Drug Discovery Approaches Targeting the PI3K/Akt Pathway in Cancer

Oncogene (2008) 27, 5511–5526 & 2008 Macmillan Publishers Limited All rights reserved 0950-9232/08 $32.00 www.nature.com/onc REVIEW Drug discovery approaches targeting the PI3K/Akt pathway in cancer

C Garcia-Echeverria1 and WR Sellers2

1Oncology Drug Discovery, Novartis Institutes for Biomedical Research, Basel, Switzerland and 2Novartis Institutes for Biomedical Research, Cambridge, MA, USA

The abnormal activation of the phosphatidylinositol 3- genous negative regulator of the PI3K pathway or (vi) kinase (PI3K)/Akt pathway has been validated by constitutive recruitment and activation by mutant forms epidemiological and experimental studies as an essential of the Ras oncogene (Carnero et al., 2008; Tokunaga step toward the initiation and maintenance of human et al., 2008; Maira et al., 2008b). The preceding tumors. Notable in this regard are the prevalent somatic alterations trigger a cascade of biological events, from genetic alterations leading to the inactivation of the tumor cell growth and proliferation to survival and migration, suppressor gene PTEN and gain-of-function mutations which drive tumor progression. In addition, for these targeting PIK3CA—the gene encoding the catalytic cell autonomous survival and proliferation functions, phosphosinositide-3 kinase subunit p110a. A number of PI3K is a key regulator of angiogenic pathways and the intracellular components of this pathway have been upregulated metabolic activity in tumors. Finally, this targeted as anticancer drug discovery activities leading to pathway also plays a major part in the resistance of the current panoply of clinical trials of inhibitors of PI3K, tumor cells to conventional cytotoxic and targeted Akt and HSP90 in man. This review summarizes current anticancer therapies. Blocking the PI3K/Akt pathway preclinical knowledge of modulators of the PI3K/Akt could therefore simultaneously inhibit the proliferation pathway in which drug discovery and development and growth of tumor cells and sensitize them toward activities have been advanced focusing on both the relevant programmed cell death. To this end, drug discovery clinical stage inhibitors and other disclosed tool com- activities have been directed over the past few years to pounds targeting PI3K, PDK1, Akt and HSP90. identify compounds that effectively and specifically Oncogene (2008) 27, 5511–5526; doi:10.1038/onc.2008.246 disrupt this pathway in tumor cells. Here, the progress in the identification and development of compounds Keywords: phosphoinositide-3kinase; Akt; PDK1; designed to target some of the intracellular components HSP90; drug design; cancer therapy of the PI3K/Akt is reviewed.

Inhibitors of the lipid kinase activity of PI3K Introduction The PI3Ks are widely expressed lipid kinases that The abnormal activation of the phosphatidylinositol 3- phosphorylate phosphoinositides at the D-3position kinase (PI3K)/Akt (Akt also known as protein kinase of the inositol ring. These enzymes function as signal (PKB)) pathway (Figure 1) has been validated by transducers downstream of cell-surface receptors. The epidemiological and experimental studies as an essential products of PI3K-catalyzed reactions, phosphatidylino- step toward the initiation and maintenance of human sitol 3,4,5-trisphosphate (PtdIns(3,4,5)P3), phosphatidy- tumors. Constitutive pathway activation can result from linositol 3,4-bisphosphate (PtdIns(3,4)P2)and the distinct and/or complementary biological events phosphatidylinositol 3-phosphate (PtdIns(3)P), are sec- including (i) constitutively active mutants or amplifica- ond messengers and have central roles in a number of tion of receptor tyrosine kinases (for example, epidermal cellular processes, including cell growth, differentiation, growth factor receptor or ErbB2) leading to constitutive mobility, proliferation and survival. recruitment and activation of PI3K and downstream The eight members of the PI3K family have been effectors; (ii) amplification of PI3K; (iii) presence of classified into three groups based on their primary activating mutations in the PIK3CA gene encoding the sequence, in vitro substrate preference, domain structure p110a catalytic subunit; (iv) overexpression of the and mode of regulation. The subject of this section is downstream kinase Akt; (v) loss or inactivating muta- class I PI3Ks, which is composed of two subgroups, tions of the tumor suppressor gene PTEN, an endo- IA and IB (Maira et al., 2008b). The class IA PI3K subgroup consists of three catalytic subunits, p110a, p110b and p110d, that form heterodimers with one of Correspondence: Dr C Garcia-Echeverria, Oncology Drug Discovery, Novartis Institutes for BioMedical Research, WKL-125.13.16, Basel five regulatory domains: p85a, p85b, p85g,p50a and CH-4002, Switzerland. p55a. As shown in Figure 1, these PI3Ks are activated E-mail: [email protected] by cell-surface receptor tyrosine kinases. The class IB Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5512

Figure 1 Schematic representation of the PI3K/Akt pathway and its main components. Upon activation by their cognate growth factors, PI3K recruitment to receptor tyrosine kinases leads to a transient increase in PIP3 levels. This phospholipid serves as an anchor point to recruit PH-domain containing proteins such as Akt. After translocation to the plasma membrane, Akt is phosphorylated on two distinct residues, Thr-308 and Ser-473, by PDK1 and PDK2 (mTORC2), respectively. Once fully activated, Akt phosphorylates and activates down stream cytosolic and nuclear effectors. In tumor cells, Akt is constitutively activated leading to dysregulated proliferation, cell growth and survival. Activation of this pathway is also involved in tumor angiogenesis, which results in the formation of a network of blood vessels that penetrates into the tumor tissue, supplying nutrients and oxygen and removing waste products. PDK1, phosphoinositide-dependent protein kinase-1; PH, plecstrin homology domain; PI3K, phosphatidylinositol 3-kinase.

PI3K consists of one member, a heterodimer of p110g improved pharmaceutical properties and therapeutic and a regulatory p101 domain, which is activated by indexes such as PWT-458 (compound 3, Figure 2; G-proteins bg subunits following the stimulation of Yu et al., 2005) and PX-866 (compound 4, Figure 2; Ihle G-protein-coupled receptors. PI3Ks IA and IB catalyze et al., 2005; Howes et al., 2007) have been reported. the formation of PtdIns(3,4,5)P3, a process that is PWT-458 is a PEGylated derivative of wortmannin that reversed by the action of the lipid phosphatase, PTEN. has a higher therapeutic index in preclinical animal Two well-known and isoform-unselective PI3K in- models compared to the parent compound. Upon hibitors are the fungal metabolite wortmannin (com- intravenous (i.v.) administration, the polyethylene-gly- pound 1, Figure 2; Brian et al., 1957) and LY294002 col (PEG) moiety is cleaved and 17-hydroxywortmannin (compound 2, Figure 2; Vlahos et al., 1994). These is released. In vivo antitumor activity has been reported compounds block the enzymatic activity of PI3Ks by in several human xenograft models (for example, different mechanisms. Thus, wortmannin is an irrever- U87MG, A549 and A498). PX-866 is a semisynthetic sible inhibitor (IC50E2nM) that forms a covalent viridin derivative that potently and irreversibly inhibits bond with a conserved lysine residue involved in the PI3Ka, g and d (IC50 ¼ 5.5, 9.0 and 2.7 nM, respectively). phosphate-binding reaction (Wymann et al., 1996), Unlike wortmannin, it shows a certain level of selectivity whereas LY294002 is a classical reversible, ATP- over PI3Kb (IC50>300 nM) and higher stability. PX-866 competitive PI3K modulator (IC50 ¼ 1.40 mM) (Vlahos suppresses the growth of several human tumor multi- et al., 1994; Davies et al., 2000). In spite of the crossover cellular spheroids (for example, U87, PC3, T47D and inhibition of other lipid and PKs and their unfavorable HCT116) and provokes a more prolonged inhibition of pharmaceutical properties, wortmannin and LY294002 PI3K than does wortmannin, The compound has have served as important research tools for more than a antitumor activity against several human solid tumor decade in elucidating the role of PI3Ks in the biology of models (for example, OvCar-3, HT-29 or A-549) when human cancers. In parallel with their preclinical use, the used alone (2 mg/kg, orally) or in combination with scaffolds of wortmannin and LY294002 have been chemotherapeutic or targeted anticancer agents the subjects of intense lead optimization efforts directed (for example, paclitaxel or gefitinib). In addition to at improving upon the pharmaceutical limitations of a decrease in body weight and leukocytosis, a the parent compounds. Thus, using wortmannin as a major toxicity of PX-866 administration to severe starting point, broad-spectrum PI3K inhibitors with combined immunodeficient mice was hyperglycemia

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5513 O O O O O O PEG O O O O S

O N H O H O O O O O O 123 Wortmannin LY294002 PWT-458 O O O O H2N NH

O NH O H O OH O O O O O H H + N OH HN O N OH H N N N O H H O O O OH 4 5 PX-866 SF-1126 O O N N N O S N N N N O N NH N N N N OH N N O N S O 678 NVP-BEZ235 PI103 GDC-0941

O NO2 O O N O S N N N N S NH N O N NH O N Br N

9 10 11 PIK-75 TGX-221 AS-605240

O

N

N N N N N

H2N 12 IC87114 Figure 2 Representative examples of inhibitors of PI3K.

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5514 with decreased glucose tolerance. These changes were pound-mediated alterations in microvessel permeability dose-dependent and reversible upon discontinuation of and significantly reduced tumor interstitial fluid pressure the drug. These wortmannin derivatives have not reflecting vascular permeability changes (Schnell et al., entered clinical trials at this time. 2008; Yuan et al., 2008). Moreover, early reduction of Similar efforts have been used to improve the permeability, as determined by Ktrans quantification, was pharmacological properties of LY29002 and derivatives found to be a predictive marker for late-stage antitumor thereof. SF-1126 (compound 5, Figure 2) (Garlich et al., activity. This later results suggest that vascular perme- 2008) is a water soluble prodrug of LY294002. The ability reduction, which can be monitored noninvasively compound, which is designed to target specific integrins by dynamic contrast-enhanced magnetic resonance within the tumor compartment by means of an imaging might be a useful pharmacodynamic measure appended peptide moiety, inhibits all paralogs of PI3K in clinical trials and might serve as an early response (IC50 ¼ 0.5–5.7 mM) and other members of the PI3K biomarker predictive of longer-term efficacy. In pre- family, including DNA-PK and mTOR. It blocks the clinical in vivo models (Engelman et al., 2008; Serra phosphorylation and activation of Akt in cellular et al., 2008; Maira et al., 2008a), NVP-BEZ235 was well settings with IC50 values in the low micromolar range tolerated at the efficacious doses when compared with (for example, IC50 ¼ 2.4 mM, MM.1S). This prodrug vehicle-treated animals, with no significant difference approach appears to enhance the delivery of the active seen in the body weight. Unlike other modulators of the PI3K inhibitor to tumors resulting in significant PI3K/Akt pathway, no statistically significant changes antitumor activity in xenograft models of brain, in blood glucose levels were observed in the animals neuroblastoma, non-small cell lung, prostate, myeloma, treated with NVP-BEZ235 after in vivo efficacy experi- renal and colon carcinoma. In addition to its direct ments in mice or rats. Moreover, no effect on insulin or activity on cancer cells, microvessel density analysis of glucose levels was reported for NVP-BEZ235 at tumor tissue demonstrated that the compound also had efficacious doses at the end of a 13-week rat tolerability significant antiangiogenic activity in vivo. The safety of study (Marrer et al., 2008). These preclinical data SF-1126 in humans is currently being tested in phase I demonstrate the feasibility of effectively blocking the trials in which patients are being treated twice per week PI3K/Akt pathway in vivo with manageable effects on by i.v. infusion over 1.5 h in treatment cycles of 4 weeks. glucose regulation. Phase I trials with NVP-BEZ235 In parallel with the synthetic efforts made to over- inhibitor were initiated in patients with solid tumors at come the pharmaceutical hurdles encountered with the end of 2006. Patients are enrolled in successive wortmannin and LY294002, high throughput screening cohorts receiving the compound orally once per day for and medicinal chemistry activities have been extensively cycles of 28 days. utilized to discover and develop new pan- or paralog XL147 and XL765 (structures not disclosed) are also selective PI3K modulators. Overall, the new generation low-molecular mass PI3K inhibitors in phase I clinical of PI3K inhibitors has improved selectivity and drug- trials. XL-147 is active against class I PI3Ks (IC50 ¼ 39, like properties when compared with the earlier com- 383, 23, 36 nM for p110a, b, g and d, respectively) pounds. A number of these compounds are in clinical without inhibiting the kinase activity of Vps34 (IC50 ¼ development and are reviewed below. 6975 nM), DNA-PK or mTOR (IC50>15 000 nM; Foster, NVP-BEZ235 (compound 6, Figure 2) is an imida- 2007; Shapiro et al., 2007). The compound inhibits PI3K zo[4,5-c]quinoline derivative that inhibits PI3K signaling in cultured tumor cells (IC50 ¼ 0.40–1 mM,p- (IC50 ¼ 4, 76, 7, 5 and 21 nM against p110a, b, g and Thr308Akt) and blocks vascular endothelial growth d) and mTOR (IC50 ¼ 21 nM) kinase activity by binding factor-induced tubule formation in cellular setting to the ATP-binding cleft of these enzymes (Maira et al., (IC50 ¼ 529 nM, HMVEC-L). Oral dosing slows tumor 2008a). The compound potently inhibits proliferation in growth or induces shrinkage in preclinical models of a broad panel of tumor cell lines (for example, breast, lung, ovarian, prostate and glioma tumors. IC50 ¼ 10 nM, U87MG cells) by specifically blocking Increased tumor growth inhibition over the correspond- the biological function of PI3K signaling components ing single agents was obtained by combining this PI3K (for example, IC50 ¼ 10±1nM, for inhibition of p- inhibitor with cytotoxic or targeted anticancer agents Ser473-Akt in U87MG cells) and inducing a G1 arrest. (for example, carboplatin, paclitaxel or erlotinib). XL- The antiproliferative activity of this imidazoquinoline 765 has a similar activity toward class I PI3K (IC50 ¼ 39, derivative in cellular settings translates well in in vivo 113, 9 and 43 nM for p110a, b, g and d, respectively) models of human cancer (for example, U87MG, PC3M as XL-147, but, in addition, also inhibits DNA-PK and A549). Thus, compound treatment results in disease (IC50 ¼ 150 nM) and mTOR (IC50 ¼ 157 nM; Laird, 2007; stasis or tumor regression when administered orally—25 Patnaik et al., 2007). As in the case of XL-147, tumor to 50 mg/kg/day—and enhanced the efficacy when used growth is delayed or arrested in in vivo efficacy in in vivo combination studies (for example, docetaxel or experiments using suitable xenograft tumor models. temozolomide). Ex vivo pharmacokinetic/pharmacody- Interim analyses of ongoing phase I studies with these namic analyses of tumor tissues upon acute dose or after compounds have shown interpatient variability with termination of in vivo efficacy studies showed a time- terminal half-life values that ranged from 2.5 and 8 days dependent correlation between compound concentration for XL147 and 2–15 h for XL765. Preliminary signs and inhibition of Akt phosphorylation. Additional of exposure-dependent pharmacodynamic modulation in vivo experiments have demonstrated marked, com- (changes in plasma insulin levels) were observed at

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5515 30/60 mg for XL147 and 15/30 mg b.i.d. for XL765. bLys782) may provide the basis for the identification Analyses of other surrogate tissues, including peripheral of inhibitors with some level of selectivity for the distinct blood mononuclear cell lysates and hair follicles, for p110 enzymes. For example, it has been recently shown target modulation are also under investigation. that modulating and fine-tuning the interactions of the Imidazopyrimidines and pyridofuropyrimidines with modified molecular template with key amino acids in the class I PI3K activity have been disclosed in several ATP-binding pocket can provide paralog-selective PI3K publications. For example, PI103(compound 7; inhibitors for cancer treatment (Redaelli et al., 2003; Figure 2) is a pan-PI3K inhibitor with enzymatic IC50 Knight et al., 2004). To this end, selectivity trends values in the 2 nM (PI3Ka)to15nM range (PI3Kg) among compound classes that target PI3K members (Fan et al., 2007; Raynaud et al., 2007). Despite its rapid have been identified by extensive biochemical profiling in vivo metabolism, the compound, which readily against all the members of this lipid kinase family distributes to tumor tissue, delays the growth of (Knight et al., 2006). According to these results, different subcutaneous human cancer xenografts. Re- inhibitors of p110b often tend to inhibit p110d, whereas cently, it was shown that PI103as with other pan-PI3K inhibitors of p110a seem to preferentially inhibit p110g. modulators also inhibits both complexes of mTOR: Interestingly, and in spite of a clear sequence homology the rapamycin-sensitive mTORC1 (IC50 ¼ 0.02 mM) and in the kinase domain, PIK-75 (compound 9, Figure 2), the rapamycin-insensitive mTORC2 (IC50 ¼ 0.083 mM). TGX-221 (compound 10, Figure 2), AS-605240 (com- The mTOR biochemical activity was further confirmed pound 11, Figure 2) and IC87114 (compound 12, in cells, where the compound blocks downstream Figure 2) are representative examples of p110a, p110b, markers of mTOR signaling (for example, phosphoryla- p110g and p110d modulators, respectively, that have tion of p70 S6 kinase IC50 o0.1 mM). PI103has shown a some level of selectivity against the other p110 paralogs. significant antitumor activity in human glioma xeno- The in vivo exploration—efficacy and tolerability stu- grafts with a significant therapeutic window. Structural dies—of compounds with different selectivity profiles is modifications of PI103and derivatives thereof resulted likely to provide important insights into the optimal in the identification of GDC-0941 (compound 8; kinase activity profile to exploit for drug discovery Figure 2). This ATP-competitive inhibitor is active (Chen et al., 2008). However, as with any other drug against class I PI3K with the following activity profile: discovery project, the ultimate proof of the ‘most IC50 ¼ 3, 33, 3 and 75 nM for p110a, p110b,p110d and desirable lipid kinase activity profile’ will come from p110g, respectively. In biochemical assays, GDC-0941 clinical trials. demonstrates selectivity over PKs and phosphoinositol Although major advances have been made in the 3-kinase related kinase (PIKK) family members (for identification of paralog-selective p110 inhibitors, it example, IC50 ¼ 1.2 and 0.6 mM for DNA-PK and remains to be seen whether p110a mutation-specific mTOR, respectively). It inhibits p-Akt-Ser473 inhibitors for cancer therapeutics can be identified. The (IC50 ¼ 28 nM, MDA-MB-361 in a dose-dependent number of these mutations and the fact that they are manner and causes a cell-cycle G1 arrest. Interestingly, significantly removed from the ATP-binding cleft repre- GDC-0941 induces apoptosis in a subset of human sents a higher medicinal chemistry hurdle, but it cannot tumor cell lines, more frequently in cells bearing p110a be rule out that alternative inhibitory approaches (for mutants. Significant in vivo antitumor activity has been example, allosteric kinase inhibitors or antagonists of observed when administered orally (75–100 mg/kg q.d.) protein–protein interactions) may provide the desirable either as a single agent or in combination with other activity and selectivity profile. Additional structural data anticancer drugs (for example, taxotere and PD0325901) that could provide detailed information on the mechan- in athymic mice bearing human tumor xenografts (for isms by which the most common PIK3CA oncogenic example, U87MG, PC3M and MDA.MB.361). mutations affect p110a catalytic activity and its physical The identification and optimization of the preceding contact with the p85a regulatory domain or other PI3K modulators has certainly benefited by having early interacting proteins will be fundamental to identify access to structural information. The resolution of the p110a mutation-specific modulators. X-ray structure for the p110 subunit of PI3Kg (Walker et al., 2000, 2001), and more recently of the human p110a/p85a complex (Huang et al., 2007) has been Inhibitors of the Ser/Thr kinase activity of PDK1 fundamental in providing a molecular map of the ATP- binding cleft of this special family of kinases providing The 3-phosphoinositide-dependent protein kinase-1 the basis for structure-based design approaches. The (PDK1) is a 556-amino-acid enzyme composed of three sequence identity between p110a and p110g is approxi- well-differentiated motifs: an N-terminal domain, a mately 35% and cross-comparison between the kinase constitutively activated serine/threonine kinase domain, domains of these enzymes shows a root mean square and a plecstrin homology (PH) domain at its C-terminus deviation of 1.8 A˚ between 288 out of 337 aligned Ca (Alessi et al., 1997; Stephens et al., 1998). Unlike PI3K atoms. These subtle, but significant conformational and Akt, only a single PDK1 isoform has been reported shifts, together with a limited number of differences in in humans. The attractiveness of PDK1 as a potential some of the amino acids that form the ATP-binding cleft anticancer target is linked to its ability to phosphorylate of these lipid kinases (for example, aHis885/bGlu858, and activate a diverse set of AGC kinase members aGln859/bAsp862, aSer773/bAsp780 and aAla775/ (named after family members: PKs A, G and C), in

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5516 particular the three Akt isoforms (PKBa/Akt1, PKBb/ hypomorphic mice expressing B10% of the normal level Akt2 and PKBg/Akt3) (Mora et al., 2004). Full of PDK1 display no obvious deleterious phenotype— activation of Akt requires phosphorylation at two sites, they are 40–50% smaller than control animals—(Lawlor one within the activation loop (for example, Thr-308 for et al., 2002), and when crossed to PTEN heterozygote Akt1), and one within the C-terminus (for example, Ser- mice abrogate the formation of PTEN deficient tumors. 473for Akt1). Phosphorylation of the critical and These latter data suggest the possibility that a selective conserved threonine residue in the activation loop (also PDK1 kinase inhibitor might provide effective antic- called the T-loop) of the three Akt isoforms is carried ancer activity with an acceptable therapeutic index. out by PDK1 at the plasma membrane (Alessi et al., One of the most potent, but nonselective PDK1 1996; Stokoe et al., 1997). In addition to Akt, other kinase inhibitor reported to date is UCN-01 (compound members of the AGC kinase subfamily including p70 13; Figure 3; IC50 ¼ 6–33 nM) (Sato et al., 2002). This ribosomal S6 kinase (S6K1), serum and glucocorticoid- staurosporine-based compound, isolated from the cul- regulated PK and PKC are phosphorylated and ture broth of Streptomyces sp, was originally developed activated by PDK1 (Casamayor et al., 1999). The as an inhibitor of calcium-dependent PKC. UCN-01 established role of PDK1 in the activation of the above inhibits a broad array of PKs, including other members proteins and downstream effectors points to its im- of the AGC subfamily of enzymes (for example, portant role in human cancer biology. Finally, PDK1 IC50 ¼ 491 nM for Akt) (Komander et al., 2003). In

H H O N H N O OH O N O

N N N N O N N O H H O O O N N HN O 13 14 15 UCN-01 PKC412 Ly333531

O O H2N S F O N F HN N N H N H N F H2N N H N N O O O O N N N NH2 H H H Br 16 17 18 BX-320BX-517 Celecoxib

NH H2N HN F N F N F

19 OSU-03013 Figure 3 Representative examples of inhibitors of PDK1.

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5517 addition to the activity reported in biochemical assays, between the urea moiety of the compound and the side UCN-01-mediated inhibition of PDK1 has also been chain of the gatekeeper residue, methionine, of PKA. In observed in in vivo murine and human tumor xenografts the case of PDK1, the gatekeeper is leucine and its (Sato et al., 2002). In addition to UCN-01, other shorter side chain may provide room to accommodate staurosporine or maleimide derivatives (for example, the urea group. The binding of the compound was compounds 14 and 15; Figure 3) have been described as confirmed by the solving of the X-ray structure with PDK1 modulators (Hill et al., 2001; Fujita and Tsuruo PDK1. Additional medicinal chemistry activities pro- 2003; Komander et al., 2003). vided PDK1 inhibitors with improved pharmacological UCN-01 has been tested in advanced cancer patients properties (Islam et al., 2007a). in phase I/II clinical trials, both as single agent and in The elucidation of the three-dimensional cocrystal combination with conventional chemotherapeutic drugs structures of PKA, Akt and PDK1 bound to ATP, (for example, carboplatin, fludarabine or topotecan); ATP analogs or kinase inhibitors (Biondi et al., 2000; however, significant antitumor activities have not yet Yang et al., 2002a; Komander et al., 2004) may provide been reported (Welch et al., 2007). In addition to initial insights into how PDK1 kinase selectivity might be pulmonary toxicity, nausea/vomiting, lactic acidosis and achieved with ATP-competitive inhibitors. We can expect transaminitis, UCN-01 induced insulin resistance in the that these structure-based design efforts combined with clinic. Owing to the relatively poor selectivity of UCN- new screening methods (for example, inverse in silico 01, it is unclear which inhibited kinase(s) is/are involved screening; Zahler et al., 2007) may provide the basis for in this latter side effect, although as shown recently in the identification and development of a new generation of rat adipose cells, the observed insulin resistance may be PDK1 modulators with the desirable activity/selectivity owing to the inhibition of Akt phosphorylation and profile and pharmacological properties. subsequent blockade of insulin-dependent GLUT4 Inhibitory activity against PDK1 has also been translocation to the plasma membrane (Kondapaka reported for compounds originally designed to antagonize et al., 2004). If UCN-01-mediated insulin resistance a different therapeutic target. For example, it has been results from PDK1 inhibition, then, contrary to what found that celecoxib (compound 18, Figure 3), which is a was observed in the PDK1 hypomorphic mice (vide cyclooxygenase-2 inhibitor, can block the activation of supra), insulin resistance may represent an important Akt (Hsu et al., 2000) in a variety of cancer cells by hurdle in the development of PDK1 inhibitors. inhibiting PDK1 enzymatic activity (Arico et al., 2002). In A class of aminopyrimidine derivates has also been an immunoprecipitation assay, the compound inhibits the reported to inhibit PDK1 kinase activity. A representa- kinase activity of PDK1 in a dose-dependent manner with tive example of this compound class is BX-320 an IC50 value of 3.5 mM. However, weaker inhibitory (compound 16, Figure 3), an ATP-competitive PDK1 activity (IC50 ¼ 30 mM) has been obtained when celecoxib inhibitor (IC50 ¼ 39nM) that displays good selectivity was tested in another cell-free assay using the recombinant over AGC family members (for example, 35- and 146- PDK1 protein. Interestingly, celecoxib inhibits PDK1 by fold for PKA and PKC, respectively) and other PKs competing with ATP for binding, and on the basis of this (Feldman et al., 2005). BX-320 blocks the growth in soft result and using a structure-based design optimization agar of a wide range of tumor cell lines (IC50 ¼ 0.093– strategy, celecoxib derivatives (for example, OSU-03013, 1.32 mM), including PTEN-negative tumor cell lines such compound 19; Figure 3) with slightly improved anti- as U87MG and PC3M. The compound is also proliferative activity (IC50 ¼ 3 mM)wereidentified(Zhu efficacious in a metastasis mouse model—LOX mela- et al., 2004). Celecoxib itself is currently being investigated noma tumors—when administered orally at 200 mg/kg in phase II/III clinical studies as a single agent or in b.i.d. The resolution of the X-ray structure of the kinase combination therapy. domain of PDK1 with BX-320 revealed a classical two H-bond interactions between the hinge region—the amino-acid backbone of Ala-162—of the enzyme and Inhibitors of the Ser/Thr kinase activity of Akt two of the aminopyrimidine nitrogens of the inhibitor. In this context, the mode of binding of BX-320 is similar Akt (also known as PKB) is a serine/threonine kinase to the one reported for aminopyridines in complex with belonging to the AGC PK subfamily. Three mammalian cyclin-dependent kinase 2/cyclin A (Sayle et al., 2003). isoforms—Akt1, Akt2 and Akt2—have been identified The BX series of molecules has not yet entered into the and are broadly expressed although isoform-specific clinical trials at the time of this review. patterns of expression exists in some tissues. The three More recently, the identification and initial medicinal enzymes have a similar organizational structure: an N- chemistry optimization of an indoline-based series of terminal PH domain, a central serine/threonine catalytic PDK1 modulators has been reported (Islam et al., domain and a short regulatory region at the C-terminal 2007a, b). Using a homology model of the initial end containing the so called hydrophobic motif (Jones indolinone hit bound to PDK1, BX-517 (compound et al., 1991a, b; Yang et al., 2002b). 17, Figure 3) was identified as a potent PDK1 inhibitor Akt catalyzes the phosphorylation of serine/threonine (IC50 ¼ 6nM) with significant selectivity over PKA residues at consensus phosphorylation sites—R-X-R-X- (IC50 ¼ 1600 nM; Islam et al., 2007b). Although not X-S/T—in molecular targets, including proteins with fully understood, the observed selectivity over PKA was key roles in the regulation of cell growth and survival. rationalized on the basis of negative, steric interactions As shown in Figure 1, Akt is a key downstream effector

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5518 of PI3K. Following the growth factor stimulation of negative interactions do not occur in the modeled PI3K, Akt is recruited at the plasma membrane, an complex of the inhibitor bound to Akt; in this protein, event accomplished by physical or direct interaction the amino acid at position 187 is leucine and not between its PH domain and the generated PtdIns phenylalanine. The activity of these azepane derivatives (3,4,5)P3 molecules (see previous section). Colocaliza- in cellular settings or in vivo models has not been tion of PDK1 at the plasma membrane allows the reported yet. phosphorylation of Thr308, located in the activation Promising Akt inhibitors have also been obtained loop, by this enzyme, which is necessary and sufficient from indazole–pyridine-based derivatives (Luo et al., for Akt activation (Alessi et al., 1996; Bellacosa et al., 2005). Exemplified by A-443654 (compound 25, Fig- 1998). However, maximal enzymatic activation requires ure 4; Ki ¼ 160 pM for Akt1), these compounds are phosphorylation of Ser473located in the C-terminal reversible, ATP competitive inhibitors, which decrease hydrophobic motif. The molecular identity/ies of the the phosphorylation of Akt downstream targets in cells kinase/s, previously referred to as PDK2, required for (for example, GSK3a/b, FOXO3, TSC2 and mTOR) the phosphorylation of Ser473has/have been contro- and in vivo in a dose-dependent manner. Interestingly, versial (Feng et al., 2004), but compelling data point to these biological effects appear to induce Akt Ser-473 the mTOR/rictor complex as the key enzyme (Sarbassov phosphorylation in human cancer cell lines, including et al., 2005). More recently, DNA-PK was also found to PTEN- and TSC2-deficient cell lines (Han et al., 2007). phosphorylate S473under conditions of DNA damage Although several possibilities were considered (for (Feng et al., 2004; Bozulic et al., 2008). example, a conformational change in the structure of Akt isoforms are overexpressed in a variety of human the compound-bound Akt that prevents Ser-473de- tumors, and, at the genomic level, are amplified in phosphorylation or changes in Akt relocalization), gastric adenocarcinomas (Akt1), ovarian (Akt2), pan- further studies will be required to delineate the creatic (Akt2) and breast (Akt2) cancers (Hill and mechanism(s) involved in this interesting observation. Hemmings, 2002; Mitsiades et al., 2004). In addition, A-443654 shows some level of selectivity against other many tumor cells display elevated levels of PI3K members of the AGC family: PKA, 40-fold; PKCg, 150- enzymatic products as a result of deletion/mutations of fold; PKCd, 200 and PDK1, >120 000-fold (Luo et al., PTEN, activation of Ras or expression of autocrine 2005). A structural comparison of A-443654 bound to growth factors (Maira et al., 2008b). The biological Akt and PKA demonstrated that the compound adopts consequences of uncontrolled Akt activation in tumor a binding conformation in Akt that differs from that cells are critical for inducing, among other effects, a with PKA (Davies et al., 2007). In particular, the survival signal that allows them to withstand apoptotic reported differences in potency of A-443654 for Akt stimuli. Thus, Akt protects tumor cells from death by compared with PKA seem to arise as a result of one of phosphorylating and inactivating components of the the three key residue differences between the ATP- intrinsic apoptosis pathway (Downward, 2006). binding clefts: namely the presence of methione-282 in Nonselective Akt inhibitors have been extensively place of leucine at the base of the Akt cleft. This Leu– used as tool compounds to elucidate the role of this Met substitution allows the inhibitor to adopt a kinase in the biology of human cancers. Akt isoforms conformation where it can form increased lipophilic are potently inhibited by promiscuous kinase inhibitors contacts with the enzyme. Moreover, the folded like staurosporine (compound 20, Figure 4; IC50 ¼ 48– conformation adopted by the compound leads to the 11 nM for Akt1) and derivatives thereof (for example, energetically favorable burial of approximately 50 A˚ 2 of compound 21, Figure 4) (Kumar et al., 2001; Li and surface area within the inhibitor itself. Overall, the Zhu, 2002). Similarly, Ro-31-8220 (compound 22, preceding results nicely show that even when there is a Figure 4) is a potent, but nonselective pan-Akt inhibitor, high sequence identity in the ATP-binding pockets of with primary activity against PKC (IC50 ¼ 10 nM) PKs, subtle amino-acid differences can be exploited to (Davis et al., 1992) and more modest activity against achieve selectivity for the intended target. Akt1 (IC50 ¼ 240 nM). In animal experiments, the compound showed anti- Exploiting the X-ray crystal structure of (À)-balanol, tumor activity as a single agent and in combination which is a potent inhibitor of AGC-kinases, in complex regimens in a number of tumor xenografts, but the with PKA, medicinal chemistry efforts were directed to dosing period was limited owing to malaise and weight impart Akt selectivity and improve its pharmaceutical loss (Shi et al., 2005). Although further studies with properties (Gassel et al., 2003; Breitenlechner et al., other chemotypes will be required to assess the general- 2004, 2005). Starting with compound 23(Figure 4) as a ity of these in vivo findings, the side effects observed in lead compound (IC50 ¼ 5nM for Akt1 and PKA), a new animal models raise concerns that the therapeutic series of potent Akt inhibitors (for example, compound application of Akt inhibitors will be limited by 24, Figure 4; IC50 ¼ 20 nM for Akt versus mechanism-based metabolic toxicities. Recently, com- IC50 ¼ 1900 nM for PKA) were obtained by exploiting pounds with improved potency, selectivity and cardio- a binding site in Akt that is narrower in PKA. The X-ray vascular safety as compared with A-443654, and structure of compound 24 bound to PKA shows that the derivatives thereof, have been reported (Zhu et al., piperidine moiety of the inhibitor adopts an energeti- 2004). A representative example of this series is cally unfavorable envelop conformation to avoid a steric compound 26 (Figure 4) (IC50 ¼ 0.6 nM, Akt1), which clash with the side chain of phenylalanine-187; these is orally bioavailable in mice (F ¼ 25%) and did not

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5519 H H H N O O N O N O

N N O N O N N N H NH O O HO HN OH H2N S 20 21 22 Staurosporine K-252b Ro-31-8220

HO F O N O O HN N HN N O F N HN N H N N . N NH2 . 2 HCl H H 3 HCl N O N H O O O H

23 24 25 A-443654

N OH

O NH2 H2N N HN N N N H N NH2 N F N N N O F F O

26 27 28 GSK690693

N N HO OH O N N N O O S H2N N N NH2 N N O O O 29 30 Lactoquinomycin Figure 4 Representative examples of inhibitors of Akt. cause hypotension when administered orally in con- reaching an 880-fold increase in relation to the baseline scious mice up to 150 mg/kg. level at 4 h postdosing. In spite of these observations, the GSK690693(compound 27, Figure 4) is an ATP- compound was well tolerated with once daily intraper- competitive, pan-Akt kinase inhibitor (for example, itoneal administration and showed significant antitumor IC50 ¼ 2nM for Akt1) that recently entered phase I activity in human cancer xenografts (for example, clinical trials (Rhodes et al., 2008). This aminofurazan SKOV-3, LNCap, BT474 and HCC-1954). The first derivative is also active (IC50o80 nM) against members phase I study in humans to investigate the safety, of the AGC, CAMK and group II PAK kinase families tolerability, PK and pharmacodynamics of this Akt that can contribute to its reported preclinical, antitumor inhibitor given weekly or twice weekly as an i.v. infusion activity. Acute intraperitoneal treatment of severe over 1–4 h is ongoing. combined immunodeficient mice bearing BT474 tumors If the identification of potent and selective Akt kinase with GSK690693(20 mg/kg) resulted in acute hypergly- inhibitors has proven difficult, the design of Akt isoform cemia with blood glucose levels returning to baseline as selective modulators represents a higher medicinal the circulating drug concentration decreases. A rapid chemistry hurdle. The kinase domain of Akt1 has a elevation of plasma insulin levels was also observed, sequence identity of 90 and 87% with Akt2 and Akt3,

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5520 respectively. Notably, the amino acids that line the Allosteric inhibition of Akt has also been reported for ATP-binding cleft of Akt1 are strictly conserved in lactoquinomycin (compound 29, Figure 4) (Toral-Barza Akt2, and only one amino acid is different in Akt2 et al., 2007), a natural product isolated from the (alanine in Akt1/2 versus valine in Akt3). A detailed fermentation broth of a microbial strain Streptomycin molecular map of the ATP-binding pocket of these sp. LL-AF101. This pyranonathpthoquinone derivative, kinases has been obtained recently by solving the X-ray which was identified by screening a natural product crystal structure of DPH-Akt2 with adenyl-imidodipho- collection, inhibits the kinase activity of Akt1 with an sphate tetralithium salt and a GSK3b-derived peptide IC50 value of 0.149 mM and shows significant selectivity (Yang et al., 2002a, b), but the structure-based design of over a panel of 45 PKs. Elucidation of its mechanism of ATP-competitive Akt isoform selective inhibitors will action confirmed the formation of a covalent, irrever- probably require access to the structures of the full Akt sible bond with the Akt, and mutational analysis isoforms. Amino acids away from the ATP-binding highlighted the T-loop cysteine residues—Cys-296 and pocket or the interaction of the kinase domain with Cys-310—as critical for kinase inhibition. These two other parts of the protein may alter the size and shape of cysteines surround the critical activating Thr-308 amino the ATP-binding cleft and hence the binding affinity and acid in the catalytic site. In cellular settings, the selectivity of an ATP-competitive modulator. compound blocks the phosphorylation and activation Clearly, other pockets besides the ATP-binding cleft of Akt and downstream effectors at concentrations can be exploited for the identification and development consistent with its Akt inhibitory activity. Although this of Akt kinase modulators. It is notable that allosteric mechanism of Akt inhibition represents an interesting inhibitors of Akt containing the 2,3-diphenylquinoxa- and intriguing approach that might also be applicable to line or 5,6-diphenyl-pyrazin-2(1H)-one scaffolds have other T-loop cysteine-containing kinases, lactoquino- been described (Barnett et al., 2005; Lindsley et al., mycin cannot be developed as a drug, among other 2005; Zhao et al., 2005). Here, a high throughput factors, owing to its nonspecific redox-related cytotoxi- homogenous time-resolved fluorescence kinase assay city that limits its pharmacological properties and was used to screen around 270 000 compounds for their therapeutic window. It remains to be seen whether the ability to inhibit Akt enzymatic activity. The identified structural features of lactoquinomycin responsible for compound exhibited a linear mixed-type inhibition its allosteric, covalent Akt kinase inhibition can be against ATP and peptide substrate, showed isozyme incorporated in other chemical analogs. selectivity, and were only active against the full-length protein (for example, compound 28, Figure 4; IC50 ¼ 2.7 mM for Akt1 versus IC50>250 mM for DPH-Akt1). Other therapeutic modes to block the PI3K/Akt pathway Although the mechanism of inhibition by these allosteric kinase inhibitors has not been fully elucidated and there In addition to directly targeting the kinase components are no structural data yet, it seems that these molecules of the PI3K/Akt pathway, alternative therapeutic modes may bind outside the ATP-binding pocket, interacting have been pursued to control aberrant pathway activity. with the PH domain and/or hinge region likely Phosphatidylinositol analogs, inositol polyphosphates, promoting the formation of an inactive conformation. phospholipids and hsp90 inhibitors have been used for It is important to note that these allosteric inhibitors are pathway interruption. Phospholipids and hsp90 inhibi- slightly less potent than the classical ATP-competitive tors are reviewed below. inhibitors in blocking the kinase activity of a recently reported Akt1 mutant found in breast cancer samples. Phospholipids—antagonists of PH domains The mutated protein contains a lysine amino acid Phospholipid-containing molecules have been used in instead of glutamic acid at position 17 in the lipid- recent years to modulate membrane function and binding pocket of Akt1 (Carpten et al., 2007). This signaling targets that use naturally occurring lipid single amino-acid mutation leads to constitutive asso- moieties as substrates or cofactors. Perifosine (NSC ciation of Akt1 with the plasma membrane, constitutive 639966/D-21266 compound 31, Figure 5) appears to activation of the pathway and induces leukemia in mice. block the activation and phosphorylation of Akt in In spite of its relatively low frequency, the sensitivity of cellular settings (Kondapaka et al., 2003; Ruiter et al., this mutant to the currently reported Akt modulators, 2003). Perifosine is an alkylphosphocholine structurally including the allosteric inhibitors reported in the related to miltefosine, but has an improved tolerability preceding paragraphs, may serve as a clinically relevant profile with regard to the dose-limiting gastrointestinal predictive marker for directing Akt drug development. side effects reported for the latter. Although the Several follow-up papers have illustrated medicinal mechanism of action of perifosine is not fully under- chemistry efforts directed to improving the potency and stood, one hypothesis is that, following insertion into drug-like properties of this compound class (Hartnett the cellular membrane, perifosine interferes with Akt et al., 2008; Zhao et al., 2008). The optimized inhibitor membrane localization by inhibiting the association of from the disclosed series (compound 29, Figure 4) its PH domain with PtIns(3,4,5)P3. In keeping with this inhibits Akt1 and Akt2 with IC50 values of 126 and notion, perifosine blocks Akt plasma membrane locali- 22 nM, respectively, has a significant selectivity profile zation (Engel et al., 2000) as assessed by immunofluor- against some AGC PKs (IC50>10 mM for PKA and escence imaging (Kondapaka et al., 2003). Perifosine has PKCa) and a desired pharmacological profile in dogs. marked cytotoxic effects on human tumor cell lines, in

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5521 O R1 O

N OH O O O H O O N+ O P O C18H37 OH O O HO OH O Cl

O NH2 O 1 31 Geldanamycin R = OCH3 32 33 Perifosine 17-AAG 1 Radicicol R = NHCH2CH=CH2 34 1 17-DMAG R = NHCH2CH2N(CH3)2 35

_ Cl OH O N+ O Cl H N N N N H OH N O H2N N N OH O N O H N O O O NH2 HO OH O 36 37 38 IPI-504 CNF2024 NVP-AUY922 Figure 5 Structure of perifosine and representative examples of hsp90 inhibitors. particular larynx carcinoma, breast cancer, small cell (grade 1–3) were the most common side effects at the lung cancer, prostate cancer and colon carcinoma. Of weekly and daily schedules. Encouraging evidence of special interest is the synergistic antiproliferative effect antitumor activity as single agent or in combination observed when perifosine was used in combination with with radiation was reported in phase I clinical trials, in UCN-01 in cell culture assays with PC3and A459 tumor particular, in the treatment of non-small cell lung cancer cells (Dasmahapatra et al., 2004), with the antiepidermal patients. However, no significant clinical activity was growth factor receptor antibody, cetuximab, in PTEN- observed in phase II studies in previously untreated deficient cancer cells (Li et al., 2006) and with the hsp90 patients with metastatic melanoma (Ernst et al., 2005), inhibitor 17-DMAG (see next section) in multiple and in patients with progressive, metastatic androgen- myeloma cells (Huston et al., 2008). Blocking simulta- independent prostate cancer (Posadas et al., 2005). An neously distinct components of the PI3K/Akt pathway extensive phase II clinical program is currently under seems to enhance the inhibition of phosphorylation of way, including trials in breast, head and neck, kidney, Akt along with activation of the apoptotic pathway. lung, prostate, glioma, leukemia, sarcoma and multiple Recently, it has been shown a statistically significant myeloma. correlation between the antitumor efficacy of perifosine and quantified downregulation of p-Akt and p-S6 Inhibitors of the ATPase activity of heat–shock protein 90 (Hennessy et al., 2007). Moreover, results obtained Heat–shock proteins are ATP-dependent molecular from a study of perifosine in MT-1 human breast chaperones involved in the folding, stability, activation xenograft suggests that immunomodulation may partly and cellular localization of a selected range of sub- contribute to the anticancer activity of this compound strates, the so-called ‘client proteins’ (Whitesell and (Safa et al., 1998). Despite these data, the specific Lindquist, 2005). The Hsp90 family of chaperones is antitumor mechanisms of perifosine and its relationship composed of four isoforms: Hsp90a, Hsp90b, GRP94 to Akt inhibition remain unclear. and TRAP-1. The 90 KDa chaperone binds to client From the results of phase I studies, a weekly dose of proteins in the presence of other partner proteins to 600 mg or 250 mg/day was selected for phase II studies produce a multiprotein complex that folds the target (Traiser et al., 1998). Nausea, vomiting and diarrhea substrate into its biologically active conformation(s).

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5522 Binding and release of Hsp90 client proteins are and combination trials with 17-AAG are currently being regulated by the activity of the N-terminal ATPase carried out. domain, which binds and hydrolyzes ATP to mediate a More recently, additional geldanamycin analogs with series of association–dissociation cycles between Hsp90 increased chemical/metabolic stability and formulation and its substrate. options have been reported. Included among these Heat–shock proteins are likely involved in dealing are 17-(2-dimethylaminoethyl)amino-17-demethoxygel- with the cellular stress associated with a hostile cancer danamycin (17-DMAG, KOS1022; alvespimycin; com- environment, as well as being essential for the proper pound 35, Figure 5; Tian et al., 2004; Hollingshead function/folding of a number of key oncogenic proteins. et al., 2005) and JPI-504 (compound 36, Figure 5; Ge Many of the proteins that interact with Hsp90 are key et al., 2008), each having entered clinical trials. players in signal transduction pathways that are Alvespimycin is a water-soluble analog (11.2 mg/ml) essential to mediate and sustain tumor cell growth and of 17-AAG that has shown in preclinical models survival. For example, functional Hsp90 is required for comparable or somewhat better antitumor activity in the correct folding and stability of Akt and PDK1. comparison to 17-AAG. The compound had been in (Basso et al., 2002; Fujita et al., 2002; Solit et al., 2003). phase I clinical trials for the treatment of hematological Given the critical roles played by these and other client malignancies and in phase II clinical studies for the proteins in tumor growth and survival, inhibition of treatment of HER2-positive metastatic breast cancer in hsp90 has emerged as a possible strategy for the combination with trastuzumab. It seems that the clinical treatment of human tumors. development of this compound has been discontinued. The chaperoning function of Hsp90 can be ‘switched- IPI-504 (compound 36, Figure 5), which is a hydro- off’ by inhibiting its ATPase activity (Chiosis et al., quinone hydrochloride derivative of 17-AAG, intercon- 2004; Dymock et al., 2004) and indeed, most current verts with 17-AAG in vitro and in vivo through an Hsp90 inhibitors act by binding to the ATP pocket in oxidation/reduction equilibrium (Ge et al., 2008). It the N-terminal domain of the protein. Although other shows a comparable binding affinity for Hsp90 inhibitory mechanism are currently under exploration (EC50 ¼ 63n M) and Grp94 (EC50 ¼ 119 nM) and po- (for example, interference with cochaperone binding and tently inhibits the growth of tumor cell lines that function in the so-called superchaperone complex or overexpress Hsp90 client protein Her2 (for example, binding to the C-terminal nucleotide binding site), this SKBR3, GI50 ¼ 22 nM and SKOV3, GI50 ¼ 52 nM). As section is focused on the compounds targeting the N- for other Hsp90 modulators, it is preferentially retained terminal ATP binding cleft (for a recent review on this in tumor tissues relative to plasma, and it is efficacious topic, see Bishop et al., 2007). alone (for example, 100 mg/kg 2xw) and in combination The initial discovery of the potential use of Hsp90 with the 20S proteasome inhibitor bortezomib when inhibitors as novel anticancer agents was based on two administered i.v. to mice bearing the RPMI-8226 natural products, geldanamycin (compound 32, Fig- multiple myeloma model (Sydor et al., 2006). On the ure 5) and radicicol (also called monorden, compound basis of its significant improvement in water solubility 33, Figure 5). These compounds bind into the ATP and pharmacological properties in relation to 17-AAG, binding cleft of the N-terminal domain of Hsp90 IPI-504 entered phase I clinical trials and its i.v. preventing the chaperone from cycling between its formulation is currently in phase I/II clinical trials in ADP and ATP-bound conformations (Stebbins et al., non-small cell lung cancer (NSCLC) and imatinib- 1997; Roe et al., 1999). Both natural products are highly resistant gastrointestinal stromal tumor (GIST) tumors. active in preclinical models (Neckers and Neckers, In the phase I trial in GIST and other sarcomas, five 2002), but, owing to their poor solubility and other metabolic responses were noted, with one true partial pharmacological issues, the clinical evaluation of response as determined by response evaluation geldanamycin and radicicol has not been pursued. In criteria in solid tumors (RECIST) criteria. Clinical the case of geldanamycin, extensive medicinal chemistry activity (as defined by positron emission tomography efforts have been made to replace the 17-methoxy (PET) response) has also been reported for IPI504 in substituent and generate analogs with improved phar- advanced nonsmall-cell lung cancer tumors and in maceutical properties. One of these derivatives, 17- gastrointestinal stromal tumors. Interestingly, one pa- allyamino-17-demethoxygeldanamycin (17-AAG, com- tient with a mutation in epidermal growth factor pound 34, Figure 5; Jia et al., 2003), a first-in-class receptor, an oncogenic client protein of Hsp90, and Hsp90 drug, has provided proof-of-concept for Hsp90 prior history of progression on targeted kinase inhibi- inhibition in patients at tolerated doses—(phase I tors experienced extended stable disease over seven studies are reviewed in Ramanathan et al., 2007). More cycles (E27 weeks). recently, KOS953(tanespimycin), which contains a Novel HSP90 inhibitor chemotypes are now emerging proprietary form of 17-AAG in a novel, optimized based on structure-based design and high throughput formulation—polyethoxylated castor oil—achieved clin- screening approaches (Janin, 2005; Bishop et al., 2007). ical proof-of-concept with trastuzumab. This combina- Representative examples of low-molecular mass Hsp90 tion was well tolerated and tumor regressions were inhibitors that have entered clinical trials are described observed exclusively in patients with human epidermal in the following paragraphs. growth factor receptor 2 (erbB2)-positive metastatic CNF2024 (originally called BIIB021; compound 37, breast cancer (Modi et al., 2007). Phase II monotherapy Figure 5) is a synthetic orally available Hsp90 mod-

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5523 ulator that entered clinical trials in October 2005; (50 mg/kg intraperitoneal or i.v.) to mice bearing human however, only minimal information is currently in the tumor xenografts with diverse oncogenic profiles pro- public domain (Lundgren et al., 2007). BIIB021 duced statistically significant growth inhibition and/or promotes the degradation of the Hsp90 client protein regression, and significant target inhibition evidenced by HER-2 with an EC50 of 32 nM in MCF-7 cells and the depletion of client proteins and induction of hsp70. inhibits tumor growth at tolerated doses in human Significant antitumor activity was also observed when tumor xenograft models expressing high levels of HER2 the compound was administered on a weekly i.v. (for example, N87) or the androgen receptor (for schedule (Jensen et al., 2008). NVP-AUY922 has example, CWR22). CNF2024 is in phase II clinical entered phase I clinical trials. trials for the treatment of GIST, phase I trials in advanced solid tumors and B-cell chronic lymphocytic leukemia and in phase Ib/II in combination with trastuzimab in patients with advanced breast cancer. Conclusions SNX-2112 (structure not disclosed) was identified using a protein affinity-displacement assay in which a The introduction of modulators of the PI3K/Akt purine-based affinity resin is used to trap purine-binding pathway as potential targeted anticancer agents is still proteins derived from cell lysates. SNX-2112 binds in an early stage and, despite the lack of complete to hsp90a and b with low nanomolar affinity specificity against the intended target, several interesting (IC50 ¼ 30nM), with moderate selectivity over the other compounds have entered clinical trials. The differences hsp90 family members (IC50 ¼ 862 and 4275 nM, respec- in therapeutic response and toxicity observed with the tively) (Chandarlapaty et al., 2008). As in the case of currently available inhibitors in preclinical models geldanamycin and derivatives thereof, SNX-2112 sig- suggest that, besides the potential biological effects nificantly inhibits the proliferation of tumor cell lines by owing to the concomitant inhibition of other off-targets inducing the degradation of relevant client proteins. and/or the intrinsic pharmacological properties of the SNX-5542 (structure not disclosed), which is a water- compound, blocking the PI3K/Akt pathway at different soluble prodrug of SNX-2112, can be administered nodes might yield different antitumor activities and orally to tumor-bearing mice and is rapidly converted to therapeutic windows. Although the likelihood of an SNX-2112 where it preferentially accumulates in tumor inhibitor with no off-target activity owing to the tissues, a feature reported for all the current hsp90 inhibition of homologous protein is low, additional modulators. Antitumor activity as a single agent (50 mg/ preclinical and clinical studies with better characterized kg, orally, Monday to Friday schedule) was reported in compounds or, if possible, more selective inhibitors will BT474 and H1650 xenografts. SNX-2112 has entered be required to determine where and how to block the phase I clinical trials. PI3K/Akt pathway to provide the desired therapeutic NVP-AUY922 (compound 38, Figure 5) is one of the outcome. For example, it is unclear whether the activity most potent synthetic small molecular mass hsp90 reported for some of the PI3K inhibitors now in the inhibitor yet described (Ki ¼ 9.0 and 8.2 nM for Hsp90a clinic for members of the lipid kinase family (for and b, respectively; Brough et al., 2008; Eccles et al., example, Vps34 or DNA-PK; Abraham, 2004; Weter- 2008; Stu¨ hmer et al., 2008). This isoxazole derivative ings and Chen, 2007; Yan and Backer, 2007) is going to inhibits the proliferation of human tumor cell lines with be beneficial or preferable to a highly PI3K- or p110a- IC50 values of 2.3–50 nM, and induces marked accumu- selective modulator. These types of preclinical studies lation in either G1 or G1 plus G2-M phases in most cell and the clinical efficacy and tolerability of modulators lines (Eccles et al., 2008). Daily dosing of NVP-AUY922 of the PI3K/Akt are eagerly awaited.

References

Abraham RT. (2004). PI3-kinase related kinases: ‘big’ players in stress- Basso AD, Solit DB, Chiosis G, Giri B, Tsichlis P, Rosen N. (2002). induced signaling pathways. DNA Repair 3: 883–887. Akt forms an intracellular complex with heat shock protein 90 Alessi DR, Andjelkovic M, Caudwell B, Cron P, Morrice N, Cohen P (Hsp90) and Cdc37 and is destabilized by inhibitors of Hsp90 et al. (1996). Mechanism of activation of protein kinase B by insulin function. J Biol Chem 277: 39858–39866. and IGF-I. EMBO J 15: 6541–6551. Bellacosa A, Chan TO, Ahmed NN, Datta K, Malstrom S, Stokoe D Alessi DR, James SR, Downes CP, Holmes AB, Gaffney PRJ, et al. (1998). Akt activation by growth factors is a multiple-step Reese CB et al. (1997). Characterization of a 3-phosphoinositide- process: the role of the PH domain. Oncogene 17: 313–325. dependent protein kinase which phosphorylates and activates Biondi RM, Cheung PCF, Casamayor A, Deak M, Currie RA, Alessi protein kinase B. Curr Biol 7: 261–269. DR. (2000). Identification of a pocket in the PDK1 kinase domain Arico S, Pattingre S, Bauvy C, Gane P, Barbat A, Codogno P et al. that interacts with PIF and the C-terminal residues of PKA. EMBO (2002). Celecoxib induces apoptosis by inhibiting 3-phosphoinosi- J 19: 979–988. tide-dependent protein kinase-1 activity in the human colon cancer Bishop SC, Burlison JA, Blagg BSJ. (2007). Hsp90: a novel target for HT-29 cell line. J Biol Chem 277: 27613–27621. the disruption of multiple signaling cascades. Curr Med Chem 7: Barnett SF, Defeo-Jones D, Fu S, Hancok PJ, Haskell K, Jones RE 369–388. et al. (2005). Identification and characterization of pleckstrin- Bozulic L, Surucu B, Hynx D, Hemmings BA. (2008). PKBa/Akt1 acts homology-domain-dependent and isoenzyme-specific Akt inhibi- downstream of DNA-PK in the DNA double-strand break response tors. Biochem J 385: 399–408. and promotes survival. Mol Cell 30: 203–213.

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5524 Breitenlechner CB, Friebe WG, Brunet E, Werner G, Graul K, Fan QW, Cheng CK, Nicolaides TP, Hackett CS, Knight ZA, Thomas U et al. (2005). Design and crystal structures of protein Shokat KM et al. (2007). A dual phosphoinositide-3-kinase kinase B-selective inhibitors in complex with protein kinase A and alpha/mTOR inhibitor cooperates with blockade of epidermal mutants. J Med Chem 48: 163–170. growth factor receptor in PTEN-mutant glioma. Cancer Res 67: Breitenlechner CB, Wegge T, Berillon L, Graul K, Marzenell K, 7960–7965. Friebe WG et al. (2004). Structure-based optimization of Feldman RI, Wu JM, Polokoff MA, Kochanny MJ, Dinter H, Zhu D novel azepane derivatives as PKB inhibitors. J Med Chem 47: et al. (2005). Novel small molecule inhibitors of 3-phosphoinositide- 1375–1390. dependent kinase-1. J Biol Chem 280: 19867–19874. Brian PW, Hemming HG, Norris GLF. (1957). Wotmannin, an Feng J, Park J, Cron P, Hess D, Hemmings BA. (2004). Identification antibiotic produced by penicillium wortmanni. Br Mycol Soc Trans of a PKB/Akt hydrophobic motif Ser-473kinase as DNA- 40: 365–368. dependent protein kinase. J Biol Chem 279: 41189–41196. Brough PA, Aherne W, Barril X, Borgognoni J, Boxall K, Cansfield JE Foster PG. (2007). Potentiating the antitumor effects of chemotherapy et al. (2008). 4,5-Diaryl-lisoxazole hsp90 chaperone inhibitors: with the selective PI3K inhibitor XL147. 19th AACR-NCI-EORTC potential therapeutic agents for the treatment of cancer. J Med Meeting (abstract C199). Chem 51: 196–218. Fujita N, Tsuruo T. (2003). Survival-signaling pathway as a promising Carnero A, Blanco-Aparicio C, Renner O, Link W, Leal JFM. (2008). target for cancer chemotherapy. Cancer Chemother Pharmacol 52: The PTEN/PI3K/Akt signalling pathway in cancer, therapeutic S24–S28. implications. Curr Cancer Drug Targets 8: 187–198. Fujita N, Sato S, Ishida A, Tsuruo T. (2002). Involvement of Hsp90 in Carpten JD, Faber AL, Horn C, Donoho GP, Briggs SL, Robbins CM signaling and stability of 3-phosphoinositide-dependent kinase-1. et al. (2007). A transforming mutation in the plecstrin homology J Biol Chem 277: 10346–10353. domain of Akt1 in cancer. Nature 448: 439–444. Garlich JR, De P, Dey N, Su JD, Peng X, Miller A et al. (2008). A vascular Casamayor A, Morrice NA, Alessi DR. (1999). Phosphorylation of targeted pan phosphoinosite 3-kinase inhibitor prodrug, SF1126, with Ser-241 is essential for the activity of 3-phosphoinositide-dependent antitumor and antiangiogenic activity. Cancer Res 68: 206–215. protein kinase-1: identification of five sites of phosphorylation Gassel M, Breitenlechner CB, Ruger P, Jucknischke U, Schneider T, in vivo. Biochem J 342: 287–292. Huber R et al. (2003). Mutants of protein kinase A that mimic Chandarlapaty S, Sawai A, Ye Q, Scott A, Silinski M, Huang K et al. the ATP-binding site of protein kinase B (Akt). J Mol Biol 329: (2008). SNX2112, a synthetic heat shock protein 90 inhibitor, has 1021–1034. potent antitumor activity against HER kinase-dependent cancers. Ge J, Normant E, Porter JR, Dembski MS, Gao Y, Georges AT Clin Cancer Res 14: 240–248. et al. (2008). Design, synthesis, and biological evaluation of Chen JS, Zhou LJ, Entin-Meer M, Yang X, Donker M, Knight ZA hydroquinone derivatives of 17-Amino-17-demethoxygeldanamycin et al. (2008). Characterization of structurally distinct, isoform- as potent, water-soluble inhibitors of Hsp90. J Med Chem 49: selective phosphoinosite 30-kinase inhibitors in combination 4606–4615. with radiation in the treatment of glioblastoma. Mol Cancer Ther Han E-H, Leverson JD, McGonigal T, Shah OJ, Woods KW, 7: 841–850. Hunter T et al. (2007). Akt inhibitor A-443654 induces rapid Akt Chiosis G, Vilenchik M, Kim J, Solit D. (2004). Hsp90: the vulnerable Ser-473phosphorylation independent of mTORC1 inhibition. chaperone. Drug Discov Today 9: 881–888. Oncogene 26: 5655–5661. Dasmahapatra GP, Didolkar P, Alley MC, Ghosh S, Sausville EA, Hartnett JC, Barnett SF, Bilodeau MT, Defeo-Jones D, Hartman GD, Roy KK. (2004). In vitro combination treatment with perifosine and Huber HE et al. (2008). Optimization of 2,3,5-trisubstitued pyridine UCN-01 demonstrates synergism against prostate (PC-3) and lung derivatives as potent allosteric Akt1 and Akt2 inhibitors. Bioorg (A549) epithelial adenocarcinoma cell lines. Clinl Cancer Res 10: Med Chem Lett 18: 2194–2197. 5242–5252. Hennessy BT, Lu Y, Poradosu E, Yu Q, Yu S, Hall H et al. (2007). Davies SP, Reddy H, Caivano M, Cohen P. (2000). Specificity and Pharmacodynamic markers of perifosine efficacy. Clinc Cancer Res mechanism of action of some commonly used protein kinase 13: 7421–7431. inhibitors. Biochem J 351: 95–105. Hill MM, Hemmings BA. (2002). Inhibition of protein kinase B/Akt Davies TD, Verdonk ML, Graham B, Saalau-Bethell S, Hamlett CCF, implications for cancer therapy. Pharmacol Ther 93: 243–251. McHardy T et al. (2007). A structural comparison of inhibitor Hill MM, Andjelkovic M, Brazil DP, Ferrari S, Fabbro D, Hemmings binding to PKB, PKA and PKA–PKB chimera. J Mol Biol 367: BA. (2001). Insulin-stimulated protein kinase B phosphorylation on 882–894. Ser-473is independent of its activity and occurs through a Davis PD, Elliott LH, Harris W, Hill CH, Hurst SA, Keech E et al. staurosporine-insensitive kinase. J Biol Chem 276: 25643–25646. (1992). Inhibitors of protein kinase C. 2. Substituted bisindolylma- Hollingshead M, Alley M, Burger AM, Borgel S, Pacula-Cox C, leimides with improved potency and selectivity. J Med Chem 35: Fiebig H-H et al. (2005). In vivo antitumor efficacy of 17-DMAG 994–1001. (17-dimethylaminoethylamino-17-demethoxygeldanamycin hydro- Downward J. (2006). PI-3kinase, Akt and cell survival. Semin Cell Dev chloride), a water-soluble geldnamycin derivative. Cancer Che- Biol 15: 177–182. mother Pharmacol 58: 115–125. Dymock BW, Drysdale MJ, McDonald E, Workman P. (2004). Howes AL, Chiang GG, Lang ES, Ho CB, Powis G, Vuori K et al. Inhibitors of Hsp90 and other chaperones for the treatment of (2007). The phosphatidylinositol 3-kinase inhibitor, PX-866, is a cancer. Expert Opin Ther Pat 14: 837–847. potent inhibitor of cancer cell motility and growth in three- Eccles S, Massey A, Raynaud FI, Sharp SY, Bax G, Valenti M et al. dimensional cultures. Mol Cancer Ther 6: 2505–2514. (2008). NVP-AUY922: a novel heat shock protein 90 inhibitor Hsu AL, Chingm TT, Wang DS, Song S, Rangnekar VM, Chen CS. active against xenograft tumor growth, angiogenesis, and metas- (2000). The cyclooxygenase-2 inhibitor celocoxib induces apoptosis tasis. Cancer Res 68: 2850–2860. by blocking Akt activation in human prostate cancer cells Engel J, Hilgard P, Klenner T, Kutscher B, Nossner G, Traiser M et al. independently of Bcl-2. J Biol Chem 275: 11397–11403. (2000). Perifosine: oncolytic, ether phospholipid. Drugs Future 25: Huang C-H, Mandelker D, Schmidt-Kittler O, Samuels Y, Veculescu 1257–1260. VE, Kinzler KW et al. (2007). The structure of a human p110a/p85a Engelman JA, Chen L, McNamara K, Upadhyay R, Crosby K, Maira complex elucidates the effects of oncogenic PI3Kamutations. M et al. (2008). Differential efficacy of PI3K inhibition in PIK3CA Science 318: 1744–1748. and K-Ras driven murine lung cancers (submitted). Huston A, Leleu X, Jia X, Moreau A-S, Ngo HT, Runnels J et al. Ernst DS, Eisenhauer E, Wainman N, Davis M, Lohmann R, Baetz T (2008). Targeting Akt and heat shock protein 90 produces et al. (2005). Phase II study of perifosine in previously untreated synergistic multiple myeloma cell cytotoxicity in the bone marrow patients with metastatic melanoma. Invest New Drugs 23: 569–575. microenvironment. Clin Cancer Res 14: 865–874.

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5525 Ihle NT, Paine-Murrieta G, Berggren MI, Baker A, Tate WR, Wipf P discovery and SAR of isozyme selective inhibitors. Bioorg Med et al. (2005). The phosphatidylionositol-3-kinase inhibitor PX-866 Chem Lett 15: 761–764. overcomes resistance to the epidermal growth factor receptor Lundgren K, Zhang H, Kamal A, Lough R, Timple N, Sensintaffar J inhibitor gefitinib in A-549 human non-small cell lung cancer et al. (2007). BIIB021 is a small molecule inhibitor of the heat shock xenografts. Mol Cancer Ther 4: 1349–1357. protein, Hsp90, that shows potent anti-tumor activity in preclinical Islam I, Brown G, Bryant J, Hrvatin P, Kochanny MJ, Phillips GB models. 19th AACR-NCI-EORTC Meeting (abstract B161). et al. (2007a). Indoline-based phosphoinositide-dependent kinase-1 Luo Y, Shoemaker AR, Liu X, Woods KW, Thomas SA, de Jong R (PDK1) inhibitors. Part 2: optimization of BX-517. Bioorg Med et al. (2005). Potent and selective inhibitors of Akt kinases slow the Chem Lett 17: 3819–3825. progress of tumors in vivo. Mol Cancer Res 4: 977–986. Islam I, Bryant J, Chou Y-L, Kochanny MJ, Lee W, Phillips GB et al. Maira S-M, Stauffer F, Brueggen J, Furet P, Schnell C, Fritsch C (2007b). Indoline-based phosphoinositide-dependent kinase-1 et al. (2008a). Identification and characterization of NVP-BEZ235, (PDK1) inhibitors. Part 1: design, synthesis and biological activity. a new orally available dual PI3K/mTor inhibitor with potent Bioorg Med Chem Lett 17: 3814–3818. in vivo antitumor activity. Mol Cancer Ther (in press); e-pub ahead Janin YL. (2005). Heat shock protein 90 inhibitors. A text book of print. example of medicinal chemistry? J Med Chem 48: 7503–7512. Maira S-M, Voliva C, Garcia-Echeverria C. (2008b). Class IA PI3 Jensen MR, Schoepfer J, Radimerski T, Massey A, Guy CT, Kinase: from their biological implication in human cancers to drug Brueggen J et al. (2008). NVP-AUY922: a small molecule HSP90 discovery. Expert Opin Ther Targets 12: 223–238. inhibitor with potent antitumor activity in preclinical breast cancer Marrer E, Maira S-M, Schnell C, Garcia-Echeverria C. (2008). models. Breast Cancer Res 10: R33. Integrative approaches to investigate the molecular basis of the Jia W, Yu C, Rahmani M, Krystal G, Sausville EA, Dent P et al. in vivo activity of NVP-BEZ235, a dual pan-PI3K/mTOR inhibitor. (2003). Synergistic antileukemic interactions between 17-AAG and Proc Am Assoc Cancer Res AACR 99th Annual Meeting (abstract 215). UCN-01 involve interruption of RAF/MEK- and AKT-related Mitsiades CS, Mitsiades N, Koutsilieris M. (2004). The Akt pathway: pathways. Blood 102: 1824–1832. molecular targets for anti-cancer drug development. Curr Cancer Jones PF, Jakubowicz T, Hemmings BA. (1991a). Molecular cloning Drug Targets 4: 235–256. of a second form of rac protein kinase. Cell Regul 2: 1001–1009. Modi S, Stopeck AT, Gordon MS, Mendelson D, Solit DB, Bagatell R Jones PF, Jakubowicz T, Pitossi FJ, Maurer F, Hemmings BA. et al. (2007). Combination of trastuzumab and tanespimycin (17- (1991b). Molecular cloning and identification of a serine/threonine AAG, KOS-593) is safe and active in trastuzumab-refractory HER- protein kinase of the second-messenger subfamily. Proc Natl Acad 2-overexpressing breast cancer: a Phase I dose escalation study. Sci USA 88: 4171–4175. J Clin Oncol 25: 5410–5417. Knight ZA, Chiang GG, Alaimo PJ, Kenski DM, Ho CB, Coan K Mora A, Komander D, van Aalten DMF, Alessi DR. (2004). PDK1, et al. (2004). Isoform-specific phosphoinositide 3-kinase inhibitors the master regulator of AGC kinase signal transduction. Sem Cell from an arylmorpholine scaffold. Bioorg Med Chem 12: 4749–4759. Dev Biol 15: 161–170. Knight ZA, Gonzalez B, Feldman ME, Zunder ER, Goldenberg DD, Neckers L, Neckers K. (2002). Heat–shock protein 90 inhibitors as Williams O et al. (2006). Pharmacological map of the PI3K novel cancer chemotherapeutic agents. Exp Opin Emerg Drugs 7: family defines a role for p110ain insulin signaling. Cell 125: 277–288. 733–747. Patnaik A, Lorusso PM, Tabernero J, Laird AD, Aggarwal SK, Komander D, Fairservice A, Deak M, Kular GS, Prescott AR, Papadopoulos KP. (2007). Biomarker development for XL765, a Peter Downes C et al. (2004). Structural insights into the regulation potent and selective oral dual inhibitor of PI3K and mTOR of PDK1 by phosphoinositides and inositol phosphates. EMBO J currently being administered to patients in a phase I clinical trial. 23: 3918–3928. EORTC-Meeting (abstract B265). Komander D, Kular GS, Bain J, Elliott M, Alessi DR, van Aalten Posadas EM, Gulley J, Arlen PM, Trout A, Parnes HL, Wright J et al. DMF. (2003). Structural basis for UCN-01 (7-hydroxystaurospor- (2005). A Phase II study of perifosine in androgen independent ine) specificity and PDK1 (3-phosphoinositide-dependent protein prostate cancer. Cancer Biol Ther 4: 1133–1137. kinase-1) inhibition. Biochem J 375: 255–262. Ramanathan RK, Egorin MJ, Eiseman JL, Ramalingan S, Kondapaka SB, Singh SS, Dasmahapatra GP, Sausville EA, Roy KK. Friedlan D, Agarwala SV et al. (2007). Phase I and pharmacody- (2003). Perifosine, a novel alkylphospholipid, inhibits protein kinase namic study of 17-(allylamino)-17-demethoxygeldanamycin in adult B activation. Mol Cancer Ther 2: 1093–1103. patients with refractory advanced cancers. Clin Cancer Res 13: Kondapaka SB, Zarnowski MJ, Yver DR, Sausville EA, Cushman SW. 1769–1774. (2004). 7-hydroxystaurosporine (UCN-01) inhibition of Akt Thr308 Raynaud FI, Eccles S, Clarke PA, Hayes A, Nutley B, Alix S et al. but not Ser473phosphorylation. A basis for decreased insulin- (2007). Pharmacologic characterization of a potent inhibitor of class stimulated glucose transport. Clin Cancer Res 10: 7192–7198. I phosphatidylinositide 3-kinases. Cancer Res 67: 5840–5850. Kumar CC, Diao R, Yin Z, Liu Y, Samatar AA, Madison V et al. Redaelli A, Lee JM, Stephens JM, Pashos CL. (2003). Epidemiology (2001). Expression, purification, characterization and homology and clinical burden of acute myeloid leukemia. Expert Rev modeling of active Akt/PKB, a key enzyme involved in cell survival Anticancer Ther 3: 695–710. signaling. Biochim Biophys Acta 1526: 257–268. Rhodes N, Heerding DA, Duckett DR, Eberwein DJ, Knick VB, Laird D. (2007). XL765 targets tumor growth, survival, and Lansing TJ et al. (2008). Characterization of an Akt kinase inhibitor angiogenesis in preclinical models by dual inhibition of PI3K and with potent pharmacodynamic and antitumor activity. Cancer Res mTOR. 19th AACR-NCI-EORTC Meeting (abstract B250). 68: 2366–2374. Lawlor MA, Mora A, Ashby PR, Williams MR, Murray-Tait V, Roe SM, Podromou C, O’Brien R, Ladbury JE, Piper PW, Pearl LH. Malone L et al. (2002). Essential role of PDK1 in regulating cell size (1999). Structural basis for inhibition of the Hsp90 molecular and development in mice. EMBO J 21: 3728–3738. chaperone by the antitumor antibiotics radicicol and geldanamycin. Li Q, Zhu GD. (2002). Targeting serine/threonine protein kinase B/ J Med Chem 42: 260–266. Akt and cell-cycle checkpoint kinases for treating cancer. Curr Top Ruiter GA, Zerp SF, Bartelin H, Van Blitterswijk WJ, Verheij M. Med Chem 2: 939–971. (2003). Anti-cancer alkyl-lysophospholipids inhibit the phosphati- Li X, Luwor R, Lu Y, Liang K, Fan Z. (2006). Enhancement of dylinositol 3-kinase-Akt/PKB survival pathway. Anticancer Drugs antitumor activity of the anti-EGF receptor monoclonal antibody 14: 167–173. cetuximab/C225 by perifosine in PTEN-deficient cancer cells. Safa O, Parkin SM, Bibby MC. (1998). Morphological changes and Oncogene 25: 525–535. cytokine gene expression in tumor xenografts following treatment Lindsley CW, Zhao Z, Leister WH, Robinson RG, Barnett SF, with alkylphosphocholines hexadecylphosphocholine and perifo- Defeo-Jones D et al. (2005). Allosteric Akt (PKB) inhibitors: sine. Drugs Today 34: 15–26.

Oncogene Drug discovery approaches targeting the PI3K pathway C Garcia-Echeverria and WR Sellers 5526 Sarbassov D, Guertin DA, Ali SM, Sabatini DM. (2005). Phosphor- Traiser M, Reichert S, Voss A. (1998). Current development status of ylation and regulation of Akt/PKB by the rictor-mTOR complex. the second generation alkylphosphochloline analog perifosine. Science 307: 1098–1101. Drugs Today 34: 67–71. Sato S, Fujita N, Tsuruo T. (2002). Interference with PDK1-Akt Vlahos CJ, Matter WF, Hui KY, Brown RF. (1994). A specific survival signaling pathway by UCN-01 (7-hydroxystaurosporine). inhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8- Oncogene 21: 1727–1738. phenyl-4H-1-benzopyran-4-one (LY294002). J Biol Chem 269: Sayle KL, Bentley J, Boyle FT, Calvert AH, Cheng Y, Curtin NJ et al. 5241–5248. (2003). Structure-based design of 2-arylamino-4-cyclohexylmethyl- Walker EH, Pacold ME, Perisic O, Stephens L, Hawkins PT, 5-nitroso-6-aminopyrimidine inhibitors of cyclin-dependent kinases Wymann MP et al. (2001). Structural determinants of phosphoino- 1 and 2. Bioorg Med Chem Lett 13: 3079–3082. sitide 3-kinase inhibition by wortmannin, LY294002, quercetin, Schnell C, Stauffer F, Allegrini PR, O’Reilly T, McSheehy PMJ, myricetin, and staurosporine. Molecular Cell 6: 909–919. Dartois C et al. (2008). Effects of the dual PI3K/mTOR inhibitor Walker EH, Perisic O, Ried C, Stephens L, Williams RL. (2000). NVP-BEZ235 on the tumor vasculature: implications for clinical Structural insights into phosphoinositide 3-kinase catalysis and imaging. Cancer Res; e-pub ahead of print. signalling. Nature 402: 313–320. Serra V, Markman B, Scaltriti M, Eichhorn P, Valero V, Guzman M Welch S, Hirte HW, Carey MS, Hotte SJ, Tsao MS, Brown S et al. et al. (2008). NVP-BEZ235, a dual PI3K/mTOR inhibitor, prevents (2007). UCN-01 in combination with topotecan in patients with PI3K signaling and inhibits growth of cancer cells with activating advanced recurrent ovarian cancer: a study of the Princess Margaret PI3K mutations. Cancer Res; e-pub ahead of print. Hospital Phase II consortium. Gynecol Oncol 106: 305–310. Shapiro GI, Edelman G, Calvo E, Aggarwal SK, Laird AD. (2007). Weterings E, Chen DJ. (2007). DNA-dependent protein kinase in Targeting aberrant PI3K pathway signaling with XL147, a potent, nonhomologous end joining: a lock with multiple keys? J Cell Biol selective, and orally bioavailable PI3K inhibitor. 19th AACR-NCI- 179: 183–186. EORTC Meeting (abstract C205). Whitesell L, Lindquist SL. (2005). Hsp90 and the chaperoning of Shi Y, Liu X, Han E, Guan R, Shoemaker AR, Oleksijew A et al. cancer. Nat Rev 5: 761–772. (2005). Optimal classes of chemotherapeutic agents sensitized by Wymann MP, Bulgarelli-Leva G, Zvelebil MJ, Pirola L, Vanhaeseb- specific small-molecule inhibitors of akt in vitro and in vivo. roeck B, Waterfield MD et al. (1996). Wortmannin inactivates Neoplasia 7: 992–1000. phosphoinositide 3-kinase by covalent modification of Lys-802, a Solit DB, Basso AD, Olshen AB, Scher HI, Rosen N. (2003). residue involved in the phosphate transfer reaction. Mol Cell Biol Inhibition of Heat Shock Protein 90 function down-regulates Akt 16: 1722–1733. kinase and sensitizes tumors to Taxol. Cancer Res 63: 2139–2144. Yan Y, Backer JM. (2007). Regulation of class III (Vps34) PI3Ks. Stebbins CE, Russo AA, Scheneider C, Rosen N, Hartl FU, Pavletich Biochem Soc Trans 35: 239–241. NP. (1997). Crystal structure of an Hsp90-geldanamycin complex: Yang J, Cron P, Good VM, Thompson V, Hemmings BA, Barford D. target a protein chaperone by an antitumour agent. Cell 89: (2002a). Crystal structure of an activated Akt/Protein Kinase B 239–250. ternary complex with GSK3-peptide and AMP-PNP. Nat Struct Stephens L, Anderson K, Stokoe D, Erdjument-Bromage H, Painter Biol 9: 940–944. GF, Holmes AB et al. (1998). Protein kinase B kinases that mediate Yang J, Cron P, Thompson V, Good VM, Hess D, Hemmings BA phosphatidylinositol 3,4,5-trisphosphate-dependent activation of et al. (2002b). Molecular mechanism for the regulation of protein protein kinase B. Science 279: 710–714. kinase B/Akt by hydrophobic motif phosphorylation. Molecular Stokoe D, Stephens LR, Copeland T, Gaffney PRJ, Reese CB, Cell 9: 1227–1240. Painter GF et al. (1997). Dual role of phosphatidylinositol-3,4,5- Yu K, Lucas J, Zhu T, Zask A, Gaydos C, Toral-Barza L et al. (2005). trisphosphate in the activation of protein kinase B. Science 277: PWT-458, a novel pegylated-17-dydroxywortmannin, inhibits phos- 567–570. phatidylinositol 3-kinase signaling and suppresses growth of solid Stu¨ hmer T, Zo¨ llinger A, Siegmund D, Chatterjee M, Grella E, Knop S tumors. Cancer Biol Ther 4: 538–545. et al. (2008). Signalling profile and antitumour activity of the novel Yuan TL, Choi HS, Matsui A, Benes C, Lifshits E, Luo J et al. (2008). Hsp90 inhibitor NVP-AUY922 in multiple myeloma. Leukemia The role of class IA PI3K in vasculogenesis and tumor angiogenesis. (advance online publication). Proc Natl Acad Sci USA (advance online publication). Sydor J, Normant E, Pien CS, Porter JR, Ge J, Grenier L et al. (2006). Zahler S, Tietze S, Totzke F, Kubbutat M, Meijer L, Vollmar AM Development of 17-allylamino-17-demethoxygeldanamycin hydro- et al. (2007). Inverse in silico screening for identification of kinase quinone hydrochloride (IPI-504), an anti-cancer agent directed inhibitor targets. Chem Biol 14: 1207–1214. against Hsp90. Proc Natl Acad Sci USA 103: 17408–17413. Zhao Z, Leister WH, Robinson RG, Stanley FB, Defeo-Jones D, Tian ZQ, Liu Y, Zhang D, Wang Z, Dong SD, Carreras CW et al. Jones RE et al. (2005). Discovery of 2,3,5-trisubstitued pyridine (2004). Synthesis and biological activities of novel 17-aminogelda- derivatives as potent Akt and Akt2 dual inhibitors. Bioorg Med namycin derivatives. Bioorg Med Chem 12: 5317–5329. Chem Lett 15: 905–909. Tokunaga E, Oki E, Egashira A, Sadanaga N, Morito M, Kakeji Y Zhao Z, Robinson RG, Barnett SF, Defeo-Jones D, Jones RE, et al. (2008). Deregulation of the Akt pathway in human cancer. Hartman GD et al. (2008). Development of potent, allosteric dual Curr Cancer Drug Targets 8: 27–36. Akt1 and Akt2 inhibitors with improved physical properties and cell Toral-Barza L, Zhang W-G, Huang X, McDonald LA, Salaski EJ, activity. Bioorg Med Chem Lett 18: 49–53. Barbieri LR et al. (2007). Discovery of lactoquinomycin and related Zhu J, Huang J-W, Tseng P-H, Yang Y-T, Fowble J, Shiau C-W et al. pyranonaphthoquinones as potent and allosteric inhibitors of Akt/ (2004). From the cyclooxygenase-2 inhibitor celecoxib to a novel PKB: mechanistic involvement of Akt catalytic action loop class of 3-phosphoinositide-dependent protein kinase-1 inhibitors. cysteines. Mol Cancer Ther 6: 3028–3038. Cancer Research 64: 4309–4318.

Oncogene