The anti-motility signaling mechanism of TGFβ3 that controls cell traffic during skin wound healing

Skin wound healing is a complex process involving collaborative efforts of multiple cell types, newly synthesized extracellular matrices and growth factors. Inflammation, re-epithelialization, granulation tissue formation and remodeling are the proposed sequential events during skin wound healing under physiological conditions (Martin, 1997; Singer and Clark, 1999). Abnormalities in any of the events could result in either delayed wound healing or a wound healed with hypertrophic scars (Tredget et al., 1997). Throughout the wound healing process, cell migration is a rate-limiting event. The epidermal keratinocytes laterally migrate across the wound bed to resurface the wound, the process known as wound re-epithelialization. The dermal fibroblasts move into the wound to produce and deposit new matrix proteins. Finally, the dermal microvascular endothelial cells that have entered the wounded tissue are responsible for building new blood vessels. It has long been recognized that epidermal and dermal cells do not enter the wound bed at the same time. Following wounding, keratinocyte migration occurs within hours, whereas migration of the dermal cells remains undetectable until several days later (Singer and Clark, 1999). These observations are in accordance with the notion that wound closure by keratinocyte migration precedes wound remodeling by the dermal cells (Clark, 1993; Falanga, 2004). However, the regulation of the sequential epidermal and dermal cell migration remained a mystery.

In intact skin, cells are bathed in interstitial fluid, largely a filtrate of plasma in blood vessels. When skin is wounded, the cells at the cut edge of the wound experience a dramatic microenvironmental transition from plasma to a new environment of serum for the first time (Henry et al., 2003). Bandyopadhyay and colleagues discovered a key factor in human serum, TGFβ3 that seems to play the “traffic control” role in migration of epidermal versus dermal cells. They showed that human serum selectively inhibits the migration of dermal fibroblasts and microvascular endothelial cells via TGFβ3 (not TGFβ1 or TGFβ2). Furthermore, they demonstrated that the differential responses of keratinocytes versus dermal fibroblasts to TGFβ3's anti-motility signal are due to the difference in the expression levels of type II TGFβ receptor (TβRII) in these cells (Bandyopadhyay et al., 2006). Therefore, this finding suggests that it is TGFβ3 in human serum that selectively halts dermal cell migration for a few days, allowing keratinocyte migration take place under the microenvironment of serum.

The three mammalian TGFβ members, TGFβ1, TGFβ2 and TGFβ3, bind to and transmit signals via the common heterodimeric complex of TβRII and TβRI serine/threonine kinases (Wrana et al., 1992; Attisano and Wrana, 1996; Piek et al., 1999; Miyazono, 2000; Derynck and Zhang, 2003; Shi and Massagué, 2003). In this classical model, TGFβ binds to TβRII, which in turn recruits, transphosphorylates and activates TβRI (Derynck and Feng, 1997; Shi and Massagué, 2003). The activated TβRII/TβRI complex then engages R-Smad-dependent and/or R-Smad-independent signaling pathways inside the cells (Miyazono, 2000; Mulder, 2000; Derynck and Zhang, 2003; Moustakas and Heldin, 2005). However, several recent studies started challenging the dogma of TβRII/TβRI signaling and suggested that TβRII and TβRI are not always required to work together to mediate TGFβ signaling. An earlier study by Zhang and colleagues provided biochemical support that TβRII under physiological expression levels are able to form homodimers in response to TGFβ stimulation (Zhang et al., 2009). Bandyopadhyay et al. showed that TβRII alone is able to mediate TGFβ signaling to ERK_1/2, in the absence of TβRI and TβRIII in primary human dermal fibroblasts (Bandyopadhyay et al., 2011). Moreover, Iwata et al. have recently reported that TβRI and TβRIII are able to mediate TGFβ signaling to TRAF6/TAK1/p38 pathway in TβRII^−/− mouse cranial neural crest cells (Iwata et al., 2012).

In the current study, we undertook a systematic approach to identify the pathway through which the anti-motility signal of TGFβ3 travels in primary human dermal fibroblasts. While our results demonstrated the necessity for the conventional TβRII/TβRI and Smad2/3/4 complexes, they also revealed a detour of the post-Smad signaling to the PKA > CREB pathway for the anti-motility signal. This is the first report on “anti-motility signaling pathway” for the TGFβ family cytokines.

TGFβ in skin can come from two sources, filtrate of blood circulation and secretion by the skin cells. It is technically difficult to establish TGFβ profiles in unwounded intact skin, since human skin isolated by biopsy is in fact “wounded” tissue. In wounded tissues, the main source for TGFβ1 comes from degranulation of the platelets, in which TGFβ1 is the only isoform present in the α-granules (Assoian et al., 1983; Roberts and Sporn, 1996). The source for TGFβ2 and TGFβ3 remains largely unknown. There have been reports that human serum, but not plasma, contains active TGFβ3 at a concentration range of 1–2 ng/ml (Hering et al., 2001), in comparison to 10–50 ng/ml of TGFβ1 in human serum (Assoian et al., 1983; Roberts and Sporn, 1996; Grainger et al., 2000). Since the half-life of TGFβ is short (2–3 minutes), the majority of TGFβ isoforms in serum must be in complexes with carrier proteins, such as latency-associated peptide (LAP) and latent TGFβ1-binding protein (LTBP) (Roberts, 1998). These TGFβ-bound carrier proteins can be cleaved in serum and new TGFβ is made available (Roberts, 1998; Khalil, 1999; Murphy-Ullrich and Poczatek, 2000). Based on the above understanding, we examined the profiles of the three TGFβ family members using frozen section of normal adult human skin. As shown in Fig. 1, anti-TGFβ1 antibody stained the dermis of the skin significantly stronger than the epidermis (panel B versus panel A). As expected, anti-TGFβ2 antibody (panel D) and anti-TGFβ3 antibody (panel F) staining of the skin sections was much weaker than anti-TGFβ1 antibody staining. These data indicate that human skin contains mainly TGFβ1 and lower levels of TGFβ2 and TGFβ3, provided that the three antibodies share similar affinity toward their antigens. To test whether or not the weaker staining by anti-TGFβ3 antibody was due to its lower affinity toward TGFβ3, we incubated the skin sections with human recombinant TGFβ3 prior to anti-TGFβ3 antibody staining. We detected a strong and almost exclusively staining of the dermis (panel H). These results indicated that 1) the weaker staining of the skin sections by anti-TGFβ3 antibody (panel F) was not due to a compromise in the antibody's affinity for TGFβ3 and 2) dermal cells are the primary targets for environmental TGFβ3, consistent with our previous finding that human dermal cells express 7–15 fold higher levels of TβRII than human keratinocytes (Bandyopadhyay et al., 2006).

Migration of human keratinocytes (HKCs) and human dermal fibroblasts (HDFs) into the wound are two critical events during wound healing. However, a major difference between these two cell types lies in distinct responses of these cells to the anti-motility signal of TGFβ3, (Bandyopadhyay et al., 2006). As shown in Fig. 2, in the absence of TGFβ3, HDF stimulus, PDGF-bb (A), and HKC stimulus, TGFα (B), strongly stimulated migration of HDFs (A, panel b versus panel a) and HKCs (B, panel b′ versus panel a′) in colloidal gold migration assay (Materials and Methods). Presence of TGFβ3, however, selectively inhibited PDGF-bb-stimulated HDF migration (A, panel c versus panel b), but not TGFα-stimulated HKC migration (B, panel c′ versus panel b′). Migration Index (MI) was used to quantitate cell migration. Under similar conditions, both PDGF-bb (C) and TGFα (D) also stimulated DNA synthesis in HDFs and HKCs (bars 2 versus bars 1). Interestingly, in contrast to the selective inhibition of HDF migration by TGFβ3, TGFβ3 inhibited growth factor-induced DNA synthesis in both HDFs and HKCs in a dose-dependent manner (bars 3–7 versus bars 2). These data suggest that the TGFβ receptors in HDFs allows both anti-proliferation and anti-motility signals of TGFβ3 to enter the cells, whereas the TGFβ receptors in HKCs only transmit the anti-proliferation, but not anti-motility, signal of TGFβ3. Therefore, comparative studies between HDFs and HKCs would provide a unique model system for defining the signaling pathway by which TGFβ3 blocks growth factor-stimulated cell migration – the anti-motility signaling mechanism by TGFβ3.

We have previously shown that it is the higher expression level of TβRII in HDFs (than in HKCs) that makes HDFs sensitive to the anti-motility signal of TGFβ3 (Bandyopadhyay et al., 2006). Therefore, we chose TβRII as the starting signaling molecule. We asked whether TβRII transmits the anti-motility signal of TGFβ3 to TβRI or independent of TβRI. We used the lentiviral shRNA-delivery system, FG12 (Qin et al., 2003), to downregulate the endogenous TβRI in HDFs. The FG12 system, as shown in Fig. 3A, offers more than 90% transduction efficiency in HDFs as indicated by a green fluorescent protein (GFP) gene expression by the vector following a single infection. Downregulation of TβRI by one of the shRNAs (shTβRI-NO. 2) is shown in Fig. 3B (panel a, lane 2 versus lane 1). In these cells, we observed that TGFβ3-stimulated phosphorylation of Smad3 became undetectable, while the levels of TβRII remained unaffected (Bandyopadhyay et al., 2011). When these cells were subjected to the colloidal gold migration assay, as shown in Fig. 3C, TGFβ3 was no longer able to block PDGF-bb-stimulated migration of the TβRI-downregulated cells (bar 6) in comparison to the parental HDFs (bar 3). These results indicated that the endogenous TβRI is required for transmitting the anti-motility signal of TGFβ3.

We then tested whether TβRII transmits the anti-motility signal of TGFβ3 in full to TβRI, i.e. whether activation of TβRI alone is sufficient to replace TGFβ3. We took an advantage of the constitutively activated mutant of TβRI, the TβRI-TD mutant (Wieser et al., 1995). We reasoned that if expression of TβRI-TD alone were able to block PDGF-bb-stimulated migration even in the absence of TGFβ3, it indicates that TβRII does not need any TβRI-independent pathways for transmitting the anti-motility signal of TGFβ3. As shown in Fig. 3D, TβRI-wt (lane 2) and the TβRI-TD mutant (lane 3) genes were expressed several fold over the endogenous TβRI in HDFs by the lentiviral vector, pRRLSin (lane 1). The constitutive kinase activity of TβRI-TD was indicated by constitutive Smad3 phosphorylation even in the absence of TGFβ3 under serum-free conditions (Fig. 3E, panel c, lane 1 versus lanes 2, 3, 4). In TβRI-wt-expressing HDFs, however, phosphorylation of Smad3 still depended upon TGFβ3-stimulation (panel a, lanes 2, 3, 4 versus lane 1). Interestingly, the constitutively activated TβRI-Smad pathway was able to replace TGFβ3's anti-motility signal. As shown in Fig. 3F, TβRI-TD-expressing HDFs failed to migrate in response to PDGF-bb stimulation even in the absence of TGFβ3 (bar 5 versus bar 2). In contrast, the HDFs expressing TβRI-wt (inactive in the absence of TGFβ3) behaved just like the parental HDFs, with a basal migration under serum-free conditions (bars 1), enhanced migration in response to PDGF-bb (bar 2) and inhibited migration in the presence of TGFβ3 (bar 3).

Furthermore, we reasoned that if TβRI-TD was capable of carrying the full scale of the anti-motility signal of TGFβ3, expression of TβRI-TD should be able to block migration of HKCs. As shown in supplementary material Fig. S1, overexpression of TβRI-TD was sufficient to block TGFα-stimulated HKC migration even in the absence of TGFβ3 (bars 5 versus bar 2). In comparison, TGFα-stimulated migration of HKCs expressing the TβRI-wt remained unaffected by TGFβ3 (bar 3), just like the parental HKCs (Fig. 1). Taken together, we concluded that TβRI activation by a higher level of TβRII in HDFs is both necessary and sufficient for transmitting anti-motility signal of TGFβ3.

The post-TβRII and TβRI signaling could use R-Smad-dependent or R-Smad-independent routes (Derynck and Zhang, 2003). We first tested the R-Smad pathway. It is known that Smad2/3 are phosphorylated by activated TβRI and in turn form a complex with Smad4. This R-Smad trimeric complex translocates to the nucleus to regulate gene expression (Shi and Massagué, 2003). To individually assess the three Smads, we designed two independent shRNAs against each Smad and introduced each of these shRNAs into HDFs by the FG-12 lentiviral shRNA system, as previously described. As shown in Fig. 4A, we achieved nearly complete downregulation of Smad 2 (panel a, lanes 1 and 2 versus lane 3), Smad 3 (panel c, lanes 1 and 2 versus lane 3), and partial downregulation of Smad4 (panel e, lanes 1 and 2 versus lane 3). The incomplete downregulation of Smad4 may reflect the fact that a fraction of Smad4 locates in the nucleus. These cells were then subjected to the colloidal gold migration assay. As shown in Fig. 3B, TGFβ3 blocked PDGF-bb-stimulated migration of HDFs infected with vector alone, as expected (bar 3 versus bar 2). However, downregulation of Smad2 prevented TGFβ3 from inhibiting PDGF-bb-stimulated HDF migration (bar 6 versus bar 5). Similarly, downregulation of Smad3 also nullified TGFβ3's anti-motility signaling (bar 9 versus bar 8). More intriguingly, even under partial downregulation of Smad4, TGFβ3 was unable to inhibit PDGF-bb-stimulated migration of the cells (bar 12 versus bar 11). Taken together, these data indicate that Smad2/3 in complex with Smad4 are required for transmitting the anti-motility signal from TGFβ3.

Where next does the R-Smad complex connect the anti-motility signal to? The possibilities include i) going into the nucleus to regulate expression of migration-suppressing genes or ii) connecting with other cytosolic signaling pathway(s). We pre-set two parameters to test these possibilities. First, we focused on primary targets of R-Smads, i.e. targets that are rapidly activated by TGFβ3 in a R-Smad-dependent and protein synthesis-independent manner. Therefore, we chose 10 minutes (maximum phosphorylation of Smad2/3) as the stimulation time for TGFβ3. Second, the to-be-identified new pathway should only be activated by TGFβ3 in HDFs, but not in HKCs. Based on these parameters, we treated serum-starved HDFs and HKCs side-by-side with the same amount of TGFβ3 for 10 minutes. Cell extracts were subjected to pathway screening with the Human Phospho-Kinase Antibody Array (ARY 003, R&D), which represents 46 well-characterized protein kinase pathways (supplementary material Fig. S2A). As shown in Fig. 5A, we found that TGFβ3 induced increase in ERK1/2 phosphorylation (open dotted squares), CREB phosphorylation (a direct substrate of protein kinase A, PKA) (open dotted circles) and Src kinase phosphorylation (open dotted triangles) in HDFs (panel b versus panel a). None of these three pathways appeared to be significantly activated by TGFβ3 in HKCs (panel d versus panel c). Quantitation by phosphoimager (Materials and Methods) of these three target proteins is shown in Fig. 5B (HDFs) and Fig. 5C (HKCs). In comparisons, we detected additional three pathways, β-catenin, p53 and eNOS, which were activated in both HDFs and HKCs in response to TGFβ3 stimulation (dotted underlines, 1, 2 and 3). We decided to focus on PKA pathway for its possible role in mediating the anti-motility signaling of TGFβ3 for the following reasons. PKA has previously been shown as an antagonist of growth factor signaling (Cook and McCormick, 1993; Tokiwa et al., 1994). In contrast, both ERK1/2 and Src are known downstream pathways of the PDGF-bb signaling and required for PDGF-bb-stimulated cell proliferation and migration (,Li et al., 2004a; Heldin and Westermark, 1996). It is, therefore, less likely that these two pathways are simultaneously involved in the anti-motility signaling of TGFβ3.

We directly confirmed that TGFβ3 activates PKA pathway in HDFs, but not HKCs. Lysates of TGFβ3-treated HDFs and HKCs were subjected to Western blot analysis with anti-phospho-CREB antibodies. As shown in Fig. 5D, TGFβ3 stimulation clearly caused an increased CREB phosphorylation in HDFs (panel a, lane 2 versus lane 1), and this induction occurred in a time dependent manner (supplementary material Fig. S2B). In contrast, a much weaker, if any, TGFβ3-stimulated CREB phosphorylation was detected in HKCs (lanes 3 and 4). This weaker induction was reproducible in time-course experiments (supplementary material Fig. S2B). More importantly, we found that the TGFβ3-stimulated CREB phosphorylation in HDFs depended upon the Smad pathway, since Smad4 downregulation prevented the TGFβ3-stimulated CREB phosphorylation (Fig. 5E, panel a, lane 4 versus lane 2). Phosphoimager scanning showed TGFβ3-induced fold increases in CREB phosphorylation in HDFs versus HKCs. Therefore, PKA appeared to be a specific downstream effector of the Smads in HDFs in response to TGFβ3.

We undertook two complementary approaches to examine whether PKA acts downstream R-Smads to mediate TGFβ3's anti-motility signal. First, we tested the effect of a chemical inhibitor, H-89, and a chemical activator, forskolin, of PKA pathway. As shown in Fig. 6A, PDGF-bb greatly enhanced HDF migration under serum-free conditions in the absence of TGFβ3 (panel b versus panel a). The presence of TGFβ3 completely nullified the stimulatory effect of PDGF-bb (panel c). However, in the presence of H89, TGFβ3 was no longer able to block PDGF-bb-stimulated HDF migration (panel d versus panel c). Consistently, H89 inhibited TGFβ3-induced phosphorylation of CREB (Fig. 6B, panel a, lane 3 versus lane 2). In comparison, H89 did not affect TGFβ3-stimulated phosphorylation of Smad3 in the same cells (panel b, lane 3 versus lane 2), consistent with the notion that R-Smads are upstream of PKA in response to TGFβ3. Similarly, the PKA activator, forskolin, recapitulated the effect of TGFβ3. As shown in Fig. 6C, the presence of forskolin caused increased phosphorylation of CREB under serum-free and TGFβ3-free conditions (insert) and treatment with forskolin alone was able to block PDGF-bb-stimulated HDF migration in a dose-dependent manner (bars 3 to 7 versus bar 2).

Second, since few chemical inhibitors/activators could claim absolute specificity, we next directly targeted the catalytic subunit of human PKA, the PKA-Cα, with shRNA technology. We designed two shRNAs against the human PKA-Cα and individually delivered them by the FG-12 lentiviral system into HDFs. It is shown in Fig. 7A that PKA-Cα was clearly detected in FG-12 vector-infected HDFs (panel a, lane 1). However, PKA-Cα expression was dramatically downregulated in cells infected with shRNA-NO. 1 or shRNA-NO. 2 against PKA-Cα (panel a, lanes 2 and 3). The effect of PKA-Cα downregulation was also confirmed by blockade of forskolin-stimulated CREB phosphorylation in these cells (Fig. 7B, panel a, lane 4 versus lane 2). In contrast, in the same cells, TGFβ3 was still able to induce Smad2/3 phosphorylation (Fig. 7C). We then tested whether TGFβ3 was still able to block PDGF-bb-stimulated migration of the cells. As shown in Fig. 7D, in control HDFs, PDGF-bb stimulation caused dramatically increased migration (panel b versus panel a), but failed to do so in the presence of TGFβ3 (panel c). However, downregulation of the PKA-Cα by either shRNA-NO. 1 or shRNA-NO. 2 completely blocked anti-motility signaling of TGFβ3 (panels f and i versus panels e and h, respectively). Quantitation of the migration, Migration Index, was shown in Fig. 7E. Taken together, results of this study demonstrated that the “TβII > TβI > R-Smads > PKA” pathway mediates TGFβ3's anti-motility signaling in HDFs, which represents a key cells traffic regulator during wound healing.

The tumor suppressing effect of the TGFβ family cytokines includes anti-proliferation and anti-motility. These two functions of TGFβ also take an important part in the process of wound healing. A cell could only choose migration or proliferation, but not both, at a time. Immediately following the injury, the anti-proliferation effect of TGFβ3 from serum halts the proliferation mode of both epidermal and dermal cells and migration mode of only dermal cells at the wound edge, in order for epidermal keratinocytes to migrate toward the wound first (Sarret et al., 1992; Bandyopadhyay et al., 2006). Following keratinocyte migration to re-epithelialize the wound, the dermal fibroblasts and dermal microvascular endothelial cells start to migrate into the wound bed. These findings provide a molecular mechanism for how keratinocyte migration occurs within hours, whereas the dermal cell migration is undetectable until 3–4 days later following the injury (reviewed by Singer and Clark, 1999). While the anti-proliferation effect of TGFβ has been studied in various cell culture models, the anti-motility signaling of TGFβ has largely been overlooked over the years. In the current study, we took an advantage of the differential responses of epidermal keratinocytes and dermal fibroblasts to the anti-motility signal of TGFβ3 to identify the pathway that mediates the anti-motility signal of TGFβ3. As schematically shown in Fig. 8, both anti-proliferation and anti-motility signals of TGFβ3 enter HDFs via the higher expression level of TβRII, whereas only the anti-proliferation, but not anti-motility, signal of TGFβ3 was able to enter keratinocytes due to a lower expression of TβRII. In HDFs, TGFβs uses the “TβII > TβI > R-Smads > PKA” pathway to execute its anti-motility signal.

Our finding of the R-Smads > PKA pathway to execute anti-motility signal of TGFβ3 is in line with several previous reports that the canonical R-Smad pathway can cross talk with the PKA pathway in cells. Zhang et al. showed that TGFβ activates PKA via interaction between Smad2/3 and the regulatory subunit (R) of PKA, resulting in release of the catalytic subunit from the PKA holoenzyme in Mv1Lu cells (Zhang et al., 2004). Chowdhury and colleagues reported that TGFβ/PKA pathway is part of TGFβ's tumor suppressing function that decreases the survival and inhibits metastasis of colon cancer cells (Chowdhury et al., 2011). Moreover, consistent with our current finding, Giannouli and Kletsas reported that TGFβ activates PKA in adult dermal fibroblasts (Giannouli and Kletsas, 2006). Two groups also reported that PKA and CREB mediate TGFβ-stimulated gene expression in the human keratinocyte cell line, HaCat, and other cell lines of epithelial origin (Schiller et al., 2003; Hayashida et al., 2006). However, we detected weak CREB phosphorylation in response to TGFβ3 stimulation and TGFβ3 did not block migration of HKCs under these conditions. The discrepancy may be due to differences between the immortalized cell lines used by others and primary HKCs used in our study. It should be pointed out that, while the migration of HKCs is unaffected by TGFβ3, HKCs, like HDFs, are equally sensitive to TGFβ3's anti-proliferation signal (Fig. 1B). Therefore, the anti-motility signaling and anti-proliferation signaling of TGFβ3 must take distinct signaling routes in the same cells. Our observation that the constitutively active TβRI-TD mutant was sufficient to convert HKCs to HDF-like cells toward the anti-motility signal of TGFβ3 suggests that the “degrees” of TβRI activation (by the upstream TβRII) are different between HKCs and HDFs, which would determine transmission of either anti-proliferation signal alone or both anti-proliferation or anti-motility signals of TGFβ3 into the cells.

Finally, how TGFβ's anti-motility signal overrides PDGF-bb's pro-motility signal remains to be studied. For instance, where exactly the two opposite signals meet with each other, where they “fight” and TGFβ's anti-motility effect wins over PDGF-bb's pro-motility effect, is the next important question for investigation. Results of our recent experiments pointed to the important focal adhesion protein, paxillin, as a center of this battlefield. It is known that extracellular signals regulate paxillin function by two ways: its phosphorylation status and its levels of expression. Changes in either of the two parameters could greatly affect cell morphology and motility (Turner, 2000). While we observed that neither PDGF-bb nor TGFβ3 significantly affect the protein levels of paxillin in HDFs, we found that PDGF-bb and TGFβ3 regulate the serine and tyrosine phosphorylation of paxillin in opposite ways (A.H. and W.L., unpublished data). To functionally define the effects of the distinct phosphorylation levels and sites awaits further and technically challenging studies. In these studies, one would need to first deplete the endogenous wild type paxillin (permanently) and then to introduce various paxillin mutants on phosphorylation sites into these cells. We are currently testing if these opposite effects in paxillin phosphorylation are the reason by which TGFβ3 executes its anti-motility effect over PDGF-bb's stimulatory effect.

Primary human neonatal HKCs and HDFs were purchased from Clonetics (San Diego, CA). HKCs were cultured in EpiLife medium with added human keratinocyte growth supplements (HKGS). HDFs were cultured in DMEM supplemented with 10% fetal bovine serum. The third or fourth passages of cells were used for experiments. rhPDGF-bb, rhTGFα, rhTGFβ3, anti-phospho-CREB (Ser-133) antibody and anti-Smad4 antibody were purchased from R&D System (Minneapolis, MN). The sources of human TβR (I and II) and other Smad reagents, including TβRI-TD, TβRII-KD cDNAs, antibodies against TβRI/Alk5 and TβRII, anti-phospho-Smad2 (ser465/467) antibody and anti-Smad2/3 antibody were as previously described (Bandyopadhyay et al., 2006; Bandyopadhyay et al., 2011). Anti-phospho-Smad2/3 or anti-phosphoSmad3 antibodies were from Abcam (Cambridge, MA). PD0325901 (PD901) was from Calbiochem (San Diego, CA). Human Phospho-Kinase Antibody Array (ARY 003) was purchased from R&D System. Anti-PKA Cα antibody, H-89 (dihydrochloride) and forskolin were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and Sigma–Aldrich (St. Louis, MO), respectively. Rat type I collagen was purchased from BD Biosciences (San Jose, CA). Anti-β actin antibody and anti-GAPDH antibody were from Cell Signaling Technology, Inc. (Beverly, Massachusetts). XL-10 Gold Ultra competent cells (XL-10 Gold) (for lentiviral vectors) were from Stratagene (La Jolla, CA).

For immunostaining with anti-TGFβ antibodies, we used a diamond cutter to make a highly visible circle around the tissue on the back of the slides with tissue specimen, washed and air-dried the slides. The tissue specimen on slides was fixed with 25 µl acetone for 5 min, washed with PBS, and blocked in 25 µl of the blocking reagent (1:10 dilution of normal goat serum, 0.05% Tween 20, 0.05 Triton X-100, and 1% BSA in PBS) for 60 min in a H₂O humidified chamber. The slides were washed with PBS and incubated with a primary antibody for two hours. The slides were washed three times (7 min each) in PBS, incubated with FITC-conjugated secondary antibody (added directly onto the tissue area) and covered to prevent exposing to light (for the rest of the procedures). Slides were subjected to final wash twice in PBS, air-dried and mounted. The results were analyzed under a microscope (100× at −20°C; TE-2000U Eclipse; Nikon) with application of green (FITC) fluorochrome in the absence of any imaging medium. The images were recorded in JPEG by an attached camera (TE-2000U; Nikon) using MetaMorph software (version-6.2rb; Universal Imaging). The detailed procedures are as previously described (Woodley et al., 2007). This experiment was repeated with tissue specimen from multiple (>6) independent donors.

The colloidal gold migration assay was reported previously (Albrecht-Buehler, 1977) and modified and described in details by us (Li et al., 2004b). Data from multiple independent experiments on HKCs or HDFs were calculated as Migration Index (MI), i.e. the percentage (%) of cell-migrated area over total areas under microscope, in response to corresponding stimulus/inhibitor over the baseline control (mean ± s.d.), P<0.01–0.05. The [H³]-thymidine incorporation assay to measure de novo DNA synthesis followed our previously published protocol (Bandyopadhyay et al., 2006).

The lentivirus-derived FG-12 system for shRNA delivery, shRNA cloning and viral infection were carried out as previously described (Qin et al., 2003) and then modified (Guan et al., 2007). The sense oligonucleotide shRNA sequence for TβRI, Smad2, Smad3, Smad4, and the PKA catalytic subunit (Cα) are as follows: Smad2, 5′-GGACTGAGTACACCAAATA-3′ and 5′-GCAGAACTATCTCCTACTA-3′, Smad3, 5′-GCCTGGTCAAGAAACTCAA-3′ and 5′-GGTGCTCCATCTCCTACTA-3′, Smad4, 5′-GGTGGAGAGAGTGAAACAT-3′ and 5′-GAATCCATATCACTACGAA-3′. PKA-Cα, 5′-GGGTCTCCATCAATGAGAA-3′ and 5′-CCATCCAGATCTATGAGAA-3′. The oligonucleotide shRNA sequence (5′-GACCACAGACAAAGTTATA-3′) for TβRI was previously reported (Bandyopadhyay et al., 2011). The usage of the lentivirus-derived vector, pRRLsinh-CMV, for overexpressing genes of interest was as described in details (Li et al., 2004a). Down- or up- regulation of genes was verified by Western blotting with corresponding antibodies (Bandyopadhyay et al., 2006).

Passage 3 to 4 subconfluently-grown primary HDFs and HKCs in 15 cm tissue culture plates were serum-starved (with 0.2% FBS) overnight. The cells were treated with TGFβ3 for 10 min under serum-free conditions. The cells were lysed (in commercially provided lysis buffer) and subjected to the Proteome Profiler Human Phospho-Kinase Array Kit (R&D Systems, Minneapolis, MN) strictly according to the manufacturer's instructions, including preparation of cell lysates (750 µl lysis buffer, scraped and centrifuged in a 1.5 ml Eppendorf tube at 13,000 g for 3 min at 4°C), time and temperature for incubation with array membrane, washing, secondary antibodies and ECL. Results were verified by Western blotting experiments. Data were quantitated using an imager, ChemiDoc^TM MP with Imagelab Version 4.0.1 software. For the phosphoarray's phosphoimager data, fold  =  +TGFβ/−TGFβ. For the Western blot data, fold  =  −TGFβ/its own GAPDH ÷ +TGFβ/its own GAPDH.

Data are presented as mean ± standard deviation (s.d.). Statistical significance for comparisons was determined by the Student's two-tailed t-test. A p value equal or less than 0.05 was considered statistically significant (Woodley et al., 2007).

This study was supported by NIH grants GM066193 and GM067100 (to W.L.), AR46538 (to D.T.W.) and AR33625 (M.C. and D.T.W.).