The Nucleotide Analog Cidofovir Suppresses Basic Fibroblast Growth Factor (FGF2) Expression and Signaling and Induces Apoptosis in FGF2-Overexpressing Endothelial Cells

ABSTRACT

Cidofovir [(S)-1-(3-hydroxy-2-phosphonylmethoxypropyl)cy- tosine; (S)-HPMPC] is an antiviral drug that has been approved for the treatment of cytomegalovirus retinitis in patients with AIDS. Cidofovir also possesses potent activity against human papillomavirus-induced tumors in animal models and patients. We have recently shown that cidofovir inhibits the development of vascular tumors induced by basic fibroblast growth factor (FGF2)-overexpressing endothelial cells (FGF2-T-MAE) in mice. Here, we demonstrate that the inhibitory activity of cidofovir in FGF2-T-MAE cells may result from the specific induction of apoptosis. Cell cycle analysis revealed that cidofovir induces accumulation of cells in the S phase and, upon prolonged treatment, a significant increase in sub-G1 cells, exhibiting a subdiploid DNA content. Moreover, annexin V binding, an early event in apoptosis induction, was increased in cidofovir-treated FGF2-T-MAE cells. Cidofovir also caused nuclear fragmenta- tion and the activation of caspase-3-like proteases, as evidenced by the cleavage of poly(ADP-ribose)polymerase. In ad- dition, cidofovir treatment of FGF2-T-MAE cells resulted in a pronounced up-regulation of the tumor suppressor protein p53. However, the expression of Bax and Bcl-2 remained un- changed, and cidofovir did not induce the release of cyto- chrome c from the mitochondria. In addition, cidofovir did not suppress the phosphorylation of protein kinase B/Akt, a trans- mitter of antiapoptotic survival signals, or its downstream reg- ulator Bad, indicating that the Akt pathway is not affected by cidofovir in FGF2-T-MAE cells. However, the compound inhib- ited the expression of FGF2 and FGF2 signaling through Erk42/ 44, as shown by Western blot analysis. Our results indicate that cidofovir inhibits the growth of FGF2-T-MAE cells via inhibition of FGF2 expression and signaling and via the induction of apoptosis. These findings suggest that the clinical use of cido- fovir might be expanded to tumors that are not induced by oncogenic viruses.

The acyclic nucleoside phosphonate analog cidofovir [(S)-1-(3-hydroxy-2-phosphonylmethoxypropyl) cytosine, (S)- HPMPC] (Fig. 1a) is a broad-spectrum anti-DNA virus agent that has been approved for the treatment of cytomegalovirus retinitis in patients with AIDS (De Clercq et al., 1986; De Clercq and Holy´, 2005). However, cidofovir also proved effec- tive in various other clinical situations, including the treatment of progressive multifocal leukoencephalopathy, recur- rent respiratory papillomatosis, pulmonary papillomatosis, anogenital condylomata acuminata, gingival papillomatous lesions, adenovirus-associated hemorrhagic cystitis, and mol- lusca contagiosa. In addition, cidofovir is active against pox- viruses and is currently the only licensed antiviral drug that has been stockpiled for possible use in the prophylactic ther- apy of smallpox (De Clercq and Holy´, 2005). The antiviral effect of cidofovir results from a selective inhibition of the viral DNA polymerase by its diphosphate metabolite. HPMPCpp acts as a competitive inhibitor/alternative sub- strate with respect to the natural substrate dCTP. HPMPC is also converted to HPMPCp-choline, which has a very long intracellular half-life of > 80 h and may serve as a reservoir from which the active metabolite HPMPCpp can be continuously generated, thus explaining the potent antiviral activity of cidofovir upon infrequent dosing (Ho et al., 1992).

Cidofovir has also shown marked antitumor activity in several animal models, including Epstein-Barr virus (EBV)- associated nasopharyngeal carcinoma and human papilloma- virus (HPV)-induced cervical carcinoma xenografts in athy- mic-nude mice, polyomavirus-induced hemangiomas in rats, and hemangiosarcoma development in mice (Andrei et al., 1998; Liekens et al., 1998, 2001c; Neyts et al., 1998; Murono et al., 2001). These tumors are all associated with oncogenic viruses, which encode for cell-transforming proteins that in- teract with products of tumor suppressor genes. Several re- ports have indicated that the antitumor effect of cidofovir may result from interference with this interaction (Andrei et al., 2000; Abdulkarim et al., 2002, 2003). However, the potent inhibitory activity of cidofovir on the growth of strongly vas- cularized tumors of endothelial cell origin (i.e., hemangiomas or hemangiosarcomas) led us to assume that the compound might also possess antiangiogenic properties (Liekens et al., 1998, 2001c).
Angiogenesis is a complex multistep process leading to the formation of new capillary blood vessels. During tumor growth, an intratumoral network of blood vessels is gener- ated to provide the growing tumor with oxygen and nutri- ents. The new capillaries prevent tumors from undergoing apoptosis and facilitate the escape of tumor cells into the circulation and subsequent metastasis to distant organs (Liekens et al., 2001a; Carmeliet, 2003). Blood vessel forma- tion requires an extensive interplay between cellular and soluble factors with either inhibitory or stimulatory function (Liekens et al., 2001a). One of the best characterized angio- genic factors is basic fibroblast growth factor (FGF2), which has been shown to induce endothelial cell proliferation, che- motaxis, and protease production in vitro and angiogenesis in vivo (Presta et al., 2005). FGF2 has also been implicated in the pathology of angiogenic diseases such as Kaposi’s sar- coma and hemangiomas and plays a crucial role in the patho- genesis of several human tumors, including gliomas and hep- atocellular carcinoma (Takahashi et al., 1990; Kandel et al., 1991; Takahashi et al., 1994; Presta et al., 2005).

Endothelial cells overexpressing FGF2 (FGF2-T-MAE cells) are tumorigenic in athymic-nude mice, giving rise to highly vascularized lesions, which histologically resemble Kaposi’s sarcoma (Gualandris et al., 1996; Sola et al., 1997). We have recently shown that intratumoral or systemic cido- fovir administration significantly inhibits the growth of sub- cutaneous, intraperitoneal, and intracerebral FGF2-T-MAE xenografts in athymic nude and SCID mice (Liekens et al., 2001b). Cidofovir also completely suppressed hemangioma formation on the chick chorioallantoic membrane induced by intra-allantoic injection of FGF2-T-MAE cells, without affect- ing the normal chorioallantoic membrane vessels. In addi- tion, the drug inhibited the proliferation of primary and transformed endothelial cells but did not affect their capacity to form tube-like structures in fibrin gel or matrigel, indicat- ing that the primary action of cidofovir is not angiostatic (Liekens et al., 2001b).

Because FGF2-T-MAE cells are not associated with an oncogenic virus, we wanted to examine how cidofovir exerts its inhibitory effect on these FGF2-overexpressing endothe- lial cells. Our results indicate that cidofovir may inhibit the growth of FGF2-T-MAE cells via the inhibition of FGF2 ex- pression and signaling and via the induction of apoptosis.

Materials and Methods

Materials. Cidofovir (Vistide; Fig. 1a) was kindly provided by Gilead Sciences (Foster City, CA). Poly(4-styrenesulfonic acid) (PSS) was provided by Dr. P. Mohan (Chicago, IL). Cell Cultures. FGF2-transfected mouse aortic endothelial (FGF2-T-MAE) cells express high levels of the Mr 18,000, Mr 22,000, and Mr 24,000 isoforms of FGF2 (Gualandris et al., 1996). FGF2-T- MAE cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Carlsbad, CA) supplemented with 10 mM HEPES (Invitrogen), 10% fetal calf serum (FCS; Harlan Sera-Lab Ltd., Loughborough, UK) and 500 µg/ml Geneticin (Invitrogen).

Cell Proliferation Assays. FGF2-T-MAE cells were seeded at various cell densities (see Results) in DMEM with 10% FCS. After 24 h, the medium was replaced and compounds (cidofovir or PSS) were added. The cell cultures were incubated for 3 days, trypsinized, and counted by a Coulter counter (Beckman Coulter, Fullerton, CA). Flow Cytometric Analysis of Cell Cycle and Apoptosis.

FGF2-T-MAE cells were seeded at 20 × 103 cells/cm2 in DMEM with 10% FCS. After 24 h, the cells were exposed to 50 µg/ml cidofovir or various concentrations of PSS (see Results). At different time points, the DNA of the cells was stained with propidium iodide (PI) using the CycleTEST PLUS DNA Reagent Kit (BD Biosciences, San Jose, CA). The DNA content of the stained cell cultures was assessed by flow cytometry on a FACSCalibur flow cytometer equipped with CellQuest software (BD Biosciences) within 3 h after staining (Hatse et al., 1999). Cell debris and clumps were excluded from the analysis by appropriate dot plot gating. Percentages of sub-G1, G1, S, and G2/M cells were estimated using appropriate region markers.
To differentiate between normal (living), apoptotic, and necrotic cells, exponentially growing FGF2-T-MAE cells (seeded at 20 × 103 cells/cm2) were incubated for 1, 2, 3, or 5 days with or without 50 or 100 µg/ml cidofovir. Cells were then simultaneously stained with annexin-V-FITC and PI using the annexin-V-FITC staining kit (Sigma, St. Louis, MO) and analyzed by flow cytometry. Early apo- ptotic cells are characterized by high annexin binding and low PI staining, whereas late apoptotic and necrotic cells stain strongly for both annexin and PI.

DNA Fragmentation Assay. FGF2-T-MAE cells were seeded at 20 × 103 cells/cm2 in DMEM with 10% FCS. After 24 h, the medium was replaced, and 200 µg/ml cidofovir was added. The cell cultures were incubated for 3 days. Next, cells were lysed, high molecular weight DNA was precipitated, and the cleared supernatants were collected. DNA was precipitated overnight at —20°C. DNA pellets were washed with 70% ethanol and treated with RNase A. The samples were heated at 65°C for 10 min, and DNA was electropho- resed in a 2% agarose gel. Induction of apoptosis is associated with the activation of endonuclease, which cleaves DNA at sites located between nucleosomal units generating mono- and oligonucleosomal DNA fragments.

Western Blot Analysis of FGF2-T-MAE Cell Extracts. FGF2- T-MAE cells were seeded at 20 × 103 cells/cm2 (unless otherwise stated under Results) in flasks containing DMEM with 10% FCS. After 24 h, the medium was replaced, and cidofovir was added. The cell cultures were incubated for 3 days, washed with ice-cold phos- phate-buffered saline (PBS) and lysed as described previously (Liek- ens et al., 1999). Lysates were cleared by centrifugation, and the protein concentration was determined. SDS-polyacrylamide gel elec- trophoresis of the cell lysates was performed as described previously (Liekens et al., 1999). After electrophoresis, proteins were trans- ferred to pretreated Hybond-P polyvinylidene difluoride membranes (GE Healthcare, Little Chalfont, Buckinghamshire, UK). The mem- branes were incubated for 1 h at room temperature in blocking buffer (5% nonfat dry milk in PBS containing 0.1% Tween 20) and subsequently for 12 h at 4°C in blocking buffer with primary antibodies raised against β -actin (1/5000; Sigma), FGF2 (clone FB-8, 1/5000; Sigma), Erk42/44 (1/1000; Cell Signaling Technology, Danvers, MA), phospho-Erk42/44 (1/2000; Cell Signaling Technology), Akt (1/1000;Cell Signaling Technology), phospho-Akt (Ser473, 1/1000; Cell Sig- naling Technology), Bcl-2 (1/1000; Cell Signaling Technology), Bax (1/1000; Cell Signaling Technology), p53 (1/2000; Cell Signaling Technology), or phospho-Bad (Ser136, 1/500; Cell Signaling Technol- ogy). To detect cleavage of poly(ADP-ribose) polymerase (PARP) by caspase-3, a primary antibody was used that detects both intact PARP (113 kDa) and the 89-kDa fragment (Roche, Mannheim, Ger- many). After washing, the membranes were incubated with the corresponding horseradish peroxidase-conjugated secondary anti- body (anti-mouse, 1/2000; anti-rabbit, 1/4000; Dako North America, Inc., Carpinteria, CA) in blocking buffer for 25′ at room temperature.

Next, the membranes were washed extensively. Immunoreactive proteins were detected by chemiluminescence (ECLplus; Bio-Rad Laboratories, Hercules, CA). Samples were collected from at least two independent experiments and run several times. All images were quantified using ChemiDoc XRS (Bio-Rad Laboratories), and statistical significance of the results was determined by means of the unpaired Student’s t test. P values < 0.05 were considered signifi- cant.

Measurement of Mitochondrial Membrane Potential. FGF2- T-MAE or HeLa cells were seeded at 20 × 103 cells/cm2 in flasks containing DMEM with 10% FCS. After 24 h, the medium was replaced, and the cells were treated with cidofovir (100 µg/ml) or actinomycin D (10 µM) for the indicated times (see Results). Then, the monolayers were trypsinized, and 106 cells were stained with the membrane-permeable fluorochrome JC-1 using the flow cytometry mitochondrial membrane potential detection kit (BD Biosciences) according to the manufacturer’s instructions. The cells were ana- lyzed on a FACSCalibur flow cytometer equipped with CellQuest software. JC-1 penetrates the plasma membrane of living cells as a monomer, and its uptake into mitochondria is driven by the mito- chondrial membrane potential. Inside the mitochondria, JC-1 aggre- gates are formed, resulting in a spectral shift toward higher levels of red fluorescence (FL2). Thus, decreased JC-1 fluorescence in the FL2 channel of the flow cytometer is indicative of depolarized mitochon- drial membrane.

The overlap coefficient, Rp, which describes the extent of overlap between image pairs (0 meaning no overlap and 1 meaning 100% overlap) was used to evaluate colocalization. This coefficient is inde- pendent of the intensity variation between channels. SCh1I and SCh2I represent the signal intensity of pixels in channel 1 and channel 2, respectively.

To prepare subcellular fractions, cells were harvested in isotonic buffer (5 mM MOPS, 250 mM sucrose, 1 mM EDTA, 1 mM dithio- threitol, and 0.1% EtOH, pH 7.4) supplemented with protease inhib- itors (1 mM phenylmethylsulfonyl fluoride and 10 µg/ml leupeptin), and homogenized by 37 strokes in a Dounce homogenizer. The ho- mogenates were centrifuged at 750g for 10 min at 4°C to remove unbroken cells and nuclei. The supernatant was then centrifuged at 10,000g for 30 min at 4°C to obtain the heavy membrane pellet enriched with mitochondria. The resulting supernatant was referred to as the cytosolic fraction. Protein concentration of the different fractions was determined, and SDS-polyacrylamide gel electrophore- sis and Western blot analysis were performed as described above, using a primary anti-cytochrome c antibody, provided by Cell Sig- naling Technology.

RNA Isolation. Total cellular RNA (from three independent cell culture experiments) was isolated using the RNeasy minikit (QIA- GEN, Venlo, The Netherlands). RNA was reverse-transcribed to cDNA by means of Moloney murine leukemia virus reverse tran- scriptase (Invitrogen, Paisley, UK) and random primers (Invitrogen) according to the manufacturer’s instructions.

Oligonucleotide Primers and TaqMan Probes. The human FGF2 primers and probe were designed and delivered by Applied Biosystems Assays-on-Demand Gene Expression Products; Assay ID, Hs00266645-m1; Foster City, CA). Mouse β-actin primers and probe were designed with Primer Express (Applied Biosystems). A 138-base-pair fragment of β-actin exon was amplified using primers actin-FW (5'-AGAGGGAAATCGTGCGTGAC-3') and actin-RV (5'-
CAATAGTGATGACCTGGCCGT-3') and detected by a 23-base-pair probe [5'-(carboxy-4,5-dichloro-2,7-dimethoxyfluorescein)CACTGC- CGCATCCTCTTCCTC(5-carboxytetramethylrhodamine)-3'].

Real-Time PCR. Quantitative real-time PCR was performed with an ABI Prism 7000 sequence detection system (Applied Biosys- tems). All samples were analyzed in triplicate reactions in a final volume of 25 µl containing PCR Master Mix (Eurogentec, Seraing, Belgium), different amounts of primers and probes (β-actin, 120 nM primers and 50 nM probe; FGF-2, 360 nM primers and 100 nM probe), and 30 ng of cDNA template. PCR amplification consisted of 40 cycles of denaturation at 95°C for 15 s and annealing/extension at 60°C for 60 s. The results were evaluated by use of SDS software (Applied Biosystems). Relative quantification of gene expression was performed using the standard curve method. For generation of stan- dard curves, plasmids containing the full-length human FGF2 cDNA (i.e., pJC 119-14, obtained from Prof. M. Presta, University of Bres- cia, Brescia, Italy) or β-actin was used. For each sample, the relative amount of FGF2 mRNA was determined and normalized to β-actin. For statistical analyses, the Student’s t test was calculated. Data are given as mean ± S.E. P values < 0.05 were considered significant.

Results

Cidofovir Treatment Suppresses FGF2-T-MAE Cell Growth and Induces Apoptosis in Vitro. The antiviral drug cidofovir (Fig. 1a) inhibited primary tumor growth of FGF2-T-MAE cells in mice (Liekens et al., 2001b). Because these tumors are not associated with a virus, we wanted to examine how cidofovir suppresses the growth of these FGF2- overexpressing endothelial cells.

We evaluated the antiproliferative effect of cidofovir on FGF2-T-MAE cells, seeded at different cell densities. All nontreated cells grew exponentially over a 4-day period and underwent at least two population doublings (data not shown). Cidofovir suppressed the proliferation of FGF2-T- MAE cells in a dose-dependent manner (Fig. 1b). The inhib- itory effect of cidofovir was more pronounced with decreasing initial cell density; i.e., the 50% inhibitory concentration (IC50) was 8, 14, 24, or 35 µg/ml when, respectively, 5, 10, 25, or 50 × 103 FGF2-T-MAE cells/cm2 were seeded. Cidofovir did not inhibit FGF2-T-MAE proliferation (IC50 > 100 µg/ml) when 105 cells/cm2 were seeded, indicating that the initial cell density is an important determinant for cidofovir activ- ity. Thus, for all subsequent experiments, cells were seeded at low density (20 × 103 cells/cm2).

Analysis of cell cycle distribution, via DNA staining with PI, showed that cidofovir (at 50 µg/ml) induces accumulation of FGF2-T-MAE cells in the S phase after 48 h and, upon longer treatment, a significant increase in sub-G1 cells (19% at day 4), exhibiting a subdiploid DNA content (Fig. 2a), characteristic of apoptotic cells. Apoptosis also results in membrane alterations, including the translocation of phos- phatidylserine from the cytoplasmic to the extracellular side of the plasma membrane, where it can be detected by an- nexin V. Thus, normal (living) cells with low annexin binding and low PI staining, early apoptotic cells with high annexin binding and low PI staining, and necrotic cells with high annexin binding and high PI staining can easily be differen- tiated (Fig. 2b). Fluorescence-activated cell sorting showed 0.25 to 4.4% of apoptotic cells in control FGF2-T-MAE cell cultures. Cidofovir treatment markedly increased the num- ber of apoptotic cells to 8.4% after 3 days (versus 0.9% in control cells) and up to 78% after 5 days of treatment at 50 µg/ml (versus 4.4% in control cells) (Fig. 2b). Treatment of FGF2-T-MAE cells with cidofovir also resulted in the activa- tion of endonuclease, which cleaves DNA at sites located between nucleosomal units generating mono- and oligonu- cleosomal DNA fragments, which are absent in DNA from nontreated cell cultures (Fig. 3). These analyses clearly sug- gest that cidofovir induces apoptosis in cultured FGF2-T- MAE cells.

Cidofovir Induces PARP Cleavage and p53 Up-Reg- ulation in FGF2-T-MAE Cells. PARP is a substrate for the protease caspase-3, which is activated during early stages of apoptosis. The antibody used for Western blot analysis rec- ognizes both uncleaved PARP (113 kDa) and the larger cleav- age fragment (89 kDa) (Fig. 4a). After 3 days of treatment with cidofovir, at 50 to 200 µg/ml, a dose-dependent decrease in intact PARP and up to 4-fold increase in the cleaved form of PARP was visible, indicating that caspase-3 is involved in apoptosis induction by cidofovir (Fig. 4a). Protein kinase B/Akt is known to play a critical role in controlling the balance between cell survival and apoptosis (Bellacosa et al., 2005). However, cidofovir suppressed neither the phosphor- ylation of Akt nor its downstream regulator Bad, indicating that the Akt pathway was not affected by cidofovir treatment of FGF2-T-MAE cells (Fig. 4b). In addition, there was no significant change in the expression of Bcl-2 or Bax after 3 days of exposure to 100 µg/ml cidofovir (Fig. 4c). However, cidofovir (100 µg/ml) induced a 3-fold increase in the expression of the tumor suppressor protein p53 (p < 0.05; Fig. 4d). Cidofovir Induces Apoptosis in FGF2-T-MAE Cells Independently of Cytochrome c Release. p53 is an important mediator of the mitochondrial pathway of apoptosis, causing cytochrome c release and/or depolarization of the mitochondrial membrane (Moll et al., 2005). To evaluate whether cytochrome c release is involved in cidofovir-induced apoptosis in FGF2-T-MAE cells, we examined the distribu- tion of cytochrome c in subcellular fractions by immunoblot- ting. The majority of cytochrome c remained in the mitochon- dria-enriched fraction, and there was no detectable increase of cytochrome c levels in the cytosolic fractions after cidofovir treatment (100 µg/ml for 2–4 days) of FGF2-T-MAE cells (data not shown). To confirm these findings, we used immu- nofluorescence microscopy to compare the localization of cy- tochrome c and complex Vα, a mitochondrial protein. Both untreated and cidofovir-treated (100 µg/ml for 2–4 days) FGF2-T-MAE cells showed a mitochondrial distribution of cytochrome c, which colocalized with complex Vα (Fig. 5). In addition, treatment with actinomycin D (10 µM, Fig. 5) or staurosporine (1 µM, data not shown), which have been shown to induce cytochrome c release in other cell types (Arnoult et al., 2002; Duan et al., 2003), did not cause a diffuse cytosolic staining of cytochrome c in FGF2-T-MAE cells, at any of the time points analyzed (8, 10, 12, 14, 18, or 24 h), although cell death was observed.

Moreover, treatment of FGF2-T-MAE cells with cidofovir (100 µg/ml for 2–4 days) did not result in depolarization of the mitochondrial membrane (data not shown). Also actino- mycin D (10 µM) did not affect the mitochondrial membrane potential in FGF2-T-MAE cells at any of the time points analyzed (14, 18, or 24 h), whereas depolarization of the mitochondrial membrane was observed in 70% of the actino- mycin D-treated HeLa cells after 18 h (data not shown).

Inhibition of FGF2 Expression by Cidofovir in FGF2- T-MAE Cells. Growth factors, including FGF2, can protect cells from undergoing apoptosis. We therefore wanted to investigate whether cidofovir has an effect on FGF2 expres- sion and signaling. FGF2-T-MAE cells express high levels of the Mr 18,000, Mr 22,000, and Mr 24,000 forms of FGF2 (Gualandris et al., 1996) (Fig. 6a). FGF2 expression de- creased in a dose-dependent manner after 3 days of treatment with cidofovir [i.e., 38 ± 2% inhibition (p < 0.05) at 50 µg/ml to 67 ± 3% inhibition (p < 0.05) at 200 µg/ml (Fig. 6a)]. Effect of Cell Density on p53 and FGF2 Expression in FGF2-T-MAE Cells Treated with Cidofovir. Cidofovir inhibited FGF2-T-MAE cell proliferation when ≤50,000 cells/ cm2 were seeded (Fig. 1b). Because the findings described above were obtained with cells, seeded at low density (20 × 103 cells/cm2), we wanted to know whether the changes in p53 and FGF2 expression, caused by cidofovir, are also influ- enced by the initial cell number (Fig. 6b). When 104/cm2 FGF2-T-MAE cells were seeded, a 6-fold increase in p53 expression, accompanied by a decrease in FGF2 protein lev- els of 64 ± 10% (p < 0.05), could be noted after 3 days of treatment with 100 µg/ml cidofovir. At higher initial cell densities, the effect of cidofovir was less pronounced (Fig. 6b). In particular, FGF2 levels were unaffected by cidofovir when the initial cell density exceeded 50,000 cells/cm2 (Fig. 6b). In contrast, cidofovir still induced a 2-fold increase in p53 ex- pression (210 ± 10%, p < 0.05) when 105 FGF2-T-MAE cells/cm2 were seeded. Thus, the effects of cidofovir on cell
proliferation, p53, and FGF2 expression are all dependent on the initial FGF2-T-MAE cell density.

In a parallel set of experiments, we quantified FGF2 gene expression levels in untreated and cidofovir-treated FGF2-T- MAE cells by real-time PCR (Fig. 6c). For each sample, the amount of FGF2 mRNA was normalized to β-actin. Statisti- cal analysis was performed on the average values of triplicate reactions from samples taken from three independent exper- iments. As shown in Fig. 6c, the amount of FGF2 mRNA remained constant as the initial cell density decreased from 105 to 2 × 104 cells/cm2. Only at an initial cell density of 104/cm2 was FGF2 transcription significantly reduced, in both untreated and cidofovir-treated cell cultures (p < 0.05). However, cidofovir did not cause a significant down-regulation of FGF2 mRNA expression at any of the cell densities used, indicating that the drug does not inhibit FGF2 tran- scription.

Inhibition of FGF2 Signaling by Cidofovir in FGF2- T-MAE Cells. Next, we studied the effect of cidofovir on intracellular proteins involved in FGF2 signal transduction. As shown in Fig. 7a, phosphorylation of the mitogen-acti- vated protein kinase Erk42/44 was reduced by 52 ± 22% (p <
0.05) in the presence of 100 µg/ml cidofovir, indicating that the drug may inhibit the growth of FGF2-T-MAE cells via inhibition of FGF2 expression and signaling.

We then investigated whether inhibition of FGF2 signaling was sufficient to induce apoptosis in FGF2-T-MAE cells. We showed previously that PSS inhibits FGF2 receptor binding, signaling, and mitogenic activity in endothelial cells at 1 µM (Liekens et al., 1999). Therefore, we treated FGF2-T-MAE cells with increasing concentrations of the FGF2 antagonist. Over a 3-day period, PSS inhibited FGF2-T-MAE cell proliferation with an IC50 value of 1.38 ± 18 µM. After 24 h of treatment, PSS (at 0.5 to 2 µM) significantly reduced Erk42/44 phosphorylation, whereas 5 µM PSS was no longer active (Fig. 7b). At 48 and 72 h, PSS inhibited FGF2 signal transduction up to concentrations of 1 µM (not shown). How- ever, the decrease in FGF2 signaling, caused by PSS, was not accompanied by increased p53 expression (Fig. 7b). In fact, PSS treatment even resulted in down-regulation of p53 pro- tein in FGF2-T-MAE cells. To evaluate the effect of PSS on FGF2-T-MAE cell survival, we performed cell cycle analysis of PSS-treated cells (0.5 to 5 µM for 24 to 72 h). PSS (at 5 µM) caused the accumulation of cells in G1 phase after 48 h (i.e., 81% of PSS-treated cells versus 66% of control cells) (Fig. 7c). However, no apoptotic cells in sub-G1 phase could be detected at any of the concentrations and time points analyzed. Thus, inhibition of FGF2 signaling is not sufficient for apoptosis induction in FGF2-T-MAE cells.

Discussion

Cidofovir is a potent inhibitor of DNA virus replication, and suppresses the growth of HPV-associated tumors in animal models and patients (De Clercq and Holy´, 2005). We have recently shown that cidofovir also inhibits primary FGF2-T-MAE tumor growth in mice (Liekens et al., 2001b). The present study was designed to investigate how cidofovir suppresses the growth of these FGF2-overexpressing endo- thelial cells that are not associated with an oncogenic virus. We found that the antiproliferative effect of cidofovir is more pronounced with decreasing initial cell density. When seeded at low density (max. 20 × 103 cells/cm2), cidofovir induced accumulation of FGF2-T-MAE cells in the S phase and, upon prolonged treatment, a significant increase in sub-G1 cells. Moreover, cidofovir caused nuclear fragmenta- tion and an increase in annexin V binding, indicating that the cytotoxic activity of cidofovir in FGF2-T-MAE cells is attributed to induction of apoptosis. Thus far, cidofovir has only been shown to cause apoptosis in tumors that are asso- ciated with oncogenic viruses, which are known to encode for cell-transforming proteins that interact with products of tu- mor suppressor genes. Indeed, HPV-associated cancers ex- press E6 and E7 oncoproteins, which bind p53 and retino- blastoma (pRb) tumor suppressor proteins, respectively, and neutralize their function. Cidofovir inhibited E6 and E7 ex- pression, and induced the accumulation of active p53 and pRb in cervical carcinoma. (Abdulkarim et al., 2002). In a number of HPV-positive cell lines, cidofovir was shown to induce apoptosis, which was associated with an increase in p53 and p21WAF-1 (Andrei et al., 2000). Also in patients, cidofovir proved very effective in the treatment of HPV-asso- ciated tumors (De Clercq and Holy´, 2005). Intratumoral or local administration resulted in complete regression or par- tial remission of the lesions. Thus, regression of papilloma- tous tumors upon cidofovir treatment in patients may be due, at least in part, to the induction of apoptosis. Intratumoral or systemic administration of cidofovir has also been shown to inhibit EBV-associated nasopharyngeal carcinoma in mice by induction of apoptosis (Neyts et al., 1998). Also in these experimental models, the apoptotic effect of cidofovir was unrelated to inhibition of the viral DNA polymerase, because the enzyme is not expressed in these latently transfected cells (Murono et al., 2001). In addition, the drug decreased the expression of EBV oncoproteins LMP1 and EBNA2 in EBV-related malignancies, resulting in a decrease of the antiapoptotic Bcl-2 and an increase in the pro-apoptotic Bax protein (Abdulkarim et al., 2003). Based on the above-men- tioned findings, it has always been assumed that the anti- tumor activity of cidofovir results from inhibiting the inter- actions of viral oncoproteins with products of tumor suppressor genes.

However, FGF2-T-MAE cells are not associated with an oncogenic virus, implying a different mechanism of action of cidofovir in these tumor cells. Indeed, in contrast to EBV- related cancers, the expression of Bcl-2 and Bax remained unchanged after cidofovir treatment of FGF2-T-MAE cells. We next examined the effect of the drug on the serine-threo- nine kinase Akt/PKB, a transmitter of antiapoptotic survival signals. Activation of the PI3K/Akt pathway has been sug- gested to be associated with the development of cancer and may allow resistance of cells to a variety of apoptosis-induc- ing treatments (Bellacosa et al., 2005). Cidofovir did not suppress the phosphorylation of Akt or its downstream reg- ulator Bad, indicating that the Akt pathway is not affected by cidofovir in FGF2-T-MAE cells.

FGF2-T-MAE cells are murine aortic endothelial cells that overexpress multiple molecular weight isoforms of FGF2 (Gualandris et al., 1996). FGF2 has been shown to stimulate endothelial cell proliferation, migration, and tube formation in vitro and angiogenesis in vivo and can protect cells from undergoing apoptosis (Presta et al., 1986; Gualandris et al., 1996; Liekens et al., 1999; Hotfilder et al., 2005; Presta et al., 2005; Vandermoere et al., 2005). Cidofovir treatment of FGF2-T-MAE cells resulted in a pronounced reduction of FGF2 protein expression and signaling through Erk42/44, as shown by Western blot analysis. However, quantitative RT- PCR did not reveal a significant down-regulation of FGF2 mRNA expression by cidofovir, indicating that the drug does not inhibit FGF2 transcription. Moreover, the FGF2 antago- nist PSS, which inhibits Erk42/44 phosphorylation and FGF2-induced proliferation (Liekens et al., 1999), did not cause apoptosis in FGF2-T-MAE cells, indicating that inhi- bition of FGF2 signaling is not sufficient for apoptosis induc- tion in these cells. In addition, addition of exogenous FGF2 (25–500 ng/ml) could not prevent cidofovir-induced FGF2-T- MAE cell death (data not shown).

Cidofovir caused a significant up-regulation of the tumor suppressor p53. p53 induces either cell cycle arrest by trans- activation of p21 or apoptosis through transcription-depen- dent and -independent mechanisms (Waldman et al., 1995; Polyak et al., 1997; reviewed in Moll et al., 2005 and Resnick- Silverman and Manfredi, 2006). Nuclear p53 may increase the expression of death receptors, pro-apoptotic proteins such as Bax, Bid, Noxa, and PUMA, and redox-related genes (Polyak et al., 1997; Resnick-Silverman and Manfredi, 2006). Cytosolic or mitochondrial p53 may directly activate Bax/Bak via transcription-independent mechanisms, and neutralize the antiapoptotic effect of Bcl-2/Bcl-XL, resulting in the per- meabilization of the mitochondrial outer membrane, the sub- sequent release of cytochrome c, and caspase-3 activation (Mihara et al., 2003; Chipuk et al., 2004; Leu et al., 2004). We did not observe changes in the expression of Bax or Bcl-2 after cidofovir treatment of FGF2-T-MAE cells. In addition, cidofovir did not cause the release of cytochrome c from the mitochondria. In addition, actinomycin D and staurosporine, which have been shown to induce cytochrome c release in various cell types (Arnoult et al., 2002; Duan et al., 2003), caused the death of FGF2-T-MAE cells without affecting cytochrome c localization. Previous reports indicate that p53 may also induce cytochrome c-independent apoptosis in cer- tain cell types (Li et al., 1999). In HeLa cells that were transduced to overexpress p53, apoptosis was induced by reactive oxygen species-mediated disruption of the mitochon- drial membrane potential, in the absence of cytochrome c release (Li et al., 1999). However, cidofovir did not affect the mitochondrial membrane potential in FGF2-T-MAE cells (not shown). In addition, actinomycin D, which strongly in- duced depolarization of the mitochondrial membrane in HeLa cells, had no effect on the mitochondrial potential in FGF2-T-MAE cells (data not shown). These findings indicate that apoptosis induction in FGF2-T-MAE cells may occur without involvement of the mitochondria. However, we did observe the cleavage of the caspase-3 substrate poly(ADP- ribose)polymerase, suggesting that apoptosis stimulation by cidofovir in FGF2-T-MAE cells is caspase-dependent.

Recent studies using knockout mouse models have pro- vided evidence for the existence of several novel p53 targets,including Noxa, Puma, and Perp (Ihrie et al., 2003; Jeffers et al., 2003; Shibue et al., 2003; Villunger et al., 2003). These studies have shown that the role of each target gene is dependent on the cell type and apoptotic stimulus. Noxa and Puma are members of the Bcl-2 family of apoptotic regula- tors, which act on mitochondria to stimulate apoptosis (Vil- lunger et al., 2003). In contrast, Perp represents a novel type of p53 effector that localizes to the plasma membrane rather than to the mitochondria (Attardi et al., 2000; Ihrie et al., 2003). Thus, the up-regulation of p53 upon cidofovir treat- ment of FGF2-T-MAE cells may lead to apoptosis via alter- native routes yet to be identified. It should be noted that cidofovir has been shown to possess potent antitumor activity against EBV-associated nasopharyngeal carcinoma in mice with xenografts that contain mutant p53 genes, indicating that cidofovir may also induce p53-independent apoptosis (Neyts et al., 1998). In fact, several reports indicate that, in certain cell types, DNA damaging agents may induce apopto- sis via Fas signaling, leading to caspase-8 activation and the direct activation of caspase-3 in the absence of mitochondrial involvement (Kasibhatla et al., 1998, Huang et al., 2003).

p53 also participates in post-transcriptional control pro- cesses. In human skin fibroblasts, an inverse correlation was shown between p53 expression and FGF2 mRNA translation efficiency (Galy et al., 2001a,b). Thus, increased expression of p53 protein by cidofovir may cause a reduction in FGF2 levels by a post-transcriptional mechanism, resulting in the inhibi- tion of FGF2 signal transduction and reduced proliferative capacity of FGF2-T-MAE cells. Our results suggest that ci- dofovir affects FGF2 translation and/or degradation. Because increased p53 expression is accompanied by reduced FGF2 levels in cidofovir-treated FGF2-T-MAE cells, p53 may con- tribute to the reduction in FGF2 levels. However, upon seed- ing of FGF2-T-MAE cells at high cell density (i.e., 105 cells/ cm2) cidofovir induced a 2-fold increase in p53 expression, whereas FGF2 mRNA and protein remained constant. Thus, under these conditions, increased expression of p53 was not sufficient to cause a decrease in FGF2 protein levels.

Taken together, our data indicate that cidofovir increased p53 expression and caused caspase-dependent apoptosis in FGF2-T-MAE cells, although the molecular mechanisms linking p53 to caspase-3 remain unclear. The marked anti- tumor effect of cidofovir in mice inoculated with FGF2-T- MAE cells is probably due to the induction of apoptosis. Cidofovir may thus also be useful in a clinical setting to prevent the progression of tumors that are not associated with an oncogenic virus,Caspase Inhibitor VI which would considerably expand the antitumor spectrum of cidofovir.