Akt is activated downstream of the growth factor receptors and oncogenes implicated in human cancer and plays a critical role in normal development, as well as in tumor pathogenesis, via effects on metabolism, survival, and proliferation. The role of Akt in cell migration and metastases is less clear because of conflicting studies suggesting either positive or negative regulatory roles (Shaw et al., 1997
; Park et al., 2001
). Previous studies have largely relied on overexpression strategies in cancer cell lines or pharmacological inhibitors of PI 3-kinase activity, which would inhibit activity of all Akt isoforms. We describe studies that reveal isoform-specific roles for endogenous Akt1 and -2 in both positive and negative regulation of processes downstream of growth factor receptors and provide insights into mechanisms that may be partly responsible for those conflicting conclusions. In addition, we provide evidence for the importance of Akt1 in a cross-regulatory control circuit between the PI 3-kinase and ERK signaling pathways, two of the major pathways that regulate diverse cellular activities. Differential regulation of signaling pathways by Akt isoforms may critically contribute to their distinct roles in normal development and disease pathogenesis.
The 3D culture system used in this study provides an in vitro model to investigate phenotypic effects that resemble events that take place during breast cancer initiation and progression, such as escape from proliferative suppression, filling of the hollow acinar luminal space, and induction of protrusive, invasive behavior (Debnath et al., 2003a
). We demonstrate that enhanced IGF-I signaling leads to the formation of constitutively proliferating structures with low apoptotic activity and filled lumen. These structures share features with noninvasive breast carcinoma in situ, including maintenance of basement membrane architecture, absence of a hollow lumen, and hyperproliferation. These results extend our understanding of the phenotypic effects of IGF-IR hyperstimulation derived from previous studies in monolayer cultures.
The use of shRNA vectors has made it feasible to investigate cellular pathways required for the distinct IGF-IR–induced phenotypic effects in our model and revealed interesting, isoform-specific functions of Akt. Akt2 down-regulation caused a near complete inhibition of the IGF-IR phenotype in 3D cultures; the structures failed to escape proliferation arrest, to suppress apoptosis of centrally localized cells, and to fill the luminal space. These results indicate that either reduction of Akt2 specifically, or reduction in total Akt levels, suppresses all of the phenotypic effects observed in 3D culture. In contrast, Akt1 down-regulation resulted in a surprising conversion of IGF-IR structures from large, misshapen, solid masses to invasive structures that displayed features associated with EMT (fibroblast-like morphology, enhanced migration, loss of epithelial markers, and acquisition of mesenchymal gene expression). These results are consistent with an inhibitory role for endogenous Akt1 in these processes.
Our data suggest that one target of the inhibitory activity of Akt1 is the ERK signaling pathway, as specific down-regulation of Akt1 enhanced ERK activation both in response to IGF-I or EGF stimulation. These data are consistent with previous studies in which overexpression of activated Akt1 decreased ERK activity (Rommel et al., 1999
; Zimmermann and Moelling, 1999
). Our studies highlight the isoform-specific nature of this effect and establish a role for endogenous Akt1 protein in modulating ERK under conditions of growth factor stimulation. The ability of endogenous Akt to cross-regulate Ras/Raf/MEK/ERK signaling may be conserved across species, as Akt down-regulation in Drosophila melanogaster
, which express only one isoform, enhances insulin-stimulated ERK activation (Freedman, A., and N. Perrimon, personal communication).
Several lines of evidence suggest that inhibition of ERK and migration is specific for Akt1 and unlikely because of the different degrees of residual Akt activity after down-regulation. First, comparable degrees of overexpression of Akt1, but not Akt2, inhibits ERK activation and migration stimulated by EGF. Second, reduction of Akt2 in the background of Akt1 shRNA expression does not enhance ERK activation or migration. Dual down-regulation of Akt2 suppresses enhanced migration in cells overexpressing Akt1 shRNA vectors. Finally, overexpression of Akt2 to levels comparable or greater than Akt1 does not prevent the effects of Akt1 down-regulation, thus supporting an isoform-specific effect of Akt1 that occurs regardless of total levels of residual Akt.
Akt1-mediated inhibition of the ERK pathway could occur at multiple levels. Constitutively active Akt1 is able to phosphorylate a residue (Ser259) of Raf, which mediates binding to 14-3-3 proteins, causing inhibition of Raf activity (Zimmermann and Moelling, 1999
). In preliminary studies, we have not observed significant changes in Ser259 phosphorylation of Raf with Akt1 down-regulation (unpublished data). PI 3/Akt kinase signaling has also been shown to regulate ERK upstream of Raf at the level of IRS-1–Grb2 complex formation (Choi and Sung, 2004
). Furthermore, constitutively active Akt1 has been reported to suppress ERK activity downstream of Raf and MEK and independent of ERK phosphorylation (Galetic et al., 2003
). Thus, there may be multiple levels of ERK regulation and studies are underway to delineate these mechanisms.
The importance of enhanced ERK activation in migration and induction of EMT is supported by studies in which ERK activity was found to be critically involved in EMT induced by other stimuli, such as Ras/TGFβ (Janda et al., 2002
), EGF/TGFβ (Grande et al., 2002
), and HGF/ErbB2 (Khoury et al., 2005
). Pharmacological inhibition of ERK signaling has been shown to decrease invasion or inhibit specific biochemical changes consistent with EMT induced by these stimuli. In our study, enhancement of ERK activation, via a constitutively active MEK2, appears to be sufficient to induce migration, conversion to an invasive phenotype in 3D cultures, repression of E-cadherin, and induction of N-cadherin expression in collaboration with IGF-I hyperstimulation. Furthermore, the enhanced migration induced by Akt1 down-regulation is sensitive to pharmacological inhibition of MEK/ERK signaling.
Interestingly, however, ERK inhibition did not restore expression of epithelial markers or significantly down-regulate mesenchymal markers in Akt1 down-regulated cells that had undergone EMT. These results are consistent with previous ones showing that, although pretreatment or concomitant treatment with a pharmacological MEK inhibitor is able to prevent invasion or the development of EMT (Grande et al., 2002
; Janda et al., 2002
; Khoury et al., 2005
), treatment after the establishment of EMT did not (Khoury et al., 2005
). The failure to completely reverse EMT may be caused by irreversible changes induced by enhanced ERK activation or to ERK-independent pathways that are sufficient to maintain the mesenchymal phenotype induced by Akt1 down-regulation. GSK3β signaling has previously been implicated in E-cadherin suppression (Zhou et al., 2004
); however, we have not observed significant changes in GSK3β phosphorylation after Akt1 down-regulation (unpublished data). Induction of EMT by both Ras and FosER has been reported to induce an autocrine TGFβ loop that stabilizes the mesenchymal phenotype (Gotzmann et al., 2002
; Janda et al., 2002
; Eger et al., 2004
). EMT induced by Akt1 down-regulation may lead to the production of a similar stabilizing soluble factor. Thus, combined inhibition of multiple signaling pathways may be required for full reversion of EMT induced by Akt1 down-regulation.
Akt2 may play a role in growth factor–stimulated migration and invasion that is distinct, if not contrasting, to that of Akt1. This is based on our observations that Akt2 down-regulation suppressed migration stimulated by EGF or Akt1 down-regulation, and Akt2 down-regulation reverted the spindle-shaped morphological changes induced by Akt1 down-regulation. These observations are consistent with previous studies that reported that Akt2 overexpression in breast cancer cell lines enhanced their invasive potential and inhibition of Akt2 (via overexpression of dominant-negative constructs) suppressed invasion and metastases triggered by ErbB2 overexpression (Arboleda et al., 2003
). Although differential localization and regulation of adhesion molecules (e.g., β1 integrin) was implicated in these Akt2 isoform-specific effects, additional studies to examine endogenous Akt2 functions are ongoing.
The present studies do not allow us to establish whether Akt2 preferentially regulates the antiapoptotic activities of Akt because the loss of Akt1 disrupted acinar morphogenesis to such an extent that analysis of apoptosis in the presumptive luminal space could not be evaluated. For similar reasons, we were unable to examine escape from proliferative arrest in Akt1 down-regulated 3D acini. However, we did observe that proliferation of IGF-I–stimulated cells in monolayer cultures was significantly impaired after down-regulation of either Akt1 or -2, indicating that both proteins contribute to IGF-I–stimulated proliferation. Several targets of Akt family proteins have been shown to regulate cell proliferation and apoptosis through effects on the expression or activity of several proteins including cyclin D, cyclin-CDK inhibitors, mTOR, and proapoptotic proteins Bad and FOXO transcription factors (for review see Brazil et al., 2004
). None of the Akt substrates that regulate these proteins have been shown to be specifically phosphorylated by Akt1 or -2. In our preliminary studies, IGF-IR–stimulated phosphorylation of GSK3, FOXO3a, and S6 is not significantly affected by Akt1 or -2 suppression (unpublished data). Previous studies in adipocytes indicated that isoform-specific loss of Akt2 has a more substantial impact on insulin-stimulated glucose uptake than does loss of Akt1 (Bae et al., 2003
; Jiang et al., 2003
; Katome et al., 2003
). Because glucose and other nutrient transporters regulate metabolic processes that affect cell proliferation, effects of Akt2 down-regulation on this pathway may contribute to the reduction in cell proliferation and survival. The precise contributions of Akt1 and -2 to IGF-IR–induced proliferation and antiapoptotic activity require further investigation. In addition, it will be important to examine whether changes in the level of expression of Akt isoforms during morphogenesis contribute in part to the effects of Akt1 and -2 down-regulation in 3D cultures.
The specific mechanisms responsible for the distinct roles of Akt1 and -2 are not known; however, there are a few explanations extrapolated from previously published studies. Differential subcellular localization or binding partners may determine isoform-specific functions. Akt2 expression was reported to be most prominent in regions of cell–matrix contact (Arboleda et al., 2003
). Preferential localization of Akt2 to areas of cell–matrix contact may therefore enable interactions with molecules required for motility and invasion. Differential localization may result from distinct protein-binding interactions. Although the Akt isoforms exhibit significant sequence homology and possess similar domain structure, the greatest variation is located within the phosphoinositide-binding PH domain. Indeed, the L-jun NH2
-terminal kinase scaffold proteins POSH and L-jun NH2
-terminal kinase interacting protein 1 interact selectively with the PH domains of Akt2 and -1, respectively (Kim et al., 2002
; Figueroa et al., 2003
). Chimeric variants of Akt1 and -2 may be useful in establishing which domains of each protein are required for the regulation of ERK activation and cell migration.
The balance between Akt isoform activation downstream of IGF-IR and other growth factor receptors may influence the invasive or metastatic potential of tumors or tumor cell lines. The relative abundance or activation of Akt isoforms may be dynamic and change depending on different cellular contexts. Whether migration or invasion is stimulated may depend on the extent to which a particular agonist activates Akt1 and the extent to which ERK is influenced by Akt1. For example, in tumor cells carrying mutations in Ras or Raf, which activate ERK constitutively, IGF-I may promote cell migration and invasion. Likewise, the ability of wild-type or constitutively activated Akt1 to modulate migration and invasion may depend on whether migration is ERK dependent. The ability of Akt to cross-regulate Ras/Raf/MEK/ERK signaling may also be influenced by mediators that are contextually expressed; e.g., differentiation status in myocytes correlated with the ability to form inhibitory Akt–Raf complexes (Rommel et al., 1999
The suppressive effects of Akt1 on invasive activity reported previously are consistent with evidence that coexpression of activated Akt1 with oncogenic ErbB2 in mouse mammary epithelial cells decreases the metastatic activity of oncogenic ErbB2 (Hutchinson et al., 2004
). However, further studies are required to investigate the precise mechanisms for this decrease. Interestingly, activated Akt1 can rescue the tumor-inducing potential of a mutant form of the polyomavirus middle T antigen that lacks a PI 3-kinase binding site, but does not rescue its invasive activity (Hutchinson et al., 2001
). This could result from Akt1-mediated suppression of ERK activation or from a critical requirement for Akt2 or other downstream PI 3-kinase targets. Acute, inducible knockout of Akt isoforms in transgenic tumor models should be informative in establishing the role of specific Akt isoforms in tumorigenesis in vivo.
Efforts are underway to develop pan- or isoform-specific Akt inhibitors as cancer therapeutics (for review see Barnett et al., 2005
). Akt2 is amplified in breast and ovarian tumors and, more recently, mutations thought to be activating have been detected in colon cancer (Bellacosa et al., 1995
; Parsons et al., 2005
). Thus, Akt2 may be a particularly attractive candidate for therapeutic inhibition. The consequences of isoform-specific inhibition will need to be carefully evaluated in different cellular contexts, especially as there may be unanticipated, differential effects on other signaling pathways, as observed in our studies. The effect of isoform-specific inhibition on different aspects of the tumorigenic phenotype may also vary. Although Akt1 and -2 both contribute to proliferation, isoform-specific pharmacological inhibition may have a differential impact on migration and invasion. Understanding these differences will be key to the development of improved targeted therapeutic strategies.