Identification of a Potent Phosphoinositide 3‐Kinase Pan Inhibitor Displaying a Strategic Carboxylic Acid Group and Development of Its Prodrugs

Abstract Activation of the phosphoinositide 3‐kinase (PI3K) pathway is a key signaling event in cancer, inflammation, and other proliferative diseases. PI3K inhibitors are already approved for some specific clinical indications, but their systemic on‐target toxicity limits their larger use. In particular, whereas toxicity is tolerable in acute treatment of life‐threatening diseases, this is less acceptable in chronic conditions. In the past, the strategy to overcome this drawback was to block selected isoforms mainly expressed in leukocytes, but redundancy within the PI3K family members challenges the effectiveness of this approach. On the other hand, decreasing exposure to selected target cells represents a so‐far unexplored alternative to circumvent systemic toxicity. In this manuscript, we describe the generation of a library of triazolylquinolones and the development of the first prodrug pan‐PI3K inhibitor.


Introduction
The discovery of phosphoinositide 3-kinases( PI3Ks) dates back to 1985 when,w ith the seminalw ork of Cantley et al., [1] the specific lipid kinases able to phosphorylate the alcoholic group selectively at the 3-position of phosphatidylinositols were reported.T hree subfamilies of PI3Ks have been identified so far, referredt oa sc lasses I, II, and III. Class Iist he most studied and comprises four isoforms, PI3Ka,P I3Kb,P I3Kd,a nd PI3Kg,t hat share somes tructural features but that are distinct for protein domains, regulatory subunits, and activation mechanism. [2] All class IP I3Ks can catalyze the in vivo conversion of phosphatidylinositol 4,5-bisphosphate (1,P IP2) into phosphatidylinositol 3,4,5-trisphosphate (2,PIP3) ( Figure 1). PIP3 production induces the recruitment to the plasma membrane of 3-phosphoinositide-dependent protein kinase-1(PDK1), which in turn activates protein kinase B( Akt) through phosphorylation on Thr308.
Thirty years of basic research on PI3Ks have unequivocally demonstrated their involvement in ap lethora of biological processes such as cell growth, proliferation, differentiation, and motility. [3] Furthermore,h yperactivation of the PI3K/Akt signaling pathway has been linked to different pathologies, such as cancer and autoimmune diseases. [4,5] Studies performed with the use of animal models and PI3K inhibitors have established PI3Ka and PI3Kb as promising targets for the treatment of human cancer. [6] Concurrently, the central role of PI3Kg and PI3Kd in leukocyte biology sustains their inhibition as apromising therapeutic approach in av ariety of inflammatory diseases. [5] These findings have prompted both academia andi ndustry research to develop inhibitors targeting all PI3K isoforms. Most of these molecules act by antagonizingt he binding of ATPt ot he PI3K catalytic pocket. To date,h undreds of panand tens of isoform-selective PI3K inhibitors have been discovered, and some of them are in clinicale valuation for tumor treatment ( Figure 2). [6][7][8] These efforts culminated in 2014 with the FDA approvalofi delalisib (8), aPI3Kd-selective inhibitor for the treatment of patients with relapsed follicularB -cell non-Hodgkinl ymphoma or relapsed small lymphocytic lymphoma (SLL). [9] Activation of the phosphoinositide 3-kinase (PI3K) pathway is a key signaling event in cancer,i nflammation, and other proliferative diseases. PI3K inhibitors are already approved for some specific clinical indications, but their systemico n-target toxicity limits their larger use. In particular,w hereas toxicity is tolerable in acute treatment of life-threatening diseases, this is less acceptable in chronic conditions. In the past, the strategy to overcomet his drawback was to block selected isoformsm ainly expressedi nl eukocytes, but redundancy within the PI3K family members challenges the effectiveness of this approach. On the other hand, decreasing exposure to selected target cells represents as o-far unexplored alternative to circumvent systemic toxicity. In this manuscript, we describe the generation of al ibrary of triazolylquinolonesa nd the development of the first prodrug pan-PI3K inhibitor.
Amore accurate analysisoft he structure of already-reported PI3K inhibitors [10] highlighted the lack of molecules displaying ac arboxylic acid group that we here report to be fundamental for binding to the PI3 kinase structural motif. Furthermore,t his functional group can undergo chemical modifications with the intent to develop useful prodrugs. Indeed, this approach might circumvent nonspecific targeting and avoid systemic side effects associated with the use of ap an-PI3K inhibitor,e specially if proposed for topical use. Therefore, in the present manuscript we report the discovery of 37,apotent pan-PI3K inhibitord isplaying as trategic and pivotal carboxylic acid group, which can be esterified to providep rodrugs. The synthesis of 37 and the metabolic stabilityo fi ts prodrugs, along with their full biological characterization, are discussed herein.

Results and Discussion
Chemistry Design, synthesis, and enzyme inhibition On the basis of the considerations discussed above, we decided to start our project by modifyingc ompound LY294002 (9, Figure 3), one of the first pan-PI3K inhibitors discovered. [11] This quercetin analogue competesw ith ATPi nt he ATP-binding site of PI3Ks, with an IC 50 value of~0.5 mm against all of the isoforms. The X-ray structure of 9 bound to PI3Kg is also avail-able, [12] and it shows that the morpholino ring partially overlaps the volume occupied by the adenine in the ATP-enzyme complex ( Figure 3).
Ah ydrogen bond is formed between the morpholino oxygen atom and the backbonea mide group of residue Val882 of the hinge region, and this bond mimics the interaction that ATPestablishes with the enzyme. There is also aputative hydrogen bond between Lys833 and the carbonyl group. The phenylr ing is located in ap ocket, usually occupiedb yt he ribose of ATP, formedb yM et804 and Trp812 on one side and Met953 on the other,a nd it protrudes toward the solvent-accessible part of the binding pocket. Interestingly, 9 does not extendi nto the phosphate-binding region.
We therefore reasoned that the phenyl ring could be replacedw ith a1 ,4-disubstituted 1,2,3-triazole by ac lick-chemistry approach. [13] In this case, the heterocycle ring would behavea samere linker,d elivering an extra site R( Figure4), that might establisha dditional interactions with the close amino-acid residues, either enhancing the potencyo f9 or impartings electivity.I ndeed, it is important to highlight that the rim of the binding site of class IP I3Ks is the regiono ft he enzymef or which there is major variability in the amino-acid composition amonga ll the four isoforms, [14] andt herefore, it is more amenable to be targeted in the search for selectivity. Owing to the versatility and functional-group tolerance of the Fokin-Sharpless reaction, [15] ac ombinatorial approachw as ex-  ploited that would allow exploration of av ast array of different functional groups in the Rp ortion ( Figure 4). It is importantt o highlight that modificationo ft he phenylr ing of 9 wasa lready attempted and described. For example, TGX155 (11), in which the phenylr ing is replaced with a4 -fluoro-2-methylphenoxy moiety,w as shown to be as elective, potent inhibitoro ft he b isoformo fP I3K and ap otential therapeutic agent for cardiovasculard iseases. [16] This example corroborates our hypothesis to replace the phenylr ing of 9.I no ur specific case, we opted to diversify the chemical nature of the Rg roup by exploiting different azides 13.T he azides were preferred as decorating groups rather than alkynes because their chemicalp reparation, startingf rom commerciallya vailable amines, is easier than the more challenging synthesis of alkynes.
Once the goal was fixed, we proceeded with the synthesis of the pivotala lkyne.T he preparationo f12 startedw ith the reactionb etween Meldrum's acid (14)a nd CS 2 (15). The intermediate was then methylated in situ to give 16. [17] Subsequently,t his compound underwent af irst nucleophilic displacementw ith 2-iodoaniline (17)a tr eflux to give 18,w hich was then treated with morpholine (19)i nT HF at reflux to give the second nucleophilic substitution to lead to compound 20. Heatingc ompound 20 in diphenyl ether at 170 8Ct riggered intramolecular aromatic nucleophilic substitution followed by a decarboxylation reaction to afford iodo intermediate 21.S onogashira coupling with trimethylsilyl acetylene gave 22 and subsequent deprotection under basic conditions afforded desired alkyne 12 (Scheme 1).
Once alkyne 12 was obtained, ac ombinatorial parallel synthesis with 30 azides was performed. To this point, before startingt he construction of the library,apreliminary modelr eaction with phenyla zide (23)w as attempted, which revealed that under the standard conditions of the Fokin-Sharpless click reaction( CuSO 4 ·5H 2 O, sodium ascorbate, tBuOH/H 2 O) unexpected compound 24 was obtained insteado fd esired triazole 25.I tw as reasonable to suppose that coppert riggered intramolecular cyclization between the N1 atom and the triple bond, with the formation of a6 -oxopyrrolquinolinicr ing (Scheme 2). [18] In light of this event,w er easoned that in the presence of an already-chelated copper ion, this transformation might be suppressed. The ligand tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA, also known as the Chan ligand) was our first choice. It was previously shown that TBTAw as indeed able to stabilizet he oxidation state of the copper(I) ion  by encapsulatingi ti ni ts structure and preventing other interactions. [19] The use of the Chan ligand wass uccessful, and under optimized reaction conditions (THF/H 2 O, Cu(OAc) 2 , sodium ascorbate,T BTA), desired 1,2,3-triazole 25 was obtained in excellent yield (Scheme 2).
With this procedurei nh and, we proceeded with our combinatorial strategy.T he azidesw ere selected to introduce both electron-donating and electron-withdrawing groups, hydrogen-bond donors and acceptors, and ionizable functions. They were already presenti no ur databaseo rp repared according to standard procedures. The feasibilityo ft his approachw as demonstrated by precipitation of the resultantp roducts,w hich were purified by filtration and subsequent water/diethyl ether washes. Thisp rocedure resulted in the formation of 30 compounds,w hich were submitted to MS and 1 HNMR spectroscopy analysis to verify their purity (> 95 %) and structural identity. Then, the 1,2,3-triazolesw ere evaluated for their ability to inhibit PI3K isoformso fc lass Ia tafixed concentration of 1 mm, as shown in Table 1. The synthesized 1,2,3-triazoles along with their ability to inhibitP I3K isoformsa t1mm concentration are  quinolin-3-yl ÀÀÀÀ 37 3-carboxybenzyl [a] PI3K inhibitiond etermined at 1 mm with at least triplicate determinations in multiple experiments; ' + ' refers to those compounds able to inhibit PI3K isoform activity by more than 50 %. reported.P ossibly,aminute concentration of copper salts might have precipitated with the productsa nd might have been retained in the purified compounds, and so, we also evaluated the PI3K inhibitory activity of theses alts. Neither copper(II) acetate nor copper(I) iodide displayed significant inhibition of PI3Ks up to 10 mm.
Only those compounds able to suppress the kinase activity by 50 %o fa tl east one isoform (i.e., 25,35,37,45,48,a nd 54)a t1mm were resynthesized, purified by column chromatography,a nd reevaluated to obtain ap recise IC 50 value.
We then proceeded to evaluate the ATPc ompetitive inhibitory nature of 37 through an enzymatick inetic study by using an in vitro assay to measuret he PI3Kd activity at various concentrations of ATPa nd 37.A ss howni nF igure 6, Lineweaver-Burk plot analysis revealed that 37 acts as an ATP-competitive inhibitorf or PI3Kd,a st he plots (straight lines) intersect on the 1/v axis (v = rate of ADP production).
To investigate the binding of 37 in the enzyme active site, the crystal structure of 37 bound to the murine d isoform of PI3K was resolved (see the Supporting Information for technical details). Compound 37 binds in the ATPb inding site in a canonical mode (Figure 7), as described for other type Ik inase inhibitors. In detail, the morpholine ring establishes at ypical hydrogen bond with Val882 in the hinge region, analogously to 9;t he quinolone ring is in ac entral pocket with an orientation very similart ot hat of 9 (Figure 7a), and the carbonyl group is involved in ap utative hydrogen-bonding interaction with Asp911.T he 1,2,3-triazole appears as fundamental to orientate the carboxylic acid group properly.I ndeed, the X-ray structure of 9 showsaperpendicular orientation of the phenyl ring to the central core, whereas 37 is characterized by copla-  . [20] ATPl oss and consequent ADP production were measured thoughaluminometric assay. This different spatiald isposition allows for ap ivotal ionic saltbridge interaction between the carboxyl moietya nd Lys708, which is consistentw ith the observed potent inhibitory activity.

Cellular inhibitory activities
After identifying 37 as ap romising candidate, we started to evaluatei ti nc ell-based assays to define its inhibitory activity on the PI3K signaling pathway.T herefore, we selected an in vitro insulin modelt oa ssess the inhibitory effect of 37 in PI3K signaling. NIH3T3c ells were treated with different concentra-tions of 37,s timulated with insulin, andt he amount of phosphorylated Akt was detected. Nonetheless, as shown in Figure 8, 37 did not affect the PI3K signaling pathway,a si td id not decrease Akt phosphorylation. We therefore reasoned that the lack of activity of 37 in cellbased experiments could be ascribed to its inability to cross cell membranes owing to the increased polarity imparted by the ionizedc arboxylic acid. For this reason, we prepared corresponding methyl ester 64.Alipid kinase assay confirmedt hat 64 was not able to inhibit PI3Ka activity at 100 nm (see the Supporting Information), which confirmed the crucial role played by the carboxyl group in the interaction with PI3Ks.
Values are the means of at least threed eterminations in multiple experiments, with 95 %c onfidence intervals; n.d.:n ot determined. See the Experimental Section for the synthesis of compound 63.  Nevertheless, given its hydrophobic nature, it was conceivable that the methyl ester could cross the membrane and, after intracellular enzymatic hydrolysis, explicate its inhibitory activity.
To demonstrate that 64 could act as ap rodrug, we treated NIH3T3 cellsw ithd ifferentc oncentration of 64,a nd we then analyzed Akt phosphorylation after insulin stimulation. As shown in Figure 8, methyl ester 64 was able to inhibit the PI3K-signalingp athway in ad ose-dependent manner.
To determine whether 64 could inhibitP I3K/Akt in tumor cells, in which the PI3K signaling is hyperactive,P C3 prostatic tumor cells were incubated with different concentrations of 64 ( Figure 9). Treatment with 64 attenuated,i nadose-dependent manner,t he levels of p-Akt, as well as the levelso fs ome Akt effectors such us p-PRAS40,p -p70S6K, p-GSK3a/b,a nd p-FOXO3A.
We next determined the effect of 64 on cell proliferation, cell cycle, and apoptosis. As shown in Figure 10 a, 72 hoft reatment with 64 dose-dependently decreased proliferation of PC3 cells. In addition, after 48 ho ft reatment, 64 was able to induce cell-cycle arrest by significantly increasingt he percentage of cells in the G0-G1 phase with concomitantr eduction in the percentage in the Sa nd G2/M phases (Figure 10 b). On the other hand, 64 did not showa ny effect in inducing cell apoptosis (see the Supporting Information).

Prodrug selectiona nd metabolic stability
The opportunity to generated ifferent prodrugs by acting on the carboxylic acid group stimulatedt he design and synthesis of ap anel of different esterified products. It is important to highlight that by browsing the literature data on PI3K inhibitors it emerged that only one other reported PI3Ka inhibitor displayed ac arboxylic acid, but in this case, the corresponding methyl ester was still ablet oi nhibitt he PI3 kinases and was only four times less active than the corresponding carboxylic acid analogue (IC 50 on PI3Ka:46.0 versus10nm). [21] Twelve esters werep repared (see the Experimental Section for their synthesis and the Supporting Information for chromatographic methods) and evaluated in terms of metabolic stability.F or this task, ac ombined approach based on microsomal and plasmatic stabilitya ssays was exploited considering the relevance of the plasmatic and hepatic tissues in esterase activity.M icrosomals tability of the selected esters was evaluated by monitoring the disappearance of the substrate incubatedi n rat liver microsomes (RLM) at 50 mm concentration, in the presence of NADPH. In Table 4, the percentage of the remaining Figure 8. Treatment with 64 prodrug inhibitedPI3K/Akt signaling. NIH3T3 cells were harvestedf or 12 hand were treatedw ith the indicatedconcentrations of 37 and 64.Cells were next stimulated with 1 mm insulin for 5min, and totalcell lysates were prepared. Immunoblot analysis was conducted for the expressionlevels of p-AKTand GAPDH. The immunoblot is representative of three independent experiments.  substrate, after incubation (t = 60 min), with respect to the initial amount is reported. Overall, the compounds underwent a relevant metabolic transformation,a nd the substrate depletion ranged from 0( for 76)t o1 00 %( for 70). This is mainly due to hydrolysiso ft he ester function, whereas the metabolites arising from the oxidation catalyzed by the NADPH-dependent monooxygenase system weren egligible from aq uantitative point of view.I ndeed, the main metabolite formed during the incubation of the esters was corresponding carboxyl derivative 37.F urthermore, the addition of NADPH did not significantly increaset he oxidative metabolic transformation (see Figure S3 in Supporting Information). As expected, the aliphatic esters were found to be hydrolytically labile, and ethyl ester 67 was less stable than methyl ester 64, [22] whereas long, linear alkanoates 75 and 76 and branched esters 69 and 74 showed greaters tability. Moreover, whereas CH 2 -pyridin-4-yl ester 70 was completelym etabolized, corresponding benzyl derivative 65 and piperonyl derivative 72 weremore hydrolytically stable, and their residual substrates were 53 and 83 %, respectively. The hydrolytic stabilityo ft he selected esters was also evaluated by monitoring the depletion of the substrate incubated at a5 0 mm concentrationi nm ousep lasma over ap eriod of 30 min. As expected, the plasmatic metabolic fate of the esters was characterizedb yt he formation of 37 as the only metabolite (see Figure S4). The rate constant k [min À1 ], derived for the substrate peak area versust ime curve (single-exponential decay equation-GraphPad Prism software, Inc.,S an Diego, CA (USA);s ee Figure S5) was used to calculate the in vitro half-life t 1/2 [min],a nd the valuesa re reported in Ta ble 4. Except for branched ester 69 and long, linear ester 76,w hich were inert to the mousep lasma hydrolases, the other compounds underwent relevant hydrolytic cleavage with t 1/2 values ranging from < 1t o1 5min. In detail, short-linear esters 67 and 73 in addition to benzyl derivative 65,C H 2 -pyridin-4-yld erivative 70,a nd piperonyl derivative 72 were rapidly hydrolyzed, and their t 1/2 valuesw ere less than 1min (see the Supporting Information).
The biological activity of each prodrug was tested by analyzing their ability to decrease PC3p roliferation as ar apid test to evaluatet he PI3Ks inhibitory activity after hydrolysis in ac ellbased assay.P C3 cells werei ncubated with 10 mm of each prodrug:c ompounds able to decreasep roliferation to an extent more than 50 %w ere defined as active, whereas those reducing to an extent less than 50 %w erec onsidered inactive. The choiceo fu sing such ac ut-off was due to the intent to identify rapidly which prodrug couldb eh ydrolyzed in the assay to release the active compound.A se xpected, ag ood correlation between rateo fh ydrolysis and cellular activity was observed, and esters 64-68 and 70-73 were determined to be good candidates for topicalu se. We decided to focus our attentiono n ester 64,a si ts half-life was ag ood compromise for ap rodrug. The EC 50 value of this ester wast hen evaluated on PC3 cells, and av alue of (2.3 AE 0.8) mm was obtained.

Conclusions
In conclusion, we report herein the discovery of 3-{[4-(2-morpholino-4-oxo-1,4-dihydroquinolin-8-yl)-1H-1,2,3-triazol-1-yl]me-thyl}benzoic acid (37), an ovel, potent pan-phosphoinositide 3kinase (PI3K) inhibitort hat bears as trategic carboxylic acid, and the design of its ester prodrugs. Esterification of 37 generated molecules that were inactivei nt he enzymatic assay but that could cross cell membranes, and after ester hydrolysis, they inhibited the PI3Ks. To date, few prodrug PI3K inhibitors have been reported.O ne is ap rostate-cancer-specific PI3K inhibitor generated by coupling ac hemically modified form of the quercetin analogue LY294002 with the peptide Mu-LEHSSKLQL, in which the internal sequence HSSKLQi sasubstrate for the prostate-specific antigen protease. [23] The second prodrug was derived by conjugation of LY294002 to the amino terminus of the Arg-Gly-Asp-Ser integrin bindingt etrapeptide through am orpholinium methylsuccinate. [24] However,i nb oth prodrugs the biologically active portion was ablet oc ross cell membranes, which raises questions on side effects mediated by uncontrolleds ystemic inhibition of PI3Ks. Recently,A bbott et al. reported the generation of tetrazole analogues of LY294002 active on the a isoform of PI3K only if delivered as a prodrug. [25] Although different cancerc ell lines were used, an EC 50 value comparable with the one displayedb yo ur compound, methyl 3-{[4-(2-morpholino-4-oxo-1,4-dihydroquinolin-8-yl Although pan-PI3K inhibitors are emerging as an ew opportunity in the treatment of non-cancer-related diseases,dataderived from clinicalt rials has demonstrated that treatmentw ith pan-PI3K inhibitors causes several side effects, including nausea,d iarrhea, rash, and increased released of insulin in plasma.T herefore, the unique properties of the prodrugs described in this manuscript might be considered of great relevance in the context of systemic toxicitym anagement. Indeed, these derivatives might explicatet heir activity exclusively within the cell, after conversion into 37,w hich might in turn decreases ystemic toxicity.T heirs hort half-lives,b oth in microsomes and plasma, highlight the therapeutic potential in topical administration (e.g.,a erosol therapy or epicutaneous treatment), which might overcome toxicitya nd avoid systemic exposure.I ndeed, severalp ieces of evidence suggest the use of pan-PI3K inhibitors in topicala dministration. [26,27] Further studies are underway and will be performed to demonstrate the efficacy of compound 64 in different mousem odelso fc hronic inflammation, for which topical administration would be beneficial, such as asthma and psoriasis, and to evaluate its systemic toxicity.

Experimental Section
Generalprocedures Commercially available reagents and solvents were purchased from Aldrich or Alfa Aesar and were used without further purification. Te trahydrofuran (THF) and toluene were distilled immediately before use from Na/benzophenone under as light positive atmosphere of dry nitrogen. Ethylenediamine was distilled immediately before use from CaH 2 under as light positive atmosphere of dry nitrogen. N,N'-Dimethylformamide (DMF) was distilled under vacuum from KOH and was stored on activated molecular sieves (4 ). Dichloromethane was dried by distillation from P 2 O 5 and was stored on activated molecular sieves (4 ). If needed, the reactions were performed in flame-or oven-dried glassware under ap ositive pressure of dry nitrogen. Melting points were determined in open glass capillaries with aS tuart scientific SMP3 apparatus. Compounds were checked by IR (FTIR THERMO-NICOLET AVATAR), 1 HNMR and 13 CNMR APT (JEOL ECP 300 MHz spectrometer), and mass spectrometry (Thermo Finnigan LCQ-deca XP-plus) equipped with an ESI source and an ion-trap detector.C hemical shifts are reported in parts per million (ppm). Flash column chromatography was performed on silica gel (Kieselgel 60, 230-400 mesh ASTM). Thin-layer chromatography (TLC) was performed on 5 20 cm plates with al ayer thickness of 0.25 mm (Merck Silica gel 60 F 254 ). If necessary,t hey were developed with KMnO 4 reagent. The purity of the tested compounds was established by elemental analyses. Elemental analyses (C, H, N) of the compounds that underwent biological evaluation were within AE 0.4 %o ft he calculated values, confirming ! 95 %purity. 17 mmol) in DMSO (300 mL). The mixture was stirred for 3hat room temperature, and then iodomethane (2 equiv) was slowly added at 0 8C. The mixture was stirred overnight at room temperature. Ice was added to the flask to promote precipitation of the product, which was then filtered to yield compound 16 as ayellow solid (34.1 g, 81 %).

2-Morpholino
The residual PI3K lipid kinase activity in the presence of each molecule at different concentrations was related to the control, intended as the recombinant protein in the presence of DMSO alone and was expressed as ap ercentage. To derive the IC 50 value, all data (% of lipid kinase activity) were plotted on ad ose-response curve (Graph Pad software), and the IC 50 was calculated by using nonlinear regression fit (equation [log agonist] versus response).
Lineweaver-Burk plot:T he linear phase of the kinetic reaction was defined with 2.4 mgmL À1 of PI3Kd in a2 0min reaction time. In the presence of increasing concentrations of 37 (5, 10 and 20 nm), the PI3K lipid kinase activity (for the assay method see "In vitro selection, lipid kinase assay" in the Experimental Section) was assayed at various concentrations of ATP ( 5,10,25,50, and 100 mm)a nd in the presence of 0.5 mg mL À1 lipid micelles (1:1 phosphatidylinositol/phosphatidylserine). AL ineweaver-Burk plot was obtained by plotting 1/v,for which v represents pmol min À1 of ADP produced in the reaction measured with the ADP-Glo Kinase Assay kit from Promega (#V9102) versus 1/[ATP] (the inverse of the ATPc oncentration). The minus-enzyme control was obtained by incubating phosphatidylinositol in the absence of PI3Kd.
NIH3T3 cells were seeded in a9 6-well plate and starved overnight (ON) with serum-free DMEM high glucose GlutaMAXTM (Gibco) supplemented with 5000 UmL À1 penicillin-streptomycin (Gibco). Cells were then incubated for 1hwith increasing concentrations of 64, 37,o rD MSO. Subsequently,c ells were stimulated for 5min with 1 mm of insulin (#91077C, Sigma-Aldrich, USA). Cells were lysed and protein extracts were analyzed by western blot to assess the phosphorylation level of Akt on Ser473.
Cell proliferation:P C3 cell proliferation was measured with a3 -(4,5dimethyl thiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)-based colorimetric assay (#114 65 007 001 Roche Diagnostic GmbH, Germany). PC3 cells were seeded in a9 6-well plate at ac oncentration of 10 4 cells mL À1 .P C3 cells were treated for 72 hw ith increasing concentrations of 64,a nd MTT was added. After ON solubilization of the MTT metabolite, formazan, the absorbance (l = 550-600 nm absorbance, with ar eference wavelength of 650 nm) of each well was read in am ulti-well reader (Glomax multi detection system, Promega, model number 9301-010). The percentage of proliferating cells was calculated by relating each absolute value to the control intended as PC3 cells treated with DMSO alone. An IC 50 value was then derived by plotting all data on ad ose-response curve (Graph Pad software) by using nonlinear regression fit (equation [log agonist] versus response).
Cell-cycle analysis:F or DNA content determination, PC3 cells were seeded at 4 10 5 cells per 6cmd ish in triplicate in the presence of DMEM high glucose GlutaMAXTM (Gibco), 10 %f etal bovine serum (Invitrogen) supplemented with 5000 UmL À1 penicillin-streptomycin (Gibco). The day after seeding, cells were treated for 48 hw ith 10 mm 64 or DMSO. Subsequently,P C3 cells were detached with 0.1 %t rypsin-EDTA( Gibco), fixed in 70 %e thanol overnight at À20 8C, and stained for 40 min at 57 8Cw ith PI solution, containing 0.05 %T riton X-100, 0.1 mg mL À1 RNase, and 25 mgmL À1 propidium iodide. Sample were analyzed for the DNA content by using a FACS-Calibur flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, USA). The percentages of cells in the G0/G1, S, and G2 phases were calculated.
Analysis of apoptosis:F or Annexin Vs taining of apoptotic cells, PC3 cells were seeded at 4 10 5 cells per 6cmd ish in triplicate. Cells were incubated 48 hi nt he presence or absence of 10 mm 64 and were then detached with 0.1 %t rypsin-EDTA, resuspended in Annexin buffer (10 mm HEPES, pH 7.4, 140 mm NaCl, 2.5 mm CaCl 2 ), and stained for Annexin V( FITC Annexin V# 640906, BioLegend, USA). Samples were analyzed by using aF ACS-Calibur flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, USA).
Additional biological experiments, crystallographic details, the pdb file for the X-ray crystal structure of 5NGB, and metabolism stability studies are available in the Supporting Information.