MAP-kinase activity necessary for TGFβ1-stimulated mesangial cell type I collagen expression requires adhesion-dependent phosphorylation of FAK tyrosine 397

The pathogenic importance of transforming growth factor β (TGFβ) in glomerulosclerosis has been supported by clinical and experimental data. Among the pathophysiological roles of this multi-functional cytokine, induction of extracellular matrix (ECM) expression is one of the crucial steps of the sclerosing process, not only in the kidney but also in fibrosis of other organs such as the lungs and skin (Leask and Abraham, 2004). However, the intracellular signals mediating TGFβ-stimulated ECM accumulation are not fully understood.

Our laboratory has been working to delineate the intracellular mechanism(s) by which TGFβ signals are transduced to stimulate ECM accumulation in cultured human kidney mesangial cells. We previously determined that Smad proteins, the major TGFβ-receptor substrate, mediate TGFβ1 stimulation of collagen production in these cells (Poncelet and Schnaper, 1998). Furthermore, TGFβ1 activates extracellular signal-regulated kinase (ERK) MAP kinase and ERK synergizes with Smad pathways to optimize mesangial cell Smad activation (Hayashida et al., 2003; Hayashida et al., 1999). We also demonstrated that TGFβ1 enhances stress-fiber assembly and focal-adhesion formation, and that disrupting cytoskeletal integrity decreases TGFβ-stimulated collagen production (Hubchak et al., 2003).

Recent reports indicate that TGFβ can activate various other signaling molecules in addition to Smad proteins (Feng and Derynck, 2005). Although each of these pathways was initially described as being isolated and linear, it has become increasingly apparent that they are mutually regulated by their interaction at various levels from the membrane to the nucleus. These interactions have been reported to enhance or inhibit Smad activity (Moustakas and Heldin, 2005). One such pathway of significant attention is ERK MAP kinase (Javelaud and Mauviel, 2005). Although initially ERK was reported to inhibit TGFβ/Smad signaling via a mechanism involving phosphorylation of the linker regions of receptor (R)-Smads with ensuing decreased Smad nuclear localization (Kretzschmar et al., 1997; Kretzschmar et al., 1999), subsequent studies have made this an area of some controversy (Cordenonsi et al., 2007; Javelaud and Mauviel, 2005; Massague, 2003), with different conclusions based upon the type of cell being investigated and the downstream effect of TGFβ that is being examined, as well as how the ERK MAP kinase is activated (Davis, 1993; Friedman and Perrimon, 2006).

Various studies, including ours, have reported that TGFβ stimulation of the ECM expression appears to be ERK dependent [cf. Hayashida et al. (Hayashida et al., 2003) and references therein]. Our laboratory has studied the effects of TGFβ1/Smad3 on human kidney glomerular mesangial cells, a mesenchymal cell type with multiple roles in kidney physiology. Using type I collagen expression as an indicator of fibrogenic activity, we found that ERK MAP kinase plays an important positive role in promoting type I collagen expression (Hayashida et al., 1999; Hayashida and Schnaper, 2004). ERK is necessary but not sufficient for the collagen response, because activating ERK without Smad activation did not lead to collagen expression (Runyan et al., 2003). The effect appears to be related to ERK-stimulated phosphorylation of the phosphoacceptor sites in the Smad3 linker region (Hayashida et al., 2003). Importantly, although ERK was activated by TGFβ1 in mesangial cells, and this activation was abrogated immediately by the addition of MEK/ERK inhibitor, the inhibitor affected ERK-dependent phosphorylation of the Smad3 linker region only when the inhibitor was present at least 40 minutes prior to the addition of TGFβ1. This result suggested that this TGFβ1-stimulated ERK is not the source of ERK activity that is crucial for type I collagen expression (Hayashida et al., 2003). Therefore, further study was required to elucidate the mechanism by which the ERK pathway is activated to interact with TGFβ/Smad signaling.

Because we previously demonstrated that TGFβ1 enhances stress-fiber assembly and focal adhesion (FA) formation (Hubchak et al., 2003), we studied whether the FA could play a role in ERK activity that is important for mesangial cell fibrogenesis. One protein in the FA that could potentially link several signaling pathways is focal adhesion kinase (FAK). This 125-kDa protein is a major component of the FA throughout its maturation (Cukierman et al., 2002). Despite its original characterization in association with cell adhesion and migration, recent studies have delineated more diverse roles for this kinase. FAK provides a `dock' for various FA proteins and related signaling molecules to transduce extracellular signals initiated by the binding of integrin to the ECM (Boudreau and Jones, 1999; Geiger et al., 2001). FAK also can be activated directly via binding of growth factor receptors such as PDGF and EGF (Sieg et al., 2000). Therefore, FAK serves as a point that converges signals from the ECM and those initiated by growth factors. Upon integrin engagement, the tyrosine residue at position 397 of FAK (FAK Y397) is auto-phosphorylated, providing a binding site for SH2-or SH3-containing molecules, a prototype for which is the Src family kinases. The Src-FAK complex facilitates further phosphorylation on several different FAK tyrosine residues by either FAK kinase activity itself (Toutant et al., 2002) or kinase activity of Src (Schaller, 2001). Motifs that include each phosphorylated tyrosine have differential affinity for different adapter proteins that are recruited to the FA, determining downstream signaling events (DeMali et al., 2003; Mitra et al., 2005; Wozniak et al., 2004).

ERK is a downstream effector of FAK (Gu et al., 1999; Ruest et al., 2001; Schlaepfer and Hunter, 1996). FAK mediates adhesion-dependent ERK activity (Schlaepfer et al., 1998; Zhu and Assoian, 1995), and ERK has been suggested to mediate FA disassembly and cell migration by recruiting a protease, calpain 2 (Carragher et al., 2003). Stimuli for FAK-mediated ERK activation could include adhesion to the ECM and/or a growth factor binding to its receptor. However, previous reports describe different mechanisms depending on cell type and culture conditions. Little is known regarding whether and how FAK and its downstream effectors play a role in TGFβ signaling. Here, we investigate how these signals might be linked. Our results suggest that FAK could provide a bridge between cytoskeletal and ERK signaling, promoting the fibrogenic response to TGFβ.

We previously reported that TGFβ1 treatment intensifies stress fibers and promotes FA rearrangement in human mesangial cells (Hubchak et al., 2003). To further investigate the mechanism by which FAs modulate TGFβ/Smad signaling in mesangial cells, we investigated FAK. We first performed western blot analysis using phospho-specific FAK antibodies to determine the phosphorylation status of the Y397 and Y925 residues. The immunoblots demonstrated that phosphorylation of Y397, which is typically associated with cell adhesion, was present under basal conditions and was minimally affected by TGFβ1 treatment. By contrast, phosphorylation of Y925 was minimal in control conditions but became apparent by 30 minutes after TGFβ1 treatment (Fig. 1A). These findings were confirmed by immunocytochemistry for phospho-Y397 FAK and phospho-Y925 FAK. Phospho-Y397 FAK was basally detected in a pattern that is characteristic for FAs and that appeared to be further intensified by TGFβ1, whereas Y925 phosphorylation became apparent only after TGFβ1 treatment (Fig. 1B). Immunostaining for either phosphorylated-or pan-FAK was localized mainly at the sites of FA, determined by co-staining for vinculin, another known FA component (supplementary material Fig. S1).

To determine whether cytoskeletal changes are related to FAK activity, we next stained the cells for FAK phosphorylation specific for the adhesion-dependent Y397 residue, and for F-actin to detect cytoskeletal structure. Some of the cells were plated on a surface coated with poly-L-lysine (pLL), which facilitates cell adhesion in an integrin-independent manner and has been used to disrupt focal adhesions (Arora et al., 1995). As shown in Fig. 2A, mesangial cells started forming spindle-like processes that were detected by vinculin staining, along with some stress fibers, as early as 3 hours after re-plating on gelatin-coated glass slides (left panels), whereas cells cultured on pLL spread little, if at all, in 3 hours, with vinculin condensed in a punctate pattern within the cell center (right panels). By 6 hours, the cells re-plated on the control surface (gelatin) had attached and spread with distinct phospo-Y397 FAK staining (Fig. 2Ba, upper-left quadrant) in a similar pattern to what we previously observed with vinculin staining for FAs in quiescent cells cultured on plastic (Hubchak et al., 2003). FA staining became more intense and arrayed in a series of linear, parallel structures after 30 minutes of TGFβ1 treatment (Fig. 2Ba, upper-right quadrant). Moreover, these intensified structures coalesced with the ends of F-actin fibers, as depicted by the yellow color on the overlay (Fig. 2Bb, upper-right quadrant) and colocalizing pixels (Fig. 2Bc, upper-right quadrant). These findings suggest that FA assembly and FAK phosphorylation are coordinated spatially and temporally with cytoskeletal structural changes upon TGFβ1 treatment. By contrast, cytoskeletal structures in cells on pLL were less prominent, and FAs detected by phospho-Y397 immunostaining were disorganized and did not develop into more mature, spindle-like structures (Fig. 2B, lower quadrants, panels a). Also, TGFβ1 treatment did not induce phospho-Y397 FAK staining to co-localize with the end of stress fibers in cells plated on pLL (Fig. 2B, lower right quadrant, panels b and c, and graph depicting quantified colocalizing pixels). These results were corroborated by immunoblotting for phospho-Y397 and phospho-Y925 FAK. Phospho-Y397 FAK was present in cells on plastic but not on pLL (Fig. 2C, top panel). TGFβ1-stimulation of Y925 FAK phosphorylation was also abrogated in cells on pLL (Fig. 2C, middle panel, cell images not shown). These findings indicate that cells on pLL manifest disorganized FAs and thus might serve as a tool to test the role of FAK in TGFβ signaling.

We next examined type I collagen gene activation by TGFβ1 in cells plated on pLL. TGFβ-induced collagen mRNA expression was decreased in cells plated on pLL (Fig. 3A). A–0.4-kb segment of the α2(I)-procollagen-gene promoter upstream of a luciferase reporter was activated by TGFβ1 by 2.18-fold over control, whereas this promoter failed to respond to TGFβ1 in cells on pLL (Fig. 3B). These results strongly suggest that proper FA formation is required for TGFβ1 to stimulate transcription of the type I collagen gene. The observed effects of pLL appeared not to reflect a change in TGFβ1 receptor capacity, which we found to be unchanged, as determined by a receptor-binding assay using ¹²⁵I-TGFβ1 (data not shown). Following ligand binding to the receptor, Smad3 is phosphorylated by the kinase activity of the type I receptor at its C-terminus, which we previously reported to be crucial for TGFβ1 induction of collagen accumulation (Poncelet et al., 1999). In cells on pLL, Smad3 C-terminal phosphorylation was not different from that in cells on a control surface (Fig. 4A, top panel). Recent reports describe several Smad linker-region serine residues that are phosphorylated via interacting signaling molecules, such as ERK, PKC and Ca²⁺-calmodulin kinases (Moustakas and Heldin, 2005). We previously showed that ERK MAP kinase activity enhances TGFβ/Smad signaling via ERK-dependent phosphorylation of the R-Smad linker-region and plays a role in TGFβ1 induction of collagen production (Hayashida et al., 2003). Phosphorylation of the Smad3 linker region, detected by an antibody that recognizes Smad3 phosphorylated at Ser207/Ser212 (Mori et al., 2004), was significantly lower in cells plated on pLL compared with those on gelatin (Fig. 4A, middle panel and solid squares in the graph on right). By contrast, C-terminal phosphorylation of Smad3 was not affected by reduced Y397 phosphorylation of FAK (Fig. 4A, top panel and solid triangles in the graph).

Because FAK activity can lead to ERK activation, we next evaluated ERK activity in cells on pLL. Elk-Gal transactivation assay, a downstream readout of ERK-mediated transcriptional activity, showed a 1.82-fold induction by TGFβ1; this induction was also abrogated by culturing cells on pLL (Fig. 4B). These results suggest that defective cell-substrate interaction in cells on pLL impairs TGFβ1-induced collagen production via an effect on ERK, possibly via stimulation of R-Smad phosphorylation. To test this possibility, we used a construct that expresses constitutively active MAP kinase kinase (caMEK) that directly activates ERK and bypasses the cell-substrate interaction that is required for ERK activity. Cells expressing caMEK or an empty vector were replated on gelatin or pLL. When caMEK was present, ERK was highly activated in cells either on pLL or control, with or without TGFβ1, as expected (Fig. 5A, left). Further results suggested that this caMEK-derived ERK activity is sufficient to support TGFβ1-stimulated type I collagen promoter activity (Fig. 5A, right), even on pLL. Although overall promoter activity was somewhat lower in cells on pLL, co-transfection of caMEK restored the ability of TGFβ1 to activate the α2(I) collagen promoter. To further support the importance of ERK activity in our model, caMEK restored phosphorylation of the Smad3 linker region (Fig. 5B). These results indicate that adhesion-dependent ERK activity is important to facilitate TGFβ-stimulation of collagen production and phosphorylation of the R-Smad linker region.

To define a specific pathway for cell-substrate interaction in type I collagen expression, we next examined the role of specific FAK tyrosines. An Elk-Gal transactivation assay was performed with cells that express either wild-type, Y397F or Y925F FAK. Both Y397F and Y925F FAK mutants inhibited TGFβ1-stimulated Elk-Gal activity (Fig. 6A). Interestingly, α2(I) collagen promoter activity was affected largely by the presence of the Y397F FAK mutant, but to a much lesser extent with Y925F FAK (Fig. 6B). The FAK mutants did not cause changes in cell spreading or FA formation (supplementary material Fig. S2). These results suggest that, although phosphorylation of several different FAK tyrosines might lead to ERK activity, adhesion-dependent FAK phosphorylation is crucial for the type I collagen response. To determine whether this Y397 FAK phosphorylation could affect collagen expression through ERK activity, we examined the α2(I) collagen promoter response in the presence of Y397F or wild-type FAK, along with caMEK or an empty vector. The inhibitory effect of Y397F FAK on promoter activation by TGFβ1 was rescued by caMEK overexpression (Fig. 7). When Y397F FAK was present, TGFβ1 induction of phosphorylation of the R-Smad linker region was abrogated, whereas C-terminal phosphorylation remained intact (Fig. 8A). When Smad 3EPSM, which lacks all the potential ERK-targeted phosphorylation sites in the Smad3 linker region (Kretzschmar et al., 1997), was co-transfected instead of wild-type Smad3, linker-region phosphorylation was no longer detectable, whereas C-terminal phosphorylation was intact, confirming the specificity of the antibodies (Fig. 8B). Therefore, these results indicate a crucial role for phospho-Y397-FAK-derived ERK activity in phosphorylation of the Smad3 linker region.

In order to evaluate the physiological relevance of our in vitro findings, we examined Smad3 and FAK phosphorylation in kidney samples of bleomycin-treated (bleo) mice. These mice have been shown to develop a scleroderma-like skin lesion (Lakos et al., 2004; Takagawa et al., 2003) and lung fibrosis (Venkatesan et al., 2004) in a TGFβ-dependent manner, and therefore serve as an animal model in which to study TGFβ-induced fibrosis. Type I collagen expression was dramatically upregulated in the kidneys of bleo mice compared with those from control mice (Fig. 9A, upper blot), indicating a fibrotic response in kidneys of bleo mice. Phosphorylation of Y397 FAK was clearly detected in bleo mice by immunostaining (Fig. 9B, upper images), as well as by immunoblotting (not shown), whereas it was minimally detected in control kidneys. Interestingly, total FAK expression levels were also increased in bleo mice (Fig. 9B, lower images). Smad3 phosphorylation at both the C-terminus and the linker region was also enhanced in bleo mice (Fig. 9A, middle blots). Levels of total Smad3 were significantly downregulated in bleo-mouse kidneys, as has been previously described in the lung of bleo mice (Zhao and Geverd, 2002) and in TGFβ-treated kidney cells (Poncelet et al., 2007).

Taken together, the present results support a crucial role for phospho-Y397-FAK-derived ERK activity and subsequent phosphorylation of the Smad3 linker-region in TGFβ1-stimulated type I collagen expression.

FAK is one of the central participants in the signaling events transduced by cell-substrate interaction (DeMali et al., 2003; Geiger et al., 2001; Mitra et al., 2005; Schaller, 2001; Wozniak et al., 2004). Over the past decade, the actions of FAK have been elucidated. Via association with specific adaptor molecules at its SH2 or SH3 docking site, FAK modulates a wide range of signaling cascades, including ERK MAP kinase, PI3 kinase and the Rho family of small GTPases. However, the biological functions of this protein are relatively unknown other than those in relation to invasive cell phenotypes.

We have been studying the mechanism(s) by which TGFβ leads to type I collagen production in cultured human mesangial cells as a model for glomerular sclerosis (Schnaper et al., 2003). We have shown that various signaling pathways, such as those involving ERK MAP kinases (Hayashida et al., 2003; Hayashida et al., 1999), PKC δ (Runyan et al., 2003) and PI3K (Runyan et al., 2004), interact with and modulate TGFβ/Smad signaling. We also showed that cytoskeletal components contribute to Smad regulation (Hubchak et al., 2003). In the present study, we demonstrate that TGFβ1 activates FAK and leads to engagement of FAs with F-actin, a process known to enhance FA assembly. Interference with FA formation by plating cells on a surface that promotes attachment independent of integrin resulted in decreased TGFβ-stimulated collagen production, suggesting a role for signal(s) derived from the FA complex. The signal could involve changes in cell shape, intracellular tension or `tensegrity' status (Ingber, 2003), or biochemical signals such as those that originate from the FAK phosphorylation event. In this article, we focus on signaling events downstream of FAK phosphorylation.

Previously, we reported that the ERK MAP kinase plays an important role in the TGFβ/Smad signaling cascade, possibly by causing phosphorylation of the linker region of Smad3 (Hayashida et al., 2003; Hayashida et al., 1999). Here, we show that mesangial cell adhesion induces phosphorylation of FAK at residue Y397, whereas TGFβ1 treatment stimulates Y925 phosphorylation. Although a point mutation at either site prevents FAK-mediated ERK activity, only Y397 mutation inhibits TGFβ1-stimulated collagen-promoter activation. Rescue of the promoter response to TGFβ by overexpressing a constitutively active form of MEK in the presence of the Y397F FAK mutant or in cells cultured on pLL confirmed a crucial role for Y397 phosphorylation in both ERK activity and subsequent collagen expression. Together, these results define a pathway from adhesion-dependent Y397 FAK phosphorylation through ERK activity for fibrogenic signaling.

Regulation of TGFβ/Smad signaling by phospho-Y397 FAK in the present study is consistent with a recent report in which TGFβ induction of smooth muscle α actin in mouse lung fibroblasts was blocked by expression of a Y397F FAK mutant (Thannickal et al., 2003). Also, in myofibroblasts derived from patients with scleroderma, Y397 FAK phosphorylation is constitutively high (Mimura et al., 2005). In both cases, the expression of the Y397F FAK mutant did not affect R-Smad phosphorylation at receptor-specific C-terminal serine residues. Our results with cells on pLL or those expressing the Y397F FAK mutant, in which linker-region but not C-terminal phosphorylation of Smad3 was decreased, further support the notion that FAK modulation of R-Smad activity occurs at sites other than at the C-terminal serine residues of R-Smad. A potential mechanism to explain this finding is the mediation of linker-region phosphorylation, which we have previously reported to be ERK-dependent (Hayashida et al., 2003).

The present finding of differential roles for Y397 and Y925 FAK in TGFβ1 induction of collagen production versus ERK activity is intriguing. Recent research into FAK function provides increasing evidence that specific FAK phosphorylation status and adapter proteins recruited could differentially regulate downstream events (Barberis et al., 2000; Brunton et al., 2005; Gu et al., 1999; Hsia et al., 2003; Mitra et al., 2005; Rodriguez-Fernandez, 1999). ERK activation via FAK could occur through distinct pathways, including the binding of GRB2 to Y925 FAK, formation of a SHC-GRB2 complex or phosphorylation of MEK by PAK1 (Schlaepfer and Mitra, 2004). Thus, the downstream events could be determined by: (i) the distinct cellular stimulus, (ii) its timing, (iii) FAK localization and/or (iv) access to the molecules leading to subsequent pathways. Y397 FAK phosphorylation is classically associated with integrin engagement and is constitutive in adherent cell culture conditions, whereas Y925 FAK phosphorylation is mainly induced after addition of a ligand. In our previous report, we showed that a MEK/ERK inhibitor needed to be present more than 30 minutes prior to TGFβ1 stimulation in order to inhibit phosphorylation of the R-Smad linker region, even though the inhibitor blocks ERK activities almost immediately (Hayashida et al., 2003). Thus, although TGFβ phosphorylates FAK at its Y925 residue and subsequently induces ERK activity, that activity is not the source of FAK/ERK activity that is important for the initial TGFβ/Smad signal. Instead, our findings suggest that adhesion-dependent FAK phosphorylation at Y397, through ERK activity, permits phosphorylation of the Smad3 linker region to support the fibrogenic signal (Fig. 10). The role of Y925 phosphorylation is not certain.

Alternative stimuli could localize FAK differentially. Hsia et al. suggested that FAK localization to focal contacts promotes integrin-stimulated cell motility, and that FAK accumulation at lamellipodia/invadopodia promotes cell invasion (Hsia et al., 2003). It is noteworthy that a hyperphosphorylated form of FAK, at the Y925 residue in particular, appears to be excluded from focal contacts (Katz et al., 2003). Thus, differential phosphorylation could localize FAK in a specific compartment of the cell, determining downstream signaling events.

Recently, putative FAK phosphorylation sites, including novel ones with undefined function, were mapped by mass spectrometry (Grigera et al., 2005). Several growth factors have been shown to phosphorylate FAK at serine residues that are implicated in cell cycle regulation (Huang et al., 2002; Hunger-Glaser et al., 2004; Ma et al., 2001). Furthermore, protein-tyrosine phosphatases might also play a role in modulation of FA (Burridge et al., 2006). These data support a complex regulation of signals from FAK via differential phosphorylation. The list of sites at which R-Smad is phosphorylated by interacting signaling molecules, other than at the receptor-dependent C-termini, is also increasing (Xu, 2006). MAP-kinase-mediated linker-region phosphorylation at Ser203, Ser207, Ser212, Thr8 and Thr178 of Smad3 has been identified (Kamaraju and Roberts, 2005; Matsuura et al., 2005; Mori et al., 2004). Regulation of these phosphorylation sites by either TGFβ or MAP kinases is poorly understood. Therefore, although we focused on the effects of Y397 FAK phosphorylation on the Smad linker region in the present study, the definitive roles of phosphorylation at individual residues of FAK and/or Smad in the regulation of TGFβ/Smad signals and fibrogenesis remain to be determined.

Cell-matrix adhesion might differ substantially in three-dimensional structure (Cukierman et al., 2001), as might be encountered in the renal glomerulus. Therefore, the present findings still need to be explored in three dimensions, as well as at a tissue level. In the present study, we showed that Y397 FAK phosphorylation was induced in the kidney in a mouse model that develops skin and lung fibrosis in a TGFβ-dependent manner (Lakos et al., 2004; Takagawa et al., 2003; Venkatesan et al., 2004). Increased Smad3 phosphorylation at both the C-terminus and linker region support the notion that the TGFβ/Smad pathway is involved. The change was associated with type I collagen accumulation. Therefore, these data support the biological relevance of our in vitro model. To date, only a few reports have suggested involvement of FAK in glomerulonephritis. In MRL/Mp-lpr/lpr mice, a lupus model that presents with proliferative glomerulonephritis, the phosphorylated form of FAK is specifically upregulated in affected glomeruli (Morino et al., 1999). Tensin, a FA component, is localized in areas of mesangial expansion in biopsies from diabetic-or IgA-nephropathy patients and FAK is co-precipitated with tensin in cultured human mesangial cells (Takahara et al., 2004; Yamashita et al., 2004). In diseased glomeruli, increased intraglomerular pressure and/or pulsatile stretch due to high systemic pressure could cause FAK activation (Boutahar et al., 2004; Shikata et al., 2005). Indeed, a recent report suggested a crucial role for FAK in myocyte contraction and cardiac fibrosis in cardiomyocyte-targeted knock-down of FAK (DiMichele et al., 2006).

The present study suggests that specific phosphorylation of FAK plays a pivotal role in regulating TGFβ/Smad signaling for collagen expression. These results emphasize the importance of further investigating how FAs and associated molecules modulate mesangial cell function in the sclerosing process.

Active, recombinant human TGFβ1 (R&D Systems, Minneapolis, MN) was reconstituted as a 4 μg/ml stock solution in 4 mM HCl with 1 mg/ml bovine serum albumin. Antibodies were purchased from the following vendors: anti-phospho-Y397-FAK and pan-FAK antibodies from BioSource International (Camarillo, CA); anti-phospho-Y925-FAK, anti-phospho-Smad1/2/3 (H-2) and anti-phospho-HA (H-3) antibodies from Santa Cruz Biochemistry (Santa Cruz, CA); anti-phospho Smad3 (423/425) antibody from Cell Signaling Technology (Beverly, MA); anti-vinculin and anti-β-actin antibodies, anti-Flag M2 affinity gel and DAPI (4′, 6-diamidino-2-phenylindole) from Sigma (St Louis, MO); anti-His⁶ antibody from Roche Molecular Biochemical (Indianapolis, IN); anti-type I collagen antibody from Southern Biotech (Birmingham, AL); rhodamine phalloidin from Cytoskeleton (Denver, CO); and Alexa-Fluor-488 goat anti-rabbit IgG from Molecular Probes (Eugene, OR). Anti-phospho-Smad3 (207/212) antibody was kindly provided by K. Matsuzaki (Mori et al., 2004).

The wild-type, Y397F-FAK and Y925F-FAK constructs were a generous gift from D. Schlaepfer (Schlaepfer and Hunter, 1996) and the Flag-tagged Smad3 wild-type and His-tagged Smad3 EPSM construct was obtained from H. F. Lodish and X. Liu (Whitehead Institute, MA) and J. Massague (Howard Hughes Medical Institute, NY), respectively. The–378COL1A2-LUC construct containing the sequence 378 bp of the α2(I) collagen promoter and 58 bp of the transcribed sequence fused to the luciferase reporter gene was constructed as previously described (Poncelet and Schnaper, 1998). pFA-Elk, pFC-MEK and pFR-Luc plasmid were purchased from Stratagene (La Jolla, CA), and CMV-SPORT β-galactosidase from Invitrogen (Carlsbad, CA).

Human mesangial cells were isolated from glomeruli by differential sieving of minced human renal cortex obtained from anonymous surgery or autopsy specimens. Cells were cultured with DMEM/Ham's F12 medium supplemented with 16% heat-inactivated newborn calf serum (HI-NBCS), glutamine, penicillin/streptomycin, sodium pyruvate, HEPES buffer and 8 μg/ml insulin as previously described (Poncelet and Schnaper, 1998), and the cells were used at passages 5-8 for the experiments. Where noted, pLL (MW 70,000-150,000, Sigma) was applied at 1 ml/25 cm² and incubated for 10 minutes at room temperature to coat the dish surface. Thereafter, the pLL was aspirated and the dishes were rinsed briefly with sterile water, then dried for 24 hours. Human kidney tubular epithelial cells (HKCs) were a kind gift from L. Racusen (John Hopkins Medical School, Baltimore, MD) and cultured in DMEM/Ham's F12 medium with 10% HI-fetal bovine serum, glutamine, penicillin/streptomycin, amphotericin B and HEPES buffer.

Cells grown to ∼80% confluence were cultured for 20 hours with serum-free medium and then treated with 1 ng/ml TGFβ1. The cells were washed twice with ice-cold phosphate-buffered saline (PBS), lysed on ice in RIPA buffer (50 mM Tris/HCl, pH 7.5; 150 mM NaCl; 1% Nonidet P-40; 0.5% deoxycholate; 0.1% SDS) containing protease and phosphatase inhibitors (1 mM PMSF; 1 mM EDTA; 1 μg/ml of leupeptin, aprotinin, and pepstatin A; 1 mM sodium orthovanadate; 50 mM sodium fluoride; 40 mM 2-glycerophosphate) and then centrifuged at 18,000 g for 10 minutes at 4°C. Protein samples were electrophoresed through a 8% SDS-PAGE gel, then transferred onto Immobilon-P (PVDF) membranes (Millipore, Bedford, MA). After immunoblotting, immunoreactive bands were visualized by chemiluminescence reagent according to the manufacturer's protocol (Santa Cruz Biotechnology). The resulting bands were densitometrically analyzed using the ImageJ 1.33 program for Macintosh.

Three μg each of Smad3-expressing vector, along with a construct that expresses either a wild-type or Y397F FAK, was transfected with Fugene6 in serum-free medium into HKCs grown 80% confluent on 100 mm dishes. Twenty-four hours later, cells were rinsed twice with PBS and treated with 1 ng/ml TGFβ1 for 30 minutes. Expressed proteins were immunoprecipitated with 20 μl of anti-Flag M2 affinity gel (Sigma) or anti-His antibody (Roche) as directed by the manufacturer. Immune complexes were washed three times with TBS and then eluted with 2× Laemmli loading buffer at 95°C for 3 minutes in non-reducing conditions, followed by immunoblotting analysis as described above.

Mesangial cells that had been serum deprived for 24 hours were lifted by nonenzymatically (0.5 mM EDTA in PBS, Accutace, Chemicon, Temecula, CA) and replated onto either gelatin-or pLL-coated dishes in serum-free media. After 3 hours where the cells had attached to the surface, 1.0 ng/ml TGFβ1 or vehicle was added to the cells. Total cellular RNA was collected with TRIzol as directed by the manufacturer (Invitrogen, Carlsbad, CA). Four μg collected RNA was subjected to 1.2% agarose-1.1% formaldehyde gel electrophoresis and transferred to MagnaGraph nylon membranes (MSI, Westborough, MA). The blots were prehybridized at 65°C for 2 hours and then hybridized overnight at 65°C with ³²P-labeled cDNAs (10⁶ cpm/ml) made with the Rediprime DNA labeling kit (Amersham Pharmacia Biotech, Piscataway, NJ). The blots were then washed with increasing stringency and exposed to X-ray film (Eastman Kodak, Rochester, NY) at–80°C. The human α1(I) collagen cDNA was obtained from Y. Yamada (NIH, Bethesda, MA). Equal loading was confirmed by rehybridizing the blots with cDNAs for bovine 28S ribosomal RNA provided by H. Sage (University of Washington, Seattle, WA).

Cells cultured on six-well plates at 5.0×10⁴/well the day before the experiments were transfected with the indicated plasmids along with CMV-SPORT β-galactosidase (Invitrogen) as a control for transfection efficiency. For inhibition experiments, the dominant-negative construct for a specific pathway was co-transfected. 0.5 μg per well of each DNA were transfected in serum-free medium using Fugene6 (2 μl/1 μg of DNA; Roche) according to the manufacturer's instructions. For re-plating experiments, transfection was performed in 100-mm dishes for 24 hours and re-plated on the gelatin-or pLL-coated plates. After 3 hours, 1.0 ng/ml TGFβ1 or vehicle was added to cultures and the cells were harvested in reporter lysis buffer (Promega) after a 24-hour incubation. Luciferase and β-galactosidase activities were measured as previously described (Hayashida et al., 1999). Each condition was tested in triplicate and experiments were repeated at least three times for statistical analyses.

Cells were re-plated on gelatin-or pLL-coated glass cover slips in six-well dishes to achieve 70% confluence after 3 or 6 hours and treated with TGFβ1 for the indicated time period. Coverslips were prepared for immunocytochemistry by fixing/permealizing with cytoskeleton fixing buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 3 mM MgCl, 0.2% Triton X-100, 3.7% formaldehyde, pH 6.1, 37°C) for 7 minutes. Primary antibodies diluted in PBS containing 20% normal goat serum were mounted and incubated for 3 hours at 37°C, followed by 30 minutes incubation with Alexa-Flour-488-conjugated secondary antibody (1:800), along with rhodamine phalloidin (1:500) when applicable. Images were acquired with a LSM510 laser scanning confocal microscope (Zeiss, Thornwood, NY). Colocalization was evaluated using CoLocalizer Express 1.1 software [Colocalization Research Software Boise, ID, (Zinchuk et al., 2005)].

Kidney samples were obtained from mice treated with bleomycin as previously described (Takagawa et al., 2003). 6-to 8-week-old female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) were studied in parallel. Briefly, bleomycin (0.2 units/mouse in 100 μl PBS) or PBS as vehicle was administered by subcutaneous daily injections for 2 weeks (total 2.8 units) and the mice were sacrificed after 28 days from initiation of the treatment. Each group contained at least five mice. Protein was extracted from frozen kidney tissue with RIPA buffer and subjected to immunoblotting analysis. A portion of renal cortex fixed with 4% paraformaldehyde/PBS was molded in OCT compound (Leica, Nussloch, Germany) and 4-μm-thick sections were obtained by cryotome (CM1850, Leica) and analyzed with immunofluorescent staining as described above under Immunostaining. Nuclei were co-stained with DAPI for 15 minutes following the immunostaining. The protocols in this study were institutionally approved and were in accordance with the animal welfare guideline of the NIH/Association of Assessment and Accreditation of Laboratory Animal Care.

Statistical differences between experimental groups were determined by analysis of variance and values of P<0.05 by Fisher's PLSD test were considered significant. All the analyses were performed using Stat View 4.02 software program for Macintosh.

This work was supported in part by grants R01 DK49362 and R21 DK68637 (to H.W.S.), K01 DK64074 (to A.-C.P.) from the National Institute of Diabetes, Digestive and Kidney Diseases and AR42309 (to J.V.) from the National Institutes of Health. We thank Susan C. Hubchak for isolating and characterizing the human mesangial cells. We appreciate generous provision of the expression vectors for HA-FAK and mutants from D. Schlaepfer; for Flag-N-Smad3 and Flag-N-Smad3A from H. F. Lodish and X. Liu; 6×His-Smad3 EPSM from J. Massague; and the provision of anti-phospho-Smad3 (207/212) antibody from K. Matsuzaki. We also thank other members of the Schnaper laboratory for helpful discussions.