Transforming growth factor-β1 causes pulmonary microvascular endothelial cell apoptosis via ALK5

Abstract

We have previously shown that transforming growth factor (TGF)-β1 protected against main pulmonary artery endothelial cell (PAEC) apoptosis induced by serum deprivation and VEGF receptor blockade through a mechanism associated with ALK5-mediated Bcl-2 upregulation. In the current study, we investigated the effect of TGF-β1 on pulmonary microvascular endothelial cell (PMVEC) apoptosis. We found that, in contrast to the results seen in conduit PAEC, TGF-β1 caused apoptosis of PMVEC, an effect that was also dependent on ALK5 activity. We noted that non-SMAD signaling pathways did not play a role in TGF-β1-induced apoptosis. Both SMAD2 and SMAD1/5 were activated upon exposure to TGF-β1. TGF-β1-induced activation of SMAD2, but not SMAD1/5, was abolished by ALK5 inhibition, an effect that associated with prevention of TGF-β1-induced apoptosis. These results suggest that SMAD2 is important in TGF-β1-induced apoptosis of PMVEC. While caspase-12 activity was not altered, caspase-8 was activated by TGF-β1, an effect that correlated with a reduction of cFLIP protein levels. Additionally, TGF-β1 decreased Bcl-2 protein levels and induced cytochrome c cytosolic redistribution. These results suggest that TGF-β1 caused apoptosis of PMVEC likely through both caspase-8-dependent extrinsic pathway and mitochondria-mediated intrinsic pathway. We noted that inhibition of ALK5 attenuated serum deprivation-induced apoptosis, an effect that correlated with increased expression and activation of CREB and its potential target genes, Bcl-2 and cFLIP. These results suggest that CREB may be important in mediating apoptosis resistance of PMVEC upon ALK5 inhibition perhaps through upregulation of Bcl-2 and cFLIP. Finally, we noted that SMAD1/5 were activated upon ALK5 inhibition in the presence of low levels of TGF-β1, an effect associated with enhanced endothelial proliferation. We speculate that imbalance of ALK1 and ALK5 may contribute to the development of pulmonary artery hypertension.

apoptosis is important in development, tissue homeostasis, and remodeling. Apoptosis also plays a fundamental role in the genesis of various diseases. For example, enhanced apoptosis has been implicated in the pathogenesis of emphysema, whereas diminished apoptosis may be important in the development of some cancers and pulmonary arterial hypertension (16).

Apoptosis occurs through two fundamental signaling pathways, the extrinsic and intrinsic pathways. The extrinsic pathway is activated through a death receptor and mediated through initiator caspases, such as caspase-8, and effector caspases, such as caspase-3. Conversely, the intrinsic pathway is triggered by internal proapoptotic signals, which promote mitochondrial outer membrane permeabilization (MOMP), resulting in the release of several apoptosis-inducing proteins, such as cytochrome c, apoptosis-inducing factor (AIF), and endonuclease G, from the mitochondria into the cytosol (19), leading to apoptosis through caspase-9-dependent and -independent pathways (58, 79). Apoptosis mediated through the mitochondria-dependent intrinsic pathway is regulated by the balance of the activities of proapoptotic and antiapoptotic Bcl-2 family members (1, 19), whereas apoptosis mediated through the extrinsic pathway is regulated by Fas-associated death domain-like inhibitory proteins (FLIPs) (3, 66). Additional signaling pathways through which apoptosis can occur upon prolonged, unresolved endoplasmic reticulum (ER) stress are caspase-12-dependent (14) and -independent pathways (64).

Transforming growth factor (TGF)-β1 triggers numerous cellular responses through receptors and various intracellular signaling pathways (10, 11, 40, 57). Data reporting the effects of TGF-β1 on a particular cellular process are often contradicted in the literature. For example, TGF-β1 has important anti-inflammatory and immunosuppressive effects in some settings and proinflammatory effects in others (31, 35). Contrasting effects of TGF-β1 have also been noted in oncogenesis (2) and wound healing (30). Similarly, the role of TGF-β1 in endothelial cell apoptosis is controversial. Several studies demonstrate that TGF-β1 causes endothelial cell apoptosis (20, 47, 69, 72), whereas others suggest that TGF-β1 promotes endothelial cell survival (33, 42, 60, 70). The disparate responses reported may be attributed to a number of factors, including the type of cells used and the cellular context (13). We have previously shown that TGF-β1 was protective against apoptosis of main pulmonary artery endothelial cells (PAEC) induced by serum deprivation and VEGF receptor blockade, yet not protective against apoptosis induced by UV light exposure or TNFα (36). Our previous work suggested that the protective effect of TGF-β1 against apoptosis is dependent on the mechanism stimulating.

It is well established that endothelial cells derived from distinct organs display a heterogeneous behavior, perhaps as a result of adaptation to the local tissue environment. For example, endothelial cells from the brain and the lung have tight endothelial barrier function, whereas glomerular capillaries are highly permeable. Endothelial heterogeneity of pulmonary macro- and microvascular endothelial cells has also been noted with regard to barrier function, response to injury, and maintenance of homeostasis (61). A recent study demonstrates a heterogeneous proliferative response of pulmonary microvascular endothelial cells (PMVEC) and PAEC (59). However, the heterogeneous response of endothelial cells from different pulmonary vascular beds to apoptotic stimuli has not been investigated. We have previously shown that TGF-β1 protected against apoptosis of PAEC (36). In the current study, we demonstrate that TGF-β1 promoted apoptosis of PMVEC. Our data indicate that endothelial cells derived from pulmonary conduit arteries and from the lung microvasculature differ in their apoptotic response to TGF-β1.

MATERIALS AND METHODS

Cell lines and reagents.

Rat PMVEC were purchased from VEC Technologies (Rensselaer, NY). PMVEC at passages 3–9 were used in this study. MCDB-131 complete media and MCDB-131 basic media were purchased from VEC Technologies and Invitrogen (Carlsbad, CA), respectively. Recombinant human TGF-β1 was obtained from R&D Systems (Minneapolis, MN). SB-431542 and Y-27632 were purchased from Tocris (Ellisville, MO) and CalBiochem (San Diego, CA), respectively. SB-203580 and SKF-86002 were obtained from SmithKline Beecham Pharmaceuticals (Philadelphia, PA). Mouse monoclonal anti-actin α-smooth muscle-Cy3 (anti-αSMA), wortmannin, LY-294002, TRITC-conjugated Bandeiraea simplicifolia (also referred as Griffonia simplicifolia) agglutinin (BS-I), and FITC-conjugated Helix pomatia agglutinin (HPA) were purchased from Sigma (St. Louis, MO). U0126, PD-98059, and antibodies directed against caspase-3, Bax, SMAD5, phospho-SMAD1/5 (Ser463/465), and phospho-SMAD2 (Ser465/467) were obtained from Cell Signaling Technology (Beverly, MA). Antibodies directed against caspase-9, caspase-12, cFLIP, phospho-Bad (Ser136), and Bad were purchased from Stressgen (Ann Arbor, MI). Antibodies directed against Bcl-2, caspase-8, and GAPDH were purchased from BD Transduction Laboratories (San Jose, CA), Stratagene (La Jolla, CA), and Covance Research Products (Richmond, CA), respectively. Antibodies directed against CREB, phospho-CREB (Ser133), and SMAD2/3 were obtained from Upstate (Lake Placid, NY). Mouse monoclonal anti-von Willebrand factor antibody was obtained from Chemicon (Australia). Antibodies directed against VE-cadherin and cytochrome c were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Caspase-8/FLICE fluorometric assay kit was purchased from BioVision Research Products (Mountain View, CA).

Characterization of pulmonary macro- and microvascular endothelial cells.

Rat PMVEC and bovine PAEC grown on coverslips were fixed with 4% paraformaldehyde. After being washed, cells were incubated with diluted fluorescence-conjugated lectins from varying origins for 1 h and then washed. Lectin binding was visualized using a Nikon Eclipse E400 fluorescence microscope at ×1,000 magnification and recorded.

Caspase enzyme activity assay.

Cells were lysed in caspase lysis buffer by freeze-thaw cycles. Caspase-3 and caspase-8 activities were assessed by measuring cleavage of substrates Ac-DEVD-AMC and IETD-AFC, respectively, using a fluorescence microplate reader. Caspase-3 activity was presented as pmol·mg⁻¹·min⁻¹, as previously described (17, 36, 37). Caspase-8 activity was presented as the fold increase in FLICE activity relative to vehicle-treated (unstimulated) cells.

Assessment of apoptosis.

Endothelial cell apoptosis was assessed by DAPI staining of pycnotic nuclei, as previously described (16). Five randomized high power fields per slide were used to examine the percentage of apoptotic cells. The data are presented as the mean of percentage of apoptotic cells over the total counted cells in each slide.

Gel electrophoresis and immunoblot analysis.

Lysates were solubilized in Laemmli buffer, and proteins were resolved by SDS-PAGE. The resolved proteins were then transferred to Immunobilon PVDF membranes and immunoblotted with the indicated antibodies, as previously described (17, 36, 37).

Immunofluorescence microscopy.

Endothelial cells grown on coverslips were treated as described, fixed with 4% paraformaldehyde, and permeabilized with 0.1% Triton X-100, as described (37). Cells were stained with primary antibody followed by Texas red-conjugated species-specific secondary antibody. Images were visualized using a Nikon Eclipse E400 fluorescence microscope at ×1,000 magnification and recorded.

Data analysis.

All experiments were performed at least in triplicate. Data are presented as means ± SE. ANOVA was performed, and significance among the groups was determined using the Tukey-Kramer post hoc test. Differences among means were considered significant when P < 0.05.

RESULTS

Characterization of pulmonary macro- and microvascular endothelial cells.

We have previously shown that TGF-β1 protected against main PAEC apoptosis (36). To elucidate the effect of TGF-β1 on PMVEC apoptosis, we first characterized the two primary cultured endothelial cells at different passages by using markers discriminating pulmonary macro- and microvascular endothelial cells. We found that bovine PAEC, but not rat PMVEC, displayed stronger staining with the lectin H. pomatia (Fig. 1). These results are consistent with the previous observations by King and colleagues (23). In addition, rat PMVEC stained strongly with G. simplicifolia (also known as B. simplicifolia) (Fig. 1), which is also similar to the report from King and coworkers (23). Unlike rat PAEC (23), bovine PAEC were strongly stained by B. simplicifolia (Fig. 1). This result is consistent with the report from Schnitzer and colleagues (56) that B. simplicifolia agglutinins recognized glycoproteins of bovine PAEC. Additionally, we noted that there was no difference in lectin staining among passages of either PMVEC (passages 3–9) or PAEC (passages 3–7) (data not shown).

Fig. 1.

Characterization of pulmonary macro- and microvascular endothelial cells. Rat pulmonary microvascular endothelial cells (PMVEC) and bovine pulmonary artery endothelial cell (PAEC) were grown to 95% confluence, and binding of lectins from Helix pomatia and Bandeiraea simplicifolia was assessed by fluorescence microscopy.

TGF-β1 caused PMVEC apoptosis.

To examine the effect of TGF-β1 on PMVEC apoptosis, PMVEC were exposed to vehicle or varying concentrations of TGF-β1 for 24 h in MCDB-131 complete medium (containing serum) or basic medium (without serum), and caspase-3 activity was assessed. We found that PMVEC exposed to TGF-β1 in complete medium demonstrated a significant increase in caspase-3 activity compared with the cells exposed to vehicle (Fig. 2A). As expected, cells incubated in the basic medium had significantly elevated caspase-3 activity compared with cells cultured in complete medium (Fig. 2A), suggesting that serum deprivation caused PMVEC apoptosis. TGF-β1 dose dependently exacerbated serum deprivation-induced caspase-3 activation (Fig. 2A). The time course studies showed that TGF-β1-induced caspase-3 activation, as measured by procaspase-3 cleavage, occurred at 4–6 h and significantly increased at 24 h of exposure (Fig. 2B). We further demonstrated that TGF-β1 caused PMVEC apoptosis, as assessed by measuring the percentage of the pycnotic nuclei (Fig. 2C). Similar results were seen in bovine PMVEC (data not shown). In addition, we noted no difference in response to TGF-β1 among passages of PMVEC (passages 3–9) or PAEC (passages 3–7) (data not shown). Thus, our data suggest that the heterogeneous response of endothelial cells to TGF-β1 was due to the phenotypic differences between endothelial cells isolated from a distinct location in the lung vasculature, but not due to differences among species or the passages of the cells.

Fig. 2.

Effects of transforming growth factor (TGF)-β1 on apoptosis of PMVEC. A: PMVEC were incubated with vehicle or indicated concentrations of TGF-β1 in MCDB-131 complete medium (CM) or basic medium (BM) for 24 h, and caspase-3 activity was assessed by measuring enzyme activity using Ac-DEVD-AMC as a substrate. Data are presented as means ± SE of 3 independent experiments. *P < 0.05 vs. respective vehicle; ξP < 0.05 vs. CM without TGF-β1. B: PMVEC were incubated with 1 ng/ml TGF-β1 in CM (top) or BM (bottom) for indicated times. Caspase-3 activation was assessed by examining procaspase-3 cleavage. Immunoblots were stripped and reprobed for GAPDH to serve as a control for protein loading. Representative immunoblots of 3 independent experiments are shown. C: PMVEC were incubated with vehicle or TGF-β1 (1 ng/ml) in CM or BM for 24 h, and apoptosis was assessed by determining the percentage of pycnotic DAPI-stained nuclei. Data are presented as means ± SE of 9 independent experiments. *P < 0.05 vs. respective vehicle; ξP < 0.05 vs. vehicle in CM.

TGF-β1 has been shown to promote transdifferentiation of bovine aortic endothelial cells into smooth muscle-like cells after 5 days of exposure in vitro (4). Recently, TGF-β1 has also been reported to induce endothelial-to-mesenchymal transdifferentiation of human coronary endothelial cells, with enhanced expression of proteins associated with myofibroblasts and diminished expression of proteins characteristic of endothelial cells (77). In this study, PMVEC exposed to TGF-β1 for 24 h expressed similar levels of endothelial cell markers, including von Willebrand factor and VE-cadherin, to that in vehicle-treated cells, and with negative staining of α-smooth muscle actin, a myofibroblast marker (data not shown). Our data suggest that TGF-β1 does not cause PMVEC transdifferentiation within the time frame used in this study.

Inhibition of ALK5 abolished TGF-β1-induced apoptosis and promoted endothelial cell survival.

Endothelial cells express two isoforms of TGF-β type I receptor, ALK1 and ALK5. To identify the receptors important in TGF-β1-induced apoptosis, PMVEC were exposed to vehicle or TGF-β1 in the absence or presence of the ALK5 inhibitor, SB-431542, in complete or basic medium for 6 and 24 h, and apoptosis was assessed by measuring caspase-3 enzyme activity, procaspase-3 cleavage, and the percentage of apoptotic nuclei. We found that inhibition of ALK5 abolished TGF-β1-induced caspase-3 activation in both complete medium (Fig. 3A) and basic medium (Fig. 3B). Consistently, ALK5 inhibition abrogated TGF-β1-induced nuclear condensation (Fig. 3C). These results suggest that TGF-β1 triggered apoptosis of PMVEC through activation of ALK5. Similar to the results shown in Fig. 2A, cells incubated with the basic medium displayed elevated levels of cleaved caspase-3 (Fig. 3B) and condensed nuclei (Fig. 3C) compared with the cells incubated with the complete medium (Fig. 3, B and C). Interestingly, caspase-3 cleavage and nuclear condensation induced by serum deprivation was attenuated by the ALK5 inhibitor SB-431542 (Fig. 3, B and C). These data suggest that ALK5 inhibition protects against serum deprivation-induced apoptosis.

Fig. 3.

Effects of ALK5 inhibition on apoptosis of PMVEC. PMVEC were treated with vehicle or TGF-β1 (1 ng/ml) in the absence or presence of the ALK5 inhibitor SB-431542 (10 μM) in CM (A–C) or BM (B and C) for 6 h (C) and 24 h (A–C). Caspase-3 activity (A), procaspase-3 cleavage (B), and the percentage of apoptotic nuclei (C) were assessed. A and C: data are presented as means ± SE of multiple independent experiments. A: n = 3, *P < 0.05 vs. vehicle. B: representative immunoblots of 3 independent experiments are shown (top). The level of cleaved caspase-3 was quantified by densitometry and is presented as means ± SE of the cleaved caspase-3 relative to GAPDH. *P < 0.05 vs. respective vehicle; ξP < 0.05 vs. vehicle in CM; ψP < 0.05 vs. vehicle in BM; #P < 0.05 vs. TGF-β1 in BM. C: n = 9, *P < 0.05 vs. respective vehicle; #P < 0.05 vs. respective TGF-β1; ξP < 0.05 vs. vehicle in CM at 24 h; ψP < 0.05 vs. vehicle in BM at 24 h.

SMAD2 signaling may mediate TGF-β1-induced apoptosis of PMVEC.

TGF-β1 activates SMAD and non-SMAD signaling pathways (26, 50). To determine if non-SMAD signaling pathways play a role in TGF-β1-induced apoptosis of PMVEC, we inhibited RhoA/ROCK, p38, PI3K/Akt, and ERK1/2 pathways using respective chemical inhibitors. The level of caspase-3 activation by TGF-β1 was not affected by inhibition of any of these non-SMAD pathways tested (Fig. 4A). Thus, these results suggest that TGF-β1-induced apoptosis of PMVEC occurs through a SMAD signaling pathway.

Fig. 4.

Signaling pathways involved in TGF-β1-induced apoptosis of PMVEC. A: PMVEC were incubated with vehicle or TGF-β1 (1 ng/ml) in the absence or presence of either of Y-27632 (10 μM), wortmannin (500 nM), LY-294002 (10 μM), SB-203580 (5 μM), SKF-86002 (5 μM), U0126 (10 μM), or PD-98059 (10 μM) in CM for 24 h, and caspase-3 activity was assessed. Data are presented as means ± SE of 3 independent experiments; *P < 0.05 vs. vehicle. B–E: PMVEC were exposed to TGF-β1 (1 ng/ml), SB-431542 (10 μM), or TGF-β1 (1 ng/ml) plus SB-431542 (10 μM) in CM (C and E) or BM (B and D) for the indicated times, and activation of SMAD2 and SMAD1/5 was assessed by measuring the level of phosphorylation of SMAD2^S465/467 or SMAD1/5^S463/465. The immunoblots were stripped and reprobed for SMAD2/3 and SMAD5. B–D: representative immunoblots of 3 independent experiments for each data set are shown. E: representative immunoblots of 3 independent experiments are shown (top). The immunoblot signals were quantified by densitometry. The densitometric values are presented as means ± SE of the phosphorylated SMAD1/5 relative to total SMAD5. *P < 0.05 vs. time 0.

In endothelial cells, TGF-β1 can activate SMAD1/5 and SMAD2/3 via ALK1 and ALK5, respectively. ALK5 expression and kinase activity are necessary for TGF-β1-induced activation of ALK1/SMAD1/5 signaling (13). We have previously shown that inhibition of ALK5 kinase activity prevented TGF-β1-induced activation of both SMAD2 and SMAD1/5 in PAEC (36). To determine if ALK5 inhibition affected TGF-β1-induced activation of SMAD2 and SMAD1/5 in PMVEC, PMVEC were treated with TGF-β1 in the absence or presence of the ALK5 inhibitor SB-431542 for various times in complete medium or basic medium. Activation of SMAD2 and SMAD1/5 was assessed by examining the levels of phosphorylation of SMAD2 at Ser465/467 and phosphorylation of SMAD1/5 at Ser463/465. Total SMAD2/3 and SMAD5 were also examined. As expected, ALK5 inhibition completely abolished TGF-β1-induced SMAD2 activation in both basic medium (Fig. 4B) and complete medium (Fig. 4C), effects associated with prevention of TGF-β1-induced apoptosis (Fig. 3, A–C). Similar to the results seen in PAEC (36), ALK5 inhibition also abrogated TGF-β1-induced SMAD1/5 activation when PMVEC were incubated in the basic medium (Fig. 4D). However, inhibition of ALK5 did not prevent TGF-β1-induced SMAD1/5 activation when PMVEC were incubated in the complete medium (Fig. 4E), a condition in which ALK5 inhibition prevented TGF-β1-induced apoptosis (Fig. 3, A and C). These results suggest that SMAD1/5 activation does not play a role in TGF-β1-induced apoptosis. Together, these results suggest that SMAD2 may be important in mediating TGF-β1-induced apoptosis.

TGF-β1 diminished protein levels of antiapoptotic proteins, cFLIP and Bcl-2.

As a dominant negative homolog of procaspase-8, cFLIP has been implicated in endothelial cell survival (62). Serum deprivation has been shown to decrease cFLIP protein levels in endothelial cells (62). Similarly, PMVEC incubated with basic medium for 24 h demonstrated a decrease in protein levels of cFLIP compared with cells incubated with complete medium (Fig. 5A). cFLIP protein levels were further diminished at 6 h of exposure to TGF-β1 and continuously declined at 24 and 48 h when cells were cultured in basic medium (Fig. 5, A and B). We have previously shown that serum deprivation slightly decreased Bcl-2 protein levels in PAEC (36). This effect was not seen in PMVEC exposed to basic medium for 24 h compared with cells cultured in complete medium (data not shown). However, TGF-β1 significantly reduced Bcl-2 protein levels at 24 and 48 h of exposure when cells were cultured in basic medium (Fig. 5C), effects that were prevented by coincubation with SB-431542 (Fig. 5D), suggesting a dependence on ALK5. We also noted that the levels of the proapoptotic proteins Bax, Bad, and phosphorylated Bad at Ser136, were not altered by TGF-β1 (Fig. 5E). These results suggest that ALK5-mediated decrease in cFLIP and Bcl-2 protein levels may play a role in TGF-β1-induced apoptosis.

Fig. 5.

Effects of TGF-β1 and ALK5 inhibition on protein levels of cFLIP and Bcl-2 family members. A: PMVEC were incubated with vehicle in CM, vehicle in BM, or indicated concentrations of TGF-β1 in BM for 24 h, and cFLIP protein levels were assessed. PMVEC were exposed to TGF-β1 (1 ng/ml) or SB-431542 (10 μM) (B and C), or TGF-β1 (1 ng/ml) or TGF-β1 (1 ng/ml) plus SB-431542 (10 μM) (D), or vehicle or TGF-β1 (1 ng/ml) (E) in BM for the indicated times (B–D) or 24 h (E). The protein levels of cFLIP and Bcl-2 family members were assessed using GAPDH to control for protein loading. Representative immunoblots of 3 independent experiments for each data set are shown. B and C: the immunoblot signals were quantified by densitometry. The densitometric values are presented as means ± SE of cFLIP or Bcl-2 relative to GAPDH. *P < 0.05 vs. time 0. ξP < 0.05 vs. time 0.

TGF-β1 activated caspase-8 and promoted MOMP.

Apoptosis occurs through a caspase-8-dependent extrinsic pathway, a mitochondria-mediated intrinsic pathway, and a caspase-12-mediated ER stress pathway. To elucidate the pathways involved in TGF-β1-induced apoptosis, PMVEC were treated with vehicle or TGF-β1 for various times, and caspase-8, -9, and -12 activities were assessed by measuring procaspase cleavage and enzyme activities. We noted that the protein levels of procaspase-8 were slightly decreased at 6 h and significantly diminished at 24 h of exposure to TGF-β1, an effect that was prevented by ALK5 inhibition (Fig. 6A). To demonstrate that the decreased procaspase-8 is due to its cleavage, thus activation of caspase-8, we assessed caspase-8 (FLICE) activity using IETD-AFC as a substrate. We found that cells exposed to TGF-β1 for 24 h in the complete medium displayed a 2.23 ± 0.85-fold increase in FLICE activity (P < 0.05, n = 3) compared with vehicle-treated cells, suggesting caspase-8 activation. This result is consistent with a decrease in cFLIP protein levels.

Fig. 6.

Effects of TGF-β1 on the pathways of apoptosis. PMVEC were incubated with TGF-β1 (1 ng/ml) or TGF-β1 (1 ng/ml) plus SB-431542 (10 μM) in CM for the indicated times (A) or were incubated with vehicle or TGF-β1 (1 ng/ml) in CM for 24 h (B and C). Activation of caspases-8 (A), -9 (C), and -12 (C) were measured by assessing the level of cleavage of procaspases-8, -9, or -12 using GAPDH to control for protein loading. Cytochrome c subcellular localization was assessed by immunofluorescence microscopy (B). Representative immunoblots (A and C) and images (B) of 3 independent experiments for each data set are shown. B: arrows indicate mitochondria localization, and arrowheads indicate cytosolic localization of cytochrome c.

Bcl-2 is one of the essential anti-apoptotic Bcl-2 family members participating in the regulation of MOMP (16). Since Bcl-2 protein levels were diminished by TGF-β1, we explored changes in MOMP by examining cytochrome c subcellular redistribution. Immunofluorescence analysis demonstrated cytochrome c located within the perinuclear region of PMVEC treated with vehicle (Fig. 6B), consistent with mitochondrial localization. Cytochrome c redistributed to the cytosol in cells treated with TGF-β1 (Fig. 6B). This result suggests an enhanced MOMP and activation of intrinsic pathway of apoptosis. We noted that caspase-9 was not activated by TGF-β1 (Fig. 6C), suggesting that TGF-β1 enhanced MOMP, leading to apoptosis through caspase-9-independent pathway(s). We also noted that caspase-12 was not altered by TGF-β1 upon 24-h exposure (Fig. 6C), suggesting that ER stress pathway may not be involved. Together, these results suggest that TGF-β1 caused apoptosis of PMVEC through both caspase-8-mediated extrinsic pathway and mitochondria-mediated intrinsic pathway.

Inhibition of ALK5 enhanced protein levels of cFLIP and Bcl-2 and activated CREB.

To understand the mechanism by which ALK5 inhibition promoted survival of PMVEC, we examined the effect of ALK5 inhibition on expression of cFLIP and Bcl-2. PMVEC exposed to the ALK5 inhibitor SB-431542 for various times demonstrated that ALK5 inhibition significantly increased cFLIP protein levels at 6 and 24 h of exposure when cells were cultured in the basic medium (Fig. 5B). Similarly, ALK5 inhibition also enhanced Bcl-2 protein levels at 6 and 24 h of exposure (Fig. 5C).

Activation of cAMP response element (CRE)-binding protein (CREB) has been implicated in promoting cell survival by upregulation of Bcl-2 (15, 71) and cFLIP (7, 49, 78). We then examined expression and activation of CREB upon TGF-β1 stimulation and ALK5 inhibition. We found that TGF-β1 did not alter either expression or activation of CREB (data not shown). However, ALK5 inhibition alone not only significantly increased CREB phosphorylation but also enhanced the total CREB protein levels at 6 and 24 h of exposure (Fig. 7). The ratio of the phosphorylated CREB to the total CREB did not change (Fig. 7), suggesting that the increase in CREB phosphorylation is due to the increase in protein expression.

Fig. 7.

Effects of ALK5 inhibition on expression and activation of CREB. PMVEC were exposed to SB-431542 (10 μM) in BM for the indicated times, and CREB activation was assessed by measuring CREB phosphorylation at Ser133. The immunoblots were stripped and reprobed for CREB and GAPDH. Representative immunoblots of 3 independent experiments are shown. The immunoblot signals were quantified by densitometry. The densitometric values are presented as means ± SE of CREB^S133∼P relative to GAPDH, CREB relative to GAPDH, and CREB^S133∼P relative to CREB. *P < 0.05 vs. respective time 0.

ALK5 inhibition activated SMAD1/5 in the presence of low levels of TGF-β1.

ALK5 kinase activity is required for TGF-β1-induced activation of ALK1/SMAD1/5 signaling (13). Consistently, we noted that inhibition of ALK5 kinase activity by SB-431542 prevented TGF-β1-induced ALK1/SMAD1/5 activation when cells were cultured in basic medium (Fig. 4D). However, ALK1/SMAD1/5 signaling was still activated in cells cultured in complete medium, with addition of ALK5 kinase inhibitor and TGF-β1 (Fig. 4E). These results suggest that something within the serum of complete medium, in combination with ALK5 inhibitor and/or TGF-β1, may activate ALK1/SMAD1/5 signaling. An unknown ALK1 ligand has been observed in serum (38). We then tested the effect of serum on SMAD2 and SMAD1/5 activation and found that phosphorylation of neither SMAD2 nor SMAD1/5 was altered in cells cultured in complete medium for 0–24 h (data not shown), suggesting that serum alone is not sufficient to activate ALK1/SMAD1/5 signaling. In addition, neither SMAD2 nor SMAD1/5 was activated when cells were cultured in basic medium (data not shown). We further noted that SMAD1/5 was activated by ALK5 inhibitor, without addition of TGF-β1, when cells were cultured in complete medium (Fig. 4E), but not in basic medium (Fig. 4D). These data suggest that presence of serum was necessary for ALK5 inhibitor to activate ALK1/SMAD1/5 signaling. These results also suggest that addition of TGF-β1 is not necessary for ALK5 inhibitor to activate ALK1/SMAD1/5 signaling as long as serum is present.

Since the serum within complete medium contains low levels of TGF-β1, we hypothesized that ALK1/SMAD1/5 signaling was activated by low levels of TGF-β1 upon ALK5 inhibition. To test this, PMVEC were treated with 10 μM SB-431542 in basic medium, with addition of 1 ng/ml, 5 pg/ml, and 0.5 pg/ml TGF-β1 for 30 min and 1 h, and SMAD1/5 activation was determined. We found that SMAD1/5 was activated upon ALK5 inhibition, with addition of pg/ml amounts, but not 1 ng/ml, of TGF-β1 into the basic medium (data not shown). These results suggest that low levels of TGF-β1, contained within serum of complete medium, contribute to activation of ALK1/SMAD1/5 signaling upon ALK5 inhibition. Since SMAD1/5 was activated upon ALK5 inhibition in the presence of serum and 1 ng/ml TGF-β1 (Fig. 4E), or in the absence of serum and presence of pg/ml amount of TGF-β1 (data not shown), but not activated upon ALK5 inhibition in the absence of serum and presence of 1 ng/ml TGF-β1 (Fig. 4D), we speculate that high levels (ng/ml) of TGF-β1 in the absence of serum may inhibit ALK1/SMAD1/5 activation upon ALK5 inhibition.

DISCUSSION

We have previously shown that TGF-β1 protected against apoptosis of main PAEC, induced by serum deprivation or VEGF receptor blockade, through a mechanism associated with ALK5-mediated Bcl-2 induction (36). In the current study, we investigated the effect of TGF-β1 on apoptosis in PMVEC and its underlying signaling mechanisms. In contrast to the protective effect seen in PAEC, TGF-β1 caused apoptosis in PMVEC. The TGF-β1 effect was associated with decreased protein levels of cFLIP and Bcl-2, activation of caspase-8, and enhanced MOMP in PMVEC, effects that were abrogated by ALK5 inhibition. TGF-β1 activated both SMAD2 and SMAD1/5. However, TGF-β1-induced activation of SMAD2, but not SMAD1/5, was abolished by ALK5 inhibition, results that were associated with the prevention of apoptosis. Non-SMAD signaling pathways did not play a role in TGF-β1-induced apoptosis. These results suggest that SMAD2 is a major contributor to TGF-β1-induced apoptosis of PMVEC. In addition, we have demonstrated that ALK5 inhibition attenuated serum deprivation-induced apoptosis, an effect associated with enhanced expression and activation of CREB and elevation of its potential target genes, cFLIP and Bcl-2. These data suggest that ALK5 inhibition-induced CREB expression, thus activation, may play an important role in protection against apoptosis of PMVEC through upregulation of cFLIP and Bcl-2.

The apoptotic effect of TGF-β1 has been shown to vary in endothelial cells. Several studies have demonstrated that TGF-β1 causes apoptosis of endothelial cells derived from human umbilical veins, bovine aorta, and retina (20, 47, 69, 72). In contrast, TGF-β1 has been reported to promote mouse capillary endothelial cell survival (70). In addition, TGF-β1 type III receptor CD105 has been shown to attenuate apoptosis of human dermal microvascular endothelial cells induced by hypoxia (33). The variability of these responses may be attributed to endothelial heterogeneity within the vasculature and organ bed. We have previously shown that TGF-β1 protected against apoptosis of PAEC induced by serum deprivation and VEGF receptor blockade (36). In contrast, in this study we demonstrated that TGF-β1 caused apoptosis of PMVEC. Our data indicate a heterogeneous apoptotic response to TGF-β1 of endothelial cells from pulmonary conduit arteries and the microvasculature.

We have shown that ALK5 inhibition abolished TGF-β1-induced apoptosis of PMVEC, suggesting that ALK5 activation mediates the effect of TGF-β1 on apoptosis. This notion is supported by a previous report showing that overexpression of constitutively active ALK5 induced apoptosis of HUVEC (44). TGF-β1 has been shown to activate SMAD1/5 and SMAD2/3 through ALK1 and ALK5, respectively, in nonpulmonary endothelial cells (28). Consistently, we observed activation of both SMAD1/5 and SMAD2 upon exposure of PMVEC to TGF-β1. TGF-β1-induced SMAD2 activation and apoptosis were completely abrogated by ALK5 inhibition. ALK5 kinase activity is required for TGF-β1-induced activation of ALK1/SMAD1/5 signaling (13). We have also shown that inhibition of ALK5 kinase activity with SB-431542 prevented TGF-β1-induced SMAD1/5 activation in conduit PAEC (36). A similar effect was also seen in PMVEC cultured in the basic medium. However, the ALK5 inhibitor did not blunt TGF-β1-induced SMAD1/5 activation when PMVEC were cultured in the complete medium, a condition in which TGF-β1-induced apoptosis was prevented. We also noted that non-SMAD signaling pathways, including RhoA/ROCK, PI3K/Akt, p38, and ERK1/2, did not play a role in TGF-β1-induced apoptosis. Together, our results suggest that activation of SMAD2, not SMAD1/5, is most likely responsible for TGF-β1-induced apoptosis of PMVEC, as depicted in Fig. 8. Molecular approaches to target SMAD2 are needed to directly address the essential role of SMAD2 in this process.

Fig. 8.

Proposed mechanism by which TGF-β1 promoted apoptosis and ALK5 inhibition attenuated apoptosis of PMVEC. Solid arrows indicate defined pathways; dashed arrows indicate speculative pathways.

Apoptosis occurs through three major pathways, a caspase-8-dependent extrinsic pathway, a mitochondria-mediated intrinsic pathway, and an ER stress pathway. Our data indicate that caspase-8, but not caspase-12, was activated by TGF-β1, suggesting activation of the extrinsic pathway, but not the ER stress pathway. As an antagonist of procaspase-8, cFLIP has been implicated in endothelial cell survival by inhibiting caspase-8 activation (62). Downregulation of cFLIP has been associated with serum deprivation-induced apoptosis (62). Consistently, we noted that cFLIP protein levels were decreased in cells incubated with basic medium (serum deprivation) for 24 h. Addition of TGF-β1 further reduced the protein levels of cFLIP in a time-dependent manner, an effect associated with caspase-8 activation and apoptosis. Our results suggest that TGF-β1 may suppress cFLIP gene expression or enhance cFLIP degradation, thereby activating caspase-8, leading to caspase-3 activation and apoptosis (Fig. 8). TGF-β1 has also been reported to induce cFLIP expression and inhibits Fas-mediated apoptosis of microglia (55). The opposite effects of TGF-β1 on cFLIP expression between studies from Schlapbach and colleagues (55) and ours may be due to the different cell types used. Downregulation of Bcl-2 has been associated with TGF-β1-induced apoptosis of HUVEC (69). Similarly, we observed a reduction in Bcl-2 protein levels upon TGF-β1 stimulation, an effect associated with enhanced MOMP, as indicated by cytochrome c redistribution. This result suggests that TGF-β1 also activated the mitochondria-mediated intrinsic pathway. However, we did not observe any visible change in caspase-9 activity in response to TGF-β1. Our data suggest that enhanced MOMP may contribute to TGF-β1-induced apoptosis through a caspase-9-independent pathway (Fig. 8). Release of AIF (63) and endonuclease G (34) from mitochondria to the cytosol has been shown to cause apoptosis independently of caspase activities. Whether AIF and endonuclease G contribute to TGF-β1-induced apoptosis remains to be elucidated. Together, our data suggest that TGF-β1 caused apoptosis of PMVEC through both a caspase-8-dependent extrinsic pathway and a mitochondria-mediated intrinsic pathway, but not an ER stress pathway (Fig. 8).

CREB was originally described as a transcription factor that binds to the CRE motif in the somatostatin gene promoter (43). CREB phosphorylation at Ser133 is indispensable for transcriptional activation (8, 12, 25). Activation of CREB has been shown to mediate phorbol ester-induced Bcl-2 transcription and subsequent inhibition of apoptosis in B cells (71). In addition, CREB dephosphorylation and Bcl-2 downregulation is correlated with endothelial cell apoptosis (15). These findings suggest that CREB activation inhibits apoptosis through induction of Bcl-2. CREB also promotes cell survival perhaps through induction of cFLIP (7, 49, 78). In this study, CREB expression and activity was significantly enhanced upon ALK5 inhibition, an effect associated with upregulation of Bcl-2 and cFLIP and with attenuation of apoptosis. Our data suggest that CREB induction and subsequent activation is important in mediating apoptosis resistance of PMVEC upon ALK5 inhibition through upregulation of cFLIP and Bcl-2, as depicted in Fig. 8.

As a nuclear located transcription factor, CREB can be activated by various signaling pathways, such as cAMP/PKA, calcium/calmodulin kinase II, PI3K/Akt, and Ras/MEK/MAP kinases (21). TGF-β1 has been shown to activate CREB independently of the above pathways (24, 48). We noted that TGF-β1 did not alter CREB expression and activation in PMVEC. However, ALK5 inhibition significantly enhanced both phosphorylated and total CREB protein levels. The ratio of the phosphorylated CREB relative to the total CREB was not changed, suggesting that the increased activation of CREB was due to elevation of gene expression. SMAD proteins can repress or activate TGF-β1 target gene expression dependent on the availability of its cosuppressor Ski and coactivator CREB-binding protein (CBP) (5). We speculate that SMAD2 may suppress CREB expression in PMVEC. Inhibition of ALK5/SMAD2 by SB-431542 allowed expression and activation of CREB, which promoted cell survival through upregulation of cFLIP and Bcl-2 (Fig. 8). This seems to be inconsistent with the findings that CREB expression and activation were not significantly altered by TGF-β1. One explanation is that TGF-β1 may suppress CREB expression/activation through ALK5/SMAD2 and also promote CREB expression/activation through ALK1/SMAD1/5. Failure to increase CREB expression and activation may lead to a decrease in cFLIP and Bcl-2, and apoptosis would then occur. We also cannot rule out a direct effect of SMAD2 on reduction in protein levels of cFLIP and Bcl-2 (Fig. 8).

We noted that ALK1/SMAD1/5 signaling was not activated by 1 ng/ml TGF-β1 in the presence of the ALK5 kinase inhibitor SB-431542 unless cells were cultured in the complete medium. These data suggest that something within the serum of complete medium, in conjunction with TGF-β1 and/or ALK5 inhibitor, is required for activating ALK1/SMAD1/5. However, ALK5 inhibitor alone, without addition of 1 ng/ml TGF-β1, also activated ALK1/SMAD1/5 when cells were cultured in complete medium. These data suggest that the ALK5 inhibitor, but not 1 ng/ml TGF-β1, in combination with serum within complete medium, activated ALK1/SMAD1/5. The presence of an unknown ALK1 ligand within serum has been reported (38). However, we noted that ALK1/SMAD1/5 signaling was not activated by complete medium alone, unless ALK5 kinase inhibitor was present. Our data suggest that ALK5 kinase inhibition is necessary for complete medium to activate ALK1/SMAD1/5 signaling. Goumans and colleagues (13) have shown that ALK5 kinase activity is required for activation of ALK1/SMAD1/5 signaling after TGF-β1 challenge. Consistently, we noted that inhibition of ALK5 kinase activity by SB-431542 prevented TGF-β1-induced ALK1/SMAD1/5 activation when cells were cultured in basic medium. Goumans and coworkers have also demonstrated that ALK1, ALK5, and TβRII form a heteromeric complex in the absence of TGF-β1, and ALK5 kinase activity is required for recruitment of ALK1 into the ALK5-TβRII complex (13). Recent studies have suggested that ALK1 is activated independently of ALK5 and TβRII (45). In addition, ALK1 activation is regulated by a number of other factors, including VE-cadherin (52), caveolin-1 (54), and endoglin (29, 46). We speculate that the ALK5 kinase inhibitor SB-431542 may inhibit the baseline level of ALK5 kinase activity, thus preventing the basal ALK1-ALK5-TβRII complex formation, allowing ALK1 to be free from the ALK5-TβRII complex, thereby activated through the non-ALK5 pathway, perhaps by low levels of TGF-β1, other ligand(s), or regulators. In support of a role of low levels of TGF-β1 present within serum of complete medium, in triggering ALK1/SMAD1/5 activation, we further observed that ALK1/SMAD1/5 signaling was activated by the ALK5 inhibitor when cells were cultured in basic medium, with addition of pg/ml amount of TGF-β1. This effect was not seen upon ALK5 inhibition when none or 1 ng/ml TGF-β1 was added into basic medium. Because 1 ng/ml TGF-β1 did not prevent ALK1/SMAD1/5 activation upon ALK5 inhibition unless cells were cultured in basic medium, we speculate that high levels (1 ng/ml) of TGF-β1, in the absence of serum, may inhibit ALK1/SMAD1/5 signaling activation, and some unknown factor(s) within serum may act against this inhibition. Activation of ALK1/SMAD1/5 signaling promotes endothelial cell proliferation (29, 46). To further support the concept that low levels of TGF-β1 contained within serum of complete medium contribute to activation of ALK1/SMAD1/5 signaling upon ALK5 inhibition, we also observed that the ALK5 inhibitor promoted proliferation of PMVEC cultured in either complete medium or in basic medium with addition of pg/ml amount of TGF-β1, but not in basic medium with addition of none or 1 ng/ml TGF-β1 (data not shown). Together, our data suggest that ALK5 inhibition may sensitize ALK1, thus activating SMAD1/5 and promoting endothelial cell proliferation, in response to low levels of TGF-β1 present in the serum of complete medium (Fig. 8). However, we cannot exclude the possibility that other factor(s) within serum of complete medium, in addition to low levels of TGF-β1, are involved in ALK1/SMAD1/5 activation upon ALK5 inhibition.

Pulmonary artery hypertension (PAH) has been associated with a decrease in pulmonary endothelial cell apoptosis. For example, compared with endothelial cells from control human lungs, lung endothelial cells of patients with idiopathic PAH displayed decreased apoptosis, which associated with persistent activation of STAT3 and increased expression of its downstream prosurvival target Mcl-1 (39). In addition, Bcl-2 was highly expressed in pulmonary endothelial cells in PAH (32, 73). There is growing evidence that incidence of PAH is linked to abnormalities of TGF-β signaling. Heterozygous germline mutations of the bone morphogenetic protein receptor II, a receptor member of the TGF-β superfamily, have been identified in patients with familial and sporadic PAH (9, 27, 67). Mutations in genes encoding ALK1 and endoglin have also been found in PAH patients with hereditary hemorrhagic telangiectasia (18, 68). In addition, the ALK5/SMAD2/3 signaling pathway was impaired (41, 76), whereas the ALK1 level was elevated (6, 22, 41) in the lungs of animal models of PAH. Impairment of TGF-β1 signaling has also been implicated in plexiform lesion formation. For example, somatic mutations of TGF-β type II receptor (TβRII) in endothelial cells within plexiform lesions of patients with PAH have been demonstrated (74). In addition, endothelial cells within plexiform lesions of PAH lack expression and activation of the key components (ALK1, ALK5, TβRII, SMAD1/5/8, SMAD2/3, and SMAD4) of TGF-β1 signaling (51, 74). These data suggest that impaired TGF-β1 signaling may cause PAH. Our data demonstrated that ALK5 inhibition blunted serum deprivation-induced apoptosis of PMVEC, an effect associated with activation of CREB. We also demonstrated that inhibition of ALK5 activated SMAD1/5 and promoted proliferation of PMVEC in the presence of low concentrations of TGF-β1. Thus, we speculate that inhibition of ALK5 may worsen PAH through SMAD1/5-mediated enhancement of proliferation and CREB-mediated suppression of apoptosis of PMVEC. In contrast, enhanced expression and activation of SMAD1/5/8 and SMAD2 has been found in endothelial cells in pulmonary arteries of idiopathic PAH lungs compared with pulmonary arteries of normal and emphysema lungs (51, 74). Moreover, active SMAD2 expression was higher in the smaller-sized pulmonary vessels than larger diameter pulmonary arteries in IPAH lungs (51, 74). In addition, overexpression of dominant negative mutant of TβRII attenuated chronic hypoxia-induced pulmonary hypertension and vascular remodeling (6). Recently, Zaiman and colleagues (75) have reported that inhibition of ALK5 with SD208 significantly attenuated the development of pulmonary hypertension induced by monocrotaline (MCT), an effect associated with prevention of SMAD2 activation and inhibition of early stage of lung microvascular endothelial apoptosis. Their results are consistent with our in vitro findings that ALK5 inhibition abrogated TGF-β1-induced SMAD2 activation and apoptosis of PMVEC. Given the notion that initial endothelial apoptosis is required for the development of PAH (53, 65), these results suggest that ALK5 activation may initiate PAH by inducing lung microvascular endothelial cell apoptosis. Thus, inhibition of ALK5 may be a beneficial treatment for early stage of PAH. Zaiman et al. also found that ALK5 inhibition with SD208 was not as effective in reversing established MCT-induced PAH, suggesting that ALK5 activation may not contribute to the progression of PAH (75). Based on our data that inhibition of ALK5 suppressed apoptosis, and, in combination with low levels of TGF-β1, promoted proliferation of PMVEC through ALK1 activation, we speculate that ALK5 inhibition may contribute to progression of PAH through promoting proliferation and apoptosis resistance of PMVEC. We also speculate that inhibition of ALK1 may be an efficacious treatment for established PAH. We noted that TGF-β1 also protects against apoptosis of PAEC (36). Thus, the role of TGF-β1 in the development of pulmonary hypertension may be more complicated than we thought.

GRANTS

This work was supported with resources and the use of facilities at the Providence Veterans Affairs Medical Center and was supported by American Lung Association Research Grant RG-1140-N, a Parker B. Francis Fellowship, an ATS/Pulmonary Hypertension Association Research Grant (Q. Lu), and National Heart, Lung, and Blood Institute Grants HL-64936 (S. Rounds) and HL-67795 (E. O. Harrington).