aPKC alters the TGFβ response in NSCLC cells through both Smad-dependent and Smad-independent pathways

Transforming growth factor β (TGFβ) signaling regulates many cellular processes including proliferation, apoptosis, and epithelial to mesenchymal transition (EMT), and, in addition to being a key pathway during embryonic development, aberrant TGFβ signaling is a hallmark of several pathological conditions, including cancer and fibrosis (Derynck and Akhurst, 2007; Derynck et al., 2001; Elliott and Blobe, 2005; Massagué, 1992; Massagué and Chen, 2000). The canonical TGFβ pathway involves the cell-surface binding of TGFβ ligand to the TGFβ type II receptor (TβRII), which then binds to and phosphorylates the TGFβ type I receptor (TβRI) (Massagué, 1998). Phosphorylation of TβRI leads to its activation and its ability to transduce intracellular signaling through the phosphorylation of substrate proteins such as the receptor-regulated Smads (R-Smads) Smad2 and Smad3 (Massagué, 1998; Massagué and Chen, 2000). Once phosphorylated, R-Smads accumulate in the nucleus, where they act as transcription factors to regulate subsequent TGFβ gene response (Attisano and Wrana, 2002; Massagué, 1998; Massagué and Chen, 2000; Shi and Massagué, 2003; Siegel and Massagué, 2003). Entry of R-Smads into the nucleus is facilitated by direct binding of these proteins to the nucleopore complex, binding to importins (for Smad3) or by the binding of the common Smad, Smad 4 (Massagué, 2003; Shi and Massagué, 2003)

Importantly, proteins that control the membrane trafficking and endocytosis of TGFβ receptors play a role in regulating the intensity and duration of TGFβ signals. For example, the efficient regulation of Smad signaling can be facilitated by adaptor proteins such as the Smad cytosolic cofactor SARA (Smad anchor for receptor activation, also known as ZFYVE9). SARA contains a Smad-binding domain, as well as a TGFβ receptor complex-interacting region, and it acts as a bridge, facilitating R-Smad presentation to the activated receptor complex (Tsukazaki et al., 1998; Wu et al., 2000). SARA contains a phosphatidylinositol 3-phosphate (PI3P)-binding FYVE domain, which links it to the early endosome, implicating receptor endocytosis and trafficking in efficient Smad signal transduction (Di Guglielmo et al., 2003; Tsukazaki et al., 1998). Once SARA-bound R-Smads are phosphorylated, they dissociate from the SARA–receptor complex, bind to Smad4 and accumulate in the nucleus to regulate transcription (Xu et al., 2000). Interestingly, although receptor endocytosis has been reported to be dispensable for the phosphorylation of R-Smads, it has been reported that endocytosis is required for the efficient dissociation of R-Smads from SARA, nuclear accumulation and subsequent transcriptional response (Runyan et al., 2005). The precise regulation of transcriptional activity of Smads in the nucleus is important for the proper execution of embryonic development by controlling tissue patterning and normal organ development, and it is also important for controlling cellular growth and apoptotic response in adult tissues; deviations are associated with various pathologies (Derynck et al., 2001; Heldin et al., 2009; Pardali and Moustakas, 2007; Shi and Massagué, 2003; Siegel and Massagué, 2003).

Although it is established that Smads are central regulators of gene response to TGFβ, multiple Smad-independent pathways are also initiated upon TGFβ receptor activation (Derynck and Zhang, 2003; Heldin et al., 2009; Massagué, 2003; Moustakas and Heldin, 2005; Pardali and Moustakas, 2007). In addition to Smads, TGFβ can also activate the mitogen-associated protein kinase family (MAPK). There are three principle classes of MAPK proteins – ERK, JNK, and p38 – each of which has a complex but apparent role in the development and progression of cancer (Wagner and Nebreda, 2009). The p38 MAPK pathway downstream of TGFβ has gained considerable interest as a pathway that regulates apoptosis. Briefly, TGFβ receptor activation leads to the recruitment and Lys63-linked autoubiquitylation and activation of TRAF6 (TNF-receptor-associated factor 6), an E3 ubiquitin ligase. This stimulates a cascade that culminates in the activation of p38 MAPK and, ultimately, apoptosis of various cell types (Edlund et al., 2003; Sorrentino et al., 2008; Yamashita et al., 2008). Interestingly, atypical protein kinase C (aPKC) isoforms interact with TRAF6 to mediate cytokine signaling (Sanz et al., 2000), but less is known about the involvement of aPKC in modulating the TGFβ–p38 MAPK pathway.

We have previously shown that aPKC isoforms can alter TGFβ signaling patterns in non-small cell lung cancer (NSCLC) cells by altering receptor trafficking, degrading specific receptor complexes and by enhancing Par6-dependent phosphorylation (Gunaratne et al., 2012; Gunaratne et al., 2013). However, we had not examined gene changes on a large scale. Furthermore, we have not examined whether aPKC isoforms alter TGFβ-induced MAPK pathways. The aPKC isoforms, which consist of PKCι and PKCζ, are a subset of the protein kinase C family that are Ca²⁺ and diacylglycerol (DAG)-independent (Griner and Kazanietz, 2007). Importantly, the aPKCs show altered expression and activities in various cancers (Huang and Muthuswamy, 2010), and PKCι has been described as an oncogene (Fields and Regala, 2007; Regala et al., 2005). Interestingly, aPKC isoforms are known to play a role in p38-MAPK-induced apoptosis, as inhibition or knockdown of aPKC sensitizes glioblastoma cells to chemotherapeutic agents through a p38-dependent mechanism (Baldwin et al., 2006).

In this report, we examine gene changes in aPKC-silenced cells by microarray analyses, and examine how knockdown of aPKC alters Smad and MAPK signaling pathways and stimulates TGFβ-dependent apoptosis of NSCLC cells.

We reported previously that aPKC gene silencing using small interfering RNA (siRNA) temporally extended TGFβ-induced Smad2 phosphorylation (Gunaratne et al., 2012) and inhibited Par6-dependent EMT (Gunaratne et al., 2013). In order to assess the effects of aPKC silencing on TGFβ-dependent transcription, we silenced PKCι and PKCζ in combination (PKCι/ζ) using siRNA and conducted microarray analysis. Table 1 summarizes fold change differences between siControl- and siPKCι/ζ-treated cells following TGFβ induction as determined by microarray analysis. This list was selected from a set of genes commonly known to be regulated by TGFβ (Siegel and Massagué, 2003). The full microarray dataset can be accessed online at the NCBI Gene Expression Omnibus website (GEO; GSE26241). We observed that several classical TGFβ-regulated genes had similar expression patterns between control and aPKC-silenced cells, including BMP4, SNAI1 and SNAI2. However, there were several genes that displayed a reduced TGFβ-dependent response in aPKC-silenced cells compared to that of control cells, including IL1A, SMURF2, MMP2 and MMP9 (shown in bold). In order to assess gene changes in further detail, we investigated several transcripts using real-time PCR (qPCR).

We first assessed the expression of PKCι, PKCζ and PKCα to confirm that our siRNA approach was efficient and specific, and we observed that siRNA treatment was successful in reducing the gene expression of the atypical PKCs (aPKCι and ζ) but not the classical PKCα (Fig. 1A). We next assessed the transcription of known TGFβ-dependent genes in response to TGFβ in control and aPKC-knockdown cells at 4 and 24 h after stimulation with TGFβ. Several genes showed a significantly reduced induction by TGFβ in aPKC-silenced cells at 4 and 24 h after ligand-mediated stimulation. These included SMURF2 (Fig. 1B) in A549 cells (but not in H1299 cells; supplementary material Fig. S1A), SERPINE1 (also known as PAI-1; Fig. 1C; supplementary material Fig. S1B) and MMP9 (Fig. 1D; supplementary material Fig. S1C). We also observed cases in which the TGFβ-dependent response was either partial (SNAI1; Fig. 1E; supplementary material Fig. S1D) or was similar in both aPKC-knockdown and control cells (CDH1; Fig. 1F). Finally, KLF10 (also known as TIEG1) was unresponsive to TGFβ in both treatment groups (data not shown). Taken together, our microarray and qPCR data suggest that TGFβ-dependent transcription is muted in aPKC-knockdown cells, despite the fact that these cells exhibited extended Smad2 phosphorylation in response to TGFβ induction (Gunaratne et al., 2012; supplementary material Fig. S2).

Activated TGFβ receptors phosphorylate receptor-regulated Smads (Smad2 and Smad3) on a C-terminal SSXS motif, which facilitates their dissociation from SARA, association with Smad4 and accumulation in the nucleus to modulate transcription (Shi and Massagué, 2003). Therefore, the reduced TGFβ-dependent transcriptional response that we observed in aPKC-silenced cells might be due to a reduced nuclear translocation of Smad2. The subcellular localization of R-Smads can be controlled by a balance between binding factors that retain them in the cytoplasm versus transcription factors that retain them in the nucleus. One such cytoplasmic retention factor is SARA, a protein enriched on early endosomes (Itoh et al., 2002; Tsukazaki et al., 1998).

SARA preferentially binds to the non-C-terminally phosphorylated form of Smad2, and it is thought that the activated receptor complex formed at the plasma membrane is captured by SARA in the early endosome, which then presents the bound R-Smad to the receptor for phosphorylation (Massagué and Chen, 2000). Smad2 then dissociates from SARA and associates with Smad4 prior to nuclear translocation and the initiation of transcription (Massagué and Chen, 2000). Furthermore, although phosphorylated R-Smads can activate transcription alone, a full TGFβ response requires complex formation by Smad2 and Smad4 (Levy and Hill, 2005; Wrana, 2009). To test this, we immunoprecipitated Smad2 from control and aPKC-silenced cells treated with TGFβ to examine whether SARA would dissociate from Smad2 upon TGFβ addition. TGFβ addition reduced SARA association with Smad2 and resulted in a concomitant increase in binding to Smad4 (Fig. 2A). Interestingly, in aPKC-silenced cells, TGFβ addition reduced SARA–Smad2 dissociation as well as the binding of Smad2 to Smad4, suggesting a deficit in this exchange. This indicated that SARA might be retaining Smad2 in the cytoplasm to a greater degree in aPKC-silenced cells, and might reduce nuclear translocation of Smad2. Because we have previously reported alterations in the membrane trafficking of TGFβ receptors upon PKC inhibition (Gunaratne et al., 2012), we next examined whether SARA levels were altered in aPKC-silenced cells. Interestingly, aPKC-silenced cells showed significantly increased total protein levels of SARA compared with control cells, although no appreciable changes were observed with TGFβ addition (Fig. 2B). We next examined whether increased SARA expression might disrupt its localization to the early endosome, which, in turn, might disrupt normal Smad signaling in aPKC-silenced cells. Using immunofluorescence microscopy, we did not detect any appreciable alteration of SARA colocalization with EEA-1 (an early endosome marker), suggesting that SARA still accessed the early endosome in aPKC-silenced cells (Fig. 2C). We next analyzed whether the increased SARA levels in aPKC-depleted cells could be retaining Smad2 in the cytoplasm.

To examine TGFβ-dependent nuclear translocation of Smad2 in control and aPKC-silenced cells, we carried out immunofluorescence microscopy analysis (Fig. 3A). As expected, in cells transfected with control siRNA, TGFβ induced an increase in Smad2 nuclear staining, suggesting nuclear accumulation of Smad2 (Fig. 3A). Consistent with the microarray and qPCR data, aPKC-silenced cells showed reduced nuclear accumulation of Smad2 in response to TGFβ (Fig. 3A). The reduction in Smad nuclear translocation in aPKC-silenced cells was also observed for another TGFβ receptor R-Smad, Smad3, and also occurred in another NSCLC cell line, H1299 NSCLC cells (supplementary material Fig. S3A,B). Because we observed an increase in SARA expression and association with Smad2 in aPKC-silenced cells (Fig. 2), we next assessed whether reducing SARA levels would reverse the cytoplasmic retention of Smad2 in aPKC-silenced cells. To investigate this possibility, we first silenced SARA using siRNA and observed no differences in Smad2 phosphorylation (supplementary material Fig. S2C) or nuclear accumulation in A549 cells (Fig. 3A). Interestingly, concomitant siRNA targeting of aPKC isoforms and SARA partially restored the nuclear accumulation of Smad2 that was absent in cells with knockdown of aPKC only (Fig. 3A).

To verify and quantify the observation of reduced nuclear Smad accumulation, we also conducted subcellular fractionation studies and immunoblotting of cellular cytosolic and nuclear fractions when cells were treated in the presence or absence of TGFβ (Fig. 3B). Consistent with our immunofluorescence microscopy analysis, TGFβ treatment stimulated an increase in nuclear Smad2 levels in cells transfected with control siRNA. In contrast, aPKC-knockdown cells contained significantly reduced nuclear Smad2 levels upon TGFβ stimulation (Fig. 3B). The amount of TGFβ-dependent nuclear accumulation of Smad2 in SARA-silenced cells was similar to that of control cells (Fig. 3B). However, consistent with the immunofluorescence analysis, the silencing of aPKC isoforms and SARA partially restored the accumulation of Smad2 in aPKC-silenced cells (Fig. 3B).

Taken together, these results suggest that an increase in SARA protein in the cytoplasm of aPKC-silenced cells might be inhibiting Smad2 from translocating to the nucleus after TGFβ stimulation, and possibly altering transcriptional activity. However, the results do not exclude the possibility that other signaling pathways might be altering Smad function upon aPKC knockdown. We therefore examined whether aPKC knockdown would affect MAPK pathways, as these pathways have been shown to crosstalk with the Smad pathway and have been reported to alter R-Smad nuclear targeting.

Smads shuttle to and from the nucleus, and their subcellular localization is primarily controlled through phosphorylation events. MAPK pathways are known to crosstalk with Smads through the phosphorylation of the Smad linker region, which can alter Smad localization and function, including nuclear exclusion (Massagué, 2003; Kretzschmar et al., 1999). Given our observation of reduced nuclear accumulation of Smad2/3 in aPKC-silenced cells, we next assessed whether MAPK pathways were altered in aPKC-silenced cells. We analyzed the levels of the three activated MAPK pathways in response to TGFβ in control and aPKC-silenced cells (Fig. 4). Interestingly, aPKC knockdown increased basal and TGFβ-induced levels of phosphorylated p38 MAPK at both 1 h and 24 h time-points, whereas no appreciable differences were observed for phosphorylated ERK or phosphorylated JNK (Fig. 4A).

We then assessed shorter timecourses of p38 MAPK activation in control and aPKC-knockdown cells and further observed that TGFβ-induced p38 MAPK phosphorylation was increased and extended in aPKC-silenced cells at 0.5, 1.5 and 4.5 h after TGFβ treatment (Fig. 4B). Because MAPK crosstalk has been reported to alter Smad nuclear–cytoplasmic shuttling dynamics (Burch et al., 2010; Kamato et al., 2013; Massagué, 2003), we reasoned that increased p38 MAPK activity might be reducing Smad2 nuclear import in aPKC-silenced cells. When we tested this hypothesis, we observed that p38 MAPK inhibition did not rescue Smad2 nuclear accumulation in aPKC-silenced cells (supplementary material Fig. S3C). This result suggested that p38 MAPK activity might not be responsible for the reduced Smad2 nuclear accumulation observed in aPKC-silenced cells, and that the Smad2 and p38 MAPK pathways might be independent of each other.

TGFβ receptors can activate the p38 MAPK pathway to stimulate apoptosis (Derynck and Zhang, 2003; Yu et al., 2002). We therefore examined whether the increased p38 MAPK signaling observed in aPKC-silenced cells sensitized cells to TGFβ-induced apoptotic response. To measure apoptosis, control and aPKC-silenced cells were treated with or without TGFβ for 48 h, and the level of apoptosis was measured through the assessment of nuclear morphology after Hoescht staining (Fig. 5A). Cells treated with control siRNA showed a modest apoptotic response to TGFβ. This was in contrast to aPKC-silenced cells, which exhibited a significant increase in cell death when treated with TGFβ (Fig. 5A). Importantly, aPKC-silenced cells treated with a p38 MAPK inhibitor showed a reduced number of apoptotic nuclei, indicating that the apoptotic response observed was downstream of p38 MAPK signaling (Fig. 5A). We observed similar results using another NSCLC cell line, H1299 cells (supplementary material Fig. S4A,B). TGFβ-induced cleaved-Parp levels (c-Parp), a marker of apoptosis, were significantly higher in aPKC-silenced cells than in control cells, and c-Parp levels were reduced with the p38 inhibitor (Fig. 5B), although the p38 inhibitor had no effect on Smad2 phosphorylation levels (supplementary material Fig. S4C). These results further highlight the importance of p38 signaling in this apoptotic response.

TGFβ-stimulated apoptosis through p38 MAPK has been previously reported to occur through the recruitment and activation of the E3 ubiquitin ligase TRAF6 (Sorrentino et al., 2008; Yamashita et al., 2008). Briefly, upon TGFβ activation, TRAF6 is recruited to TβRI of the TGFβ receptor complex. This causes TRAF6 to become auto-ubiquitylated, which activates TAK1 (also known as MAP3K7; a MAP3K), which triggers the MAPK cascade to p38 activation (Sorrentino et al., 2008; Yamashita et al., 2008). We have shown previously that aPKC expression can alter the binding and degradation patterns of TβRI and its substrates (Gunaratne et al., 2012; Gunaratne et al., 2013). We suspected that the increased TGFβ–p38 MAPK signals that we observed in aPKC-silenced cells might be due to increased levels of TRAF6 and TGFβ receptor complexes when aPKC was depleted. We tested this by immunoprecipitating endogenous TRAF6 from control and aPKC-silenced cells, followed by immunoblotting for TβRI (Fig. 6A). Interestingly, TβRI associated to a greater degree with TRAF6 in the absence of aPKC expression (Fig. 6A). This finding suggested that TβRI–TRAF6 complexes were more stable in aPKC-knockdown cells. We reasoned that this increase in TβRI–TRAF6 complexes was responsible for the increased TGFβ-induced p38 MAPK signals in aPKC-silenced cells. To test this, we next examined whether siRNA-mediated TRAF6 knockdown could abrogate p38 MAPK signaling in aPKC-silenced cells. We used siRNA to knock down the aPKCs (siPKCι/ζ), TRAF6 alone (siTRAF6) or aPKC and TRAF6 together (siPKCι/ζ+TRAF6) (Fig. 6B). As we had observed before, knockdown of aPKC enhanced TGFβ-induced phosphorylated p38 MAPK levels (Fig. 6B, lanes 3 and 4). Interestingly, in cells where aPKC and TRAF6 were knocked down simultaneously, TGFβ-induced p38 MAPK phosphorylation was abrogated (Fig. 6B, lanes 7 and 8). Furthermore, the p38 signaling patterns were reflective of the level of apoptotic stimulation, as TRAF6 siRNA also reduced the TGFβ-induced apoptotic response seen in aPKC-silenced cells (Fig. 6C). These results suggested that the enhanced p38 MAPK signaling and apoptosis that we had observed in aPKC-silenced cells were TRAF6 dependent.

Finally, given that aPKC knockdown increased TRAF6-dependent p38 MAPK activity and decreased TGFβ-dependent gene transcription, we assessed whether the two observations were linked, i.e. whether TRAF6 modulation would alter TGFβ-dependent gene transcription. To do this, we performed qPCR analysis of SMURF2, SERPINE1, MMP9, CDH-1 and SNAI-1 mRNA levels in control and TRAF6-silenced cells following TGFβ stimulation for 4 and 24 h (Fig. 7). Unlike our previous results regarding SMURF2 mRNA levels in control versus aPKC-silenced cells (Fig. 1B), we observed no differences in SMURF2 transcript levels between control and siTRAF6-treated cells (Fig. 7A). However, consistent with the results observed with aPKC silencing (Fig. 1), TRAF6 silencing decreased TGFβ-dependent SERPINE1, MMP9 and SNAI1 levels (Fig. 7A). Interestingly, the reduction of gene transcription in TRAF6-silenced cells was independent of Smad2 phosphorylation or nuclear accumulation (Fig. 7B,C). Taken together, our results suggest that aPKC isoforms not only regulate TGFβ-dependent gene regulation but also direct NSCLC cells towards EMT versus apoptotic responses using multiple signaling pathways.

TGFβ pathways regulate many developmental and homeostatic processes, and aberrant signaling is associated with various pathologies such as fibrosis and cancer. Here, we have found that aPKC knockdown alters both Smad-dependent and the Smad-independent p38 MAPK signaling pathways and regulates whether NSCLC cells undergo TGFβ-dependent apoptosis.

In this report, we examined the transcriptional changes associated with TGFβ signaling in an aPKC-silenced background. In this context, we found that several TGFβ-stimulated genes showed reduced transcriptional activity and this was a result of reduced Smad2 nuclear accumulation. Furthermore, we found that the knockdown of aPKC increased the basal protein levels of SARA. This is an important finding, as it has been reported previously that increased SARA expression is associated with reduced TGFβ receptor degradation (Di Guglielmo et al., 2003), consistent with our previous findings that aPKC knockdown reduces TGFβ receptor degradation (Gunaratne et al., 2012). Moreover, increased SARA expression is associated with the maintenance of epithelial cell phenotype (Runyan et al., 2009), consistent with our findings that aPKC knockdown reduces TGFβ-induced EMT (Gunaratne et al., 2013). Given our previous finding that aPKC alters membrane trafficking of the TGFβ receptors (Gunaratne et al., 2012), it would be interesting to explore whether aPKC alters the function or localization of receptor-bound SARA to control the context under which Smads are signaling. Interestingly, Runyan and colleagues showed that although inhibiting TGFβ receptor internalization from the membrane only slightly altered phosphorylated Smad2 levels, it did significantly impact on the ability of Smad2 to dissociate from SARA (Runyan et al., 2005). Moreover, Smad2–Smad4 complex formation has also been reported to occur in the early endosome (Chen et al., 2007). This suggests that the coordinated function and subcellular localization of SARA and associated Smads are important for mediating TGFβ-dependent transcription properly.

A recent report has implicated SARA in general endocytic processes through classical ESCRT complex machinery (Kostaras et al., 2013). More specifically, the correct subcellular trafficking of the EGFR from the early endosome to late endosomes to regulate EGFR degradation was dependent on SARA, implicating SARA with a more general role in endocytic trafficking than was appreciated previously (Kostaras et al., 2013). This might have important implications with respect to our findings that aPKC knockdown reduces TGFβ receptor degradation and stabilizes particular TGFβ receptor–protein complexes (Gunaratne et al., 2012; Gunaratne et al., 2013). One possibility is that depletion of aPKC leads to an accumulation of SARA and TGFβ receptor complexes. Indeed, aPKC has been reported previously to be involved in the trafficking of membrane proteins, as well as being involved in the passage of EGFR to lysosome-targeted endosomes through the anchoring protein p62 (also known as SQSTM1) (Sanchez et al., 1998). Whether the knockdown of aPKC in our model is causing a reduced passage of receptors to lysosomes is an important area for future study.

Here, we made the novel finding that the knockdown of aPKC increases and extends TGFβ-induced p38 MAPK activation, which sensitizes NSCLC cells to undergo apoptosis. We found that knockdown of aPKC stabilized TβRI–TRAF6 complexes, and that knockdown of TRAF6 in aPKC-silenced cells returned p38 MAPK activation back to control levels. In line with our results, increases in p38 MAPK activity have been reported before upon aPKC silencing (Baldwin et al., 2006), indicating that aPKC might be attenuating p38 MAPK signaling in multiple tumor cell types. Interestingly, when aPKC is knocked down, p38 MAPK is able to elicit an apoptotic response, indicating that in some situations aPKC might be a viable therapeutic target. However, the role of p38 MAPK in cancer is also complex, and context dependent, and in addition to sensitizing cells to a death response, p38 activity is also associated with cancer cell survival and both the stimulation and suppression of EMT (Bakin et al., 2002; Strippoli et al., 2010; Wagner and Nebreda, 2009). We also report here that the enhanced p38 MAPK activation we observed in aPKC-silenced cells was not responsible for the reduction in Smad2 nuclear accumulation. The role of Smad2 linker phosphorylation by MAPK members in TGFβ signaling has yielded mixed results. The original reports show that linker phosphorylation by MAPK blocked Smad2 nuclear accumulation (Grimm and Gurdon, 2002; Kretzschmar et al., 1999); however, nuclear stabilization of Smad2 by linker phosphorylation has also been reported (Alarcón et al., 2009; Burch et al., 2010), suggesting that Smad linker phosphorylation is more complex than originally thought, and requires further examination.

In conclusion, we have found that aPKC plays multiple roles in TGFβ signaling, and the localization and expression patterns of aPKC might dictate how a cell responds to TGFβ. This is especially important given that aPKC isoforms have recently been implicated in cancer progression (Huang and Muthuswamy, 2010; Kojima et al., 2008) and aPKCι has been classified as a human oncogene (Fields and Regala, 2007; Murray et al., 2011). Whether aPKC might be a viable therapeutic target in TGFβ-driven tumor progression remains to be examined.

Primary antibodies were as follows: anti-β-actin (Sigma, A2668), anti-PKCι (BD Transduction Laboratories, 610175), anti-PKCζ (Cell Signaling Technology, 9372), anti-phospho-Smad2 (Cell Signaling Technology, 3101), anti-Smad2/3 (BD Transduction Laboratories, 610842), anti-tubulin (Sigma, T4026), anti-H3-histone (Millipore, 05-499), anti- phospho-p38 (Cell Signaling Technology, 9211), anti-p38 (Cell Signaling Technology, 9212), anti phospho-ERK (Cell Signaling Technology, 4370), anti-phospho-JNK (Cell Signaling Technology, 9255s), anti-Smad4 (Abcam, AB40759), anti-SARA (Santa Cruz Biotechnology, sc-9135), anti-Flag (Sigma Aldrich, F3165), anti-Traf6 (Cell Signaling Technology, 8028s), anti-EEA1 (BD Transduction Laboratories, 610457). Horseradish-peroxidase-conjugated secondary goat anti-rabbit-IgG (Thermo Scientific, 31460) and goat anti-mouse-IgG (Thermo Scientific, 31430) were used for immunoblot analysis. Fluorescently conjugated donkey anti-mouse-IgG (Life Technologies, A21206) and donkey anti-rabbit-IgG (Life Technologies, A31572) were used for immunofluorescence studies. Human siRNA constructs were purchased from Life Technologies (siPKCζ, siPKCι and siControl catalog numbers were 10620319-HSS183348, 10620319-HSS183318 and 4390844, respectively). TRAF6 siRNA was purchased from Life Technologies (product number s14389- 4390824). p38 MAPK inhibitor was purchased from Calbiochem (506126).

A549 and H1299 NSCLC cell lines were maintained in F12K and RPMI-1640 medium, respectively, supplemented with 10% fetal bovine serum. Cells were kept in a humidified tissue culture incubator at 37°C under 5% CO₂. siRNA and DNA transfections were conducted using Lipofectamine RNAi max and Lipofectamine LTX (Life Technologies) according to the manufacturer's protocol. TGFβ treatments (250 pM) were conducted in low-serum medium (0.2% FBS) for the indicated times after cells were serum deprived overnight.

Cells were lysed (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride and a mixture of protease inhibitors) and centrifuged at 20,000 g at 4°C for 10 min. Aliquots of supernatants were collected for analysis of total protein concentration. Protein concentrations were determined using the Lowry method (Fisher). Immunoprecipitations and immunoblotting were conducted as described previously (Gunaratne et al., 2012; Gunaratne et al., 2013).

Cytoplasmic and nuclear cellular fractions were isolated using the Thermo Scientific Kit NE-PER® kit (78833) according to the manufacturer's protocols.

Cells were fixed with 4% paraformaldehyde, permeabilized with 0.25% Triton X-100, and incubated with primary antibodies at 4°C overnight. Following incubation with the appropriate fluorescent-probe-conjugated secondary antibodies, the probes were visualized by immunofluorescence microscopy using an inverted IX81 Microscope (Olympus, Canada).

To conduct the microarray analyses, siRNA-treated cells (siControl and siPKCι/ζ double knockdown) were treated with TGFβ for 1 h, followed by washout and further incubation of cells for 24 h in low-serum medium. Total RNA was extracted and processed for microarray analysis as described below.

RNA and GeneChips were processed at the London Regional Genomics Centre (Robarts Research Institute, London, Ontario, Canada; http://www.lrgc.ca) as described previously (Guo et al., 2011). Using Partek, any batch effect due to scan date was removed and an ANOVA (Y_ijk = μ+Condition×Time_ij+e_ijk) using Method of Moments (Eisenhart, 1947) was run to determine gene-level P-values. Fold change comparisons are expressed relative to untreated siControl cells and represent the average of three separate experiments (three separate GeneChips per condition). A fold change of ±1.6 was considered as the cutoff for induction.

RNA extraction, reverse transcription and real-time PCR were conducted as described previously (Gunaratne et al., 2013). Primer sequences are shown in supplementary material Table S1. Gene expression in each treatment is expressed relative to the control (siControl, no TGFβ) and is an average of three to six independent experimental trials.

A549 and H1299 cells transfected with the appropriate siRNA constructs were serum deprived (0.2% FBS) and then incubated with or without TGFβ in low-serum medium in the presence or absence of a p38 MAPK (1 µM) inhibitor for 48 h. After 48 h, apoptosis of A549 and H1299 cells was analyzed by examining nuclear morphology after Hoechst 33342 staining. Hoescht stain (1 µg/ml) was added directly to the medium and samples were incubated for 30 min at 37°C. The cells were then visualized using a fluorescent microscope (Olympus IX71), and ten random images were acquired per condition. Normal and apoptotic nuclei were counted and the apoptotic nuclei (characterized by condensed chromatin) were scored as a proportion of normal healthy cells.

One-way or two-way ANOVA analyses followed by post-hoc Bonferonni's test were used to evaluate the significance of the results. Statistical analyses were performed using GraphPad Prism® Software 5.0 and P-values of <0.05 were considered to be statistically significant.

The authors would like to thank the members of the Di Guglielmo laboratory for advice and support.