All cell lines used were maintained in high glucose Dulbecco's Modified Eagle Medium (DMEM; Invitrogen, Carlsbad, CA) and purchased from American Type Culture Collection repository (Mannassas, VA; NIH-3T3, CRL-1658; Mv1Lu, CCL-64; HEK-293A, CRL-1573; NMuMG, CRL-1636). The murine embryonic fibroblast cell line, AKR-2B, was grown in DMEM supplemented with 5% Fetal Bovine Serum (FBS; PAA Labs Inc, Etobicoke, ON)), while NIH-3T3 cells were grown in DMEM supplemented with 10% Newborn Calf Serum (NBCS; Invitrogen, Carlsbad, CA). Pak2 flox/flox MEF parental cell line and the Cre/Pak2 knockout derivative (kind gift of Dr. Jonathan Chernoff, Fox Chase Cancer Centre, OH) were maintained in DMEM supplemented with 10% FCS, as were Mv1Lu epithelial cells, while NMuMG growth media also contained 10 µg/ml bovine Insulin (Sigma Biochemicial, St. Louis, MO) and 5 ng/ml EGF (Cell Signaling Technologies; Pickerington, ON). All buffer salts, bovine serum albumin (BSA) and acrylamide were purchased from ThermoFisher Biotechnology.

Mesenchymal cell lines were plated 24 h prior to serum depletion (0.1% NBCS/DMEM) 18 h prior to experimentation. Epithelial cell lines were treated approximately 18 hours after plating (Mv1Lu), or medium was replaced with 10% FCS/DMEM, 18 hours prior to treatment to remove interference from supplemental growth factors (NMuMG). Cells were stimulated with 2 ng/ml TGF-β1 (US Biological, Swampscott, MA) for indicated time periods. Inhibitors to PI3K (LY294002; Upstate Biologicals; Billerica, MA), MEK1/2 (U0126; Cell Signal Technologies; Pickerington, ON), proteosomal inhibitor (MG132; Tocris Bioscience; Ellisville, MO) were used at 10 µM dissolved in DMSO, the TGF-β receptor inhibitor (LY364947, Tocris Bioscience; Ellisville, MO) was used at 500 nM and also dissolved in DMSO, while the Ras inhibitor (FPT II; EMD Biosciences, Gibbstown, NJ) was used at 20 µM dissolved in water.

Total cellular protein was obtained by lysing cells with RIPA lysis buffer [29] and quantified for total protein by BCA assay with a standard curve generated using a BSA standard (Pierce/ThermoFisher Scientific; Rockford, IL). Aliquots of equivalent total protein were separated by polyacrylamide gel electrophoresis and transferred to PVDF (Millipore; Billerica, MA) or Nitrocellulose (Pall Life Sciences; Pensacola, FL) membranes prior to antibody detection of each specific protein of interest. Primary antibody binding was detected using a goat anti-rabbit IgG-Horseradish peroxidase secondary antibody (Santa Cruz Biotechnology: Santa Cruz, CA), visualized with Supersignal West Pico Chemiluminscent Substrate (Pierce/ThermoScientific; Rockford, IL) by exposing Hyperfilm (GE Healthcare Bio-Sciences; Piscataway, NJ). All primary antibodies used were from Cell Signal Technologies and included p44/22 MAP Kinase, Phospho-p44/22 (Thr202/Tyr204), Phospho-Akt (Ser473), and Akt, Phospho-smad2 (Ser245/250/255), Phospho-smad2 (Ser465/467), smad2, Phospho-c-Raf (Ser338), and Pak2. Anti-phospho-smad3 (Ser423/425) antiserum was the kind gift of Dr. Ed Leof (Mayo Clinic, Rochester, Minnesota; [10]).

The thymidine incorporation assay was a modification of previously described methods [30]. Briefly, NIH 3T3 cells were plated at 40 000 cells/well in a 24 well plate and allowed to attach for 24 h. Cells were then serum depleted for 24 h prior to treatment (TGF-β, 5 ng/ml), supplemented with or without U0126 for 18 hours. Tritiated Thymidine (1 µCi; PerkinElmer; Shelton, CT) was added to each well and incubated for 2 h at 37°C. Macromolecular incorporated radioactivity was precipitated with ice cold 10% trichloroacetic acid, solubilized (0.2 N NaOH, 200 µg/ml ssDNA) then quantified using a Beckman Coulter LS6500 Liquid Scintillation Counter.

Dominant negative Pak2-EGFP (Ad-dnEGFP-Pak2) fusion protein and control EGFP (Ad-EGFP) expressing adenoviruses were generously provided by Dr. Ed Leof (Mayo Clinic, Rochester, Minnesota) and Dr. Mark Natchigal (Dalhousie University, Halifax, NS), respectively. Adenovirus constructs were amplified by infecting HEK-293A cells. Fibroblast cells (AKR-2B or NIH 3T3) were plated 8 h prior to infection with the indicated virus at a multiplicity of infection (MOI) of 125∶1 for 24 h, then either serum depleted for pathway analysis (AKR-2B), or replated for thymidine incorporation assays (NIH 3T3).

Smad signaling duration was determined using a variation of a pulse-chase treatment of AKR-2B fibroblasts. Cells were plated and serum depleted as previously described. Cells were pretreated with U0126 (10 µM) 30 minutes prior to the addition of TGF-β (2 ng/ml) for 10 minutes. Medium was replaced with prewarmed medium (37°C, 0.1%NBCS/DMEM) with or without supplimentation of U0126. The indicated treatment times are relative to the addition of TGF-β to the medium. Protein lysates were prepared as previously described while RNA was isolated from separate experiments using Trizol (Invitrogen; Carlsbad, CA) following the manufacturer's protocol. Total RNA was evaluated and quantified using a 2100 Bioanalyzer (Agilent Technologies; Waldbronn, Germany). Samples with a RIN value greater than 8.5 were used. Target gene mRNA levels were evaluated using TaqMan assays on-demand (Applied Biosystems; Foster City, CA) with 18S rRNA as the internal standard, Mm00435860_m1 for plasminogen activator-1 (PAI-1), or Mm00484742_m1 for Smad 7. Relative expression levels were determined using the ΔΔCt method [31] where target gene expression was evaluated relative to 18S rRNA levels. Untreated control levels were arbitrarily set at 1 to which the treatment groups were compared. Differences between mean levels of treatment group's target gene expression were analyzed with Graphpad Prism software by 1-way ANOVA using a Tukey's Multiple Range test for Post-hoc analysis. Differences in band intensities of western blots were determined using Image J software to define the amount of phospho-proteins, relative to their respective loading control. Decay rates for C-terminal phospho-smad2 and 3 were determined using Graphpad Prism software to calculate exponential decay curves for each experiment (smad 2, n = 3; smad 3, n = 4). Exposure differences between experiments were equalized by setting the 60 min. value at 100% and calculating the intensities relative to this value.

NIH 3T3 cells were plated on 4 chamber slides (Lab-Tek; ThermoFisher Scientific; Rockford, IL), treated as previously described. Cells were fixed with 4% Paraformaldehyde/PBS, permeabilized with 0.2% Triton X-100, blocked [PBS, 5% BSA, 10% normal goat serum (Sigma/Aldrich Chemical Company; St. Louis, MO)], prior to incubation in primary antibody overnight at 4°C in a humidified chamber. Immune complexes were detected with Rhodamine X conjugated goat anti-rabbit secondary antibody (BioCan Scientific; Mississauga, ON), and coverslipped with Vectashield (Vector Laboratories; Burlingame, CA). Photomicrographs were obtained using a Leica DMII microscope with a TX2 filter cube and Open Lab software. Digital images were all adjusted for size, contrast and brightness equally using Photoshop software to preserve the original relative magnification and intensities.

Since TGF-β has been known to stimulate fibroblast replication and had been shown to activate PI3K [10], [12] we wished to further define activation of the Erk pathway in normal (untransformed) cell lines. Temporal changes in Erk phosphorylation as the result of fibroblast cell lines treated with TGF-β over the course of 4 h were determined (Figure 1). Taking into account protein loading (total Erk) between lanes, we found Erk phosphorylation increased significantly above background between 60–90 minutes after TGF-β addition (Figure 1B). Previous reports have shown Erk activation at earlier time points [7], [13]. We likewise saw an increase in phospho-Erk earlier, but only when cells were allowed to cool when taken from the incubator during addition of TGF-β. Cells maintained near 37°C, demonstrated no significant activation of Erk prior to the 60–90 minute window. Earlier activation was consistent with Erk signaling being indicative of an induced stress response [32], [33]. Since TGF-β activation of PI3K has been shown to be cell type dependent [10], [12] the previous experiment was repeated using Mv1Lu and NMuMG epithelial cells. No increase in Erk phosphorylation was identified at any time point (Figure 1C) following TGF-β treatment. Together these results confirm that activation of Erk upon TGF-β treatment occurs in phenotypically normal cells of mesenchymal origin, but not epithelial cells.

Having shown Erk activation, via TGF-β, as a cell type specific response, we next wanted to confirm the mechanism through which the TGF-β signal was propagated. Previous studies have shown that TGF-β induced activation of Pak2 and/or Ras [10], [12], [13], we sought to identify the pathway resulting in Erk activation. Using LY294002, a specific inhibitor of PI3K, Erk phosphorylation was inhibited (Figure 2). Additionally, the MEK1/2 inhibitor U0126, blocked Erk phosphorylation, demonstrating that not only is PI3K necessary but also the MAPKK, MEK is involved in TGF-β activation of Erk. There was also a temporal increase in c-Raf phosphorylation (Figure 2B) dependent on TGF-β stimulation and subsequent to PI3K activation as shown by the phosphorylation of S338, a known Pak activation site [34], [35] and its inhibition by the PI3K inhibitor LY294002. In the canonical Erk pathway, activated Ras initiates a cascade resulting in Erk activation. Using small molecular weight inhibitors, we evaluated the activation of two different PI3K activated pathways, Erk and Akt (Figure 2C). Although both pathways were activated by TGF-β through PI3K, only Erk phosphorylation was sensitive to U0126 and neither pathway was inhibited by the Ras farnysylation inhibitor, FPT II [36]. TGF-β induced Erk phosphorylation was significantly increased, above control levels (untreated and FTPII), with no significant difference found between the two TGF-β treated groups (TGF-β alone and TGF- β+FTPII).

Since inhibition of Ras farnysylation did not appear to effect TGF-β induced Erk activation, and previous data had shown Pak to be downstream of PI3K and phosphorylate Raf [34], [35], we wanted to determine where Pak2 acted within this signaling pathway. Initially we assessed S338 phosphorylation of c-Raf in fibroblasts expressing a dn-Pak2. When fibroblasts were treated with TGF-β, both c-Raf and Erk phosphorylation was dramatically reduced (Figure 3A). Additionally, a dramatic reduction in phosphorylation of Erk (Figure 3B) was seen when Pak2 was knocked out in MEFs. The low levels of Erk activation in both dnPak2 expressing fibroblasts and Pak2 KO-MEFs, suggests Erk activation can occur via Ras, albeit at a greatly reduced amount. This data is consistent with the hypothesis that TGF-β induced phosphorylation of Erk primarily follows the pathway of PI3K/Pak2/c-Raf/MEK/Erk, with a secondary contribution of Ras, similarly to that described for PDGF [21].

Having shown a direct TGF-β/Erk signaling pathway, our next goal was to show an association existed between Erk and smad signaling, both under the direct control of TGF-β. AKR-2B fibroblasts were treated with TGF-β and probed for phosphorylation of smad2 at the Erk phosphorylation sites (Figure 4A; Ser 245, 250, and 255). We observed a temporal increase in smad2 linker region phosphorylation that was dependant on MEK activitivation of Erk, as treatment with U0216 abolished linker region phosphorylation (Figure 4A,C). Similarly, inhibition of TGF-β receptor kinase activity (LY364947) blocked smad2 linker phosphorylation (Figure 4C). Smad2 phosphorylation showed prominent C-terminal phosphorylation (Ser465/467) at all times tested, independent of MEK activity (Figure 4B; TGF-β+U0126). The linker phosphorylation seen at the 30 min. time point (Figure 4A) was not significantly above background levels (0 min.). However, rapid activation of Erk by EGF did result in significant smad2 linker phosphorylation (Figure 4D). Together this data indicates TGF-β stimulation causes phosphorylation of smad 2 via two distinct mechanisms, within the linker region, through Erk, and at the C-terminus, by TGF-β receptors.

Although these results indicate a direct cross-talk between smad2 and Erk, in light of the controversy between the Erk and smad linker region phosphorylation, we wished to examine their spatial relationship. Since smad2 is a transcription factor that translocates quickly into the nucleus following receptor mediated phosphorylation, we fractionated TGF-β treated fibroblasts into cytoplasmic and nuclear fractions and examined smad2 phosphorylation. Linker phosphorylation was seen primarily in the nuclear fraction (Figure 5A). Similarly, receptor phosphorylated smad2 was primarily in the nucleus, with small amounts in the cytoplasmic fraction, following TGF-β treatment. Total smad2 was present in both the cytoplasm and the nucleus. Since our concern was cytoplasmic protein contamination of the nuclear fractions, both cytoplasmic and nuclear fractions were probed for GAPDH to demonstrate purity of nuclear fractions relative to cytoplasmic protein contamination (Figure 5A).

In order to substantiate our findings, immunofluorescent localization of phospho-smad2 linker region was determined. Nuclear localization of phospho-smad2 linker region was observed, in agreement with previous biochemical data that phosphorylated smad2 linker region localizes primarily in the nucleus of fibroblasts (Figure 5B). These data suggests Erk phosphorylation of the smad2 linker region is likely limited to a specific subcellular location via TGF-β induction of both smad C-terminal phosphorylation/translocation and activation of the PI3K/Pak2/Erk pathway.

To ascertain the function of linker region phosphorylation of smads, initially we treated fibroblasts with the proteosomal inhibitor MG132 with and without TGF-β (Figure 5A,B). Nuclear levels of both receptor and linker region phosphorylated smad 2 increased with inhibition of proteosomal degradation, consistent with previous data showing nuclear proteosomal degradation of smads [37], but suggested that linker region phosphorylation may be associated with directing smads to proteosomal degradation [28]. However, TGF-β stimulated fibroblasts treated with U0126 alone or with MG132 did not show an increase in C-terminal phospho-smad2 or 3, even after 3 h, as would be expected if Erk mediated linker phosphorylation targeted smads for proteosomal degradation (data not shown). To the contrary, treatment of fibroblast with TGF-β, in the absence or presence of active Erk (Figure 5C), demonstrated that linker region phosphorylation (TGF-β alone) resulted in sustained levels of smad 2 and 3 (C-terminal phosphorylated), suggesting that linker phosphorylation is associated with an increased stability of C-terminal phospho-smads 2 and 3. Using the band densities of multiple experiments, we calculated decay rates for the C-terminal phospho-smad proteins (Figure 5D). The time to decrease the signal intensity by 50% differed between treatment groups for both smad 2 and 3 (116 vs. 135 min. for smad2, and 100 vs 127 min. for smad3; P<0.05, for TGF-β+U0126 vs. TGF-β alone, respectively). Although only linker region phosphorylation of smad2 is shown in previous figures (Figures 4 and 5), there are equivalent putative Erk phosphorylation sites found on smad 3 [24] but the equivalent phospho-linker region antibody for smad 3 is not commercially available. Based on the data shown in figure 5C, smad 3 linker region is similarly phosphorylated by Erk, in that U0126 has the same effect on inhibiting the decay of C-terminal phosphorylated smads 2 and 3.

Since nuclear localized, C-terminal phospho-smads bind DNA and regulate gene expression [38], this implies that increased stability would likely increase duration of function. By transiently treating fibroblasts with TGF-β (10 min.), we wished to assess the stability of smad mediated transcription. The 3 h time point chosen to assess mRNA stability was based on the differences in western blot phospho-smad densities in figure 5C. As shown in figure 5C/D, smads remained phosphorylated longer when Erk was active. Likewise, smad mediated gene transcription of both PAI-1 and Smad7 remained elevated greater than 2 fold (P>0.001; Figure 6A) in the presence of active Erk, three hours after removal of TGF-β treatment.

In addition to the described phosphorylation of nuclear translocated smads, Erk functions by phosphorylating a variety of cytoplasmic and nuclear targets, many of which are critical in cell cycle progression [39]–[42]. Having established a functional connection between Erk and the Smad signaling pathway, we sought to determine if Erk also plays a role in the proliferative effects of TGF-β in fibroblasts. The biological consequences of TGF-β induced Erk activation were addressed using a thymidine incorporation assay (Fig. 6B). Treatment with TGF-β yielded a 6 fold increase in DNA synthesis as compared to untreated fibroblasts. When cells were treated with U0126 to inhibit Erk activation prior to TGF-β addition, this growth stimulation was attenuated (P<0.001). Consistent with our previous data showing Pak2's role in Erk activation, expression of dnPak2 dramatically decreased TGF-β induced growth stimulation (Ad-EGFP+TGF-β vs Ad-dnPAK2+TGF-β; P<0.001). This reduction, rather than a complete block, may be due to the high endogenous Pak2 levels in these cells and dnPak2 being a competitive inhibitor over 20 h, in addition to the previously described minor contribution of Ras. Thus, the ability of TGF-β to induce replication in fibroblasts appears to depend primarily on the function of Pak2 acting through Erk.