Inhibition of capillary morphogenesis and associated apoptosis by dominant negative mutant transforming growth factor-beta receptors.

Transforming growth factor-β1 (TGF-β1) induces angiogenesis in vivo and capillary morphogenesis in vitro. Two receptor serine/threonine kinases (types I and II) have been identified as signal transducing TGF-β receptors. We explored the possibility of inhibiting TGF-β- mediated events in glomerular capillary endothelial cells using a TGF-β type II receptor (TβR-II) transdominant negative mutant. A mutant TGF-β type II receptor (TβR-IIM), lacking the cytoplasmic serine/threonine kinase domain, was produced by polymerase chain reaction using rat TβR-II cDNA as template. Since TβR-II and TGF-β type I receptor (TβR-I) heterodimerize for signal transduction, the mutant receptor competes for binding to wild-type TβR-I, hence acting in a dominant negative fashion. Glomerular capillary endothelial cells were stably transfected with TβR-IIM, and four independent clones were expanded. That the TβR-IIM mRNA was expressed was shown by reverse transcriptase-polymerase chain reaction, RNase protection assay, and Northern analysis. Presence of cell surface TβR-IIM protein was shown by affinity cross-linking with 125I-TGF-β1. In wild-type endothelial cells, TGF-β1 (2 ng/ml) significantly inhibited [3H]thymidine incorporation to 63 ± 10% of control (n = 4). In transfected endothelial cells carrying TβR-IIM, TGF-β1 stimulated [3H]thymidine incorporation to 131 ± 9% of control (n = 4, p < 0.005). Also, in wild-type endothelial cells, endogenous and exogenous TGF-β1 induced apoptosis and associated capillary formation. Both apoptosis and capillary formation were uniformly and entirely absent in transfected endothelial cells carrying TβR-IIM. This represents the first demonstration that capillary morphogenesis in vitro is associated with apoptosis, and that interference with TβR-II signaling inhibits this process in glomerular capillary endothelial cells.

Angiogenesis, the process of new blood vessel formation, is an integral part of development, wound repair, and tumor growth. The formation of capillary networks requires a complex series of cellular events, in which endothelial cells locally de-grade their basement membrane, migrate into the connective tissue stroma, proliferate at the migrating tip, elongate and organize into capillary loops (1). In response to angiogenic stimuli, endothelial cells in culture develop networks of capillary-like tubes.
In the early stages of angiogenesis, proteases are required for extracellular matrix (ECM) proteolysis to facilitate endothelial cell migration (9). TGF-␤1 induces endothelial cell secretion of plasminogen activator (PA) which activates plasmin, a protease that degrades ECM proteins (10,11). Increased production of PA has been associated with the invasive properties of cultured endothelial cells in response to angiogenic stimuli (10,11). In addition, plasmin activates latent TGF-␤1 (12), in an autocrine fashion. Furthermore, TGF-␤1 is a potent chemoattractant for macrophages and fibroblasts (13,14), which are postulated to release angiogenic peptides in vivo, such as basic fibroblast growth factor (bFGF), platelet-derived growth factor, or tumor necrosis factor-␣ (15).
Two transmembrane serine/threonine kinases, types I and II, have been identified as signal transducing TGF-␤ receptors. TGF-␤ type II receptor (T␤R-II), a constitutively active kinase, directly binds TGF-␤1, and this ligand binding results in the recruitment and phosphorylation of TGF-␤ type I receptor (T␤R-I) to produce a heteromeric signaling complex (16). T␤R-I alone is unable to bind TGF-␤1, and T␤R-II is unable to signal without T␤R-I (17).
We explored the possibility of inhibiting TGF-␤1-mediated events in renal glomerular capillary endothelial cells using a T␤R-II transdominant negative mutant. A mutant T␤R-II construct (T␤R-II M ), lacking the cytoplasmic serine/threonine kinase domain, but with full transmembrane spanning and extracellular domains, was produced by polymerase chain reaction (PCR) using rat T␤R-II cDNA (18) as template. Since T␤R-II and T␤R-I heterodimerize for signal transduction, the mutant receptor competes for binding to wild-type T␤R-I, hence acting in a dominant negative fashion (19 -21). When the trans-dominant negative mutant construct was stably expressed in glomerular capillary endothelial cells, capillary morphogenesis and associated apoptosis were entirely blocked in these cells.

EXPERIMENTAL PROCEDURES
Cell Culture-Glomerular capillary endothelial cells were isolated from bovine kidney cortex as described previously (22), with the following modifications. After collagenase digestion, the cells were plated at low density on gelatin-coated plates, in RPMI 1640 medium containing 15% fetal bovine serum (FBS) to which 8 ng/ml acidic fibroblast growth factor (aFGF) (R & D Systems), 0.1 g/ml heparin, and 5 units/ml penicillin, and 5 g/ml streptomycin were added. The aFGF was used to stimulate endothelial cell proliferation, and the heparin both to increase the affinity of aFGF for endothelial cell FGF receptors and to inhibit mesangial cell growth. Colonies of endothelial cells were subjected to two rounds of cloning to establish cell lines free of contaminating mesangial cells. Once the cells were established in culture, they were maintained in RPMI 1640 with 15% FBS and 5 units/ml penicillin, and 5 g/ml streptomycin. Cells between passages 5 and 15 were used for experiments described herein. That the cells are endothelial cells was documented for each isolation by labeling with fluorescent acetylated low density lipoprotein (Biomedical Technologies Inc.).
To induce capillary tube formation, cells grown to confluence were placed in RPMI medium containing 0.5% FBS, in the presence or absence of exogenous 2 ng/ml TGF-␤1 (Collaborative Biomedical Products). For transmission electron microscopy, the cells were fixed in 3% glutaraldehyde in phosphate-buffered saline, postfixed in 1% osmium tetroxide, dehydrated in a graded series of ethanol, and embedded in epon directly on the culture dish, in order to preserve morphology of the capillary tubes. Thin sections were cut, stained with uranyl acetate and lead citrate, and subjected to electron microscopy (Paragon Biotech Inc., Baltimore, MD). In experiments with neutralizing antibody to TGF-␤1, cells grown to confluence were placed in RPMI medium containing 0.5% FBS in the presence or absence of 10 ng/ml turkey anti-human TGF-␤1 IgG (Collaborative Biomedical Products).
Mutant T␤R-II Construct-A truncated TGF-␤ type II receptor construct (T␤R-II M ), lacking the cytoplasmic serine/threonine kinase domain, but with full transmembrane spanning and extracellular domains, was generated by PCR using a rat T␤R-II cDNA as the template (Fig. 1). Primer sequences were as follows: sense primer 5Ј-GTTAAG-GCTAGCGACGGGGGCTGCCATG-3Ј; antisense primer 5Ј-GGCGGTC-GACTAGACACGGTAACAGTAGAAG-3Ј; and contained the sequences for the restriction enzymes NheI and SalI, respectively (underlined), for directional cloning, and a stop codon in the antisense primer. Each 100-l PCR reaction mixture contained 1 ng of template DNA, 0.25 M primers, 0.05 mM dNTPs, 0.75 mM MgCl 2 , 1 ϫ PCR buffer II (Perkin-Elmer), and 2 units of Taq polymerase (Perkin-Elmer). Amplification consisted of initial denaturation at 95°C for 1 min, followed by 25 cycles (15 s at 95°C, 15 s at 50°C, and 15 s at 72°C) in GeneAmp PCR System 9600 (Perkin-Elmer). This reaction product was gel-purified and cloned with NheI and SalI into pMAMneo (CLONTECH), a glucocorticoidinducible mammalian expression vector. That the clone contained correct directionality and in-frame sequences of the PCR product were verified by restriction mapping with EcoRI, BamHI, and HindIII, and sequencing by dideoxy chain termination technique using Sequenase 2.0 (U. S. Biochemical).
Stable Transfection of Glomerular Capillary Endothelial Cells-To generate clones that stably expressed T␤R-II M , glomerular capillary endothelial cells were transfected by using Lipofectin (Life Technologies, Inc.) as follows. Cells grown to approximately 50% confluency on 6-well plates were washed with RPMI, then incubated with 1-5 g of DNA (T␤R-II M ligated in pMAMneo) in RPMI and 5-10 l of Lipofectin suspension for 5 h at 37°C in a 5% CO 2 atmosphere. Control cells were incubated with pMAMneo vector (not containing T␤R-II M ) and Lipofectin. Following a 5-h incubation, medium containing 20% FBS in RPMI was added to each well to make a final concentration of 10% FBS, and incubated further for 48 h. Then the medium was changed to 10% FBS in RPMI (no antibiotics) and incubated for another 24 h. To select for stable transfectants, cells were treated with 400 g/ml Geneticin (Life Technologies, Inc.) in RPMI medium containing 15% FBS, and the medium was changed every 2-3 days. Clones emerged at approximately 14 days after lipofection. Stably transfected clones were subcloned using ring cylinders, expanded, and maintained in RPMI medium containing 15% FBS, 200 g/ml Geneticin, 5 units/ml penicillin, and 5 g/ml streptomycin. Four independent, stably transfected clones containing T␤R-II M and 4 clones containing empty vector were expanded. Non-transfected glomerular endothelial cells similarly treated with Lipofectin served as additional controls.
Solution Hybridization/RNase Protection-RNase protection analysis was done using the RPA II kit (Ambion) according to the manufacturer's instructions. The 32 P-labeled antisense RNA probe was prepared from the linearized plasmid containing a fragment of the rat T␤R-II cDNA using T7 RNA polymerase, yielding a probe 488 nucleotides (nt) long. 20 g of total RNA from wild-type and transfected cells were hybridized with the 32 P-labeled probe. Hybridization was for 16 -18 h at 42°C in 50% formamide, 5 ϫ SSPE, 0.1 M Tris, pH 7.4, and 50 g/ml salmon sperm DNA. The samples were then digested with RNase A/T1, and resolved on a 6% acrylamide, 7.7 M urea sequencing gel. A sample of 32 P-labeled 1-kb ladder DNA was loaded in adjacent lanes as the molecular size marker.
Northern Blot Analysis-Total RNA from cells grown in the absence or presence of 1 M dexamethasone (Sigma) was isolated by lysis with TRI reagent (Molecular Research Center, Inc.) according to the manufacturer's instructions, and size fractionated (30 g/lane) on a 1% agarose, 2% formaldehyde gel in 20 mM MOPS, 5 mM sodium acetate, and 1 mM EDTA, pH 7.2. Messenger RNA was transferred to a nylon membrane (Nytran, Schleicher & Schuell) and UV linked to the membrane. The blot was prehybridized at 65°C using 1% bovine serum albumin (Sigma), 7% SDS, 0.5 M phosphate buffer, 1 mM EDTA, pH 8.0, and 100 g/ml heat-denatured salmon sperm DNA for 2 h, hybridized in the same solution containing the appropriate 32 P-labeled cDNA at 65°C overnight, followed by two 30-min washes at 65°C with 0.5% bovine serum albumin, 5% SDS, 40 mM phosphate buffer, 1 mM EDTA, pH 8.0, then four 15-min washes with 1% SDS, 40 mM phosphate buffer, 1 mM EDTA, pH 8.0, at 65°C. The membrane was then exposed to Kodak X-AR 5 film for 25-48 h. The T␤R-II probe is a 2.8-kb rat T␤R-II full-length cDNA (17) which was labeled with [ 32 P]dCTP using random primer labeling system (Life Technologies, Inc.). The cross-linking reaction was quenched by washing three times with cold 250 mM sucrose, 10 mM Tris, pH 7.4, 1 mM EDTA. The cells were lysed with 100 l of 1% Triton X-100, 10 mM Tris, pH 7.4, 1 mM EDTA, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 50 g/ml aprotinin, 10 g/ml pepstatin, and subjected to centrifugation at 13,000 ϫ g for 30 min to remove particulate matter. Sample loading buffer (sucrose, 0.01% bromphenol blue, 2% ␤-mercaptoethanol, 5 mM EDTA) was added (1:1, v/v) and boiled for 5 min, followed by 10% SDS-polyacrylamide gel electrophoresis. The gel was stained with Coomassie Brilliant Blue (Bio-Rad) to visualize equivalence in protein loading, and destained prior to autoradiography.

Covalent Labeling of TGF-␤ Receptors-Cells
[ 3 H]Thymidine Incorporation-10 4 cells were plated in 24-well dishes and incubated in medium containing 15% FBS and grown to subconfluence. Medium was then changed to serum-free RPMI for 24 h, followed by incubation in 0.5% FBS in the presence or absence of TGF-␤1 (2 ng/ml). After 45 h, the medium was removed, and cells were exposed for 3 h to 1 Ci/ml [ 3 H]thymidine in RPMI 1640 medium containing 2% bovine platelet-poor plasma derived serum, at 37°C. The cells were washed three times with RPMI, and then extracted three times with ice-cold 6% trichloroacetic acid, followed by solubilization in 1 N NaOH and counted in a Packard Liquid scintillation counter. For determination of the time course of [ 3 H]thymidine incorporation, similar methods were utilized and the cells were extracted at the various time points, with 3-h [ 3 H]thymidine exposure prior to each period.
Genomic DNA Isolation and Analysis-Cells were plated on 150-mm plates, and grown to confluence in medium containing 15% FBS, then incubated in the presence or absence of 1 M dexamethasone for 24 h. The medium was then changed to 0.5% FBS and incubated in the presence or absence of TGF-␤1 (2 ng/ml) at 37°C for 5 days. Genomic DNA isolation was performed using Puregene (Gentra) according to the manufacturer's instructions. Briefly, cells were lysed directly on the plate by removing the culture medium and adding the cell lysis solution, followed by incubation with RNase A, then protein precipitation solution. The samples were centrifuged at 2,000 ϫ g for 10 min and supernatant transferred to new tubes. Precipitated DNA was resuspended in DNA hydration solution, and quantitated by UV spectrophotometer. 20 g of DNA was analyzed with a 1.5% agarose gel electrophoresis.

RESULTS
Expression of T␤R-II M mRNA-To demonstrate that the transfected glomerular endothelial cells expressed T␤R-II M mRNA, RNase protection assay was performed using antisense RNA prepared from the rat T␤R-II cDNA with some flanking vector sequence. The rat T␤R-II probe contained 488 nt of authentic rat T␤R-II sequence, which included only 211 nt of T␤R-II M , and predicted to hybridize fully with rat (but not bovine) T␤R-II mRNA. Based on the size of the T␤R-II cDNA probe, the expected size of the protected fragment produced from transfected rat T␤R-II M cDNA sequence was 211 nt. As shown in Fig. 2A, hybridization with the rat T␤R-II antisense riboprobe protected a fragment 211 nt in length from RNase digestion in transfected cells. No protected fragment was seen in the wild-type bovine glomerular endothelial cells or mock transfected cells with vector alone. Expression of T␤R-II M mRNA in glomerular endothelial cells was also demonstrated by RT-PCR utilizing the sense and antisense primers to rat cDNA sequence used in producing the mutant construct (data not shown), and four independent transfected clones expressing T␤R-II M mRNA were propagated for all subsequent experiments.
Northern blot analysis of total RNA isolated from wild-type and transfected glomerular endothelial cells, probed with T␤R-II cDNA, showed a 5.5-kb band in all cells (Fig. 2B), corresponding to wild-type T␤R-II. A new 1.8-kb band, corresponding to T␤R-II M , was observed only in cells transfected with T␤R-II M , and not present in wild-type or mock transfected cells, and was strongly induced by dexamethasone. Also, a modest induction of the 5.5-kb wild-type T␤R-II mRNA was seen with dexamethasone.
Cell Surface Expression of T␤R-II M -Affinity cross-linking with 125 I-TGF-␤1 in wild-type glomerular endothelial cells detected two distinct bands with molecular masses of approximately 89 and 70 kDa corresponding with T␤R-II and T␤R-I, respectively (Fig. 3)  Effect of Serum Deprivation and TGF-␤1 on Capillary Morphogenesis-In cultured wild-type glomerular endothelial cells, with serum deprivation in the presence or absence of dexamethasone pretreatment, many of the cells detached from their substratum while remaining cells organized into capillary-like structures (Fig. 5A). Furthermore, treatment with exogenous TGF-␤1, in the presence or absence of dexamethasone, also induced cell detachment and formation of capillary-like structures (Fig. 5B). These events were accelerated by approximately 48 h when compared to those cells under serum deprivation alone. Both cell detachment and formation of capillarylike structures were observed with serum deprivation in mock transfected cells carrying empty vector. In contrast, cell detach-ment and formation of capillary-like structures were uniformly and entirely absent in transfected cells carrying T␤R-II M treated either with serum deprivation (Fig. 5C) or exogenous TGF-␤1 (Fig. 5D). When examined with a high power phasecontrast objective, after 5 days of culture the cellular cords formed by wild-type glomerular endothelial cells appeared as tubes that contained a central translucent lumen-like space along their length (Fig. 6A), similar to the in vitro angiogenesis models described by Ingber and Folkman (23) and Montesano et al. (11). Lumen formation was confirmed by transmission electron microscopy, which revealed groups of endothelial cells that were joined by interdigitated cell processes and enclosed a central lumenal space, as shown in Fig. 6B. Amorphous material within the lumen, also previously described by Ingber and Folkman (23), likely represents matrix and debris. Clathrincoated pits and vesicles, as well as cell junctional complex are also observed. Fig. 7 shows genomic DNA size analysis. In wild-type glomerular endothelial cells treated either with serum deprivation or exogenous TGF-␤1, DNA fragmentation was observed. This occurred both with and without dexamethasone pretreatment. Genomic DNA fragmentation was absent in transfected glomerular endothelial cells carrying T␤R-II M . Fig. 8 shows serum-deprived wild-type glomerular endothelial cells, incubated in the absence (panel A) or presence (panel B) of neutralizing antibody to TGF-␤1. Cell detachment and formation of capillary-like structures were observed with serum deprivation alone. However, with the addition of neutralizing antibody to TGF-␤1, both cell detachment and formation of capillary-like structures were not observed. Additionally, neutralizing antibody to TGF-␤1 inhibited DNA fragmentation in serum-deprived wild-type glomerular capillary endothelial cells (data not shown).

DISCUSSION
This study sought to explore the role of TGF-␤ receptors in capillary morphogenesis, using renal glomerular capillary endothelial cells stably transfected with a T␤R-II construct designed to inhibit T␤R-II dependent signals through a transdominant negative action. TGF-␤ receptors types I and II are co-expressed by most cells (24), and heterodimerize upon ligand binding (16,17). Heterodimer formation was recently shown to induce phosphorylation of T␤R-I at the GS domain, an effect dependent on T␤R-II kinase activity. Phosphorylation of T␤R-I by T␤R-II is thought to be essential for the propagation of TGF-␤1 signals (16). Mutant T␤R-II lacking the cytoplasmic signaling domain can be predicted to inhibit T␤R-II dependent signals by virtue of competition for T␤R-I binding, as long as the ability to heterodimerize is conserved. Chen et al. (19) previously showed that mutant T␤R-II lacking the cytoplasmic kinase domain overexpressed in a dominant negative fashion selectively blocked T␤R-II signaling for inhibition of cell proliferation. Wieser et al. (20) demonstrated that truncated T␤R-II lacking the cytoplasmic domain was able to bind TGF-␤, and form a complex with T␤R-I, but failed to inhibit cell proliferation, activate extracellular matrix synthesis, or activate transcription from a promoter containing TGF-␤-responsive elements. Moreover, mutant T␤R-II with kinase domain deleted was shown to confer resistance to TGF-␤ control of developmentally regulated cardiac genes (21). In this study, essentially the same construct was utilized, and stably expressed in cultured renal glomerular capillary endothelial cells. Expression of T␤R-II M mRNA in transfected glomerular capillary endothelial cells was demonstrated by reverse transcriptase-PCR, RNase protection assay, and Northern blot analysis. Although the T␤R-II M construct was under a glucocorticoid-regulated promoter, expression was observed even in uninduced conditions, although at lower levels than that in the presence of dexamethasone. That such promoters "leak" during uninduced conditions has previously been observed by others (25).
To demonstrate cell-surface expression and ligand-binding by the mutant receptor, intact cells were incubated with 125 I-TGF-␤1 followed by affinity cross-linking and analysis of labeled proteins by SDS-polyacrylamide gel electrophoresis. In untransfected or mock transfected cells, only the wild-type T␤R-I and T␤R-II, approximately 70 and 89 kDa, respectively, were observed. In transfected cells, two additional bands of 48 and 36 kDa in size were also observed. Mutant receptor has a predicted molecular mass of 23 kDa. The 48-kDa band is interpreted to represent the mutant receptor with TGF-␤ dimer bound to it, the 36-kDa band could represent the same mutant receptor with TGF-␤1 monomer or possibly a degradation product of the same receptor. These data are interpreted to show that the mutant receptor is expressed at the cell surface and that it can bind TGF-␤1.
Cross-linking to wild-type T␤R-I was less in the transfected cells carrying T␤R-II M (Fig. 3), when compared to untransfected or mock transfected cells. A plausible explanation is that the mutant receptor competes and heterodimerizes with wildtype T␤R-I and is then quickly degraded, internalized, or secreted. Alternatively, the mutant T␤R-II could interfere with TGF-␤ binding to T␤R-I. Inhibition of binding to wild-type T␤R-II was not observed under these conditions. This is not unexpected since wild-type T␤R-II can bind TGF-␤1 in the absence of dimerization with T␤R-I (16,17).
Previous in vitro studies have demonstrated that TGF-␤1 inhibits proliferation of many cell types (26,27), including endothelial cells (28). In wild-type or mock transfected glomerular endothelial cells, TGF-␤1 (2 ng/ml) significantly inhibited  (Fig. 4B). In comparison, 5% FBS maximally stimulated [ 3 H]thymidine incorporation between 12 and 24 h (data not shown). These findings suggest that the stimulation may be a secondary rather than a direct mitogenic effect. Our results of stimulation of DNA synthesis are consistent with previous observations that when cells are released from the negative growth regulatory control of TGF-␤1, cell proliferation occurs and the potential for tumorigenesis can emerge (27,29).
TGF-␤1 inhibits proliferation by arresting cells in the G 1 phase and thus interrupting progression through the cell cycle. Laiho et al. (30) observed that TGF-␤1 prevented phosphorylation of the retinoblastoma gene product and arrested cells in late G 1 phase of the cell cycle. The underphosphorylated retinoblastoma gene product has growth-suppressive function. When progression through the cell cycle is prevented, cells may remain quiescent or withdraw from the cell cycle and undergo terminal differentiation. Indeed, Zentella et al. (31) showed that TGF-␤1 inhibited cell cycle progression of skeletal myoblasts through the G 1 phase and induced terminal differentiation.
In addition to the inhibition of [ 3 H]thymidine incorporation in wild-type glomerular endothelial cells, cell detachment was observed. Cells that remained on the plate tended to organize into capillary-like structures, a process previously described by others (22,32,33). A possible explanation for the observed decreased [ 3 H]thymidine incorporation and cell detachment may be cell cycle arrest and withdrawal, and entry into a suicide program, or apoptosis. In support of this, proto-oncogene c-myc, an important positive regulator of cell growth induced during the G 0 /G 1 phase of cell cycle, can induce apoptosis under conditions of growth arrest, such as presence of a negative growth regulator, TGF-␤1 (34).
Indeed, TGF-␤1 has been shown to inhibit cell proliferation and induce apoptosis in rat hepatocytes in vivo (35) and in rabbit uterine epithelial cells in vitro (36). We observed in wild-type glomerular endothelial cells, serum deprivation induced apoptosis as shown by genomic DNA fragmentation and associated capillary formation. Exogenous TGF-␤1 treatment accelerated these events. In contrast, genomic DNA fragmentation and associated capillary formation were not observed in transfected endothelial cells expressing T␤R-II M , either with serum deprivation or exogenous TGF-␤1 treatment. Since serum deprivation acted as if exogenous TGF-␤1 had been added, and since effects of serum deprivation were not seen in cells carrying the transdominant negative mutant receptor, it is plausible that endogenous TGF-␤1 might mediate capillary morphogenesis with serum deprivation. In support of this hypothesis, we observed that neutralizing antibody to TGF-␤1 abolished both cell detachment and capillary-like tube formation in serum deprived wild-type endothelial cells as well as DNA fragmentation.
In the process of capillary morphogenesis, our studies show that endothelial cells, in response to TGF-␤1, undergo a programmed cell death and detach from the substratum, a phenomenon called anoikis (37). The term anoikis is derived from the Greek word for homelessness. Anoikis implies that once cells lose contact with underlying matrix, they undergo programmed cell death, thus preventing these detached cells from establishing themselves in another location. Thus, the phenomenon of anoikis is a mechanism for homeostasis that maintain a certain correct cell number in the body by balancing cell production with cell death. Tumor cells escape this regulation by blocking the apoptotic response. Anoikis may also be important in cell positioning. For instance, in the normal maturation of the skin, the cells that are in contact with the basement membrane proliferate and the cells that migrate away from it into the more superficial layers undergo apoptosis (38).
Since integrins are primarily responsible for cell adhesion to ECM, integrin-mediated signaling has been implicated in controlling apoptosis. Frisch and Francis (37) demonstrated that apoptosis was induced by disruption of the interactions between normal epithelial cells and ECM. In endothelial cells, Meredith et al. (39) showed that cells incubated in suspension and denied interactions with ECM rapidly underwent apoptosis. Furthermore, when endothelial cells were plated on an integrin ␤ 1 monoclonal antibody, apoptosis was suppressed. Therefore, regulation of apoptosis may be mediated by disruption of cell-matrix interactions and altered cell-cell interactions. Given this body of evidence, it is not unreasonable to propose that our findings of cell detachment and apoptosis may reflect TGF-␤1-mediated matrix degradation. TGF-␤1 induces endothelial cell secretion of PA, which is an enzyme that cleaves the proenzyme plasminogen to form active plasmin. Plasmin is a protease which degrades ECM proteins. Moreover, plasmin activates latent TGF-␤1 (12), and thus providing a mechanism for autoamplification loop. ECM proteolysis by proteases such as plasmin facilitates endothelial cell migration and angiogenesis in vivo (11). However, even though apoptosis may be mediated by PA-induced matrix degradation by TGF-␤1, this may not be the sole mechanism, in view of the fact that mitogenic growth factors which can prevent apoptosis, such as bFGF, also induces PA.
Angiogenesis is regulated by a number of cytokines in vivo, including bFGF (40), vascular endothelial growth factor (41), and TGF-␤1 (6,7). In our studies, we were able to isolate and delineate the effects of TGF-␤1 in vitro from other angiogenic growth factors. Our findings raise the intriguing possibility that apoptosis is a phenomenon necessary in the process of capillary morphogenesis and that both are dependent on TGF-␤ receptor signaling.