|Home | About | Journals | Submit | Contact Us | Français|
The activity of protein kinase B, also known as Akt, is commonly elevated in human malignancies and plays a crucial role in oncogenic transformation. The relationship between Akt and the mitogen-activated protein kinase cascade, which is also frequently associated with oncogenesis, remains controversial. We report here examples of cooperation between Akt and cRaf in oncogenic transformation, which was accompanied by elevated activity of extracellular signal-regulated mitogen-activated protein kinases. The effect of Akt on extracellular signal-regulated kinases depended on the status of p21-activated kinase (PAK). Importantly, disruption of the function of PAK not only uncoupled the activation of Akt from that of extracellular signal-regulated kinases, but also greatly reduced the capacity of Akt to act as a transforming oncogene. For the malignancies with hyperactive Akt, our observations support the role for PAK-1 as a potential target for therapeutic intervention.
The classical mitogen-activated protein (MAP) kinase cascade and the PI-3-K-Akt pathway are two signaling mechanisms that are commonly found activated in human malignancies, are actively pursued as therapeutic targets, and are the subjects of an immense body of research literature reviewed elsewhere (for example, Kandel and Hay, 1999; McCubrey et al., 2006; Roberts and Der, 2007; Yuan and Cantley, 2008). Both mechanisms are engaged after activation of various growth factor receptors and, at least in some cells, both could be turned on by activated Ras.
The MAP kinase cascade typically ensues from recruitment and activation at the plasma membrane of a ‘MAP kinase kinase kinase’, such as cRaf. Consequently, cRaf phosphorylates and activates dual specificity MEKs (‘MAP/ERK Kinases’, also known as ‘MAP kinase kinases’), which, in turn, phosphorylate and activate MAP kinases such as extracellular signal-regulated kinases (ERKs). Activated ERKs control the function of various transcription factors, such as the ones belonging to the Ets family, and additional kinases (for example, ribosomal protein S6 kinases). Several other factors have been implicated in the control of this pathway. For example, the function of cRaf may be influenced by the status of 14-3-3 proteins and by p21-activated kinases PAK-1 and PAK-3 (King et al., 1998; Chaudhary et al., 2000). In various models, continuous engagement of this signaling pathway contributes to the survival and proliferation of cancer cells.
Similarly, a multitude of effectors are capable of elevating the phosphoinositol-3-kinase activity in a cell, leading to accumulation of phospholipids that act to recruit members of the Akt family to the plasma membrane. The membrane-bound Akt undergoes full activation after phosphorylation by phosphotidyl inositol-dependent kinases (PDKs) and is capable of phosphorylating a plethora of cellular proteins. The precise number, identity and biological relevance of such targets are a matter of intense research and debate. It is clear that transcription, translation, carbohydrate and lipid metabolism, cell adhesion, motility and death are all influenced by the status of this pathway. Constitutive activation of Akt in cancer is achieved through amplification or mutation of the corresponding genes, overexpression of the catalytic subunit of PI-3-K, mutation or overexpression of various growth factor receptors, and the loss of negative regulator PTEN. As an oncogene, hyperactive Akt may enhance resistance to growth-arresting and pro-apoptotic impacts (Kennedy et al., 1999; Mirza et al., 2000), and facilitate acquisition of additional mutations in some conditions of genotoxic stress (Kandel et al., 2002).
The interplay between the two signaling cascades remains controversial. As both pathways could be simultaneously engaged and, apparently, contribute to the same features of cancer cells, it would appear that a positive cooperation between the two might exist. Unexpectedly, an early report claimed that Akt directly phosphorylates and inactivates cRaf (Zimmermann and Moelling, 1999). Based on the stated direct nature of this interaction, one might think that an increase in Akt activity would be always inhibitory to Raf, yet the same group has reported that the effect of Akt activation on Raf varies dramatically depending on the conditions of treatment (Moelling et al., 2002). Others have suggested that Akt interferes with ERK activation, but the point of interference is downstream of Ras, Raf and MEK (Galetic et al., 2003). In contrast, several reports described cooperation between the two pathways in acquiring growth factor independence and in cell-cycle progression (McCubrey et al., 2001; Sheng et al., 2001; Mirza et al., 2004).
We examined the status of cRaf protein in mouse embryonic fibroblasts that did or did not express a constitutive form of mouse Akt1 (mAkt) (Figure 1a). We observed that the levels of endogenous cRaf increased upon mAkt expression. This was seen even in the cells in which cRaf expression was greatly elevated through the introduction of human cRaf. Importantly, co-expression of mAkt and cRaf resulted in a noticeable increase in the activity of MAP kinase cascade, as is evidenced by the increase in phosphorylation of ERK kinases. Interestingly, ERK activation was readily achieved by expression of an activated Harvey Ras protein, but without an increase in the level of cRaf. Thus, the specifics of ERK activation may differ depending on whether this is achieved by activation of Ras or through cooperation of mAkt and cRaf. Importantly, in these experiments we assay the steady-state condition of the signaling pathways in genetically engineered cells, which resemble cancerous cells harboring activated oncogenes, but may be distinct from the cells transiently treated with growth factors.
We hypothesized that an apparently similar effect on the exogenous and endogenous proteins, which are being expressed from different promoters in the context of different transcripts, may point to a post-transcriptional effect on cRaf abundance. We also hypothesized that such a change may reflect participation of cRaf in a different set of interactions, which may become visible as intercellular redistribution of the protein. To test this hypothesis, we investigated the localization of cRaf as a function of Akt activity. We observed distinct high-intensity staining at the periphery of the cells that expressed mAkt. Some accumulation of cRaf at this location was occasionally seen in the cells that were fed with serum-containing medium, but not in the presence of LY294002 (‘LY’), a compound that inhibits PI3 kinase and endogenous Akt. Therefore, Akt activity in these cells facilitates targeting of a subset of cRaf molecules to a specific compartment at the cell surface. This observation may be significant, because peripheral localization of cRaf has been associated with elevated oncogenic activity of this kinase before (Leevers et al., 1994; Stokoe et al., 1994) and resembles the Akt-dependent localization of Rac-1 (Somanath and Byzova, 2009).
The cooperation between overexpressed cRaf and mAkt extended beyond activation of ERKs. Although none of the protein was able to relieve our mouse embryonic fibroblast culture of contact inhibition, this was readily achieved upon their co-expression (Figures 1b–f). The metabolic shutdown, which is characteristic of contact-inhibited cells (Bereiter-Hahn et al., 1998), was also lost when mAkt and cRaf were co-expressed; and the doubly infected cells rapidly acidified growth medium. Similar results were obtained with rat intestinal epithelial cells RIE-1 (data not shown). We concluded that the observed biochemical cooperation between Akt and cRaf correlates with multiple biological features of oncogenic transformation.
Unlike the earlier report that observed cooperation between the activated forms of the two kinases (Sheng et al., 2001), cRaf used in our system was a wild-type protein and on its own was unable to significantly affect the activity of ERKs. It appears that at least two Akt-dependent signals, one of them requiring Raf activation, are necessary for transformation in these cells. Indeed, the activated form of MEK kinase, which mimics the effect of activation by Raf, relieved contact inhibition, but only in cooperation with activated Akt (data not shown). This is similar to the observation that activated MEK and Akt cooperatively relieve hematopoietic cells of cytokine dependence (Shelton et al., 2004).
The biological consequences of activation of ERKs ensue, in part, from activation of the Ets family of transcription factors which, at least in some case, is required for the transformed phenotype (Galang et al., 1996). We examined the regulation of this pathway using transient transfection of an Ets-dependent reporter construct in HEK293T cells. It has been suggested that PI3K may affect cRaf through activation of PAK-1, which phosphorylates cRaf on serine 338 (Chaudhary et al., 2000; Sun et al., 2000). We observed that phosphorylation of both PAK-1 and ERKs was inhibited by LY (Figure 2a). PAK-1 itself is known to be regulated by Rac-1. Predictably, phosphorylation of ERKs was sensitive to dominant-negative forms of Rac-1 (RacS17N) and PAK-1 (PAK K299R), and was induced by the constitutively active forms of these proteins (Rac-1 Q61 L and PAK T493E, respectively) (Figures 2b and c), whereas the status of serine 338 on cRaf was PAK-1-dependent (Figure 2c). Ets activity also responded profoundly to the manipulations of PAK-1 and Rac-1 (Figure 2d). Inhibition of endogenous Akt by LY or by expression of the dominant-negative form of this kinase (Akt1 K179D) reduced Ets activity, whereas mAkt activated it (Figure 2d). The effect of dominant-negative Akt was partially alleviated by simultaneous expression of the constitutively active form of PAK-1. On the other hand, dominant-negative PAK-1 abolished induction of Ets activity by mAkt (Figure 2e). Importantly, dominant-negative PAK-1 was able to ablate the Ets-stimulating activity even in the case of ddAkt (Figure 2f). This Akt variant has the PDK-sensitive serine and threonine of wild-type protein substituted for aspartates to mimic constitutive phosphorylation by PDKs. It lacks the constitutive membrane localization signal of mAkt and retains the native PH domain. Although we cannot rule out multiple mutual interactions between PAK-1 and the PI3K-Akt pathway, our data suggest that there is at least one PAK-1-dependent step in the process of ERK activation by Akt. Overall, our observations are compatible with signaling from activated Akt to the Rac-1/PAK-1 complex to the MAP kinase cascade and on to the respective target transcription factors.
PAK-1 is essential for Ras-induced upregulation of cyclin D1 (Nheu et al., 2004) and is an important intermediary in Ras-mediated oncogenic transformation of Rat-1 fibroblasts (Tang et al., 1997). These cells are also prone to transformation by expression of activated Akt (Mirza et al., 2000). Our observations predict that transformation of these cells by Akt also depends on the function of PAK-1. To test this prediction, we expressed mAkt in this cell line, alone or with dominant-negative PAK-1. Activated Akt elevated the levels of active ERKs, and this effect was abolished by co-expression of dominant-negative PAK-1 without a significant loss of phosphorylation on either of the PDK sites (Figure 3a). When grown past confluence, original Rat-1 form a flat monolayer, while their transformed derivatives continue proliferation, forming multilayer foci that could be easily detected on methylene blue staining. The loss of contact inhibition, a classic feature of oncogenic transformation in fibroblasts (Weinberg, 2007), was readily induced by mAkt, but was dramatically reduced in the presence of dominant-negative PAK-1 (Figure 3b). Similar results were obtained in experiments that used different sets of expression constructs for mAkt and dnPAK (data not shown), and the ability of mAkt to facilitate tumor growth was significantly diminished by expression of dnPAK (Figure 3c). Transforming ability of mAkt was also reduced by co-expression of an shRNA against PAK-1. Although the use of RNA interference bares some risk of artifacts (Gartel and Kandel, 2006), the concordance between the dnPAK and shRNA experiments proves that Akt-mediated transformation of these cells is PAK-1 dependent.
Our observations confirm that at least in some cells constitutive activation of Akt can cooperate with the MAP kinase cascade in oncogenic transformation, thus explaining the co-activation of both pathways in human cancers. Furthermore, we have observed that in our models the hyperactivation of elements of the MAP kinase cascade is necessary, albeit not sufficient, for full oncogenic activity of Akt. Importantly, we have observed that disrupting the function of PAK-1 uncoupled activation of Akt from that of ERKs, and prevented Akt from acting as a transforming oncogene. Considering the prevalence of Akt activation in human malignancies, out findings predict that PAK-1 is a strong candidate target for therapeutic intervention.
We thank Dr Nikolay Neznanov for the Ets-reporter construct, Dr Channing Der for RIE-1 cells, Dr George Stark for the MEF cells and Dr Nissim Hay for various Akt-expressing constructs. This work was supported by the Howard Temin Award to ESK, NIH Grant HL071625 to TB and American Heart Association grant 0830326N to PRS.
Conflict of interest
The authors declare no conflict of interest.