Delayed stress fiber formation mediates pulmonary myofibroblast differentiation in response to TGF-β

Abstract

Myofibroblast differentiation induced by transforming growth factor-β (TGF-β) and characterized by de novo expression of smooth muscle (SM)-specific proteins is a key process in wound healing and in the pathogenesis of fibrosis. We have previously shown that TGF-β-induced expression and activation of serum response factor (SRF) is required for this process. In this study, we examined the signaling mechanism for SRF activation by TGF-β as it relates to pulmonary myofibroblast differentiation. TGF-β stimulated a profound, but delayed (18–24 h), activation of Rho kinase and formation of actin stress fibers, which paralleled SM α-actin expression. The translational inhibitor cycloheximide blocked these processes without affecting Smad-dependent gene transcription. Inhibition of Rho kinase by Y-27632 or depolymerization of actin by latrunculin B resulted in inhibition TGF-β-induced SRF activation and SM α-actin expression, having no effect on Smad signaling. Conversely, stabilization of actin stress fibers by jasplakinolide was sufficient to drive these processes in the absence of TGF-β. TGF-β promoted a delayed nuclear accumulation of the SRF coactivator megakaryoblastic leukemia-1 (MKL1)/myocardin-related transcription factor-A, which was inhibited by latrunculin B. Furthermore, TGF-β also induced MKL1 expression, which was inhibited by latrunculin B, by SRF inhibitor CCG-1423, or by SRF knockdown. Together, these data suggest a triphasic model for myofibroblast differentiation in response to TGF-β that involves 1) initial Smad-dependent expression of intermediate signaling molecules driving Rho activation and stress fiber formation, 2) nuclear accumulation of MKL1 and activation of SRF as a result of actin polymerization, and 3) SRF-dependent expression of MKL1, driving further myofibroblast differentiation.

idiopathic pulmonary fibrosis (IPF) is a progressive, fatal disease characterized by parenchymal fibrosis and structural distortion of the lungs, and it poses a vexing clinical challenge given the lack of proven efficacious therapy. IPF is characterized by an abnormal wound healing response (29, 55, 65–67, 80, 90) mediated in part by interstitial lung fibroblasts (34, 57, 81). In this context, fibroblasts are activated by both soluble and mechanical environmental signals (18, 34, 59, 81), of which transforming growth factor-β1 (TGF-β1) is the most established (2, 6, 41). Upon stimulation with TGF-β1, fibroblasts respond by altering their morphology and ultrastructure, with formation of prominent actin stress fibers (16, 88) and modified focal adhesion complexes, which provide mechanical coupling to the surrounding matrix (32). TGF-β1-induced morphological changes in myofibroblasts are accompanied by robust changes in gene expression profile, with de novo expression of contractile proteins normally found within smooth muscle cells, focal adhesion proteins (32), and components of the extracellular matrix (44, 57, 81). The totality of these changes defines the myofibroblast, displaying an intermediary phenotype between fibroblasts and smooth muscle cells (15, 34, 59). Smooth muscle-α-actin (SM α-actin) is the most established marker for myofibroblast differentiation (15, 20, 34, 43, 58, 81).

TGF-β1 signals through its transmembrane receptor serine/threonine kinase to phosphorylate the receptor-associated Smad proteins (Smad2/3), followed by their heteromerization with coactivator Smad4, nuclear translocation of the Smad2/3/4 complex, and activation of Smad-binding elements on the promoters of TGF-β1 target genes (14, 24, 44), including SM α-actin (37, 60, 77). We and others have observed that TGF-β1-induced expression of SM α-actin in fibroblasts is delayed, occurring 24–48 h after TGF-β1 stimulation (63, 79), which is different from the rapid (as early as at 3–6 h) expression of SM α-actin in response to G protein-coupled stimuli (5). The mechanism of G protein-mediated SM gene expression is well understood and involves rapid activation of small GTPase Rho by Gα12/13 proteins, Rho/Rho kinase-dependent actin polymerization, resultant depletion of monomeric G-actin, and a release (from sequestration by G-actin) of megakaryoblastic leukemia (MKL)/myocardin-related transcription factors (MRTF) that translocate to the nucleus and activate serum response factor (SRF), driving gene transcription through CArG elements on the promoters of SM genes (47, 51, 64, 74, 89).

While it was reported that TGF-β1 can induce Rho activation and actin polymerization in various cell types, these processes were largely associated with alterations in cell morphology during the transition toward a myofibroblast phenotype (4, 19, 45, 69, 85). We have recently reported that TGF-β1 stimulates a delayed expression and activation of SRF, which is required for TGF-β1-induced SM α-actin expression in human pulmonary fibroblasts (HLF) (63). Therefore, in this study, we sought to examine if and how Rho/F-actin signaling is recruited by TGF-β1 for SRF activation and SM α-actin expression using primary cultured HLF.

MATERIALS AND METHODS

Isolation and primary culture of HLF.

HLF were isolated as described previously (63). Briefly, tissue samples from explanted lungs from patients undergoing lung transplantation for pulmonary fibrosis were obtained and placed in DMEM with 100 U/ml streptomycin, 250 ng/ml amphotericin B, and 100 U/ml penicillin. Alveolated lung tissue was minced, washed in PBS, and plated on 10-cm plates in growth media containing DMEM supplemented with 10% FBS, 2 mM l-glutamine, 100 U/ml streptomycin, 250 ng/ml amphotericin B, and 100 U/ml penicillin. Expanded populations of fibroblasts were subsequently subcultured after 4–5 days, resulting in the development of a homogenous fibroblast population. All primary cultures were used from passage 5 to 10 and maintained on tissue culture plastic until the time of experiments.

DNA transfection/transduction.

Cells were maintained in DMEM supplemented with 10% FBS, 2 mM l-glutamine, 100 U/ml streptomycin, 250 ng/ml amphotericin B, and 100 U/ml penicillin for up to 10 passages. In all experiments, cells were grown on collagen-coated plates for 1 day and were serum-deprived overnight in DMEM containing 0.1% BSA and 2 mM l-glutamine before stimulation with desired agonists. Transient DNA transfections were performed using GenDrill reagent (BamaGen BioScience, Gaithersburg, MD) following the standard manufacturer's protocol. Adenovirus-mediated gene transduction was performed by incubating cells with desired adenoviruses (100 plaque-forming units/cell) in the presence of GeneJammer reagent (Stratagene, La Jolla, CA) to facilitate the efficiency of transduction (25).

Reagents.

The firefly luciferase reporter driven by two copies of CArG elements (SRF-Luc) was used previously (12, 35, 63, 78). The firefly luciferase reporter driven by four copies of Smad-binding elements (SBE-Luc) was provided by Dr. Bert Vogelstein and was used previously (63). The thymidine kinase promoter (TK)-driven Renilla luciferase was from Promega (Madison, WI). Recombination-deficient adenovirus expressing short-hairpin RNA against SRF (Ad-shSRF) or the control adenovirus expressing shRNA against green fluorescent protein (Ad-shGFP) were kindly provided by Dr. Joseph Miano (University of Rochester) (8). TGF-β1, Y-27632, latrunculin B, jasplakinolide, and cycloheximide were from EMD Biosciences (Gibbstown, NJ). CCG-1423 was from Cayman Chemical (Ann Arbor, MI). SB-431542 was from Sigma (St. Louis, MO). Antibodies against β-actin and SM α-actin were from Sigma. SRF, Smad4, and 14–3-3β antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Phospho-Smad2 and Smad2 antibodies were from Cell Signaling (Danvers, MA). MKL1 antibodies were from Bethyl Laboratories (Montgomery, TX). Lamin A/C antibodies were from BD Transduction Laboratories (Lexington, KY).

Cell lysis and Western blotting.

After stimulation of quiescent cells with desired agonists, cells were lysed in RIPA buffer containing 25 mM HEPES (pH 7.5), 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 2 mM EDTA, 2 mM EGTA, 10% glycerol, 1 mM NaF, 200 μM sodium orthovanadate, and protease inhibitor cocktail (Sigma). Cells were scraped, sonicated for 5 s, and boiled in Laemmli buffer for 5 min. The samples were subjected to polyacrylamide gel electrophoresis, analyzed by Western blotting with desired primary antibodies and corresponding horseradish peroxidase-conjugated secondary antibodies, and developed by an enhanced chemilumeniscence reaction (Pierce). The digital chemilumeniscent images were taken by a Luminescent Image Analyzer LAS-4000 (Fujifilm).

In vitro isolation of stress fibers.

Stress fibers were isolated using a modified method described previously (40). All of the procedures were performed on ice using the buffers containing protease inhibitor cocktail (Sigma). After stimulation with desired agonists, cells were washed with PBS and then extracted with a buffer containing 2.5 mM triethanolamine (pH 8.2) for 30 min with six buffer changes, followed by extraction with 0.05% Nonidet P-40 (pH 7.2) for 5 min and subsequent extraction with 0.5% Triton X-100 (pH 7.2) for an additional 5 min. Cells were then immediately washed with cold PBS, scraped, and suspended in PBS, followed by centrifugation at 100,000 g for 1 h. Supernatant was removed, and the pellet was sonicated in 0.5% Triton X-100, 50 mM NaCl, 20 mM HEPES (pH 7.0), and 1 mM EDTA. Laemmli buffer was added, and samples were boiled for 5 min before further Western blot analysis as described above.

In vitro isolation of nuclear and cytoplasmic fractions.

Preparation of nuclear and cytoplasmic fractions was performed using the NE-PER nuclear and cytoplasmic reagents (Thermo Scientific) following the manufacturer's protocol. Briefly, after stimulation with desired agonists, cells were trypsinized and washed with PBS to remove trypsin by centrifugation at 300 g for 3 min. Cell pellets were suspended in cytoplasm extraction reagent for 10 min, pelleted again at 16,000 g for 5 min, and the supernatant (cytoplasmic fraction) was collected. The pellets were suspended in the nuclear extraction reagent for 40 min, centrifuged at 16,000 g for 10 min, and the supernatant (nuclear fraction) was collected. Laemmli buffer was added, and samples were boiled for 5 min before further Western blot analysis as described above. Immunoblotting of nuclear lamin A/C and cytoplasmic 14–3-3β was performed to confirm the purity of the nuclear and cytoplasmic fractions, respectively.

Luciferase reporter assay.

Cells grown on 24-well plates were cotransfected with 500 ng of desired Firefly luciferase reporter plasmid and 20 ng of constitutively active thymidine kinase promoter (TK)-Renilla luciferase reporter plasmid. Cells were placed in growth media overnight and then serum-starved for 24 h, followed by stimulation with the desired agonists for the desired time points as indicated in the legends for Figs. 1–9. Cells were then washed with PBS and lysed in protein extraction reagent (Pierce). The lysates were assayed for Firefly and Renilla luciferase activity using the dual luciferase assay kit (Promega). To account for differences in transfection efficiency, Firefly luciferase activity of each sample was normalized to Renilla luciferase activity.

Fig. 1.

Transforming growth factor (TGF)-β1-induced stress fiber formation and smooth muscle (SM)-α-actin expression in human lung fibroblasts (HLF). HLF were stimulated with 1 ng/ml TGF-β1 for desired time points as indicated. A: stress fiber formation and SM α-actin expression were examined by phalloidin staining and by immunofluorescence with SM α-actin antibodies, respectively. B: in vitro isolated stress fibers or total protein lysates were analyzed by Western blotting with antibodies against β-actin and SM α-actin as indicated. Representative Western blots from a total of 3 independent experiments are shown. C: the levels of SM α-actin mRNA were examined by quantitative real-time PCR.

Fig. 2.

TGF-β1-induced stress fiber formation and SM α-actin expression are dependent on Rho kinase activity. HLF were stimulated with 1 ng/ml TGF-β1 in the presence or absence of 10 μM Y-27632 for indicated times. A: time course of cofilin phosphorylation in response to TGF-β1 as assessed by Western blotting of total lysates with phospho-cofilin (Ser-3) antibodies. max, Maximum. B: inhibition of TGF-β1-induced cofilin phosphorylation, stress fiber formation, and SM α-actin expression by Y-27632. C and D: Y-27632 has no effect on TGF-β1-induced Smad2 phosphorylation (P) as examined by Western blotting of total cell lysates with phospho-Smad2 antibodies (C) or on Smad-dependent gene transcription as assessed by SBE-luciferase (Luc) reporter (D). TK-RL, thymidine kinase-Renilla luciferase. Representative Western blots from a total of 2 (A) or 3 (B and C) independent experiments are shown, unless otherwise specified.

Fig. 3.

Inhibition of TGF-β1-induced cofilin phosphorylation, stress fiber formation, and SM α-actin expression by cycloheximide. HLF were stimulated with 1 ng/ml TGF-β1 or 500 nM jasplakinolide (Jasp) in the presence or absence of 1 μM cycloheximide (Chx) for indicated times. A: cycloheximide attenuates TGF-β1-induced cofilin phosphorylation and stress fiber formation without affecting stabilization of stress fibers by jasplakinolide. Representative Western blots from a total of 3 independent experiments are shown, unless otherwise specified. B: cycloheximide inhibits TGF-β1-induced transcription of SM α-actin mRNA, but not of plasminogen activator inhibitor 1 (PAI-1) mRNA, as assessed by corresponding real-time quantitative PCR (qPCR) assays.

Fig. 4.

TGF-β1-induced stress fiber formation and SM α-actin expression is dependent on Smad signaling. HLF were stimulated with 1 ng/ml TGF-β1 in the presence or absence of 10 μM SB-431542 for indicated times. Cell lysates or stress fiber fractions were analyzed by Western blotting as indicated. A: SB-431542 inhibits TGF-β1-induced stress fiber formation and SM α-actin expression. B and C: SB-431542 inhibits TGF-β1-induced Smad2 phosphorylation (B), without affecting sphingosine 1-phosphate (S1P)-induced extracellular signal-regulated kinase (ERK) 2 phosphorylation (C). Representative Western blots from a total of 2 (B and C) or 3 (A) independent experiments are shown.

Fig. 5.

Control of TGF-β1-induced SM α-actin expression and serum response factor (SRF) activation by actin dynamics. HLF were stimulated with 1 ng/ml TGF-β1 or 500 nM jasplakinolide in the presence or absence of 500 nM latrunculin B (Lat-B) for indicated times as indicated. A–C: latrunculin B inhibits TGF-β1-induced stress fiber formation and SM α-actin protein expression (A), SM α-actin mRNA synthesis (B), and SRF-luciferase reporter activity (C). D–F: latrunculin B has no effect on TGF-β1-induced Smad2 phosphorylation (D), Smad2/Smad3 nuclear translocation (E), or Smad-dependent gene transcription as assessed by SBE-luciferase reporter (F). Cyto, cytoplasmic; Nuc, nuclear. G–I: stabilization of stress fibers by jasplakinolide drives the expression of SM α-actin protein (G) and SM α-actin mRNA synthesis (H) without affecting TGF-β1-induced Smad2 phosphorylation (I). Representative Western blots from a total of 3 independent experiments are shown. *Statistical significance (P < 0.05) by Student's t-test.

Fig. 6.

TGF-β1-induced stress fiber formation is not dependent on SRF activity. HLF were treated with 10 μM CCG-1423 (A) or were transduced with adenovirus expressing short-hairpin RNA against SRF (Ad-shSRF) or the control adenovirus expressing shRNA against green fluorescent protein (Ad-shGFP) as indicated (B), followed by stimulation with 1 ng/ml TGF-β1 for 24 h. The stress fiber fractions or total cell lysates were analyzed by Western blotting with the desired antibodies as indicated. Representative Western blots from a total of 2 (B) or 3 (A) independent experiments are shown. C, control.

Fig. 7.

Regulation of TGF-β1-induced nuclear accumulation of megakaryoblastic leukemia-1 (MKL1) by latrunculin B. HLF were stimulated with 1 ng/ml TGF-β1 for 24 h in the presence or absence of 500 nM latrunculin B, as indicated. A: left, immunofluorescent staining of HLF with MKL1 antibodies; middle, DAPI staining of nuclei; right, merged images of MKL1 immunofluorescence (green), SM α-actin immunofluorescence (red), and DAPI images modified by Adobe Photoshop “Solarization” function to highlight the borders of nuclei (white). B: Western blotting of MKL1 in the cytoplasmic and nuclear fractions. Immunoblotting of nuclear lamin A/C and cytoplasmic 14–3-3β was performed to confirm the purity of the nuclear and cytoplasmic fractions, respectively. Representative Western blots from a total of 3 independent experiments are shown. C: MKL1 subcellular localization was quantitated via densitometry of Western blots from 3 independent experiments. MKL1 levels were normalized to the loading controls 14–3-3β (cytosolic) or lamin A (nuclear). The nos. in the bars are the mean values of the combined densitometry (from 3 experiments) of the chemiluminescence signal (detected by Western blotting of MKL1), for each condition (expressed as fold of control). * and #Statistical significance (P < 0.05) by Student's t-test with Bonferroni's correction for multiple comparisons.

Fig. 8.

TGF-β1 promotes MKL1 expression in an F-actin/SRF-dependent manner. HLF were treated with 500 nM latrunculin B (A and B) or 10 μM CCG-1423 (C) or transduced with Ad-shSRF or with control Ad-shGFP virus (D) as indicated, followed by stimulation with 1 ng/ml TGF-β1 for 48 h. Cell extracts were analyzed for MKL1 mRNA expression by real-time qPCR (A) or by Western blotting with desired antibodies (B–D). Representative Western blots from a total of 2 (D) or 3 (B and C) independent experiments are shown, unless otherwise specified.

Fig. 9.

Proposed model for a triphasic mechanism of SM α-actin expression by TGF-β1: 1) TGF-β1 stimulates a Smad-dependent expression of signaling molecules (X) driving activation of Rho (the candidate molecules are mentioned in the discussion); 2) Rho-induced stress fiber formation promotes MKL1 nuclear accumulation and SRF-dependent transcription of SM α-actin, as well as 3) of SRF itself and MKL1, both driving the SRF-dependent gene transcription further. ROCK, Rho-associated kinase.

Reverse transcription-quantitative real-time PCR.

Total RNA was harvested using RNA STAT-60 (Tell-Test) following the manufacturer's protocol. One microgram of total RNA was used as a template for random-primed reverse transcription using an iScript cDNA synthesis kit (Bio-Rad) according to the manufacturer's protocols. Real-time PCR analysis was performed using iTaq SYBR Green supermix with ROX (Bio-Rad) in a MyIQ single-color real-time PCR detection system (Bio-Rad). PCR primers contained the following sequences: human SM α-actin, forward AAAGACAGCTACGTGGGTGACGAA and reverse TTCCATGTCGTCCCAGTTGGTGAT; human plasminogen activator inhibitor 1 (PAI-1), forward GTGGTCTGTGTCACCGTATC and reverse GTAGTTGAATCCGAGCTGCC; and human MKL1, forward GGCTTGAGGAACCATTTTCCT and reverse ACCCCCACTTCTCTGCTGAA. The 18S transcript was used to normalize the amount and quality of the extracted RNA, using the following primer set: 18S forward GATTAAGTCCCTGCCCTTTG and reverse GTTCACCTACGGAAACCTTG.

Indirect immunofluorescence microscopy.

Cells were washed two times with PBS and fixed either in 4% paraformaldehyde/PBS for 15 min at room temperature followed by permeabilization with 0.2% Triton X-100/PBS for 5 min (Fig. 1A) or fixed in 100% methanol for 15 min at −20°C and dried for 1 h (see Fig. 7). Cells were then incubated with 2% BSA in PBS for 1 h, followed by incubation with desired primary antibodies at 4°C overnight, five washes with PBS within 15 min, incubation with corresponding rhodamine- or fluorescein-conjugated secondary antibodies for 1 h at room temperature, and five washes with PBS within 15 min. The cover slips were mounted using Vectashield mounting medium containing DAPI nuclear stain (Vector Laboratories, Burlingame, CA). The immunofluorescence images were obtained under the Leica fluorescent microscope.

Statistical analysis.

All of the data represent the results of at least three independent experiments. Quantitative data were analyzed by the Student's t-test with Bonferroni's correction for multiple comparisons when appropriate. Values of P < 0.05 were considered as statistically significant.

RESULTS

TGF-β1 induces delayed stress fiber formation in HLF by a dual mechanism.

A key morphological feature of differentiated myofibroblasts is the presence of actin-containing stress fibers (16, 43, 88). However, the data on the time course of TGF-β1-induced stress fiber formation in fibroblasts are conflicting: some studies reported rapid (within minutes-hours) F-actin formation (54, 69, 73, 85) while others showed a much more delayed effect (hours-days) (19, 21, 39, 88), largely using phalloidin staining for detection of stress fibers. In our experiments, we did not observe significant changes in phalloidin staining of primary cultured HLF within up to 6 h of TGF-β1 stimulation (data not shown). However, stimulation of HLF with TGF-β1 for 24 h resulted in a clearly visible increase in stress fiber formation, as assessed by phalloidin staining (Fig. 1A). Likewise, SM α-actin was not detected by immunofluorescence in significant quantities under serum starvation or within several hours of TGF-β1 stimulation (data not shown), but its staining was increased substantially after 24 h of TGF-β1 treatment, also showing a predominant colocalization with phalloidin-labeled stress fibers (Fig. 1A). From these data, it was unclear whether TGF-β1 promoted a delayed F-actin formation in HLF or whether increased phalloidin staining simply reflected increased SM α-actin expression in response to TGF-β1.

Because phalloidin does not discriminate between actin isoforms and because of the semiquantitative nature of immunofluorescent staining, we next performed in vitro isolation of stress fibers (40), allowing for a precise biochemical quantification of various actin isoforms in stress fibers. As shown in Fig. 1B, TGF-β1 stimulates a substantial but delayed incorporation of β-actin and γ-actin in stress fibers, whereas the total amounts of these actin isoforms were unchanged. The kinetics of β-actin and γ-actin polymerization parallel the delayed induction of SM α-actin mRNA synthesis (Fig. 1C) and expression of SM α-actin protein, which also largely accumulates in the stress fiber fraction (Fig. 1B). Together, these data suggests that 1) TGF-β1-induced stress fiber formation is mediated by both polymerization (β-actin, γ-actin) and expression (SM α-actin) of corresponding actin isoforms; and 2) because of time courses that exhibit similarly delayed kinetics, these two processes may be mutually dependent. To explore this possibility, we then examined the signaling mechanism for these processes in response to TGF-β1.

Rho signaling is required for delayed stress fiber formation and SM α-actin expression in response to TGF-β1.

Actin stress fiber formation in response to serum or to G protein-coupled receptor agonists is promoted rapidly via the activation of small GTPases of the Rho family and their target Rho-associated kinase (ROCK). The latter activates LIM kinase (LIMK), which, in turn, inactivates the actin-depolymerizing factor cofilin through its phosphorylation at the serine-3 site, leading to stabilization of F-actin (46, 56, 61). Therefore, we used phosphorylation of cofilin at serine-3 as a reporter for Rho/ROCK/LIMK signaling to determine whether TGF-β1-induced actin stress fiber formation was mediated by this pathway. Western blot analysis with phospho-cofilin (Ser-3) antibodies showed that cofilin phosphorylation in response to TGF-β1 also occurred in a delayed manner (Fig. 2A) and paralleled that of stress fiber formation (Fig. 1B), suggesting a role for the Rho/ROCK pathway in this process. To confirm this, we used a specific pharmacological inhibitor of ROCK1, Y-27632, that does not affect the activity of many other kinases that could affect cell growth or differentiation (11, 38, 83). As shown in Fig. 2, 10 μM Y-27632 fully inhibited TGF-β1-induced cofilin phosphorylation, stress fiber formation, and SM α-actin expression (Fig. 2B) without decreasing TGF-β1-induced Smad phosphorylation (Fig. 2C) or Smad-dependent gene transcription, as assessed by SBE-Luc reporter (Fig. 2D).

Stress fiber formation and SM α-actin gene transcription in response to TGF-β1 require de novo, Smad-dependent protein expression.

Given the delayed stress fiber formation in response to TGF-β1 (Fig. 1), we hypothesized that de novo protein expression was required for these processes. To test this, we used the inhibitor of protein synthesis cycloheximide. As shown in Fig. 3A, cycloheximide fully inhibited cofilin phosphorylation and actin stress fiber formation in response to TGF-β1. This effect was not due to the inability of actin to polymerize, since stress fiber formation induced by the direct stabilizer of actin filaments jasplakinolide was unaffected by treatment with cycloheximide (Fig. 3A). Interestingly, jasplakinolide completely abolished the basal phosphorylation of cofilin, suggesting that there could be a reverse regulation of the Rho/ROCK/LIMK pathway by actin dynamics. Importantly, cycloheximide attenuated SM α-actin mRNA induction in response to TGF-β1 while having no effect on PAI-1 mRNA production, which occurs through canonical Smad signaling (13) (Fig. 3B). This suggests that SM α-actin gene transcription requires additional translational events beyond Smad-mediated transcriptional regulation.

Smad-dependent signaling in response to TGF-β1 occurs rapidly (Fig. 2C), much earlier than stress fiber formation (Fig. 1B). Given the requirement for intermediate protein synthesis in TGF-β1-induced stress fiber formation and SM α-actin expression (Fig. 3), we examined if Smad-dependent gene transcription is responsible for these responses using the inhibitor of the TGF-β receptor kinase activity SB-431542, which blocks TGF-β1-induced phosphorylation and activation of Smad2 and Smad3 (28). As shown in Fig. 4A, SB-431542 fully inhibits TGF-β1-induced stress fiber formation and SM α-actin expression. Control experiments confirmed inhibition of TGF-β1-induced Smad2 phosphorylation by SB-431542 (Fig. 4B) and demonstrated its specificity, since SB-431542 did not affect phosphorylation of the extracellular signal-regulated kinases 1/2 in response to the G protein-coupled agonist sphingosine 1-phosphate (S-1-P) (Fig. 4C).

Together, these data suggest that TGF-β1-induced delayed stress fiber formation and SM α-actin gene transcription occur through a biphasic signaling mechanism involving Smad-dependent expression of intermediate proteins that further drive these processes.

TGF-β1-induced actin stress fiber formation is required for SRF activation and myofibroblast differentiation.

We have recently shown that TGF-β1-induced SM α-actin gene transcription and protein expression in HLF require SRF (63), the activation of which is known to be induced by serum or G protein-coupled receptor agonists through modulation of actin dynamics (48, 51). On the other hand, SM α-actin expression is rapidly incorporated into stress fibers, augmenting intracellular force generation and the development of enlarged focal adhesion complexes (33). This process results in elevated tension across the cell, which could further promote stress fiber formation (68). Conversely, depletion of SM α-actin with a monoclonal antibody has been observed to disrupt stress fiber integrity (71). Finally, SRF itself can control the expression of cytoskeletal proteins, including those that promote actin polymerization and bundling (53). Given these previous observations, we considered two hypotheses that could explain the similar kinetics between stress fiber formation and SM α-actin expression in response to TGF-β1 (Fig. 1): 1) SRF activation and SM α-actin expression could require the polymerization of ubiquitous β-actin into stress fibers; and/or 2) conversely, β-actin stress fiber formation could depend on SRF and SM α-actin expression.

To test the first possibility, we used the actin-filament disrupting agent latrunculin B. Pretreatment of HLFs with latrunculin B fully inhibited TGF-β1-induced incorporation of β-actin into stress fibers without affecting the total levels of β-actin (Fig. 5A). This effect of latrunculin B was accompanied by inhibition of TGF-β1-induced SM α-actin expression at both protein (Fig. 5A) and mRNA (Fig. 5B) levels as well as by inhibition of SRF-dependent gene transcription, as assessed by SRF-Luc reporter (Fig. 5C). In contrast, latrunculin B had no effect on TGF-β1-induced Smad2 phosphorylation (Fig. 5D), Smad2/3 nuclear translocation (Fig. 5E), or Smad-dependent gene transcription (Fig. 5F). Furthermore, stabilization of filamentous actin by jasplakinolide was sufficient for stimulation of SM α-actin protein expression (Fig. 5G) and SM α-actin mRNA synthesis (Fig. 5H) in the absence of TGF-β1. In control experiments, jasplakinolide had no effect on TGF-β1-induced Smad2 phosphorylation (Fig. 5I).

To test the second possibility (that β-actin stress fiber formation could depend on SRF activity and on SM α-actin expression in response to TGF-β1), we used 1) the pharmacological SRF inhibitor CCG-1423 (22), which blocks TGF-β1-induced SM α-actin expression without affecting Smad signaling (63), and 2) the knockdown of SRF, using adenovirus-mediated transduction of short-hairpin DNA against SRF transcript (Ad-shSRF) (8, 63). Figure 6A shows that, while CCG-1423 inhibited SM α-actin expression as expected, it had no effect on TGF-β1-induced incorporation of β-actin into stress fibers. Likewise, knockdown of SRF by Ad-shSRF transduction (but not of control Ad-shGFP) completely abolished the expression of SRF and, as a result of SM α-actin expression, without affecting stress fiber formation in response to TGF-β1 (Fig. 6B).

Together, these data suggest that 1) TGF-β1-induced transcription of SM α-actin in HLF is mediated by β-actin stress fiber formation (Fig. 5) that is induced through the Smad-dependent expression of intermediate signaling molecules (Figs. 3 and 4), whereas 2) neither of these processes requires SRF activation or the expression of SM α-actin (Fig. 6).

Stress fiber formation mediates TGF-β1-induced SRF activation through the expression and nuclear accumulation of MKL1/MRTF-A.

SRF activation in response to serum or to G protein-coupled receptor agonists is known to be controlled by actin dynamics through the regulation of the subcellular localization of the coactivator of SRF MKL1/MRTF-A (27, 31, 51, 87). Briefly, monomeric actin sequesters MKL1 in the cytosol via a direct interaction (27). Actin polymerization results in the release of MKL1, and, because of its nuclear localization signal, MKL1 readily translocates to the nucleus and potentiates SRF transcriptional activity (51). Having established the dependency of TGF-β1-induced SM α-actin expression on stress fiber formation (Fig. 5) and on SRF activity (63), and given that actin dynamics do not affect the signaling and gene transcription mediated by Smads in HLF (Fig. 5, D-F and I), we examined if this role of stress fiber formation in mediating TGF-β1-induced myofibroblast differentiation is related to a nuclear accumulation of MKL/MRTF, focusing on the endogenous MKL1/MRTF-A isoform.

Immunofluorescence microscopy of HLF showed that, under basal conditions, endogenous MKL1 is largely localized in the cytoplasm (Fig. 7A). We did not observe significant changes in MKL1 localization during the first 2 h of TGF-β1 treatment (data not shown), as was reported for serum- or G protein-mediated rapid MKL1 translocation (31, 51), consistent with our observation of delayed stress fiber formation in response to TGF-β1 (Fig. 1B). However, after 24 h treatment with TGF-β1, the increase in nuclear staining of MKL1 was clearly visible (Fig. 7A). Importantly, latrunculin B treatment prevented TGF-β1-induced nuclear staining of MKL1 (Fig. 7A).

Given the semiquantitative nature of immunofluorescent microscopy, we then examined MKL1 translocation by in vitro nuclear/cytoplasmic fractionation. TGF-β had no effect on MKL1 distribution within a 2-h period (data not shown) but induced a substantial accumulation of MKL1 in the nuclear fraction after 24 h of stimulation, which was blocked by treatment with latrunculin B (Fig. 7, B and C). Notably, we did not observe a reduction of MKL1 in the cytoplasmic fraction upon TGF-β1 treatment but rather saw a slight increase that was also attenuated by latrunculin B (Fig. 7, B and C). This suggested that TGF-β1 may also promote the expression of MKL1 in a manner dependent on actin polymerization.

To test this possibility, we examined the total levels of MKL1 in TGF-β1-treated HLF. Indeed, TGF-β1 stimulated a significant increase in MKL1 mRNA and protein expression that was inhibited by latrunculin B (Fig. 8, A and B). Given that latrunculin B attenuated the activation of SRF (Fig. 5C) but not of Smad-dependent gene transcription (Fig. 5, D-F), we examined whether TGF-β1-induced MKL1 expression was dependent on SRF activation. As shown in Fig. 8C, the induction of MKL1 expression in response to TGF-β1 was attenuated by the pharmacological inhibitor of SRF CCG-1423, which does not affect Smad signaling (63) or TGF-β1-induced β-actin stress fiber formation (Fig. 6A). Furthermore, the knockdown of SRF using adenovirus-mediated expression of SRF-shRNA also attenuated MKL1 expression in response to TGF-β1 (Fig. 8D) without affecting β-actin polymerization (Fig. 6D). These data suggest that both nuclear accumulation and SRF-dependent MKL1 expression contribute to TGF-β1-induced myofibroblast differentiation.

DISCUSSION

The data described in this study support a triphasic model for a sustained myofibroblast differentiation in response to TGF-β1 that involves 1) initial Smad-dependent expression of intermediate signaling molecules driving Rho activation and stress fiber formation followed by 2) nuclear accumulation of MKL1 and activation of SRF as a result of actin polymerization and 3) SRF-dependent expression of MKL1 and of SRF itself (63, 75), driving further myofibroblast differentiation (Fig. 9).

Consistent with prior observations by some investigators (21, 39, 88), but not by others (54, 69, 73, 85), we see a formation of actin stress fibers in response to TGF-β1 only at delayed time points (Fig. 1). According to our data, the formation of actin stress fibers (but not of Smad-dependent gene transcription) is Rho kinase-dependent (Fig. 2) while both the Rho signaling and F-actin formation depend on Smad-mediated gene expression (Figs. 3 and 4). This may explain the delayed kinetics of Rho activation and actin polymerization in response to TGF-β1 and places Smad signaling upstream of these processes. Some of the direct Smad target genes that are rapidly induced by TGF-β1 and may mediate Rho activation and stress fiber formation at various signaling levels (Fig. 9) have been suggested by previous studies. This may include the rapid expression of endothelin-1 (62) that can activate Rho through G protein signaling (26); of sphingosine kinase-1 (42, 91) that produces S-1-P, a G protein-coupled receptor agonist activating Rho signaling (50); of Rho guanine exchange factors NET1, which directly activates Rho (45, 69), or GEF-H1/Lfc (82); or of Rho family members themselves, including RhoB (85, 86).

Rho-dependent signaling is now increasingly implicated as a central mediator of myofibroblast differentiation from other cell sources, such as during epithelial-mesenchymal transition in response to various environmental cues such as TGF-β (49, 52), G protein-coupled receptor agonists, inflammatory mediators (7), and cell-cell contact disruption (23). Our investigations help elucidate the mechanistic linkage between Smads and Rho-dependent signals in the TGF-β signaling cascade during myofibroblast differentiation. Interestingly, previous studies support an interplay between Smads and Rho signaling during the epithelial-mesenchymal transition, with inhibition of both Smad- and Rho-dependent signals required to fully revert mesenchymal cells to an epithelial phenotype (10). These investigations support the hypothesis that, although these pathways are linked, they may not be redundant in their effects.

The other important finding of this study is that TGF-β1-induced myofibroblast differentiation is mediated by actin polymerization (Fig. 5), which was originally hypothesized by us based on a striking similarity between the kinetics of stress fiber formation and SM α-actin expression in response to TGF-β1 (Fig. 1). Although this role of actin polymerization is well-defined for G protein-mediated SM α-actin expression, it was not previously demonstrated for TGF-β1-induced myofibroblast differentiation, and it would not be predicted based on the previously published data on the role of Smads, p38 mitogen-activated protein kinase, and krüppel-like factors in the transcriptional regulation of the SM α-actin gene by TGF-β1 (1, 9, 17, 30, 36, 70, 73, 84). In our experiments, disruption of actin stress fiber formation, either directly by latrunculin B or indirectly via inhibition of protein synthesis with cycloheximide, fully inhibited SM α-actin mRNA expression by TGF-β1 (Figs. 3 and 5), confirming the requirement for an intermediate gene expression step, not only for actin filament formation but also for the expression of SM α-actin. Importantly, disruption of the actin cytoskeleton did not affect Smad-dependent gene transcription (Figs. 3 and 5), which is upstream of actin polymerization in the TGF-β1 signaling pathway.

Actin dynamics mediates G protein-induced SM gene expression through activation of SRF via a rapid nuclear localization of the SRF coactivator MKL1/MRTF-A (47, 48, 51, 74). We have shown previously the requirement of SRF for TGF-β1-induced SM α-actin expression in pulmonary fibroblasts (63), whereas others have shown the requirement of MKL1/MRTF-A for this process as well as TGF-β1-induced nuclear translocation of endogenous MKL1 in kidney epithelial cells (53) and of ectopically expressed MKL1 in cardiac fibroblasts (72). However, the role of actin dynamics in mediating TGF-β1-induced MKL1 translocation was not examined. Consistent with these studies, our data demonstrate a delayed nuclear accumulation of endogenous MKL1 in the response of pulmonary fibroblasts to TGF-β1, which is inhibited by latrunculin B and therefore requires stress fiber formation (Fig. 7).

More importantly, we show for the first time that TGF-β1 also stimulates expression of MKL1 at both mRNA and protein levels in a manner dependent on stress fiber formation and SRF activity (Fig. 8). Together with SRF-dependent SRF expression (63, 75, 76), this may serve to amplify TGF-β1 signaling and provide a sustained expression of SM α-actin. At this time, we know that TGF-β1 promotes MKL1 expression at the mRNA level (Fig. 8A); however, the mechanism remains to be determined, and this is the focus of ongoing studies.

In summary, there are several important points to be taken from this study. First, our data support a versatile role for the actin cytoskeleton in TGF-β1-induced myofibroblast differentiation. While the formation of actin stress fibers is a defining morphological feature of the myofibroblast, mediating such functions of myofibroblast as cell contraction and focal adhesion, our data demonstrate that actin stress fiber formation also simultaneously augments the gene expression characterizing myofibroblast differentiation via transcriptional regulation of SRF-dependent genes. This provides an elegant mechanism by which cell structural changes lead to a positive-feedback loop via upregulation of structural and contractile genes. Second, delayed regulation of SRF-dependent gene expression likely accounts for the delayed kinetics seen with TGF-β1-induced SM α-actin expression, reaching a maximum after 72 h of TGF-β1 stimulation (our unpublished observations). From this perspective, the later actin/MKL1/SRF signaling may serve as an important amplifier of the TGF-β1 response through further upregulation of SRF and MKL1. Finally, actin signaling may serve as an intracellular integrator of divergent “profibrotic” inputs such as matrix stiffness, local force application, and other G protein-coupled stimuli (3, 26, 42, 92). Thus, this signaling mechanism may be an attractive target for disrupting in vivo fibrogenesis.

GRANTS

This study was supported by National Institutes of Health Awards R01 HL-71755 (N. O. Dulin), R01 GM-85058 (N. O. Dulin), K08 HL-093367-01A1 (N. Sandbo), and T32 HL-07605 (D. Yau.); American Lung Association Dalsemer Award DA-161895-N (N. Sandbo); a Central Society for Clinical Research Early Investigator Award (N. Sandbo); and American Heart Association Fellowships 0825868G (D. Yau) and 10PRE4190120 (J. Kach).

DISCLOSURES

No conflicts of interest are declared by the authors.