Doxorubicin Induces Apoptosis and CD95 Gene Expression in Human Primary Endothelial Cells through a p53-dependent Mechanism*

Regulation of the homeostasis of vascular endothelium is critical for the processes of vascular remodeling and angiogenesis under physiological and pathological conditions. Here we show that doxorubicin (Dox), a drug used in antitumor therapy, triggered a marked accumulation of p53 and induced CD95gene expression and apoptosis in proliferating human umbilical vein endothelial cells (HUVECs). Transfection and site-directed mutagenesis experiments using the CD95 promoter fused to an intronic enhancer indicated the requirement for a p53 site for Dox-induced promoter activation. Furthermore, the p53 inhibitor pifithrin-α (PFT-α) blocked both promoter inducibility and protein up-regulation of CD95 in response to Dox. Up-regulated CD95 in Dox-treated cells was functional in eliciting apoptosis upon incubation of the cells with an agonistic CD95 antibody. However, Dox-mediated apoptosis was independent of CD95/CD95L interaction. The analysis of apoptosis in the presence of PFT-α and benzyloxycarbonyl-Val-Ala-dl-Asp-fluoromethylketone revealed that both p53 and caspase activation are required for Dox-mediated apoptosis of HUVECs. Finally, Dox triggered Bcl-2 down-regulation, cytochrome c release from mitochondria, and the activation of caspases 9 and 3, suggesting the involvement of a mitochondrially operated pathway of apoptosis. These results highlight the role of p53 in the response of primary endothelial cells to genotoxic drugs and may reveal a novel mechanism underlying the antitumoral properties of Dox, related to its ability to induce apoptosis in proliferating endothelial cells.

The regulation of apoptosis in endothelial cells is critical for the integrity of endothelium. Processes such as vascular re-modeling and angiogenesis involve both proliferation and apoptosis of vascular endothelial cells (1,2). In addition, endothelial injury that results in apoptosis appears to play an important pathogenic role in the progression of atherosclerotic lesions and many inflammatory disorders (3)(4)(5)(6)(7)(8).
The CD95 (Fas/Apo-1) receptor, a member of the TNF 1 /nerve growth factor receptor family (9,10), triggers a potent apoptotic signal when bound to its natural specific ligand CD95L (11). This apoptotic signal eventually results in the activation of the caspase cascade (12). However, a number of stress agents may also elicit the activation of downstream caspases through a different apoptotic pathway, which involves the release of cytochrome c (13).
Genotoxic stress by chemotherapeutic drugs such as doxorubicin (Dox) has been shown to activate apoptosis in a number of different cell types (14). Dox has been used to treat a variety of cancers, and, as in the case of other DNA-damaging agents, it induces the accumulation of the p53 tumor suppressor protein. This results in cell cycle arrest and may lead to apoptosis (15,16), a mechanism that ensures the elimination of these dangerous cells from the organism. Recently, p53 has been shown to regulate CD95 gene expression through interactions with p53 binding motifs located within the CD95 gene promoter and an enhancer situated within the first intron of the CD95 gene (17). In several tumor cell lines the p53-mediated up-regulation of CD95 has been reported to be required for genotoxic druginduced apoptosis (17,18). CD95 and CD95L have been shown to be expressed on the surface of endothelial cells (19 -23). In the vascular endothelium, CD95L appears to negatively regulate extravasation by inducing apoptosis of leukocytes as they cross the vessel wall (24). Under homeostatic conditions, CD95 is expressed at lower levels in endothelial cells that are resistant to CD95-mediated apoptotic cell death (19 -22, 24, 25). Recently, a number of cellular stresses, including matrix detachment and exposure of cells to oxidized low density lipoprotein or hydrogen peroxide have been reported to up-regulate endothelial cell surface expression of CD95 (20,21,25) and to induce apoptosis through CD95/CD95L interaction (20,25). Despite this, very little is known about the mechanisms that regulate CD95 gene expression in endothelial cells. To address this issue we have searched for stimuli that induce CD95 expression and found that Dox promoted apoptosis in human primary endothelial cells and was a potent inducer of CD95. In these cells, Dox induced a sustained accumulation of p53, which regulated both CD95 gene expression and the apoptotic process. However, although Dox-induced CD95 protein was able to trigger apoptotic signals, it was not involved in the apoptosis induced by the drug. Our findings show that Dox induced apoptosis through a mitochondrially operated p53-dependent pathway, and it is noteworthy that significant apoptosis was only seen in subconfluent endothelial cells These findings support a major role for p53 in the regulation of CD95 and apoptosis of primary endothelial cells exposed to genotoxic drugs.
The preferential effect of Dox inducing apoptosis on proliferating endothelial cells may reveal a novel mechanism underlying the antitumoral activity of Dox. If Dox is able to selectively trigger apoptosis in proliferating cells "in vivo," this could result in selective killing by the drug of endothelial cells involved in neovascularization, including that required for the growth and dissemination of tumors. We also discuss the potential advantages and disadvantages that would be derived from the use of p53 inhibitors for the prevention of the side effects of Dox in wound healing or ovulation.
Determination of Apoptotic Cells-Subconfluent HUVECs were plated in 24-well tissue culture plates (4 ϫ 10 4 cells/plate) precoated with 0.5% gelatin. Attached cells were detached from the culture plates with a trypsin solution containing 3 mM EDTA (tryspin/EDTA). These cells were then collected together with floating cells, washed once with cold phosphate-buffered saline (PBS), and incubated with 200 l of staining buffer (0.1% sodium citrate, 0.02 mg/ml RNase, 0.3% Nonidet P-40, 0.05 mg/ml propidium iodide) for 30 min on ice. Then, 300 l of staining buffer (without RNase and propidium iodide) was added and hypodiploid apoptotic cells were determined by cytofluorometric analysis of DNA content in a FACScan cytofluorometer (Becton Dickinson).
Flow Cytometry-HUVECs were plated in six-well tissue culture plates (3 ϫ 10 5 cells/plate) precoated with 0.5% gelatin. They were then treated as described under "Results," detached with 3 mM EDTA, washed once with cold PBS, and then incubated with one of the following antibodies for 30 min on ice: anti-CD95 mouse monoclonal IgG antibody DX2 (1 g/ml), the anti-CD95L mAb NOK-1 (IgG1) (2 g/ml), or the anti-ICAM-1 mAb RR1/1. After this incubation, cells were washed once with cold PBS and then incubated with fluorescein isothiocyanate-conjugated rabbit anti-mouse Ig (1/50, Dako) for 30 min on ice. Cells were again washed with cold PBS and resuspended in PBS. Flow cytometry was performed on a FACScan cytometer and analyzed with CellQuest software (Becton Dickinson).
Plasmid Constructs and Transient-transfection Assays-Genomic fragments of the CD95 upstream regulatory region spanning 391 bp from the ATG site of the CD95 gene was amplified by PCR of human genomic DNA. These sequences were then cloned into the BglII site of the pXP2 luciferase reporter plasmid to generate the pCD95 391 Luc construct. This plasmid was used as the parental construct to generate the pI-CD95 391 Luc by cloning a 500-bp enhancer region from the first intron of the CD95 gene (17) into the BamHI site of the pXP2 polylinker (upstream of the promoter region). The 500-bp fragment was also amplified from genomic cDNA using the primers ENHFAS5Ј: cgggtccGT-GAGCCCTCTCCTGCCCGGGT and ENHFAS3Ј: cgggatcCCTGAAG-GCTGCAGGCTCTCTCC. The pmI-CD95 391 Luc plasmid, harboring mutations within the p53 sites of the intronic enhancer, was generated by site-directed mutagenesis of the pI-CD95 391 Luc with a QuikChange kit from Stratagene. Mutations were introduced using the following sense oligonucleotides: 5Ј-AACTCCTGGAGGGGCCCTGA-CAAG-3Ј and 5Ј-CCCTGACAATAAAAGCCAAAGGT-3Ј and their respective antisense oligonucleotides (mutated nucleotides are in boldface). The pBHA-941Luc plasmid including a 941-bp fragment of the upstream regulatory region of the ICAM-1 promoter has been described previously (26). The reporter plasmid pG13 Luc (p53 Luc), containing 13 tandem copies of the p53-responsive element, was kindly donated by Dr. B. Vogelstein (Johns Hopkins University School of Medicine, Baltimore, MD) (27). The pCDNA3-hBCL-2 plasmid, containing the cDNA for the human Bcl-2 human gene, was kindly provided by Dr. Jacint Boix (University of Lleida, Spain). The pEGFP-spectrin expression vector was kindly provided by Dr. R. F. Kalejta (Princeton University, Princeton, NJ) (28). The pKBF-Luc construct includes a trimer of the NF-B motif from the H-2K b gene, placed upstream of a herpes simplex virus thymidine kinase minimal promoter driving the luciferase reporter gene (29).
For transient transfection experiments, HUVECs were plated in 100-mm tissue culture dishes (1.5 ϫ 10 6 cells/plate) the day before transfection. Cells were transfected in 4 ml of Dulbecco's minimal essential medium containing 10% FCS, with 10 g of the indicated luciferase reporter plasmid by the calcium phosphate procedure as previously described (30,31), with some modifications. Briefly, cells were incubated with precipitated DNA until a mild cytopathic effect was observed (4.5-9 h). Cells were then washed twice with PBS and detached with trypsin/EDTA from the culture dishes. After centrifugation, cells were resuspended in OPTI-MEM (Invitrogen) containing 0.5% FCS, and split among six-well (35-mm) tissue culture plates precoated with 0.5% gelatin. After 24 h, transfected cells incubated with or without z-VAD-FMK were treated with different agents for an additional period of 14 h. Cells were then detached with trypsin/EDTA, washed with PBS, and lysed. Luciferase activity was measured according to the instructions of the Luciferase System kit (Promega) in a Sirius luminometer (Berthold, Germany). The expression of Renilla luciferase was used as an internal control for the efficiency of transfection. A total of 0.1 g of Renilla luciferase expression vector pRLCM (Promega) was used in co-transfection experiments. In these experiments, 1/10 of cells co-transfected with both types of luciferase plasmid were plated in 24-well tissue culture plates, and the cells were treated in the same way as those plated on six-well tissue culture plates. After lysis with passive lysis buffer, Renilla luciferase activity was measured with the Dual luciferase assay kit (Promega).
Subcellular Fractionation and Western Blot Analysis-After treatment, subconfluent attached HUVECs grown in 35-mm culture dishes were detached with trypsin/EDTA and collected together with floating cells. They were then washed with cold PBS, resuspended in 40 l of Laemmli buffer, and sonicated. For the detection of cytochrome c release, cells were detached, and washed with PBS, and the pellet was resuspended in 50 l of cold lysis buffer (25 mM Tris-HCl pH 6.8, 250 mM sucrose, 1 mM EDTA, 0.005% digitonin, 1 mM DTT, 0.1 mM PMSF, 1 mM benzamidine, and 1 g/ml each of aprotinin, leupeptin, and pepstatin) for 30 s on ice. Lysates were centrifuged, and the supernatant containing the cytosolic fraction was removed and separated from mitochondria. Cytosolic proteins were mixed with Laemmli buffer and p53 Regulates CD95 Expression and Apoptosis in HUVECs sonicated. For detection of p53 protein, 10 g of whole cell extract were resolved on 8% SDS-PAGE minigels. For all other proteins, 60 g of extract (cytosolic or whole) was resolved on 12% SDS-PAGE minigels. Proteins were transferred onto Immobilon membranes (Millipore), and the blots were blocked with 5% w/v skimmed milk in PBS/0.05% Tween 20 (PBST) for 1 h at room temperature. Blots were then washed three times for 10 min with PBST and incubated in PSBT, 1% w/v milk at 4°C overnight with the indicated antibodies: anti-Bcl-2 (1:500, DAKO), anticaspase-3 (1:1000, New England BioLabs), anti-PARP (1:4000, Roche Molecular Biochemicals), anti-caspase-9 (1:250, New England BioLabs), anti-p53 (1:3000, Ab-7, Calbiochem), or anti-cytochrome c mAb (1 g/ ml, PharMingen). Blots were again washed three times for 10 min with PBST and then incubated with the corresponding horseradish peroxidase-coupled goat anti-rabbit or anti-mouse secondary antibody (1: 2000, DAKO). After three washes with PBST and once with PBS, blots were visualized with the Amersham Biosciences, Inc. enhanced chemiluminescence (ECL) detection reagents. Detection of ␣-tubulin (anti-␣tubulin, 1:40000, Sigma) was used to control for loading of protein.
Electrophoretic mobility shift assays (EMSAs) were performed by incubating nuclear proteins (2-3 g) with 1 g of poly(dI-dC) DNA carrier and 3 l of 5ϫ DNA binding buffer (10% (w/v) polyvinylethanol, 12.5% (v/v) glycerol, 50 mM Tris (pH 8), 2.5 mM EDTA, 2.5 mM DTT) in a final volume of 15 l on ice for 10 min. Then 2 l (1 ng/l) of 32 P-labeled double-stranded oligonucleotide (10 8 cpm/g) was added to the reaction mixture, and it was incubated at room temperature for 30 min. For competition experiments, a 30-fold molar excess of unlabeled oligonucleotide was added before the addition of the probe. Where indicated, nuclear extracts were incubated at room temperature for 10 min before addition of the probe with the following antibodies: 2-4 l of anti-p53 antiserum PAb421; 1 l of antiserum 1226, raised against the p65 NFB subunit; or 1 l of antiserum 1141, raised against the p50 NFB subunit. DNA-protein complexes were resolved by electrophoresis on 4% nondenaturing polyacrylamide gels. The sequences of the oligonucleotides (5Ј to 3Ј) used in these experiments were as follows: ctagCTCCCCAACCCGGGCGTTCCCCAGCGAGG (human NFB sequence Ϫ306 to Ϫ278 of the 391-bp fragment of CD95 promoter (32)) and gatcCTCCTGGACAAGCCCTGACAAGCCAAGCCA (human p53 sequence located within of the intronic enhancer of the CD95 human gene (17)).
Immunofluorescence Experiments-To analyze the effect of Bcl-2 expression on the viability of HUVECs treated with Dox, cells were co-transfected with 200 ng of the pEGFP-spectrin expression plasmid together with 200 ng of either pCDNA3-hBcl-2 or the control pCMV␤galactosidase expression vector (33). The pGL3 Basic vector (2.1 g) was added as a DNA carrier in a total volume of 0.140 ml, and transfection was performed by the calcium phosphate procedure in 35-mm tissue culture dishes. After treatment, the cells were washed with PBS, fixed with 3.7% formaldehyde for 15 min, and washed for a further 10 min with 50 mM NH 4 Cl blocking solution in PBS. Cells were then washed with PBS, permeabilized with a 0.1% Triton X-100 for 10 min, washed again with PBS, and stained with 1 g/ml 4Ј,6-diamidino-2phenyl-indole solution for 2 min. The cells were examined under a fluorescence microscope, and GFP-positive cells were scored after counting a minimum of 1000 total cells for each condition. The efficiency of transfection in Bcl-2-and ␤-galactosidase-expressing cells, determined in aliquots of transfected cells just before the addition of Dox, was similar (10 -12%).
Data Analyses-Data are presented as means Ϯ S.E. of several determinations. Differences between groups were tested for significance using Student's t test. Analysis of cell cycle by flow cytometry was performed by counting a total population of at least 5000 cells. In these assays similar results were obtained in at least three independent experiments, and the significance determined by the 2 test.

Doxorubicin Induces Apoptosis in Subconfluent HUVECs-
Previous studies have shown that different genotoxic drugs induce apoptosis in many types of cells (34 -39). To analyze whether Dox affected the viability of primary endothelial cells, we performed FACS analysis on propidium iodide-stained HU-VECs and determined the fraction of hypodiploid apoptotic cells following Dox treatment. These experiments showed the presence of apoptotic cells after 36-h treatment, with a further increase by 48 h. In addition, this cell cycle analysis demonstrated a marked accumulation of Dox-treated cells in the G 2 /M phase after a 24-h treatment (Fig. 1A). Because we observed FIG. 1. Induction of apoptosis in Dox-treated subconfluent endothelial cells. A, HUVECs were incubated with 500 ng/ml Dox, and apoptosis was determined by FACS analysis after propidium iodide staining of nuclei at the indicated times after treatment. B, the effect of cell confluence on the sensitivity of HUVEC to Dox was analyzed by cell cycle analysis as above, using cells plated at the indicated confluence and then treated with 500 ng/ml Dox for 30 or 48 h. Results are expressed as the percentage of cells displaying a sub-G 1 DNA content. The basal level of apoptosis (Control) was monitored in parallel cultures of untreated HUVECs after 48 h. The data are representative of three independent experiments, and the significance level ( 2 ) was p Ͻ 0.01 in cells at either 60% or 40% confluence when comparing Dox-treated cells for either 30 or 48 h versus nontreated cells at the same confluence and treatment times, and in cells at 100% confluence when comparing Dox-treated cells for 48 h with nontreated cells. variability in the percentage of apoptotic cells induced by Dox in different experiments, we analyzed whether cell confluence could influence the sensitivity of HUVECs to Dox. As shown in Fig. 1B, cell confluence significantly affected the strength of Dox-mediated apoptosis in HUVECs. Dox failed to induce apoptosis of confluent cells (100%) after a 30-h treatment, and only a slight increase in the number of sub-G 1 apoptotic cells (10 -12%) was observed by 48 h under these conditions. However, in cells plated at low cell density (40% confluence) apoptosis was observed after 30 h, and an extensive number of sub-G 1 apoptotic cells (up to 70%) was detected after 48 h (Fig. 1B).
Doxorubicin Induces CD95 Expression and Apoptosis through a p53-dependent Mechanism in HUVECs-Dox has been shown to up-regulate CD95 expression in a number of cells that can become sensitized to death through CD95/CD95L interactions (34,35,40,41). In some cases, the cellular response to DNA damage results in p53 accumulation, which is required for the up-regulation of CD95 expression. To determine whether Dox could induce CD95 protein expression at the cell surface of primary endothelial cells, we analyzed CD95 expression levels by flow cytometry in Dox-treated HUVECs. Dox induced a marked increase in the number of CD95-expressing HUVECs ( Fig. 2A). This increase took place in a dose-dependent manner (data not shown) and was consistent with that detected in the CD95 mRNA levels by RT-PCR. In these experiments, the low levels of CD95 mRNA found in unstimulated cells were clearly up-regulated by Dox after 10-h treatment, and further maintained for at least 24 h. As a control, the CD95 mRNA levels of Jurkat JHM1 T-lymphocytes were analyzed in parallel (Fig. 2B). To further investigate the mechanisms involved in the up-regulation of CD95 by Dox, we analyzed the effect of the drug on CD95 promoter activity in HUVECs. We performed transient transfection experiments to determine the transcriptional activity of a luciferase reporter plasmid driven by a 391-bp fragment of the CD95 promoter (pCD95 391 Luc). This plasmid was further modified to contain a 500-bp p53-responsive enhancer element from the first intron of the CD95 gene, inserted upstream of the promoter fragment (pI-CD95 391 Luc). As shown in Fig. 2C, Dox induced by 2-to 3-fold the transcriptional activity of CD95 reporter plasmid that harbored the p53 enhancer. The promoter fragment alone was not responsive to Dox, and the intronic fragment containing the functional p53 binding site (17) was required to achieve a significant promoter induction by Dox (Fig. 2C). Together these results indicate that Dox regulates CD95 expression in HUVECs and suggest that p53 could be involved in the CD95 up-regulation by Dox through transcriptional mechanisms.
To further analyze the role of p53 in CD95 expression we next tested whether Dox regulated p53 protein expression in endothelial cells. To this end, we performed Western blot experiments using whole cell extracts of HUVECs obtained at different times after treatment with Dox. These assays showed that p53 protein, expressed at low levels in unstimulated cells, was already induced at 4 h after Dox treatment, and reached higher levels of expression by 12 h that were maintained for at least 36 h (Fig. 3A).
In agreement with the results of the Western blot experiments, EMSAs with nuclear extracts of HUVECs and a probe, including the p53 sequence of the CD95 intronic enhancer showed that p53 binding activity was significantly induced after Dox treatment (Fig. 3B, left). The presence of p53 in the retarded complex was shown by the addition of the PAb421 antibody, which completely shifted the specific complex. In addition, p53 complex formation was efficiently competed by an excess of cold homologous oligonucleotide (Fig. 3B, right).
Of note, flow cytometric analysis of CD95 expression in HU-VECs indicated that the specific p53 inhibitor PFT-␣ (42) blocked both constitutive and Dox-induced expression of CD95 (Fig. 4A, upper panels). By contrast, the TNF-␣-induced upregulation of ICAM-1 cell surface expression (analyzed as a control in parallel HUVEC cultures) was not affected by the treatment with PFT-␣ (Fig. 4A, lower panels). Therefore, the PFT-␣-mediated inhibition of CD95 expression in HUVECs

p53 Regulates CD95 Expression and Apoptosis in HUVECs
was not due to a toxic or nonspecific effect of the inhibitor. Control Western blot experiments, performed with extracts from aliquots of the same cells used for FACS, showed that PFT-␣ efficiently inhibited both the basal and inducible levels of p53 protein (Fig. 4B).
To evaluate the functional contribution of the intronic p53 site to the transcriptional response of CD95 to Dox, we tested the effect of PFT-␣ in transient transfection experiments with the pCD95 391 Luc and pI-CD95 391 Luc luciferase reporter plasmids. As shown in Fig. 5A, pretreatment of cells with PFT-␣ completely inhibited the induction of pI-CD95 391 Luc by Dox. Consistently with the FACS and Western blot experiments, PFT-␣ also inhibited the basal activity of the CD95 promoter constructs regardless of the presence of the intronic enhancer. Again, the activity of PFT-␣ was controlled for by Western blot analysis of cell lysates from the treated cells. (Fig.  5A, inset). As with the TNF-␣-mediated cell surface up-regulation of ICAM-1, the TNF-␣-induced luciferase reporter activity of a 941-bp fragment of the ICAM promoter was refractory to inhibition by PFT-␣ (data not shown). Evidence for the functional involvement of the p53 site located within the CD95 intronic enhancer in the Dox-induced activation of the CD95 promoter was provided by site-directed mutagenesis experiments. Mutation of this p53 site resulted in a reduction of the basal and inducible transcriptional activity of the pmI-CD95 391 Luc construct that was similar to that displayed by the enhancerless pCD95 391 Luc (Fig. 5B).
Dox has been shown to activate NFB in a number of cell types (43)(44)(45)(46). Because NFB is also implicated in the transcriptional regulation of CD95 (32,47,48), we next analyzed whether NFB was also involved in the Dox-mediated activation of CD95 in HUVECs. As shown in Fig. 6A, Dox failed to induce NFB binding activity to the Ϫ306/Ϫ278 functional B site of the CD95 promoter, whereas the binding of p65 and p50 NFB subunits was induced by TNF-␣. Similar results were obtained with a different site; a B motif of the interleukin-2 promoter that was also used as a probe in parallel EMSA experiments. Furthermore, the activity of an NFB reporter plasmid, which was activated by TNF-␣ treatment of HUVECs, was not induced by Dox, whereas in parallel experiments Dox efficiently induced the transcriptional activity of a p53-dependent promoter (Fig. 6B).
Given the involvement of p53 in the Dox-mediated apoptosis in many different cell types (17, 49 -51), we next examined whether the induction of p53 expression was related to the apoptosis triggered by the drug. For this purpose, we analyzed the effect of PFT-␣ on the cell cycle of HUVECs exposed to Dox for 48 h. PFT-␣ treatment completely prevented the appearance of sub-G 1 apoptotic cells, indicating the requirement for p53 in the Dox-induced apoptosis in HUVECs (Fig. 7). The efficient inhibition of Dox-mediated p53 expression by PFT-␣ was confirmed by control Western blot experiments using whole cell extracts from the cells analyzed in these cell cycle experiments (Fig. 7, inset).
CD95/CD95L Interaction Does Not Mediate Dox-induced Apoptosis in HUVEC-Because p53 was implicated in Doxinduced cell surface expression and gene promoter activation of CD95, we tested whether signals delivered through CD95 were involved in the apoptosis induced by Dox. We first evaluated whether Dox induced the expression of CD95L, which could be involved in apoptosis through interaction with CD95. As shown in Fig. 8A, CD95L was not detected in the surface of untreated HUVECs by FACS. Low levels of CD95L were detected by HUVECs exposed to KB8301, an inhibitor of matrix metalloproteinase previously shown to inhibit the cleavage of membrane-bound CD95L (52). But Dox failed to induce the expression of CD95L in the presence or absence of KB8301 (Fig. 8A). Moreover, Dox also failed to induce the expression of CD95L mRNA in HUVECs, whereas in parallel experiments CD95L mRNA expression was efficiently amplified in JHM1 Jurkatderived cells treated with carbachol or phorbol 12,13-dibutyrate plus ionophore (data not shown and Ref. 53). Furthermore, incubation of Dox-treated HUVECs with either the CD95 antagonistic antibody DX2 or the CD95L blocking antibody NOK-1 failed to inhibit Dox-mediated apoptosis (Fig. 8B). Control experiments using carbachol and CD95 antibody in JHM1 cells demonstrated the ability of NOK1 and DX2 antibodies to efficiently block apoptosis triggered through CD95 activation (Fig. 8B). Nonetheless, the Dox-induced CD95 protein was able to trigger death signals as demonstrated by the marked induction of cell death displayed by the CD95 agonistic antibody CH11 in Dox-treated HUVECs (Fig. 8B). Therefore, on the one hand, these results indicate that Dox induces apoptosis in HUVECs through a CD95/CD95L-independent mechanism, whereas, on the other hand, they also show that the CD95 receptor expressed after genotoxic drug treatment is able to elicit functional apoptotic signaling in HUVECs.
Involvement of a p53-dependent Caspase Activation in the Doxorubicin-induced Apoptosis of HUVECs-Because CD95/ CD95L interaction was not involved in Dox-mediated apoptosis in HUVECs, we next analyzed the effect of Dox on the activa-

FIG. 3. Dox induces p53 protein expression and p53-DNA binding activity in HUVECs.
A, cells were treated either with or without Dox (500 ng/ml) for the indicated times. Ten g of total cell lysate was analyzed by Western blot probed with an anti-p53 antibody, or with an anti-␣-tubulin antibody as a control for protein loading. B, nuclear extracts from HUVECs stimulated for 1 h with 500 ng/ml Dox were analyzed by EMSA with a probe containing the p53 site of the human CD95 intronic enhancer. EMSAs were performed in the presence or absence of the anti-p53 antiserum PAb421. The specific DNA-p53 complex is indicated by the arrow. A 30-fold molar excess of unlabeled p53 oligonucleotide was added to the binding reaction to determine the specificity of binding. A representative result of three independent experiments is presented. tion of the caspase cascade, and the potential involvement of p53 in this process. We first determined the effect of the general caspase inhibitor z-VAD-FMK on Dox-induced apoptosis. Cell cycle analysis of HUVECs exposed to 100 M z-VAD-FMK efficiently blocked apoptosis induced by Dox (Fig. 9). Further confirmation of the involvement of caspases was obtained by Western blot analysis of extracts of HUVECs treated with Dox for 30 -36 h. As shown in Fig. 10A, Dox led to the activation of executioner caspases, as revealed by the proteolytic cleavage of the PARP nuclear substrate and the activation of caspase-3. At earlier time points we failed to detect this activation (data not shown). In addition, exposure of HUVECs to Dox resulted in release of cytochrome c from the mitochondria and caspase-9 activation (Fig. 10B). Moreover, Bcl-2 protein levels (reported to regulate caspase activation through the inhibition of cytochrome c release (54, 55)) were down-regulated by Dox at times at which cytochrome c release was detected (Fig. 10, C and B). In addition, we performed experiments where GFP was coexpressed with Bcl-2, or with ␤-galactosidase as a control, in HUVECs that were then treated with Dox. When the cells were treated with Dox for 24 h, the number of viable, GFP-expressing cells was reduced by 53-67% in two independent experiments (data not shown) in cultures co-expressing the ␤-galactosidase control gene. In contrast, co-expression of Bcl-2 completely prevented this loss of viability upon Dox treatment. This suggests that down-regulation of Bcl-2 is a critical step in Dox-mediated apoptosis of HUVEC, and together these data support the involvement of a mitochondrially operated pathway of apoptosis, triggered by Dox in endothelial cells. Although PARP and caspase-3 cleavage were already detected after 30 h of treatment, they reached higher levels by 36 h. However, maximal activation of caspase-9 was observed after a 30-h treatment. This probably reflects the activation of caspase-3 by caspase-9 (13).
Because the inhibition of p53 resulted in the blockade of apoptosis induced by Dox (Fig. 7), we determined whether inhibition of p53 accumulation affected caspase activation by Dox. As shown in Fig. 10D, treatment of HUVECs with PFT-␣ completely prevented activation of caspase-3 by Dox. This suggests that p53 could be mediating Dox-induced apoptosis via activation of the caspase pathway in HUVECs.
Taken together, these findings support the hypothesis that Dox-induced apoptosis in HUVECs is regulated by a p53-dependent mechanism, involving the activation of downstream caspases, in a mitochondrially operated apoptotic pathway. DISCUSSION Dox is a chemotherapeutic drug widely used in the treatment of a variety of cancers, including leukemias, sarcomas, and breast cancer (36,37,56). Although Dox has been shown to induce programmed cell death, the mechanisms by which Dox operates appear to be different depending on the cell type analyzed. Thus, CD95/CD95L interactions have been reported to mediate the drug-induced apoptosis in several tumor cell lines (34,35,40) but not in others (36,37,56). Similarly, apoptosis by genotoxic drugs appears to be dependent on p53 in hepatoma cells (18) but not in various breast tumor xenografts (57).
Several recent reports (20,21,25) have shown that treatment of endothelial cells with different stimuli such as oxidized low density lipoprotein, hydrogen peroxide, or matrix detachment result in the up-regulation of CD95 cell surface expression by endothelial cells. However, the mechanisms that regulate CD95 gene expression in endothelial cells remain poorly understood. In this study, we have looked for stimuli that induce CD95 expression and found that Dox induced both apoptosis and CD95 up-regulation in human primary endothelial cells through a p53-dependent mechanism. This up-regulation of CD95, however, was not involved in the Dox-induced FIG. 4. p53 is required for the Doxmediated up-regulation of CD95 expression in HUVECs. A, HUVECs were pretreated with or without 30 M PFT-␣ for 24 h and subsequently treated with or without 500 ng/ml Dox for an additional 48 h. Cell surface expression of CD95 was determined by flow cytometry. As a control, the effect of the same dose of inhibitor was tested on the cell-surface expression of ICAM in HUVECs stimulated with TNF-␣ (50 ng/ml) for 24 h. In all instances, fresh PFT-␣ was added every 24 h. B, the inhibitory effect of PFT-␣ was confirmed by Western blot analysis using anti-p53 and anti-␣-tubulin antibodies. Whole cell extracts from aliquots of cells used in A were analyzed. Results representative of three independent experiments are presented.
p53 Regulates CD95 Expression and Apoptosis in HUVECs activation of the caspase cascade that led to the apoptosis of HUVECs.
We have found that Dox efficiently up-regulates p53 protein expression and DNA binding activity, and the transcriptional activity of the pG13 Luc p53-dependent promoter in HUVECs. Because NFB has been shown to transcriptionally regulate CD95, we conducted parallel experiments to analyze the effect of Dox on the activation of NFB in HUVECs. However, in these experiments we did not detect any effect of Dox on the binding or transcriptional activation by NFB. Although the doses of Dox we used here are lower than those shown to activate NFB in some previous reports (44 -46), other authors have observed activation at lower doses (43). It is likely that the activation of NFB by Dox and the role of this in CD95 gene induction may depend on the cell type. In this regard, Dox has been shown to activate NFB and CD95 expression in hepa-toma cells but this expression was not transcriptionally mediated by NFB (46).
The involvement of p53 in the regulation of CD95 gene expression in HUVECs was demonstrated by transfection experiments, which showed that transcriptional activity of the gene promoter was dependent on the presence of an intact p53 site within the first intron of the CD95 enhancer. Previous studies have reported the importance of the intronic region for transcriptional activation of the CD95 promoter in response to p53 in hepatoma cells (17). We have shown that mutation of a critical p53 site within this enhancer completely blocked Doxmediated transcriptional activation of promoter construct in HUVECs. Furthermore, the p53 inhibitor PFT-␣ blocked both the activity of the promoter and the expression of CD95 at the cell surface induced by Dox, thus suggesting that p53 regulates CD95 at the transcriptional level.
Although anti-CD95 or anti-CD95L blocking antibodies did not inhibit Dox-mediated apoptosis, the CH11 agonistic antibody to CD95 was able to trigger apoptotic signals in Doxtreated cells, showing that the cellular machinery required for CD95 apoptotic signaling, inactive in untreated HUVECs, was activated by Dox. Because CD95L is expressed by activated circulating lymphocytes, the presence of functional CD95 on the surface of endothelial cells might represent a potential risk for the endothelium, and it is clear that tight regulatory mechanisms must operate to maintain vascular integrity. In fact, despite the basal expression of CD95 found in endothelial cells, these cells have been reported to be particularly resistant to CD95-mediated apoptosis (19 -22, 24, 25), and we have shown here that resting HUVECs fail to undergo apoptosis after CD95 ligation. In view of these results, it will be very important to investigate whether the sensitization to CD95 ligation and the apoptosis induced by Dox that we have observed in HUVECs "in vitro " take place "in vivo." It is important to note that cardiotoxicity, impairment of wound healing, and renal and liver complications are common side effects frequently found in patients treated with Dox (58 -62). Because Dox was not able to induce significant apoptosis of confluent HUVECs, it would also be of great interest to address whether the selective sensitivity of HUVECs to Dox under subconfluent conditions is reflected in the proliferating cells of the endothelium in vivo. If  Dox is able to selectively trigger apoptosis of proliferating (but not resting) endothelial cells in vivo, it is possible that part of its antitumoral activity is mediated through this effect. In such a case, Dox could exert selective apoptosis in endothelial cells involved in neovascularization, which would result in disruption of blood vessel formation in growing tumors.
Our experiments clearly show that the caspase inhibitor z-VAD-FMK prevented Dox-mediated apoptosis of HUVECs. A number of recent reports have revealed a role for oxidative stress and the involvement of mitochondria in Dox-mediated apoptosis in different cell systems, including bovine aortic endothelial cells and myocytes (38,58,63,64). We found that after 30 -36 h of treatment, Dox induced cytochrome c release from mitochondria into the cytosol, caspase-9 activation, and a concomitant down-regulation of Bcl-2 protein levels. These results support the involvement of a mitochondrially operated apoptotic pathway induced by Dox in endothelial cells. In this context, it is important to note that Bcl-2 has been shown to prevent both the disruption of the inner mitochondrial membrane potential and the release of cytochrome c from the mitochondria (54,55). Although the precise molecular mechanisms by which Bcl-2 prevents apoptosis are not completely clear and appear to be different depending on the cell type analyzed, Bcl-2 has been reported to act by an antioxidant mechanism (65,66), and by interaction with different proapoptotic members of the family (67,68). Because p53 has been reported to mediate Bcl-2 down-regulation (69,70), it is possible that the reported Dox-mediated activation of free radicals and toxic metabolites (63,64) could initially trigger the activation of p53, leading to a p53-dependent down-regulation of Bcl-2 and activation of a mitochondrially regulated caspase cascade. This scenario would be consistent with our experiments showing that Dox-mediated loss of cell viability is blocked in cells expressing Bcl-2. Although p53 may regulate genes other than Bcl-2 involved in apoptosis (69,71), it is possible that the effect of PFT-␣ in preventing apoptosis of HUVECs is mediated, at least in part, by preventing the Dox-mediated down-regulation of Bcl-2 levels.
The involvement of p53 in the regulated expression of CD95 and apoptosis by Dox, and the prevention of apoptosis by PFT-␣, point to the potential use of p53 inhibitors for the treatment of the impaired wound healing and ovulation that occur as side effects of Dox in treated patients. However, the inhibition of p53 may have undesirable effects. Not only might this block the potential beneficial effect of Dox on proliferating endothelial cells, it could also interfere with apoptosis of tumor cells that sense DNA damage in response to genotoxic stress and trigger an apoptotic response upon accumulation of p53. A careful analysis by evaluation of the effects of Dox and p53 inhibitors in vivo will be required to elucidate these important issues.
Acknowledgments-We are very grateful to Dr. S. Bartlett for critical reading of the manuscript and editorial assistance. We also thank to Drs. Edmundo Ferná ndez, Yolanda Rodríguez, Jesú s Vazquez, Miguel Campanero, Manuel Izquierdo, and Emilia Mira for providing reagents and advice.