive control (ϩ). E, whole cell lysates from EPC cultured with ox-dmLDL were transfected as above and subjected to SDS-PAGE. The filter was immunoblotted with anti-pRb and anti-p16 antibodies. Tubulin was used as loading control. Four different experiments were performed with similar results. p21 and p53 expression and Rb phosphorylation were also quantified by densitometric analysis, as described under "Experimental Procedures." *, p Ͻ 0.05. FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 of senescence (9 -11, 40). To assess whether the activation of the Akt/ p53/p21 signaling pathway translates into an accelerated onset of senescence, EPC cultured for 15 days in the presence of ox-dmLDL or dmLDL were assayed for acidic ␤-galactosidase. The result of a representative experiment depicted in Fig. 5, A and B, demonstrates an increased proportion of SA-␤-gal-positive EPC cultured with ox-dm-

TABLE 2 Effects of ox-dmLDL on cell cycle events in EPC depleted of endogenous p53
EPC cultured with 10 g/ml VEGF plus ox-dm-LDL were transfected with p53 siRNA or the scramble sequence (control) and processed by FACS. 10 M/liter LY294002 was added where indicated.

Percentage of cells
a p Ͻ 0.05 ox-dmLDL versus untreated (control siRNA). b p Ͻ 0.05 ox-dmLDL ϩ LY294002 (control siRNA) versus ox-dmLDL (control siRNA) and untreated, ox-dmLDL, ox-dmLDL ϩ LY294002 p53 siRNA expressing cells versus ox-dmLDL (control siRNA). In lane 2 EPC were transiently (10 min) treated with VEGF. Total cell extracts were subjected to SDS-PAGE, and the filter was immunoblotted with an anti-pAkt and reprobed with an anti-Akt antibodies. B, EPC cultured with VEGF and dmLDL were treated for 10 min with ox-dmLDL as indicated. EPC cultured with VEGF and ox-dmLDL were pretreated or not with LY294002 (Ly) or SB203580 (SB) (for the last 24 h). Cell extracts were subjected to SDS-PAGE, and the filters were immunoblotted with anti-pAkt antibody and reprobed with an anti-Akt antibody. C, cell extracts from EPC treated with dm-LDL or ox-dmLDL (alone or with LY294002 or NAC) were subjected to SDS-PAGE, and the filters were immunoblotted with anti-p53 and anti-p21 antibodies. Tubulin and ␤-actin were used as loading controls. D, whole cell lysates from EPC cultured as indicated were processed for Western blot analysis. The filters were immunoblotted with anti-MnSOD and anti-catalase antibodies. Tubulin was used as loading control. E, DCF was added to the samples for the analysis of ROS production. Time 0 represents basal DCF fluorescence. TNF, tumor necrosis factor. F, EPC cultured with ox-dmLDL were transfected with the scrambled sequence (control siRNA) or with Akt siRNA for the indicated intervals and lysed. Total cell extracts were subjected to SDS-PAGE. The filter was immunoblotted with an anti-Akt antibody. ␤-Actin was used as loading control. G, the rate of ROS production, measured by lucigenin-derived chemiluminescence, was evaluated in EPC transfected with the scrambled sequence (control siRNA) or with Akt siRNA. EPC were treated with saline alone (c), dmLDL, or with ox-dmLDL. In selected experiments ox-dmLDL treatment was combined with NAC, DPI, or LY294002 treatment as indicated. H, EPC cultured with VEGF and ox-dmLDL were transfected with p53 siRNA or with the scrambled sequence (control siRNA) and lysed. Total cell extracts were subjected to SDS-PAGE, and the filter was immunoblotted with anti-MnSOD and anticatalase antibodies and reprobed with an antitubulin antibody. Four different experiments were performed with similar results. p21, p53, MnSOD, and Akt protein expression and Akt activation were also quantified by densitometric analysis, as described under "Experimental Procedures." * and §, p Ͻ 0.05.

p53-mediated EPC Senescent Growth Arrest in Diabetes
LDL (B) compared with EPC cultured in the presence of dmLDL (A). The results obtained upon NAC and LY294002 treatment (Fig. 5C) sustain the role of Akt and ROS production in the signaling that regulates EPC life span. The involvement of p53 in regulating EPC life span was also evaluated in EPC endogenously depleted of p53. Consistent with the results obtained by the use of NAC and LY294002 the results reported in Fig. 5D demonstrate that Akt activation does not occur in ox-dmLDL-cultured EPC depleted of endogenous p53 and that this event is associated with a reduced proportion of senescent cells. The potential role of this signaling pathway was also evaluated in a diabetic setting. To this end EPC recovered from sex-and age-matched normal and diabetic subjects (in good metabolic control: HbA1c Յ7%) were analyzed. As in EPC cultured in the presence of ox-dmLDL, also in EPC from diabetic patients Akt is activated (Fig. 5H), the level p53 and p21 is increased (H), the level of MnSOD is decreased, and cells are prevalently senescent (Fig. 5, F and G). It has been shown that several stress-causing signals can accelerate the onset of senescence via p53-mediated signals (40). Our findings that, after p53 silencing, diabetic EPC are more resistant to oxidative stress (expression of MnSOD) (Fig. 5I, right), that the proportion of senescent cells is reduced (Fig. 5L), and that the ability to differentiate and to form tube-like structures is recovered (Fig. 5M) provide evidences that a p53-mediated pathway may account for the accelerated onset and progression of diabetic vasculopathy.

DISCUSSION
In the present study we investigated the molecular mechanisms involved in EPC dysfunction in diabetes. We demonstrate that when EPC are cultured in a diabetic milieu they show a senescent-like growth arrest that depends upon p53-mediated signaling pathway.
Inhibition of cell proliferation and increased p53 expression are known to occur in endothelial cells as well as in smooth muscle cells and macrophages within the atherosclerotic lesions (41,42). These findings as well as the appearance of multinucleated and enlarged EC both in the aging vasculature and in vascular lesions such as atherosclerosis and postangioplasty restenosis (42) appear to be part of the protective mechanisms that cells develop to limit the extent of proliferation in response to injury. Indeed, evidences accumulated over the last years lent support to the notion that p21, p53, p57, and p27 can limit neointimal thickening in animal models of atherosclerosis and angioplasty (13)(14)(15). On the basis of immunohistochemical studies, these proteins have also been considered as negative regulators of cell proliferation in human atherosclerotic and restenosis lesions, and their overexpression has been generally proposed as a promising approach to treat or prevent these diseases (43). However, the common perception of p21 as an atheroprotective factor is now challenged by the unexpected acceleration of atherosclerosis in p21 Ϫ/Ϫ /apoE ϩ/ϩ mice (36).
Depending on the severity of damage to the genome, p53 can activate genetic programs that halt cell proliferation transiently (G 1 and G 2 cell cycle arrest), permanently (senescence), or eliminate the cell altogether (apoptosis). Evidence for the role of p53 in senescence comes from several studies. Abrogation of the p53/p21 pathway by various strategies can bypass senescence in human and mouse cells (44 -46). Moreover, enforced expression of p53 or p21 in certain cell types can induce a senescent-like phenotype (9,47). It has been proposed that p21 induction by p53 leads to inhibition of CDKs, which in turn results in hypophosphorylation of pRb, which very likely mediates cell cycle arrest during senescence (9). Indeed, our study demonstrates that the microenvironment containing ox-dmLDL causes an increase of p53 and of the

Cell cycle phases
Percentage of cells a p Ͻ 0.05 ox-dmLDL and ox-dmLDL ϩ SB203580 versus VEGF. b p Ͻ 0.05 ox-dmLDL ϩ NAC or ϩ LY294002 versus ox-dmLDL. CDK inhibitor p21, which maintains Rb in an underphosphorylated state, thereby causing EPC senescent-like growth arrest. Moreover, the results obtained by depletion of endogenous p53 provide strong evidence that p21 induction and the resulting hypophosphorylation of Rb can regulate ox-dmLDL-mediated senescent-like growth arrest. It is well established that pRb remains constitutively hypophosphorylated in senescent cells (48), suggesting deregulation of the activity of pRb kinases during senescence. Moreover, the CDK4 and CDK6 inhibitor p16 is up-regulated during senescence in different human tissues (48 -50), suggesting that high p16 may account for hypophosphorylation of pRb in senescent cells. It has been recently suggested that the increased expression of p16 during senescence is not nearly as universal as thought previously (51) and that on the level of p16 depends the propensity to reverse senescence by inactivating p53 (52). In agreement with these data the results of the p16 knock-down strategy have shown that down-regulation of p16 may not be functionally equivalent to pRb inactivation (53). Consistent with these findings and with data obtained by McConnell et al. (54) are the observations that in EPC undergoing senescence p16 is not up-regulated and that depletion of endogenous p53 rescues ox-dmLDL-mediated senescence.
Regulation of endothelial cell life span depends upon an Akt-mediated p53/p21 signaling pathway (17). Moreover, hyperinsulinemia, a common feature of type 2 diabetic patients, seems to be associated with an activated state of Akt in endothelial cells, suggesting that a senescence-like growth arrest may commonly occur in diabetic vasculopathy (17). We herein demonstrate that (i) Akt is activated in EPC cultured in the presence of ox-dmLDL and in EPC recovered from diabetic patients; (ii) inhibition of Akt activity is temporally related to a decreased p21 expression and p53 accumulation; (iii) Akt is stably activated in EPC undergoing senescence; (iv) inhibition of Akt activity prevents growth arrest and senescence. Moreover, in ox-dmLDL-cultured EPC depleted of endogenous p53 the senescent-like growth arrest is prevented and Akt is not constitutively activated. Taken together our findings indicate that a deranged regulation of EPC life span occurs in diabetes and that this effect may also depend on the constitutive activation of Akt.
High glucose concentrations in vitro and hyperglycemia in vivo are well known stimuli for the production of free radicals and the generation of oxidative stress with a corresponding increase in the expression and activity of antioxidant enzymes that act as a defensive system against cell damage (55). Genetic analyses have demonstrated that the reduction-of-function mutations in the insulin/phosphatidylinositol 3-kinase/Akt signaling pathway of insulin/insulin-like growth factor I extends longevity by altering the balance between pro-oxidant and antioxidant enzymes in organisms ranging from yeast to mice (37,38). Consistently we found that both normal EPC cultured in the presence of oxidized diabetic LDL and EPC from type 2 diabetic patients have a FIGURE 5. Diabetic milieu accelerates the onset of senescence. A and B, senescence was evaluated by acidic ␤-gal activity on EPC cultured for 15 days with VEGF plus dmLDL (A) or ox-dmLDL (B). C, SA-␤-gal-positive cells were quantified on EPC cultured with VEGF plus dmLDL (c) or ox-dmLDL. ox-dmLDL-cultured EPC were also pretreated with NAC or LY294002. D, Akt activation (lower panel) and SA-␤-gal-positive cells (upper panel) were evaluated on EPC transfected with p53 siRNA or with the scramble sequence (control siRNA) and cultured with VEGF plus ox-dmLDL or dmLDL. For Akt activation total cell extracts were subjected to SDS-PAGE, and the filters were immunoblotted with an anti-pAkt antibody and reprobed with an anti-Akt antibody. Positive control for activated Akt is indicated as c. E and F, EPC from age-and sex-matched controls and diabetic patients cultured for 10 days were evaluated for acidic ␤-gal activity (E, normal; F, diabetic). G, senescence was evaluated by acidic ␤-gal activity. SA-␤-gal-positive cells were quantified on EPC obtained from normal (N) and diabetic (D) subjects. H, whole cell lysates from normal (N) and diabetic (D) samples were subjected to SDS-PAGE, and the filters were immunoblotted with anti-p21, anti-p53, anti-pAkt, and anti-MnSOD antibodies. ␤-Actin and tubulin were used as loading controls. I, EPC obtained from diabetic patients were transfected with p53 siRNA or with the scramble sequence (control siRNA). After 60 h the cells were lysed. Total cell extracts were subjected to SDS-PAGE, and the filters were immunoblotted with an anti-p53 antibody (left panel) or with anti-p21 or anti-MnSOD antibody (right panel). Tubulin was used as loading control. L, SA-␤-gal-positive cells were quantified on EPC obtained from diabetic patients transfected with p53 siRNA or the scramble sequence (control siRNA). M, EPC from diabetic subjects, transfected as above, were processed for in vitro angiogenesis. The quantification of tube-like structure formation in Matrigel assay, as described under "Experimental Procedures," was expressed as a percentage Ϯ S.D. of the tube-like structure area to the total Matrigel area. Data in C and D (upper) and G and L are the mean Ϯ S.D. *, p Ͻ 0.05, control versus experimental groups; §, p Ͻ 0.05, ox-dmLDL versus ox-dmLDL ϩ NAC and ox-dmLDL ϩ LY294002. Three different experiments were performed with similar results.

p53-mediated EPC Senescent Growth Arrest in Diabetes
reduced protective effect against oxidative damage and an accelerated onset of senescence and that both events can be rescued by depletion of endogenous p53. These observations, together with the finding that depletion of endogenous Akt abrogates ox-dmLDL-mediated ROS production and MnSOD down-regulation (data not shown), provide evidence that the signaling pathway leading to Akt activation and MnSOD down-regulation might be recapitulated in a human disease.
Preservation of a quiescent multipotential stem cell pool capable of intermittently giving rise to highly proliferative progenitors is essential for maintaining tissue homeostasis. In patient with diabetes both the number and function of EPC are altered (18 -20). In the present study we first provide evidence that EPC cultured in a diabetic milieu as well as EPC recovered from diabetic patients undergo a senescent-like growth arrest by a p53/p21-mediated pathway. Our results suggest that this event can contribute to the impairment of the ischemia-induced neovascularization in a diabetic setting. Moreover, the results presented here point to a more critical approach to cell-based clinical strategies to enhance tissue perfusion in diabetic patients with coronary and peripheral artery diseases and/or to the definition of EPC functional tests for further therapeutic approaches to obtain optimal cells for transplantation.