Oncogene (2005) 24, 74937501

& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00
www.nature.com/onc

AKT crystal structure and AKT-specic inhibitors

Chandra C Kumar*,1 and Vincent Madison2
1 2

Department of Tumor Biology, Schering-Plough Research Institute, 2015, Galloping Hill Road, Kenilworth, NJ 07033, USA; Department of Structural Chemistry, Schering-Plough Research Institute, 2015, Galloping Hill Road, Kenilworth, NJ 07033, USA

AKT kinases are attractive targets for small molecule drug discovery because of their key role in tumor cell survival/proliferation and their overexpression/activation in many human cancers. This review summarizes studies that support the rationale for targeting AKT kinases in new drug discovery efforts. Structural features of AKT kinase in its inactive and active states, as determined by crystal structure analysis, are described. Recent efforts in the development and biological evaluation of small molecule inhibitors of AKT, and the challenges remaining are summarized. Inhibitors targeting the ATP binding site, PH domain and protein substrate binding site, as well as isoform selective allosteric inhibitors are reviewed. Structure-based design using PKA mutants as surrogates and computer modeling in the discovery of selective inhibitors is discussed. The issues and challenges facing the development of different classes of inhibitors as therapeutics are also discussed. Oncogene (2005) 24, 74937501. doi:10.1038/sj.onc.1209087 Keywords: AKT crystal structure; Akt inhibitors; ATP competitive; PH domain; peptide substrate; apoptosis; chemoresistance

considers inhibitors of other components of the AKT pathway. Rationale for targeting AKT for new drug discovery AKT kinases have been selected as targets for anticancer therapeutics at many academic and pharmaceutical institutions based on the rationale summarized below:  Reintroduction of wild-type PTEN into tumor cells that are mutant for PTEN resulted in inhibition of AKT activation leading to either cell cycle arrest (glioblastomas, renal cell carcinoma lines) or apoptosis (breast and prostate tumor cell lines) (Li et al., 1998; Di Cristofano and Pandol, 2000; Saito et al., 2003). Inducible expression of PTEN in Jurkat T cells inhibited activation of AKT leading to an increase in apoptosis and susceptibility to apoptosis by various stimuli (Xu et al., 2002). AKT expression protected cells from PTEN-mediated apoptosis and conferred resistance to proapoptotic cytotoxic agents (Li et al., 1998).  Expression of antisense AKT2 RNA in PANC-1 cells signicantly reduced their tumorigenicity in nude mice (Cheng et al., 1996).  Expression of a kinase-dead (KD) mutant of AKT using an adenoviral vector system induced apoptosis selectively in tumor cells expressing activated AKT but not in normal cells or tumor cells expressing low levels of activated AKT. In addition, the growth of tumors in a mouse model was also inhibited by intratumoral injection of AKT-KD expressing adenovirus (Jetzt et al., 2003). Similarly, adenoviralmediated expression of PTEN was shown to induce apoptosis in several tumor cells lacking PTEN expression but not in tumor cells expressing wild-type protein or in normal cells (Jetzt et al., 2003). These studies showed that tumor cells, unlike normal cells, are dependent on AKT kinases for survival and the specicity suggested that inhibition of AKT signaling may not be toxic to normal cells.  Expression of constitutively active AKT was shown to enhance IGF-1-mediated neuronal survival, whereas expression of a dominant-negative allele of AKT enhanced apoptosis and blocked the antiapoptotic effect of IGF-1 (Dudek et al., 1997; Kulik and Weber, 1998).  Introduction of cell-permeable anti-AKT1 single chain antibodies into tumor cells resulted in inhibition

Introduction It is now well established that hyperactivation of AKT kinases is a common event in many human cancers, and this activation results in tumor cell survival and enhanced resistance to apoptosis through multiple mechanisms (for a recent review see Bellacosa et al., 2005). Activation of AKT kinases occurs by an assortment of mechanisms, such as loss or downregulation of the PTEN tumor suppressor, which are described in detail by Altomare and Testa in this volume. Our review summarizes studies that support the rationale for targeting AKT kinases in new drug discovery efforts. We highlight structural features of AKT kinase in its inactive and active states as well as current efforts to design structure-based small molecule inhibitors of AKT. Another review in this volume, by Cheng et al.,

*Correspondence: Chandra C Kumar; E-mail: chandra.kumar@ spcorp.com

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7494

of AKT kinase activation and reduced phosphorylation of Gsk3a/b, leading to apoptosis in three cancer cell lines and inhibition of tumor growth in nude mice (Shin et al., 2005). A peptide termed AKT-in comprising a portion of an AKT-interacting protein TCL-1 blocked membrane translocation and activation of AKT leading to inhibition of cell proliferation and antiapoptosis in cells and in vivo tumor growth (Hiromura et al., 2004). Expression of a phosphatase termed PHLPP (PH domain/leucine-rich repeat protein phosphatase) that dephosphorylates AKT triggered apoptosis in a tumor cell line and inhibited tumor growth in mice (Gao et al., 2005). Treatment of lung and breast cancer cell lines with PI-3 kinase inhibitors such as Wortmannin and LY 294002 was shown to result in inhibition of AKT activation leading to an increase in apoptosis selectively in cells expressing high levels of activated AKT, but not in others expressing low levels of activated AKT (Brognard et al., 2001; Clark et al., 2002). Recent studies indicate that blocking AKT activation underlies the therapeutic efcacy seen with a number of approved drugs including Gleevec (Bcr-Abl inhibitor, Novartis), Trastuzumab (Her2/neu antibody, Genentech), Getinib (EGFR inhibitor, Astra Zeneca) and Tarceva (EGFR inhibitor, Genentech/OSI). Transformation mediated by amplication of Her2/ neu is the result of enhanced signaling through PI3K/ AKT signaling, and interruption of AKT signaling was essential for the cytotoxic effect of Trastuzumab. (Bacus et al., 2002; Yakes et al., 2002). Constitutively active AKT (myr-AKT) was shown to confer resistance to Trastuzumab in breast cancer cells (Clark et al., 2002). Sensitivity of breast tumor cells to Trastuzumab can be overcome by expression of constitutively active AKT. Recent studies have shown that Trastuzumab treatment causes rapid activation of PTEN tumor suppressor leading to down regulation of AKT activity and the antitumor effects are impaired in tumors lacking PTEN leading to hyperactive AKT kinases (Pandol, 2004). Thus, loss of PTEN leading to AKT activation is predicted to correlate with resistance to Trastuzumab. Getinib, which targets EGFR, induces dramatic clinical responses in non-small-cell lung cancers with activating mutations within the EGFR kinase domain. These mutant EGFRs selectively activate AKT and STAT signaling pathways, which promote cell survival but have no effect on extracellular signalregulated kinase (ERK2) signaling (Paez et al., 2004; Sordella et al., 2004). It was concluded that mutant EGFRs selectively activate AKT and STAT survival signaling pathways on which tumors become dependent and that inhibition of these signals by Getinib may contribute to the drugs efcacy. PTEN loss leading to constitutive AKT activation resulted in resistance to Getinib, which could be reversed by reintroduction of PTEN or pharmacological downregulation of the PI3K/AKT pathway (She et al., 2003).

 Similarly, Gleevec was shown to decrease phosphorylated AKT levels in PDGF-transformed ovarian cancer cells, gastrointestinal stromal tumor cells (Frolov et al., 2003; Ohashi et al., 2004) and Bcr-abl transformed chronic myelogenous leukemia cells (Kawauchi et al., 2003). These studies highlight the importance of the activation of AKT kinases in the establishment of malignant transformation and resistance to chemotherapy. When AKT is inhibited, malignant cells lose their resistance to apoptosis and expression of activated AKT can restore this resistance.

Sequence and structural analysis of AKT AKT kinases belong to the AGC kinase family (related to AMP/GMP kinases and protein kinase C) and consist of three conserved domains, an N-terminal pleckstrin homology (PH) domain, a central kinase catalytic (CAT) domain and a C-terminal extension (EXT) containing a regulatory hydrophobic motif (HM) (Table 1). Among the AKT isoforms, the PH domains are B80% identical and B30% identical to PH domains in pleckstrin and other proteins. The linker (LINK) region connecting the PH domain to the CAT domain is poorly conserved among the AKT isoforms (1746% identical) and has no signicant homology to any other human protein. The consensus CAT domain is B90% identical among the AKT isoforms and is closely related the PKC, PKA, SGK and S6 subfamilies of the AGC kinase family. The C-terminal extension (EXT) is B70% identical among the AKT isoforms and is most closely related to the PKC family. The N-terminal 3040 residues of EXT are homologous in the AKT, S6, SGK, PKA and c-GMP kinase families.

Table 1 Pairwise % identity in AKT domainsa

T308 1 107 149 408 S473 480

PH domain

Catalytic domain

C-terminal extension

PAIR AKT1/AKT2 AKT1/AKT3 AKT2/AKT3

PH 80 84 76

LINK 46 40 17

CAT 90 88 87

EXT 66 76 70

a Domain denitions using AKT1 residue numbers. PH Pleckstrin Homology domain, AKT1 (1107), B30% identical to pleckstrin and other PH domains. LINK Linker region, AKT 1(108148), no signicant homology to other proteins. CAT Kinase Catalytic domain, AKT1 (149408), homologous to all kinases, B50% identical to the PKC, PKA, SGK and S6 families. EXT C-terminal Extension, AKT1 (409480) is only B15% identical to the PKA family and B3540% identical to the rest of the AGC family members

Oncogene

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7495

Figure 1 Composite of crystal structures of the AKT1 PH (1UNQ) and AKT2 kinase (1O6K) domains. The PH domain was positioned arbitrarily relative to the kinase domain and a connector was modeled. Also a segment was added to connect two fragments of EXT. Segments are color coded: PH (yellow), InsP4 (orange, red), LINK (magenta), CAT (cyan), EXT (red), AMPPNP (atom colors), peptide substrate (green). Figure constructed by Jose Duca using PyMOL (DeLano, WL The PyMOL Molecular Graphics System (2002); DeLano Scientic, San Carlos, CA, USA. http://www.pymol.org.)

Figure 1 shows the three-dimensional arrangement of the conserved sequence domains as a composite of PH (1UNQ) and kinase domain (1O6K) structures. The PH domain was positioned arbitrarily relative to the kinase domain and connected by a modeled peptide. There are no structures that encompass all of LINK; The structure of the 25 residue gap between PH and CAT remains unknown. Due to low sequence homology with known structures, models of LINK would be highly speculative. Several three-dimensional structures have been determined for the separate AKT1 (PDB codes 1UNP, 1UNQ, 1UNR, 1H10) and AKT2 (1P6S) PH domains (Thomas et al., 2001, 2002; Milburn et al., 2003; Auguin et al., 2004). These structures are very similar and also

match the consensus core of the sequence related PH domains from TAPP1 (1EAZ) and pleckstrin (1XX0). The C-terminal helix of the AKT PH domains extends about three turns beyond the consensus core (AKT1 (1 107) to Arg121 (1UNP), see Figure 1). Inositol-(1,3,4,5)tetrakisphosphate binds to a shallow pocket in the AKT1 PH domain largely mediated through several salt bridges between the phosphate groups and basic residues in the protein (1UNQ, 1H10). This binding site is unattractive for drug development due to the fact that the pocket is shallow and charge/charge interactions dominate the binding. AKT kinases are activated by phosphorylation at a Thr residue in the activation T loop (T308 in AKT1, T309 in AKT2 and T305 in AKT3) and a Ser residue in the C-terminal HM (S473 in AKT1, S474 in AKT2 and S472 in AKT3). Phosphorylation of AKT1 at Thr308 stimulates enzymatic activity by at least 100fold, whereas phosphorylation of Ser473 increases AKT activity by another 10-fold (Alessi et al., 1996). Together these two phosphorylations augment AKT kinase activity by 1000-fold. Elucidation of three-dimensional structures of inactive and active AKT2 kinase domains helps to explain how these phosphorylations lead to conformational changes to increase enzymatic activity. The crystal structures of AKT2 kinase domain in its inactive (unphosphorylated) form were determined independently by Barfords group at ICRF in London and by Amgen Cambridge Research Center in Cambridge, USA (Yang et al., 2002b; Huang et al., 2003). The structure of activated AKT2 (phosphorylated at Thr309) in complex with glycogen synthase kinase-3 (Gsk-3) peptide and 50 -adenylylimidodiphoshate (AMPPNP) was also described (Yang et al., 2002a). The protein constructs include the entire kinase domain (AKT2 (143481) PDB codes 1MRV, 1MRY or AKT2 (146481) 1GZN, 1O6K, 1O6L) or a shortened version (AKT2 (146460) 1GZK, 1GZO). In either case, the structurally dened region is bounded by residue 146 at the N-terminus and ca. residue 440 at the C-terminus. Structures of apo-AKT2 (1GZN) and the ternary AKT2/AMP-PNP/peptide substrate complex (1O6K) are compared in Figure 2. In both cases, the CAT domain assumes the consensus kinase core conformation. For the unliganded form extending to either residue 460 or 481 and regardless of phosphorylation state of the activating residues, there are several signicant structural differences. These include different conformations of the Gly-rich loop, the activation loop and EXT as well as disordered segments in the B and C-helices, the activation loop and EXT. AKT2 is less ordered than most kinases after phosphorylation, implying a larger entropic penalty for substrate and inhibitor binding. In apo-AKT2 the DFG motif at the beginning of the activation loop is ipped out towards the front, opening an internal cavity that comprises part of the Gleevec binding site in ABL. This conformational change results in disruption of the ATP binding site. Only in the ternary complex can the exible elements assume a well ordered conformation closely mimicking that of other activated kinases, especially the closely
Oncogene

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7496

Figure 2 Backbone traces from crystal structures of (a) AKT2 (1GZN) and (b) AKT2/AMP-PNP/Peptide Substrate ternary complex (1O6K). For each structure, the main body is colored cyan and segments that differ are highlighted in different colors. G-Loop (Glyrich loop) magenta, B- and C-helices green, T-loop (activation loop) yellow, EXT red, substrate magenta, ATP (AMP-PNP) atom colors. The locations of the DFG and motifs and the PIF-Pocket are also shown

related PKA and PKC (Knighton et al., 1991; Xu et al., 2004). Phosphorylation of Ser474 and subsequent binding of the C-terminal HM is an important part of forming the ordered, active conformation. In each of the AGC family members, EXT wraps around the front of the ATP binding pocket making the cavity smaller than in other kinases, and then continues over the top of the Gly-rich loop terminating in an HM. HM binds to a pocket at the back of the kinase domain (the PIF pocket) forming HM/C-helix contacts. The EXT sequence domain is not an independently folded 3D domain, but a C-terminal extension of CAT. Functionally, EXT is very similar in AKT2 and PKA even though the sequences are only 17% identical over its full length and the conformations of the HM are different. The AKT2 structures have been discussed thoroughly in several original papers and reviews (Pearl and Barford, 2002; Scheid and Woodgett, 2003). Figure 3 shows the bound substrates on the AKT2 surface in the ternary complex (1O6K). The peptide substrate binds at a typical rather at, extended protein/ protein interface while AMP-PNP extends deeply into a taco-shaped cavity. Small molecule inhibitors of AKT The common domain organization and high sequence homology of the AKT isoforms was discussed above. Earlier, homology modeling predicted that the three
Oncogene

AKT isoforms would share the three-dimensional structure of other AGC family members such as PKA and that the 25-residue ATP binding site is 96100% identical in the three isoforms. The only residue that differs is Ala230 in AKT1 and Ala232 in AKT2 vs Val228 in AKT3 (Kumar et al., 2001). The homology models have been conrmed by subsequent crystal structures of AKT2. Extending the binding site analysis to the larger 32-residue balanol (a non-specic ATP competitor) site still reveals only the single Ala (AKT1, AKT2)-Val (AKT3) variant. The most homologus other AGC kinases are 7077% identical in this site (Table 2). While differentiation of AKT isoforms through binding to the balanol site would be very difcult, specicity between AKT and PKA has been demonstrated (Breitenlechner et al., 2005). In addition to the compounds that target the ATP binding site, inhibitors that target the PH, LINK and the protein substrate sites have been reported. In principle, the PIF pocket and the Gleevec-like pocket of the inactive form could also be targeted. Compounds that target the ATP binding pocket Several small molecule leads that compete with ATP for binding to its pocket are being used as the basis for developing potent and selective inhibitors of AKT. A number of these leads were discovered as inhibitors of PKA or PKC. One PKA inhibitor that has been subjected to optimization for AKT selectivity is H-89,

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7497

Figure 3 Crystal structure of AKT2/AMP-PNP/Peptide Substrate ternary complex (1O6K). The substrates are shown as ball-andstick models on the protein surface colored by atom type with carbons shown as green for the peptide substrate and cyan for AMP-PNP

Table 2

Inhibitors in Figures 46 target the ATP binding site

Crystal structures have been determined for balanol and its analogs (Figure 5) complexed with PKA and PKA variants with AKT residues substituted into the binding site (Breitenlechner et al., 2005). The balanol binding site comprises 32 residues within 5A contact around the inhibitor. Table 2 gives the 18 residues for which at least one of the other sequences differs from that of AKT1; shading highlights these differences.

which inhibits PKA and AKT with IC50s of 35 and 2500 nM, respectively. The crystal structure of the PKA/H-89 complex shows that the isoquinoline nitrogen mimics N1 of adenosine in accepting a H-bond from the peptide backbone, while the sulfonamide chain extends in the direction of the triphosphate chain of ATP (1YDT). Libraries around the lipophilic amino terminus led to the identication of NL-71101, which inhibits AKT1 2.4-fold better than PKA (IC50s 3.7 and 9 mM respectively) (Reuveni et al., 2002). This compound induced apoptosis in OVCAR-3 ovarian carcinoma cells at high concentrations (>25 mM). A more potent and moderately selective analog of H-89, which inhibits AKT and PKC with IC50s of 0.17 and 1.4 mM (Figure 4, compound 3), respectively, has also been described (Garrett, 2004). Synthetic efforts to develop novel analogs of balanol, a nonselective kinase inhibitor, were undertaken recently. Based on a crystal structure of a balanol analog binding to PKA, several analogs were synthe-

O2S

H N

N H Br

O2 S

H N

N H

H-89 AKT IC50 2.5 M PKA IC50 0.035 M

NL-71-101 AKT IC50 3.7 M AKT IC50 9 M

Figure 4 PKA inhibitors as leads for AKT inhibitors

sized to improve plasma stability (Breitenlechner et al., 2004). Compound 2 was developed that is highly potent against AKT1 and PKA (IC50s 4 and 2 nM respectively) and has improved plasma stability (Figure 5, compounds 1, 2). This study was extended to derive AKTselective inhibitors using cocrystallization with PKA and mutants of PKA in which the amino acids that were
Oncogene

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7498
N HN OMe F O O O NH OMe F N HN HN O
N

O NH
O

O N

N N H

O
NH

H N N

N N

O NH

OH

OH

Compound 1 AKT1 IC50 5 nM PKA IC50 5 nM T1/2 in plasma <1 min.

N HN N HN O O NH

Compound 2 AKT1 IC50 4 nM PKA IC50 2 nM T1/2 in plasma 69 h.

N HN N HN O O NH

Compound 7 AKT1 selective AKT1 IC50 0.76 M AKT2 IC50 24 M AKT3 IC50 >50 M
N N N H N N N O NH

Compound 8 AKT2 selective AKT1 IC50 21 M AKT2 IC50 0.32 M AKT3 IC50 21 M

OH

OH

Compound 9 Dual AKT1/AKT2 selective AKT1 IC50 0.15 M AKT2 IC50 0.33 M AKT3 IC50 > 50 M

Compound 3 AKT1 IC50 20 nM PKA IC50 1900 nM

Compound 4 AKT1 IC50 40 nM PKA IC50 1700 nM

Figure 7 Allosteric AKT1-, AKT2- and dual AKT1/AKT2selective inhibitors from Merck

Figure 5 Balanol analogs designed to improve selectivity against PKA

N NH

N NH

N NH2 O

N NH2

N H Compound 5 A-443654 AKT1 Ki = 160 pM Compound 6 A-674563 AKT1 Ki = 11 nM Orally Available

indazole-pyridine based compounds that inhibit AKT1 kinase (Figure 6, compounds 5, 6). These compounds were active in cell-based assays and blocked Gsk-3 phosphorylation induced by the overexpression of all three isoforms of AKT (Luo et al., 2005). These compounds slowed the growth of tumors when used as monotherapy or in combination with paclitaxel or rapamycin. The therapeutic window for these compounds was very narrow and signicant toxicities were observed with these compounds, which included malaise, weight loss and increased plasma insulin levels. It is not clear if the toxicities seen with these compounds are solely due to inhibition of AKT kinases or due to inhibition of other kinases. This class of compounds represents the most potent AKT inhibitors described so far and most advanced preclinically. Allosteric inhibitors Recently the group from Merck research laboratories described novel allosteric inhibitors that are isoform selective and highly specic against closely related members of kinases (Barnett et al., 2005; Lindsley et al., 2005; Zhao et al., 2005). These compounds were not competitive with ATP or peptide substrate and are only active against the full-length enzyme requiring the PH domain for full enzyme inhibition. Structures of AKT1-selective, AKT2-selective and dual AKT1/AKT2 inhibitors are shown in Figure 7 (compounds 7, 8 and 9). Published speculation that these inhibitors bind to LINK is consistent with the sequence diversity of this region, note that the largest isoform difference is AKT2 vs AKT3, with only 17% identity. These compounds were only modestly active in inducing caspase-3 activation in LnCaP tumor cells, which express AKT1 and AKT2 but not AKT3. However, when treated with a 1 : 1 mixture of AKT1-selective and AKT2-selective

Figure 6 ATP-competitive AKT kinase inhibitors from Abbott

different from AKT were altered. These efforts have led to the identication of potent AKT inhibitors (IC50s 20 and 40 nM, respectively) with signicant improvements in selectivity against PKA (IC50s 1900 and 1700 nM) (Figure 5 compounds 3, 4) (Breitenlechner et al., 2005). Selectivity against PKA was improved by introducing bulkier substituents, exploiting an enlarged subsite in AKT vs PKA that arises largely from the smaller Leu296 in AKT2 compared to Phe187 in PKA (Table 2). This is a good example of structure-based inhibitor optimization. Recently, the group from Abbott laboratories described a series of potent and moderately selective
Oncogene

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7499

compounds or the dual AKT1/AKT2 inhibitor, a dramatic increase in caspase-3 activation was observed in combination with chemotherapeutic agents (DeFeo-Jones et al., 2005). These studies indicate that inhibition of AKT1 and AKT2 is required to reduce the apoptotic threshold and these inhibitors are only modestly active in inducing apoptosis in tumor cells, but synergize with chemotherapeutic agents to induce apoptosis. Importantly it was shown that inhibition of AKT1 and AKT2 sensitized only tumor cells but not normal cells to apoptotic stimuli, suggesting a potential therapeutic window. However, these compounds have poor solubility and pharmacokinetic properties that precluded their evaluation in animal tumor models. Targeting the PH domain Another approach to inhibiting AKT kinases is to target the PH domain of AKT and interfere with PtdIns(3,4,5)P3 (PIP3) binding and membrane translocation. Based on molecular modeling of the PH domain of AKT, a number of phosphatidylinositol ether lipid analogs (PIAs) were synthesized to inhibit this interaction (Castillo et al., 2004; Gills and Dennis, 2004). Five PIAs were identied (Figure 8, PIAs 5, 6, 23, 24 and 25), which inhibited AKT with IC50s less than 5 mM. These compounds inhibited AKT activation and phosphorylation of several downstream substrates of AKT in tumor cells without affecting the activities of upstream kinases such as PI3-K and PDK1 or members of other signaling pathways such as ERK2. Signicantly these PI analogs increased apoptosis 20- to 30-fold in tumor cell lines expressing high levels of endogenous activated AKT and were only modestly active in tumor cells expressing low levels of activated AKT. Major issues with this class of molecules are limited solubility, moderate potency against AKT kinases, aggregation and poor pharmacokinetics, which limit their usefulness as small molecule drug leads. Moreover, it is not clear how specic these compounds are towards blocking AKT translocation to the membrane, since PH domains are present in several proteins. As noted above, the crystal structure has been solved for the complex of AKT1 PH domain with inositol (1, 3, 4, 5) tetraphosphate (1H10, 1UNQ). As

already noted, this binding pocket is not an ideal drug target because it is shallow and highly charged. Pseudosubstrate inhibitors A recent study identied a 14-mer peptide of the sequence, ARKRERTYSFGHHA (AKTide-2T), which binds to the substrate binding region of AKT1 and inhibits the enzyme with a Ki of 12 mM. A hybrid between this sequence and a segment of the FOXO3 forkhead transcription factor (sequence VELDPEFEPRARERTYSFGH, 20-mer) inhibited AKT1 with a Ki of 1.1 mM. Substitution of a phosphorylatable serine residue with an alanine resulted in a further 10-fold improvement in potency (Ki of 0.11 mM) and these peptides were highly selective versus several closely related kinases (Luo et al., 2004). These peptides bind to AKT only in the presence of Mg-ATP. Fusion peptides constructed to allow cellular uptake and demonstrated dose-dependent inhibition of Gsk-3 phosphorylation and viability of tumor cells. An interesting series of bi-substrate inhibitors consists of a 7-mer peptide sequence linked to the isoquinoline sulfonyl moiety found in H-89. These compounds, while not as selective as the peptidic pseudosubstrate inhibitors, are potent against AKT1 and PKA (IC50s 20 and 12 nM respectively). The protein substrate binding site is highly extended and has only shallow indentations. Similar to other protein/protein interaction sites, it offers poor prospects for developing potent small molecule pseudosubstrate or bisubstrate inhibitors. Summary and outlook Substantial preclinical data and studies of tumor specimens have established AKT kinases as major contributors to the development of many human cancers (see reviews by Altomare and Testa in this volume). AKT activation is also linked to the resistance of many cancers to cytotoxic agents and targeted therapeutics that have been approved recently. Although inhibition of AKT activation or signaling has been demonstrated to induce apoptosis in several experimental models, the best use of AKT inhibitors may be in combination with other chemotherapeutics. Small molecule inhibitors targeting the AKT signaling pathway hold great promise as novel therapeutics against cancer. However, signicant challenges remain for the development of potent and selective inhibitors of AKT kinases that are suitable as drug candidates. Despite intense efforts at many pharmaceutical companies, there are no reports of clinical trials with an AKT inhibitor to date. Recent success in the elucidation of crystal structures of the AKT2 kinase domain, together with the use of PKA mutant proteins as surrogates, opens the possibility of structure-based design of potent, selective inhibitors.
Acknowledgements We thank Drs Tin-Yau Chan and Jose Duca for help in the preparation of gures and our colleagues in the AKT project team for stimulating discussions.
Oncogene

OMe HO HO HO O O P O HO OMe OC18H37

HO HO HO

O O P O HO

OMe OC18H37

PIA5

PIA6

O HO HO HO O O P O HO
OH O O P O HO OMe OC 18H37

OMe OC18H37

HO HO HO

O O P O HO

OMe OC18H37

PIA23
HO HO

PIA24

PIA25

Figure 8 PIP analogs that target the PH domain of AKT

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7500 References Alessi DR, Caudwell FB, Andjelkovic M, Hemmings BA and Cohen P. (1996). FEBS Lett., 399, 333338. Altomare DA and Testa JR. (2005). Oncogene Rev., 24, 74557464. Auguin D, Barthe P, Auge-Senegas MT, Stern MH, Noguchi M and Roumestand C. (2004). J. Biomol. NMR, 28, 137155. Bacus SS, Altomare DA, Lyass L, Chin DM, Farrell MP, Gurova K, Gudkov A and Testa JR. (2002). Oncogene, 21, 35323540. Barnett SF, Bilodeau MT, Lindsley CW, DeFeo-Jones D, Fu S, Hancock PJ, Haskell KM, Leander KR, McAvoy E, Robinson RG, Duggan ME, Zhao Z, Huber HE, Jones RE, Leister WH, Hartman GD, Huff JR, Kahana JA, Kral AM, Leander K, Lee LL, Malinowski J, McAvoy EM and Nahas DD. (2005). Curr. Top. Med. Chem., 5, 109125. Bellacosa A, Kumar CC, Di Cristafano A and Testa JR. (2005). Advances in Cancer Research. Vande Woude G and Klein G (eds). Vol. 94.Academic press: New York, pp. 2987. Breitenlechner CB, Friebe WG, Brunet E, Werner G, Graul K, Thomas U, Kunkele KP, Schafer W, Gassel M, Bossemeyer D, Huber R, Engh RA and Masjost B. (2005). J. Med. Chem., 48, 163170. Breitenlechner CB, Wegge T, Berillon L, Graul K, Marzenell K, Friebe WG, Thomas U, Schumacher R, Huber R, Engh RA and Masjost B. (2004). J. Med. Chem., 47, 13751390. Brognard J, Clark AS, Ni Y and Dennis PA. (2001). Cancer Res., 61, 39863997. Castillo SS, Brognard J, Petukhov PA, Zhang C, Tsurutani J, Granville CA, Li M, Jung M, West KA, Gills JG, Kozikowski AP and Dennis PA. (2004). Cancer Res., 64, 27822792. Cheng JQ, Ruggeri B, Klein WM, Sonoda G, Altomare DA, Watson DK and Testa JR. (1996). Proc. Natl. Acad. Sci. USA, 93, 36363641. Cheng JQ, Lindsley CW, Cheng GZ, Yang H and Nicosia SV. (2005). Oncogene Rev., 24, 74287492. Clark AS, West K, Streicher S and Dennis PA. (2002). Mol. Cancer Ther., 1, 707717. DeFeo-Jones D, Barnett SF, Fu S, Hancock PJ, Haskell KM, Leander KR, McAvoy E, Robinson RG, Duggan ME, Lindsley CW, Zhao Z, Huber HE and Jones RE. (2005). Mol. Cancer Ther., 4, 271279. Di Cristofano A and Pandol PP. (2000). Cell, 100, 387390. Dudek H, Datta SR, Franke TF, Birnbaum MJ, Yao R, Cooper GM, Segal RA, Kaplan DR and Greenberg ME. (1997). Science, 275, 661665. Frolov A, Chahwan S, Ochs M, Arnoletti JP, Pan ZZ, Favorova O, Fletcher J, von Mehren M, Eisenberg B and Godwin AK. (2003). Mol. Cancer Ther., 2, 699709. Gao T, Furnari F and Newton AC. (2005). Mol. Cell, 18, 1324. Garrett M. (2004). Eur. J. Cancer (Suppl.16th EORTC-NCIAACR Symp. Mol. Targets Cancer Ther. Sept 28Oct 1, Geneva 2004), 2, Abstr 322. Gills JJ and Dennis PA. (2004). Expert Opin. Invest. Drugs, 13, 787797. Hiromura M, Okada F, Obata T, Auguin D, Shibata T, Roumestand C and Noguchi M. (2004). J. Biol. Chem., 279, 5340753418. Huang X, Begley M, Morgenstern KA, Gu Y, Rose P, Zhao H and Zhu X. (2003). Structure (Camb), 11, 2130.
Oncogene

Jetzt A, Howe JA, Horn MT, Maxwell E, Yin Z, Johnson D and Kumar CC. (2003). Cancer Res., 63, 66976706. Kawauchi K, Ogasawara T, Yasuyama M and Ohkawa S. (2003). Blood Cells Mol. Dis., 31, 1117. Knighton DR, Zheng JH, Ten Eyck LF, Ashford VA, Xuong NH, Taylor SS and Sowadski JM. (1991). Science, 253, 407414. Kulik G and Weber MJ. (1998). Mol. Cell Biol., 18, 67116718. Kumar CC, Diao R, Yin Z, Liu Y, Samatar AA, Madison V and Xiao L. (2001). Biochim. Biophys. Acta, 1526, 257268. Li J, Simpson L, Takahashi M, Miliaresis C, Myers MP, Tonks N and Parsons R. (1998). Cancer Res., 58, 5667 5672. Lindsley CW, Zhao Z, Leister WH, Robinson RG, Barnett SF, Defeo-Jones D, Jones RE, Hartman GD, Huff JR, Huber HE, Duggan ME, Fu S, Hancock PJ, Haskell KM, Kahana JA, Kral AM, Leander K, Lee LL, Malinowski J, McAvoy EM and Nahas DD. (2005). Bioorg. Med. Chem. Lett., 15, 761764. Luo Y, Shoemaker AR, Liu X, Woods KW, Thomas SA, de Jong R, Han EK, Li T, Stoll VS, Powlas JA, Oleksijew A, Mitten MJ, Shi Y, Guan R, McGonigal TP, Klinghofer V, Johnson EF, Leverson JD, Bouska JJ, Mamo M, Smith RA, Gramling-Evans EE, Zinker BA, Mika AK, Nguyen PT, Oltersdorf T, Rosenberg SH, Li Q, Giranda VL, Shen J, Hutchins C, Richardson P and Holzman T. (2005). Mol. Cancer Ther., 4, 977986. Luo Y, Smith RA, Guan R, Liu X, Klinghofer V, Shen J, Hutchins C, Richardson P, Holzman T, Rosenberg SH and Giranda VL. (2004). Biochemistry, 43, 12541263. Milburn CC, Deak M, Kelly SM, Price NC, Alessi DR and Van Aalten DM. (2003). Biochem. J., 375, 531538. Ohashi A, Kinoshita K, Isozaki K, Nishida T, Shinomura Y, Kitamura Y and Hirota S. (2004). Int. J. Cancer, 111, 317321. Paez JG, Janne PA, Lee JC, Tracy S, Greulich H, Gabriel S, Herman P, Kaye FJ, Lindeman N, Boggon TJ, Naoki K, Sasaki H, Fujii Y, Eck MJ, Sellers WR, Johnson BE and Meyerson M. (2004). Science, 304, 14971500. Pandol PP. (2004). N. Engl. J. Med., 351, 23372338. Pearl LH and Barford D. (2002). Curr. Opin. Struct. Biol., 12, 761767. Reuveni H, Livnah N, Geiger T, Klein S, Ohne O, Cohen I, Benhar M, Gellerman G and Levitzki A. (2002). Biochemistry, 41, 1030410314. Saito Y, Swanson X, Mhashilkar AM, Oida Y, Schrock R, Branch CD, Chada S, Zumstein L and Ramesh R. (2003). Gene Therapy, 10, 19611969. Scheid M and Woodgett J. (2003). FEBS Lett., 27349, 15. She QB, Solit D, Basso A and Moasser MM. (2003). Clin. Cancer Res., 9, 43404346. Shin I, Edl J, Biswas S, Lin PC, Mernaugh R and Arteaga CL. (2005). Cancer Res., 65, 28152824. Sordella R, Bell DW, Haber DA and Settleman J. (2004). Science, 305, 11631167. Thomas CC, Deak M, Alessi DR and van Aalten DM. (2002). Curr. Biol., 12, 12561262. Thomas CC, Dowler S, Deak M, Alessi DR and van Aalten DM. (2001). Biochem. J., 358, 287294. Xu Z, Stokoe D, Kane LP and Weiss A. (2002). Cell Growth Differ, 13, 285296. Xu ZB, Chaudhary D, Olland S, Wolfrom S, Czerwinski R, Malakian K, Lin L, Stahl ML, Joseph-McCarthy D,

AKT crystal structure and AKT-specic inhibitors CC Kumar and V Madison

7501 Benander C, Fitz L, Greco R, Somers WS and Mosyak L. (2004). J. Biol. Chem., 279, 5040150409. Yakes FM, Chinratanalab W, Ritter CA, King W, Seelig S and Arteaga CL. (2002). Cancer Res., 62, 41324141. Yang J, Cron P, Good VM, Thompson V, Hemmings BA and Barford D. (2002a). Nat. Struct. Biol., 9, 940944. Yang J, Cron P, Thompson V, Good VM, Hess D, Hemmings BA and Barford D. (2002b). Mol. Cell, 9, 12271240. Zhao Z, Leister WH, Robinson RG, Barnett SF, Defeo-Jones D, Jones RE, Hartman GD, Huff JR, Huber HE, Duggan ME and Lindsley CW. (2005). Bioorg. Med. Chem. Lett., 15, 905909.

Oncogene