Loss of epithelial cell-cell adhesions upon exposure to PKC-activating carcinogens is commonly used to model scattering of epithelial cells during tumor metastasis [19, 22, 25]. In this study, we investigate molecular mechanisms underlying tumor promoter-induced disassembly of epithelial apical junctions by challenging HPAF-II human pancreatic epithelial cell monolayers with OI-V or TPA. HPAF-II cells have been derived from highly metastatic pancreatic adenocarcinoma. These cells polarize in culture and develop a high resistance paracellular barrier with well-defined TJs and AJs. In addition, these cells are amenable to transfection with siRNAs. Together, these features make HPAF-II cells an excellent in vitro model to study disruption of epithelial cell-cell adhesions during tumor progression/metastasis.
Exposure of HPAF-II cell monolayers to OI-V or TPA induced rapid opening of the paracellular barrier (Figure 1) and disassembly of AJs and TJs (Figure 2). Since both agents are known to activate PKC, it is logical to suggest that such a junctional disassembly is triggered by activation of PKC. Our pharmacological inhibition and biochemical analyses confirmed the role of PKC activation in disruption of epithelial AJs and TJs and indicated that classical PKC isoenzymes are responsible for this biological effect (Figure 3 and Additional File 2). Different classes of PKC have been previously implicated in phorbol ester-mediated opening of epithelial barriers. Our data are consistent with results of two studies that have implicated classical PKC in TPA-induced increase in paracellular permeability in LLC-PK1  and T84 cell monolayers . However, some other studies have suggested the role of novel PKC isoforms in TPA-induced internalization of AJ protein E-cadherin in MDCK cells  and disruption of claudin-4-based TJ in OVCA433 ovarian epithelial cells . Since TPA and OI-V activate both classical and novel PKC, distinct roles of these subfamilies in AJ/TJ disassembly in different epithelia is likely to reflect cell-specific differences in their expressional levels and/or junctional association. It is noteworthy, that pharmacological inhibitors do not allow identification of the individual PKC isoform, which is responsible for AJ/TJ disruption in HPAF-II cell monolayers. Further studies involving more selective RNA interference or dominant-negative PKC mutant approaches are required to answer this important question.
A key finding of this study is a critical role of NM II in disassembly of AJs and TJs upon PKC activation in HPAF-II epithelial cells. It should be noted that the involvement of actomyosin contractility in PKC-dependent disruption of cell-cell adhesions has been addressed in previous publication, which yielded conflicting results. Thus, the increased F-actin tension/contraction has been implicated in phorbol-ester-induced disruption of endothelial junctions in one , but not another  study. Furthermore, TPA was shown to either increase [76, 77], have no effect  or decrease [47, 75, 78] RMLC phosphorylation in endothelial and epithelial cell monolayers. Our conclusion that PKC activation disrupts epithelial AJs and TJs via stimulating NM II activity is based on several lines of evidence. First, OI-V triggered NM II activation in HPAF-II cells at the onset of junctional disassembly (Figure 5). Second, NM II relocalized to disassembling junctions in parallel with cell rounding in OI-V-treated cells (Figure 4B, 5C). Finally, inhibition of NM II significantly attenuated OI-V-induced disassembly of AJs and TJs (Figure 6).
Our study provides the first direct evidence implicating NM II activity in the disruption of epithelial apical junctions by PKC-activating tumor promoters. Such activation of NM II is likely to serve as a trigger for junctional disassembly by either breaking adhesive contacts formed by transmembrane junctional proteins, or by destabilizing perijunctional F-actin bundles. This may activate endocytosis of AJ/TJ proteins, thus leading to complete disintegration of apical junctions [19, 25]. Although the role of NM II in the breakdown of epithelial barrier in inflammation is generally accepted [79, 80], it has not been explored whether similar mechanism mediates disruption of epithelial cell-cell adhesions in tumorigenesis. Based on the present in vitro data, we hypothesize that stimulation of actomyosin contractility can also be involved in the loss of epithelial cell-cell contacts during metastatic scattering of tumor cells in vivo.
Although PKC was shown to directly phosphorylate RMLC at Ser1/2 residues [65, 66], such phosphorylation cannot be responsible for the increased level of Ser19/Thr18-phosphorylated RLMC and stimulation of actomyosin contractility. In a search for an intermediate signaling step, which links PKC activation and stimulation of NM II, we identified ROCK as a critical regulator of junctional disassembly in OI-V-challenged HPAF-II cells (Figures 7 &8). Importantly, selective siRNA-mediated knock-down of ROCK-I and ROCK-II showed a unique role of the latter isoform in the PKC-dependent disruption of epithelial junctions (Figure 8). These results are in a good agreement with our recent study, which demonstrated the involvement of ROCK-II, but not ROCK-I in AJ/TJ disassembly induced by the depletion of extracellular calcium in intestinal epithelial cells . Although ROCK-I and ROCK-II are highly homologous (~65% of sequence identity and 92% identity in their kinase domain) , these isoforms can be differentially regulated, and can activate distinct actomyosin-dependent processes [82, 83]. For example, ROCK-II, but not ROCK-I, binds to inositol phospholipids, and is activated by phosphoinositol-3-kinase . These different regulatory mechanisms are likely to underline a selective involvement of ROCK-II in the dynamics of epithelial apical junctions. Importantly, our pharmacological analysis failed to observe the involvement of MLCK or ERK-mediated signaling in PKC-dependent AJ/TJ disassembly (Figure 9), which further supports a unique role of ROCK in stimulating NM II-dependent contractility that disrupts epithelial junctions.
Inhibition of NM II and ROCK while substantially attenuating OI-V-induced disintegration of AJ/TJ structure either did not affect, or only modestly attenuated the decrease in TEER (Figures 6 &7). This suggests that PKC activation compromises integrity of the epithelial barrier via several pathways including, but not limited to ROCK and NM II. Additional mechanisms may involve PKC-dependent changes in phosphorylation of different junctional components, which perturbs normal protein-protein interactions within the AJ/TJ complexes. Indeed, altered phosphorylation of occludin and p120 catenin was observed in epithelial and endothelial cells exposed to PKC-activating tumor promoters 
One intriguing findings of this study is that the signaling cascade, which is triggered by PKC activation and mediated by ROCK-II does not involve RhoA GTPase (Figure 10). This implies that PKC can bypass RhoA in stimulating ROCK-II activity and is consistent with a recent report demonstrating PKC-dependent ROCK activation downstream of RhoA in human endothelial cells . Our findings also agree with data obtained in other experimental systems which demonstrated that RhoA inhibition did not prevent TPA-induced contraction in cerebrovascular smooth muscle  or TPA-mediated reorganization of NM II in CHO cells . Lack of involvement of RhoA in OI-V-induced junctional disassembly indicates that this process may be either regulated by other Rho isoforms such as RhoC and RhoE, or may involve Rho-independent mechanisms.
Recent studies have unraveled a complexity of mechanisms regulating ROCK activity which also involves Rho-independent activatory modes. For example, a Rho-independent cleavage of ROCK, that renders this enzyme constitutively active, was reported by several groups [88–91]. Particularly, ROCK-II was shown to be cleaved by caspase-2 in thrombin-stimulated endothelial cells  or by granzyme B in apoptotic lymphocytes . However, this mechanism is not responsible for ROCK-II activation by PKC-targeting tumor promoters in model pancreatic epithelium, since OI-V did no induce ROCK-II cleavage in HPAF-II cells (data not shown), and caused junctional disassembly in caspase-independent fashion (Additional File 5). Further studies are required to elucidate mechanisms of PKC-dependent activation of ROCK-II in epithelial cells.