Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

In 1997, PTEN (phosphatase and tensin homolog deleted on chromosome 10) was identified as the frequently lost tumor suppressor gene in a region of human chromosome 10 (10q23) that was known to be highly susceptible to deletion in malignant tumors of the prostate and the brain (Li et al. 1997; Steck et al. 1997).

Soon after its discovery, the work of Maehama and Dixon (1998) unveiled the biochemical function of PTEN as a plasma-membrane lipid phosphatase that hydrolyzes the 3-phosphate on the second-messenger molecule phosphatidylinositol-3,4,5-triphosphate (PIP3) to generate phosphatidylinositol-4,5-bisphosphate (PIP2). In the following years, several groups (Di Cristofano et al. 1998; Podsypanina et al. 1999; Stambolic et al. 1998) showed that PTEN exerts its function as a tumor suppressor at least in part through negative regulation of the crucial cell survival serine/threonine kinase AKT (PKB).

Since then, PTEN has been shown to affect pleiotropic cellular processes such as cell cycle progression, cell proliferation, senescence, chemotaxis, apoptosis, aging, muscle contractility, DNA damage response, angiogenesis, and cell polarity. In line with its role in multiple crucial cellular processes, PTEN has a role in the pathogenesis of numerous diseases such as diabetes, autism, and cancer.

Indeed, PTEN is one of the most frequently mutated, deleted, and silenced tumor suppressor genes in human cancer. The importance of PTEN as a tumor suppressor is supported by the observations that germline PTEN mutations in humans can result in autosomal dominant syndromes collectively referred to as the PTEN hamartomas tumor syndromes (PHTS), characterized by developmental defects, neurological deficits, multiple hamartomas in various tissues including skin, breast, intestine and brain, and an increased risk of breast, thyroid and endometrial cancers (Liaw et al. 1997; Marsh et al. 1997; Zhou et al. 2000).

The identification of PTEN as an important tumor suppressor gene led to a rapid outburst of several mouse models aimed at understanding the consequences of Pten loss. During this time, these mouse models have been further refined to study specific organs and specific cell lineages. This has allowed us to faithfully recapitulate some features of human cancers and to reconstruct the intricate connections between the PTEN signaling cascade and other oncogenic or tumor suppressive pathways.

Overall, this chapter will focus on the role of PTEN as a critical player in human diseases and, specifically, on the faithful mouse models generated to dissect the roles of this phosphatase in tumor initiation and progression in different organs. Additionally, a particular relevance will be given to the work carried out in vivo in the mice to identify the network of signaling pathways enabling PTEN to exert its tumor suppressive function. Finally, emphasis will be given to the differential outcomes observed in different contexts as a consequence of loss of Pten.

2 Spectrum of Human Diseases Associated with Loss of PTEN

Over the last decade a multitude of important studies have identified PTEN gene mutations in a wide range of sporadic malignancies and at a high frequency in cancer-susceptibility syndromes.

Sequencing of the PTEN gene has revealed that this non-redundant, evolutionary conserved phosphatase is one of the most commonly mutated tumor suppressors in human malignancies (Cairns et al. 1998; Dahia et al. 1997; Duerr et al. 1998; Rasheed et al. 1997; Shao et al. 1998; Tashiro et al. 1997; Wang et al. 1997).

Genetic alterations of the PTEN gene include various types of abnormalities ranging from point mutations (encoding mostly unstable and/or catalytically inactive proteins) to large chromosomal deletions (Georgescu et al. 1999, 2000; Li et al. 1997; Steck et al. 1997). PTEN mutations can affect both alleles in various cancers with the following frequencies: endometrial (∼50%), glioblastoma (∼30%), melanoma (∼12%), prostate (∼10%), and breast (∼5%) (Ali et al. 1999; Birck et al. 2000; Cairns et al. 1997; Celebi et al. 2000; Chiariello et al. 1998; Duerr et al. 1998; Haluska et al. 2006; Lin et al. 1998; Saal et al. 2005; Shao et al. 1998; Steck et al. 1997; Tashiro et al. 1997; Wang et al. 1997; Zhou et al. 2002). Loss of one PTEN allele is frequently observed in the following malignancies: glioma (∼75%), breast (∼40%), colon (∼20%), lung (∼37%), prostate (∼42%) (Bose et al. 1998; Feilotter et al. 1998; Lin et al. 1998; Rubin et al. 2000; Teng et al. 1997).

Importantly, PTEN expression is regulated not only genetically but also at the transcriptional/translational level. DNA methylation, transcriptional repression, and microRNA-directed mRNA degradation and translational abrogation have been reported to be important mechanisms in reducing PTEN expression in several cancers(Wiencke et al. 2007; Yang et al. 2008; Poliseno et al. 2010).

Overall, these findings imply that loss of function of PTEN is a common event in cancer, which is accomplished though several layers of control and mechanisms.

Germline deletion/mutation of PTEN is associated with several autosomal dominant tumor predisposition syndromes including Cowden syndrome, Bannayan-Riley-Ruvalcaba syndrome, Lhermitte-Duclos disease, Proteus syndrome, and Proteus-like syndrome (Liaw et al. 1997; Marsh et al. 1997; Zhou et al. 2000, 2001). These patients often suffer from hamartomas in multiple organs with the risk of progression to malignant cancer transformation. In addition to hamartoma development, patients affected by Cowden syndrome and Bannayan–Riley–Ruvalcaba also develop macrocephaly (Buxbaum et al. 2007; Herman et al. 2007). This observation led to an association of PTEN mutation with autism, which is characterized by patients with macrocephaly (Butler et al. 2005). This observation expanded the role of PTEN to suppress human diseases of non-neoplastic nature. Recently, PTEN loss has also been associated with neurological diseases such as Parkinson's (Gasser 2007) and metabolic syndromes such as diabetes. The latter implication is supported by studies in animal that have demonstrated that Pten deletion causes an insulin sensitivity phenotype (Stiles et al. 2004, 2006). This finding supports the notion that patients affected by PHTS characterized by Pten mutations also present increased insulin sensitivity (Iida et al. 2000).

Together these data point to a role for PTEN as a key regulator of several cellular processes and that deregulation of its function can causes a wide spectrum of human diseases.

3 Modeling PTEN Loss in Specific Murine Organs

Extensive mouse modeling has been performed to elucidate the importance of PTEN and the consequences of its loss in human health and disease.

Homozygous deletion of Pten in the mouse embryo is lethal and is characterized by developmental defects in the mesoderm, endoderm and ectoderm (Di Cristofano et al. 1998). Heterozygous Pten mice develop multiple neoplasias in a wide spectrum of tissues including prostate, thyroid, colon, lymphatic system, mammary gland, and endometrium (Di Cristofano et al. 1998; Podsypanina et al. 1999; Stambolic et al. 2000; Suzuki et al. 1998). These mouse models also recapitulate some of the features of the PTEN-associated hamartoma syndromes in humans.

To further analyze the consequences of Pten loss in other organs, several tissue specific models of Pten deletion have been developed using a conditional gene-targeting approach (Table 1). In this chapter, we will examine several mouse models of Pten conditional inactivation as examples of the human tissues where loss of Pten is observed, such as brain, prostate, and breast.

Table 1 Tissue-specific mouse models of Pten deletion
Full size table

3.1 Brain

The first three mouse models of tissue-specific inactivation of Pten were generated in 2001 (Backman et al. 2001; Groszer et al. 2001; Kwon et al. 2001) with the brain chosen as a target organ. The choice was most likely dictated by the fact that: (1) PTEN is very highly frequently mutated in glioblastoma (30%), the most aggressive primary brain tumor in humans (Knobbe et al. 2002) and (2) syndromes associated with germline mutation of PTEN are characterized by neurological abnormalities.

For instance, patients with Lhermitte–Duclos disease (LDD) develop dysplastic gangliocytoma, which is described clinically as a benign overgrowth of neurons in the cerebellum that causes increased intracranial pressure, ataxia and seizure (Zhou et al. 2003). Although patients affected by LDD have inherited only one normal copy of PTEN, the dysplastic cells have either completely lost PTEN expression or express only the mutant allele due to loss of heterozygosity (LOH); both events are characterized by an increase in the phosphorylation of AKT (Abel et al. 2005; Iida et al. 1998; Zhou et al. 2003). Two of these mouse models generated faithfully recapitulated the features of LDD. In these models, Pten flox/flox mice were crossed with transgenic mice in which Cre recombinase expression is under the control of the glial fibrillary acidic protein (Gfap) promoter. In these mice, Pten is deleted late in the development of granule neurons of the cerebellum and results in a cell-autonomous loss of size regulation (Backman et al. 2001; Kwon et al. 2001). As a consequence, the size of Pten-deficient granule neurons progressively increases without evidence of abnormal proliferation. This observation is reminiscent of the focal lesions in LDD that rarely contain proliferative cells. Furthermore, LDD is characterized by dysplastic neurons ectopically placed in the molecular layer, which is similar to the ectopically positioned granule neurons resulting from a neuronal migration defect in mouse models (Abel et al. 2005; Backman et al. 2001; Kwon et al. 2001). Indeed, several studies have established that deletion of Pten in different neuronal types during development results in marked defects in migration and patterning in brain (Backman et al. 2001; Kwon et al. 2001; Marino et al. 2002; Yue et al. 2005). Another similarity between humans and the mouse models is that abnormalities in synaptic structure have been identified in LDD patients as well as in Pten conditional knockout mice (Fraser et al. 2008; Kwon et al. 2006). Overall, these mouse models suggest that the abnormalities observed in LDD can be attributed to key roles of PTEN in neuronal migration, size regulation, and specialized subcellular structure. Notably, this nonproliferative disease resulting from PTEN inactivation, although not malignant, is often associated with premature morbidity.

The third mouse model developed in 2001 utilized the neural stem cell specific nestin promoter (NesCre) to deliver Cre (Groszer et al. 2001) in neural stem cells, thereby resulting in Pten deletion throughout the entire brain. These mice succumb to an early postnatal death, presumably due to a continuous increase in brain size with individual cells being larger than those from wild-type mice brains. In contrast to the other two mouse models previously described, these mutant mice showed increased cell proliferation and decreased cell death. Using this model, Groszer et al. concluded that Pten most likely negatively regulates neural stem/progenitor cells self-renewal capability by modulating G0–G1 cell cycle entry (Groszer et al. 2006).

Other neurological abnormalities observed in patients with germline mutations of PTEN were also modeled in mice. One manifestation of inherited PTEN mutation includes macrocephaly; several studies have identified autism or autistic behaviors in macrocephalic PHTS patients (Butler et al. 2005; Goffin et al. 2001). Moreover, it is interesting to note that neurological phenotypes associated with PHTS are very variable. As outlined above, the development of LDD is characterized by a second hit that inactivates the wild-type allele of PTEN in the lesions of the cerebellum. It is possible that other neurological deficits observed in PHTS patients, such as macrocephaly, mental retardation, and autism, are also associated with second hits that occur stochastically during development. In such a scenario, the timing and specific cell populations in which PTEN function is lost during development would determine the specific neurological outcome observed. For instance, a mouse model where Pten was deleted in subsets of differentiated neurons in the cerebral cortex and hippocampus showed anxiety-like behavior and decreased learning, that may recapitulate the autistic features of some PHTS patients (Kwon et al. 2006).

Although PTEN is frequently inactivated in malignant human brain tumors, PHTS is not associated with an increased incidence of brain tumors, and mice with heterozygous loss of Pten fail to develop brain tumors. Brain tumors are also not observed in conditional knockouts targeting Pten deletion in the brain, indicating that cooperating mutations in other genes are required for the neoplastic process (Backman et al. 2001; Fraser et al. 2004; Groszer et al. 2001; Kwon et al. 2001; Marino et al. 2002) (see following paragraphs).

3.2 Prostate

Prostate cancer and glioblastoma cell lines were the first cellular models where deletion of the chromosomal region containing PTEN was reported. These findings led to the identification of PTEN as a tumor suppressor gene (Li et al. 1997; Steck et al. 1997). It is reported that the majority of primary prostate cancers show loss of only one allele of PTEN, whereas homozygous inactivation of PTEN is generally associated with advanced cancer and metastasis (Gray et al. 1998).

Although human and mouse prostates are structurally dissimilar, prostate cancer progression in mice and humans is strikingly similar. In both species, epithelial hyperplasia is followed by low-grade prostatic intraepithelial neoplasia (PIN), which can progress to high-grade PIN. As the lesion becomes more neoplastic and aggressive, the prostate epithelium invades through the basement membrane into the surrounding stroma, thus establishing a localized yet invasive adenocarcinoma (De Marzo et al. 2003; Marandola et al. 2004).

In an effort to define the role of PTEN loss in prostate tumorigenesis, a series of Pten loss mouse models, the so called “hypomorphic Pten allelic series” (Pten heterozygous, Pten hypomorphic, and Pten conditional knock-out), have been generated (Trotman et al. 2003). Specific deletion of Pten in the prostate was achieved by crossing Pten loxP/loxP mice with Probasin-Cre (PB-Cre) transgenic mice. PB-Cre transgenic mice express Cre recombinase under the control of the ARR 2 Probasin promoter specifically in the prostate epithelium post-puberty (Wu et al. 2001). The generation of the “hypomorphic Pten allelic series” has revealed that the prostatic epithelium is exquisitely vulnerable to subtle variations of PTEN expression levels. For instance, loss of one allele of Pten is associated with the development of high-grade PIN with incomplete penetrance after a long latency (9 months), whereas when the level of Pten is reduced to ∼30% (hypomorphic mouse model), mice developed invasive prostatic adenocarcinoma albeit with incomplete penetrance (Trotman et al. 2003). Furthermore, complete loss of Pten results in the development of high-grade PIN (HG-PIN) as early as 8 weeks of age, together with the concomitant activation of cellular senescence response (see below) (Chen et al. 2005). HG-PIN lesions progress to invasive prostate cancer with complete penetrance at 6 months of age, once the senescence response has been evaded (Chen et al. 2005; Trotman et al. 2003). These analyses imply that (1) loss of PTEN is critical for prostate cancer initiation and that (2) the level of PTEN expression is inversely associated with prostate tumorigenesis.

Wang et al. also used PB-Cre transgenic mice to generate mice with conditional inactivation of Pten in the prostate, which also results in invasive prostate cancer (Wang et al. 2003). Additionally, these mice developed metastatic prostate cancer of the lymph nodes and lung, which is not observed in other mouse models of Pten conditional inactivation in the prostate (Abate-Shen et al. 2003; Chen et al. 2005; Wang et al. 2003). This may be due to the different genetic background strain of the mice, which is known to influence cancer susceptibility.

In a later report, by crossing Pten loxP/loxP mice with MMTV-Cre transgenic mice, Backman et al. inactivated Pten in the prostate during development (Backman et al. 2004). Deletion of Pten in the prostate before puberty resulted in the onset of neoplastic lesions at a very early time point, with mice displaying high-grade PIN by the age of 2 weeks at complete penetrance that frequently progressed to invasive adenocarcinomas by 7–14 weeks (Backman et al. 2004). These data show that, if Pten has already been deleted in the prostate during development, the incidence, penetrance, and progression of neoplasia are much greater than if Pten is lost during or after puberty.

In 2005, yet another model of complete Pten inactivation in the prostate was generated using Pten loxP/loxP mice crossed with prostate-specific antigen (PSA)-Cre transgenic mice (Ma et al. 2005c). The onset of prostatic neoplastic lesions is significantly delayed in these mice which show focal PIN at the age of 4–5 months. By 7–9 months, focal microinvasion was observed which progressed to frank invasive adenocarcinoma at 10–14 months (Ma et al. 2005c).

Recently, two groups have generated two mouse models where Pten deletion is temporally controlled through the use of Pten loxP/loxP mice crossed with tamoxifen-inducible Cre recombinase transgenic mice (Luchman et al. 2008; Ratnacaram et al. 2008). Like other models before, deletion of Pten results in the development of PIN lesions that later progress to invasive adenocarcinoma.

Together, models of conditional Pten inactivation in the prostate clearly demonstrate the sensitivity of the prostatic epithelium to alterations of Pten and they recapitulate the sequential stages of the human disease from PIN to invasive prostate cancer where time to progression is dictated solely by the developmental time of Pten excision and the remaining dose of functional Pten.

3.3 Breast

A characteristic feature of Cowden disease is the development of benign breast hamartomas that are accompanied by a higher risk of breast cancer. Although somatic PTEN mutations are detected only in a smaller fraction of breast cancer cases (Dahia 2000), LOH at the PTEN locus (10q23) is frequently found (40%) (Bose et al. 1998; Garcia et al. 1999). Furthermore, immunohistochemical studies suggest that loss of PTEN protein expression is a common event in breast cancer (33–48%), with strong correlation with lymph node metastasis, loss of estrogen receptor staining, and disease related death (Depowski et al. 2001; Perren et al. 1999). Thus, epigenetic mechanisms are hypothesized to be responsible for a number of cases in which PTEN levels are downregulated or even totally ablated in the absence of a detectable mutation.

The relevance of Pten in breast tumorigenesis was initially highlighted in the mouse model of Pten germline heterozygous loss generated by Stambolic et al. (2000). Female Pten +/− developed mammary tumors at incomplete penetrance, with most of them having features of well-differentiated adenocarcinoma. Similar to CS patients (Schrager et al. 1998), breast lesions in Pten +/− mice displayed marked proliferation of the stroma. The authors observed an increased penetrance of the breast tumors with age in Pten +/− mice thereby suggesting a requirement for additional hits for tumor progression in this tissue.

After that, to fully understand the role of Pten in breast tumorigenesis, in 2002 Li et al. crossed Pten loxp/loxp mice with transgenic mice expressing MMTV-Cre transgenes in order to achieve Pten deletion in the mammary epithelium (Li et al. 2002). The deletion of Pten in mammary epithelium triggered increased cell proliferation, hyper-branched ductal structure, precocious development, delayed involution and severely impaired apoptosis. Pten-deficient mammary epithelium also displayed remarkable neoplastic changes. Females with mammary-specific Pten deletion develop tumors as early as 2 months. Histological features of the tumors varied from benign fibroadenomas to pleiomorphic adenocarcinomas. Furthermore, immunohistochemistry analysis revealed up-regulation of cytokeratins 5 and 6 in these mice (Li et al. 2002). Interestingly, this finding nicely correlates with overexpression of these two cytokeratins in human breast tumors of the basal subtype (Sorlie et al. 2001). The basal subtype often occurs in patients with germline BRCA1 mutations and is associated with a poor prognosis. Importantly, it has been recently shown that heterozygous inactivation of Pten leads to the formation of basal-like mammary tumors in mice, and that loss of PTEN expression is significantly associated with this subtype of breast cancer in human sporadic and BRCA1-associated hereditary breast cancers (Saal et al. 2008). In addition, Saal and colleagues have identified frequent gross PTEN mutations, involving intragenic chromosome breaks, inversions, deletions and micro copy number aberrations, specifically in BRCA1-deficient tumors (Saal et al. 2008).

It has recently been shown that even a subtle reduction in Pten dose determines breast cancer susceptibility (Alimonti et al. 2010). Indeed, Pten hypomorphic mice, expressing 80% normal levels of Pten, develop a spectrum of tumors, with breast occurring at the highest penetrance (Alimonti et al 2010). Overall, all these observations underscore the essential role of PTEN during normal mammary gland development and in suppressing breast cancer formation.

4 In Vivo Deconstruction of the PI3K-AKT-mTOR Axis

The numerous mouse models generated to study the PI3K-AKT-mTOR pathway have been valuable tools to shed light on the role of various components of the PI3K signaling cascade in disease and tumorigenesis. Overall, these mouse models have defined the mTOR pathway as a crucial converging node downstream PI3K-AKT signals required for oncogenic transformation driven by loss of PTEN.

4.1 PI3K-PDK-AKT

The PI3K (phosphatidylinositol-3-kinase) pathway starts at the plasma membrane where the binding of ligands to the growth factor receptor tyrosine kinases activate PI3K, which phosphorylates PIP2 to produce PIP3, thereby directly antagonizing PTEN (Klinghoffer et al. 1996).

Class I PI3K contains four p110 isoforms, α, β, γ, and δ. The association between p110α and tumorigenesis is well established and has been corroborated by the occurrence of gain of function p110α mutations in human cancer (Samuels and Velculescu 2004). Recently, the p110β isoform has also been connected to oncogenesis (Ciraolo et al. 2008; Jia et al. 2008). With regards to PTEN-loss driven cancer, conditional inactivation of PIK3CB, the gene encoding p110β, blocked prostate tumorigenesis mediated by loss of PTEN whereas prostate-specific knockout of the α-isoform did not alter tumor formation. The observations of Jia and co-workers identify a previously unknown role for p110β in cancer, specifically in PTEN mutated tumors. These studies collectively suggest that p110β may represent a potential “druggable” target, specially in PTEN null cancers.

PI3K signaling induces a series of growth-promoting events through the activation of the protein kinases PDK1 and AKT, which directly bind to and are activated by PIP3 (Alessi et al. 1997; Currie et al. 1999). Upon PIP3 binding, PDK1 induces AKT kinase activity 30-fold by phosphorylating it on residue T308 in addition to the phosphorylation of numerous other target proteins within the T loop (such as Serum and Glucocorticoid-regulated kinases, SGK) enabling their activation (Alessi et al. 1997). AKT, in turn, phosphorylates multiple targets to activate the cell cycle, prevent apoptosis and trigger cellular growth (Manning and Cantley 2007). In vivo studies in the mice have genetically highlighted the important epistasis of Pten and Pdk1 and Akt. Compound mutant mice have provided clear genetic evidence for the roles of Akt and Pdk1 as mediators of cancer phenotypes identified upon heterozygous Pten-loss in mice. Specifically, Akt1 deficiency suppresses tumor development in Pten +/− mice (Chen et al. 2006). Similarly, the hypomorphic expression of Pdk1 (levels that are 80–90% reduced compared with normal) inhibits tumor formation in Pten +/− mice (Bayascas et al. 2005). Therefore, these mouse models have validated AKT1 and PDK1 as critical players in mediating tumorigenesis upon PTEN-loss.

4.2 TSC1/2-Rheb-mTOR

Among the many downstream targets of AKT, the mammalian target of rapamycin (mTOR) has been demonstrated to be an essential effector in promoting cell proliferation and susceptibility to oncogenic transformation. mTOR is a serine/threonine kinase that regulates protein synthesis, cell growth, and proliferation in response to pleiotropic inputs including growth factors, nutrients, energy, and stress (Wullschleger et al. 2006). mTOR differentially regulates PI3K/AKT signaling by acting as a key component of two multiprotein complexes: mTOR complex 1 (mTORC1) which is activated downstream of AKT, and mTOR complex 2 (mTORC2) which has been demonstrated to phosphorylate AKT on Ser 473 (Sarbassov et al. 2005). This modification, in conjunction with the phosphorylation on Thr308 by PDK1 triggers full activation of AKT in response to mitogenic stimuli (Sarbassov et al. 2005). Moreover, in many cell types, mTORC1 has been reported to elicit a negative feedback regulation on the PI3K pathway through the ability of its downstream target ribosomal S6 kinase 1 (S6K1) to inhibit IRS-1 (reviewed in Guertin and Sabatini 2007). The elusive cross-talk between the AKT and mTOR pathways was uncovered by the finding that tuberous sclerosis complex 1 (TSC1) and 2 (TSC2) negatively regulates mTORC1 (Gao et al. 2002; Tapon et al. 2001). These studies demonstrated that AKT phosphorylates and inactivates the TSC1/TSC2 complex, and as a consequence results in mTORC1 activation (Jaeschke et al. 2002; Tee et al. 2002). Specifically, Ras homologue enriched in brain (Rheb), a small guanosine triphosphate (GTP)-binding protein, was discovered as a novel substrate for TSC2, which could also lead to the activation of mTOR (Garami et al. 2003; Inoki et al. 2003; Zhang et al. 2003). TSC2 was shown to display a GTPase activating protein (GAP) activity towards the Rheb GTPase; this event stimulates the intrinsic GTP-hydrolysis activity of Rheb to promote its transition from an active GTP-bound to an inactive guanosine diphosphate (GDP)-bound form (Garami et al. 2003; Zhang et al. 2003). Conversely, inactivation of the TSC1/TSC2 complex by AKT phosphorylation, results in GTP loading and activation of Rheb, which ultimately promotes the activation of mTORC1. AKT also promotes mTORC1 activity through phosphorylation of PRAS40, which prevents its inhibitory function on mTORC1 (reviewed in Guertin and Sabatini 2007). When active, mTORC1 promotes cell growth through phosphorylation of various regulators of translation including the well-characterized ribosomal S6K1 which activates the S6 ribosomal protein (S6), and the eukaryotic translation initiation factor 4E binding protein 1 (4EBP1) which leads to its uncoupling from the elongation initiation factor 4E (eIF4E; Wullschleger et al. 2006).

Several mouse models have validated the importance of the mTOR pathway in further promoting tumorigenesis driven by Pten loss. For instance, oncogenic events proximal to mTOR activation such as Tsc2 heterozygosity as well as Rheb overexpression cooperate with Pten haploinsufficiency to accelerate tumorigenesis (Ma et al. 2005b; Manning et al. 2005; Nardella et al. 2008) (Fig. 1a, b). A definitive proof of principle for a major role of mTOR in prostate tumorigenesis driven by Pten-loss was uncovered by the conditional inactivation of mTor in Pten null prostates. In these prostates, mTor deletion markedly suppressed the tumor initiation and progression observed in Pten-null mice (Nardella et al. 2009) (Fig. 1c). These findings are corroborated by the work of Guertin et al, which showed that Rictor, one of the components of the mTORC2 multiprotein complex, is also required for Pten-loss induced tumorigenesis in the mice (Guertin et al. 2009).

Fig. 1
figure 1

Analysis of the contribution of the PI3K pathway to Pten loss-driven disease, through the combination of genetic events in the mouse. Pten-loss driven disease is accelerated by Tsc2 heterozygous loss (a) as well as Rheb transgenic expression (b), whereas it is opposed by loss of mTOR (c)

Full size image

Collectively, these data have important therapeutic application in the treatment of cancer triggered by loss of PTEN, since mTOR is a kinase with activity that is amenable to pharmacological inhibition. Indeed, efforts have been placed to develop and improve upon drugs that inhibit mTOR activity. The first generation of mTOR inhibitors directed solely against mTORC1, such as Rapamycin have had limited success in the clinic. We now await the development and testing of drugs targeting both mTORC1 and mTORC2 for the treatment of tumors triggered by PTEN deficiency and aberrant PI3K-AKT-mTOR signaling.

5 PTEN Network: Linking the PI3K Signaling Cascade to Other Oncogenic Pathways Through In Vivo Genetic Analysis

It is now well established that the PI3K pathway is intimately linked to several other oncogenic events including activated MAPK signaling, ETS-related gene (ERG) overexpression, and loss of the tumor suppressor gene p53. Below, we will describe the efforts to generate faithful mouse models of human cancers which recapitulate critical cooperative events identified in human cancer.

5.1 PTEN-MAPK Pathway

The RAS oncogene and the PTEN tumor suppressor are upstream of two of the most predominant oncogenic signaling pathways, MAPK and PI3K, respectively (Dhillon et al. 2007; Engelman et al. 2006). The signaling emanating from these two pathways is however complicated by a remarkable number of interconnections (Carracedo et al. 2008).

Ras is the upstream regulator of the MAPK pathway and frequently activated in cancer through mutations (G12D, G12V) which are sufficient to initiate cancer in the mouse (Fisher et al. 2001; Johnson et al. 2001). Ras mutations lead to hyperactivation of the MAPK pathway, which cross-talks with the PI3K cascade through the regulation of common targets such as BAD and TSC2 (Datta et al. 1996; Fang et al. 1999; Gupta et al. 2007; Ma et al. 2005a). In addition, activated Ras leads to loss of PTEN expression through c-Jun-mediated transcriptional events (Vasudevan et al. 2007). Consistent with this notion, mutations in RAS and PTEN in cancer tend to be mutually exclusive RAS mutations are prevalent in pancreatic, lung, and colon cancers, but not in glioblastomas whereas the opposite is true for PTEN mutation (Liu et al. 1997; Simpson and Parsons 2001).

Downstream of RAS, RAF (BRAF, CRAF, and ARAF) serine/threonine protein kinases regulate MAPK signaling (Balmanno and Cook 2009; Moodie et al. 1993). BRAF is also frequently found mutated in cancer (principally V600E mutation), most frequently in melanoma where BRAF mutation is observed in 50–70% of cell lines and tumors, and does not overlap with RAS mutations (Halilovic and Solit 2008). Unlike RAS, concomitant genetic alterations in PTEN and BRAF have been found in melanoma and shown to be cooperative events in the mouse (see below), therefore reinforcing the complexity of the interaction between components within these two pathways (PTEN-PI3K and MAPK) in cancer.

Combining PTEN and BRAF mutations to model human metastatic melanoma in mouse. Mutant activated BRAF (BRAFV600E) can induce senescence in cultured melanocytes providing an explanation for the high frequency of BRAF mutations in benign nevi (Denoyelle et al. 2006; Dhomen et al. 2009; Michaloglou et al. 2005; Taube et al. 2009). Hence, overcoming oncogene-induced senescence may be critical for melanomagenesis. Progression to malignant melanoma is invariably accompanied by silencing of one or more tumor suppressor genes, most commonly PTEN or CDKN2A (Chin et al. 2006; Garraway et al. 2005). Additionally, the combination of mutated BRAF and silencing of PTEN expression is observed in 20% of human melanomas (Backman et al. 2004). While Pten-loss or BRAFV600E activation alone did not have a dramatic consequence for melanoma onset and progression in mice, compound BRafV600E-Pten-Null mice succumbed to an aggressive form of metastatic melanoma (Dankort et al. 2009) (Fig. 2a). Of note, combinatorial inhibition of MAPK and mTORC1 led to the reduction of melanoma formation, implying that this pharmacological approach may be an effective therapeutic avenue in the treatment of this type of cancer. Therefore, the Pten −/−;BRafV600E melanoma model, together with other recently developed genetically modified melanoma mouse models (Goel et al. 2009), represent an invaluable tool for modeling of melanoma in the mouse and the evaluation of therapeutic approaches in the treatment of this deadly disease. Importantly, whether PTEN is required for the bypass and senescence induced by BRAF mutation in melanoma remains to be determined, and might add complexity to the already diverse tissue-specific outcomes of Pten-loss in vivo (see below).

Fig. 2
figure 2

Genetic interventions in the mouse that have allowed the generation of faithful mouse models resembling human disease. Faithful mouse models of human melanoma (a), prostate cancer onset and progression (b), glioblastoma multiforme, bladder cancer, and advanced prostate cancer (c)

Full size image

5.2 Pten and Transcriptional Regulators: Erg and Myc

As mentioned above, PTEN is frequently lost or downregulated in prostate cancer (Salmena et al. 2008). Recently, the translocation of an ETS transcription factor gene (ERG or ETV1) to the TMPRSS2 gene promoter region, which contains androgen responsive elements, has been identified in prostate tumors (Tomlins et al. 2005). TMPRSS2-ERG is the first recurrent translocation event to be described in human tumors. It occurs in approximately 40% of prostate tumors and results in an aberrant androgen-regulated expression of ERG (Perner et al. 2007). In mice, transgenic ERG expression in the prostate leads to an unremarkable phenotype (Carver et al. 2009a), suggesting that ERG overexpression is not an initiating event in prostate cancer, in line with the notion that TMPRSS22-ERG translocation is rarely found in early lesions (Balmanno and Cook 2009; Carver et al. 2009b).

On the other hand, genetic lesions such as amplification and polymorhisms at 8q24, where c-MYC is located, are robustly associated with prostate cancer risk (Amundadottir et al. 2006; Bubendorf et al. 1999; El Gedaily et al. 2001; Gudmundsson et al. 2007; Haiman et al. 2007; Qian et al. 1997; Tsuchiya et al. 2002; Witte 2007; Yeager et al. 2007). Furthermore, recent studies have shown that over-expression of c-MYC is not restricted to advanced/late prostate cancer lesions, but occurs also in early lesions (Gurel et al. 2008). Additional studies are needed in order to determine the mechanisms underlying the frequent over-expression of c-MYC in prostate cancer, and to assess whether this is also a consequence of 8q24-polymorphism-relates transcriptional events or may in fact be the result of post-transcriptional mechanisms (Pomerantz et al. 2009).

Seeking a faithful model of prostate cancer in the mouse: Combinatorial mutation of PTEN with ERG and MYC. Loss of PTEN is frequently accompanied by the translocation of TMPRSS22-ERG in prostate cancer (Balmanno and Cook 2009; Carver et al. 2009a). Modeling the compound loss of Pten and Erg overexpression (Probasin-transgenic Erg, ErgTG) in the mouse prostate has uncovered a strong cooperativity between these two genetic events (Balmanno and Cook 2009; Carver et al. 2009a) (Fig. 2b). Whereas Pten heterozygosity leads to high-grade prostate intraepithelial neoplasia (HGPIN) lesions starting at the age of 9 months, compound Pten +/−-ErgTG mutants develop HGPIN by 2 months of age, which progresses to invasive cancer by the age of 6 months.

c-MYC transgenic expression in the prostate leads to hyperproliferation and PIN (Ellwood-Yen et al. 2003; Kim et al. 2009; Zhang et al. 2000). Moreover, Pten-heterozygous loss cooperated with c-MYC to induce high-grade prostatic intraepithelial neoplasia (HGPIN)/cancer lesions, which harbor loss of the wild type Pten allele (Kim et al. 2009) (Fig. 2b).

Overall, Pten +/−-ErgTG and Pten +/−; cMYCTG prostate cancer models likely represents the most faithful models of prostate cancer initiation and progression to date, since they recapitulate the precise sequence of mutagenic events occurring in a large fraction of human prostate cancers. However, unlike human prostate cancer, which exhibits a highly metastatic tropism to the bone in later stages, mouse models have thus far failed to recapitulate these late events faithfully. Hence, additional genetic studies and novel combinatorial efforts in the mouse are required to generate better models of human prostate cancer progression to metastasis.

5.3 Pten/p53

In terms of overall frequency, p53 is undoubtedly the most frequently mutated tumor suppressor gene in human cancers (Levine et al. 2004; Vogelstein et al. 2000) with PTEN following in second (Cantley and Neel 1999; Simpson and Parsons 2001). The spectrum of human cancers associated with p53 and PTEN mutation are very different. (Fujisawa et al. 2000; Kato et al. 2000; Koul et al. 2002; Kurose et al. 2002). Mutations of p53 occur at high frequencies in lung, colon and breast cancers, whereas PTEN mutations are mostly found in glioblastoma, endometrial cancer, malignant melanoma, and prostate cancer. However, compound loss of PTEN and p53 has been reported in glioblastoma (Han et al. 2008; Salmena et al. 2008), bladder cancer (Puzio-Kuter et al. 2009) and advanced/metastatic prostate cancer (Chen et al. 2009).

Mouse models of combined loss of Pten and p53 in glioblastoma, bladder and prostate cancer. In glioblastoma, the impact of combined Pten and Trp53 loss has been recently reported in a model of concomitant deletion of Pten (in heterozygosity) and Trp53 in the GFAP+ cell lineage. These mice develop an acute and lethal form of glioblastoma multiforme (Fig. 2c). Importantly, this mouse model displays features reminiscent of the pathological lesions observed in human glioblastoma multiforme (Zheng et al. 2008). Mechanistically, compound loss of Pten and Tp53 in neural stem cells leads to increased cell renewal and decreased differentiation in a MYC-dependent fashion (Zheng et al. 2008).

Bladder cancer is a major cause of cancer morbidity and mortality (Jemal et al. 2005). Combined p53 and PTEN losses have been identified in invasive bladder cancer and are reportedly causal factors that predict poor outcome (Puzio-Kuter et al. 2009). Indeed, combined loss of Trp53 and Pten in mice results in lesions with characteristics of human carcinoma in situ with complete penetrance at 6 months of age(Puzio-Kuter et al. 2009) (Fig. 2c).

In prostate cancer, partial loss of the PTEN tumor suppressor gene is a prevalent event (see above). However, complete loss of PTEN is infrequent in early lesions and is restricted to advanced cancers. Through the analysis of acute complete conditional loss of Pten in the prostatic epithelium, we found that one plausible explanation for this phenomenon is the fact that complete acute loss of Pten elicits a p53-dependent failsafe senescence response which opposes tumor progression, (Chen et al. 2005). In agreement with this notion and the fact that in human prostate cancer p53 loss is a late event observed prevalently in advanced lesions, compound loss of Pten and Trp53 in the mouse prostate leads to a lethal form of advanced prostate cancer where the senescence response has been evaded (Chen et al. 2005). In spite of the local aggressiveness of these tumors, Pten/Trp53 compound mutants, surprisingly, do not develop metastasic prostate cancer (Chen et al. 2005).

Although p53 and PTEN represent the most frequently lost of all tumor suppressors, further studies are required to precisely determine the frequency and the timing of their loss, and the specific tissues where it occurs. However, modeling these mutations in the mouse has already allowed the generation of faithful models of advanced prostate, bladder cancers, and glioblastoma that will prove extremely valuable to study the biology of these cancers and to test novel therapeutic modalities in preclinical studies.

6 Context-Dependent Differential Outcomes Triggered by Loss of PTEN

The large number of studies reporting phenotypes of Pten conditional knockout mice has highlighted the function of Pten in different cell and tissue types.

Although the PI3K pathway is ubiquitous, PTEN-mediated regulation of the PI3K/AKT pathway results in cell context-dependent outcomes such as cell size, proliferation, survival and senescence. Furthermore, there is a growing body of evidence suggesting that differential outcomes can be due to differential timing of Pten loss in specific stages of the development within the same tissue.

Cell size. Conditional knock-out mice show that loss of Pten may influence cell size or cell number depending on the specific context. The brain is an example of where selective deletion of Pten in specific cell types such as granule neurons of the cerebellum and dentate gyrus, cerebellar precursor cells and Purkinje neurons results in a cell-autonomous size increase in Pten-deficient cells (Backman et al. 2001; Kwon et al. 2001; Marino et al. 2002).

Cell number. In other settings, the consequences of PTEN loss determines changes in cell number, due to the combined effects of proliferation and cell survival, rather than aberrant cell size. Conditional deletion of Pten caused increased proliferation, decreased apoptosis and tumorigenesis, as exemplified in keratinocytes (Backman et al. 2004; Suzuki et al. 2004), prostatic epithelium (Backman et al. 2004; Wang et al. 2003), mammary epithelium (Li et al. 2002), germ cells (Kimura et al. 2003) and hepatocytes (Horie et al. 2004). Overall, these examples suggest that the cellular context strongly influences the specific outcome of PTEN deficiency.

Cellular senescence. PTEN does not exert its tumor suppressive function in isolation, but cross-talks extensively with other tumor suppressors, including p53. Therefore, the status of the p19ARF/p53 network in the different tissues can also affect the differential context-dependent outcomes dictated by the loss of Pten.

This is nicely exemplified by the response of the prostatic epithelium upon complete inactivation of Pten. Surprisingly, Chen et al. showed that complete acute loss of PTen in the prostate did not provide a proliferative advantage as would be expected, but instead promoted a strong p53-dependent senescence response that opposed tumor progression (Chen et al. 2005). As predicted from these findings, combined inactivation of Pten and Trp53 leads to unconstrained tumor growth as demonstrated by the generation of massive invasive prostate tumors. This implies that complete ablation of PTEN can be detrimental to tumor growth in the absence of p53 mutations and highlights the importance of haploinsufficiency or partial PTEN impairment in tumor progression. Clinically, these findings provide an explanation as to why complete PTEN loss is not frequently observed at cancer presentation.

Yilmaz et al. (2006) suggested that the hematopoietic stem cell (HSC) compartment may also be a tissue where complete loss of Pten triggers a senescence response (Yilmaz et al. 2006). Deletion of Pten in the adult HSCs results in a different outcome in normal hematopoietic stem cells versus leukemia-initiating cells (Yilmaz et al. 2006). Specifically, the authors show that deletion of Pten results in the generation of leukemic stem cells and concomitant depletion of normal HSCs. The mechanism responsible for the depletion of Pten-deficient HSCs remains to be elucidated but it has been speculated that Pten deficiency induces a senescence response in HSCs whereas the leukemia-initiating cells might acquire secondary mutations that inactivate the senescence response (Yilmaz et al. 2006).

Differential outcomes. In certain tissues, p53 mutations are not required for tumors to progress upon Pten loss because p53 is repressed through different mechanisms, and consequently cellular senescence is not observed. This is well exemplified by the deletion of Pten in smooth muscle, which results in the development of leiomyosarcomas with very high penetrance (80%) (Hernando et al. 2007). In response to loss of Pten the authors observed a substantial upregulation of p19Arf in the sarcoma cells, without concomitant induction of p53. In addition, they observed no evidence of cellular senescence either in the hyperplastic tissue or in the sarcomas. However, marked Mdm2 levels in leiomyosarcoma cells compared to normal smooth muscle of Pten-null mice, which kept p53 functionally repressed, thus reducing the need for p53 mutations usually required for tumor progression. In sum, Mdm2 stabilization promoted by Akt phosphorylation seems to prevail over Mdm2 inhibition by p19Arf in Pten-null smooth muscle cells, resulting in p53 functional inactivation and thereby tumor development (Hernando et al. 2007).

7 Conclusion

Tremendous technology advances have allowed us to gain powerful insight into the molecular and genetic determinants that drive cancer. Mouse models have been at the forefront of this revolution of information that has allowed us to faithfully recapitulate the features of tumor initiation and progression observed in human cancer. Mouse models of Pten loss have shed light on the critical roles of Pten in tumor suppression, specifically as a regulator of cell size, proliferation rate, and failsafe responses, such as senescence, in specific tissues. As one of the “most modeled” of all human cancer genes, Pten mouse models are exemplary of the power of genetic modeling and the success that can be achieved through such studies.

Further insight into the function of PTEN genetic mutations will rely upon the generation of specific point mutations knock-in mice models, which can inform us not only about canonical PTEN function, but the ever-increasing role of PI3K and AKT independent functions of PTEN. These models will provide further understanding of the regulatory mechanisms that affect the role of this protein in normal development and tumorigenesis.

Translation of the information acquired in mice has been and will be extremely useful for the preclinical evaluation of targeted therapeutic anti-cancer agents thereby dramatically improving our ability to cure this and other diseases.