Antisense inhibition of basic fibroblast growth factor induces apoptosis in vascular smooth muscle cells.

Basic fibroblast growth factor (bFGF), a potent mitogen for many cell types, is expressed by vascular smooth muscle cells and plays a prominent role in the proliferative response to vascular injury. Basic FGF has also been implicated as a survival factor for a variety of quiescent or terminally differentiated cells. Autocrine mechanisms could potentially mediate both proliferation and cell survival. To probe such autocrine pathways, endogenous bFGF production was inhibited in cultured rat vascular smooth muscle cells by the expression of antisense bFGF RNA. Inhibition of endogenous bFGF production induced apoptosis in these cells independent of proliferation, and apoptosis could be prevented by exogenous bFGF but not serum or epidermal growth factor. The induction of apoptosis was associated with an inappropriate entry into S phase. These data demonstrate that interruption of autocrine bFGF signaling results in apoptosis of vascular smooth muscle cells, and that the mechanism involves disruption of normal cell cycle regulation.

Basic fibroblast growth factor (bFGF), a potent mitogen for many cell types, is expressed by vascular smooth muscle cells and plays a prominent role in the proliferative response to vascular injury. Basic FGF has also been implicated as a survival factor for a variety of quiescent or terminally differentiated cells. Autocrine mechanisms could potentially mediate both proliferation and cell survival. To probe such autocrine pathways, endogenous bFGF production was inhibited in cultured rat vascular smooth muscle cells by the expression of antisense bFGF RNA. Inhibition of endogenous bFGF production induced apoptosis in these cells independent of proliferation, and apoptosis could be prevented by exogenous bFGF but not serum or epidermal growth factor. The induction of apoptosis was associated with an inappropriate entry into S phase. These data demonstrate that interruption of autocrine bFGF signaling results in apoptosis of vascular smooth muscle cells, and that the mechanism involves disruption of normal cell cycle regulation.
Basic fibroblast growth factor (bFGF) 1 belongs to a multigene family with important roles in embryonic mesodermal and neuroectodermal development as well as angiogenesis and wound repair in the mature organism (1,2). It is a potent mitogen for many cell types, including vascular smooth muscle cells (3). Basic FGF is found in quiescent smooth muscle cells of the arterial media (4), and its release following vascular injury has been implicated in driving the initial proliferative response of medial smooth muscle cells, at least by a paracrine mechanism (5). The concomitant expression of both bFGF and cell surface FGF receptors by smooth muscle cells suggests that autocrine FGF signaling has important phenotypic consequences. Autocrine FGF signaling has been reported to contribute to smooth muscle cell migration (6) and may also participate in proliferation, as has been reported for other cell types (7)(8)(9)(10). Other consequences of autocrine FGF signaling in vascular smooth muscle cells, such as the promotion of cell survival, have not been explored.
A variety of cell types that are either terminally differenti-ated or normally remain in a quiescent state in the adult organism express FGF, but the autocrine functions of FGF in this context have not been elucidated. There is good experimental evidence that FGF can reside in the extracellular matrix around nondividing cells as a storage form, with its mitogenic activity available in the event of tissue damage and a requirement for repair or scar formation (11)(12)(13). However, the widespread expression and intracellular location of FGF in nondividing cells suggests other functions, and has spurred studies of the ability of FGF to promote cell survival. The survivalpromoting activity appears to be distinct from the mitogenic effects of FGF, as it has been documented in nonproliferating cells (14,15), including terminally differentiated cells of neuronal origin (16,17). How FGFs promote survival through an autocrine mechanism in this context has not been defined but may be through the prevention of apoptosis, or programmed cell death. That FGFs are autocrine survival factors has been suggested by studies in vitro (15) and in vivo (18). The notion that exogenous FGFs promote survival by preventing apoptosis, at least through a paracrine mechanism, has been suggested by studies in PC12 cells (19) and endothelial cells (20,21). However, these studies do not distinguish between the survival function of FGFs and either their ability to modulate phenotype or stimulate mitogenesis. These observations, coupled with the presence of bFGF in medial smooth muscle cells of the arterial media, prompted the hypothesis that autocrine bFGF supports smooth muscle cell proliferation, survival, or both. Vascular smooth muscle contributes to normal vessel function, the development of vascular diseases, and the response to vascular injury. There is a growing appreciation that smooth muscle cell apoptosis contributes to vascular development, physiologic adaptation, and vascular remodeling in response to the natural progression or invasive treatment of human vascular disease (22)(23)(24)(25). The role of autocrine FGF signaling in smooth muscle cell proliferation or survival has not been clearly defined. In order to probe the autocrine function of bFGF in vascular smooth muscle cells, endogenous expression was inhibited using an antisense strategy, and the functional consequences were examined.

EXPERIMENTAL PROCEDURES
Recombinant Adenoviruses-The Ad.ASbFGF virus was constructed by replacement of the ␤-galactosidase cDNA with a 1.1-kilobase rat bFGF cDNA containing the entire coding sequence plus 5Ј-untranslated region sequences (gift of Dr. Andrew Baird, Whittier Institute; Gen-Bank accession number M22427) (26)  Recombinant adenovirus was identified by PCR amplification of putative recombinant plaques using one vector-specific and one insertspecific primer. The vector-specific primer (5Ј-AGACATGATAAGATA-CAT-3Ј) corresponds to a region of the shuttle vector upstream of the insert cloning site but contains no adenoviral genomic sequence. The insert-specific primer (5Ј-GCTTCTTCCTGCGCATCC-3Ј) corresponds to codons 37-42 of the 18-kDa rat bFGF coding sequence. The recombinant virus was twice plaque-purified, and expression of the antisense transcript was confirmed by reverse transcription-PCR (RT-PCR). Total cellular RNA (1 g) isolated from uninfected smooth muscle cells or cells infected with either Ad.ASbFGF or Ad5/RSV/GL2 was heat-denatured (65°C for 5 min) and reverse-transcribed using 0.5 pmol of a sequence-specific primer. This primer contains 18 bases (5Ј-ACT-TCGCTTCCCGCACTG-3Ј) of bFGF sequence complementary to the antisense strand (corresponding to codons 7-12 of 18-kDa rat bFGF) at the 3Ј end, as well as a 30-base random sequence at the 5Ј end (5Ј-CTTATACGGATATCCTGGCAATTCGGACTT-3Ј). The 5Ј end random sequence tag permits subsequent PCR amplification with a primer corresponding to this tag sequence only, thus rendering the overall amplification both strand-and RNA-specific (30). The 5Ј primer used for the PCR amplification (5Ј-GCACACACTCCCTTGATGGACAC-3Ј) corresponds to codons 71-78 of 18-kDa rat bFGF. As controls, each of the cellular RNAs was also reverse-transcribed using random hexamer primers (1 pmol), and these cDNA products were subsequently amplified using primers (5Ј-primer: 5Ј-CCTGAAGGGTGGTGCAAAAG-3Ј; 3Јprimer: 5Ј-CCATCCACAGTCTTCTGAGTG-3Ј) for glyceraldehyde-3phosphate dehydrogenase (GAPDH). Reverse transcription was performed in a final reaction volume of 20 l using 20 units of avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim) according to the supplier's recommendations, and controls included AS-bFGF and sense GAPDH RNA transcripts generated in vitro using cDNA templates and bacteriophage T7 RNA polymerase as well as a no-template negative control. The ASbFGF in vitro transcript yields a 246-base pair RT-PCR amplification product that is indistinguishable from that generated from the ASbFGF recombinant adenoviral transcript; the GAPDH cDNA used to generate the in vitro transcript contains a 106-base deletion relative to the native mRNA, which is flanked by the PCR primer sites and thus generates a smaller RT-PCR amplification product (132 base pairs versus 238 base pairs, respectively). Ten percent of each cDNA product was amplified using 1 pmol of each pair of relevant primers, with one primer in each reaction 5Ј-endlabeled using T4 polynucleotide kinase (New England Biolabs, Beverly, MA) and [␥-32 P]ATP (6000 Ci/mmol, Amersham Corp.). The PCR was performed in a 20-l volume using 1 unit/reaction of Taq polymerase (Perkin-Elmer) in an Ericomp thermal cycler (San Diego, CA) for 25 cycles using 1-min steps. Denaturation was done at 94°C, annealing at 60°C, and extension at 72°C. Products were separated on a 6% polyacrylamide, 7 M urea gel and detected by autoradiography using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Cell Culture and Western Blotting for Basic FGF-Cultured rat aortic smooth muscle cells (gift of Dr. Gary Owens, University of Virginia) were cultured in DF12 (equal parts Dulbecco's modified Eagle's and Ham's F12 media) supplemented with 10% fetal bovine serum (FBS) and antibiotics. Subconfluent cells were infected with recombinant adenoviruses and 3 days later harvested with trypsin, homogenized, and partially purified by heparin-agarose affinity chromatography (Bio-Rad, Richmond, CA) essentially as described (31). Lysates from equal numbers of cells were loaded onto the heparin affinity matrix, eluted by boiling in SDS-PAGE sample buffer, separated by electrophoresis in 15% SDS-polyacrylamide gels, and electrophoretically transferred to a polyvinylidene difluoride membrane (Immobilon P; Millipore Corp., Bedford, MA) in Tris-glycine buffer containing methanol and SDS. The three isoforms of rat bFGF (32) were determined by probing with a monoclonal anti-bovine bFGF antibody (Type II, Upstate Biotechnology, Lake Placid, NY) detected by enhanced chemiluminescence (ECL; Amersham Corp.) using a secondary anti-mouse antibody conjugated with peroxidase, as recommended by the manufacturer. Similar results were obtained by immunoblotting whole cell lysates without prior affinity enrichment, and reprobing these blots with an antibody directed against ␤-actin demonstrated equivalent protein loading (data not shown). For the bFGF and epidermal growth factor (EGF) rescue experiments, 25 ng/ml recombinant human bFGF (Upstate Biotechnology) or 50 ng/ml human EGF (Upstate Biotechnology) was added directly to each well daily. Tissue culture reagents were from Life Technologies, Inc. unless otherwise specified. Electrophoresis and other chemicals were from Boehringer Mannheim, Sigma, or Fisher (Fairlawn, NJ).
Assessment of Apoptosis-For assessment of cellular and nuclear morphology, cells were grown on glass coverslips, infected with recombinant adenoviruses, and 48 -72 h later fixed in 4% paraformaldehyde. Coverslips were imaged without staining under phase contrast optics or stained with 0.015% propidium iodide (Sigma) and imaged with epifluorescence illumination. The percentage of apoptotic cells was estimated by counting the numbers of apoptotic (condensed or fragmented) nuclei in at least 1000 cells imaged in contiguous fields on each coverslip, and the mean Ϯ S.D. was calculated from determinations made on three independent coverslips. For 5-bromo-2Ј-deoxyuridine (BrdU) labeling, cells were pulse-labeled with 10 M BrdU (Sigma) for 2 h prior to harvest, fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 (Sigma), and stained for BrdU using a fluorescein isothiocyanate-coupled monoclonal anti-BrdU antibody (Becton-Dickenson, San Jose, CA) according to the manufacturer's instructions. The BrdUpositive cells were imaged by epifluorescence microscopy, the percentage of positive cells was counted in at least 1000 cells in contiguous fields on each coverslip, and the mean Ϯ S.D. was calculated from determinations made on three independent coverslips. For the detection of DNA strand breaks, cells were grown and infected on glass coverslips, fixed in 4% paraformaldehyde, labeled using the fluoresceinbased ApopTag kit (Oncor, Gaithersburg, MD) according to the manufacturer's instructions, and imaged by epifluorescence microscopy. Photomicrographs were obtained using Kodak Ektachrome film (Rochester, NY) and either a Zeiss Optiphot or an Olympus AX-70 microscope at ϫ 200 -400 magnification using phase contrast optics or epifluorescence with appropriate filters. For the detection of oligonucleosomal fragmentation, cells were collected with trypsin and subjected to hypotonic lysis in buffer containing Triton X-100, and a low molecular weight DNA fraction was isolated as described (33). The low molecular weight fraction from equal numbers of cells was separated by electrophoresis through agarose, stained with ethidium bromide, and visualized by transillumination with UV light using an Eagle Eye apparatus (Stratagene, La Jolla CA). Total cellular DNA isolated from v-myc-transformed rat embryo fibroblasts exposed to 10 Gy of ␥-radiation (gift of E. Bernhard) was used as a positive control for oligonucleosomal fragmentation.
Cell Cycle Phase Determination and Tritiated Thymidine Labeling-For flow cytometric analysis of cell cycle phase distribution, cells were infected with recombinant adenoviruses and harvested with trypsin, and nuclei were prepared and stained with propidium iodide as described (34). Flow cytometric analysis was performed using a FACScan cytometer (Becton-Dickenson, San Jose, CA) and analyzed using Mod-FitLT curve-fitting software (Verity Software, Topsham, ME) running on a Macintosh computer (Apple, Cupertino, CA). For tritiated thymidine labeling, sham-or adenovirus-infected cells were labeled 2 days following infection for 12 or 18 h in serum-free medium containing 1-2 Ci/ml [methyl-3 H]thymidine (Amersham Corp.), fixed in trichloroacetic acid, solubilized in dilute NaOH, and collected by filtration, and incorporation was determined by scintillation counting (Beckman Instruments, Columbia, MD).

Antisense Inhibition of Endogenous bFGF Triggers Smooth
Muscle Cell Death-Expression of an antisense bFGF RNA (Fig. 1A) by a recombinant adenovirus (Ad.ASbFGF) in cultured smooth muscle cells markedly reduced cellular bFGF content (Fig. 1B). To test the hypothesis that inhibition of endogenous bFGF expression reduces either smooth muscle cell proliferation or survival, cell number was monitored following Ad.ASbFGF infection (Fig. 2). The expected accumulation of smooth muscle cells cultured in 10% FBS was inhibited by Ad.ASbFGF in a dose-dependent fashion (panel A) and could be prevented by exogenous bFGF (panel B). In order to distinguish between an inhibition of proliferation and promotion of cell death, quiescent smooth muscle cells were infected with the antisense bFGF virus and maintained in low (0.5%) serum. The number of antisense-infected cells decreased in a dose-dependent fashion (panel C), indicating cell death, which was subsequently identified as apoptosis by accepted morphologic and biochemical criteria.
Criteria for Apoptosis in Cells Infected with the Antisense bFGF Adenovirus-Cells infected with the antisense bFGF virus appeared normal for the first 24 h, but thereafter alter-ations in cellular and nuclear morphology and evidence of DNA fragmentation were observed (Fig. 3). Cells became rounded, developed vesiculations around the cell borders (panels A and B), and displayed condensation of nuclear chromatin, nuclear fragmentation (panels C and D), and evidence of DNA strand breaks (panels E-H). Agarose gel electrophoresis of the low molecular weight DNA (soluble cytoplasmic fraction) isolated from equal numbers of smooth muscle cells showed prominent oligonucleosomal fragmentation in Ad.ASbFGF-infected cells, which was barely detectable in control virus-infected cells and was undetectable in uninfected cells (Fig. 4). Nonspecific viral toxicity as a cause of cell death was excluded by the normal appearance of cells infected with similar doses of control viruses, in contrast to the rapid killing of cells (Ͻ12 h) intentionally infected at a very high multiplicities of infection, which was characterized by cell swelling and lysis. The delay of Ͼ24 h between infection with Ad.ASbFGF and the appearance of apoptosis suggests that cells are initially protected by existing bFGF protein and that the content of bFGF must fall below a critical level before apoptosis is initiated.
Apoptosis can be induced in some cell types by deprivation of specific growth factors (19,20,(35)(36)(37). Other cell types, including vascular smooth muscle cells (38), survive well when cultured in serum-free medium but undergo apoptosis when forced into S phase in the absence of serum by constitutive expression of cell cycle transcription factors like c-myc (39,40) or E2F (41,42). Apoptosis triggered by antisense inhibition of bFGF production might resemble a growth factor withdrawal model if it could be prevented by heterologous growth factor signaling pathways (43). Alternatively, it might conform to the second type of apoptotic mechanism involving inappropriate S phase entry. To distinguish between these two possibilities, the ability of a different growth factor to prevent apoptosis following infection with Ad.ASbFGF was assessed, and the degree of S phase entry was estimated by flow cytometry, rates of DNA synthesis, and BrdU labeling of apoptotic nuclei in Ad.AS-bFGF-infected cells.
Epidermal Growth Factor Does Not Prevent Apoptosis Triggered by Ad.ASbFGF-EGF, similar to exogenous bFGF, was able to prevent the reduction in the number of Ad.ASbFGFinfected cells over time in either high or low serum (data not shown). However, the ability of added growth factors to prevent  1, -C). Each of the cellular RNAs was also analyzed for a control mRNA transcript, GAPDH, using a synthetic GAPDH transcript containing a short deletion (GAPDH del ) as a positive control. This experiment was repeated at least five times using independent RNA samples and alternative primer sets with similar results. the reduced accumulation of cells over time does not distinguish between the prevention of cell death and the stimulation of proliferation in the surviving cells. In order to make this distinction, the extent of apoptosis in the presence or absence of growth factors was assessed by the percentage of apoptotic nuclei visualized by propidium iodide staining. In cells cultured in 0.5% FBS in the absence of growth factors and infected at an MOI of 1000 with either Ad.ASbFGF or control virus, the percentage of apoptotic nuclei in control virus-infected cells was 0.5 Ϯ 0.5% (mean Ϯ S.D. of triplicate samples) compared with 18.9 Ϯ 4.4% in Ad.ASbFGF-infected cells, which was also associated with a 20% decrease in cell number. In Ad.ASbFGFinfected cells treated with bFGF, the percentage of apoptotic nuclei was only 8 Ϯ 1.1% compared with 15.5 Ϯ 4.7% in cells treated with EGF, despite a similar (2-fold) increase in cell number stimulated by either growth factor in both control virus-or Ad.ASbFGF-infected cells. These data show that despite a similar stimulation of proliferation in non-apoptotic cells, bFGF but not EGF was able to reduce the proportion of apoptotic cells. This result suggests that apoptosis triggered by inhibition of autocrine bFGF synthesis does not correspond to a growth factor withdrawal model of apoptosis or that it involves signaling pathways that are not shared by bFGF and EGF. An alternative hypothesis, that apoptosis triggered by inhibition of autocrine bFGF synthesis involves disruption of normal cell cycle regulation, was tested by examining several complementary measures of cell cycle activity.

Basic FGF cell content was determined by Western (immunoblot) analysis (B). Both uninfected cells (lane 1) and cells infected with
Apoptosis Induced by Ad.ASbFGF Involves Inappropriate S Phase Entry-Analysis of cell cycle activity in Ad.ASbFGFversus control virus-infected smooth muscle cells suggested that an inappropriate entry into S phase participates in the mechanism of apoptosis due to inhibition of autocrine bFGF synthesis. By flow cytometric analysis, Ad.ASbFGF-infected cells displayed a marked increase in the proportion of cells in S phase at the expense of the G 0 /G 1 population (Fig. 5), and a time course experiment demonstrated an abrupt transition from a normal cell cycle phase distribution to one characterized by an increase in S phase between 16 and 20 h after infection (data not shown). These findings were confirmed by measuring rates of DNA synthesis by tritiated thymidine labeling (Fig. 6), which were greater in Ad.ASbFGF-infected versus control cells. The magnitude of the difference in thymidine labeling appeared to be dose-related (Fig. 6A), and at the higher MOI thymidine incorporation was about 2-fold the value in control cells. This result could be interpreted to mean that twice as many Ad.ASbFGF-infected cells were incorporating thymidine as control cells, an interpretation consistent with the flow cytometry results. This suggests that the excess DNA synthesis in the Ad.ASbFGF-infected cells was due to apoptosis, and this was made more apparent by thymidine labeling in quiescent cells (Fig. 6B). Under these circumstances, the thymidine incorporation in the control cells is very low, and the difference in labeling is apparently due to DNA synthesis in the proportion of cells undergoing apoptosis. This conclusion was supported by BrdU pulse labeling and immunofluorescence imaging of cells that had been infected with either Ad.ASbFGF or control viruses (Fig. 7), which showed that approximately 20% of cells were BrdU-labeled and that about half of these labeled cells had apoptotic (condensed or fragmented) nuclei, demonstrating that the increased DNA synthesis was associated with apoptosis and was not simply a compensatory response of the surviving cells. These data demonstrate that apoptosis induced by inhibition of bFGF synthesis involves an inappropriate entry into S phase, similar to other models of apoptosis involving disrupted cell cycle regulation, and suggests that autocrine FGF signaling participates in cell cycle regulation as part of its survival function. DISCUSSION This study demonstrates that inhibition of endogenous basic FGF synthesis using an antisense strategy triggers apoptosis in cultured vascular smooth muscle cells, supporting an essential survival function for autocrine FGF signaling in this cell type. Prior studies implicating fibroblast growth factors as survival factors in mesodermal and neuroectodermal cells (14,16,17,19,20,31,44) suggest that this survival function is related to prevention of apoptosis (19,20,44) but do not distinguish between mitogenic and survival-promoting activities.
Overexpression of FGF can lead to a transformed phenotype (45)(46)(47)(48)(49)(50), supporting the idea that autocrine FGF signaling can drive proliferation, but interactions between FGFs and transmembrane FGF receptors may be required for FGF-stimulated mitogenesis (45)(46)(47)(48)(49). On the other hand, selective overexpression of the high molecular weight isoforms of bFGF can either promote or inhibit proliferation without evidence of receptor interaction (50,51). None of these studies establish that autocrine FGF signaling is either necessary or sufficient for proliferation in cells capable of such autocrine loops. The mechanism of apoptosis in the present study appears to be independent of proliferation, as it occurs in cells cultured in either high (10%) or low (0.5%) serum. The ability of exogenous bFGF to prevent apoptosis may be related to activation of receptor-mediated signal transduction pathways or to internalization of ligand per se, although the inability of epidermal growth factor to block this mechanism of apoptosis suggests that receptor-mediated signaling may be less important. Thus the survival function of endogenous bFGF in smooth muscle cells appears to involve signal transduction pathways that may be distinct from those mediating mitogenesis. Experiments in transfected skeletal myoblasts (52) and fibroblasts (51) support important differences between FGF signaling pathways controlling mitogenesis and other cell functions.
The data presented establish that the pathway of apoptosis triggered by inhibition of autocrine FGF signaling involves an inappropriate entry into S phase. The possibility that DNA repair mechanisms contribute to the observed increase in total DNA synthesis has not been excluded, but the cell cycle phase analysis suggests that S phase entry is the predominant, if not exclusive, mechanism responsible. In this aspect it resembles other models of apoptosis that are due to disruption of normal cell cycle regulation. Unregulated expression of c-myc (39,40) or E2F (41,42) induces an inappropriate S phase entry and triggers apoptosis in the absence of serum. These pathways of apoptosis can be understood, at least in the case of E2F, by alteration in the regulatory balance between E2F and the retinoblastoma gene product, normally regulated by cyclins and cyclin-dependent kinase inhibitors. It is possible that expression of adenoviral genes in conjunction with the antisense bFGF transcript could be contributing to the observed phenotype. The adenoviral protein E1A interacts with cell cycle regulators such as the retinoblastoma protein and is a recognized inducer of apoptosis (22,53). Although the recombinant viruses used in this study are specifically E1-deleted, contamination with E1-expressing virus, although highly unlikely, has not been rigorously excluded. Although a direct relationship between FGF signaling and the activities of cell cycle regulatory proteins has not been demonstrated, the present study provides more direct evidence of the ability of autocrine FGF signaling to influence cell cycle regulation than has previously been reported.
Antisense inhibition of endogenous bFGF expression leads to an interruption of autocrine FGF signaling that appears to alter cell cycle regulation, resulting in apoptosis of vascular smooth muscle cells. These observations provide a mechanism that could explain the survival function of FGFs expressed in quiescent or terminally differentiated cells. The expression of FGFs is widely detectable in mesodermal and neuroectodermal lineages of adult organisms, and autocrine FGF signaling may well play a role in the cellular homeostasis and survival of these cell types. The present work provides insights into the autocrine functions of FGF and the signaling pathways whereby endogenous FGF prevents apoptosis are currently being explored.