A novel inhibitor of the PI3K/Akt pathway based on the structure of inositol 1,3,4,5,6-pentakisphosphate

Background: Owing to its role in cancer, the phosphoinositide 3-kinase (PI3K)/Akt pathway is an attractive target for therapeutic intervention. We previously reported that the inhibition of Akt by inositol 1,3,4,5,6-pentakisphosphate (InsP5) results in anti-tumour properties. To further develop this compound we modified its structure to obtain more potent inhibitors of the PI3K/Akt pathway. Methods: Cell proliferation/survival was determined by cell counting, sulphorhodamine or acridine orange/ethidium bromide assay; Akt activation was determined by western blot analysis. In vivo effect of compounds was tested on PC3 xenografts, whereas in vitro activity on kinases was determined by SelectScreen Kinase Profiling Service. Results: The derivative 2-O-benzyl-myo-inositol 1,3,4,5,6-pentakisphosphate (2-O-Bn-InsP5) is active towards cancer types resistant to InsP5 in vitro and in vivo. 2-O-Bn-InsP5 possesses higher pro-apoptotic activity than InsP5 in sensitive cells and enhances the effect of anti-cancer compounds. 2-O-Bn-InsP5 specifically inhibits 3-phosphoinositide-dependent protein kinase 1 (PDK1) in vitro (IC50 in the low nanomolar range) and the PDK1-dependent phosphorylation of Akt in cell lines and excised tumours. It is interesting to note that 2-O-Bn-InsP5 also inhibits the mammalian target of rapamycin (mTOR) in vitro. Conclusions: InsP5 and 2-O-Bn-InsP5 may represent lead compounds to develop novel inhibitors of the PI3K/Akt pathway (including potential dual PDK1/mTOR inhibitors) and novel potential anti-cancer drugs.

Phosphoinositide 3-kinase (PI3K) isoforms catalyse the phosphorylation of the 3-hydroxyl group within the inositol ring of phosphoinositides generating lipid products, which in turn mediate the activation of several proteins (Maffucci and Falasca, 2001;Vanhaesebroeck et al, 2001). The best characterised PI3K effector is the Serine/Threonine kinase protein kinase B (PKB)/ Akt, which regulates a plethora of intracellular processes, including cell survival, growth, proliferation, migration and regulation of cell size (Vivanco and Sawyers, 2002;Manning and Cantley, 2007). Upon PI3K activation, interaction between Akt pleckstrin homology (PH) domain and the PI3K product phosphatidylinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P 3 ) recruits Akt to the plasma membrane, where it is activated through phosphorylation at its residues Thr308 and Ser473. Phosphorylation of Thr308 is mediated by 3-phosphoinositide-dependent protein kinase 1 (PDK1), which itself possesses a PH domain able to bind PtdIns(3,4,5)P 3 (Komander et al, 2004). Mutations in PDK1 PH domain that abolish PtdIns(3,4,5)P 3 binding strongly inhibit Akt activation in homozygous knock-in embryonic stem cell and knock-in mice (McManus et al, 2004;Bayascas et al, 2008). On the other hand, binding of Akt PH domain to PtdIns(3,4,5)P 3 is critical to induce a conformational change that allows PDK1-dependent phosphorylation (Calleja et al, 2007). Among other targets, Akt activates the multi-protein complex mTORC1 containing the enzyme mammalian target of rapamycin (mTOR), which regulates several intracellular functions including cell growth, cell cycle progression and autophagy (Wullschleger et al, 2006). It is interesting to note that a second mTOR-containing complex (mTORC2) is involved in the phosphorylation of Akt at its residue Ser473 as well as its activation (Sarbassov et al, 2005). The mechanism of mTORC2-dependent Akt phosphorylation at Ser473 is still not completely understood, but it does not seem to involve phosphoinositides, as in the case of PDK1, since mTOR does not appear to possess phosphoinositide-binding domains.
Deregulation of PI3K-dependent signalling pathways is linked to the development of cancer (Maehama and Dixon, 1999;Shayesteh et al, 1999;Vivanco and Sawyers, 2002;Bader et al, 2005;Shaw and Cantley, 2006;Vogt et al, 2007) and to increased resistance to treatment with chemotherapeutic agents (Clark et al, 2002;Liang et al, 2003;She et al, 2003). Accumulation of PtdIns(3,4,5)P 3 either due to gain of function of PI3K activity (Vivanco and Sawyers, 2002;Vogt et al, 2007) or loss of the enzyme phosphatase and tensin homolog deleted on chromosome 10 (PTEN), which specifically dephosphorylates PtdIns(3,4,5)P 3 (Maehama and Dixon, 1999), has been detected in almost 50% of all tumour types . Reducing the levels of PDK1 in PTEN þ /À mice strongly protects them from developing a wide range of tumours (Bayascas et al, 2005). Furthermore, it has recently been reported that PDK1 is overexpressed in several human breast cancers and that increased copy number of the gene encoding for PDK1 is associated with upstream pathway lesions and patient survival (Maurer et al, 2009), highlighting the importance of PDK1 in cancer development. Elevated Akt activity has been found in several forms of cancer (Sun et al, 2001;Bacus et al, 2002;Altomare et al, 2004) and evidence suggest that mTORC1 is one of the key effectors in PI3K/Akt-mediated tumourigenesis (Guertin and Sabatini, 2007). The crucial role of mTORC2 in tumorigenesis driven by Pten loss has also been reported (Guertin et al, 2009). The PI3K/Akt pathway is, therefore, at present being considered to be an attractive target for therapeutic intervention, and several compounds targeting the different components of the pathway have been developed or are in development (Vivanco and Sawyers, 2002;Luo et al, 2003;Hennessy et al, 2005;Guertin and Sabatini, 2007;Liu et al, 2009), with some of them currently in clinical trials for cancer treatment (Liu et al, 2009). Toxicity, low therapeutic index, insolubility and aqueous instability have prevented the use of generic PI3K inhibitors wortmannin and LY294002 as anti-cancer agents despite their anti-tumour activity (Hu et al, 2000;Ng et al, 2001). More specific approaches are needed to selectively block only the deregulated rather than all the PI3Ks-dependent pathways. Small molecule inhibitors of PDK1 have recently been developed, most of which target the ATP-binding site of PDK1, but they often possess poor physicochemical properties and inadequate selectivity profiles (Peifer and Alessi, 2008). Similarly several Akt inhibitors have been designed including compounds targeting its ATP-binding domain or allosteric inhibitors and pseudosubstrates (Luo et al, 2005;Crowell et al, 2007;Lindsley et al, 2008). However, some of these agents show toxic side effects either because of non-specific effects or blockade of all Akt isoforms, thus resulting in the alteration of normal glucose homeostasis. The in vitro and in vivo effects on Akt of chemopreventive compounds, such as the rotenoid deguelin have also been reported (Lee et al, 2005). Finally, several mTOR inhibitors are at present available, whose effects have been investigated in many solid tumours (Guertin and Sabatini, 2007;Fasolo and Sessa, 2008). Despite several efforts, there is still a need to develop novel, more potent inhibitors of the PI3K/Akt pathway to overcome problems of lack of specificity and chemoresistance.
A few years ago we were the first to propose an alternative mechanism to block Akt activation based on the inhibition of its PH domain-mediated translocation to the plasma membrane . The critical role of the PH domain in Akt-driven tumourigenesis has recently been highlighted by the detection of a somatic mutation in Akt1 PH domain resulting in Akt1 activation in breast, colorectal and ovarian cancers (Carpten et al, 2007). It is interesting to note that this mutant is able to induce leukaemia in mice (Carpten et al, 2007). Our strategy was based on the hypothesis that specific exogenous inositol polyphosphates can compete with PtdIns(3,4,5)P 3 by binding to Akt PH domain, and thus prevent recruitment to the plasma membrane and activation of Akt . Indeed we reported that inositol 1,3,4,5,6pentakisphosphate (InsP 5 ) specifically blocks Akt activation and possesses pro-apoptotic (Razzini et al, 2000;Piccolo et al, 2004), anti-angiogenic and anti-tumour activity in vivo (Maffucci et al, 2005). In addition, some of us recently demonstrated the targeting of the Akt PH domain with an unusual inositol polyphosphate mimic (Mills et al, 2007). Other phosphatidylinositol-based Akt inhibitors also act by inhibiting Akt targeting to the plasma membrane, including ether lipid analogues and PH domain-targeting inhibitors (Kozikowski et al, 2003;Gills et al, 2006;Crowell et al, 2007) such as perifosine, the most developed Akt inhibitor currently available (Kondapaka et al, 2003).
In order to explore early structure-activity relationships for InsP 5 and possibly obtain more potent and specific inhibitors of the PI3K/Akt pathway, we synthesised novel compounds based on the InsP 5 structure. Here we show that the derivative 2-O-benzylmyo-inositol 1,3,4,5,6-pentakisphosphate (2-O-Bn-InsP 5 ) exhibits more efficient and potent activity than InsP 5 not only in inhibiting Akt phosphorylation but also in inducing apoptosis in different human cancer cell lines. It is interesting to note that 2-O-Bn-InsP 5 promotes apoptosis in cell lines normally resistant to treatment with InsP 5 and markedly inhibits the in vivo growth of InsP 5 -resistant xenografts. Kinase profiling analysis reveals that 2-O-Bn-InsP 5 strongly inhibits PDK1 activity in vitro with an IC 50 in the low nanomolar range. This is mirrored by the inhibition of Akt phosphorylation at its residue Thr308 in 2-O-Bn-InsP 5 -treated cells and in tumours from 2-O-Bn-InsP 5 -treated mice. Furthermore, the effect of 2-O-Bn-InsP 5 is highly specific, as this compound only inhibits PDK1 and to a lesser extent mTOR in a panel of almost 60 kinases. These data represent the first attempt to exploit InsP 5 as a potential lead compound for the development of potent small molecule inhibitors of the PI3K/Akt pathway.

Cell lines
SKOV-3 and PC3 were cultured in RPMI 1640; all other cell lines were cultured in DMEM. Media were supplemented with 10% FBS, penicillin/streptomycin and glutamine.

Cell survival and apoptosis assays
Cells seeded in a 24-well plate were treated with the indicated compounds in serum free DMEM or DMEM containing 0.5% FBS (PC3). After 72 h, the number of surviving cells was assessed by manual cell counting or by using the cell counter CDA-500 (Sysmex, Milton Keynes, UK). Alternatively, after 48 h, the number of apoptotic cells was assessed by acridine orange/ethidium bromide assay as described (Piccolo et al, 2004;Maffucci et al, 2005). SRB test was carried out in SKOV-3 and PC3 seeded in 96well plate (3800 cells per well or 5800 cells per well, respectively) after 72 h of treatment as described (Cappella et al, 2001).

In vivo studies
Male nude athymic CD-1 nu/nu mice (8-weeks old) were obtained from Harlan (San Pietro al Natisone, Italy) and maintained under specific pathogen-free conditions with food and water provided ad libitum. The general health status of the animals was monitored daily. Procedure involving animals and their care were conducted in conformity with the institutional guidelines that are in compliance with national and international laws and policies.

Toxicity assay
Male nude CD-1 mice were treated with a single dose of 750 mg kg À1 InsP 5 or 2-O-Bn-InsP 5 /mouse administered intraperitoneally (i.p.). Each group consisted of 2 -3 mice. Body weight, deaths and any other sign of toxicity and changes in behaviour (such as motility, eating and drinking habits) were recorded.

Anti-tumour activity assay
Exponentially growing PC3 cells were harvested, washed twice and resuspended in PBS at a concentration of 2.5 Â 10 7 cells ml À1 . A suspension of 5 Â 10 6 PC3 cells was injected subcutaneously (s.c.) into the left flank of the recipient mice. When tumours reached a size of B70 mm 3 (approximately 15 days after tumour cell implant), mice were divided into seven groups (n ¼ 7). InsP 5 and 2-O-Bn-InsP 5 were administered by daily i.p. injections at different doses of 12.5 -25 -50 mg kg À1 day -1 for 14 consecutive days. Control mice were treated with water in an equal volume. The diameters of s.c. growing tumours were measured with a caliper twice a week and the experiment was ended at day 28 after the implantation.

Data analysis and in vivo tumour parameters
The volume of s.c. growing tumours was calculated by the formula: Tumour weight (mg) ¼ (length Â width 2 )/2. Differences in s.c tumour growth between the treatment groups were evaluated with a one-way ANOVA followed by Fisher's test using the StatView statistical package (SAS Institute, Cary, NC, USA). The percentage of tumour growth was calculated as T/C% ¼ (RTV-treated animals/ RTV-control animals) Â 100, where RTV was the mean relative tumour volume calculated as RTV ¼ V t /V 0 . V t was the tumour volume on the day of measurement and V 0 was the tumour volume at the beginning of the treatment. The percentage of tumour weight inhibition (TWI%) was calculated using the formula: TWI% ¼ 100 -T/C%. The log cell kill (LCK) was calculated using the formula: LCK ¼ TÀC/3.32 Â Td, where TÀC is the tumour growth delay calculated as the difference in median time (in days) required for the tumours in the treatment (T) and control group (C) to reach a predetermined size (i.e., 1000 mg). Td is the tumour volume doubling time in days, determined in the exponential growth phase of the control group from a best-fit straight line. Median doubling time was 3 days in control animals.

Western blot
Mice with s.c. growing tumours were treated with a single dose of InsP 5 and 2-O-Bn-InsP 5 (50 mg kg À1 ) or vehicle. Animals were killed 24 h after treatment and tumour samples were collected and snap frozen. Frozen specimens of tumour tissue were homogenised with a Polytron homogeniser in a lysis buffer (ratio 1 : 1 w/v) containing 50 mM Tris-HCl (pH 7.4), 5 mM EDTA, 0.1% Nonidet NP-40, 250 mM NaCl, 50 mM NaF and proteases and phosphatase inhibitors. After centrifugation at 13 000 r.p.m. for 10 min at 41C, 80 mg of protein was separated on SDS -PAGE and transferred to a polyvinylidene difluoride membrane (Millipore, MA, USA). Membranes were probed with the indicated antibodies.

Protein kinase profiling
Effect of the indicated compounds on the activity of various kinases was assessed by SelectScreen Kinase Profiling Service (Invitrogen-Life Technologies, Paisley, UK). Assays were performed using 1 mM of the tested compounds and ATP concentration as indicated in the corresponding tables. In the case of InsP 5 a screen using 10 mM of the compound was also carried out, as indicated in the corresponding table.

Synthesis of novel potential inhibitors of the PI3K/Akt pathway and in vitro screening
We have recently reported that InsP 5 is a novel inhibitor of the PI3K/Akt pathway, which possesses pro-apoptotic, anti-angiogenic and anti-tumour activity (Razzini et al, 2000;Piccolo et al, 2004;Maffucci et al, 2005). To explore the design of novel inhibitors of the PI3K/Akt pathway, potentially more active than InsP 5 , we decided to modify the structures of either Ins(1,3,4,5)P 4 or InsP 5 . Different strategies were used to modify the parent molecules and several compounds were synthesised and tested. For Ins(1,3,4,5)P 4related compounds, modifications were made at C-6 because X-ray structures of the Akt (Thomas et al, 2002) and PDK1 (Komander et al, 2004) PH domains in complex with Ins(1,3,4,5)P 4 indicated that the 6-hydroxyl group of Ins(1,3,4,5)P 4 is not directly involved in binding. Furthermore, the X-ray structure of Akt PH domain showed that a tyrosine residue near the 6-OH of bound Ins(1,3,4,5)P 4 might interact with an aromatic group. In the case of InsP 5 analogs, modifications were on either the 2-O-atom or the 5-phosphate, thus maintaining the symmetry of the parent molecule. The derivatives were first tested for their ability to inhibit Akt activation in cell lines characterised by constitutive activation of the PI3K/Akt pathway and with a reported sensitivity to InsP 5 , namely ovarian cancer cells SKOV-3 and breast cancer cells SKBR3 (Piccolo et al, 2004;Maffucci et al, 2005). In this original screening we observed that the InsP 5 derivative 2-Obenzyl-myo-inositol 1,3,4,5,6-pentakisphosphate ( Figure 1A, named 2-O-Bn-InsP 5 ) showed the highest efficiency in inhibiting Akt activation in all cell lines tested (data on the other derivatives will be published elsewhere). More specifically, we found that 2-O-Bn-InsP 5 inhibited Akt phosphorylation at its residue Ser473 more efficiently than InsP 5 in SKOV-3, being already active after 8 h of treatment and at a concentration of 20 mM ( Figure 1B). Inhibition of Akt phosphorylation at residue Thr308 was also detected ( Figure 1B). More interestingly, we found that 2-O-Bn-InsP 5 was able to block Akt phosphorylation in cell lines resistant to InsP 5 , such as prostate cancer cells PC3 ( Figure 1C) and pancreatic cancer cells ASPC1 (results not shown). Taken together these data indicate that structural modification at the C-2 of InsP 5 can enhance its inhibitory properties towards Akt activation.

Analysis of the biological activity of 2-O-Bn-InsP 5
We next compared the effects of 2-O-Bn-InsP 5 and InsP 5 on proliferation/survival of cancer cells in vitro. Treatment with 2-O-Bn-InsP 5 strongly reduced the number of surviving SKBR3 ( Figure 2A) and SKOV-3 ( Figure 2B) assessed by cell counting. In particular, 2-O-Bn-InsP 5 was more active than InsP 5 in both the cell lines. Acridine orange/ethidium bromide assay confirmed that the percentage of apoptotic cells was higher in 2-O-Bn-InsP 5treated compared with InsP 5 -treated SKBR3 ( Figure 2C) and SKOV-3 ( Figure 2D). Based on the data on Akt phosphorylation we then decided to analyse the effect of 2-O-Bn-InsP 5 on the survival of cell lines normally very resistant to InsP 5 treatment. 2-O-Bn-InsP 5 was more potent than InsP 5 at a concentration of 50 mM in pancreatic cancer cells BxPc-3 ( Figure 3A), whereas it was more active than InsP 5 at almost all concentrations tested in pancreatic cancer cells ASPC1 ( Figure 3B). A stronger activity of 2-O-Bn-InsP 5 compared with InsP 5 was also observed in breast cancer cells MDA-MB-468 ( Figure 3C) and in PC3 ( Figure 3D), consistent with data on Akt phosphorylation. Higher activity of 2-O-Bn-InsP 5 in PC3 cells was also observed in SRB assays ( Figure 3E). It is important to note that although InsP 5 had no effect in PC3 at concentrations up to 50 mM, concentrations of 200 -300 mM were eventually able to mimic the effect of 2-O-Bn-InsP 5 in PC3 cells ( Figure 3F), thus suggesting that 2-O-Bn-InsP 5 is acting on the same intracellular pathway as InsP 5 . Taken together these data demonstrate that addition of a benzyl group to the axial 2-O atom of InsP 5 potentiates the pro-apoptotic properties of the compound not only in cells sensitive to InsP 5 but also in cells normally very resistant to treatment with the parent inositol compound.
In vivo anti-tumour activity of 2-O-Bn-InsP 5 on InsP 5 -resistant xenografts We then decided to test the therapeutic efficacy of 2-O-Bn-InsP 5 in human tumour xenografts characterised by the activation of PI3K/ Akt pathway and higher sensitivity to 2-O-Bn-InsP 5 compared with InsP 5 . We specifically implanted PC3 cells in nude mice and 15 days after the implantation we treated groups of mice with different concentrations (12.5, 25 and 50 mg kg À1 ) of InsP 5 or 2-O-Bn-InsP 5 for 14 consecutive days (from day 15 to day 28). Tumour growth was followed for further 12 days after the end of the treatment (upto day 40). Data revealed that 2-O-Bn-InsP 5 at doses of 12.5 and 25 mg kg À1 clearly decreased the growth of tumours compared with untreated mice, although the differences were statistically significant only on the last day of measurement ( Figure 4A and C). A strong reduction in tumour growth was obtained in the group treated with 50 mg kg À1 2-O-Bn-InsP 5 , with a statistically significant difference vs controls detectable from day 22 after tumour cells implant onwards ( Figure 4A and C). Data on in vivo anti-tumour activity parameters relative to 2-O-Bn-InsP 5 are shown in Figure 4C, bottom table. More than 50% inhibition of tumour weight was achieved in the 50 mg kg À1 -treated group, with a tumour growth delay (TÀC) of almost 9 days between this group and the untreated (control) group. In agreement with our in vitro data, we observed that InsP 5 had no effect on concentrations up to 50 mg kg À1 (Figure 4B). At the end of the experiment, western blot analysis revealed that 24 h-treatment with 2-O-Bn-InsP 5 markedly reduced Akt phosphorylation at its residue Ser473 in all 2-O-Bn-InsP 5 -treated mice ( Figure 4D). Furthermore, a clear inhibition of Akt phosphorylation at its residue Thr308 was detected in five out of seven 2-O-Bn-InsP 5 -treated mice ( Figure 4D). It is noteworthy that no evidence of toxicity was observed in different groups of mice at all the tested doses of either 2-O-Bn-InsP 5 or InsP 5 and the body weight of the treated animals was not different from the untreated mice throughout the entire experiment ( Figure 4E). Moreover, a single treatment with a very high dose of 2-O-Bn-InsP 5 or InsP 5 (750 mg kg À1 ) did not cause any major toxic effect ( Figure 4F). Taken together these data demonstrate that 2-O-Bn-InsP 5 is able to inhibit growth of InsP 5 -resistant tumours through a more efficient blockade of Akt phosphorylation in vivo.
In vitro kinase profiling of InsP 5 and 2-O-Bn-InsP 5 To determine the mechanism responsible for the higher activity of 2-O-Bn-InsP 5 , we decided to carry out a protein kinase activity screen for InsP 5 and 2-O-Bn-InsP 5 (SelectScreen Kinase Profiling Service, Invitrogen-Life Technologies). Among almost 60 protein kinases screened, 2-O-Bn-InsP 5 (1 mM) showed a very high inhibitory activity towards PDK1 (79% inhibition) and a lower activity towards mTOR (  (Table 1B). In contrast to 2-O-Bn-InsP 5 , InsP 5 did not inhibit mTOR, even when tested at a concentration of 10 mM (Supplementary Table 3). Comparing the effect of 1 mM of different natural inositol polyphosphates on PDK1, InsP 5 possessed the highest inhibitory activity towards PDK1 (71% inhibition) with only Ins(1,3,4,5)P 4 also showing some effect (56% inhibition). None of the other polyphosphates had any significant effect (Table 1C). These data indicate that 2-O-Bn-InsP 5 and InsP 5 inhibit PDK1 very specifically, with 2-O-Bn-InsP 5 possessing the highest inhibitory activity towards PDK1. Indeed results from SelectScreen Kinase Profiling Service (Invitrogen-Life Technologies) 10-point titration revealed that the IC 50 of InsP 5 towards PDK1 was 613 nM whereas the corresponding IC 50 of 2-O-Bn-InsP 5 was a striking 26.5 nM (Table 1A and B). These data clearly indicate that 2-O-Bn-InsP 5 is a novel, potent and highly selective PDK1 inhibitor. This is consistent with the detected inhibition of Thr308 phosphorylation in 2-O-Bn-InsP 5 -treated SKOV-3 and PC3 cells ( Figure 1B and C) and 2-O-Bn-InsP 5 -treated mice ( Figure 4D). Furthermore, 2-O-Bn-InsP 5 , but not InsP 5 , is able to inhibit mTOR selectively in vitro with an IC 50 of 1.3 mM (Table 1A).

In vitro effects of 2-O-Bn-InsP 5 in combination with anti-cancer compounds
Parallel RNAi and compound screens have recently revealed that PDK1 is a critical determinant of sensitivity to tamoxifen in breast  cancer cells MCF7 (Iorns et al, 2009). Based on this result we decided to investigate whether inhibition of PDK1 by 2-O-Bn-InsP 5 was able to sensitise MCF7 to the pro-apoptotic effect of tamoxifen. Our data revealed that treatment with 4-OH tamoxifen (the active metabolite of tamoxifen) for 72 h reduced the number of surviving cells, whereas 2-O-Bn-InsP 5 had little effect ( Figure 5A). It is interesting to note that the combination of 2-O-Bn-InsP 5 and 4-OH tamoxifen strongly enhanced the effect of 4-OH tamoxifen or 2-O-Bn-InsP 5 alone ( Figure 5A). We then tested the effects of 2-O-Bn-InsP 5 in combination with several natural anti-cancer compounds. The concentrations of the different compounds used in these experiments were minimally effective based on preliminary dose-response experiments (results not shown). A combination of 2-O-Bn-InsP 5 and curcumin, a component of turmeric (Curcuma longa), strongly reduced the number of surviving PC3 cells, resulting in a more than additive effect ( Figure 5B) and was able to enhance the effect of curcumin in ASPC1 ( Figure 5C)  InsP 5 and natural anti-cancer compounds results in additive or more than additive effects, and therefore suggest that 2-O-Bn-InsP 5 can potentially be used in combination with natural compounds to increase their anti-cancer activity.

DISCUSSION
The in vivo anti-tumour activity of InsP 5 together with the lack of toxicity observed using this compound (Maffucci et al, 2005), Days after cell implant suggested that InsP 5 might represent a lead compound to design novel inhibitors of the PI3K/Akt pathway to be eventually brought into clinical testing. InsP 5 possesses very few sites for chemical modification, the axial 2-hydroxyl group being the most realistic possibility. Here we describe one InsP 5 derivative, 2-O-Bn-InsP 5 , which possesses enhanced pro-apoptotic and anti-tumour activity compared with the parent molecule. In this respect 2-O-Bn-InsP 5 represents a first step towards the development of novel efficient anti-cancer drugs targeting the PI3K/Akt pathway and based on the InsP 5 structure. Kinase profiling assays revealed that 2-O-Bn-InsP 5 potently and specifically inhibits PDK1 in vitro and the PDK1-dependent phosphorylation of Thr308 Akt in cell lines and in vivo. These results are particularly important considering that, to our knowledge, no specific and selective PDK1 inhibitors are at present available and they make 2-O-Bn-InsP 5 an interesting new molecule to use as a model for designing novel specific PDK1 inhibitors. In addition 2-O-Bn-InsP 5 is able to inhibit mTOR at least in vitro (albeit to a lesser extent than PDK1). It is noteworthy that PDK1 and mTOR were the only enzymes to be inhibited by 2-O-Bn-InsP 5 in a screen of almost 60 different kinases, indicating that 2-O-Bn-InsP 5 may represent an interesting lead compound to design novel and potent dual PDK1 and mTOR inhibitors. 2-O-benzyl-myo-inositol 1,3,4,5,6-pentakisphosphate was selected in a screen of several different compounds that we synthesised and first tested for their ability to inhibit Akt phosphorylation in cells sensitive to InsP 5 . In this original screen, 2-O-Bn-InsP 5 was not only more efficient than the parent molecule in inhibiting Akt phosphorylation and inducing apoptosis in InsP 5 -sensitive cell lines but it was also able to induce apoptosis in InsP 5 -resistant cell lines including pancreatic cancer cells. The pro-apoptotic activity of 2-O-Bn-InsP 5 detected in pancreatic cancer cells represents an extremely important result taking into account the high resistance of this cancer to chemotherapeutic treatment and the urgent need for novel therapeutics in clinical treatment. Furthermore, 2-O-Bn-InsP 5 was able to inhibit the in vivo growth of InsP 5 -resistant prostate cancer xenografts. It is noteworthy that, although 2-O-Bn-InsP 5 acted in the micromolar range in in vitro studies, it was able to inhibit tumour growth in vivo at 12.5, 25 and 50 mg kg À1 , doses commonly used to test the in vivo effect of potential anti-tumour compounds. In particular, 2-O-Bn-InsP 5 induced a tumour weight inhibition of 52% in prostate cancer xenografts when dosed at 50 mg kg À1 once daily for 14 days.
We then decided to investigate in more detail the mechanisms of action of InsP 5 and 2-O-Bn-InsP 5 , and to explain the higher activity of 2-O-Bn-InsP 5 compared with the parent molecule. Our previous and current data demonstrated that the in vitro and in vivo properties of InsP 5 and 2-O-Bn-InsP 5 were because of reduced Akt phosphorylation and activation (Piccolo et al, 2004;Maffucci et al, 2005). Results from the kinase profiling assays here show that InsP 5 and 2-O-Bn-InsP 5 do not inhibit Akt kinase activity itself in vitro, whereas they are both able to directly inhibit PDK1 kinase activity (albeit with different potency), thus indicating that the detected Akt inhibition is due to blockade of the activity of its upstream regulatory kinase. This is consistent with the observed enhanced activity of 2-O-Bn-InsP 5 compared with InsP 5 , likely due to its higher inhibitory activity towards PDK1. Furthermore, this difference could explain the strong effect of 2-O-Bn-InsP 5 in InsP 5resistant cell lines such as the PTEN mutant cells MDA-MB-468 and PC3, and would be consistent with the proposed key role of PDK1 in tumourigenesis driven by Pten loss (Bayascas et al, 2005). It should be noted that the resistance to InsP 5 treatment in these cells is consistent with the delayed onset of inhibition in cells with mutant PTEN observed using the ether lipid analogues (Castillo et al, 2004). The observation that higher concentrations of InsP 5 are eventually able to mimic the pro-apoptotic effect of 2-O-Bn-InsP 5 in these cells further supports the conclusion that the different activity is because of different potency of the two compounds towards PDK1. In this respect, it is interesting to notice that the addition of the benzyl group to InsP 5 confers such a higher inhibitory activity towards PDK1 on the derivative 2-O-Bn-InsP 5 . It would be interesting to investigate whether the resulting small increase in the hydrophobicity of the molecule enhances its activity possibly by improving its binding to PDK1. Moreover, it is noteworthy that such a modification confers to 2-O-Bn-InsP 5 a selective inhibitory activity towards mTOR in vitro, providing crucial information to develop novel specific dual PI3K/mTOR inhibitors. Indeed one intriguing possibility is that 2-O-Bn-InsP 5 is more active than InsP 5 because of its unique capability to inhibit simultaneously and very specifically PDK1 and mTOR. In particular the dual activity of 2-O-Bn-InsP 5 can explain its effect  in prostate cancer cells PC3, consistent with the recently reported key role of mTORC2 in the development of loss of Pten-driven prostate cancer (Guertin et al, 2009). Furthermore, a potential 2-O-Bn-InsP 5 -mediated inhibition of mTORC2 would also increase the inhibitory activity towards Akt, by preventing its Ser473 phosphorylation, indicating that 2-O-Bn-InsP 5 represents a useful compound to be tested in cancer types specifically dependent on mTOR activation. Studies have recently revealed the existence of a negative-feedback loop by which mTORC1 inhibition leads to upregulation of Akt (Manning, 2004;O'Reilly et al, 2006) and/or ERK/MAPK pathway , activating proliferative and anti-apoptotic signals in certain cancer types. The possibility of blocking PDK1 and mTOR simultaneously in these tumours is likely to be very effective therefore, because of its dual inhibitory activity towards both enzymes, it would be interesting to investigate the effect of 2-O-Bn-InsP 5 in these cellular contexts.
Our future strategies to design novel compounds will take into consideration the possibility that, besides its direct inhibitory activity towards PDK1 and possibly mTOR, binding of 2-O-Bn-InsP 5 to non-catalytic domains may affect the activity of kinases or their mechanism of activation in vivo. Indeed in our previous work we proposed that InsP 5 can inhibit Akt activation by binding to Akt PH domain and preventing Akt recruitment to the plasma membrane . It was also proposed that inositol phosphates can bind PDK1 PH domain and retain this kinase in the cytosol, preventing Akt phosphorylation at Thr308 (Komander et al, 2004). These data suggest that the detected inhibitory effect of both 2-O-Bn-InsP 5 and InsP 5 on Akt activation in vivo can result from a combination of a direct effect on PDK1 kinase activity and effect on Akt/PDK1 recruitment to the plasma membrane. Similarly, the possibility that in vivo the inositol polyphosphates can bind and increase the activity of phosphatases which regulate Akt, such as PH domain leucine-rich repeat protein phosphatases 1 and 2 (Gao et al, 2005;Brognard et al, 2007) will be taken into consideration in our future strategies. In this respect it would be interesting to develop binding experiments of cellular lysates to immobilized 2-O-Bn-InsP 5 and InsP 5 to determine whether the compounds only bind PDK1, as indicated by the kinase profiling assays, or the in vivo mechanisms of action is more complex. These experiments would also give more information of whether the compounds may indirectly act on other kinases without directly affecting their catalytic activity.
It must be noted that, like InsP 5 , 2-O-Bn-InsP 5 is a water soluble compound and it is well tolerated in vivo even at concentrations 15 times higher the active dose. In addition, combination of 2-O-Bn-InsP 5 with other anti-cancer compounds including natural compounds results in additive or more than additive effects, indicating that such a compound (or derivatives) may prove particularly useful in combinatorial therapies. In particular 2-O-Bn-InsP 5 increases the effect of tamoxifen in breast cancer cells MCF7, consistent with the reported role of PDK1 inhibition in tamoxifen sensitisation (Iorns et al, 2009). It is worth mentioning that our in vitro assays revealed that InsP 5 itself is able to inhibit PDK1 (although less than 2-O-Bn-InsP 5 ). This raises the interesting possibility that the endogenous intracellular InsP 5 may act as an endogenous PDK1 inhibitor and regulator of the PI3K/Akt pathway. This hypothesis is currently being investigated in our laboratory.
In conclusion, here we have described the in vitro and in vivo properties of 2-O-Bn-InsP 5 , a derivative of InsP 5 , which possesses similar solubility and lack of toxicity in vivo but enhanced proapoptotic and anti-tumour activity compared with the parent molecule. In particular 2-O-Bn-InsP 5 possesses specific inhibitory activity towards PDK1. Data also indicate that 2-O-Bn-InsP 5 can inhibit mTOR, at least in vitro. It is interesting to note that InsP 5 does not possess such an inhibitory activity towards mTOR, thus suggesting that comparison of the two molecules can give useful information towards developing specific dual PDK1/mTOR inhibitors. Taken together these data indicate that InsP 5 and 2-O-Bn-InsP 5 may represent promising models for further development of novel anti-cancer drugs.