PI3K/mTORC1 activation in hamartoma syndromes

PI3K

PI3Ks are a family of lipid kinases that catalyze the addition of phosphate molecule specifically to the 3′-OH position of the inositol ring of phosphoinositides converting phosphatidylinositol-4,5-biphosphate (PIP₂) to phosphatidylinositol-3,4,5-trisphosphate (PIP₃) at the membrane.^30–33 PI3K is activated by multiple inputs such as growth factors, insulin, cytokines, cell-cell and cell-matrix interactions provided by both receptor and non-receptor tyrosine kinases and adhesion molecules.³⁴ PI3K isoforms can be divided into three classes based on their structure and substrate specificity.^35,36 Class IA PI3Ks are cytoplasmic heterodimers composed of a 110-kDa (p110α, β or δ) catalytic subunit and an 85-kDa (p85, p55 or p50) adaptor protein. Catalytic subunits p110α and p110β are ubiquitously expressed in mammalian cells. Catalytic subunit p110δ is expressed predominantly in lymphocytes and lymphoid tissues and therefore, may play a role in PI3K-mediated signaling of immune responses.³⁷ Class IA isoforms are mainly activated by receptor and nonreceptor tyrosine kinases while class IB p110γ, first identified during the screening of human bone marrow cDNA, is activated by Gβγ subunits of G protein coupled receptors (GPCRs)³⁸ and by T cell receptor tyrosine kinase activated protein.³⁹ Class II isoforms are mainly associated with the phospholipid membranes and are present in the endoplasmic reticulum and Golgi apparatus.⁴⁰ Class III isoforms, structurally related to yeast vesicular sorting protein Vps34p,⁴¹ are mammalian homologues that use only membrane phosphatidylinositol as a substrate and generate phosphatidylinositol-3-monophosphate.

Identification of PI3K isoforms and elucidation of their cellular functions revealed the critical role of PI3K signaling in normal cell functions; meanwhile, growing evidence also linked dysregulation of PI3K signaling to cancer. Mutations in the PI3KCA gene, which is located in common amplicon at 3q26 and encodes the p110α, catalytic subunit of PI3K, were found in more then 30% of various solid tumor types including ovary, cervix and head and neck cancers.⁴² Functional analysis of these mutations showed that they increased PI3K enzymatic activity and stimulated Akt signaling thus allowing growth factor-independent growth as well as increased cell migration and metastasis.³⁵ Somatic mutations in the PI3KR1 gene encoding the p85α regulatory subunit of PI3K leading to the PI3K constitutive activation, had been detected in human ovarian and colon tumors and human lymphoma cells.^43,44 While mutation of PI3Ks has been predominantly associated with different types of cancers,^22,36 specific PI3K mutations had not been linked to hamartoma syndromes. However, in hamartoma syndromes PI3K signaling is activated due to PTEN inactivating mutations, which makes PI3K a critical target for potential therapeutic intervention in hamartoma syndromes, which will be discussed below.

mTORC1

The relevance of mTORC1 in hamartoma syndromes arises from the discovery of the constitutive activation of this canonical pathway in pulmonary LAM¹² and TSC^45,46 due to the inactivating mutations of tumor suppressor complex TSC1/TSC2. Because mTORC1 is a downstream effector of PI3K and forms a canonical PI3K-mTORC1 signaling cascade, these findings also underscore the impact that mTORC1 over-activation has on the PTEN-associated hamartoma syndromes and identified mTORC1 as a molecular target for cancer treatment. mTOR, a highly conserved serine-threonine kinase, was identified as a molecular target of rapamycin, a macrolide extracted from the fungi Streptomyces hydroscopicus.⁴⁷ Rapamycin forms a cytosolic complex with FK506 binding protein 12 (FKBP12), which inhibits the catalytic activity of mTOR.^48,49 Specific inhibition of mTORC1 activity by rapamycin and the fact that mTORC1 is constitutively activated not only in hamartoma syndromes but in many different types of cancer, led to multiple rapamycin clinical trials; results of some of these trials have already been published^6,50,51 and are discussed in this review.

mTORC1 consists of the serine-threonine kinase mTOR, regulatory-associated protein of mTOR (raptor),^52,53 mLST8 (also known as GβL)⁵⁴ and the proline-rich Akt substrate of 40 kDa (PRAS40) (Fig. 1). In mTORC1, mTOR provides mTORC1 the catalytic activity, and raptor functions as a positive regulator of mTOR activity and expression.^52,53,55 The activity of raptor towards mTOR can be regulated by p90 ribosomal protein S6 kinase-dependent phosphorylation⁵⁶ and energy-stress conditions through AMP-activated protein kinase (AMPK).⁵⁷ PRAS40 acts as an inhibitor of mTOR kinase activity in an Akt-dependent phosphorylation manner.^58–60 Interestingly, some studies have demonstrate that PRAS40 can also be phosphorylated by activated mTOR at Akt-independent sites, which promotes the release of PRAS40 from mTORC1 and increases mTORC1 activity.^61,62 The function of mLST8 in mTORC1 remains to be established.⁶³

Active mTORC1 directly phosphorylates S6K1 and 4E-BP1, which leads to SK61 activation and 4E-BP1 inhibition followed by protein translation, ribosome biogenesis and cell cycle progression.^64,65 mTORC1 mediates inputs from three major classes of molecules: growth factors, amino acids and ATP, and is a master regulator of cell growth and metabolism.⁶⁶ In normal cells, mTORC1 signaling is tightly controlled; however, in hamartoma syndromes and in cancer, mutations of the pathway result in benign and neoplastic growth.²²

mTORC1 is negatively regulated by tumor suppressor complex TSC1/TSC2. TSC2 functions as a GTPase activating protein (GAP) for the small GTPase Rheb (Ras-related small G protein Ras homologue enriched in brain). Growth factor- or insulin-induced activation of PI3K, which converts PIP₂ to PIP₃ at the membrane,^30–33 leads to Akt activation which, in turn, leads to the Akt-dependent inactivating phosphorylation of TSC2. It inhibits TSC2 Rheb GAP activity followed by the activation of Rheb, which directly binds to and activates mTORC1 (Fig. 1).^67–71 Insulin-induced activation of Akt, in addition to TSC1/TSC2 inhibition, also leads to the direct phosphorylation of PRAS40, which is thus unbound from mTORC1 and increases mTORC1 activity.^58–60 In addition to FKBP12, another member of the FK506-binding protein family, FKBP38, a mitochondrial membrane protein, was identified as an inhibitor of mTORC1 activation.⁷² Apparently, in a quiescent state, FKBP38 is bound to mTOR; then, upon cell growth signaling, GTP-bound small GTPase Rheb (Fig. 1) binds to FKBP38 and induces the release and activation of mTOR.⁷³ FKBP38 binds to mTOR and inhibits its activity in a manner similar to that of the FKBP12-rapamycin complex. Interestingly, FKBP38 binds to a region of mTOR that apparently overlaps with another mTOR region that binds the FKBP12-rapamycin complex, suggesting that the later may preclude binding of FKBP38 to mTOR.⁷³ However, the precise role of FKBP38 in the canonical mTORC1 signaling pathway remains to be established. The activity of mTOR can also be regulated by targeted ubiquitination and consequent degradation by binding to the novel tumor suppressor protein F-box and WD repeat domain containing 7 (FBXW7).⁷⁴ Interestingly, tumor cells with loss or mutation of FBXW7 are particularly sensitive to rapamycin treatment.⁷⁴

In addition to the canonical rapamycin-sensitive mTORC1 signaling, mTOR also forms a physically and functionally distinct complex—the rapamycin-insensitive mTORC2,⁷⁵ consisting of mTOR, rictor,⁷⁵ GβL⁷⁶ and mSIN1 (mammalian stress-activated protein kinase-interaction protein 1).⁷⁷ In vivo⁷⁸ and in vitro⁷⁹ studies established the role of mTORC2 as the long-sought phosphoinositide-dependent kinase 2 (PDK2), which phosphorylates Akt at Ser473 (Fig. 1).⁷⁹

TSC1/TSC2

Tumor suppressors TSC1 and TSC2 form a physical and functional complex, in which TSC1 functions as the regulatory component stabilizing TSC2, and facilitating the TSC2 catalytic function as a GAP for the small GTPase Rheb, a negative regulator of mTORC1. The role of TSC1/TSC2 in regulating the mTORC1 signaling pathway has been intensively studied, and it has been clearly demonstrated that the TSC1/TSC2 complex is a key regulator of mTORC1 signaling. Comprehensive reviews of the manner in which dysregulation of TSC1/TSC2-mTORC1 modulates intracellular signaling, for example, the negative feedback inhibition of insulin signaling and Akt activation, are published in refs.^5,21,80,81

TSC1/TSC2 activity is regulated by the PI3K signaling cascade, a highly connected and conserved signal transduction network activated by growth-promoting signals that regulate cell proliferation and survival.^23–26 PI3K-Akt regulates the activity of TSC1/TSC2 due to Akt-dependent TSC2 phosphorylation at multiple sites (Fig. 1).⁵ Interestingly, while the multiple Akt-dependent sites that inactivate TSC2 Rheb GAP activity leading to mTORC1 activation have been identified, the physiological significance of Akt-dependent TSC2 phosphorylation have not been completely elucidated. Conflicting evidence comes from studies in Drosophila showing that mutation of Akt phosphorylation sites on TSC2 neither prevented S6 activation by insulin nor affected Drosophila development.⁸² Rheb activity is also positively regulated by the guanine nucleotide exchange factor activity of translationally controlled tumor protein (TCTP);⁸³ TCTP appears to act downstream or parallel to TSC1/TSC2. The precise role of TCTP in growth-promoting activation of Rheb and mTORC1 in mammalians remains to be elucidated.⁸⁴

TSC1/TSC2 activity is also negatively regulated by extracellular signal-regulated kinases (ERK). ERK-dependent TSC2 phosphorylation leads to dissociation of the TSC1 and TSC2 complex, which results in mTOR/S6K1 activation, increased cell proliferation and transformation.⁸⁵

In addition to the growth-promoting signaling mediated by PI3K-Akt and ERK, TSC1/TSC2 activity is regulated by cellular energy, that is, the availability of glucose and ATP. Energy starvation, including glucose deprivation and other stresses, such as attenuation of ATP synthesis during hypoxia or increased ATP consumption during physical exertion, results in the activation of AMPK by an upstream kinase LKB1.^86,87 Activated AMPK directly phosphorylates TSC2; this is then followed by inhibition of mTORC1 leading to inhibition of translation and attenuation of protein synthesis.^88,89 Recent data, however, indicates that, under energy stress conditions, AMPK may regulate mTORC1 activity in a TSC2-independent manner via the direct phosphorylation of raptor, which leads to mTORC1 inhibition and cell cycle arrest.⁹⁰ Additionally, genotoxic stress negatively regulates TSC1/TSC2-mTORC1 signaling in a p53-AMPK-dependent manner.⁹¹ Genotoxic stress activation of p53 leads to the expression of sestrin1 and sestrin2, two novel proteins, which activate AMPK leading to AMPK-dependent inhibiting phosphorylation of TSC2 thereby inhibiting mTORC1 and consequently cell growth and proliferation.⁹²

AMPK phosphorylation primes TSC2 for its subsequent phosphorylation by glycogen synthase kinase 3 (GSK3).⁹³ GSK3 activity is negatively regulated by the Wnt signaling pathway, which plays a critical role in cellular proliferation, especially during development.⁵⁷ Wnt-dependent activation of S6K1 requires the inhibition of GSK3 and TSC1/TSC2 activity.⁹³ Inhibition of GSK3 activity by Wnt is specific and distinct from its inhibition by Akt and RSK1.⁹⁴ Thus, TSC2 integrates growth, energy and Wnt signaling in coordinated inhibition of mTORC1 signaling, protein translation and cell growth.

Mutational inactivation of tumor suppressor proteins TSC1 or TSC2 are linked, respectively, to tumor development and the pathophysiology in hamartoma syndrome TSC.^5,10–12 Positional cloning and linkage analysis in multigenerational families with TSC were used to map both TSC1 and TSC2 genes.^10,11 Patients with TSC develop hamartomas and benign tumors in the brain, heart and kidney; they can exhibit cognitive defects, epilepsy and autism.^3,95

At the same time, another disease, pulmonary lymphangioleiomyomatosis (LAM), that could be sporadic or manifested in association with TSC, is also linked to the TSC1 and TSC2 genes.^4,5,12 Patients with LAM develop lung tumors characterized by abnormal smooth muscle-like cell growth, cystic destruction of the lungs, and loss of pulmonary function.^1,2 Extensive genetic analysis of TSC1 or TSC2 genes in TSC revealed a wide spectrum of mutations; however, no particular region was linked to higher mutation rate.³ Importantly, mutations of TSC2 account for more than 60% of the cases, and are associated with the severe clinical TSC phenotype.^3,5 Loss of function mutations or loss of heterozygosity (LOH) of either TSC1 or TSC2 lead to the constitutive activation of the mTORC1 signaling pathway and abnormal cell growth in TSC hamartoma syndrome and pulmonary LAM.^5,12,96,97

In addition to constitutive mTORC1-S6K1 activation, TSC1/ TSC2 loss of function causes the deregulation of other mTORC1-dependent and independent signaling pathways. It was shown that TSC2-related upregulation of mTORC1 activity causes endoplasmic reticulum stress in TSC2-deficient cell and tumor models⁹⁸ and deregulated glucose transporter Glut4 trafficking.⁹⁹ Additionally, expression of the platelet-derived growth factor (PDGF) receptor, one of the targets of negative feedback regulation in cells with activated mTORC1, is suppressed in TSC2-null cells.¹⁰⁰ Furthermore, constitutive mTORC1 activation due to loss of TSC2 amplifies p53 activation by stimulating p53 translation, which results in the increased accumulation of p53 and apoptosis in response to stress conditions.¹⁰¹ Importantly, loss of TSC2 function also leads to mTORC1-independent constitutive activation of the small GTPase RhoA, a well known positive regulator of cell proliferation and motility, in LAM¹⁰² and TSC2-null ELT3 cells.^103,104 While additional studies are needed to establish the contribution of ER stress and Rho GTPase in the mTORC1-related hamartomas, it is possible that some of these TSC1/TSC2-related signaling abnormalities may contribute to benign tumor formation.

PTEN

Tumor suppressor PTEN negatively regulates PI3K-mTORC1 signaling by specifically dephosphorylating the 3′-OH position of the inositol ring of the lipid second messengers PIP₃ (Fig. 1).^105–107 Loss of tumor suppressor PTEN leads to upregulation of the PI3K signaling pathway which results in the elevation of PIP₃ levels and contributes to oncogenesis (Fig. 2).¹⁰⁸

The germline inactivating mutations of PTEN are associated with a predisposition to autosomal-dominant hamartoma syndromes^42,109,110 with different phenotypic features including Cowden syndrome (CS),^15,16 Bannayan-Riley-Ruvalcaba syndrome (BRRS),¹⁷ Proteus syndrome (PS),¹⁸ and Proteus-like syndrome (PLS).^19,111 In over 85% of cases, CS results from germline PTEN mutations in promoter region, and are characterized by multiple hamartomas of the hair follicle, the mucocutaneous membranes, breast, thyroid and intestinal tissues, and an increased risk of breast, thyroid and endometrial cancer.⁴² The characteristic features of BRRS are lipomatosis, macrocephaly, high birth weight, and in males, speckled penis and no association with an increased risk of malignancy.^109,112 Germline PTEN mutations such as deletion of 10q23 and translocation involving 10q23 has been found in more then 60% of BRRS patients.¹¹¹ Interestingly, that mutational spectra of BRRS and CS overlaps. PS is a disorder of patchy or mosaic postnatal overgrowth of unknown etiology such as asymmetric limb overgrowth and length discrepancy, macrodactyly, vertebral abnormalities, hyperostosis, abnormal and asymmetric fat distribution, asymmetric muscle development, connective-tissue nevi and vascular malformations such as deep venous thrombosis and pulmonary embolism.¹¹³ Because of mosaic pattern of distribution, patients with PS manifest variable phenotype. The germline PTEN mutations have been detected in 10–20% cases of PS.¹¹¹ PLS is diagnosed in patients with overgrowth which does not meet established diagnostic criteria for PS.¹¹¹ 60% patients with PLS carry germline PTEN mutations¹¹¹ including R335X and mosaic R130X.¹⁹

LKB1

The tumor suppressor LKB1, also known as STK11 (serine-threonine kinase 11), is a serine/threonine kinase that is mutationally inactivated in PJS hamartoma syndrome.^13,27 LKB1 serves to activate AMPK by direct phosphorylation of Thr172 in its activation loop, which is essential for AMPK catalytic activity. AMPK, a cellular energy sensor, functions as a master regulator of cellular energy metabolism. Depletion of intracellular ATP levels due to either physiological stimuli such as exercise and muscle contraction or pathological stresses such as hypoxia, oxidative stress and glucose deprivation activate AMPK.⁸⁷ AMPK functions to restore intracellular ATP levels by inhibiting ATP-consuming processes such as protein translation and cell growth and promoting ATP-generating processes such as gluconeogenesis and lipogenesis.¹¹⁴ Thus, under energy starvation AMPK inhibits cell proliferation by directly phosphorylating TSC2 and enhancing its ability to switch off mTORC1 signaling.¹¹⁵ Mutational inactivation of LKB1 results in hyperactivation of mTORC1 signaling under low energy conditions,^116,117 suggesting that LKB1 is required for repression of mTORC1 in a AMPK- and TSC2-dependent manner.

Germline mutations of LKB1 are documented in more than 70–80% of PJS patients; up to 15% of cases have germline deletions of all, or part of LKB1. Hamartoma syndrome in PJS manifests by mucocutaneous pigmentation defects and hamartomatous polyps in the gastrointestinal tract.²⁷ Another key feature of PJS is a high predisposition to the early onset of a variety of carcinoma including small intestine, colorectal, esophageal, pancreatic, breast cancers and adenocarcinoma.¹¹⁸ The majority of patients have a familial history of PJS; however, 10–20% of PJS cases result from spontaneous germline mutations. The majority of disease state due to LKB1 gene mutations results in large truncations of the catalytic domain, which impairs LKB1 catalytic activity. A significant number of point mutations in the kinase domain and in the C-terminal noncatalytic region have been also identified in PJS patients.²⁷

Loss of heterozygocity (LOH), haploinsufficiency and epigenetic inactivation

The prevailing opinion for hamartoma syndromes is that the disease develops in accordance with the classic Knudson’s two-hit tumor suppressor gene model.¹¹⁹ For example, a mutation in either TSC1 or TSC2 is followed by a second hit referred to as loss of heterozygosity (LOH) leading to the loss of function of either TSC1 or TSC2 proteins. LOH in TSC1 or TSC2 has been identified in TSC lesions including angiomyolipomas,³ LAM cells,¹² and ependymal giant-cell tumors; however, LOH in TSC1 or TSC2 were rarely found in cerebral cortical tubes, a neurological manifestation of familial TSC.³ Interestingly, in about 15–20% patients with clinical symptoms of TSC, no identifiable mutations in either TSC1 or TSC2 have been found.³ In TSC skin hamartoma, TSC1 or TSC2 two-hit deletions were also not detected in the proliferating epidermal cells.¹²⁰ In CS-related tumors, hemizygous inactivation of PTEN occurs very frequently, however, loss of the second allele is rare.⁴² There is the possibility that either LOH or gene mutations were not detected due to the low sensitivity of used technical approaches; or loss of one allele may have a partial tumor promoting effect.⁴² Importantly, PTEN gene mutations are not the only cause of PTEN inactivation, which can also be due to other mechanisms such as epigenetic inactivation of tumor suppressor genes, transcriptional regulation and posttranslational modifications of tumor suppressor proteins. Thus, epigenetic inactivation of the PTEN was first reported in prostate cancer xenografts due to methylation of PTEN gene promoter and loss of PTEN protein expression.¹²¹ Furthermore, nonsense mutations leading to degradation of PTEN mRNA¹⁷ or regulation of PTEN levels by TGFβ through p53,^122,123 had been linked to the downregulation of PTEN tumor suppressor function. PTEN and TSC2 can also be post-translationally inactivated due to phosphorylation. It was shown that the phosphorylation of the C-terminus of PTEN by CK2,¹²⁴ regulates protein stability and function.¹²⁵ ERK-dependent TSC2 phosphorylation leads to dissociation of TSC1 and TSC2 complex, which results in mTOR/ S6K1 activation, increased cell proliferation and transformation.⁸⁵ Collectively, the evidence demonstrates that hamartoma syndromes can develop not only due to LOH but also due to epigenetic inactivation of tumor suppressor genes, transcriptional regulation and posttranslational modifications of tumor suppressor proteins.

Targeting the mTORC1

There are currently no approved therapies for PHTS, TSC, LAM and PJS in clinic. However, rapamycin, a specific inhibitor of mTORC1, discovered more then 30 years ago, attracts a renewed interest. Rapamycin (sirolimus is the official generic name) is a prototypical inhibitor of mTORC1 signaling.¹²⁶ It was discovered in early 1970s during the screening for new anti-fungal agents, and was first named rapamycin because it was isolated from a soil sample from island Rapa Nui, the indigenous name of Easter Island in the South Pacific. Rapamycin is a macrocyclic lactone produced by Streptomyces hydroscopicus.¹²⁶ Rapamycin forms a complex with its intracellular receptor FKBP12 and binds directly to mTORC1, which suppresses mTORC1-mediated signaling.⁸¹ However, the precise mechanism of rapamycin-induced inhibition of mTORC1 signaling is not completely understood, and at least two mechanisms of rapamycin action have been proposed. It was shown that under some experimental conditions FKBP12-rapamycin destabilizes the interaction of mTOR with raptor, which is required for mTOR activity.⁵² Separate studies demonstrate that the FKBP12-rapamycin complex suppresses mTORC1 autophosphorylation, which also may inhibit mTORC1 activity⁷⁶ suggesting the possibility that the rapamycin-dependent mTORC1 inhibition may involve more then one mechanism.

Because mTORC1 signaling is a highly conserved pathway that regulates protein synthesis and cell growth in all eukaryotes and because of its activation not only in hamartoma syndromes, but in many types of cancer, there is a growing interest in rapamycin and its analogs. In 1999 rapamycin was approved by the Food and Drug Administration (FDA) for the prevention of renal allograft rejection. Currently, clinical studies demonstrate that rapamycin and its analogs, such as CCI-779 (cell cycle inhibitor 779, also known as temsirolimus; Wyeth), RAD001 (also known as everolimus; Novartis), AP23573 (Ariad Pharmaceuticals) have shown anti-cancer activity in variety of malignancies including non-small cell lung carcinoma, advanced neuroendocrine carcinomas, breast, cervical, renal and uterine carcinomas, sarcoma, mesothelioma, metastatic melanoma, relapsed mantle cell lymphoma, glioblastomas and hepatocellular and cholangiocellular cancers.^8,127–141 Furthermore, a number of rapamycin analogs are in preclinical development and they are on the way towards the clinic; numerous patents of compounds as PI3K, Akt and mTORC1 inhibitors also have been filed.

Originally, Dr. Krymskaya’s group reported that the mTOR kinase inhibitor rapamycin selectively blocks the constitutive activation of S6K1 and proliferation of cells with mutated TSC2 from LAM patients.^12,142 These initial reports have now been confirmed by other laboratories using Eker rat and the TSC1 and TSC2 mouse tumors models.^143–145 These pre-clinical studies led to a Phase 2 rapamycin clinical trial for LAM patients with TSC at the Children’s Hospital Medical Center in Cincinnati; the results of these clinical study had been recently published.⁶ The nonrandomized, open-label Cincinnati Angiomyolipoma Sirolimus Trial for patients with TSC and sporadic LAM demonstrated that treatment with rapamycin leads to a 50% reduction in angiomyolipomas (AML) tumor size and approximately 10% improvement of the lung function in patients with LAM over the first year of treatment.⁶ Additionally, a case report provided by the Crestani group from Bichat-Claude Bernard Hospital, France, demonstrated that rapamycin treatment of TSC patient with giant angiolipomas induced a dramatic reduction in bilateral renal tumors.⁹ Although the improvements in AML volume and lung function waned after therapy withdrawal, the initial response suggests that targeting the mTORC1 pathway has promise in the treatment of TSC and LAM.⁶ Regression of TSC-associated subependymal giant-cell astrocytomas in response to off-label therapy with rapamycin was also reported by Franz group from Cincinnati Children’s Hospital Medical Center.⁷ On the basis of these studies, a 3-year randomized controlled MILES (Multicenter International LAM Efficacy of Sirolimus) Trial opened in 2006 in Cincinnati. Furthermore, a number of mTOR inhibitor trials for TSC and LAM patients are open for enrollment at the Dana-Farber Cancer Institute, Boston (Phase II clinical trial of sirolimus), Children’s Hospital Medical Center, Cincinnati (Phase I and phase II clinical trials of RED001, Novartis), and the United Kingdom (Phase II clinical trial of sirolimus, Cardiff University, University of Nottingham, St. Georges Hospital Medical School, Royal Sussex County Hospital, Wales Gene Park, The Tuberous Sclerosis Association, and Wyeth).¹⁴⁶

Currently, the phase II clinical trial of rapamycin in patients with Cowden syndrome is open for enrollment at the Warren Grant Magnuson Clinical Center, Maryland (NCI, ClinicalTrials.gov Identifier NCT00722449).

Small molecule PI3K inhibitors

Specific, drug-targeted inhibition of PI3K may have beneficial effects for the combination treatment of hamartoma syndromes because it may alleviate the influence of prolonged rapamycin-induced inhibition of mTORC1 on PI3K/Akt activation. Extensive studies of the PI3K family demonstrated that four different, but closely related, PI3K isoforms have distinct biological functions suggesting that specific drug targeting of PI3K may have differential physiological effects.¹⁴⁷ PI3Kα plays an important role in tumorigenesis, insulin signaling and glucose metabolism,^35,36 PI3Kβ plays a role in thrombosis-related diseases,¹⁴⁸ while PI3Kγ and PI3Kδ play key roles in inflammation and the immune system.^37,39

While the first generation of pan-PI3K inhibitors, LY294002 (Lilly Research Laboratories),¹⁴⁹ wortmannin, a fungal product isolated from Penicillium wortmanni (KY12420, Kyowa Hakko Kogyo Co., Ltd., Japan)¹⁵⁰ and wortmannin analogue, PX-866, (Arizona Cancer Center, University of Arizona),¹⁵¹ had been reported to show anticancer effects in vivo, these molecules were not used in clinical trials because of their low specificity and high toxic effects.^151,152 During the last 5 years, great progress has been made in the pharmaceutical development of PI3K inhibitors. Novel pan-PI3K inhibitors, which have been developed for cancer therapy, are shown to have anti-tumor activity in cellular and xenograft models (e.g., NVP-BEZ235 (Novartis) possessed strong antiproliferative activity in human prostate tumor PC3M and glioblastoma U87MG cells,¹⁵³ glioblastoma tumor model,¹⁵³ and breast cancer xeno-grafts;¹⁵⁴ PI-103 (Piramed Pharma) showed significant anti-tumor activity in orthotopic breast and ovarian cancer xenograft models¹⁵⁵ as well as in glioma cell models and xenografted EGFR-driven glioma tumors;¹⁵⁶ SF1126 (Semafore Pharmaceuticals, Inc.,) inhibited U87MG and PC3 tumor growth;¹⁵⁷ ZSTK474 (Zenyaku Kogyo, Japan) inhibited growth of mouse B16F10 melanoma tumors and A549, PC-3 and WiDr human xenografts, which originated from a non-small-cell lung cancer, a prostate cancer and a colon cancer, respectively,¹⁵⁸ and XL765 and XL147 (Exelixis) have shown anti-tumor activity in several preclinical xenograft models (information from http://www.exelixis.com). Importantly, novel pan-PI3K inhibitors demonstrate modest toxicity in pre-clinical studies, even though the molecules may cross-react with other members of the PI3K family and PI3K-related kinases.^29,147 Because of promising pre-clinical data, some of the second generation pan-PI3K inhibitors have recently entered clinical studies. BEZ235 (Novartis) is now in phase I clinical trial in solid tumors and advanced breast cancer; BGT266 (Novartis) is in ongoing phase I/II trial as monotherapy for breast cancer, solid tumors and Cowden Syndrome. XL765 and XL147 (Exelixis) entered phase I trials in adults with solid tumors and malignant gliomas.

Although broad-range PI3K inhibitors show promising results as anti-cancer agents, selectively targeting individual PI3K isoforms may be also beneficial. Pre-clinical studies suggest that simultaneous inhibition of all PI3K isoforms, which have different functions,^{35–37,39,148} may lead to a number of pathologies elicited from the modification of normal isoform activity (reviewed in ref. 159), and inactivation of individual PI3K isoforms reduce potential undesirable effects of pan-PI3K inhibitor compounds.

The possibility of developing isoform-selective PI3K inhibitors have been demonstrated by determination of the crystal structure of PI3Kγ. It was shown that minor differences exist in the ATP binding pockets; this allows the production of selective kinase inhibitors.¹⁶⁰ Recently, a number of patent specifications have been published which represent promising lead compounds for the future development of isoform-specific PI3K inhibitors: several imidazopyridine derivatives which exhibit excellent PI3K inhibitory activities, especially against p110α, were reported by Yamanouchi;¹⁶¹ quinolone and pyridopyrimidine compounds that are approximately 100-fold more potent against p110α and p110β isoforms compared to PI3Kγ was reported by Thrombogenix;¹⁶² number of p110δ quinazoline inhibitors, which inhibit p110δ with 50-fold selectivity over the other class I PI3K isoforms, were reported by ICOS Corporation.¹⁶³ TG100-115 (TargeGen), a PI3Kγ/δ isoforms-specific inhibitor, was shown to counteract edema and inflammation via blocking VEGF signaling events central to late tissue damage after myocardial infarction. Importantly, preclinical studies show that TG100-115 selectively blocked PI3K/Akt/mTORC1/S6K1 signaling, but not c-Fos and ERK1/2, which were blocked by pan-PI3K inhibitors. Additionally, TG100-115 had no influence on growth, while pan-isoform inhibitors were profoundly anti-proliferative.¹⁶⁴ Currently, TG100-115 is in clinical phase I/II trials for the reduction of acute myocardial infarction.¹⁶⁴ Given the key role of PI3Kγ in cell signaling, selective inhibitors of PI3Kγ might be also useful in preventing inflammatory cell recruitment in a range of inflammatory diseases, including asthma, rheumatoid arthritis, multiple sclerosis and inflammatory bowel disease. While much of the information concerning more recently developed inhibitors is confined to the patent literature, a side-by-side comparison and activity mapping of PI3K inhibitors is presented in ref.²⁸ The availability of selective PI3K inhibitors provides opportunities for the combination therapy of hamartoma syndromes, supported by recent data. Thus, PI3K inhibitors have been shown to reverse resistance to trastuzumab (herceptin, a monoclonal antibody directed against ERBB2), which was caused by the loss of PTEN.¹⁶⁵ Furthermore, novel design-based approach was used to develop dual inhibitor PP121 that targets both PI3K and tyrosine kinase activity.^166,167

PI3K/mTORC1 activation: Therapeutic prospects

While the critical role of the mTORC1 signaling pathway proves it is a promising therapeutic target to treat hamartoma syndromes associated with PTEN, LKB1 and TSC1/TSC2 germline mutations, recent studies suggest that caution should be taken in the use of mTOR inhibitors to treat tumors that result from the activation of the mTORC1 signaling pathway due to signaling feedbacks and crosstalks (reviewed in ref. 168). The continuous activation of mTORC1 in cultured cells suppresses insulin-dependent activation of PI3K by downregulating IRS1 (insulin receptor substrate-1) gene expression or by inhibitory phosphorylation of IRS-1.⁸⁰ Similarly, in vivo, tumors that arise in mice due to loss of TSC2 exhibit constitutive activation of the mTORC1 pathway but not activation of the PI3K pathway.¹⁶⁹ This feedback appears to explain why tumors that lose TSC2 are predominantly benign. However, the constitutive activation of the mTORC1 signaling pathway due to TSC2 loss in the context of PI3K activation induces much more aggressive tumors leading to the death of the mice,¹⁶⁹ and prolonged treatment with rapamycin may lead to the development of drug-resistance.^170,171 Additionally, the substantial pre-clinical data in testing the effects of rapamycin for the treatment of different types of cancer also indicates that rapamycin and its analogs, while having anti-tumor effects, have not shown universal anti-tumor activity in clinical trials.⁸¹

Since constitutive activation of mTORC1 provides negative feedback to PI3K/Akt, mTORC1 inhibition may lead to reactivation of PI3K-Akt and higher tumor severity. While pre-clinical study demonstrates that prolonged rapamycin treatment suppresses Akt/PKB signaling due to the destabilization of mTORC2 in some types of cultured cancer cells,¹⁷² the results from recent clinical trials provide evidence about PI3K-Akt activation in patient tumors treated with rapamycin or the rapamycin analog RAD001,^50,51 Thus, Dr. Rosen’s group demonstrated that mTOR inhibition with RAD001 treatment induced IRS-1 expression and Akt activation in patient tumors.⁵⁰ Another study shows that inhibition of mTORC1 with RAD001 leads to activation of the MAPK pathway through a PI3K-dependent feedback loop in biopsy-assessed tumor samples.⁵¹

Collectively, this evidence demonstrates that the efficiency of the rapamycin family of compounds in preclinical and clinical studies is less then satisfying. Given that over-activation of mTORC1 results in negative feedbacks, it is evident that the use of rapamycin as a single therapeutic agent is limited. Therefore, combinational targeting of PI3K, mTORC1 and MAPK pathways may represent a major direction in treatment of hamartoma syndromes. Given the critical importance of PI3K, mTORC1 and MAPK signaling in normal cellular functions, combinational therapy targeting this pathways may have some limitations due to high toxicity of targeting critical functions in non-cancerous cells. Another possibility could be development of so-called “dirty drugs” that can simultaneously target multiple molecules. For example Novartis and Exelixis developed compounds NVP-BEZ235 and XL765, respectively, for combined targeting of mTOR kinase and PI3K activities.¹⁷³ However, preclinical studies and the clinical efficacy and tolerability of modulators of PI3K and mTORC1 signaling are eagerly awaited.