AKT/PKB (protein kinase B) kinases mediate signaling pathways downstream of activated tyrosine kinases and phosphatidylinositol 3-kinase. AKT kinases regulate diverse cellular processes including cell proliferation and survival, cell size and response to nutrient availability, tissue invasion and angiogenesis. Many oncoproteins and tumor suppressors implicated in cell signaling/metabolic regulation converge within the AKT signal transduction pathway in an equilibrium that is altered in many human cancers by activating and inactivating mechanisms, respectively, targeting these inter-related proteins. We review a burgeoning literature implicating aberrant AKT signaling in many sporadic human cancers as well as in several dominantly inherited cancer syndromes known as phakomatoses. The latter include disorders caused by germline mutations of certain tumor suppressor genes, that is, PTEN, TSC2/TSC1, LKB1, NF1, and VHL, encoding proteins that intersect with the AKT pathway. We also review various pathogenic mechanisms contributing to activation of the AKT pathway in human malignancy as well as current pharmacologic strategies to target therapeutically components of this pathway.
The AKT/PKB (protein kinase B) kinases, which include AKT1, AKT2, and AKT3, are key intermediates of signaling pathways that regulate cellular processes controlling cell size/growth, proliferation, survival, glucose metabolism, genome stability, and neo-vascularization (reviewed in Bellacosa et al., 2005). The biochemical mechanisms leading to AKT activation are well defined and have been recently reviewed in detail elsewhere (reviewed in Scheid and Woodgett, 2003; Brazil et al., 2004; Bellacosa et al., 2005).
A large body of literature has documented frequent hyperactivation of AKT kinases in a wide assortment of human solid tumors and hematological malignancies (reviewed in Bellacosa et al., 2005). Moreover, genetically modified mice have been used as in vivo models to demonstrate that aberrant AKT signaling can contribute to malignancy, either alone or in cooperation with other genetic alterations (Luo et al., 2003; Bjornsti and Houghton, 2004). Since the AKT signaling cascade is frequently deregulated in many types of cancer and, in some malignancies, has implications with regard to tumor aggressiveness (see e.g., Mitsiades et al., 2004), there is potential utility in molecularly targeting components of the AKT pathway for cancer therapy and, possibly, cancer prevention.
AKT is now known to be a central node in a signaling pathway consisting of many components that have been implicated in tumorigenesis, including upstream phosphatidylinositol 3-kinase (PI3K), PTEN (Phosphatase and Tensin homologue deleted on chromosome Ten), NF1 and LKB1, and downstream tuberous sclerosis complex 2 (TSC2), Forkhead Box Class O (FOXO) and eukaryotic initiation factor 4E (eIF4E). Several of these proteins (AKT, eIF4E, and both the p110α catalytic and p85α regulatory subunits of PI3K) can behave as oncoproteins when activated or overexpressed, while others (PTEN, FOXO, LKB1, TSC2/TSC1, NF1, and VHL) are tumor suppressors. Somatic genetic and/or epigenetic changes involving genes encoding these AKT pathway components have been reported in various sporadic cancers. Moreover, germline mutations in PTEN, LKB1, TSC2/TSC1, NF1, and VHL are linked with five different dominantly inherited cancer syndromes characterized by numerous scattered hamartomas, which are benign tumors with normal differentiation but disrupted architecture, and predisposition to certain malignancies (Boudeau et al., 2003; Eng, 2003; Kwiatkowski, 2003; Johannessen et al., 2005). Each of these tumor suppressors is a negative regulator of the AKT-mTOR pathway, which, when deregulated, results in altered translation of cancer-related mRNAs that regulate cellular processes such as cell cycle progression, autocrine growth stimulation, cell survival, invasion, and communication with the extracellular environment (Mamane et al., 2004).
In this review, we summarize a large body of evidence implicating hyperactivation of the AKT pathway in many types of human cancer and dominantly inherited cancer syndromes. We also summarize various pathogenic mechanisms contributing to activation of the AKT signal transduction pathway in human cancer, and we briefly summarize current efforts to target molecularly various components of the AKT pathway for cancer therapy.
AKT regulates many cellular processes implicated in tumorigenesis
AKT is now known to play a central role in many cellular processes that, when deregulated, can contribute to the development or progression of cancer. Many of these processes are highlighted in detail in other reviews presented in this issue of Oncogene Reviews and will be discussed only briefly here.
Activated AKT is a well-established survival factor, exerting antiapoptotic activity in part by preventing the release of cytochrome c from the mitochondria (reviewed in Whang et al., 2004). AKT also phosphorylates and inactivates the proapoptotic factors BAD and procaspase-9 (reviewed in Downward, 2004). Moreover, AKT phosphorylates and inactivates the FOXO transcription factors, which mediate the expression of genes critical for apoptosis, such as the Fas ligand gene. AKT also activates IκB kinase (IKK), a positive regulator of NF-κB, which results in the transcription of antiapoptotic genes (reviewed in Pommier et al., 2004). In another mechanism to thwart apoptosis, AKT promotes the phosphorylation and translocation of Mdm2 into the nucleus, where it downregulates p53 and thereby antagonizes p53-mediated cell cycle checkpoints (Mayo and Donner, 2002; Zhou and Hung, 2002).
AKT activation mediates cell cycle progression by phosphorylation and consequent inhibition of glycogen synthase kinase 3β to avert cyclin D1 degradation (reviewed in Liang and Slingerland, 2003). AKT also directly antagonizes the action of the cell cycle inhibitors p21WAF1 and p27Kip1 by phosphorylating a site located near the nuclear localization signal to induce cytoplasmic retention of these cell cycle inhibitors (reviewed in Testa and Bellacosa, 2001; Bellacosa et al., 2005). Moreover, phosphorylation of AKT/mTOR kinases also results in increased translation of cyclin D1, D3, and E transcripts (Muise-Helmericks et al., 1998).
AKT activates the downstream mTOR kinase by inhibiting a complex formed by the tumor suppressor proteins TSC1 and TSC2, also known as hamartin and tuberin (see review in this issue by Astrinidis and Henske). mTOR broadly mediates cell growth and proliferation by regulating ribosomal biogenesis and protein translation (see review in this issue by Ruggero and Sonenberg) and can regulate response to nutrients by restricting cell cycle progression in the presence of suboptimal growth conditions (see accompanying review by Plas and Thompson). In brief, mTOR stimulates protein synthesis by phosphorylating p70 S6 kinase (p70 S6K) and eIF4E binding proteins 1, 2, and 3 (4E-BPs). In turn, p70 S6K phosphorylates the ribosomal protein S6 to increased translation of mRNAs with 5′-terminal oligopolypyrimidine (5′TOP) tracts, and phosphorylation of 4E-BPs releases the initiation factor eIF4E to promote cap-dependent translation of messages such as those encoding cyclin D1, MYC, and vascular endothelial growth factor (VEGF) (Ruggero and Pandolfi, 2003; Bjornsti and Houghton, 2004). The hypoxia-inducible transcription factor (HIF1) is also targeted by mTOR. Interestingly, mTOR inhibition by the rapamycin derivative RAD001 (everolimus, Novartis) was able to reverse completely AKT-dependent prostate intraepithelial neoplasia (PIN) through regulation of apoptotic and HIF1-dependent pathways (Majumder et al., 2004; also see review by Majumder and Sellers, in this issue).
AKT signaling also contributes to other cellular processes considered to be cancer hallmarks (Hanahan and Weinberg, 2000). For example, VEGF exhibits multiple biological activities in endothelial cells, and VEGF effects on cell survival have been shown to be mediated by the Flk1/VEGFR2-PI3K-AKT pathway (reviewed in Shiojima and Walsh, 2002). Moreover, tumor cell migration is, in part, linked to AKT signaling (reviewed in Lefranc et al., 2005). AKT has also been shown to contribute to tumor invasion and metastasis by promoting the secretion of matrix metalloproteinases (Thant et al., 2000) and the induction of epithelial–mesenchymal transition (EMT) (see review in this issue by Larue and Bellacosa). In other cellular processes, AKT has been shown to phosphorylate human telomerase reverse transcriptase, which stimulates telomerase activity and replication (Liu, 1999). Collectively, these findings implicate upregulation of the AKT pathway in many aspects of tumorigenesis.
Human tumors exhibit frequent alterations of AKT
Amplification and overexpression of AKT
In 1992, the first recurrent alterations of an AKT gene were identified in human cancer, with the demonstration of amplification and overexpression of AKT2 in a subset of ovarian carcinomas (Cheng et al., 1992). AKT2 was shown to be amplified and overexpressed in two of eight ovarian carcinoma cell lines and two of 15 primary ovarian tumors. A multicenter study verified these findings on a larger series of tumors, demonstrating AKT2 amplification in 16 of 132 (12%) ovarian carcinomas but in only three of 106 (3%) breast carcinomas (Bellacosa et al., 1995). Overexpression of AKT2 mRNA also was observed in three of 25 fresh ovarian carcinomas that were negative for AKT2 amplification. Overall, AKT2 amplification was more frequent in undifferentiated ovarian tumors (four of eight, P<0.02), suggesting that AKT2 alterations may be associated with tumor aggressiveness. Amplification/overexpression of AKT2 was proposed to influence the malignant phenotype by permitting a tumor cell to become overly responsive to ambient levels of growth factors that normally would not enhance proliferation (Hanahan and Weinberg, 2000; Testa and Bellacosa, 2001). Amplification of the chromosome region 19q13.1–q13.2, the native location of the AKT2 gene, was also reported in other primary ovarian tumors, and amplification and overexpression of AKT2 was demonstrated in several ovarian cancer cell lines (Thompson et al., 1996). In addition, AKT2 amplification has been reported in a non-Hodgkin's lymphoma (Arranz et al., 1996).
We and others have reported amplification and/or overexpression of AKT2 in 10–20% of primary pancreatic carcinomas and pancreatic cancer cell lines (Cheng et al., 1996; Miwa et al., 1996; Ruggeri et al., 1998). The pancreatic cancer cell lines PANC1 and ASPC1 exhibited 30- and 50-fold amplification of AKT2, respectively, and high levels of AKT2 RNA and protein (Cheng et al., 1996). To address the potential importance of molecularly targeting the AKT pathway, PANC1 cells were transfected with an antisense AKT2 construct. AKT2 expression and tumorigenicity of PANC1 cells in nude mice was found to be markedly inhibited following transfection with antisense AKT2 but not with a control, sense AKT2 construct (Cheng et al., 1996). In a rat tracheal xenotransplantation assay, ASPC1 and PANC1 cells expressing antisense AKT2 RNA remained confined to the tracheal lumen, whereas the respective untransfected cells invaded the tracheal wall. In contrast, no difference was seen in the growth pattern between control and antisense-treated COLO 357 cells that did not exhibit endogenous amplification or overexpression of AKT2. Collectively, these early experiments suggested that overexpression of AKT2 contributes to the growth and invasiveness of a subset of human ductal pancreatic cancers, and that selective targeting of AKT2 could have important therapeutic implications.
Elevated expression of AKT2 protein, but not AKT1, has been reported in nearly 40% of hepatocellular carcinomas, and AKT2 overexpression was found to be an independent prognostic marker (Xu et al., 2004). While an immunohistochemical analysis of human colorectal tissues conducted with a pan-AKT antibody revealed low levels of AKT in normal colonic mucosa and hyperplastic polyps, intense AKT immunoreactivity was observed in 57% of colorectal cancers (Roy et al., 2002). Overexpression of AKT also was implicated as an early event during colon tumorigenesis, since it was detected in 57% of adenomas. Moreover, an antibody specific for AKT2 appeared to duplicate the results seen with the pan-AKT antibody, suggesting that AKT2 was the principal AKT family member involved in colorectal cancers (Xu et al., 2004).
Unlike AKT2, amplification of AKT1 has not been reported as a frequent event in any tumor type. AKT1 amplification was initially detected in a single gastric carcinoma (Staal, 1987), and a more recent investigation of 103 malignant glial tumors revealed a single case (a gliosarcoma) with amplification and overexpression of AKT1 (Knobbe and Reifenberger, 2003). High-level amplification of AKT3 has not been described in human cancers, although small increases in AKT3 copy number are expected in a subset of tumors, since AKT3 resides within the long arm of chromosome 1 (1q), and an extra copy of 1q is common in many tumor types. Despite the absence of amplification specifically targeting the AKT3 locus, expression of AKT3 mRNA has been shown to be upregulated in estrogen receptor-negative breast carcinomas (Nakatani et al., 1999).
Likewise, AKT1 protein levels have been reported to be elevated in some types of cancer, even though the gene is rarely amplified. For example, an immunohistochemical analysis of a series of breast cancers revealed elevated AKT1 staining in 24% of tumors, while strong AKT2 staining was evident in only 4% of tumors (Stal et al., 2003). However, in another series of breast cancers, HER-2/neu expression correlated with elevated expression of AKT2, but not AKT1, and AKT2 protein was upregulated in a breast cancer cell line by ectopic expression of HER-2/neu (Bacus et al., 2002). Moreover, in vitro experiments with human breast and ovarian cancer cells demonstrated that increased invasion and metastasis is associated with overexpression of AKT2, but not AKT1 or AKT3 (Arboleda et al., 2003).
The role of AKT overexpression in cell transformation and malignancy has been addressed in several studies. We have shown that overexpression of wild-type AKT2 can transform NIH3T3 fibroblasts (Cheng et al., 1997). Wild-type AKT1 was unable to transform NIH3T3 cells, but NIH3T3 cells stably expressing constitutively activated AKT1 (Myr-Akt) were shown to grow in soft agar and form tumors in nude mice (Sun et al., 2001). Expression of v-Akt and Myr-Akt in squamous cell carcinomas of the tongue was associated with EMT and increased invasiveness in a rat tracheal xenotransplant assay (Grille et al., 2003). Overexpression of AKT2 was shown to upregulate β1 integrins and increase invasiveness/metastasis in human breast and ovarian cancer cells (Arboleda et al., 2003). Unlike AKT1, AKT2 protein localized primarily adjacent to the collagen IV matrix during cellular attachment. Overexpression of AKT2, but not AKT1 or AKT3, also was sufficient to replicate the invasive phenotype observed in PI3K-transfected breast cancer cells. Moreover, expression of kinase-dead AKT2, in contrast to kinase-dead AKT1 or AKT3, blocked invasion induced by PI3K activation or HER-2/neu overexpression. Altogether, these data indicate that AKT2, among members of the AKT family, may have particular importance in mediating PI3K-dependent effects on cellular adhesion, motility, invasion, and metastasis.
Hyperactivation of AKT in human cancer
Many human cancers, including carcinomas, glioblastoma multiforme, and various hematological malignancies, exhibit frequent activation of AKT (Table 1). Numerous investigators have reported correlations between tumor AKT activity and various clinicopathologic parameters (reviewed in Bellacosa et al., 2005). In particular, AKT activation has been shown to correlate with advanced disease and/or poor prognosis in some tumor types. For example, in one study, approximately 40% of breast and ovarian cancers and more than 50% of prostate carcinomas exhibited increased AKT1 kinase activity; and nearly 80% of tumors with activated AKT1 were high grade and stage III//IV carcinomas (Sun et al., 2001). In other studies, activation of the AKT2 kinase was observed in 30–40% of pancreatic and ovarian cancers (Yuan et al., 2000; Altomare et al., 2003). Moreover, elevated AKT3 activity has been reported in estrogen receptor-deficient breast cancer and androgen-insensitive prostate cancer cell lines (Nakatani et al., 1999), suggesting that AKT3 may contribute to the aggressiveness of steroid hormone-insensitive carcinomas.
Many recent studies have employed phospho-specific pan-AKT antibodies, which recognize the phosphorylated (active) form of all three AKT family members, to examine AKT activity in various types of cancer. Nearly all of these studies have revealed frequent activation of AKT. In some reports, activation of the AKT pathway was confirmed by the demonstration of frequent phosphorylation of other pathway components, such as the downstream effectors mTOR and FOXO. Some of these investigations have shown elevated AKT activity to be especially prevalent in high grade, late stage and/or metastatic tumors, and several reports have linked AKT activation with reduced patient survival or tumor radioresistance. However, in some reports, elevated AKT activity did not correlate with tumor stage or grade. Moreover, increased expression of phospho-AKT has been reported in preneoplastic lesions such as bronchial dysplasias, suggesting that AKT activation can be an early event in tumor progression and, thus, may represent a potentially important target for chemoprevention in individuals at high risk of lung cancer (Tsao et al., 2003; Balsara et al., 2004).
Pathogenic mechanisms resulting in perturbation of the PI3K/AKT/mTOR pathway in human cancer
Alterations of the AKT signaling pathway in malignant tumors
Various mechanisms contribute to activation of the AKT pathway in human tumors, including perturbation of the upstream PTEN and PI3K (Figure 1). Other mechanisms include activation of PI3K due to autocrine or paracrine stimulation of receptor tyrosine kinases (Yuan et al., 2000; Tanno et al., 2001; Sun and Steinberg, 2002), overexpression of growth factor receptors such as the epidermal growth factor receptor in glioblastoma multiforme (Schlegel et al., 2002) and HER-2/neu in breast cancer (Bacus et al., 2002), and/or Ras activation (Liu et al., 1998).
In some instances, AKT activation may result from overexpression of a wild-type growth factor receptor. In such cases, the abundance of a given receptor may enable a tumor cell to become hypersensitive to ambient levels of growth factors (Hanahan and Weinberg, 2000). Mutant receptors, on the other hand, can give rise to constitutive activation of the AKT signal transduction pathway (Sordella et al., 2004). Furthermore, in certain hematological malignancies, constitutive activation of AKT can result from a chromosomal translocation that triggers permanent activation of an upstream tyrosine kinase. One example is the aberrant BCR-ABL fusion protein produced by a (9;22) translocation seen in more than 95% of patients with chronic myeloid leukemia (Skorski et al., 1997). Another is the NPM-ALK protein, which is encoded by a fusion gene formed by a (2;5) translocation seen in anaplastic large cell lymphomas (Slupianek et al., 2001).
Upstream alterations in a given tumor are expected to significantly affect the activity of any AKT family member expressed in the malignant cells. For instance, we found that the pattern of AKT2 kinase activity was similar to that of AKT1 and AKT3 in a series of pancreatic carcinomas (S Tanno and J Testa, unpublished data). Although theoretically some tumors may show activation of a single AKT family member due to a point mutation, for example, in the kinase domain, AKT mutations presently have not been reported. For instance, in skin cancers, DNA sequence analysis uncovered no mutations in the regions encoding either the AKT1 PH domain or the AKT1 activation-associated phosphorylation sites at codons 308 and 473 (Waldmann et al., 2001).
Whereas c-Akt is the cellular homologue of a viral oncogene (Staal, 1987; Bellacosa et al., 1991), the avian sarcoma virus 16 contains a transforming gene derived from a cellular counterpart encoding the p110α catalytic subunit of PI3K (Chang et al., 1997). The human homologue, dubbed PIK3CA, also has been implicated as an oncogene in some human cancers. For example, ovarian carcinomas have been reported to have frequent amplification of PIK3CA, increased PIK3CA expression, and consequent increased PI3K activity (Shayesteh et al., 1999). Increased PIK3CA gene copy number was also reported in nearly 40% of head and neck squamous cell carcinomas (Pedrero et al., 2005) and 35% of primary gastric carcinomas (Byun et al., 2003). In the latter investigation, amplification was detected primarily in tumors that retained PTEN expression, suggesting that PIK3CA and PTEN alterations are mutually exclusive pathogenetic events in gastric cancers. Similar to ovarian cancer cell lines, high PIK3CA copy number in gastric cell lines was strongly associated with increased expression of PIK3CA and elevated AKT activity.
More recently, somatic missense mutations of the PIK3CA gene have been reported in 25–35% of gastric and colorectal carcinomas and glioblastomas (Samuels et al., 2004), as well as in about 25% breast cancer specimens and cell lines, with a similar frequency seen in tumor stages I–IV (Saal et al., 2005). The locations of the mutations within PIK3CA coincided with their potential to increase PI3K activity, because expression of a ‘hot spot’ p110α mutant in NIH3T3 cells conferred more lipid kinase activity than did expression of the wild-type protein (Samuels et al., 2004). Moreover, recently the growth-regulatory and signaling properties of the three most common PIK3CA mutations seen in tumors, namely E542K, E545K and H1047R, were each found to induce oncogenic transformation with high efficiency (Kang et al., 2005). Mutant-transformed cells showed constitutive phosphorylation of AKT, p70 S6K and 4E-BP1, although cellular transformation induced by the mutants, was inhibited by rapamycin, suggesting that mTOR and its downstream targets are essential components of the transformation process.
The PIK3R1 gene, which encodes the p85α regulatory subunit of PI3K, was shown to be mutated in a human T-cell lymphoma cell line from a patient with Hodgkin's disease (Jucker et al., 2002). The mutant protein truncated most of the C-terminal SH2 domain, but retained the inter-SH2 domain, and contributed to a constitutively active form of PI3K. DNA analysis also revealed somatic mutations of the PIK3R1 gene in a subset of human colorectal and ovarian tumors as well as in several cancer cell lines (Philp et al., 2001). Somatic mutations were identified in three of 12 colon cancer cell lines, one of 60 colon carcinomas, one of two ovarian cancer cell lines, and three of 80 ovarian carcinomas. All of the identified PIK3R1 mutations encoded deletions in the inter-SH2 region proximal to the Ser 608 autoregulatory site (Philp et al., 2001).
The PTEN tumor suppressor is the most widely studied negative regulator of the AKT pathway (see review by Lian and Di Cristofano in this issue). PTEN is a major lipid phosphatase that dephosphorylates phosphatidylinositol (3,4,5) triphosphate (PIP3) and phosphatidylinositol (3,4) diphosphate to inhibit AKT activation and thereby suppresses tumor formation by restraining PI3K/AKT signaling (reviewed in Di Cristofano and Pandolfi, 2000). Loss of PTEN function leads to an elevated concentration of the PIP3 substrate, and consequent constitutive activation of downstream components of the PI3K pathway, including the AKT and mTOR kinases (Di Cristofano and Pandolfi, 2000).
Somatic mutation and biallelic inactivation of PTEN occur frequently in high-grade glioblastoma, melanoma, and cancers of the prostate and endometrium (reviewed in Sansal and Sellers, 2004). In one investigation, approximately 35% of endometrial cancers were negative for PTEN based on immunohistochemical staining, while Western blot analysis revealed a significant inverse correlation between expression of PTEN and phosphorylated AKT (Terakawa et al., 2003). Pten knockout mouse models have been generated to better understand the function of PTEN in vivo, and heterozygous Pten (+/−) mice have been shown to develop spontaneous tumors of various histologic origins (Di Cristofano et al., 1998).
Emerging evidence suggests that FOXO factors downstream of AKT play a tumor suppressor role in a variety of cancers (see review by Greer and Brunet in this issue). To date, none of the genes encoding FOXO transcription factors have been found to be mutated in a dominantly inherited cancer syndrome, although FOXO degradation has been linked to cell transformation and expression of FOXO reduces tumorigenicity. It is also noteworthy that genes encoding three of the four FOXO family members have been found at chromosomal breakpoints in human tumors (rhabdomyosarcomas for FOXO1; acute mixed lineage leukemias for FOXO3 and FOXO4), further implicating FOXO factors in human malignancy.
In contrast to the tumor suppressor-like functions of the FOXO transcription factors, the eukaryotic initiation factor of translation eIF4E is downstream in the mTOR pathway and has properties of an oncoprotein in vivo (see accompanying review by Ruggero and Sonenberg). It has been established that the eIF4E gene cooperates with other cancer genes to induce tumor formation. For example, eIF4E has been shown to cooperate with c-Myc in B-cell lymphomagenesis (Ruggero et al., 2004; Wendel et al., 2004). In a transgenic mouse model, c-Myc was found to override eIF4E-induced cellular senescence, whereas eIF4E antagonized c-Myc-dependent apoptosis (Ruggero et al., 2004). Moreover, tumors observed in these mice recapitulated those observed in human cancers (e.g., lung adenocarcinomas, lymphomas) characterized by eIF4E overexpression (reviewed in Ruggero et al., 2004). Collectively, these and other findings suggest that eIF4E is an important mediator of oncogenic AKT/mTOR signaling.
AKT/mTOR pathway activation in hereditary cancer syndromes
Dominantly inherited disorders classified as phakomatoses include neurofibromatosis 1 and 2, tuberous sclerosis 1 and 2, von-Hippel–Lindau disease, nevoid basal cell carcinoma syndrome, Cowden disease, Peutz–Jeghers syndrome, familial adenomatous polyposis, and juvenile polyposis (Tucker et al., 2000). These disorders are characterized by scattered hamartomatous or adenomatous ‘two-hit’ lesions that have a low probability of becoming malignant. Each of the genes involved in the phakomatoses encodes a tumor suppressor protein that functions in signal transduction and is important in mammalian development (Tucker et al., 2000). Mutations of the phakomatosis genes are not directly transforming, but instead deregulate signal transduction pathways to consequently stimulate abnormal proliferation of target cells. As outlined below, the AKT/mTOR pathway is now known to play a central role in at least five of the phakomatoses.
Germline PTEN mutations have been found to occur in 80% of individuals with Cowden disease, a heritable multiple hamartoma syndrome with a high risk of breast, thyroid and endometrial carcinomas (reviewed in Eng, 2003). Several different genetic and epigenetic mechanisms of PTEN inactivation can occur in carcinomas derived from different tissues, but a favored mechanism appears to occur in a tissue-specific manner. Inappropriate subcellular compartmentalization and increased/decreased proteosome degradation have been proposed to be two novel mechanisms of PTEN inactivation (Eng, 2003). The resulting decreased or absent expression of PTEN results in constitutive activation of the AKT pathway.
Germline mutations of the tumor suppressor genes TSC1 and TSC2 each give rise to the hereditary disorder known as tuberous sclerosis complex (TSC) (see review by Astrinidis and Henske in this issue). This disorder is characterized by the presence of hamartomas arising most notably in the central nervous system, kidney, heart, lung, and skin, with occasional tumors progressing to malignancy (i.e., renal cell carcinoma). In TSC tumor cells, biallelic inactivation of TSC2 or TSC1 results in constitutive mTOR activity that is independent of the AKT activation status. Indeed, studies with rodent models demonstrated a significant reduction in growth factor stimulation of AKT activity in cells lacking TSC1 or TSC2 (Zhang et al., 2003). Primary tumors from TSC patients and the Eker rat model of TSC express high levels of phosphorylated mTOR and its effectors p70 S6K, S6 ribosomal protein, 4E-BP1, and eIF4G (Kenerson et al., 2002). Moreover, in the Eker rat, inhibition of mTOR by the drug rapamycin induced apoptosis and reduced tumor cell proliferation.
Another way of inactivating TSC2 and contributing to constitutive mTOR activation is through the LKB1 tumor suppressor/AMPK pathway (Figure 1). AMPK (AMP-activated protein kinase) is a master regulator of lipid and glucose metabolism and protein synthesis (Carling, 2004). AMPK activation reduces anabolic pathways, such as lipogenesis, cholesterol biosynthesis and protein synthesis, the latter effect mediated in part by phosphorylation of tuberin that enhances its inhibitory effect on mTOR (Inoki et al., 2003; Shaw et al., 2004). AMPK kinase has recently been identified as LKB1, which is encoded by the gene inactivated in Peutz–Jeghers syndrome, a disorder characterized by multiple gastrointestinal hamartomatous polyps. LKB1's tumor suppressing activity is related to its ability to phosphorylate and activate AMPK, and in both Lkb1-null mouse embryo fibroblasts and Peutz–Jeghers polyps, mTOR signaling is elevated, which suggests that mTOR inhibitors may be useful in the treatment of Peutz–Jeghers syndrome (Shaw et al., 2004).
A recent report showed that the NF1 tumor suppressor also regulates the AKT/mTOR pathway (Johannessen et al., 2005). Loss-of-function mutations in the NF1 tumor suppressor gene underlie the familial cancer syndrome neurofibromatosis type I, which is characterized by benign neurofibromas and malignant peripheral nerve sheath tumors (MPNSTs), as well as hamartomatous lesions of the iris, myeloid malignancies, gliomas, and pheochromocytomas. The NF1-encoded protein, neurofibromin, functions as a Ras-GAP, and deregulation of Ras due to NF1 inactivation is postulated to contribute to tumor development in patients with this disorder. Activated Ras signals to PI3K, resulting in activation of AKT. Tuberin is then phosphorylated/inactivated by AKT, which leads to constitutive activation of the downstream effector mTOR (Figure 1).
Germline inactivation of the von-Hippel–Lindau tumor suppressor gene (VHL) causes hamartomatous tumors associated with the von-Hippel–Lindau syndrome, although somatic mutations of this gene have also been found in hemangioblastomas and clear-cell renal carcinomas (reviewed in Kim and Kaelin, 2004). The VHL protein is involved in the ubiquitination and degradation of the HIF-1α transcription factor, downstream of mTOR (reviewed in Inoki et al., 2005). Loss of the VHL protein results in stabilization of HIF-1α, and increased expression of many HIF-1α target genes, such as VEGF, that are important in regulating angiogenesis, cell growth, or cell survival (Figure 1). Hence, the AKT/mTOR signaling pathway and activation of downstream HIF-1α-regulatable genes play an essential role in the progression of VHL-associated cancers.
Rationale for targeting AKT/mTOR signaling in human cancers
Components of the AKT signal transduction pathway are appealing targets for therapeutic intervention, because AKT signaling promotes cell survival, proliferation and invasion, and blocking this pathway could impede the proliferation of tumor cells by either inducing apoptosis or sensitizing tumors to undergo apoptosis in response to other cytotoxic agents. Justification for targeting the AKT signaling pathway to identify novel anticancer therapies is based on extensive prior studies, as reviewed by Kumar and Madison in this issue.
Chemoresistance is a major obstacle to successful cancer therapy. AKT has been shown to play a significant role in the therapeutic resistance of tumor cells, because it works against apoptotic mechanisms to promote cell survival. For example, ovarian cancer cell lines with either constitutive AKT1 activity or AKT2 gene amplification were highly resistant to paclitaxel, in contrast to cells with low AKT levels (Page et al., 2000). In vitro and in vivo ovarian cancer models combining the PI3K inhibitor LY294002 with paclitaxel had increased efficacy with regard to reducing tumor growth and dissemination compared to single agents alone (Hu et al., 2002). Moreover, PI3K inhibition increased apoptosis selectively in tumor cells expressing elevated levels of activated AKT, but not in tumor cells with low levels of activated AKT (Brognard et al., 2001; Altomare et al., 2004). Recent data also suggest that blockage of the AKT/mTOR signaling pathway by rapamycin restores the susceptibility of breast cancer cells to tamoxifen (deGraffenried et al., 2004).
Strategies for targeting the AKT signaling pathway are described in detail elsewhere in this issue (see reviews by Cheng et al. and by Kumar and Madison), and include selective inhibition of upstream receptor tyrosine kinases, as well as PI3K, PDK1, AKT, and mTOR kinases. In particular, mTOR inhibitors, such as RAD001 (Novartis) and CCI-779 (Wyeth Research), are currently being assessed in the treatment of advanced cancer patients. These rapamycin derivatives bind to the immunophilin FK506 binding protein, FKBP12, which then binds to mTOR and thereby prevents phosphorylation of downstream S6K and 4E-BP1 (Bjornsti and Houghton, 2004; Sansal and Sellers, 2004). In preclinical models, mTOR inhibition induced G1 arrest in various tumor cell lines at low nanomolar concentrations and diminished in vivo tumor growth in Pten heterozygous mice and mice xenografted with PTEN-null human cancer cells (Neshat et al., 2001; Podsypanina et al., 2001). Moreover, in a transgenic mouse model expressing activated AKT in the prostate, RAD001 treatment for 2 weeks eradicated all evidence of PIN lesions by inducing programmed cell death (Majumder et al., 2003), suggesting that mTOR inhibition in preneoplastic cells may have more profound effects than in fully malignant cells. Importantly, preliminary findings from several clinical trials indicate that rapamycin derivatives have promising antitumor activity, even as a monotherapy, with minor toxicity (reviewed in Bellacosa et al., 2005). Collectively, these studies suggest that mTOR inhibitors, such as RAD001, may be practical for long-term usage as chemopreventive agents in patients with a heritable hamartoma syndrome, such as TSC and Peutz–Jeghers syndrome, characterized by perturbation of the AKT/mTOR pathway. Of potential concern is whether the long-term use of mTOR inhibition will affect a feedback loop involving mTOR, p70 S6K, and insulin receptor substrate (see review by Plas and Thompson in this issue). As an alternate strategy, a recent report has shown that RAD001 can enhance the apoptotic effects of conventional therapies (Beuvink et al., 2005). RAD001 was found to enhance cisplatin-induced apoptosis by inhibiting p53-induced p21 expression; moreover, lower doses of DNA-damaging agents were shown to synergize with RAD001. Thus, it may be possible to reduce side effects of conventional therapies while maintaining antitumor efficacy (Beuvink et al., 2005).
As a cautionary note, it has been suggested that pharmaceutical inhibition of AKT may impact glucose metabolism, largely because of the critical role of AKT in insulin signaling and maintenance of glucose homeostasis (Whiteman et al., 2002). To lessen the adverse effects on normal cellular processes and to enhance clinical efficacy, tumor cells need to have an increased sensitivity to inhibitors when compared to normal cells. Numerous studies indicate that tumor cells, unlike normal cells, may be dependent on activated AKT for survival, and that tumor cells exhibiting elevated AKT activity are sensitive to inhibition of the AKT pathway (reviewed in Bellacosa et al., 2005). The specificity for the inhibition of AKT signaling may prevent toxicity to normal cells. Recent studies using a kinase-dead AKT mutant suggest that disruption of AKT signaling can induce selective apoptosis in tumor cells expressing activated AKT in comparison to normal cells and tumor cells expressing low levels of activated AKT (Boulay et al., 2004). Similar selective killing of tumor cells was observed using the AKT inhibitor API-2 (Yang et al., 2004). Collectively, these experimental findings suggest that blocking AKT signaling may not induce severe toxicity in normal tissues (reviewed in Bellacosa et al., 2005), although preclinical findings may not necessarily be predictive of clinical utility.
Interest in the role of AKT in cancer has increased enormously over the past decade, and it is now evident that activation of the AKT pathway is one of the most common molecular alterations in human malignancy. Importantly, many consequences of hyperactive AKT signaling are considered as hallmarks of cancer. Activation of the AKT pathway can occur by diverse mechanisms, and many components of this pathway are now known to act as oncoproteins or tumor suppressors. Intriguingly, several tumor suppressors (PTEN, TSC1/TSC2, LKB1, NF1, and VHL) involved in hereditary hamartomatous syndromes are now known to intersect with the oncogenic PI3K/AKT/mTOR/eIF4E axis.
Since AKT is frequently activated in a wide variety of human cancers and has been shown to confer resistance to conventional cancer therapies, there is considerable rationale for targeting the AKT pathway for new drug discovery. However, because of its involvement in diverse physiological processes, such as glucose metabolism, it will be necessary to address potential problems of toxicity associated with targeting AKT. Given the widespread interest in the AKT kinases, it is reasonable to expect that there will continue to be many important mechanistic and medical insights regarding the AKT pathway, which could lead to novel therapeutic strategies of potential benefit to many cancer patients.
This work was supported by NIH Grants CA77429, CA83638 (SPORE in Ovarian Cancer), and CA06927, and by an appropriation from the Commonwealth of Pennsylvania to the Fox Chase Cancer Center.