Targeting the Phosphoinositide 3-Kinase Pathway in Cancer

REVIEWS

Targeting the phosphoinositide 3‑kinase pathway in cancer

Pixu Liu, Hailing Cheng, Thomas M. Roberts and Jean J. Zhao Abstract | The phosphoinositide 3‑kinase (PI3K) pathway is a key signal transduction system that links oncogenes and multiple receptor classes to many essential cellular functions, and is perhaps the most commonly activated signalling pathway in human cancer. This pathway therefore presents both an opportunity and a challenge for cancer therapy. Even as inhibitors that target PI3K isoforms and other major nodes in the pathway, including AKT and mammalian target of rapamycin (mTOR), reach clinical trials, major issues remain. Here, we highlight recent progress that has been made in our understanding of the PI3K pathway and discuss the potential of and challenges for the development of therapeutic agents that target this pathway in cancer.

Germline mutation Since its discovery in the 1980s, the family of lipid Class IA PI3Ks. These are heterodimers consisting of A heritable change in the DNA kinases termed phosphoinositide 3‑kinases (PI3Ks) has a p110 catalytic subunit and a p85 regulatory subunit that occurred in a germ cell or been found to have key regulatory roles in many cell‑ (FIG. 2a). The regulatory subunit mediates receptor the zygote at the single-cell ular processes, including cell survival, proliferation and binding, activation, and localization of the enzyme. stage. When transmitted to the differentiation1–3. As major effectors downstream of In mammals, the genes PI3K regulatory subunit 1 next generation, a germline mutation is incorporated in receptor tyrosine kinases (RTKs) and G protein‑coupled (PIK3R1), PIK3R2 and PIK3R3 encode p85α (and its every cell of the body. receptors (GPCRs), PI3Ks transduce signals from various splice variants p55α and p50α), p85β and p55γ regu‑ growth factors and cytokines into intracellular messages latory subunits, respectively. This group of subunits Somatic mutation by generating phospholipids, which activate the serine– is collectively called p85 (reviewed in ReFs 2,3). In Also referred to as an ‘acquired mutation’, this is an alteration threonine protein kinase AKT (also known as protein response to growth factor stimulation and the sub‑ in DNA that occurs in a somatic kinase B (PKB)) and other downstream effector path‑ sequent activation of RTKs, PI3K is recruited to the cell, in contrast to a mutation ways (FIG. 1). The tumour suppressor PTEN (phosphatase membrane by direct interaction of its p85 subunit with in a germ cell. and tensin homologue) is the most important negative tyrosine phosphate motifs on activated receptors (for regulator of the PI3K signalling pathway4,5. Recent example, platelet‑derived growth factor receptor) or human cancer genomic studies have revealed that many to adaptor proteins associated with the receptors (for components of the PI3K pathway are frequently targeted example, insulin receptor substrate 1 (IRS1)). The acti‑ by germline mutations or somatic mutations in a broad vated p110 catalytic subunit generates phosphatidyl‑

range of human cancers. These findings, and the fact inositol‑3,4,5‑trisphosphate (PtdIns(3,4,5)P3), which that PI3K and other kinases in the PI3K pathway are activates multiple downstream signalling pathways amenable to pharmacological intervention, make this (FIGs 1,2b). pathway one of the most attractive targets for therapeutic intervention in cancer6. Class IB PI3Ks. The class IB PI3K that has been fully characterized to date is a heterodimer composed of Pathway background the catalytic subunit p110γ and the regulatory subunit (ReF. 8) (FIG. 2a) Departments of Cancer PI3Ks are divided into three classes according to their p101 . Two other regulatory subunits of 7,8 Biology, Dana–Farber Cancer structural characteristics and substrate specificity class IB PI3Ks, p84 and p87 PI3K adaptor proteins, Institute, Pathology, Harvard (FIG. 2a). Of these, the most commonly studied are the have recently been described9,10. p110γ is activated Medical School, Boston, class I enzymes that are activated directly by cell surface directly by GPCRs through interaction of its regu‑ Massachusetts 02115, USA. receptors. Class I PI3Ks are further divided into class IA latory subunit with the Gβγ subunit of trimeric Correspondence to J.J.Z. 8 e‑mail: enzymes, which are activated by RTKs, GPCRs and certain G proteins . p110γ is mainly expressed in leukocytes but [email protected] oncogenes such as the small G protein RAS, and class IB is also found in the heart, pancreas, liver and skeletal doi:10.1038/nrd2926 enzymes, which are regulated exclusively by GPCRs. muscle11–13.

Insulin or growth factors LPA or RTKs chemokines

PtdIns(4,5)P2 PtdIns(3,4,5)P3 PtdIns(4,5)P2

GPCRs RAS PTEN PTEN RAS p110 p110 PtdIns PtdIns PtdIns PtdIns p110β,δ (4,5)P (3,4,5)P3 (3,4,5)P3 (4,5)P p85 p85 2 2 p85 Gβγ p101 AKT PDPK1 p110γ Class IA PI3K Class IB PI3K

RAC1 SGK PKC MDM2 FOXO1 NFκB BAD GSK3β mTOR S6K

• Survival p53

• Transformation • Apoptosis • Growth • Translation

• Motility • Apoptosis • Cell cycle regulation • Transformation • Cell cycle regulation • Glucose metabolism • DNA repair Adaptor

Figure 1 | The class i Pi3K signalling pathway. Following growth factor stimulation and subsequent activation of receptor tyrosine kinases (RTKs), class IA phosphoinositide 3‑kinases (PI3Ks), consisting of p110α–p85, p110β–p85 and p110δ–p85, are recruited to the membrane by direct interaction of the p85 subunit withNatur thee Reactivatedviews | Drug receptors Discovery (for example, platelet‑derived growth factor receptor) or by interaction with adaptor proteins associated with the receptors (for example, insulin receptor substrate 1). The activated p110 catalytic subunit converts

phosphatidylinositol‑4,5‑bis phosphate (PtdIns(4,5)P2) to phosphatidylinositol‑3,4,5‑trisphosphate (PtdIns(3,4,5)P3) at the membrane, providing docking sites for signalling proteins that have pleckstrin homology domains, including putative 3‑phosphoinositide‑dependent kinase 1 (PDPK1) and serine–threonine protein kinase AKT (also known as protein kinase B). PDPK1 phosphorylates and activates AKT, which elicits a broad range of downstream signalling events. The class IB PI3K (p110γ–p101) can be activated directly by G protein‑coupled receptors (GPCRs) through interaction with the Gβγ subunit of trimeric G proteins. The p110β and p110δ subunits can also be activated by

GPCRs. PTEN (phosphatase and tensin homologue) antagonizes the PI3K action by dephosphorylating PtdIns(3,4,5)P3. BAD, BCL2‑associated agonist of cell death; FOXO1, forkhead box O1 (also known as FKHR); GSK3β, glycogen synthase kinase 3β; mTOR, mammalian target of rapamycin; NF‑κB, nuclear factor‑κB; PKC, protein kinase C; RAC1, RAS‑related C3 botulinum toxin substrate 1; SGK, serum and glucocorticoid‑regulated kinase; S6K, ribosomal protein S6 kinase; LPA, lysophosphatidic acid.

Class II PI3Ks. These consist of a single catalytic sub‑ indicating a potential role in regulating cell growth14. unit, which preferentially uses phosphatidylinositol or Interestingly, it has also been implicated as an important phosphatidylinositol‑4‑phosphate (PtdIns4P) as sub‑ regulator of autophagy (reviewed in ReF. 14), a cellular strates2,3 (FIG. 2a). There are three class II PI3K isoforms response to nutrient starvation. — PI3KC2α, PI3KC2β and PI3KC2γ — which can be

activated by RTKs, cytokine receptors and integrins; PTEN. The phospholipid PtdIns(3,4,5)P3, which is gen‑ however, the specific cellular functions of this family erated by activated class I PI3Ks, is the key second mes‑ remain unclear. senger that drives several downstream signalling cascades that regulate cellular processes (FIG. 1). The cellular levels

Class III PI3K. The class III PI3K consists of a single of PtdIns(3,4,5)P3 are tightly regulated by the opposing catalytic subunit, vPS34 (homologue of the yeast vacu‑ activity of PTEN. PTEN, an important tumour suppres‑ olar protein sorting‑associated protein 34; also known sor, functionally antagonizes PI3K activity through its as PIK3C3). vPS34 only produces PtdIns3P, which intrinsic lipid phosphatase activity that reduces the cellular

is an important regulator of membrane trafficking pool of PtdIns(3,4,5)P3 by converting PtdIns(3,4,5)P3 back ReF. 3 (reviewed in ). vPS34 has been shown to function to phosphatidylinositol‑4,5‑bisphosphate (PtdIns(4,5)P2) as a nutrient‑regulated lipid kinase that mediates signal‑ (FIG. 2b). loss of PTEN results in unrestrained signalling by ling through mammalian target of rapamycin (mTOR), the PI3K pathway, leading to cancer (reviewed in ReF. 4).

a Class IA p85 BD RAS BD C2 Helical domain Catalytic domain (p110α, p110β and p110δ)

SH3 BHD SH2 iSH2 SH2 p85α or p85β SH2 iSH2 SH2 p55α or p55γ p85 regulatory domain SH2 iSH2 SH2 p50α

Class IB RAS BD C2 Helical domain Catalytic domain (p110γ) p110γ BD Gβγ BD p101 Regulatory domain p110γ BD Gβγ BD p84 or p87

Class II RAS BD C2 Helical domain Catalytic domain PX C2 (PIK3C2α, PIK3C2β, PIK3C2γ) Class III C2 Helical domain Catalytic domain (VPS34) b

Class I PI3Ks

PTEN

O O O O O O O O 3 3 1 2 1 2 O O –O P O –O P O HO HO O 6 O 6 2 1 O O 2 1 O 4 OH 4 OH HO O P O– –O P O O P O– 3 O 5 3 O 5 O– O– O– –O P O –O P O PtdIns(4,5)P2 O– PtdIns(3,4,5)P3 O– Figure 2 | The Pi3K family and phosphatidylinositol-3,4,5-trisphosphate generation. a | Phosphoinositide 3‑kinases (PI3Ks) are divided into three classes according to their structural characteristics and substrateNa specificity.ture Reviews Class | Drug IA DiscPI3Ksove arery heterodimers consisting of a p110 catalytic subunit and a p85 regulatory subunit. There are three p110 catalytic isoforms: p110α, p110β and p110δ. Whereas the expression of p110δ is largely restricted to the immune system, p110α and p110β are ubiquitously expressed3,8. The p110 catalytic isoforms are highly homologous and share five distinct domains: an amino‑terminal p85‑binding domain (p85 BD), which interacts with the p85 regulatory subunit; a RAS‑binding domain (RAS BD), which mediates activation by members of the RAS family of small GTPases; a putative membrane‑binding domain, C2; the helical domain; and the carboxy‑terminal kinase catalytic domain. There are also three p85 isoforms: p85α (and its splice variants p55α and p50α), p85β and p55γ. They share three core domains, including a p110‑binding domain called the inter‑Src homology 2 (iSH2) domain, flanked by two SH2 domains. The two longer isoforms, p85α and p85β, have an SH3 domain and a BCR homology domain (BHD) located in their extended N‑terminal regions. In the basal state, p85 binds to the N‑terminus of the p110 subunit through its iSH2 domain, inhibiting its catalytic activity7,8. Class IB PI3K is a heterodimer composed of the catalytic subunit p110γ and the regulatory subunit p101. p110γ is mainly expressed in leukocytes and can be activated directly by G protein‑coupled receptors8. Class II PI3Ks are monomers with a single catalytic subunit. There are three class II PI3K isoforms: PI3KC2α, PI3KC2β and PI3KC2γ, each of which has a divergent N‑terminus followed by a RAS‑binding domain, a C2 domain, a helical domain and a catalytic domain, with PX (Phox homology) and C2 domains at the C‑termini (reviewed in ReFs 2,3). The class III PI3K consists of a single catalytic subunit, VPS34 (homologue of the yeast

vacuolar protein sorting‑associated protein 34). b | Phosphatidylinositol‑3,4,5‑trisphosphate (PtdIns(3,4,5)P3) is an important lipid second messenger that regulates many cellular processes. Class I PI3Ks phosphorylate the inositol ring of

phosphatidylinositol‑4,5‑trisphosphate (PtdIns(4,5)P2) on the 3 position, to generate PtdIns(3,4,5)P3. PTEN (phosphatase and

tensin homologue) is a lipid phosphatase that removes phosphate on the 3 position of PtdIns(3,4,5)P3 and converts it back to

PtdIns(4,5)P2. Gβγ BD, G protein βγ subunit‑binding domain.

Table 1 | Incidence of genetic alterations in the PI3K pathway in cancer genetic alteration cancer type incidence of tumors with refs alterations p110a (PIK3CA) Mutations Breast 27% (468/1766) * Endometrial 24% (102/429) * Colon 15% (448/3024) * Upper digestive tract 11% (38/352) * Gastric 8% (29/362) * Pancreas 8% (8/104) * Ovarian 8% (61/787) * Liver 6% (19/303) * Brain 5.9% (59/996) *,30,31 Oesophageal 5% (13/239) * Lung 3% (28/962) * Melanoma 9% (24/278) * Urinary tract 17% (28/162) * Prostate 2% (1/57) * Thyroid 2% (7/394) * Amplifications Lung (squamous cell) 53% (40/75) 139–141 Lung (adenocarcinoma) 12.5% (15/120) 139–141 Lung (small cell) 21.4% (3/14) 139–141 Lung (non‑small‑cell) 12.0% (11/92) 142 Cervical 69% (11/16) 143 Breast 8.7% (8/92) 144,145 Head and neck 32.2% (52/161) 145–147 Gastric 36% (20/55) 148 Thyroid 9% (12/128) 149 Oesophageal 6% (5/87) 150 Cervical 9% (2/22) 151 Endometrial 10% (3/29) 151 Ovarian 11.9% (16/134) 152,153 Glioblastoma 6.1% (21/344) 30,154 p110b (PIK3CB) Amplifications Ovarian 5% 57 Breast 5% 57

Increase in activity and Colon 70% (7/10) 56 expression Bladder 89% (8/9) 56

AKT. This serine–threonine protein kinase is expressed has been phosphorylated and activated, it phosphorylates as three isoforms — AKT1, AKT2 and AKT3 — which many other proteins — for example, glycogen synthase are encoded by the genes PKBα, PKBβ and PKBγ, respec‑ kinase 3 and FOXOs (the forkhead box family of transcrip‑ tively (reviewed in ReFs 1,15). The three isoforms share a tion factors) — thereby regulating a wide range of cell‑ similar structure: an amino‑terminal pleckstrin homology ular processes involved in protein synthesis, cell survival, domain, a central serine–threonine catalytic domain, and proliferation and metabolism (reviewed in ReFs 15,19). a small carboxy‑terminal regulatory domain. AKT activa‑ tion is initiated by translocation to the plasma membrane, mTOR. mTOR plays a crucial part in the regulation of cell which is mediated by docking of the pleckstrin homology growth and proliferation by monitoring nutrient avail‑ domain in the N‑terminal region of AKT to PtdIns(3,4,5) ability, cellular energy levels, oxygen levels and mitogenic ReF. 20 P3 on the membrane. The resulting conformational change signals (reviewed in ). It belongs to a group of in AKT exposes two crucial amino‑acid residues for serine–threonine protein kinases of the PI3K superfamily, phosphorylation16,17. Both phosphorylation events — on referred to as class Iv PI3Ks, including ATm, ataxia T308 by 3‑phosphoinositide‑dependent protein kinase 1 telangiectasia and RAD3‑related protein (ATR), DNA‑PK (PDPK1) and on S473 by putative PDPK2 — are required and SmG1 (SmG1 homologue, phosphatidylinositol for full activation of AKT1,16,17. A number of potential 3‑kinase‑related kinase). mTOR exists in two distinct PDPK2s have been identified, including IlK (integrin‑ complexes — mTORC1 and mTORC2. The mTORC1 linked kinase), protein kinase C b2, DNA‑dependent complex is composed of the mTOR catalytic subunit, protein kinase (DNA‑PK), ATm (ataxia telangiectasia regulatory associated protein of mTOR (raptor), proline‑ mutated) and AKT itself 15; however, it is currently thought rich AKT substrate 40 kDa (PRAS40) and the protein that the mTOR–rictor (rapamycin insensitive companion mlST8 (reviewed in ReFs 20,21). mTORC2 is composed of of mTOR) complex (mTORC2) is the primary source of mTOR, rictor, mammalian stress‑activated protein kinase PDPK2 activity under most circumstances18. Once AKT interacting protein 1 (mSIN1) and mlST8 (ReF. 21).

Table 1 (cont.) | Incidence of genetic alterations in the PI3K pathway in cancer genetic alteration cancer type incidence of tumors with refs alterations PDPK1 Amplifications and Breast 20% 57 overexpression AKT AKT1 mutation (E17K) Breast 3.7% (31/845) 44,58,155,156 Colon 2.8% (4/139) 58,155 Ovarian 2% (1/50) 58 Lung 1.9% (2/105) 157 AKT1 amplifications Gastric 20% (1/5) 158 AKT2 amplifications Ovarian 14.1% (30/213) 143,159,160,152 Pancreas 20% (7/35) 161 Head and neck 30% (12/40) 146 Breast 3% (3/106) 159 AKT3 mutation (E17K) Skin 1.5% (2/137) 59 AKT3 amplifications Glioblastoma 2% (4/205) 30 p85a (PIK3R1) Mutations Glioblastoma 9.9% (9/91); 8% (8/105) 30,31 Ovarian 4% (3/80) 58 Colon 2% (1/60) 58 PTEN Loss of heterozygosity Gastric 25.3% (84/332) 148,162,163 Breast 24.9% (99/398) 164–167 Melanoma 37% (53/143) 168–171 Prostate 30% (70/230) 172–176 Glioblastoma 28% (113/404) 30,31,177–179 Mutations Endometrial 38% (604/1569) * Brain 21% (611/2913) *,30,31 Skin 17% (96/555) * Prostate 14% (51/371) * Colon 13% (53/416) * Ovary 9% (55/645) * Breast 6% (34/561) * Haematopoietic and lymphoid tissue 6% (54/866) * Stomach 6% (28/499) * Liver 5% (20/372) * Kidney 5% (14/294) * Vulva 65% (17/26) * Urinary tract 9% (13/142) * Thyroid 5% (27/591) * Lung 9% (48/548) * * The values were obtained from http://www.sanger.ac.uk/genetics/CGP/cosmic/. PDPK1, 3‑phosphoinositide‑dependent protein kinase 1; PI3K, phosphoinositide 3‑kinase; PIK3CA, PI3K catalytic subunit α‑isoform; PIK3CB, PI3K catalytic subunit β‑isoform; PIK3R1, PI3K regulatory subunit‑α; PTEN, phosphatase and tensin homologue.

AKT can activate mTOR by phosphorylating both may provide tumour cells with a growth advantage by PRAS40 and tuberous sclerosis 2 protein (TSC2; also promoting protein synthesis. when bound to rictor in known as tuberin) to attenuate their inhibitory effects the mTORC2 complex, mTOR functions as a PDPK2 to on mTORC1 (ReFs 22–24). The discovery of the associa‑ phosphorylate AKT18. tion of mTORC1 with the bipartite complex TSC1 and TSC2 of tumour suppressor proteins provided a molecu‑ Linking the PI3K pathway to human cancers lar link between mTOR and cancer (reviewed in ReF. 25). Although PI3K was originally characterized two decades The most extensively characterized downstream targets ago through its binding to oncogenes and activated of mTORC1 are ribosomal protein S6 kinase 1 (S6K1; RTKs (reviewed in ReF. 27), its association with human also known as p70S6K) and eukaryotic translation cancer was not established until the late 1990s, when it initiation factor 4E‑binding protein 1 (4EBP1), both was shown that the tumour suppressor PTEN acts as of which are crucial to the regulation of protein syn‑ a phosphatase that is specific for the lipid products of thesis (reviewed in ReF. 26). Thus, activation of mTOR PI3K. Recent comprehensive cancer genomic analyses

have revealed that many components of the PI3K to constitutive PI3K activity30,53,54. In contrast to PIK3CA, pathway are frequently mutated or altered in common cancer‑specific mutations have not been found in the human cancers28–33, underscoring the importance of PIK3CB gene, which encodes p110β, even though sev‑ this pathway in cancer. eral groups have shown that it can act as an oncogene in model systems2,45. A recent study has shown that it The discovery of the PTEN tumour suppressor links may be more difficult to activate p110β than p110α by PI3K to human cancer. Germline mutations in the missense mutation45, perhaps because p110β has a lower PTEN gene cause various inherited cancer predisposi‑ lipid kinase activity than p110α55. However, PIK3CB has tion syndromes, including Cowden’s syndrome and been found to be amplified in some primary tumours Bannayan–Riley–Ruvalcaba’s syndrome34. Somatic loss and cancer cell lines56,57. of PTEN by gene mutation or deletion frequently occurs in common human tumours (TABLe 1). The discovery AKT and PDPK1 in human cancer. Amplification of that the tumour suppressor PTEN works by antagoniz‑ AKT1 and AKT2 has been reported in various tumour ing PI3K established the first direct link between PI3K types (TABLe 1). Recently, an activating mutation in the activation and human cancer. Although PTEN possesses pleckstrin homology domain of AKT1 (E17K) — which protein tyrosine phosphatase activity35, it is also a lipid results in growth factor‑independent membrane trans‑ phosphatase capable of specifically removing the 3′ location of AKT and increased AKT phosphorylation 58,59 phosphate from PtdIns(3,4,5)P3. This action is essen‑ levels — was identified in melanoma, breast, colorec‑ tial to its function as a tumour suppressor (reviewed in tal and ovarian cancers44,58,59. Interestingly, the equivalent ReFs 36, 37). unsurprisingly, PI3K signalling was found mutation was also detected in AKT3 in clinical speci‑ to be hyperactive in PTEN‑null tumour cell lines and mens of melanoma as well as in melanoma cell lines59. primary tumours5. The human disease phenotypes asso‑ unlike PI3K and AKT, there is only a single PDPK1 ciated with PTEN loss have been recapitulated in genetic isoform in mammals (reviewed in ReF. 60). Although mouse models (reviewed in ReFs 36, 38). Heterozygous mutations in PDPK1 are rarely found in human cancer loss or tissue‑specific homozygous loss of PTEN in the (two cases in colorectal cancer and one in glioma have mouse leads to hyperplastic proliferation and neoplastic been reported thus far61), amplification or overexpres‑ transformation in several tissues39–43. sion of PDPK1 was found in ~20% of breast cancers57.

Mutations of class IA PI3Ks frequently occur in human Current targeting of nodes in the PI3K pathway tumours. The importance of PI3Ks in cancer was con‑ Activation of the PI3K signalling pathway contributes to firmed by the discovery that the PI3K catalytic subunit cell proliferation, survival and motility as well as angio‑ α‑isoform gene (PIK3CA), which encodes p110α, is fre‑ genesis, which are responsible for all the important quently mutated in some of the most common human aspects of tumorigenesis. For this reason, many phar‑ tumours29–32,44 (TABLe 1). These genetic alterations of maceutical companies and academic laboratories are PIK3CA consist exclusively of somatic missense muta‑ actively developing inhibitors that target PI3K and other tions clustered in two ‘hotspot’ regions in exons 9 and key components in the pathway (FIGs 3,4; TABLe 2). 20, corresponding to the helical and kinase domains of p110α, respectively. Two of the most frequent PIK3CA Targeting PI3K. wortmannin and lY294002 are two mutations, E545K and H1047R, have been shown to well‑known, first‑generation PI3K inhibitors. wortmannin

increase PtdIns(3,4,5)P3 levels, activate AKT signalling is a natural product isolated from Penicillium wortmannin and induce cellular transformation2,45–48. Although the that binds irreversibly to PI3K enzymes by covalent mod‑ exact molecular mechanisms by which these mutations ification of a lysine residue that is necessary for catalytic activate p110α have not been determined, current data activity. lY294002 was the first synthetic drug‑like small‑ suggest an ablation of the inhibitory effect that is brought molecule inhibitor to be capable of reversibly targeting about by interaction of p110α with p85 (ReFs 45,49,50). PI3K family members at concentrations in the micro‑ This notion was supported by two recent structural molar range. However, both wortmannin and lY294004 studies of the p110α–p85α complex51,52. show little or no selectivity for individual PI3K isoforms Recent cancer genomic analysis of human glioblas‑ and have considerable toxicity in animals (reviewed in tomas showed that the PIK3R1 gene, which encodes the ReFs 62,63). Despite their limitations, the preclinical p85α regulatory subunit, was mutated in up to 10% of studies of these broad‑spectrum PI3K inhibitors have tumours analysed, making it one of the most frequently greatly contributed to our understanding of the biological altered glioblastoma cancer genes30,31. Interestingly, importance of PI3K signalling and provided a platform although PIK3CA mutations were also found in ~7% for the discovery of novel PI3K inhibitors. of glioblastomas in the same cohort, they were mutu‑ Numerous PI3K inhibitor chemotypes, some of ally exclusive of PIK3R1 mutations30. The presence which show differential isoform selectivity, have been of somatic mutations in PIK3R1 was also previously described63,64. A recent study64 presented a comparison reported in primary human colon and ovarian tumours of isoform selectivity profiles among a collection of and in one patient with glioblastoma53,54. Notably, most potent and structurally diverse PI3K inhibitors. The data of these mutations are located within the inter‑Src underlined a crucial role for p110α in insulin signalling homology 2 (iSH2) domain of p85α and are thought to and also provided important insights for the develop‑ disrupt the inhibitory contact of p85α with p110, leading ment of isoform‑selective PI3K inhibitors. Subsequently,

Allosteric inhibitor PI‑103, a p110α‑specific inhibitor, was shown to have a Numerous PI3K‑targeted compounds are being A molecule that inhibits an potent effect in blocking PI3K signalling in glioma cells introduced into clinical trials (FIG. 4; TABLe 2), many of enzyme by binding to a site through its ability to inhibit both p110α and mTOR65. which are dual PI3K–mTOR inhibitors. BEZ235 is an other than the active site, The seemingly off‑target effects of PI‑103 on mTOR imidazoquinazoline derivative that inhibits multiple causing a conformational change in the active site of the complexes opened a new avenue in the search for an class I PI3K isoforms and mTOR kinase activity by 66 enzyme and thereby inhibiting effective cancer therapy strategy that uses a combined binding to the ATP‑binding pocket . Preclinical data its catalytic function. inhibition of mTOR and PI3K. show that BEZ235 has strong anti‑proliferative activity against tumour xenografts that have abnormal PI3K signalling, including loss of PTEN function or gain‑ of‑function PI3K mutations67. BEZ235 has entered a EGFR HER2 MET RTKs Phase I clinical trials in patients with solid tumours (reviewed in ReF. 68). BGT226 is another potent pan‑ PI3K–mTOR inhibitor that has entered Phase I clinical trials. BKm120, unlike BEZ235 and BGT226, is selec‑ tive for class I PI3K enzymes and has no mTOR inhibi‑ P P P P P P P P RTK tory activity, and has just entered Phase I clinical trials. inhibitor P P P P P P P P The class I PI3K inhibitors Xl765 (which also targets mTOR) and Xl147 are currently under Phase I clinical investigation for treatment of solid tumours. Both are PI3K PI3K RAS inhibitor derivatives of quin oxaline, as revealed by their recently disclosed structures63. GDC0941 is a derivative of PI‑103 RAF RAF inhibitor that is active against all isoforms of class I PI3Ks in the AKT AKT nanomolar range. It displayed potent antitumour activity inhibitor MEK MEK in preclinical xenograft tumours and is under Phase I inhibitor trial in patients with advanced solid tumours or lym‑ mTOR phoma. GSK1059615, another clinical candidate that ERK inhibitor mTOR targets PI3K, has recently entered clinical trial in patients with solid tumours or lymphoma (FIG. 4; TABLe 2). b • Trastuzumab SF1126 is a covalent conjugate of lY294002 with • Mubritinib an arg–gly–asp peptide, designed for increased solu‑ • Pertuzumab • Canertinib bility and enhanced delivery of the active drug to the • Lapatinib • Ertumaxomab tumour69. In preclinical studies, SF1126 was found to • EKB-569 • Vandetanib • HKI-272 have potent inhibitory effects on cell growth, prolifera‑ • Gefitinib • Bevacizumab* tion and angiogenesis, and had lower toxicity than the • Erlotinib • Pazopanib parent lY294002. SF1126 has entered a Phase I clinical • Cetuximab EGFR HER2 VEGFR • BKM120 trial as a PI3K–mTOR inhibitor in a wide range of solid • Panitumumab • XL147 tumour cancers (FIG. 4; TABLe 2). • Matuzumab RTK • GDC0941 Numerous compounds that preferentially target • Nimotuzumab • GSK1059615 • MAb 806 • PX-866 selected isoforms of class I PI3Ks are also under develop‑ • MDX-447 PI3K • CAL101 ment. For example, PX‑866 targets p110α, p110δ and

• Perifosine p110γ with half‑maximal inhibitory concentration (IC50) • MK2206 values in the low nanomolar range70, and CAl‑101 is • BEZ235 • VQD-002 a p110γ‑selective inhibitor under Phase I clinical study • BGT226 • XL418 AKT • XL765 in patients with relapsed or refractory haematological • SF1126 • Rapamycin malignancies (FIG. 4; TABLe 2). • CCI-779 • RAD001 • AP23573 Targeting AKT. AKT is the most crucial proximal node mTOR • AZD8055‡ downstream of the RTK–PI3K complex, and is there‑ ‡ • OSI-027 fore another attractive therapeutic target. Several AKT Figure 3 | Targeting the Pi3K pathway in cancer. a | Inhibitors that target key nodes inhibitors have been developed, which can be grouped in the phosphoinositide 3‑kinase (PI3K) signalling pathway, including receptor tyrosine Nature Reviews | Drug Discovery into various classes, including lipid‑based phosphatidyl‑ kinases (RTKs), PI3K, AKT and mammalian target of rapamycin (mTOR), have reached inositol analogues, ATP‑competitive inhibitors, and clinical trials. Dual inhibitors that target both PI3K and RTK or PI3K and mTOR allosteric inhibitors. The most clinically advanced inhibitor, may provide more potent therapeutic effects in suppressing the PI3K signalling. perifosine, is a lipid‑based phosphatidylinositol analogue Combinations of PI3K and RAF–mitogen‑activated protein kinase (MAPK) inhibitors may that targets the pleckstrin homology domain of AKT, achieve more effective clinical results. b | Inhibitors in clinical development that target the PI3K or related pathways are shown. EGFR, epidermal growth factor receptor; which prevents AKT from binding to PtdIns(3,4,5)P3 71 ERK, extracellular signal‑regulated kinase; HER2, human epidermal growth factor and undergoing membrane translocation . It is currently receptor 2 (also known as ERBB2); MEK, mitogen‑activated protein kinase kinase; in clinical trials as a single agent or in combination with VEGFR, vascular endothelial growth factor receptor. *Bevacizumab targets VEGFA various drugs to treat multiple types of cancers (FIG. 4; instead of VEGFR directly. ‡Both AZD8055 and OSI‑027 are ATP‑competitive mTOR TABLe 2). Other AKT pleckstrin homology domain inhibi‑ inhibitors that target the mTOR complexes mTORC1 and mTORC2. tors, including PX316 (ReF. 72) and phosphatidylinositol

a O N S N H N H 2 N

O O O N N N NH N O2S NH HN N NH NN O N O2S NVP-BEZ235 XL-765 XL-147

O + O O O H22N NH O O O CH3O2S + N HN HO N O N O O O OH H N S N HN N N H O O O HN O NH O OH N N – O – O O O GDC0941 SF1126 PX-866

b CH3 N N H2N + NH N O N 2 HO O N P HO P N O O O N O N – O O HN N HO OH Perifosine MK-2206 VQD-002

c OH HO O OH O O

O O O N OH OH O O O O O O OH O O N P HO O O O O O O O O O HO O O O O

Rapamycin (Rapamune) O O O O OH N OH Temsirolimus N (Torisel/CCI-779) O O O O O O O O O O HO HO O O O O

Everolimus (RAD001 /Affinitor) AP23573

Figure 4 | selected chemical structures of inhibitors of the Pi3K pathway. For information relating to clinical trials Nature Reviews | Drug Discovery of these compounds, see TABLe 2. a | Phosphoinositide 3‑kinase (PI3K) inhibitors. b | AKT inhibitors. c | Mammalian target of rapamycin (mTOR) inhibitors.

Table 2 | Summary of drugs targeting the PI3K pathway in clinical trials for cancer treatment* Agent Target sponsor Phase cancer type or condition PI3K inhibitors BEZ235 Class I PI3K and Novartis Phase I–II Advanced solid tumours; advanced breast cancer mTOR BGT226 Class I PI3K and Novartis Phase I–II Solid tumours; advanced breast cancer; Cowden’s mTOR syndrome BKM120 Class I PI3K Novartis Phase I (in the first Solid tumours quarter of 2009) XL765 Class I PI3K and Exelixis Phase I Solid tumours; non‑small‑cell lung cancer; malignant mTOR gliomas XL147 Class I PI3K Exelixis Phase I Advanced solid tumours; endometrial carcinoma; ovarian carcinoma; non‑small‑cell lung cancer GDC0941 Class I PI3K Genentech Phase I Advanced solid tumours; non‑Hodgkin’s lymphoma SF1126 Pan‑PI3K and Semafore Phase I Advanced solid tumours mTOR GSK1059615 Pan‑PI3K GlaxoSmithKline Phase I Advanced solid tumours; metastatic breast cancer; endometrial cancer; lymphoma PX‑866 PI3K (α, δ and γ Oncothyreon Phase I Advanced solid tumours isoforms) CAL‑101 PI3K (δ isoform) Calistoga Phase I Chronic lymphocytic leukaemia; acute myeloid leukaemia; non‑Hodgkin’s lymphoma AKT inhibitors Perifosine AKT Keryx Phase I–II Advanced solid tumours; multiple myeloma; ovarian (also known as cancer; soft‑tissue sarcoma; malignant melanoma KRX‑0401) MK2206 AKT Merck Phase I Advanced solid tumours VQD‑002 AKT VioQuest Phase I Haematological malignancies; leukaemia; (also known as API‑2 non‑small‑cell lung cancer and TCN) XL418 AKT and S6K Exelixis Phase I‡ Solid tumours mTOR inhibitors Rapamycin/sirolimus mTORC1 Wyeth Phase I–II Advanced solid tumours; metastatic breast cancer; (Rapamune) myeloid leukaemia Approved Advanced renal cell carcinoma Temsirolimus mTORC1 Wyeth Phase I–III Advanced solid tumours; multiple myeloma; ovarian (CCI‑779/Torisel) cancer; endometrial cancer; mantle cell lymphoma; brain tumours; non‑small‑cell lung cancer; malignant melanoma Approved Advanced renal cell carcinoma Everolimus mTORC1 Novartis Phase I–III Advanced solid tumours; advanced hepatocellular (RAD001/Afinitor) carcinoma; bladder cancer; head and neck cancer;

glioma and astrocytoma; advanced prostate cancer; brain tumours; advanced gastric cancer; metastatic breast cancer; metastatic pancreatic cancer Approved Soft‑tisse and bone sarcomas AP23573 mTORC1 Merck/Ariad Phase I–III Advanced malignancies; relapsed haematological (also known as malignancies; progressive glioma; endometrial deforolimus and cancer; metastatic breast cancer; brain tumours; MK‑8669) non‑small‑cell lung cancer; prostate cancer AZD8055 mTORC1 and AstraZeneca Phase I–II Advanced solid tumours; endometrial carcinoma; mTORC2 lymphoma OSI‑027 mTORC1 and OSI Phase I Solid tumours; lymphoma mTORC2 mTOR, mammalian target of rapamycin; mTORC1, mTOR complex 1; PI3K, phosphoinositide 3‑kinase; S6K1, ribosomal protein S6 kinase 1 (also known as p70S6K). *Information presented is compiled from company websites and from www.clinicaltrials.gov and www.fda.gov. ‡The trial has been suspended due to low drug exposure.

ether lipid analogues (reviewed in ReF. 73), have shown However, preliminary results with mTOR inhibitors in inhibitory effects on the growth of tumour cells that have many other tumour types, including advanced breast high PI3K and AKT activity74. cancer and glioma, indicated a low response rate80. most ATP‑competitive small‑molecule AKT inhibitors mTOR also presents a potential second target are non‑selective, targeting all three AKT isoforms. — namely, its mTORC2 complex that functions as a GSK690693 is an ATP‑competitive AKT kinase inhibitor, PDPK2 and phosphorylates the C‑terminus of AKT at which targets all three AKT isoforms at low nanomolar ser473, an obligatory event for full activation of AKT18. concentrations and is active against additional kinases Although the clinical importance of PDPK2 function from the AGC (cyclic AmP‑dependent, cGmP‑dependent in cancer is unknown, a recent study showed that and protein kinase C) kinase family75. mTORC2 is required for the development of prostate To address a major issue regarding the potential benefits tumours that are induced by PTEN loss84. Therefore, of isoform specificity, a number of allosteric AKT inhibitors a kinase inhibitor of mTOR that can target both have recently been identified through screening of com‑ mTORC1 and mTORC2 would be expected to block pound libraries and application of an iterative analogue activation of the PI3K pathway more effectively than library synthesis approach. These allosteric AKT inhibitors rapamycin. Recent studies have described torkinibs and have shown some level of isoform selectivity (reviewed in torin1 — potent and selective ATP‑competitive inhibi‑ ReF. 76). AKTi‑1/2, a naphthyridinone allosteric dual inhibi‑ tors of mTOR that inhibit both mTORC1 and mTORC2 tor of AKT1 and AKT2, has potent antitumour activity in complexes and impair cell growth and proliferation more tumour xenograft models, and its analogue mK2206 is in a effectively than rapamycin85,86. Interestingly, however, the Phase I trial in patients with locally advanced or metastatic enhanced activity of these mTOR kinase inhibitors may solid tumours (FIG. 4; TABLe 2). not be due to mTORC2 inhibition. Instead, the enhanced Xl418, a small molecule that inhibits AKT and S6K activity seems to be due to more complete inhibition and has shown antineoplastic activity in preclinical studies, of mTORC1 activity, as indicated by measurements of is in Phase I clinical trials in patients with advanced solid mTORC1‑dependent and rapamycin‑independent 4EBP1 tumours (FIG. 4; TABLe 2). vQD‑002 (triciribine phosphate phosphorylation and cap‑dependent translation85,86. monohydrate), a water‑soluble tricyclic nucleotide, is Two ATP‑competitive mTOR inhibitors, OSI‑027 and currently being tested in Phase I clinical trials in patients AZD8055, are currently in clinical trials in patients with both solid and haematological malignancies (FIG. 4; with advanced solid tumours and lymphoma (FIG. 4; TABLe 2). It was recently reported that this compound TABLe 2). could play a part in reversing drug resistance in ovarian It should also be noted that several kinases in the PI3K cancer in patients who had previously undergone chemo‑ pathway are client proteins for the heat shock protein 90 therapy77; however, its mechanism of action is unclear. (HSP90)87,88. Therefore, compounds that inhibit HSP90, such as geldanamycin and its analogues, may have thera‑ Targeting mTOR. Although mTOR was only recently peutic effects mediated at least in part through inhibition defined as a member of the PI3K pathway, it is the of the PI3K pathway (reviewed in ReFs 68,89,90). first node of the pathway to be targeted in the clinic. Rapamycin (also known as sirolimus) (Rapamune; Ongoing issues and challenges wyeth), the prototypical mTOR inhibitor, is a bacterially Unravelling the specific roles of PI3K isoforms: implications derived natural product that was originally used as an for drug development. The two isoforms of class I PI3Ks antifungal agent78. It was subsequently found to have that are most widely expressed outside of the immune immunosuppressive79 and, more recently, anti‑neoplastic system in mammals are p110α and p110β, both of which properties (reviewed in ReFs 21,80–82). Rapamycin asso‑ are expressed in almost all tissue and cell types. As both ciates with its intracellular receptor, FK506‑binding pro‑ isoforms need to form a complex with the p85 adaptor tein 12 (FKBP12), which then binds directly to mTORC1 to bind to RTKs, and both use the same substrates and and suppresses mTOR‑mediated phosphorylation of its generate the same lipid products, it was long thought downstream substrates, S6K and 4EBP1 (ReFs 21,80). that they functioned redundantly in cellular physiology. Analogues of rapamycin, such as temsirolimus (CCI‑ However, over a decade ago, it was found that mice with 779/Torisel; wyeth) and everolimus (RAD001/Afinitor; homozygous germline deletion of either p110α or p110β Novartis) have been developed as anti‑cancer drugs. die early during embryonic development91,92, suggesting These rapamycin analogues, sometimes referred to as distinct roles for each isoform during embryogenesis. rapalogues, inhibit mTOR through the same mechanism more recently, several groups have created mice with as does rapamycin, but have better pharmacological conditional knockout of p110α and p110β and mice properties for clinical use in cancer. The results of many with germline knock‑in of alleles that encode kinase mTOR inhibitor studies in patients with cancer have dead p110α or p110β93–98 (TABLe 3). Studies with these been described (reviewed in ReF. 80). AP23573 has been mice have revealed that the two PI3K isoforms have designated orphan drug status for the treatment of soft‑ markedly different roles in cellular signalling, growth tissue and bone sarcomas. Results from recent clinical and oncogenic transformation (FIG. 5a). The p110α iso‑ studies with temsirolimus and everolimus used as form performs most of the functions that are commonly single agents showed that these drugs improved survival assigned to PI3K in the literature. For example, p110α in patients with advanced renal cell carcinoma, leading is responsible for most of the signalling downstream of to approval of these compounds for this indication80,83. RTKs and oncogenes such as RAS and polyoma middle

Table 3 | Mouse models with class I PI3K mutations genotype Phenotypes references p110α p110α–/– Embryonic lethality (E10.5) 92 p110αD933A–D933A Embryonic lethality (E10.5); severe vascular abnormalities at E10.5 97,98 p110α+/D933A Defective growth and metabolic regulation associated with 98 hyperinsulinaemia and glucose intolerance p110αRBD–RBD Defective lymphatic development; a small fraction survived into adulthood, 126 but had proliferative defects and altered growth factor signalling to PI3K; protected from Kras‑driven tumorigenesis in a lung cancer model Endothelial p110α–/– Severe vascular abnormalities at E10.5 and death before E12.5 97 Prostate p110α–/– Normal for prostate development; not protected from PTEN‑loss‑induced 93 high‑grade PIN p110b p110b–/– Embryonic lethality (E3.5) 91 p110bK805R–K805R Some survived to adulthood but had retarded growth and mild insulin 95 resistance with age; attenuated Erbb2‑driven mammary tumour development Liver p110b–/– Impaired insulin sensitivity and glucose homeostasis 93 Prostate p110b–/– Normal for prostate development; protected from PTEN‑loss‑induced high 93 grade PIN p110δ p110δ –/– Mice were viable but had impaired B cell and NK cell development and 180–186 functions; decreased immunoglobulin levels and a defective humoral response; and impaired neutrophil chemotaxis p110δD910A–D910A Mice were viable but had defective B cell, NK cell and mast cell development 180–186 and function; impaired antigen receptor signalling in B cells and T cells; and attenuated immune and allergic responses p110γ p110γ–/– Mice were viable but had reduced insulin secretion; increased insulin 13, 187–192 sensitivity and b‑cell mass; impaired mast cell functions and inflammatory response; reduced neutrophil and macrophage migration and oxidative burst; and increased heart contractility p110γKD–KD Mice were viable but had reduced inflammatory reactions, with no alterations 12 in cardiac contractility p110δ–/– p110γ–/– Mice were viable but had severe defects in T cell and NK cell development and 193–195 functions p85 p85α–/– Hypoglycaemia and hypoinsulinaemia; impaired B cell development and 196,197 functions but normal T cell activation p55α–/– p50α–/– Mice were viable but had enhanced insulin sensitivity 198 p85α–/– p55α–/–p50α–/– Perinatal death; liver necrosis and hypoglycaemia; increased insulin 199,200 sensitivity; and impaired B cell development and functions p85b–/– Improved insulin sensitivity, and increased T cell proliferation and 201 accumulation in response to various stimuli Liver p85α–/– p55α–/– Defects in glucose and lipid homeostatis; hyperinsulinaemia and 202 p50α–/– p85b–/– hypolipidaemia Muscle p85α–/– p55α–/– Mice were viable but had reduced muscle growth and insulin response, and 203 p50α–/– p85b–/– hyperlipidaemia Endothelial p85α–/– Acute embryonic lethality at E11.5 owing to haemorrhaging 129 p55α–/–p50α–/– p85b–/– Endothelial p85α+/– Mice were viable but had localized vascular abnormalities when challenged 129 p55α–/– p50α–/– p85b–/– with pathological insults E, embryonic day; KD, kinase dead; NK, natural killer; PI3K, phosphoinositide 3‑kinase; PIN, prostate intraepithelial neoplasia; PTEN, phosphatase and tensin homologue; RBD, RAS‑binding domain.

NATuRE REvIEwS | Drug Discovery vOlumE 8 | AuGuST 2009 | 637 © 2009 Macmillan Publishers Limited. All rights reserved REVIEWS a • Angiogenesis There is considerable evidence indicating that targeting a single isoform of PI3K or other pathway members may RTK p110α • Metabolism be sufficient to block a particular tumour type, suggesting the potential desirability of generating isoform‑specific • Cell growth inhibitors. Drugs that target single isoforms might • Proliferation avoid toxicity to the immune system, which is largely Cancer GPCR p110β • Inflammation dependent on p110δ and p110γ for function. Similarly, because p110α and p110β seem to have distinct roles in • Vascular trafficking multiple cellular processes (FIG. 5a), it is possible that a drug aimed at either target would have fewer side effects • Platelet thrombus formation than one that inhibits both. As p110α is important for the growth and maintenance of numerous tumours that b feature PI3K activation, several companies are already PIK3CA generating p110α isoform‑specific inhibitors (reviewed in ReFs 63,68,101). It is hoped that these compounds will RAS RTK GPCR bypass the problems of inhibiting p110γ and p110δ while targeting the PI3K pathway in many tumour types. Interestingly, recent studies on genetically engineered p110α p110α p110β p110β inhibitor inhibitor mouse models and chemical inhibitors, as well as limited short hairpin RNA experiments, suggest that tumours PtdIns(4,5)P PtdIns(3,4,5)P PtdIns(4,5)P 2 3 2 driven by PTEN loss may be sensitive to inhibition of PTEN loss PTEN loss p110β rather than p110α93,102,103. The exact mechanism by which p110β drives PTEN‑null tumours has not been elucidated. Perhaps ligands such as lysophosphatidic acid that activate GPCRs drive PI3K activation in PTEN‑null tumours, or perhaps PTEN loss allows a p110β‑specific Cancer mechanism of basal PtdIns(3,4,5)P3 synthesis to drive Figure 5 | Function and therapeutic targeting of p110 isoforms. a | Differential tumour formation93. It may be beneficial to consider the functions of p110α and p110β isoforms. p110α is the major effector downstream generation of p110β‑selective compounds, especially of receptor tyrosine kinases (RTKs) and p110 is an effectorNatur fore Re bothviews RTKs | Drug and Disc overy β as p110β seems to have a smaller role in insulin action G protein‑coupled receptors (GPCRs). Their differential roles in many biological 93,95 functions are indicated. Many of these roles are associated with cancer. than p110α . There are a few inhibitors — for example, An association between chronic inflammation and cancer is indicated. b | Schematic TGX‑115, TGX‑286 and TGX‑221 — which are selec‑ of phospho inositide 3‑kinase (PI3K) isoform‑selective inhibition in the treatment of tive for p110β relative to all other PI3K isoforms except cancers that feature specific oncogenic alterations. Recent studies suggest that p110δ64,100. Among them, TGX‑221 is perhaps the p110β is a primary target for PTEN (phosphatase and tensin homologue)‑deficient p110β‑selective tool compound that is most commonly cancers. However, in the case of oncogenic alterations — such as RTK amplification used to investigate p110β functions. It was shown to or mutation, RAS mutation or activating mutations in the PI3K catalytic subunit inhibit platelet aggregation and thrombosis100 and is also α‑isoform gene (PIK3CA) — the PI3K signalling largely depends on p110α, perhaps capable of suppressing the activation of PI3K and prolif‑ even in the absence of PTEN. PtdIns(3,4,5)P , phosphatidylinositol‑3,4,5‑tris‑ 3 eration of PTEN‑null cancer cells103. Further preclinical phosphate; PtdIns(4,5)P , phosphatidylinositol‑4,5‑bisphosphate. 2 development of p110β‑selective inhibitors is necessary to improve their pharmacological properties. However, there is evidence that not all tumours that are driven by T antigen94,99. Ablation of p110α resulted in substantially PTEN loss are dependent on p110β, and the presence of reduced AKT phosphorylation in response to stimulation other genetic alterations is likely to change the PI3K iso‑ by various growth factors, including insulin, epidermal form‑dependence of these PTEN‑null tumours (FIG. 5b). growth factor (EGF) and insulin‑like growth factor94. For example, a number of tumour cell lines that feature Cells that are deficient in p110α are resistant to onco‑ loss of PTEN in conjunction with activating mutations genic transformation induced by oncogenic alleles of in p110α are sensitive to loss of p110α and not to a loss RTKs94. Conversely, ablation of p110β has little effect on of p110β103,104. AKT phosphorylation in response to RTK signalling93. Given the essential roles of p110α in cellular physio‑ Instead, p110β, like p110γ, preferentially transduces logy, the development of inhibitors specific for the signals from GPCRs through a mechanism that has yet to mutant form of p110α that is found in tumours would be be elucidated93,95,96. As p110γ expression is largely limited a particularly attractive route to therapy. Such inhibitors to leukocytes whereas p110β has a broad tissue distribu‑ would presumably minimize side effects (that is, altera‑ tion, p110β may play an essential part in coupling GPCR tion in insulin signalling) that will almost certainly be signals to the PI3K pathway in cells or tissues outside of associated with inhibition of the wild‑type p110α, espe‑ the immune system. In addition, p110β has been shown cially during prolonged treatment. It remains a great to have an important role in integrin‑mediated platelet challenge for researchers to identify mutant‑specific thrombosis100 Thrombosis adhesion and arterial . Interestingly, p110β small‑molecule inhibitors that target the catalytic centre The formation or presence of also seems to possess important kinase‑independent of the mutant but not the wild‑type kinase. The structure a blood clot in a blood vessel. functions93,95. of a complex between wild‑type p110α and the iSH2

domain of p85 has recently been reported52. It reveals those that indicate drugs are reaching intended targets many features that are common to well‑characterized in the PI3K pathway, and those that can be used to protein kinases, including a hydrophobic ATP‑binding accurately predict which patients are likely to respond. pocket. most small‑molecule inhibitors targeting PI3Ks Readily obtainable tissues, such as skin, hair follicles and that are currently in development work by binding to peripheral blood mononuclear cells, have been used as the ATP‑binding pocket and thus competing with ATP, surrogate tissues to assess the effect of PI3K inhibi‑ which is the mechanism of most protein kinase inhibi‑ tors that are currently undergoing clinical trials113. The tors. The structure of wild‑type p110α suggests that the molecular markers of PI3K inhibition that are commonly H1047R mutation in the kinase domain improves sub‑ used in the clinic are the levels of phosphorylated AKT strate binding through a direct effect on the conforma‑ and phosphorylated S6K1 in biopsies of these surrogate tion of the activation loop. Given the proximity of the tissues and, when possible, tumour tissue. However, activation loop to the ATP‑binding pocket, the genera‑ there is great variability in the robustness and reproduc‑ tion of an inhibitor that is specific for a kinase domain ibility of the use of biomarkers for measuring the efficacy mutant might be feasible using existing drug scaffolds. of PI3K pathway inhibitors. Thus, identification of new It could be more difficult to target the mutated helical and potentially more robust biomarkers is a crucially domain of the kinase (E542K or E545), as these mutations important part of the preclinical development of PI3K seem to affect protein–protein interaction by eliminating inhibitors and in the conduct and interpretation of clinical an auto‑inhibitory contact with the iSH2 domain of p85 studies using these inhibitors. (ReF. 51). Preclinical studies have indicated that PI3K and AKT inhibition can be assessed by measuring blood insulin The effects of feedback loops, pathway crosstalk and levels, which are increased owing to disruption of insulin signalling circuitry on therapeutic outcomes. A hallmark signalling through the PI3K–AKT pathway64,98,114,115. of signalling networks is the presence of multiple nodes Studies in mice in which PI3Ks or AKT is genetically with feedback loops and crosstalk between pathways. inactivated93,95,98,114 also suggested that changes in blood Two negative feedback loops have been described, glucose levels in response to PI3K inhibition may be involving S6K and Jun N‑terminal kinase (JNK), which exploited clinically. However, early clinical results sug‑ attenuate insulin‑induced PI3K activation through gest that these effects might be transient and difficult to IRS1 (ReFs 105–107). S6K‑ or JNK‑knockout mice show measure. The PI3K pathway plays a major part in insulin‑ increased insulin sensitivity in response to high‑fat mediated glucose transport and metabolism, and so inhi‑ diets compared with wild‑type mice108,109. Perturbing bition of the PI3K pathway in tumours can be measured these feedback loops can have dramatic effects on drug by scanning positron emission tomography with fluoro‑ responses, as exemplified by the response of certain deoxy‑d‑glucose as a probe. A recent study showed that tumours to rapalogues. when mTOR is activated, it can the effect of BEZ235 in decreasing tumour vasculature initiate a signalling cascade through S6K1 that results permeability could be monitored by dynamic contrast‑ in a feedback loop which downregulates PI3K and AKT enhanced magnetic resonance imaging116. These types of activity. Therefore, when tumours in which this loop molecular imaging offer a minimally invasive approach is activated are treated with rapalogues, the net effect to determine the efficacy of targeted PI3K inhibition117, can be increased AKT activity, which can ultimately and could be predictive of clinical outcome. enhance tumour growth110. PI3K or AKT inhibitors should not cause this problem, but may suffer from Sensitivity and resistance to PI3K-targeted therapies. issues that arise from crosstalk with other pathways. One of the key lessons drawn from targeted therapies Specifically, in cancers bearing mutant RTKs, or onco‑ has been the importance of matching the therapy to genes such as RAS that activate both the RAF–mitogen‑ the patient. The presence of an activating mutation or activated protein kinase (mAPK) and PI3K pathways, other genomic alteration in the targeted kinase is per‑ blocking the PI3K pathway can actually upregulate haps the most accurate predictor of potential success. signalling of the RAF–mAPK pathway because the two Examples include the use of imatinib (Gleevec; Novartis) pathways have cross‑inhibitory effects111. RAF–mAPK to treat patients with chronic myelogenous leukae‑ pathway signalling can in turn drive tumour growth, mia (Cml) featuring BCR–ABL fusions; trastuzumab counteracting the effect of the PI3K pathway inhibi‑ (Herceptin; Genentech/Roche) and lapatinib (Tyverb; tion. It was recently shown that mTORC1 inhibition GlaxoSmithKline) to treat patients with breast tumours led to the activation of mAPK in a PI3K‑dependent expressing human epidermal growth factor receptor 2 manner, providing another example of such signalling (HER2; also known as ERBB2); and gefitinib (Iressa; feedback loops and crosstalk112. As discussed below, AstraZeneca) and erlotinib (Tarceva; Genentech/OSI/ combined inhibition of both pathways for therapy may Roche) for patients with lung cancer bearing mutations Biomarker alleviate this problem. and/or amplification of the EGF receptor. This logic sug‑ A characteristic that can be gests that PI3K inhibitors will be effective in tumours objectively measured and Identifying biomarkers that predict drug responses. For that feature activating mutations in p110α or loss of evaluated as an indicator of PI3K pathway inhibitors, as with all targeted therapies, PTEN, and AKT mutations will sensitize a tumour to normal biological processes, biomarkers pathogenic processes, or it is crucial to develop clinically tractable that AKT inhibitors. It also seems likely that targeting a path‑ pharmacological responses to are predictive of drug response. Biomarkers for PI3K way immediately downstream of the genetic alteration a therapeutic intervention. pathway inhibitors can be divided into two categories: will be effective. Therefore, inhibitors of PI3K and AKT

could be useful for the treatment of tumours that feature kinases and PI3Ks has been described125. A single agent activated RTKs or oncogenic RAS (perhaps in combina‑ with dual specificity may have the added advantage of tion with inhibitors of RAF and mAPK). Finally, inhibi‑ being less likely to induce drug resistance than mono‑ tors of mTOR and kinases further downstream in the specific agents. Clinical resistance to a kinase inhibi‑ pathway might also be effective in tumours that feature tor often arises through second‑site mutations in the PI3K activation. However, it is not clear how distant a targeted kinase. Targeting two kinases simultaneously mutation can be from the targeted node in the pathway greatly diminishes the possibility of resistance, as it is and it still be sensitive to the inhibition. unlikely that a given tumour can generate two resistant Because most tumours are genetically complex, it is kinases during the course of a single drug treatment. Of likely that alterations in genes other than those encoding course, this assumes that both kinases are essential for PI3K will predispose a tumour to be sensitive or, even tumour survival and/or growth. more likely, resistant to PI3K‑targeted therapy. An obvi‑ ous source of resistance is a mutation or amplification Combining PI3K pathway inhibitors with drugs that of a downstream pathway component. Just as mutations target other pathways. Although knockout of PI3K iso‑ of p110α or loss of PTEN can render HER2‑positive forms can block oncogenic transformation that is driven tumours resistant to trastuzumab118,119, it is likely that by various activated RTKs and oncogenes93,94, targeting tumours with amplifications or mutations in various PI3K may not be sufficient to cause regression of estab‑ downstream kinases will block the action of inhibitors lished tumours. For example, mice in which p110α was that target their upstream components. It is therefore mutated to ablate its binding to RAS are resistant to important to identify and select patients who are likely lung tumour development driven by activated KRAS126. to respond to PI3K‑targeted cancer therapy. However, the same KRAS lung tumour model is insen‑ These resistance events can be either primary or sitive to PI3K inhibition by BEZ235 after the tumours acquired during therapy. For example, so‑called ‘gate‑ have formed. In this case, a combination of PI3K and keeper’ mutations may occur after targeted protein RAF pathway inhibition by BEZ235 plus a mEK1 kinase therapy (reviewed in ReF. 120). These mutations (mAPK kinase 1) and mEK2 inhibitor (ARRY142886; occur in the kinase domain of the targeted kinase, at an also known as ADZ6244), effectively induced tumour amino‑acid residue known as the gatekeeper, and block regression in mice127. Consistent with these findings, binding of the inhibitor while allowing catalysis to pro‑ activation of the mEK effector pathway downstream of ceed. A functional screen against a structurally diverse RAS was found to be responsible for resistance to PI3K panel of PI3K inhibitors has identified a potential ‘hot inhibitors in tumour cell lines harbouring RAS muta‑ spot’ for resistance mutations in p110α, but surprisingly tions. Therefore, the combination of mEK inhibitors found a lack of resistance mutations at the gatekeeper and PI3K inhibitors synergistically blocked growth of residues121. Other known resistance mechanisms tumour cells that expressed oncogenic RAS102,128. Recent include the activation of alternative pathways, such as comprehensive genomic studies in cancer have revealed the induced gefitinib resistance through reactivation of that tumours with PI3K mutations or PTEN loss often HER3 signalling in lung cancer as a result of incomplete harbour other genetic alterations that can act independ‑ HER2 inhibition or amplification of hepatocyte growth ently to promote tumour development (reviewed in factor receptor (also known as mET)122–124. A compa‑ ReF. 129). The presence of these or other genetic altera‑ rable mechanism in the case of PI3K inhibitors would tions is likely to change the sensitivity of tumours to PI3K be the activation of the RAF–mAPK pathway discussed inhibition. Therefore, combinations of PI3K inhibitors above. Current preclinical research and clinical studies with other targeted drugs may achieve optimal clinical will undoubtedly reveal more resistance mechanisms benefits (FIG. 3a). and facilitate the development of therapeutic strategies to overcome drug resistance. Angiogenesis as a specific target of the PI3K pathway. Interestingly, inhibition of the PI3K pathway may be Emerging candidates and approaches effective against tumours by two distinct mechanisms: The effectiveness of simultaneously targeting two kinases by directly blocking tumour cell growth and by inhib‑ in the pathway. A number of the early PI3K inhibitor iting tumour angiogenesis. It is notable that the PI3K clinical candidates are dual‑specificity drugs, targeting pathway plays an important part in the production of not only several PI3K isoforms but also the kinase activity vascular endothelial cell growth factor (vEGF) and in of mTOR (FIG. 4; TABLe 2). It was easy to generate this the signalling of the vEGF receptor (vEGFR), which class of compounds because mTOR is in the PI3K super‑ is important for angiogenesis. Rapamycin and its ana‑ family and so bears considerable structural similarity to logues have been studied most extensively in the clinic class I PI3Ks. A strong argument can be made that by as anti‑angiogenic agents. Rapamycin was found to targeting two nodal points in the pathway concurrently, have anti‑angiogenic activities associated with mark‑ a compound may be more efficacious than if it has only edly reduced production of vEGF, and it completely Tumour angiogenesis a single target (FIG. 3). For example, PI‑103 was found to abrogated the response of vascular endothelial cells to The formation of new blood be a potent inhibitor of both PI3K and mTOR, and was stimulation by vEGF130. when used at low (minimally vessels that grow into the tumour, supplying nutrients unexpectedly effective at blocking the growth of aggres‑ immunosuppressive) doses, rapamycin significantly inhib‑ 65 130 and oxygen to assist tumour sive glioma cells in vivo and in vitro . A second class ited the growth of established vascularized tumours . growth. of inhibitors that selectively target particular tyrosine This work indicated that rapamycin might affect tumour

Cancer chemoprevention growth primarily through its anti‑angiogenic properties, Conclusions and future directions The use of chemical leading to the suggestion that it could be particularly The PI3K pathway clearly presents both a great thera‑ compounds to intervene in the effective in treating highly vascularized tumours. peutic opportunity and a tremendous challenge for can‑ early precancerous stages Indeed, rapamycin significantly inhibits the progres‑ cer therapy. As compounds that target PI3K (or AKT of carcinogenesis, thereby preventing tumour formation. sion of Kaposi’s sarcoma in which the driving oncogenic and mTOR) progress through clinical trials, potential alteration is presumed to be activated vEGF–vEGFR issues associated with toxicity and resistance can be signalling131. In renal cell carcinoma, loss of the von expected. PI3K mutants that are resistant to a kinase Hippel–lindau tumour suppressor (VHL) that normally inhibitor may arise during treatment, as has been seen inhibits HIF1A (hypoxia‑inducible factor 1, α subunit) for BCR–ABl inhibition by imatinib in the treatment of leads to enhanced vascularization and sensitizes cancer Cml. Oncogenic changes in other components in the cells to the mTOR inhibitor CCI‑779 (ReF. 132). From a PI3K pathway, or other parallel and/or interconnected mechanistic perspective, rapamycin inhibits mTORC1‑ pathways, may also render cancer cells resistant to PI3K dependent translation and action of HIF1A, and thereby inhibition. It is therefore important to identify new thera‑ decreases vEGF production133,134. peutic targets for the development of drugs that may be Data from preclinical studies in tissue culture and used either in place of PI3K inhibitors or to enhance the mouse models suggest that an anti‑angiogenic effect efficacy of PI3K inhibitors at subtoxic doses. may well contribute to the anti‑tumour effects of inhib‑ Finally, it is worth noting that reduced expression of iting PI3K or AKT97,116,129,135,136. Recently, it was shown PDPK1 (ReF. 137), AKT1 (ReF. 138) or p110α (T.m.R. and that genetic ablation of class IA PI3K specifically in J.J.Z., unpublished observations) can suppress tumour the endothelium resulted in impaired vessel integrity formation in animal models. This suggests that it might during development as well as tumour angiogenesis be possible to use low levels of inhibitors of these enzymes in mice129. Furthermore, it has been shown that angio‑ to block tumorigenesis in certain circumstances. Familial genesis specifically requires the p110α isoform, as it diseases such as Cowden’s syndrome, which features is crucial in mediating vEGF signalling and control‑ germline loss of PTEN, might be treated in this manner. ling endothelial cell migration97. Indeed, inhibition of If the inhibitors are present before tumour initiation, tumour vasculature development in xenograft tumour cancer chemoprevention could be possible. Similarly, a chemo‑ models by BEZ235, a dual PI3K–mTOR inhibitor, was preventative approach might work to block outgrowth of correlated with inhibition of PI3K and AKT, but not cells that have metastasized to distant sites after successful with inhibition of mTORC1 (ReFs 116,129). It is quite treatment of a primary tumour. Further experimentation possible that PI3K or AKT inhibitors may be as effec‑ will be required to determine whether there are doses that tive as rapamycin analogues in the treatment of highly are sufficient to block tumour outgrowth but engender vascularized tumours. only minimal side effects during long‑term therapy.

1. Vivanco, I. & Sawyers, C. L. The phosphatidylinositol 11. Chang, J. D. et al. Deletion of the phosphoinositide 23. Manning, B. D., Tee, A. R., Logsdon, M. N., Blenis, J. & 3‑kinase AKT pathway in human cancer. Nature Rev. 3‑kinase p110γ gene attenuates murine Cantley, L. C. Identification of the tuberous sclerosis Cancer 2, 489–501 (2002). atherosclerosis. Proc. Natl Acad. Sci. USA 104, complex‑2 tumor suppressor gene product tuberin as 2. Bader, A. G., Kang, S., Zhao, L. & Vogt, P. K. 8077–8082 (2007). a target of the phosphoinositide 3‑kinase/akt pathway. Oncogenic PI3K deregulates transcription and 12. Patrucco, E. et al. PI3Kγ modulates the cardiac Mol. Cell 10, 151–162 (2002). translation. Nature Rev. Cancer 5, 921–929 response to chronic pressure overload by distinct 24. Vander Haar, E., Lee, S. I., Bandhakavi, S., Griffin, T. J. (2005). kinase‑dependent and ‑independent effects. Cell & Kim, D. H. Insulin signalling to mTOR mediated by 3. Engelman, J. A., Luo, J. & Cantley, L. C. The evolution 118, 375–387 (2004). the Akt/PKB substrate PRAS40. Nature Cell Biol. 9, of phosphatidylinositol 3‑kinases as regulators 13. Sasaki, T. et al. Function of PI3Kγ in thymocyte 316–323 (2007). of growth and metabolism. Nature Rev. Genet. 7, development, T cell activation, and neutrophil 25. Crino, P. B., Nathanson, K. L. & Henske, E. P. 606–619 (2006). migration. Science 287, 1040–1046 (2000). The tuberous sclerosis complex. N. Engl. J. Med. A recent comprehensive review on PI3K. 14. Backer, J. M. The regulation and function of Class III 355, 1345–1356 (2006). 4. Cully, M., You, H., Levine, A. J. & Mak, T. W. Beyond PI3Ks: novel roles for Vps34. Biochem. J. 410, 1–17 26. Hay, N. & Sonenberg, N. Upstream and downstream PTEN mutations: the PI3K pathway as an integrator (2008). of mTOR. Genes Dev. 18, 1926–1945 (2004). of multiple inputs during tumorigenesis. Nature Rev. 15. Scheid, M. P. & Woodgett, J. R. PKB/AKT: functional 27. Zhao, J. J. & Roberts, T. M. PI3 kinases in cancer: Cancer 6, 184–192 (2006). insights from genetic models. Nature Rev. Mol. Cell Biol. from oncogene artifact to leading cancer target. 5. Cantley, L. C. & Neel, B. G. New insights into tumor 2, 760–768 (2001). Sci. STKE 2006, pe52 (2006). suppression: PTEN suppresses tumor formation 16. Alessi, D. R. et al. 3‑Phosphoinositide‑dependent 28. Wood, L. D. et al. The genomic landscapes of human by restraining the phosphoinositide 3‑kinase/AKT protein kinase‑1 (PDK1): structural and functional breast and colorectal cancers. Science 318, 1108–1113 pathway. Proc. Natl Acad. Sci. USA 96, 4240–4245 homology with the Drosophila DSTPK61 kinase. (2007). (1999). Curr. Biol. 7, 776–789 (1997). 29. Thomas, R. K. et al. High‑throughput oncogene 6. Hennessy, B. T., Smith, D. L., Ram, P. T., Lu, Y. & 17. Stephens, L. et al. Protein kinase B kinases that mediate mutation profiling in human cancer. Nature Genet. Mills, G. B. Exploiting the PI3K/AKT pathway for phosphatidylinositol 3,4,5‑trisphosphate‑dependent 39, 347–351 (2007). cancer drug discovery. Nature Rev. Drug Discov. activation of protein kinase B. Science 279, 710–714 30. Comprehensive genomic characterization defines 4, 988–1004 (2005). (1998). human glioblastoma genes and core pathways. 7. Fruman, D. A., Meyers, R. E. & Cantley, L. C. 18. Sarbassov, D. D., Guertin, D. A., Ali, S. M. & Nature 455, 1061–1068 (2008). Phosphoinositide kinases. Annu. Rev. Biochem. 67, Sabatini, D. M. Phosphorylation and regulation of 31. Parsons, D. W. et al. An integrated genomic analysis 481–507 (1998). Akt/PKB by the rictor‑mTOR complex. Science 307, of human glioblastoma multiforme. Science 321, 8. Katso, R. et al. Cellular function of phosphoinositide 1098–1101 (2005). 1807–1812 (2008). 3‑kinases: implications for development, homeostasis, 19. Manning, B. D. & Cantley, L. C. AKT/PKB signaling: 32. Samuels, Y. et al. High frequency of mutations of the and cancer. Annu. Rev. Cell Dev. Biol. 17, 615–675 navigating downstream. Cell 129, 1261–1274 PIK3CA gene in human cancers. Science 304, 554 (2001). (2007). (2004). 9. Voigt, P., Dorner, M. B. & Schaefer, M. 20. Wullschleger, S., Loewith, R. & Hall, M. N. TOR signaling This study showed that PIK3CA is frequently Characterization of p87PIKAP, a novel regulatory in growth and metabolism. Cell 124, 471–484 (2006). mutated in human cancer. subunit of phosphoinositide 3‑kinase gamma that is 21. Sabatini, D. M. mTOR and cancer: insights into a 33. Ding, L. et al. Somatic mutations affect key pathways highly expressed in heart and interacts with PDE3B. complex relationship. Nature Rev. Cancer 6, 729–734 in lung adenocarcinoma. Nature 455, 1069–1075 J. Biol. Chem. 281, 9977–9986 (2006). (2006). (2008). 10. Suire, S. et al. p84, a new Gβγ‑activated regulatory 22. Inoki, K., Li, Y., Zhu, T., Wu, J. & Guan, K. L. TSC2 is 34. Sansal, I. & Sellers, W. R. The biology and clinical subunit of the type IB phosphoinositide 3‑kinase phosphorylated and inhibited by Akt and suppresses relevance of the PTEN tumor suppressor pathway. p110γ. Curr. Biol. 15, 566–570 (2005). mTOR signalling. Nature Cell Biol. 4, 648–657 (2002). J. Clin. Oncol. 22, 2954–2963 (2004).

35. Steck, P. A. et al. Identification of a candidate tumour 59. Davies, M. A. et al. A novel AKT3 mutation in 84. Guertin, D. A. et al. mTOR complex 2 is required for suppressor gene, MMAC1, at chromosome 10q23.3 melanoma tumours and cell lines. Br. J. Cancer the development of prostate cancer induced by Pten that is mutated in multiple advanced cancers. Nature 99, 1265–1268 (2008). loss in mice. Cancer Cell 15, 148–159 (2009). Genet. 15, 356–362 (1997). 60. Vanhaesebroeck, B. & Alessi, D. R. The PI3K‑PDK1 85. Feldman, M. E. et al. Active‑site inhibitors of mTOR 36. Salmena, L., Carracedo, A. & Pandolfi, P. P. Tenets connection: more than just a road to PKB. Biochem. J. target rapamycin‑resistant outputs of mTORC1 and of PTEN tumor suppression. Cell 133, 403–414 (2008). 346, 561–576 (2000). mTORC2. PLoS Biol. 7, e38 (2009). 37. Parsons, R. Human cancer, PTEN and the PI‑3 kinase 61. Hunter, C. et al. A hypermutation phenotype and 86. Thoreen, C. C. et al. An ATP‑competitive mammalian pathway. Semin. Cell Dev. Biol. 15, 171–176 (2004). somatic MSH6 mutations in recurrent human target of rapamycin inhibitor reveals rapamycin‑ 38. Knobbe, C. B., Lapin, V., Suzuki, A. & Mak, T. W. malignant gliomas after alkylator chemotherapy. resistant functions of mTORC1. J. Biol. Chem. 284, The roles of PTEN in development, physiology and Cancer Res. 66, 3987–3991 (2006). 8023–8032 (2009). tumorigenesis in mouse models: a tissue‑by‑tissue 62. Knight, Z. A. & Shokat, K. M. Chemically targeting the 87. Fujita, N., Sato, S., Ishida, A. & Tsuruo, T. Involvement survey. Oncogene 27, 5398–5415 (2008). PI3K family. Biochem. Soc. Trans. 35, 245–249 (2007). of Hsp90 in signaling and stability of 3‑phospho‑ 39. Li, J. et al. PTEN, a putative protein tyrosine 63. Marone, R., Cmiljanovic, V., Giese, B. & Wymann, M. P. inositide‑dependent kinase‑1. J. Biol. Chem. 277, phosphatase gene mutated in human brain, breast, Targeting phosphoinositide 3‑kinase: moving towards 10346–10353 (2002). and prostate cancer. Science 275, 1943–1947 (1997). therapy. Biochim. Biophys. Acta 1784, 159–185 88. Solit, D. B., Basso, A. D., Olshen, A. B., Scher, H. I. & 40. Di Cristofano, A., De Acetis, M., Koff, A., (2008). Rosen, N. Inhibition of heat shock protein 90 function Cordon‑Cardo, C. & Pandolfi, P. P. Pten and p27KIP1 64. Knight, Z. A. et al. A pharmacological map of the down‑regulates Akt kinase and sensitizes tumors to cooperate in prostate cancer tumor suppression in PI3‑K family defines a role for p110α in insulin Taxol. Cancer Res. 63, 2139–2144 (2003). the mouse. Nature Genet. 27, 222–224 (2001). signaling. Cell 125, 733–747 (2006). 89. Solit, D. B. & Rosen, N. Hsp90: a novel target for cancer 41. Di Cristofano, A., Pesce, B., Cordon‑Cardo, C. & A comparative analysis of the roles of PI3K therapy. Curr. Top. Med. Chem. 6, 1205–1214 (2006). Pandolfi, P. P. Pten is essential for embryonic isoforms using small-molecule inhibitors. 90. Workman, P., Burrows, F., Neckers, L. & Rosen, N. development and tumour suppression. Nature Genet. 65. Fan, Q. W. et al. A dual PI3 kinase/mTOR inhibitor Drugging the cancer chaperone HSP90: combinatorial 19, 348–355 (1998). reveals emergent efficacy in glioma. Cancer Cell 9, therapeutic exploitation of oncogene addiction and 42. Wang, S. et al. Prostate‑specific deletion of the murine 341–349 (2006). tumor stress. Ann. NY Acad. Sci. 1113, 202–216 Pten tumor suppressor gene leads to metastatic 66. Maira, S. M. et al. Identification and characterization (2007). prostate cancer. Cancer Cell 4, 209–221 (2003). of NVP‑BEZ235, a new orally available dual 91. Bi, L., Okabe, I., Bernard, D. J. & Nussbaum, R. L. 43. Stambolic, V. et al. High incidence of breast and phosphatidylinositol 3‑kinase/mammalian target of Early embryonic lethality in mice deficient in the endometrial neoplasia resembling human Cowden rapamycin inhibitor with potent in vivo antitumor p110β catalytic subunit of PI 3‑kinase. Mamm. syndrome in pten+/– mice. Cancer Res. 60, 3605–3611 activity. Mol. Cancer Ther. 7, 1851–1863 (2008). Genome 13, 169–172 (2002). (2000). 67. Serra, V. et al. NVP‑BEZ235, a dual PI3K/mTOR 92. Bi, L., Okabe, I., Bernard, D. J., Wynshaw‑Boris, A. & 44. Stemke‑Hale, K. et al. An integrative genomic and inhibitor, prevents PI3K signaling and inhibits the Nussbaum, R. L. Proliferative defect and embryonic proteomic analysis of PIK3CA, PTEN, and AKT growth of cancer cells with activating PI3K mutations. lethality in mice homozygous for a deletion in the mutations in breast cancer. Cancer Res. 68, Cancer Res. 68, 8022–8030 (2008). p110α subunit of phosphoinositide 3‑kinase. J. Biol. 6084–6091 (2008). 68. Garcia‑Echeverria, C. & Sellers, W. R. Drug discovery Chem. 274, 10963–10968 (1999).

45. Zhao, J. J. et al. The oncogenic properties of mutant approaches targeting the PI3K/Akt pathway in cancer. 93. Jia, S. et al. Essential roles of PI3K‑p110β in cell growth, p110α and p110β phosphatidylinositol 3‑kinases in Oncogene 27, 5511–5526 (2008). metabolism and tumorigenesis. Nature (2008). human mammary epithelial cells. Proc. Natl Acad. Sci. 69. Garlich, J. R. et al. A vascular targeted pan This study showed that p110β, not p110α, USA 102, 18443–18448 (2005). phosphoinositide 3‑kinase inhibitor prodrug, SF1126, contributed to PTEN-deficiency-induced prostate 46. Isakoff, S. J. et al. Breast cancer‑associated PIK3CA with antitumor and antiangiogenic activity. Cancer cancer in mouse genetic models. mutations are oncogenic in mammary epithelial cells. Res. 68, 206–215 (2008). 94. Zhao, J. J. et al. The p110α isoform of PI3K is Cancer Res. 65, 10992–11000 (2005). 70. Ihle, N. T. et al. The phosphatidylinositol‑3‑kinase essential for proper growth factor signaling and 47. Samuels, Y. et al. Mutant PIK3CA promotes cell inhibitor PX‑866 overcomes resistance to the oncogenic transformation. Proc. Natl Acad. Sci. USA growth and invasion of human cancer cells. Cancer Cell epidermal growth factor receptor inhibitor gefitinib in 103, 16296–16300 (2006). 7, 561–573 (2005). A‑549 human non‑small cell lung cancer xenografts. 95. Ciraolo, E. et al. Phosphoinositide 3‑kinase p110β 48. Bader, A. G., Kang, S. & Vogt, P. K. Cancer‑specific Mol. Cancer Ther. 4, 1349–1357 (2005). activity: key role in metabolism and mammary gland mutations in PIK3CA are oncogenic in vivo. Proc. Natl 71. Hilgard, P. et al. D‑21266, a new heterocyclic cancer but not development. Sci. Signal 1, ra3 (2008). Acad. Sci. USA 103, 1475–1479 (2006). alkylphospholipid with antitumour activity. A study on the role of p110β in metabolism and 49. Lee, J. Y., Engelman, J. A. & Cantley, L. C. Eur. J. Cancer 33, 442–446 (1997). Her2-induced breast cancer using a knock-in mouse Biochemistry. PI3K charges ahead. Science 317, 72. Meuillet, E. J. et al. In vivo molecular pharmacology model. 206–207 (2007). and antitumor activity of the targeted Akt inhibitor 96. Guillermet‑Guibert, J. et al. The p110β isoform 50. Huang, C. H., Mandelker, D., Gabelli, S. B. & PX‑316. Oncol. Res. 14, 513–527 (2004). of phosphoinositide 3‑kinase signals downstream Amzel, L. M. Insights into the oncogenic effects of 73. Gills, J. J. & Dennis, P. A. The development of of G protein‑coupled receptors and is functionally PIK3CA mutations from the structure of p110α/p85α. phosphatidylinositol ether lipid analogues as inhibitors redundant with p110γ. Proc. Natl Acad. Sci. USA Cell Cycle 7, 1151–1156 (2008). of the serine/threonine kinase, Akt. Expert Opin. 105, 8292–8297 (2008). 51. Miled, N. et al. Mechanism of two classes of cancer Investig. Drugs 13, 787–797 (2004). 97. Graupera, M. et al. Angiogenesis selectively requires mutations in the phosphoinositide 3‑kinase catalytic 74. Gills, J. J. et al. Spectrum of activity and molecular the p110α isoform of PI3K to control endothelial cell subunit. Science 317, 239–242 (2007). correlates of response to phosphatidylinositol ether migration. Nature 453, 662–666 (2008). 52. Huang, C. H. et al. The structure of a human p110α/ lipid analogues, novel lipid‑based inhibitors of Akt. This work demonstrated a specific role of p110α p85α complex elucidates the effects of oncogenic Mol. Cancer Ther. 5, 713–722 (2006). in endothelial cell migration. PI3Kα mutations. Science 318, 1744–1748 (2007). 75. Rhodes, N. et al. Characterization of an Akt kinase 98. Foukas, L. C. et al. Critical role for the p110α References 51 and 52 report crystal structural inhibitor with potent pharmacodynamic and antitumor phosphoinositide‑3‑OH kinase in growth and analyses of p110α and p85 iSH2 domains to help activity. Cancer Res. 68, 2366–2374 (2008). metabolic regulation. Nature 441, 366–370 (2006). elucidate the effects of the cancer-associated 76. Lindsley, C. W., Barnett, S. F., Yaroschak, M., This study showed the role of p110α in insulin mutations in p110α. Bilodeau, M. T. & Layton, M. E. Recent progress in signaling and metabolism using an inactive-kinase 53. Mizoguchi, M., Nutt, C. L., Mohapatra, G. & Louis, D. N. the development of ATP‑competitive and allosteric knock-in mouse model. Genetic alterations of phosphoinositide 3‑kinase Akt kinase inhibitors. Curr. Top. Med. Chem. 7, 99. Utermark, T., Schaffhausen, B. S., Roberts, T. M. & subunit genes in human glioblastomas. Brain Pathol. 1349–1363 (2007). Zhao, J. J. The p110α isoform of phosphatidylinositol 14, 372–377 (2004). 77. Yang, H. et al. MicroRNA expression profiling in 3‑kinase is essential for polyomavirus middle T 54. Philp, A. J. et al. The phosphatidylinositol 3’‑kinase human ovarian cancer: miR‑214 induces cell survival antigen‑mediated transformation. J. Virol. 81, p85α gene is an oncogene in human ovarian and colon and cisplatin resistance by targeting PTEN. Cancer 7069–7076 (2007). tumors. Cancer Res. 61, 7426–7429 (2001). Res. 68, 425–433 (2008). 100. Jackson, S. P. et al. PI3‑kinase p110β: a new target for 55. Beeton, C. A., Chance, E. M., Foukas, L. C. & 78. Vezina, C., Kudelski, A. & Sehgal, S. N. Rapamycin antithrombotic therapy. Nature Med. 11, 507–514 Shepherd, P. R. Comparison of the kinetic properties (AY‑22,989), a new antifungal antibiotic. I. Taxonomy of (2005). of the lipid‑ and protein‑kinase activities of the the producing streptomycete and isolation of the active 101. Yap, T. A. et al. Targeting the PI3K–AKT–mTOR p110α and p110β catalytic subunits of class‑Ia principle. J. Antibiot. (Tokyo) 28, 721–726 (1975). pathway: progress, pitfalls, and promises. Curr. Opin. phosphoinositide 3‑kinases. Biochem. J. 350, 79. Yatscoff, R. W., LeGatt, D. F. & Kneteman, N. M. Pharmacol. 8, 393–412 (2008). 353–359 (2000). Therapeutic monitoring of rapamycin: a new 102. Torbett, N. E. et al. A chemical screen in diverse breast 56. Benistant, C., Chapuis, H. & Roche, S. A specific immunosuppressive drug. Ther. Drug Monit. 15, cancer cell lines reveals genetic enhancers and function for phosphatidylinositol 3‑kinase α 478–482 (1993). suppressors of sensitivity to PI3K isoform‑selective (p85α‑p110α) in cell survival and for 80. Faivre, S., Kroemer, G. & Raymond, E. Current inhibition. Biochem. J. 415, 97–110 (2008). phosphatidylinositol 3‑kinase β (p85α‑p110b) in development of mTOR inhibitors as anticancer agents. 103. Wee, S. et al. PTEN‑deficient cancers depend on PIK3CB. de novo DNA synthesis of human colon carcinoma Nature Rev. Drug Discov. 5, 671–688 (2006). Proc. Natl Acad. Sci. USA 105, 13057–13062 (2008). cells. Oncogene 19, 5083–5090 (2000). 81. Guertin, D. A. & Sabatini, D. M. Defining the role 104. Oda, K. et al. PIK3CA cooperates with other 57. Brugge, J., Hung., M. C. & Mills, G. B. A new of mTOR in cancer. Cancer Cell 12, 9–22 (2007). phosphatidylinositol 3’‑kinase pathway mutations to mutational AKTivation in the PI3K pathway. Cancer 82. Hay, N. The Akt‑mTOR tango and its relevance to effect oncogenic transformation. Cancer Res. 68, Cell 12, 104–107 (2007). cancer. Cancer Cell 8, 179–183 (2005). 8127–8136 (2008). 58. Carpten, J. D. et al. A transforming mutation in 83. Atkins, M. B. et al. Randomized phase II study of 105. Aguirre, V., Uchida, T., Yenush, L., Davis, R. &

the pleckstrin homology domain of AKT1 in cancer. multiple dose levels of CCI‑779, a novel mammalian White, M. F. The c‑Jun NH2‑terminal kinase promotes Nature 448, 439–444 (2007). target of rapamycin kinase inhibitor, in patients with insulin resistance during association with insulin This paper describes the identification of an advanced refractory renal cell carcinoma. J. Clin. receptor substrate‑1 and phosphorylation of Ser307. activating AKT1 mutation in human cancer. Oncol. 22, 909–918 (2004). J. Biol. Chem. 275, 9047–9054 (2000).

106. Harrington, L. S., Findlay, G. M. & Lamb, R. F. 130. Guba, M. et al. Rapamycin inhibits primary and 156. Kim, M. S., Jeong, E. G., Yoo, N. J. & Lee, S. H. Restraining PI3K: mTOR signalling goes back to the metastatic tumor growth by antiangiogenesis: Mutational analysis of oncogenic AKT E17K mutation membrane. Trends Biochem. Sci. 30, 35–42 (2005). involvement of vascular endothelial growth factor. in common solid cancers and acute leukaemias. 107. Lee, Y. H., Giraud, J., Davis, R. J. & White, M. F. c‑Jun Nature Med. 8, 128–135 (2002). Br. J. Cancer 98, 1533–1535 (2008). N‑terminal kinase (JNK) mediates feedback inhibition 131. Stallone, G. et al. Sirolimus for Kaposi’s sarcoma in 157. Malanga, D. et al. Activating E17K mutation in the of the insulin signaling cascade. J. Biol. Chem. 278, renal‑transplant recipients. N. Engl. J. Med. 352, gene encoding the protein kinase AKT1 in a subset 2896–2902 (2003). 1317–1323 (2005). of squamous cell carcinoma of the lung. Cell Cycle 7, 108. Hirosumi, J. et al. A central role for JNK in obesity and 132. Thomas, G. V. et al. Hypoxia‑inducible factor 665–669 (2008). insulin resistance. Nature 420, 333–336 (2002). determines sensitivity to inhibitors of mTOR in kidney 158. Staal, S. P. Molecular cloning of the akt oncogene and 109. Um, S. H. et al. Absence of S6K1 protects against cancer. Nature Med. 12, 122–127 (2006). its human homologues AKT1 and AKT2: amplification age‑ and diet‑induced obesity while enhancing insulin 133. Hudson, C. C. et al. Regulation of hypoxia‑inducible of AKT1 in a primary human gastric adenocarcinoma. sensitivity. Nature 431, 200–205 (2004). factor 1α expression and function by the mammalian Proc. Natl Acad. Sci. USA 84, 5034–5037 (1987). 110. O’Reilly, K. E. et al. mTOR inhibition induces upstream target of rapamycin. Mol. Cell Biol. 22, 7004–7014 159. Bellacosa, A. et al. Molecular alterations of the AKT2 receptor tyrosine kinase signaling and activates Akt. (2002). oncogene in ovarian and breast carcinomas. Cancer Res. 66, 1500–1508 (2006). 134. Bernardi, R. et al. PML inhibits HIF‑1α translation and Int. J. Cancer 64, 280–285 (1995). 111. Jun, T., Gjoerup, O. & Roberts, T. M. Tangled webs: neoangiogenesis through repression of mTOR. Nature 160. Cheng, J. Q. et al. AKT2, a putative oncogene evidence of cross‑talk between c‑Raf‑1 and Akt. 442, 779–785 (2006). encoding a member of a subfamily of protein‑serine/ Sci. STKE 1999, PE1 (1999). 135. Phung, T. L. et al. Pathological angiogenesis is induced threonine kinases, is amplified in human ovarian 112. Carracedo, A. et al. Inhibition of mTORC1 leads to by sustained Akt signaling and inhibited by rapamycin. carcinomas. Proc. Natl Acad. Sci. USA 89, 9267–9271 MAPK pathway activation through a PI3K‑dependent Cancer Cell 10, 159–170 (2006). (1992). feedback loop in human cancer. J. Clin. Invest. 118, 136. Hamada, K. et al. The PTEN/PI3K pathway governs 161. Ruggeri, B. A., Huang, L., Wood, M., Cheng, J. Q. & 3065–3074 (2008). normal vascular development and tumor angiogenesis. Testa, J. R. Amplification and overexpression of the 113. Williams, R. et al. The skin and hair as surrogate Genes Dev. 19, 2054–2065 (2005). AKT2 oncogene in a subset of human pancreatic tissues for measuring the target effect of inhibitors 137. Bayascas, J. R., Leslie, N. R., Parsons, R., Fleming, S. ductal adenocarcinomas. Mol. Carcinog. 21, 81–86 of phosphoinositide‑3‑kinase signaling. Cancer & Alessi, D. R. Hypomorphic mutation of PDK1 (1998). Chemother. Pharmacol. 58, 444–450 (2006). suppresses tumorigenesis in PTEN+/– mice. Curr. Biol. 162. Oki, E. et al. Impact of loss of heterozygosity of 114. Cho, H. et al. Insulin resistance and a diabetes 15, 1839–1846 (2005). encoding phosphate and tensin homolog on the mellitus‑like syndrome in mice lacking the protein 138. Chen, M. L. et al. The deficiency of Akt1 is sufficient to prognosis of gastric cancer. J. Gastroenterol. Hepatol. kinase Akt2 (PKBβ). Science 292, 1728–1731 (2001). suppress tumor development in Pten+/– mice. Genes 21, 814–818 (2006). 115. Brachmann, S. M., Ueki, K., Engelman, J. A., Kahn, Dev. 20, 1569–1574 (2006). 163. Li, Y. L., Tian, Z., Wu, D. Y., Fu, B. Y. & Xin, Y. Loss of R. C. & Cantley, L. C. Phosphoinositide 3‑kinase 139. Massion, P. P. et al. Genomic copy number analysis of heterozygosity on 10q23.3 and mutation of tumor catalytic subunit deletion and regulatory subunit non‑small cell lung cancer using array comparative suppressor gene PTEN in gastric cancer and deletion have opposite effects on insulin sensitivity genomic hybridization: implications of the precancerous lesions. World J. Gastroenterol. 11, in mice. Mol. Cell Biol. 25, 1596–1607 (2005). phosphatidylinositol 3‑kinase pathway. Cancer Res. 285–288 (2005). 116. Schnell, C. R. et al. Effects of the dual 62, 3636–3640 (2002). 164. Feilotter, H. E. et al. Analysis of the 10q23 phosphatidylinositol 3‑kinase/mammalian target 140. Massion, P. P. et al. Early involvement of the chromosomal region and the PTEN gene in human of rapamycin inhibitor NVP‑BEZ235 on the tumor phosphatidylinositol 3‑kinase/Akt pathway in lung sporadic breast carcinoma. Br. J. Cancer 79, 718–723 vasculature: implications for clinical imaging. Cancer cancer progression. Am. J. Respir. Crit. Care Med. (1999). Res. 68, 6598–6607 (2008). 170, 1088–1094 (2004). 165. Freihoff, D. et al. Exclusion of a major role for the 117. Bomanji, J. B., Costa, D. C. & Ell, P. J. Clinical role of 141. Okudela, K. et al. PIK3CA mutation and amplification PTEN tumour‑suppressor gene in breast carcinomas. positron emission tomography in oncology. Lancet in human lung cancer. Pathol. Int. 57, 664–671 Br. J. Cancer 79, 754–758 (1999). Oncol. 2, 157–164 (2001). (2007). 166. Garcia, J. M. et al. Allelic loss of the PTEN region 118. Eichhorn, P. J. et al. Phosphatidylinositol 3‑kinase 142. Kawano, O. et al. PIK3CA gene amplification in (10q23) in breast carcinomas of poor pathophenotype. hyperactivation results in lapatinib resistance that is Japanese non‑small cell lung cancer. Lung Cancer Breast Cancer Res. Treat. 57, 237–243 (1999). reversed by the mTOR/phosphatidylinositol 3‑kinase 58, 159–160 (2007). 167. Tokunaga, E. et al. Coexistence of the loss of inhibitor NVP‑BEZ235. Cancer Res. 68, 9221–9230 143. Ma, Y. Y. et al. PIK3CA as an oncogene in cervical heterozygosity at the PTEN locus and HER2 (2008). cancer. Oncogene 19, 2739–2744 (2000). overexpression enhances the Akt activity thus leading 119. Berns, K. et al. A functional genetic approach 144. Wu, G. et al. Somatic mutation and gain of copy to a negative progesterone receptor expression in identifies the PI3K pathway as a major determinant of number of PIK3CA in human breast cancer. breast carcinoma. Breast Cancer Res. Treat. 101, trastuzumab resistance in breast cancer. Cancer Cell Breast Cancer Res. 7, R609–R616 (2005). 249–257 (2007). 12, 395–402 (2007). 145. Woenckhaus, J. et al. Genomic gain of PIK3CA and 168. Pollock, P. M. et al. PTEN inactivation is rare in 120. Zhang, J., Yang, P. L. & Gray, N. S. Targeting cancer increased expression of p110α are associated with melanoma tumours but occurs frequently in with small molecule kinase inhibitors. Nature Rev. progression of dysplasia into invasive squamous melanoma cell lines. Melanoma Res. 12, 565–575 Cancer 9, 28–39 (2009). cell carcinoma. J. Pathol. 198, 335–342 (2002). (2002). 121. Zunder, E. R., Knight, Z. A., Houseman, B. T., Apsel, B. 146. Pedrero, J. M. et al. Frequent genetic and biochemical 169. Celebi, J. T., Shendrik, I., Silvers, D. N. & Peacocke, M. & Shokat, K. M. Discovery of drug‑resistant and drug‑ alterations of the PI 3‑K/AKT/PTEN pathway in head Identification of PTEN mutations in metastatic sensitizing mutations in the oncogenic PI3K isoform and neck squamous cell carcinoma. Int. J. Cancer 114, melanoma specimens. J. Med. Genet. 37, 653–657 p110α. Cancer Cell 14, 180–192 (2008). 242–248 (2005). (2000). 122. Engelman, J. A. et al. ErbB‑3 mediates 147. Fenic, I., Steger, K., Gruber, C., Arens, C. & 170. Birck, A., Ahrenkiel, V., Zeuthen, J., Hou‑Jensen, K. & phosphoinositide 3‑kinase activity in gefitinib‑sensitive Woenckhaus, J. Analysis of PIK3CA and Akt/protein Guldberg, P. Mutation and allelic loss of the PTEN/ non‑small cell lung cancer cell lines. Proc. Natl Acad. kinase B in head and neck squamous cell carcinoma. MMAC1 gene in primary and metastatic melanoma Sci. USA 102, 3788–3793 (2005). Oncol. Rep. 18, 253–259 (2007). biopsies. J. Invest. Dermatol. 114, 277–280 (2000). 123. Engelman, J. A. et al. MET amplification leads to 148. Byun, D. S. et al. Frequent monoallelic deletion of 171. Reifenberger, J. et al. Allelic losses on chromosome gefitinib resistance in lung cancer by activating PTEN and its reciprocal associatioin with PIK3CA arm 10q and mutation of the PTEN (MMAC1) ERBB3 signaling. Science 316, 1039–1043 (2007). amplification in gastric carcinoma. Int. J. Cancer tumour suppressor gene in primary and metastatic 124. Sergina, N. V. et al. Escape from HER‑family tyrosine 104, 318–327 (2003). malignant melanomas. Virchows Arch. 436, kinase inhibitor therapy by the kinase‑inactive HER3. 149. Wu, G. et al. Uncommon mutation, but common 487–493 (2000). Nature 445, 437–441 (2007). amplifications, of the PIK3CA gene in thyroid 172. Cairns, P. et al. Frequent inactivation of PTEN/ 125. Apsel, B. et al. Targeted polypharmacology: tumors. J. Clin. Endocrinol. Metab. 90, 4688–4693 MMAC1 in primary prostate cancer. Cancer Res. 57, discovery of dual inhibitors of tyrosine and (2005). 4997–5000 (1997). phosphoinositide kinases. Nature Chem. Biol. 4, 150. Miller, C. T. et al. Gene amplification in esophageal 173. Feilotter, H. E., Nagai, M. A., Boag, A. H., Eng, C. & 691–699 (2008). adenocarcinomas and Barrett’s with high‑grade Mulligan, L. M. Analysis of PTEN and the 10q23 126. Gupta, S. et al. Binding of ras to phosphoinositide dysplasia. Clin. Cancer Res. 9, 4819–4825 (2003). region in primary prostate carcinomas. Oncogene 16, 3‑kinase p110α is required for ras‑driven 151. Miyake, T. et al. PIK3CA gene mutations and 1743–1748 (1998). tumorigenesis in mice. Cell 129, 957–968 (2007). amplifications in uterine cancers, identified by 174. Pesche, S. et al. PTEN/MMAC1/TEP1 involvement in 127. Engelman, J. A. et al. Effective use of PI3K and MEK methods that avoid confounding by PIK3CA primary prostate cancers. Oncogene 16, 2879–2883 inhibitors to treat mutant Kras G12D and PIK3CA pseudogene sequences. Cancer Lett. 261, 120–126 (1998). H1047R murine lung cancers. Nature Med. 14, (2008). 175. Gray, I. C. et al. Mutation and expression analysis of 1351–1356 (2008). 152. Nakayama, K. et al. Amplicon profiles in ovarian the putative prostate tumour‑suppressor gene PTEN. This study showned that a combined treatment serous carcinomas. Int. J. Cancer 120, 2613–2617 Br. J. Cancer 78, 1296–1300 (1998). with PI3K and MEK inhibitors is necessary to block (2007). 176. Wang, S. I., Parsons, R. & Ittmann, M. Homozygous oncogenic Kras-induced lung cancer in a murine 153. Nakayama, K. et al. Sequence mutations and deletion of the PTEN tumor suppressor gene in a model. amplification of PIK3CA and AKT2 genes in purified subset of prostate adenocarcinomas. Clin. Cancer Res. 128. Yu, K., Toral‑Barza, L., Shi, C., Zhang, W. G. & Zask, A. ovarian serous neoplasms. Cancer Biol. Ther. 5, 4, 811–815 (1998). Response and determinants of cancer cell 779–785 (2006). 177. Bostrom, J. et al. Mutation of the PTEN (MMAC1) susceptibility to PI3K inhibitors: combined targeting 154. Kita, D., Yonekawa, Y., Weller, M. & Ohgaki, H. tumor suppressor gene in a subset of glioblastomas of PI3K and Mek1 as an effective anticancer strategy. PIK3CA alterations in primary (de novo) and but not in meningiomas with loss of chromosome arm Cancer Biol. Ther. 7, 307–315 (2008). secondary glioblastomas. Acta Neuropathol. 113, 10q. Cancer Res. 58, 29–33 (1998). 129. Yuan, T. L. et al. Class 1A PI3K regulates vessel 295–302 (2007). 178. Wang, S. I. et al. Somatic mutations of PTEN in integrity during development and tumorigenesis. 155. Bleeker, F. E. et al. AKT1E17K in human solid tumours. glioblastoma multiforme. Cancer Res. 57, 4183–4186 Proc. Natl Acad. Sci. USA 105, 9739–9744 (2008). Oncogene 27, 5648–5650 (2008). (1997).

179. Smith, J. S. et al. PTEN mutation, EGFR amplification, 190. Li, Z. et al. Roles of PLC‑β2 and ‑β3 and PI3Kγ in 201. Ueki, K. et al. Increased insulin sensitivity in mice and outcome in patients with anaplastic astrocytoma chemoattractant‑mediated signal transduction. lacking p85β subunit of phosphoinositide 3‑kinase. and glioblastoma multiforme. J. Natl Cancer Inst. 93, Science 287, 1046–1049 (2000). Proc. Natl Acad. Sci. USA 99, 419–424 (2002). 1246–1256 (2001). 191. Laffargue, M. et al. Phosphoinositide 3‑kinase γ is an 202. Taniguchi, C. M. et al. Divergent regulation of hepatic 180. Kim, N. et al. The p110δ catalytic isoform of PI3K essential amplifier of mast cell function. Immunity 16, glucose and lipid metabolism by phosphoinositide is a key player in NK‑cell development and cytokine 441–451 (2002). 3‑kinase via Akt and PKCλ/ζ. Cell. Metab. 3, 343–353 secretion. Blood 110, 3202–3208 (2007). 192. Rodriguez‑Borlado, L. et al. Phosphatidylinositol (2006). 181. Clayton, E. et al. A crucial role for the p110δ subunit 3‑kinase regulates the CD4/CD8 T cell 203. Luo, J. et al. Loss of class IA PI3K signaling in of phosphatidylinositol 3‑kinase in B cell development differentiation ratio. J. Immunol. 170, 4475–4482 muscle leads to impaired muscle growth, insulin and activation. J. Exp. Med. 196, 753–763 (2002). (2003). response, and hyperlipidemia. Cell Metab. 3, 182. Jou, S. T. et al. Essential, nonredundant role for the 193. Webb, L. M., Vigorito, E., Wymann, M. P., Hirsch, E. & 355–366 (2006). phosphoinositide 3‑kinase p110 in signaling by Turner, M. Cutting edge: T cell development requires δ Acknowledgements the B‑cell receptor complex. Mol. Cell Biol. 22, the combined activities of the p110γ and p110δ We thank N. Gray and Q. Liu for providing compound structures 8580–8591 (2002). catalytic isoforms of phosphatidylinositol 3‑kinase. and helpful discussions. We thank the reviewers for their help‑ 183. Puri, K. D. et al. Mechanisms and implications of J. Immunol. 175, 2783–2787 (2005). ful suggestions. We apologize to colleagues whose primary phosphoinositide 3‑kinase delta in promoting 194. Swat, W. et al. Essential role of PI3Kδ and PI3Kγ papers were not cited owing to space constraints. This work neutrophil trafficking into inflamed tissue. Blood in thymocyte survival. Blood 107, 2415–2422 was supported in part by the National Institutes of Health 103, 3448–3456 (2004). (2006). (CA030002, CA089021 and CA050661 to T.M.R. and 184. Okkenhaug, K. et al. Impaired B and T cell antigen 195. Tassi, I. et al. p110γ and p110δ phosphoinositide CA134502‑01 to J.J.Z.), the Department of Defense for Cancer receptor signaling in p110δ PI 3‑kinase mutant mice. 3‑kinase signaling pathways synergize to control Research (BC051565 to J.J.Z.), the V Foundation (J.J.Z.) and Science 297, 1031–1034 (2002). development and functions of murine NK cells. the Claudia Barr Program (J.J.Z.). 185. Ali, K. et al. Essential role for the p110δ Immunity 27, 214–227 (2007). phosphoinositide 3‑kinase in the allergic response. 196. Terauchi, Y. et al. Increased insulin sensitivity and Competing interests statement Nature 431, 1007–1011 (2004). hypoglycaemia in mice lacking the p85α subunit The authors declare competing financial interests: see web 186. Guo, H., Samarakoon, A., Vanhaesebroeck, B. & of phosphoinositide 3‑kinase. Nature Genet. 21, version for details. Malarkannan, S. The p110δ of PI3K plays a critical 230–235 (1999).

role in NK cell terminal maturation and cytokine/ 197. Suzuki, H. et al. Xid‑like immunodeficiency in chemokine generation. J. Exp. Med. 205, 2419–2435 mice with disruption of the p85α subunit of DATABASES (2008). phosphoinositide 3‑kinase. Science 283, 390–392 Entrez Gene: 187. Hirsch, E. et al. Central role for G protein‑coupled (1999). http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene phosphoinositide 3‑kinase γ in inflammation. Science 198. Chen, D. et al. p50α/p55α phosphoinositide 3‑kinase PIK3CA | PIK3R1 | PIK3R2 | PIK3R3 287, 1049–1053 (2000). knockout mice exhibit enhanced insulin sensitivity. OMIM: 188. MacDonald, P. E. et al. Impaired glucose‑stimulated Mol. Cell Biol. 24, 320–329 (2004). http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM insulin secretion, enhanced intraperitoneal insulin 199. Fruman, D. A. et al. Impaired B cell development and Bannayan–Riley–Ruvalcaba’s syndrome | Cowden’s syndrome tolerance, and increased β‑cell mass in mice lacking proliferation in absence of phosphoinositide 3‑kinase UniProtKB: http://www.uniprot.org the p110γ isoform of phosphoinositide 3‑kinase. p85α. Science 283, 393–397 (1999). AKT | mTOR | PTEN Endocrinology 145, 4078–4083 (2004). 200. Fruman, D. A. et al. Hypoglycaemia, liver necrosis Jean J. Zhao’s homepage: 189. Crackower, M. A. et al. Regulation of myocardial and perinatal death in mice lacking all isoforms of http://www.jeanzhao.dfci.harvard.edu contractility and cell size by distinct PI3K‑PTEN phosphoinositide 3‑kinase p85α. Nature Genet. 26, All linKs Are AcTive in The online PDF signaling pathways. Cell 110, 737–749 (2002). 379–382 (2000).