1521-0111/88/5/935–948$25.00 http://dx.doi.org/10.1124/mol.115.100131MOLECULAR PHARMACOLOGY Mol Pharmacol 88:935–948, November 2015Copyright ª 2015 by The American Society for Pharmacology and Experimental Therapeutics

Identification of a Highly Conserved Allosteric BindingSite on Mnk1 and Mnk2 s

Sunita K.C. Basnet, Sarah Diab, Raffaella Schmid, Mingfeng Yu, Yuchao Yang,Todd Alexander Gillam, Theodosia Teo, Peng Li, Tom Peat, Hugo Albrecht,and Shudong WangCentre for Drug Discovery and Development, Sansom Institute for Health Research, School of Pharmacy and Medical Sciences,University of South Australia, Adelaide, Australia (S.K.C.B., S.D., R.S., M.Y., Y.Y., T.A.G., T.T., P.L., H.A., S.W.); and CSIROBiosciences Program, Parkville, Victoria, Australia ( T.P.)

Received May 27, 2015; accepted August 11, 2015

ABSTRACTElevated levels of phosphorylated eukaryotic initiation factor4E (eIF4E) have been implicated in many tumor types, andmitogen activated protein kinase-interacting kinases (Mnks)are the only known kinases that phosphorylate eIF4E at Ser209.The phosphorylation of eIF4E is essential for oncogenic trans-formation but is of no significance to normal growth and devel-opment. Pharmacological inhibition of Mnks therefore providesa nontoxic and effective strategy for cancer therapy. However,a lack of specific Mnk inhibitors has confounded pharmaco-logical target validation and clinical development. Herein, wereport the identification of a novel series of Mnk inhibitors andtheir binding modes. A systematic workflow has been estab-lished to distinguish between type III and type I/II inhibitors. A

selection of 66 compounds was tested for Mnk1 and Mnk2inhibition, and 9 out of 20 active compounds showed type IIIinteraction with an allosteric site of the proteins. Most of thetype III inhibitors exhibited dual Mnk1 and Mnk2 activities anddemonstrated potent antiproliferative properties against theMV4-11 acute myeloid leukemia cell line. Interestingly, ATP-/substrate-competitive inhibitors were found to be highly se-lective for Mnk2, with little or no activity for Mnk1. Our studysuggests that Mnk1 andMnk2 share a common structure of theallosteric inhibitory binding site but possess different structuralfeatures of the ATP catalytic domain. The findings will assist inthe future design and development of Mnk targeted anticancertherapeutics.

IntroductionThe eukaryotic translation initiation factor 4E (eIF4E) is

phosphorylated by mitogen activated protein kinase-interacting kinases (Mnks), and elevated phosphorylationlevels are associated with cell proliferation and malignancy.There is growing interest in targeting Mnks as anticancertargets (Altman et al., 2013; Hou et al., 2012; Diab et al.,2014a). Specifically, it has been shown in animal models thatphosphorylation of eIF4E at Ser209 is crucial for tumorformation (Furic et al., 2010; Ueda et al., 2010), whereas micewith a Ser209Ala mutation are resistant to tumor develop-ment (Furic et al., 2010). Furthermore, mutated and consti-tutively active Mnk1 is known to drive tumor formation, andthis corresponds with enhanced eIF4E phosphorylation (Wen-del et al., 2007). Finally, phosphatase and tensin homolog2/2

mice crossed with Mnk 1/2 double knockout mice showreduced tumor burden compared with phosphatase and tensinhomolog2/2 mice (Ueda et al., 2010). Both rat sarcoma/rapidlyaccelerated fibrosarcoma/extracellular signal-regulated ki-nase and phosphatidylinositol 3-kinase/Akt/mammalian tar-get of rapamycin signaling pathways play important roles inoncogenesis, and they are known to synergistically activatethe oncogenic eIF4E (Konicek et al., 2011; Hou et al., 2012;Diab et al., 2014a). Therefore, inhibiting eIF4E phosphoryla-tion presents an opportunity to concurrently target twosignaling pathways. Several studies have suggested thateIF4E phosphorylation is of no importance for normal cellgrowth and development (Ueda et al., 2004; Konicek et al.,2011). Despite the fact thatMnks could potentially be effectivetargets for cancer therapywith better outcomes and fewer sideeffects, only limited advancement has been made in develop-ing pharmacological inhibitors. For example, CGP57380 hasbeen described as Mnk inhibitor and exhibited activity incancer cells at micromolar concentrations (Diab et al., 2014b;Teo et al., 2015b; Yu et al., 2015), but it also inhibited severalother kinases with similar potency (Bain et al., 2007). AnotherMnk inhibitor, cercosporamide (Sussman et al., 2004), was

This work is supported by the Australia Government National Health andMedical Research Council [Grant 1050825] and South Australian Health andMedical Research Institute, Beat Cancer Project Principal Cancer ResearchFellowship to S.W.

dx.doi.org/10.1124/mol.115.100131.s This article has supplemental material available at molpharm.

aspetjournals.org.

ABBREVIATIONS: DMSO, dimethylsulfoxide; eIF, eukaryotic initiation factor; GI50, concentration with 50% growth inhibition; IC50, half-maximalinhibitory concentration; IMAP, immobilized metal ion affinity particle; Ki, inhibition constant; Km, Michaelis-Menten’s constant; Mnk, mitogenactivated protein kinase-interacting kinase; Ras, rat sarcoma; Rfu, relative fluorescence units; S/B, signal-to-background ratio; TAMRA,tetramethylrhodamine; Tb, terbium; TR-FRET, time resolved fluorescence resonance energy transfer.

935

http://molpharm.aspetjournals.org/content/suppl/2015/08/11/mol.115.100131.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

recently shown to have antitumor activity in human tumorxenografts (Konicek et al., 2011). However, it also exertednonspecific effects against other kinases. Such off-targeteffects compromise the pharmacological proof of concept forMnks as anticancer targets.Kinase inhibitors have been categorized according to dis-

tinctmechanisms. Type I inhibitors typically form one to threehydrogen bonds with the hinge region of the kinase (Liu andGray, 2006), bind to active and inactive conformations (Nagaret al., 2002; Tokarski et al., 2006; Vogtherr et al., 2006;Wodicka et al., 2010), are ATP competitive, and often havelimited selectivity (Zhang et al., 2009; Blanc et al., 2013) due tointeraction with the highly conserved ATP binding pocket. Inmost cases, type II inhibitors not only form one to threereversible hydrogen bonds at the hinge region of the ATPactive site but also interact with the adjacent DFG/D pocket. Ithas been shown that some type II compounds display strongerbinding affinity for kinases when the activation loop adoptsthe inactive DFG/D-out conformation (Nagar et al., 2002;Tokarski et al., 2006; Vogtherr et al., 2006; Wodicka et al.,2010). Type III inhibitors (colloquially known as allostericinhibitors) bind exclusively outside of the ATP binding pocket,presumably to nonconserved structural motifs, giving rise toa potentially higher degree of selectivity. These bindingmodalities are non-ATP-competitive and reversible (Eglenand Reisine, 2011; Blanc et al., 2013).The aim of the current study was to identify compounds that

would potentially be used for the pharmacological validationofMnks as anticancer targets. Our particular interests were 1)discovery of dual and subtype specific Mnk1/2 inhibitors; 2)identification of type III inhibitors lacking substrate compe-tition with ATP and an eIF4E-derived peptide; and 3)selection of Mnk inhibitors with antiproliferative effects onleukemia cells. We report the systematic profiling of com-pounds that were derived from three distinct series. Weestablished a screening cascade that consists of a nonradio-metric fluorescence resonance energy transfer (FRET)-basedimmobilized metal ion affinity particle (IMAP) assay forprimary testing, followed by an ADP-Glo assay for dose-response and substrate competition analysis. A total of 20compounds showed inhibitory activities against Mnk2 withIC50,2mM.Among of them, 13 compounds are dual inhibitorsagainst Mnk1 and Mnk2 and 7 compounds are subtypespecific, being .10-fold more potent for Mnk2 over Mnk1.No Mnk1 specific inhibitors were identified. All Mnk inhib-itors with IC50 values ,2 mM were investigated for substratecompetition. In total, nine inhibitors were found to be non-competitive with both substrates and most of them exertedantiproliferative activities onMV4-11 acutemyeloid leukemiacells. Interestingly, eight of the nonsubstrate-competitivecompounds were dual inhibitors, suggesting a common allo-steric site between Mnk1 and Mnk2.

Materials and MethodsKinase Assays

The IMAP screening kit (Molecular Devices, Sunnyvale, CA) wasused for primary Mnk inhibitor screening with a specially designedeIF4E-derived peptide substrate.Hit compound characterization (e.g.,IC50 determination and substrate competition testing) was conductedwith the ADP-Glo kit for ATP concentrations up to 1mMand the ADP-

Glo max kit for higher ATP concentrations up to 6 mM (PromegaCorporation, Madison, WI).

Peptide Substrates

The peptide substrate eIF4E202-214 comprised the 13 C-terminalamino acids of human eIF4E (D202TATKSGSTTKNR214). Customsynthesiswas carried out byMimotopes Pty. Ltd. (Victoria, Australia),and this peptide was used for the ADP-Glo assays. The peptidesubstrate TAMRA-eIF4E202-214 was used for the IMAP technology.For this purpose eIF4E202-214 was N-terminally linked with tetrame-thylrhodamine (TAMRA).

Mnk1/2 Proteins

Recombinant glutathione s-transferaseMnk1/2 andHis-Mnk2werepurchased from Life Technologies (Grand Island, NY) and MerckMillipore (Dundee, UK), respectively.

Compounds

Cercosporamide and CGP57380 (N3-(4-fluorophenyl)-1H-pyrazolo-[3,4-d]pyrimidine-3,4-diamine) were supplied by BioAustralis (Smithfield,NSW, Australia) and Sigma (Castle Hill, NSW, Australia), respectively.These compounds were used as references for kinase assays. Com-pounds of series 1 (N-phenylthieno[2,3-d]pyrimidin-4-amines), series2 (3,4-dimethyl-5-(2-(phenylamino)pyrimidin-4-yl)thiazol-2(3H)-ones),and series 3 (N-phenyl-1H-pyrazolo [3,4-d]pyrimidin-3-amine) weresynthesized in our laboratory as reported previously (Diab et al.,2014b; Teo et al., 2015a,b; Yu et al., 2015) and were dissolved indimethylsulfoxide at 10 mM concentration and stored at 220°C.

IMAP TR-FRET Kinase Assay

The kinase reactionwas performed in 1� IMAP reaction buffer withTween-20, dithiothreitol, and distilled H2O according to manufac-turer’s protocol using the TAMRA-eIF4E202-214 peptide substrate. Thetest compounds were added to the above kinase reaction mixture at 1and 10 mM final concentrations with DMSO content of 0.5%. TheneachMnkkinasewas dilutedwith 1� IMAP reaction buffer containingTween-20 (see manufacturer’s protocol) and added to the reactionmixture. The kinase reaction was started by final addition of ATP andincubated at 30°C for 90 minutes in a total assay volume of 20 ml. Thereaction was stopped by adding 60 ml of progressive binding solution[30% binding buffer A, 70% binding buffer B, progressive bindingreagent 1:600 and terbium (Tb)-donor (1:400)], and the plate was thenread in an EnVision Multilabel plate reader (PerkinElmer, Bucking-hamshire, UK) after 5 hours of incubation in the dark. The excitationwavelength for the Tb-donor was set at 330 nm, and the emissions ofthe Tb-donor and the TAMRA-eIF4E202-214 substrate (TR-FRET signal)were recorded at 545 and 572 nm, respectively. The emission wasmeasured in a time resolved mode with a delay of 200 ms. Positive andnegative controls were carried out in 0.5% DMSO in the presence andabsence of Mnk kinases, respectively. For dose-response experimentswith controlMnk inhibitors (CGP57380 and cercosporamide), the kinasereaction was run in a similar way with an inhibitor dilution factor of 1:3for eight concentrations ranging from 10 mM to 4.5 nM in 0.5% DMSO.

ADP-Glo Kinase Assay

Assay kits from Promega Corporation were used according toinstructions and adapted as outlined in the Results. Test compoundswere generally prepared with 1:3 serial dilutions for 8 concentrations(from 10mMto 4.5 nM); 10 concentrations were used (15mMto 0.7 nM)for substrate competition experiments. The kinase reaction wasperformed with 1� kinase reaction buffer (40 nM Tris base pH 7.5,20 mM MgCl2, 0.4 mM dithiothreitol), 0.1 mg/ml bovine serumalbumin, distilled H2O, eIF4E202-214 substrate, and Mnk kinases ina total assay volume of 15 ml after the manufacturer’s protocol. Inbrief, the kinase reactions were started by addition of ATP, incubated

936 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

for 45 minutes at 30°C, and then stopped by adding 15 ml of ADP Gloreagent. After incubation at room temperature in the dark for 40minutes, 30 ml of kinase detection reagent was added per well andincubated for 30–60 minutes depending on the ATP concentrationused in the kinase reaction (10–100 mM: 30 minutes, 100–500 mM: 40minutes, 500–1000mM: 60minutes). For higher concentrations of ATP(up to 6 mM), the ADP-Glo max kit was used, which is similar to ADP-Glo kinase assay but adapted for high ATP concentrations as used inATP competition experiments. The incubation time after addition ofthe detection reagent was 60 minutes for this assay. Luminescencewas measured using an EnVision multilabel plate reader (Perki-nElmer) with an integration time of 1 second per well. Positive andnegative controls were performed in 0.5% DMSO in the presence andabsence of Mnk kinases, respectively. Standard curves with definedADP/ATP ratioswere routinely performed and used to convert relativelight units into reaction velocities (e.g., Vmax calculation).

Data Analysis

Microsoft Excel (Redmond, WA) or Graphpad Prism software(version 6.0; San Diego, CA) was used for data analysis.

The normalized TR-FRET signal was calculated by dividing therelative fluorescence units (Rfu) measured from the TR-FRET emis-sion (572 nm) by the Tb emission (545 nm) and multiplied by 10,000:

TR-FRET5TR-FRET Rfu

Tb Rfu� 10;000

Determination of the maximal velocity (Vmax) and Michaelis-Mentenconstant (Km) was carried out using Michaelis-Menten curve fitting.

Dose-response curves (IC50 and GI50) were calculated using a four-parameter logistic nonlinear regression model.

The Z9-factor (Zhang et al., 1999) was routinely calculated toquantify the assay performance according to the following equation:

Z95123� ððp 1 ðnÞn

ðmp 2mnÞ

Where sp5 standard deviation (S.D.) of positive controls; sn 5 S.D. ofnegative controls; mp 5 mean of positive controls; mn 5 mean ofnegative controls.

The Pearson correlation was calculated with the following formula:

CorrelðX;YÞ5 + ðx2 �xÞðy2 �yÞffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi+ ðx2 �xÞ2+ ðy2 �yÞ2

q

Where x and y represent the mean of IC50 and GI50, respectively.

Kinase Profiling

Inhibition of CDK2 and CDK9 were measured using a radiometricassay by Eurofins Pharma Discovery Services (Dundee, UK) usingATP concentration within 15 mM of Km for each kinase. The half-maximal inhibitory concentration (IC50) values were determined from10 data point dose-response experiments, and the inhibitory constants(Ki) were calculated from IC50 values and Km (ATP) values for eachkinase.

Cell Culture

MV4-11 cells were kindly provided by Prof. Richard D’Andrea(University of South Australia) and maintained in RPMI 1640 with10% fetal bovine serum.

Proliferation Assay

Resazurin (Sigma) assay was carried out in MV4-11 cells aspreviously reported (Diab et al., 2014b). The compound concentrationrequired to inhibit cell proliferation by 50% (GI50) was calculatedusing nonlinear regression analysis.

Western Blotting

MV4-11 cells were seeded at 8�105 cells/10 ml of medium. After 24hours cells were treated with eight different concentrations ofinhibitors for 1 hour. Protein concentration was determined by DCassay (BioRad, Hercules, CA), and a total of 20 mg protein wasseparated by SDS-PAGE and transferred onto nitrocellulose mem-branes. Primary antibodies against Ser209 phosphorylated eIF4E(p-eIF4ESer209), eIF4E, or b-actin were used (Cell Signaling Technol-ogies, Danvers, MA). A secondary anti-rabbit immunoglobulin G (IgG)horseradish peroxidase-conjugated antibody was used (Dako, Glo-strap, Denmark) to develop the Western blot with enhanced chem-iluminescence reagents (GE Life Sciences, Piscataway, NJ).

ResultsIdentification of Dual Mnk1/2 and Mono-specific Mnk2Inhibitors

The initial step to establish the compound profiling work-flow (Scheme 1) required developing nonradioactive kinaseassays based on the IMAP technology. To achieve this, a 13-amino acid peptide, i.e., DTATKSGS209TTKNR (eIF4E202-214),

Scheme 1. Screening cascade of Mnk inhibitor profiling. All compoundswere tested with the IMAP assays at 1 and 10 mM concentrations.Compounds that showed Mnk1 or Mnk2 inhibition .25% at 1 mM and/or.90% inhibition at 10 mMwere further evaluated with the ADP-Glo assayto determine their IC50 values. Dual and Mnk2-specific inhibitors with anIC50 value ,2 mM were tested for substrate competition with increasingATP and eIF4E-derived peptide concentrations using the ADP-Glo assay.Competition effects were indicated as positive when the IC50 valuesincreased by more than fivefold at elevated ATP or eIF4E peptideconcentrations (see IC50 ratios in Table 2).

Allosteric Inhibition of Mnks 937

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

which contains the Mnk1/2 phosphorylating residue Ser209and is a fragment based on the C-terminal sequence of humaneIF4E, was custom synthesized. The peptide was N-terminally linked to the fluorescent FRET acceptor TAMRA,leading to the IMAP kinase substrate TAMRA-eIF4E202-214

(Fig. 1A). The fluorescently labeled peptide is phosphorylatedby the Mnks, and the IMAP binding reagent stops the kinasereaction and binds the phosphorylated substrate throughhigh-affinity interaction of trivalent metal-containing (MIII)nanoparticles. The phosphorylated and bound substrate canbe detected by TR-FRET between the Tb-donor and theTAMRA acceptor on the peptide. Optimization of assayconditions started with incubation of the peptide substrateTAMRA-eIF4E202-214 at 5 mMwith different amounts of Mnk1orMnk2 in a standard kinase reaction using 100mMATP (Fig.1B). To obtain a sufficient signal-to-background ratio (S/B)and to minimize enzyme consumption, 100 ng of Mnk1 (S/B52.3) and 50 ng of Mnk2 (S/B5 2.5) were used in all subsequentexperiments. The second step for optimization was to test theTAMRA-eIF4E202-214 substrate at three concentrations, andsufficient S/B ratios were detected at 5mM(S/B5 2.4 forMnk1

and S/B 5 2.6 for Mnk2) (Fig. 1C). Assay conditions werefurther optimized by testing ATP at various concentrations inthe presence of Mnk1 or Mnk2, and increasing signals weredetected with both kinases to approximately 200 mM ATP,followed by a rapid decrease of the signal (Fig. 1D). The firstfive data points were used to determine suitable ATP concen-trations for the primary compound profiling. Iterative fittingto theMichaelis-Mentenmodel was carried out with the Prismsoftware, and we estimated that half-maximal signals will bereached at 165 and 80 mM for Mnk1 and Mnk2, respectively.These ATP concentrations were applied to perform dose-response experiments with two known inhibitor compoundsCGP57380 and cercosporamide using 100 ng of Mnk1 or 50 ngof Mnk2 and 5 mMTAMRA-peptide substrate. CGP57380 andcercosporamide showed an IC50 value of 1.51 and 0.67mMwithMnk1. The corresponding values for Mnk2 were 1.87 and 0.16mM (Fig. 1E).A collection of 66 compounds was generated for pharmaco-

logical target validation of Mnk1 and Mnk2. 29 compounds(series 1, Supplemental Table 1) were derived from patentedN-phenylthieno[2,3-d]pyrimidin-4-amines (WO2010/023181)

Fig. 1. Optimization of the IMAP assay conditions. (A) IMAP assay principle (adapted from Molecular Devices, LLC; www.moleculardevices.com). TheTAMRA labeled peptide substrate is depicted as a chain of circles with residue Ser209marked in black. TheMnk-phosphorylated peptide will specificallybind to trivalent metal ions of the IMAP binding reagent. This will bring the TAMRA acceptor into close proximity of the Tb-donor (Tb-sensitizercomplex). (B) Different quantities of recombinant Mnk1 or Mnk2 were incubated with 100 mMATP and 5 mMTAMRA-eIF4E202-214 to determine optimalassay conditions. Control reactions were run without the substrate. (C) Determination of TAMRA-eIF4E202-214 substrate concentration in the IMAPassay. 100 ng of Mnk1 or 50 ng of Mnk2 were incubated with 100 mM ATP and TAMRA-eIF4E202-214 peptide substrate at different concentrations.Reactions carried out without kinase acted as a control. (D) ATP titration forMnks in the IMAP assay. One hundred nanograms ofMnk1 or 50 ng ofMnk2was incubated with 5 mM TAMRA-eIF4E202-214 substrate at different ATP concentrations. Control reactions without Mnk1 or Mnk2 were performed atthe same time and used for background correction. (E) The control inhibitors CGP57380 and cercosporamidewere tested in the IMAP assay using optimalparameters obtained from (B), (C), and (D) in 20 ml assay volume. Dose-response curves were calculated by applying a four-parameter logistic nonlinearregression model.

938 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

(Teo et al., 2015a,b), 31 compounds (series 2, SupplementalTable 1) were developed based on our privileged structuresof 5-(2-(phenylamino)pyrimidin-4-yl)thiazol-2(3H)-one (Diabet al., 2014b), and 6 compounds were the derivatives of N-phenyl-1H-pyrazolo[3,4-d]pyrimidin-3-amine (series 3, Sup-plemental Table 1). All compounds were tested with the IMAPassays at two concentrations (1 and 10 mM) in duplicate, andthe averages of relative activities were calculated for bothconcentrations (Fig. 2, A andB). Allover, 31 compounds causeda reduction of Mnk1 and/or Mnk2 kinase activity to less than75% at 1 mM or 10% at 10 mM. Under these criteria, 18 activeswere considered asMnk2 specific, 11 were dual inhibitors, andonly 2 were identified as Mnk1 specific (Scheme 1); the lattershowed only weak efficacy with residual activities .65% and.70% at 10 and 1 mM, respectively. Assay robustness wasroutinely checked by monitoring the correlation of duplicatedata points (Fig. 2, C and D) and by calculating Z9-factors fromfour positive and negative controls, which were included in allexperiments (Fig. 2E). The calculated Z9-factors were all

larger than 0.5, which indicated both assays to be sufficientlyrobust.

Mnk1 and Mnk2 Show Different Kinetic Properties andLigand-binding Profiles

The IMAP assays can be run at relatively low cost and arevery robust; however, they are not suitable for kineticcharacterization of the interactions between kinases andligands. This is due to the limited compatibility with ATPconcentrations (Fig. 1D). Therefore, we established the ADP-Glo assay, which measures ATP depletion during the kinasereaction (Fig. 3A). With this assay format, the kinase reactionis conducted with Mnk and eIF4E202-214 peptide and stoppedafter the optimized incubation time. Addition of the ADP-Gloreagent stops the reaction and drives the enzymatic removal ofthe leftover ATP. In the next step, the kinase detectionreagent is used to enzymatically convert ADP back intoATP, which then is used to drive a luciferase reaction. The

Fig. 2. Compound profiling with theIMAP assay. A collection of 66 compoundswas analyzed for their inhibitory activi-ties against Mnk1 (A) and Mnk2 (B) usingthe IMAP assays. The compound concen-trations used were 1 and 10 mM induplicate, and residual kinase activitieswere calculated as percentages relative tonegative and positive controls (n = 4). Thedata are represented as means of tworeplicates. The lines indicate the cut-offvalues of 75% and 10% residual kinaseactivity at 1 and 10 mM inhibitor concen-trations, respectively. To assess assayrobustness, duplicate values were plottedagainst each other and linear regressionwas carried out (C and D). The calculatedR2 values were close to 0.9, indicatinga good correlation between two duplicatesand validating the sufficient robustness ofthe assays. The data were collected from14 independent experiments for bothMnk1 and Mnk2. All experiments in-cluded four positive and four negativecontrols that were used to calculate theZ9-factors, which were .0.5 in all cases,indicating the sufficient robustness of theassays.

Allosteric Inhibition of Mnks 939

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

Fig. 3. ADP-Glo assay validation with Mnks. (A) ADP-Glo assay principle (adapted from Promega Corporation). ATP is converted into ADP whenexposed to Mnk and eIF4E202-214. The ADP-Glo reagent is used to stop the reaction and enzymatically remove remaining ATP. Subsequently, the kinasedetection reagent is applied to convert ADP into ATP, which then will drive a luciferase reaction turning over beetle luciferin to generate light. (B)Different amounts of Mnk1 or Mnk2 as indicated in (B) were used in kinase reactions with fixed substrate concentrations of 100 mMATP and 100 mM ofeIF4E202-214 peptide substrate. The obtained luminescence signals were compared with control reactions without the peptide substrate. The S/B ratioswith 100 ng Mnk1 and 30 ng Mnk2 were 6.3 and 3.4, respectively. For all further experiments 100 ng Mnk1 and 30 ng Mnk2 were applied. (C) The VmaxandKm values were determined for the eIF4E202-214 peptide substrate. The ATP concentration was adjusted to 2mMand 100 ngMnk1 or 30 ngMnk2wasused in all reactions that were tested at different eIF4E202-214 peptide substrate concentrations. All data were baseline corrected, with signals obtainedfrom kinase control reactions (without peptide substrate) before curve fitting using the Michaelis-Menten model. (D) The determination of Vmax and Kmvalues for ATP was carried out under identical conditions as described for (C), with the difference that the eIF4E202-214 peptide substrate concentrationwas fixed at 1.2 mM while different ATP concentrations were used. Control reactions without kinase were performed at the same time and used forbackground correction before fitting with the Michaelis-Menten model. (E) The control inhibitors CGP57380 and cercosporamide were used for finalvalidation of the ADP-Glo assays using above parameters in a final assay volume of 15 ml. The ATP and eIF4E202-214 peptide substrate concentrationswere adjusted to 0.5�Km andKm forMnk1 andMnk2, respectively. Dose-response curves were calculated by iterative fitting to a four-parameter logisticnonlinear regression model. For all further experiments 0.5�Km values for ATP and peptide were used for Mnk1 andKm values were used for Mnk2. (F)

940 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

luminescence signal is directly proportional to the amount ofADP generated during the kinase reaction. The assays cangenerally be used with ATP concentrations up to 6 mM andtherefore are suitable for determination of Km and Vmax

values. Additionally, the assays are appropriate to distinguishbetween ATP competitive and non-ATP-competitive inhibi-tors by varying concentrations of ATP in the kinase reaction.In the first step of the assay development, we exploredsuitable Mnk concentrations. Various amounts of activatedrecombinantMnk1 orMnk2 were incubated with 100 mMATPand 100 mMeIF4E202-214 substrate. The kinase reactions wereperformed without the substrate as controls. Sufficientlyrobust signals were achieved with 100 ng Mnk1 and 30 ngMnk2, which resulted in S/B ratios of 6.2 and 3.4, respectively(Fig. 3B). These amounts of enzyme were applied for allfurther experiments. In the next step, Km and Vmax valueswere determined for the eIF4E202-214 substrate (Table 1). Forthis purpose, the ATP concentration was fixed at 2 mM andtested against eIF4E202-214 substrate at a range of concen-trations (Fig. 3C). Michaelis-Menten curve fit with baselinecorrected data (obtained from the kinase reaction withouteIF4E202-214) revealed a Km value of 1.272 mM, and thedetermined Vmax value was used to calculate the maximalspecific activity (eIF4E-derived peptide turn over per minute)of 2172 nmol/min/mg for Mnk1. The respective values forMnk2 were 608.5 mM for Km and 1846 nmol/min/mg for thespecific activity. Subsequently we determined Km and Vmax

values for ATP using 100 ng Mnk1 or 30 ng Mnk2 with theeIF4E-derived peptide substrate at a fixed concentration of 1.2mM (Fig. 3D). Control reactions were run without kinase andused for background correction. Michaelis-Menten curve fitresulted in a Km value of 2.094 mM and a specific activity of1852 nmol/min/mg for Mnk1 under Vmax conditions. Thecorresponding values for Mnk2 were significantly lower with215.8 mM for Km and 1075 nmol/min/mg for the specificactivity (Table 1). Dose-response testing with enzyme inhib-itors is often carried out using substrate concentrations thatcorrespond to theKm value.Mnk1 showed very highKm valuesfor ATP and the peptide substrate; therefore, we reduced ATPand peptide concentrations correspondent to 50% of the Km

values to cut consumable costs. The S/B ratio under theseconditions was 14.2 and seemed sufficient for dose-responseexperiments (data not shown). For all further tests, the ATPand peptide concentrations were adjusted to 50% of the Km

values for Mnk1 and Km values were used for Mnk2. Finally,we investigated the dose-response of the control inhibitorsCGP57380 and cercosporamide with the optimized ADP-Gloassay on both Mnks by using the aforementioned parametersin a final assay volume of 15 ml. IC50 values were calculated byapplying a four-parameter logistic nonlinear regressionmodel. CGP57380 and cercosporamide inhibited Mnk1 withKi values of 1.01 mM and 0.507 mM, respectively. Thecorresponding values against Mnk2 were 0.877 mM and0.079 mM, respectively (Fig. 3E and Table 1).All compounds causing residual kinase activities of,75% at

1 mM and/or of ,10% at 10 mM, as determined in the primaryIMAP assay (Fig. 2, A and B), were selected as hits and weresubjected to further analysis of enzyme kinetics using the

ADP-Glo kinase assay. Dose-response experiments wereperformed to determine IC50 values (Table 2). At this stage,particular emphasis was put on the identification of dual andMnk subtype-specific inhibitors. Seven compounds (MNKI-4,-12, -28, -67, -85, -4–61, and -5–17) were identified as Mnk2specific inhibitors with IC50 values being at least 10 timeslower for Mnk2 than for Mnk1; 13 compounds (MNKI-5, -6, -7,-8, -15, -19, -37, -57, -83, -2–22, -2–25, -2–30, and -7–50) wereconsidered as dual inhibitors, whereas no Mnk1-specificinhibitors were identified (Fig. 3F and Scheme 1).

Identification of Dual Mnk1/2 Inhibitors with Type IIIMechanism

The main objective of this work was to select specificchemical probes that would be useful for the future pharma-cological validation of Mnks as anticancer drug targets. Toachieve this, it is imperative to identify compounds with highspecificity. Type III inhibitors are believed to have a tendencyto achieve higher specificity due to the fact that the stronglyconserved ATP binding pocket is not involved in the bindingmodalities. To achieve our goal, a total of 20 actives wereinvestigated using the ADP-Glo assay for competing effects atelevated ATP concentrations. All compounds were tested at 10concentrations ranging from 0.7 nM to 15 mM with 100 ngMnk1 using 150, 2000 or 6000 mM ATP. In parallel, the samedose range was tested with 30 ng Mnk2 using 50, 200, or2000 mM ATP. The eIF4E202-214 peptide substrate concentra-tions were fixed for Mnk1 and Mnk2 at 1.2 and 0.6 mM,respectively. The IC50 values were systematically calculated forall conditions, and the ratios between the highest and lowestATP concentrations are given in Table 2 (IC50 ratio). Sub-sequently, a similar set of experiments were conducted toinvestigate competition effects of elevated eIF4E202-214 con-centrations. The peptide substrate was tested at 150, 600, and3000 mM with an ATP concentration fixed at 2 mM for Mnk1and200mMforMnk2. The IC50 valueswere routinely calculated,and competing effects were assessed by the ratios obtained atthe highest substrate concentration versus the one at thelowest substrate concentration (Table 2). We arbitrarily set aratio .5 as a significant competing effect. With this criterion,we proposed different inhibitory mechanisms as outlined below:Non-substrate- and non-ATP-competitive Inhibitors

(Mode 2/2). These compounds did not show significantchange in the IC50 values when ATP or the eIF4E202-214

substrate was used at increased concentrations. Compoundswith this type of inhibitory properties are depicted with mode2/2 in Table 2.ATP-competitive Inhibitors (Mode 1/2). These com-

pounds showed significant increase in the IC50 values atelevated ATP concentrations but showed no or at the mostmarginally changed IC50 values with increased eIF4E202-214

concentrations.Dual Substrate- and ATP-competitive Inhibitors

(Mode 1/1). IC50 values were affected by increased ATPand peptide substrate levels. Only one compound (MNKI-5-17)showed this property and was active only against Mnk2.Substrate-competitive Inhibitors (Mode 2/1). Only

one compound (MNKI-2-22) showed a moderate competitive

Dose-response experiments were used to calculate IC50 values, which led to the identification of 13 nonspecific Mnk1/2 inhibitors and 7 Mnk2 specificinhibitors (formore details, see Table 2). The latter were considered absoluteMnk2 selective inhibitors with the calculated ratio of IC50Mnk1/IC50Mnk2. 10(note that for compounds where no IC50 fit could be achieved, the value was set to = 15 mM). No Mnk1 specific inhibitors were found.

Allosteric Inhibition of Mnks 941

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

effect with eIF4E202-214 peptide on Mnk2 (Table 2). Surpris-ingly, when tested on Mnk1, the same compound showeda moderate competition effect with ATP but no effect witheIF4E202-214.A selection of compounds was retested with the same

conditions as outlined above, using an extended range ofATP concentrations (Fig. 4, A–C), whereas the eIF4E202-214

substrate concentrations were fixed at Km. Data from individ-ual dose-response curves were analyzed using Prism 6 Graph-pad software and plotted as a function of log transformedinhibitor concentration. The representative examples areshown in Figs. 4, A–C. A similar set of experiments wascarried out to evaluate effects of the eIF4E202-214 peptide,where the ATP concentrations were fixed at Km value (Fig. 4,D–F).The data were further used to calculate apparent Km and

Vmax values at various concentrations of inhibitors. A repre-sentative example for mode2/2 on Mnk2 is given in Fig. 5, Aand D. Increasing compound concentrations led to lower Vmax

values for ATP and eIF4E202-214, whereas the Km valuesremained unchanged, as it would be expected for a type III

noncompetitive inhibitory mechanism. Additional examplesare given for modes1/2 and1/1 in Fig. 5, B and E and C andF, respectively. Mode 1/2 showed increased Km values forATP combined with unchanged Vmax values, which is anexpected characteristic of competitive enzyme inhibition(Gumireddy et al., 2005). Finally, for mode 1/1, elevated Km

values were calculated with increased ATP and eIF4E202-214

concentrations. No major changes of the Vmax values wereobserved.

Cytotoxic Effects on MV4-11 Acute Leukemia Cells

Antiproliferative activity of the inhibitors was evaluated inMV4-11 cells using the resazurin assay. Cells were exposed toeach compound for 72 hours, and the values of concentrationthat causes 50% growth inhibition (GI50) were summarized inTable 2. All compounds with a GI50 value ,15 mM werefurther tested for inhibitory activity against CDK2A andCDK9T1; if the resulting residual kinase activities were,60% at 10 mM, a Ki value determination was carried out.Finally, all compounds with Ki values,15 mM for either CDK

TABLE 2Summary of Mnk inhibitory activity, binding mode and cellular antiproliferative activity of compoundsData shown are the mean values derived from at least two replicates.

Compound IC50 (nM)IC50 ratio

MV4-11 cellsc

Mnk1 Mnk2

Code Series Mnk1 Mnk2 ATP(6 mM/0.15 mM)

eIF4E(3 mM/0.15 mM) Mode ATP

(2 mM/0.05 mM)eIF4E

(3 mM/0.15 mM) Mode GI50 (mM) S.D.

MNKI-6 1 47 20 2.2 1.9 2/2 1.9 1.1 2/2 0.58 0.02MNKI-7 1 83 44 1.5 1.2 2/2 1.0 1.0 2/2 0.83 0.03MNKI-5 1 147 50 1.8 1.0 2/2 1.1 0.9 2/2 27.77 3.96MNKI-57 1 159 36 2.4 1.7 2/2 4.1 1.5 2/2 6.00 0.79MNKI-8 1 164 52 1.8 1.5 2/2 4.1 0.9 2/2 1.15 0.57MNKI-19 1 279 136 1.7 1.0 2/2 1.6 1.3 2/2 7.30 0.75MNKI-83 1 480 328 2.3 2.8 2/2 2.4 0.8 2/2 15.30 2.62MNKI-67b 1 1400 36 2.2 1.4 2/2 1.5 1.0 2/2 8.84 1.58MNKI-37 1 1530 219 1.9 1.1 2/2 1.1 0.8 2/2 6.08 1.99MNKI-15 1 2850 328 14.5 1.5 +/2 6.4 2.0 +/2 78.26 2.34MNKI-2-22 1 3090 480 7.2 1.9 +/2 3.5 7.6 2/+ 59.58 3.48MNKI-2-30 1 3225 1810 10.2 1.4 +/2 15.2 1.2 +/2 88.99 4.11MNKI-2-25 1 3495 678 50.2 1.8 +/2 37.8 0.9 +/2 13.96 2.21MNKI-28b 1 7230 704 N.T. N.T. 11.6 1.1 +/2 15.18 1.91MNKI-85b 1 . 10,000 62 N.T. N.T. 15.2 1.2 +/2 7.44 0.77MNKI-4b 1 . 10,000 452 N.T. N.T. 12.9 1.0 +/2 8.88 0.54MNKI-12b 1 . 10,000 1,000 N.T. N.T. 18.6 2.3 +/2 89.39 9.58MNKI-7-50 3 4800 1540 N.T. N.T. 35.7 1.4 +/2 N.A.a

MNKI-5-17b 2 . 10,000 618 N.T. N.T. 76.6 37.7 +/+ N.A.a

MNKI-4-61b 2 . 10,000 1230 N.T. N.T. 16.4 1.3 +/2 N.A.a

N.A., not active; N.T., not tested; Series, compound series; eIF4E, eIF4E202-214; Mode, ATP/eIF4E202-214 competition. Data shown are the mean valuesderived from at least two replicates.aKi value ,15 mM on CDK2A or CDK9T1.bMono-specific Mnk2 inhibitor.cBy 72 hour resazurin assay.

TABLE 1Kinetic constants for Mnks

Kinase Mnk1 Mnk2

Substrate ATP eIF4E202-214 ATP eIF4E202-214

Km (mM) 2094 1272 215.8 608.5Vmax (nmol/min) 0.18523 0.21718 0.03225 0.05537Specific activity (nmol/min/mg) 1852 2172 1075 1846Compound CGP57380 Cercosp. CGP57380 Cercosp.Ki value (mM) 1.010 0.507 0.877 0.079

942 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

were excluded from further analysis (Table 2). The remaining17 compounds were analyzed for the correlation between IC50

values on Mnks and GI50 values on MV4-11 cells (Fig. 6).Mnk1 inhibitors with an IC50 value .10 mM were excluded,resulting in a total of 14 compounds. The discrepancy betweenbiochemical and cellular potencies is expected, because cellu-lar activity may be influenced by physicochemical properties

of compounds, which may affect permeability across the cellmembrane, intracellular degradation, and so forth. Neverthe-less a clear correlation was observed for both Mnk1 and Mnk2inhibitors. The Pearson correlation was calculated to 0.4006and 0.7276 for Mnk1 andMnk2, respectively. The correspond-ing one-tailed P values were 0.078 and 0.0005. The cellularMnk inhibition was further confirmed with Western blotting

Fig. 4. Competition of inhibitor activity with increased ATP and eIF4E peptide substrate concentration on Mnk2. 30 ng of Mnk2 was mixed with 10different concentrations of Mnk inhibitors, and the ADP-Glo assay was performed in the presence of increasing concentrations of ATP (A, B, and C) andwith increasing concentrations of eIF4E202-214 peptide substrate (D, E, and F). All experiments were run in duplicate, and the values from individualsamples were analyzed using the Prism 6 Graphpad software. The three compounds were selected as examples for the different inhibitionmodes (MNKI-6:2/2; MNKI-4-61: +/2; MNKI-5-17: +/+). (G) Compound structures.

Allosteric Inhibition of Mnks 943

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

analysis of MV4-11 cells after treatment with two Mnkinhibitors i.e., MNKI-7 or MNKI-19 for 1 hour. A dose-dependent downregulation of phosphorylation at Ser209 ofeIF4E was detected (Fig. 6, C and D).

DiscussionTo identify Mnk inhibitors, we established a screening

cascade (Scheme 1). The IMAP Mnk1/2 assays enabled us torapidly identify primary hits whereas the ADP-Glo assayfacilitated the dose-response analysis. The sequential use of

these mechanistically different assay technologies (identifica-tion of phosphorylated peptide versus ATP depletion) ex-cluded any false positives.For the IMAP assay, we successfully used a TAMRA-labeled

eIF4E-derived peptide (eIF4E202-214) to detect kinase activi-ties of Mnk1 and Mnk2 in the presence of tested compounds.However the assay was incompatible with high ATP concen-trations above 200 mM (Fig. 1D). This effect was most likelydue to the displacement of the phosphorylated substrate fromthe IMAP nanobeads by ATP phosphate groups, leading toa reduced FRET signal. This inherent property of the IMAP

Fig. 5. The same compounds as shown in Fig. 4 were used to determine the apparent Km and Vmax values at different compound concentrations. TheMichaelis-Menten saturation curves are given for a range of ATP (A–C) and a range of eIF4E202-214 concentrations (D–F). Selected compounds were usedat the concentrations as indicated in the figure. The calculated kinetic parameters are given beneath the shown graphs.

944 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

assay limits its suitability for experiments requiring high ATPconcentrations (e.g., ATP competition testing). However, atrelatively low ATP concentrations, the assay was shown to bevery robust and cost effective for initial compound profiling.Finally, the IMAP assays were validated with the knownMnkinhibitors CGP57380 and cercosporamide (Fig. 1E) and theobtained inhibitory activities against Mnk1/2 were consistentwith the published data (Buxade et al., 2005; Bain et al., 2007;Konicek et al., 2011). One advantage of using low ATPconcentrations in the assays was to identify weak ATP-competitive and type II inhibitors that would be lost at higherATP concentrations.Three different chemical series of compoundswere profiled at

the concentrations of 1 and 10 mM by IMAP assays. Figure 2, Aand B, shows the averages of the measured residual kinaseactivities for Mnk1 and Mnk2, respectively. Figure 2, C and D,are scatterplots of duplicate data points with a trend line. Thecalculated R2 correlations of 0.897 and 0.879 indicated suffi-cient robustness of the assays, and this was further confirmedby the routinely calculated Z9-factors that were always .0.5.Among the total of 31 primary hits, only 2 were Mnk1 specific,11 were dual inhibitors, and 18 were Mnk2 specific.In contrast to the IMAP assay, the ADP-Glo assay can be

run at much higher ATP concentrations and thus is suited todetermine kinetic parameters such asKm andVmax values.Weestablished the ADP-Glo assay for Mnk1 and Mnk2 as out-lined in Fig. 3. Surprisingly, theKm value for ATPwas at 2.094mM for Mnk1 almost 10 times higher than for Mnk2 (Fig. 3Dand Table 1). This strongly indicated a lower affinity of ATPfor Mnk1 than Mnk2, which is most likely due to structuraldifferences between Mnk1 and Mnk2. The longer C terminusof Mnk1 is known to affect its basal kinase activity and to

repress phosphorylation of the activation loop (Parra et al.,2005; Goto et al., 2009). Dose-response curves with kinaseinhibitors are routinely run with the substrate concentrationsadjusted to Km. Because of the fact that Km values were veryhigh for Mnk1, we decided to reduce the substrate concen-trations to 50% of the Km values for ATP and the peptidesubstrate. For Mnk2, the concentrations were adjusted to Km.The Ki values for CGP57380 and cercosporamide (Fig. 3E andTable 1) were determined as a final validation step, and theobtained results were comparable to literature values (Buxadeet al., 2005; Bain et al., 2007; Konicek et al., 2011).All compounds causing residual kinase activities of,75% at

1 mM or of,10% at 10 mM in the IMAP assay (Fig. 2, A and B)were subjected to systematic determination of IC50 valueswith the ADP-Glo assay. In total, 20 inhibitors were identifiedwith IC50 values of ,2 mM. Seven inhibitors were Mnk2specific with a ratio of IC50Mnk1/IC50Mnk2 .10, and 13 inhib-itors were dual inhibitors (Fig. 3F and Table 2). Out of the 20compounds, 17 belonged to series 1, whereas only 2 and 1 werefrom series 2 and 3, respectively. Interestingly, we did notidentify any Mnk1 subtype-specific inhibitors, further indi-cating significant structural differences between Mnk1 andMnk2. Analysis of crystal structures of the Mnk1 and Mnk2kinase domains (Jauch et al., 2005; 2006) suggests that Mnk2has a larger ATP binding pocket than Mnk1. Our Mnk2-specific inhibitors provide a new opportunity to dissect biologicfunctions of Mnk1 and Mnk2. It is currently not knownwhether the individual or combined inhibition of Mnk1/2would be efficacious in induction of cancer cell apoptosis.It is generally believed that non-ATP-competitive ligandsmay

provide a selectivity advantage over their ATP-competitivecounterparts. We tested our compounds for ATP- and/or

Fig. 6. Cytotoxic effects correlate with Mnk enzymeinhibition. MV4-11 cells were exposed to compounds for72 hours. The resazurin assay was used to test cellviability in dose-response experiments and GI50 valueswere calculated. Compounds with activities againstCDK2A or CDK9T1 (Ki ,15) were excluded. (A) TheIC50 values for Mnk1 were plotted against the GI50values. Compounds with IC50 values .10 mM againstMnk1 were excluded, and the data are given for 14compounds. (B) A similar plot was generated with theMnk2 data, with 17 compounds in total. The trend linesare shown, and the Pearson correlations were deter-mined for Mnk1 (r = 0.4006; n = 14; one-tailed P = 0.078)and Mnk2 (r = 0.7276; n = 17; one-tailed P = 0.0005).Western blot analysis was conducted with MV4-11 cellsafter 1 hour exposure to MNKI-7 (C) or MNKI-19 (D) atthe concentrations indicated.

Allosteric Inhibition of Mnks 945

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

eIF4E202-214 competition before investigating their toxicity oncancer cell lines. IC50 ratios of dose-response experiments athigh and low ATP or eIF4E202-214 concentrations are given inTable 2. We identified two distinct groups: nine compounds didnot showany competitionwith either of the substrates (mode2/2)and nine compoundswere only competitive with ATP (mode1/2).In addition, two compounds were competitive with ATP andeIF4E202-214 (mode1/1). Interestingly, MNKI-2-22 showedthe inhibitory activities in an ATP competitive manner inMnk1 but is substrate dependent against Mnk2 (Scheme 1 and

Table 2). All compounds with mode 2/2 belonged to series 1,showed IC50 values ,1.6 mM, and were associated with dualinhibition of Mnk1 andMnk2 (except MNKI-67). These findingsindicate that these inhibitors act via an allosteric type IIImechanism, and we thus conclude that an allosteric bindingsite is structurally conserved between Mnk1 and Mnk2. Incontrast, all compoundswithmode1/2had IC50 values.2.8mMon Mnk1 and 6 of them were Mnk2 specific. Most intrigu-ingly, very subtle structural changes resulted in the switchbetween activity modes. For example, MNKI-4-61 of series 2

Fig. 7. (A) Structures of some selected com-pounds. Very subtle changes of MNKI-5 inducea switch frommode2/2 tomode +/2 (MNKI-15)or dramatically reduce potency by more thanfactor 10 against Mnk1 (MNKI-37). The efficacyof MNKI-37 against Mnk1 was significantlyimproved throughmodification of the acetamidegroup as shown for MNKI-83 (B) Hypotheticalworkingmodel for compound binding at reactivesite. The Mnks switch from the inactive DFD-out form (I) to the active DFD-in form (II) uponphosphorylation of the activation segment andsubsequently bind the ATP substrate. (III) Thecartoon depictsMNKI-15 as a type II kinase inhib-itor binding partly inside and partly adjacent tothe ATP binding pocket. The black diamondrepresents the common core scaffold betweenMNKI-15 and -5 that occupies the postulated al-losteric binding site. (IV) Hypothetical bindingmodalities of MNKI-5: The larger side chain isforced through steric hindering (dashed struc-ture) to adopt an alternative conformation withnon-ATP-competitive binding mode.

946 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

turned frommode1/2 tomode1/1when 2,4-dimethoxyanilinewas replaced with 2-(difluoromethoxy)aniline of MNKI-5-17 (Fig. 4). In the case of MNKI-5, the removal of the 1,3-difluoropropan-2-ol group resulted in a change from mode 2/2to mode 1/2 of MNKI-15, and the potency against Mnk1 wasalmost reduced by a factor of 20 (Fig. 7A, Table 2). In contrast,when only the fluorine atoms were removed from the 1,3-difluoropropan-2-ol group (MNKI-37) the mode 2/2 was main-tained but reduction of Mnk1 activity by almost 10-fold wasobserved. Clearly the 1,3-difluoropropan-2-ol group preventscompetition with ATP. Finally, when the acetamide on thedihydrothiophene ring of MNKI-37 was replaced with N-(3-methoxypropyl)-acetamide (MNKI-83), Mnk1 inhibitory activitywas enhanced. There is currently no structural data available foractivated Mnk1 (Kumarasiri et al., 2015) and these inhibitorymodes remain tobeelucidated.Thevery subtle structural changesresulting in the different inhibitory modes led us to believe thatthebinding site on thekinases is very close to the targeting sites ofATP and the peptide substrate. We propose the hypotheticalmodels as shown in Fig. 7B. The inactive DFD-out conformation(I) switches into the activeDFD-in (II) through phosphorylation ofthe activation segment by extracellular signal-regulated kinase1/2 or p38 kinases. The activated kinase coordinates the ATPand the peptide substrate (eIF4E) in contact with catalyticallyimportant residues on the activation segment, the glycine-richloop and the catalytic loop. MNKI-5 is allosteric but shares thesame core scaffold (indicated as black diamond in diagrams IIIand IV) with MNKI-15. The latter may have significant bindingaffinities adjacent to and inside of the ATP binding pocket so toexhibit its type II mechanism (Fig. 7, B, III]. The loss of ATPcompetition ofMNKI-5maybe caused by its 1,3-difluoropropan-2-ol-mediated conformational changes so that it moves away frominteracting with the ATP binding site (Fig. 7, B, IV).We previously showed that inhibition of Mnks caused cell type-

specific cytoxicity (Diab et al., 2014b; Yu et al., 2015). Mostpronounced cytotoxic effects have been observed in Fms-liketyrosine kinase 3 overexpressing leukemia cells (Altman et al.,2013; Lim et al., 2013; Diab et al., 2014b; Teo et al., 2015a).Accordingly the cytotoxic assay in MV4-11 cells was included inthe screening cascade. To determine whether Mnk inhibition wasresponsible for the observed cytotoxic effects and to assesspotential and off-target effects, we first carried out additionalkinase testing against CDK2A and CDK9T1, because somederivatives of the chemical series have been shown to be CDK2/9modulators (Shao et al., 2013; Diab et al., 2014b). Three com-pounds were excluded from further analysis because of activitiesonCDKs.The studywith the remaining inhibitors showedpositivecorrelations (Pearson correlation r 5 0.4006 and r 5 0.7276 forMnk1 and Mnk2, respectively) between Mnk inhibitory activityand cytotoxic effects, confirming the Mnk-specific activity of thecompounds. Our Western blot analysis also demonstrated a dose-dependent reduction in eIF4E phosphorylation after 1 hourexposure to MNKI-7 or MNKI-19 (Fig. 6, C and D).Taken together, our investigation suggests significant struc-

tural differences in the ATP catalytic domain between Mnk1and Mnk2, leading to very different inhibitory activities withATP-competitive inhibitors. However, both enzymes sharea common structural feature of the allosteric binding site thatis located in close proximity of the ATP and peptide substratetargeting sites. These findings will be invaluable for the futuredesign, optimization, and development of highly specific Mnkinhibitors as therapeutic agents.

Authorship Contributions

Participated in research design: Basnet, Schmid, Albrecht, Wang.Conducted experiments: Basnet, Diab, Schmid, Yu, Yang, Gillam,

Teo, Li.Performed data analysis: Basnet, Albrecht, Wang.Wrote or contributed to the writing of the manuscript: Basnet, Diab,

Yu, Teo, Peat, Albrecht, Wang.

References

Altman JK, Szilard A, Konicek BW, Iversen PW, Kroczynska B, Glaser H, Sassano A,Vakana E, Graff JR, and Platanias LC (2013) Inhibition of Mnk kinase activity bycercosporamide and suppressive effects on acute myeloid leukemia precursors.Blood 121:3675–3681.

Bain J, Plater L, Elliott M, Shpiro N, Hastie CJ, McLauchlan H, Klevernic I, ArthurJSC, Alessi DR, and Cohen P (2007) The selectivity of protein kinase inhibitors:a further update. Biochem J 408:297–315.

Blanc J, Geney R, and Menet C (2013) Type II kinase inhibitors: an opportunity incancer for rational design. Anticancer Agents Med Chem 13:731–747.

Buxadé M, Parra JL, Rousseau S, Shpiro N, Marquez R, Morrice N, Bain J, Espel E,and Proud CG (2005) The Mnks are novel components in the control of TNF abiosynthesis and phosphorylate and regulate hnRNP A1. Immunity 23:177–189.

Diab S, Kumarasiri M, Yu M, Teo T, Proud C, Milne R, and Wang S (2014a) MAPkinase-interacting kinases–emerging targets against cancer. Chem Biol 21:441–452.

Diab S, Teo T, Kumarasiri M, Li P, Yu M, Lam F, Basnet SK, Sykes MJ, Albrecht H,and Milne R et al. (2014b) Discovery of 5-(2-(phenylamino)pyrimidin-4-yl)thiazol-2(3H)-one derivatives as potent Mnk2 inhibitors: synthesis, SAR analysis and bi-ological evaluation. ChemMedChem 9:962–972.

Eglen R and Reisine T (2011) Drug discovery and the human kinome: recent trends.Pharmacol Ther 130:144–156.

Furic L, Rong L, Larsson O, Koumakpayi IH, Yoshida K, Brueschke A, Petroulakis E,Robichaud N, Pollak M, and Gaboury LA et al. (2010) eIF4E phosphorylationpromotes tumorigenesis and is associated with prostate cancer progression. ProcNatl Acad Sci USA 107:14134–14139.

Goto S, Yao Z, and Proud CG (2009) The C-terminal domain of Mnk1a plays a dualrole in tightly regulating its activity. Biochem J 423:279–290.

Gumireddy K, Reddy MV, Cosenza SC, Boominathan R, Baker SJ, Papathi N, JiangJ, Holland J, and Reddy EP (2005) ON01910, a non-ATP-competitive small mole-cule inhibitor of Plk1, is a potent anticancer agent. Cancer Cell 7:275–286.

Hou J, Lam F, Proud C, and Wang S (2012) Targeting Mnks for cancer therapy.Oncotarget 3:118–131.

Jauch R, Cho MK, Jäkel S, Netter C, Schreiter K, Aicher B, Zweckstetter M, JäckleH, and Wahl MC (2006) Mitogen-activated protein kinases interacting kinases areautoinhibited by a reprogrammed activation segment. EMBO J 25:4020–4032.

Jauch R, Jäkel S, Netter C, Schreiter K, Aicher B, Jäckle H, and Wahl MC (2005)Crystal structures of the Mnk2 kinase domain reveal an inhibitory conformationand a zinc binding site. Structure 13:1559–1568.

Konicek BW, Stephens JR, McNulty AM, Robichaud N, Peery RB, Dumstorf CA,Dowless MS, Iversen PW, Parsons S, and Ellis KE et al. (2011) Therapeutic in-hibition of MAP kinase interacting kinase blocks eukaryotic initiation factor 4Ephosphorylation and suppresses outgrowth of experimental lung metastases.Cancer Res 71:1849–1857.

Kumarasiri M, Teo T, andWang S (2015) Dynamical insights of Mnk2 kinase activationby phosphorylation to facilitate inhibitor discovery. Future Med Chem 7:91–102.

Lim S, Saw TY, Zhang M, Janes MR, Nacro K, Hill J, Lim AQ, Chang CT, FrumanDA, and Rizzieri DA et al. (2013) Targeting of the MNK-eIF4E axis in blast crisischronic myeloid leukemia inhibits leukemia stem cell function. Proc Natl Acad SciUSA 110:E2298–E2307.

Liu Y and Gray NS (2006) Rational design of inhibitors that bind to inactive kinaseconformations. Nat Chem Biol 2:358–364.

Nagar B, BornmannWG, Pellicena P, Schindler T, Veach DR, Miller WT, Clarkson B,and Kuriyan J (2002) Crystal structures of the kinase domain of c-Abl in complexwith the small molecule inhibitors PD173955 and imatinib (STI-571). Cancer Res62:4236–4243.

Parra JL, Buxadé M, and Proud CG (2005) Features of the catalytic domains and Ctermini of the MAPK signal-integrating kinases Mnk1 and Mnk2 determine theirdiffering activities and regulatory properties. J Biol Chem 280:37623–37633.

Shao H, Shi S, Huang S, Hole AJ, Abbas AY, Baumli S, Liu X, Lam F, Foley DW,and Fischer PM et al. (2013) Substituted 4-(thiazol-5-yl)-2-(phenylamino)pyrimidinesare highly active CDK9 inhibitors: synthesis, X-ray crystal structures, structure-activity relationship, and anticancer activities. J Med Chem 56:640–659.

Sussman A, Huss K, Chio L-C, Heidler S, Shaw M, Ma D, Zhu G, Campbell RM, ParkT-S, and Kulanthaivel P et al. (2004) Discovery of cercosporamide, a known anti-fungal natural product, as a selective Pkc1 kinase inhibitor through high-throughput screening. Eukaryot Cell 3:932–943.

Teo T, Lam F, Yu M, Yang Y, Basnet SK, Albrecht H, Sykes MJ, and Wang S (2015a)Pharmacologic inhibition of Mnks in acute myeloid leukemia.Mol Pharmacol 88:380–389.

Teo T, Yu M, Yang Y, Gillam T, Lam F, Sykes MJ, and Wang S (2015b) Pharmaco-logic co-inhibition of Mnks and mTORC1 synergistically suppresses proliferationand perturbs cell cycle progression in blast crisis-chronic myeloid leukemia cells.Cancer Lett 357:612–623.

Tokarski JS, Newitt JA, Chang CYJ, Cheng JD, Wittekind M, Kiefer SE, Kish K, LeeFYF, Borzillerri R, and Lombardo LJ et al. (2006) The structure of Dasatinib(BMS-354825) bound to activated ABL kinase domain elucidates its inhibitoryactivity against imatinib-resistant ABL mutants. Cancer Res 66:5790–5797.

Ueda T, Sasaki M, Elia AJ, Chio IIC, Hamada K, Fukunaga R, and Mak TW (2010)Combined deficiency for MAP kinase-interacting kinase 1 and 2 (Mnk1 and Mnk2)delays tumor development. Proc Natl Acad Sci USA 107:13984–13990.

Allosteric Inhibition of Mnks 947

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from

Ueda T, Watanabe-Fukunaga R, Fukuyama H, Nagata S, and Fukunaga R (2004)Mnk2 and Mnk1 are essential for constitutive and inducible phosphorylation ofeukaryotic initiation factor 4E but not for cell growth or development.Mol Cell Biol24:6539–6549.

Vogtherr M, Saxena K, Hoelder S, Grimme S, Betz M, Schieborr U, Pescatore B, RobinM, Delarbre L, and Langer T et al. (2006) NMR characterization of kinase p38 dy-namics in free and ligand-bound forms. Angew Chem Int Ed Engl 45:993–997.

Wendel HG, Silva RLA, Malina A, Mills JR, Zhu H, Ueda T, Watanabe-Fukunaga R,Fukunaga R, Teruya-Feldstein J, and Pelletier J et al. (2007) Dissecting eIF4Eaction in tumorigenesis. Genes Dev 21:3232–3237.

Wodicka LM, Ciceri P, Davis MI, Hunt JP, Floyd M, Salerno S, Hua XH, Ford JM,Armstrong RC, and Zarrinkar PP et al. (2010) Activation state-dependent bindingof small molecule kinase inhibitors: structural insights from biochemistry. ChemBiol 17:1241–1249.

Yu M, Li P, K C Basnet S, Kumarasiri M, Diab S, Teo T, Albrecht H, and Wang S(2015) Discovery of 4-(dihydropyridinon-3-yl)amino-5-methylthieno[2,3-d]pyrimidine derivatives as potent Mnk inhibitors: synthesis, structure-

activity relationship analysis and biological evaluation. Eur J Med Chem95:116–126.

Zhang J-H, Chung TD, and Oldenburg KR (1999) A simple statistical parameter foruse in evaluation and validation of high throughput screening assays. J BiomolScreen 4:67–73.

Zhang J, Yang PL, and Gray NS (2009) Targeting cancer with small molecule kinaseinhibitors. Nat Rev Cancer 9:28–39.

Address correspondence to: Shudong Wang, Centre for Drug Discovery andDevelopment, Sansom Institute for Health Research, and School of Pharmacyand Medical Sciences, University of South Australia, Adelaide, SA 5001,Australia. E-mail: shudong.wang@unisa.edu.au or Hugo Albrecht, Centre forDrug Discovery and Development, Sansom Institute for Health Research, andSchool of Pharmacy and Medical Sciences, University of South Australia,Adelaide, SA 5001, Australia. E-mail: Hugo.Albrecht@unisa.edu.au

948 Basnet et al.

at ASPE

T Journals on M

arch 8, 2021m

olpharm.aspetjournals.org

Dow

nloaded from