Discovery of Dual Inhibitors for Wild Type and D816V Mutant of c-KIT

Article pubs.acs.org/jnp

Discovery of Dual Inhibitors for Wild Type and D816V Mutant of c‑KIT Kinase through Virtual and Biochemical Screening of Natural Products Hwangseo Park,*,† Soyoung Lee,‡ and Sungwoo Hong*,‡ †

Department of Bioscience and Biotechnology & Institute of Anticancer Medicine Development, Sejong University, 209 Neungdong-ro, Kwangjin-gu, Seoul 143-747, Korea ‡ Center for Catalytic Hydrocarbon Functionalizations, Institute for Basic Science (IBS) & Department of Chemistry, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, Korea ABSTRACT: Although stem cell factor receptor (c-KIT) kinase is responsible for various malignant human cancers, the presence of constitutively active gain-of-function mutants has made it difficult to discover new anticancer agents using c-KIT as the target protein. To identify the common inhibitors of wild-type c-KIT and the most abundant gain-of-function mutant (D816V), the virtual screening of natural products was performed for the two target proteins in parallel with the scoring function improved by implementing a sophisticated solvation free energy term. As a result, four common inhibitors of natural origin are found with biochemical potencies ranging from low micromolar to submicromolar levels. The results of extensive docking simulations show that although the naturalproduct inhibitors establish weaker hydrophobic interactions with the D816V mutant than with the wild type, they exhibit a little higher inhibitory activity for the former than the latter by strengthening the hydrogen-bond interactions to a sufficient extent. Of the four natural-product inhibitors, (Z)-6-hydroxy-2-(4-methoxybenzylidene)benzofuran-3(2H)-one (3) is anticipated to serve as a new molecular core for the structure−activity relationship studies to optimize the biochemical potencies because it exhibits good inhibitory activity against both the wild type and D816V mutant despite its low molecular weight (268.3 amu).

S

region.10 Furthermore, extensive molecular dynamics (MD) simulation studies indicated that the D816V mutant promotes the separation of the juxtamembrane region (JMR) from the kinase domain, which is the first step in the structural transition of c-KIT from the inactive to the active state.11 Besides the tumorigenic effects, the D816V mutant exhibits almost complete resistance to some anticancer drugs at clinically achievable doses.12,13 The small-molecule inhibitors equipotent to the wild-type and D816V mutant c-KIT are thus anticipated to have the potential to develop into new anticancer drugs. Since the approval of imatinib to treat c-KIT-positive human cancers, much effort has been directed toward the discovery of small-molecule c-KIT inhibitors. Products of this work include 7-azaindole,14,15 3,5-diamino-1,2,4-triazole,16 amidobenzisoxazoles,17 nilotinib,18 quinazoline−pyrazolourea hybrids,19 ellipticine,20 aryl pyridones,21 and 4-methyl-N-phenylisophthalamide22 as the structural cores. Besides the ATP-competitive inhibitors, several allosteric c-KIT inhibitors have also been discovered based on novel design methods.23 Computational investigations for c-KIT-inhibitor complexes have been actively pursued by means of MD simulations,20,24 free energy perturbation methods,25 and 3D pharmacophore modeling26

tem cell factor receptor (c-KIT) is a receptor tyrosine kinase that becomes activated due to the dimerization stimulated by binding of stem cell factor. In the activated form, c-KIT phosphorylates various intracellular substrates and thereby triggers the cell signaling pathways required for the proliferation, differentiation, and survival of cells.1,2 The deregulation of c-KIT kinase activity is therefore related to the pathogenesis of a variety of human cancers such as acute myeloid leukemia, malignant melanomas, gastrointestinal stromal tumors, small-cell lung carcinoma, and colorectal cancer.3,4 The impediment of c-KIT kinase activity with small-molecule inhibitors proved to be effective in the treatment of some cancers caused by c-KIT.5 This confirmed the usefulness of c-KIT as a target for the development of new anticancer medicines. Many c-KIT-dependent cancers have sustained mutations producing constitutive c-KIT kinase activity.6 Indeed, several mutations in the kinase domain of c-KIT can stabilize the activation loop (A-loop) in its extended conformation, which leads to the maintenance of the active site configuration in the active form.7,8 Among the abnormal c-KIT variants, the substitution of Val for Asp in residue 816 of the A-loop has appeared most frequently in human cancers.9 This gain-offunction mutant was shown to induce downstream oncogenic signaling in the absence of ligand binding in the receptor © XXXX American Chemical Society and American Society of Pharmacognosy

Received: October 8, 2015

A

DOI: 10.1021/acs.jnatprod.5b00851 J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products

Article

to establish rationales for the biochemical potency and specificity of the known inhibitors. However, our understanding of the mechanisms by which D816V mutant inhibitors work lags behind that of the wild-type c-KIT inhibitors despite the fact that simultaneous inhibition of the wild type and the mutant is required to cope efficiently with human cancers caused by c-KIT. Natural products have frequently served as the structural cores of small-molecule libraries constructed to acquire a desired pharmacological activity because they account for a large portion of approved drugs especially among antibiotic and anticancer drugs.27 Nonetheless, only a few natural-product cKIT inhibitors have been reported so far including rapamycin28 and ellipticine derivatives.29 The present study was designed to identify and biochemically evaluate new classes of natural products that are equipotent inhibitors of c-KIT wild type and D816V mutant using molecular-docking-based virtual screening. Virtual screening based on molecular docking has often been unsuccessful due to the imperfections in protein−ligand binding free energy function to score putative inhibitors.30 This is responsible for a poor correlation between the computational prediction and experimental measurement for biochemical potency. In recent years, a potential-based scoring function implementing a sophisticated molecular solvation free energy term was proposed and validated on the basis of the extended solventcontact model.31 The outperformance of this new scoring function was demonstrated in estimating the biochemical potencies of inhibitors of various target enzymes.32,33 We demonstrate that virtual screening with a modified scoring function is useful for enriching chemical libraries with natural products that have good inhibitory activity against c-KIT.

Figure 1. Flowchart for the identification of common inhibitors of the wild type and D816V mutant of c-KIT through the two parallel virtual and high-throughput screening processes.

■

RESULTS AND DISCUSSION A total of approximately 40 000 molecules were screened with molecular docking with respect to the wild type and D816V mutant of c-KIT to select 100 top-scoring natural products as virtual hits for each target protein. The two sets of virtual hits contained 37 natural products in common, as indicated in Figure 1, which were commercially available (InterBioScreen Ltd., Chernogolovka, Russia). These 37 putative common inhibitors were then tested for inhibitory activity against wildtype and D816V mutant c-KIT at 100 μM based on the highthroughput binding assays (KINOMEscan, Ambit Biosciences, San Diego, CA, USA).34 As a consequence of combining the two parallel virtual and high-throughput screening processes, four natural products were identified as dual inhibitors for the wild type and D816V mutant of c-KIT. All these naturalproduct inhibitors exhibited good potency for both target proteins with percent of control (POC) values lower than 20 and were therefore selected to measure IC50 values. The chemical structures and biochemical potencies of the four common inhibitors of the wild type and D816V mutant are shown in Figure 2 and Table 1, respectively. It is interesting to note that all four natural-product inhibitors are derived from plant sources. For example, 1 is a constituent of various plant organisms including Olearia paniculata (Akiraho), Eupatorium semiserratum de Candolle, Millingtonia hortensis (Bignoniaceae), and many other plants, while 2 originates from Scutellaria ramosissima (Lamiaceae). Compounds 3 and 4 are isolated from soja and Xanthoceras sorbifolia Bunge, respectively. It is a common structural feature for 1−4 that multiple hydrogenbonding groups are situated on the fused phenyl and six- or

Figure 2. Chemical structures of the four common natural-product inhibitors of the wild type and D816V mutant of c-KIT.

Table 1. POC and IC50 (in μM) Values of the Four Common Natural-Product Inhibitors against the Wild Type and D816V Mutant of c-KIT POC

IC50

inhibitor

wild type

D816V mutant

wild type

D816V mutant

1 2 3 4

0 1 0 12

3 0 15 15

1.5 7.8 15.6 21.9

0.6 0.5 1.4 15.7

five-membered rings in the vicinity of the other terminal phenyl ring. This implies that both hydrogen-bond and van der Waals interactions could serve as significant binding forces for the natural-product inhibitors to be stabilized in the ATP-binding sites of the wild type and D816V mutant. None of these natural products have been reported as a c-KIT inhibitor in the literature or patents. Furthermore, no additional biological activity was found for 1−4 in the two publicly accessible chemical databases, ChEMBL and PubChem. As can be seen in Table 1, compounds 1 and 2 exhibit low micromolar and B



Article

type to D816V mutant stems from the corresponding decrease of ΔGbgas values because each natural-product inhibitor has a unique ΔGsol value. This implies that the enzyme−inhibitor interactions would be stronger in the ATP-binding site of the D816V mutant than in that of the wild type. Judging from the even higher ΔGbgas values of 3 than those of 1, 2, or 4 with respect to both wild type and D816V mutant, the interactions between c-KIT and the inhibitors seem to get weaker in proportion to the molecular size of the inhibitors. Nonetheless, the ΔGbaq value of 3 becomes similar to those of the larger inhibitors because of its relatively high ΔGsol value. Consistent with the lowest inhibitory activity, 4 appears to have the highest ΔGbaq values for the wild type and D816V mutant. When its ΔGbgas, ΔGsol, and ΔGbaq values are compared to those of 1−3, it can be recognized immediately that the decrease in biochemical potency on going from 1−3 to 4 stems from the larger decrease in ΔGsol than ΔGbgas. Although the ΔGbgas value of 4 is lowered significantly by the presence of many −OH groups, the strengthening of enzyme−inhibitor interactions seems to be insufficient to overcome the increased desolvation cost. Thus, the results of ΔGbaq decomposition analyses confirm that the biochemical potency enhancement for c-KIT inhibitors can be achieved not only by strengthening the interactions in the ATP-binding site but also by simply reducing the desolvation cost of the inhibitor. Although the computational ΔGbaq values compare reasonably well with the IC50 data when either the receptor protein or the inhibitor is the same, it seems to be difficult to elucidate the relative activities of the four inhibitors if both are different. For example, the ΔGbaq value of 1 for the wild type (−12.5 kcal/ mol) is predicted to be very similar to that of 2 for the D816V mutant (−12.6 kcal/mol), in contrast to the 3-fold difference between the corresponding IC50 values (Table 1). This discrepancy indicates that despite the implementation of a proper solvation free energy term, the present scoring function requires further modifications to be extensively used in the structure-based drug design. To gain structural insight into the activities of the newly identified natural-product inhibitors, their interactions with the ATP-binding sites of the wild type and D816V mutant were investigated in the comparative fashion. Figure 3 shows the lowest-energy binding conformations of 1−4 calculated with docking simulations using the modified scoring function. Although some differences are observed in the detailed binding

submicromolar level inhibitory activities against the wild type and D816V mutant, respectively, while 3 and 4 have relatively moderate biochemical potencies for the two target proteins. All four natural-product inhibitors seem to deserve further development to optimize the inhibitory and anticancer activities by structure−activity relationship (SAR) analysis because they were also screened computationally for good physicochemical properties that should be satisfied by drug candidates. In this regard, the calculated LogP values of 1, 2, 3, and 4 amount to 3.001, 2.109, 3.696, and 0.771, respectively. Although the biochemical potency of 3 is lower than those of 1 and 2 with respect to both the wild type and D816V mutant, it is nonetheless anticipated to serve as a promising molecular core from which even more potent common inhibitors can be derivatized through chemical synthesis because of its low molecular weight (268.3 amu). We now address the energetic features relevant to binding of the newly discovered natural-product c-KIT inhibitors in the ATP-binding pockets of the wild type and D816V mutant. Table 2 lists the computed binding free energies of 1−4 with Table 2. Calculated Binding Free Energies of 1−4 in the Gas Phase (ΔGbgas) and in Water (ΔGbaq) with Respect to the Wild-Type and D816V Mutant c-KIT in Comparison with the Solvation Free Energies (ΔGsol)a ΔGbgas

a

ΔGbaq

inhibitor

wild type

D816V

ΔG

1 2 3 4

−25.2 −24.8 −19.7 −26.8

−26.0 −25.7 −20.3 −27.5

−12.7 −13.1 −8.4 −16.2

sol

wild type

D816V

−12.5 −11.7 −11.3 −10.6

−13.3 −12.6 −11.9 −11.3

All energy data are given in kcal/mol.

respect to the wild type and D816V mutant. Actually, the binding free energy to form a protein−ligand complex in water (ΔGbaq) can be obtained approximately by subtracting the ligand solvation free energy (ΔGsol) from the corresponding binding free energy in the gas phase (ΔGbgas). The ΔGbaq values were therefore decomposed into the two components to estimate their relative contributions. The standard deviations (SDs) of ΔGbgas values of 1−4 for the wild type and D816V mutant and the corresponding ΔGsol values are 3.1, 3.2, and 3.6 kcal/mol, respectively. In comparison, the SDs of ΔGbaq values for the wild type and D816V mutant are 0.79 and 0.87 kcal/ mol, respectively. The similarity in the SD values indicate that ΔGbgas and ΔGbsol terms would contribute to ΔGbaq to a comparable extent, which implies the necessity of a hydration free energy term in the scoring function for estimating the relative biochemical potencies of c-KIT inhibitors. Therefore, it should be kept in mind that in order to design a new c-KIT inhibitor with increased potency, the chemical modifications should be made in such a way to maximize the strength of cKIT-inhibitor interactions and simultaneously to minimize the desolvation cost for binding of the inhibitor in the ATP-binding site. The binding free energies of 1−4 calculated with eq 1 are in reasonably good agreement with the experimentally measured inhibitory activities. For instance, ΔGbaq values of 1−4 for the wild type appear to be a little higher than those for the D816V mutant, which is consistent with their relatively lower biochemical potencies for the former than for the latter (Table 1). The decrease of ΔGbaq values on going from the wild

Figure 3. Comparative view of the binding modes of 1−4 in the ATP binding sites of (a) wild type and (b) D816V mutant. Carbon atoms of 1, 2, 3, and 4 are indicated in green, cyan, pink, and gray, respectively. The positions of Cys673 and residue 816 are also indicated. C



Article

corresponding role of the hydrogen bond receptor is played by the side-chain carboxylate ion of Asp677 located at the entrance of the ATP-binding site in the D816V-1 complex. Both third hydrogen bonds seem to have the effect of positioning the phenolic −OH and carbonyl groups of 1 to establish the bidentate hydrogen bonds with the backbone atoms of Cys673 in the ATP-binding sites because they reside in proximity to Cys673. Although c-KIT-1 and D816V-1 complexes have the same number of hydrogen bonds, the hydrogen-bond interactions of the latter seem to be stronger than those of the former because of the involvement of a charged group (Asp677). This strengthening of the hydrogen bond may be invoked to explain a little higher inhibitory activity of 1 for the D816V mutant than for the wild type. It is also a common feature of the calculated binding modes of 1 in the ATP-binding sites of the wild type and D816V mutant that its nonpolar moieties establish van der Waals contacts with the hydrophobic side chains of Leu595, Val603, Ala621, Leu644, Val654, Tyr672, and Leu799. However, 1 can establish the additional hydrophobic contact with the side chain of Phe811 in the c-KIT-1 complex, in which the DFG (Asp810Phe811-Gly812) motif of the A-loop resides in the autoinhibited off state with Phe811 flipped over and occluding the ATP-binding site. This interaction could not be observed in the D816V-1 complex because the active form of the D816V mutant was adopted in which the DFG motif resides in the active conformation with Phe811 staying distant from the ATPbinding site to allow binding of the adenine moiety. The lack of van der Waals interaction with Phe811 seems not to have a significant impact on the binding affinity of 1 for the D816V mutant because it can form the stronger hydrogen bond with Asp677 than with Leu595 due to the decrease in steric hindrance by Phe811. Taken together, the respective low micromolar and submicromolar biochemical potencies of 1 for the wild type and D816V mutant can be attributed to the multiple hydrogen bonds and van der Waals interactions established concurrently in the ATP-binding site. It is interesting to note that the change of the inhibitor from 1 to 3 leads to nearly a 10-fold decrease in the biochemical potency against the wild type as compared to only a little change in the IC50 value with respect to the D816V mutant (Table 1). Despite the reduced inhibitory activity, 3 is anticipated to serve as a good molecular core from which even more potent inhibitors may be derivatized due to the low molecular weight (268.3 amu). Figure 5 compares the lowestenergy binding modes of 3 for the wild type and D816V mutant, which were obtained from docking simulations of 3 in the ATP-binding sites. Consistent with a more than 10-fold

configurations, 1−4 appear to be stabilized in a similar way with respect to the two receptor proteins. For example, the polar moieties on the fused phenyl and six- or five-membered ring establish the hydrogen bonds with the backbone groups of Cys673 in the hinge region of the ATP-binding site, while the terminal phenyl ring is directed toward the glycine-rich phosphate-binding loop (P-loop). The necessity of the interactions with Cys673 and the P-loop for tight binding to c-KIT was also implicated in X-ray crystallographic studies and molecular dynamics simulations of c-KIT in complex with potent inhibitors.14,35 As a check on the availability of the other stable binding modes, additional extensive docking simulations of 1−4 were conducted with respect to the wild type and D816V mutant using 3D grid maps constructed to include the whole kinase domain of c-KIT. The results for clustering analyses of 100 total docking runs for all protein−ligand complexes indicate that the binding modes shown in Figure 3 represent the lowest ΔGbaq and simultaneously the most probable cluster with more than 63% of the total population. This supports the reliability of the calculated binding modes of 1−4 with respect to the wild type and D816V mutant in the active and inactive conformations, respectively. As can be inferred from a little difference in IC50 values of 1−4 for the wild type and D816V mutant, however, there seems to be some marked differences in their detailed patterns of complexation in the ATP-binding sites of the two receptor proteins. To address the structural relevance to good inhibitory activities of the newly identified natural-product inhibitors with respect to the wild type and D816V mutant, their calculated binding modes were analyzed in detail. The lowest-energy conformations of 1 in the ATP-binding sites of the wild-type and D816V mutant c-KIT are compared in Figure 4. It is a

Figure 4. Calculated binding modes of 1 in the ATP-binding sites of (a) wild type and (b) D816V mutant of c-KIT. Carbon atoms of c-KIT and 1 are indicated in cyan and green, respectively. Dotted lines indicate the hydrogen bonds.

common structural feature in the c-KIT-1 and D816V-1 complexes that one of the phenolic −OH moieties and the vicinal carbonyl oxygen of 1 donates and receives a hydrogen bond to the aminocarbonyl oxygen and from the backbone amidic nitrogen of Cys673, respectively. The appearance of these bidentate hydrogen bonds is consistent with the X-ray crystallographic analyses of c-KIT-inhibitor complexes, in which the capability to establish two hydrogen bonds with the backbone atoms of Cys673 proved to be necessary for the inhibitor to bind tightly in the ATP-binding site.46 On the other hand, the third hydrogen bond is established in different ways in the two enzyme−inhibitor complexes. In the c-KIT-1 complex, the second phenolic −OH group of 1 forms a hydrogen bond with the backbone aminocarbonyl oxygen of Leu595 that is a component of the P-loop, whereas the

Figure 5. Calculated binding modes of 3 in the ATP-binding sites of (a) wild type and (b) D816V mutant of c-KIT. Carbon atoms of c-KIT and 3 are indicated in cyan and green, respectively. Dotted lines indicate the hydrogen bonds. D



Article

was employed as the receptor model. Thus, the two receptor models were prepared to perform the molecular-docking-based virtual screening to identify the potent and common inhibitors of the wildtype and D816V mutant c-KIT. To prepare the all-atom models for the two target proteins, hydrogen atoms were attached to each protein atom. To this end, the protonation states of the ionizable residues (Asp, Glu, His, and Lys) were assigned according to the hydrogen-bonding patterns in the original X-ray crystal structure of wild-type c-KIT and those in the homology-modeled structure of the D816V mutant. The quality of the homology-modeled structure of the D816V mutant was evaluated with the PROCHECK program. The results show that the backbone Φ and Ψ dihedral angles of 84.4%, 12.3%, and 2.6% of the residues are located within most favorable, additionally allowed, and generously allowed regions of the Ramachandran plot, respectively. This good stereochemical property indicates that the structure of the D816V mutant obtained with homology modeling may serve as a good receptor model for virtual screening. Prior to conducting the molecular docking for virtual screening, a chemical library was constructed for c-KIT comprising about 40 000 natural products from the latest version (March 2015) of the database distributed by InterBioScreen Ltd. For this purpose, all 56 700 natural products included in the original chemical database were filtrated according to Lipinski’s “rule of five” using the ISIS/BASE program to select only the molecules that possess desirable physicochemical properties as a potential drug candidate.39 To avoid the redundancies in the initially selected molecules, the structurally similar natural products whose Tanimoto coefficient exceeded 0.8 were clustered to a single representative. All these filtrated natural products were then processed with the CORINA program to obtain their threedimensional atomic coordinates, which was followed by the calculations of atomic charges based on the Gasteiger−Marsilli method.40 The modified version of the AutoDock program41 was used for the virtual screening of common inhibitors of the wild-type and D816V mutant c-KIT because the outperformance of its protein− ligand binding free energy function had been demonstrated in various target proteins.31−33 Virtual Screening of Natural Products with Docking Simulations. Although the ligand solvation effects proved to be critically important in the formation of a stable protein−ligand complex,42,43 the scoring function of the original AutoDock program contained a crude form of the solvation free energy function that could cope with only six atom types. Therefore, the natural products included in the docking library were screened with a modified scoring function constructed by substituting a sophisticated solvation free energy term for the original one. This improved scoring function can be written as follows.

difference in IC50 values, some different patterns of interaction are observed in c-KIT-3 and D816V-3 complexes. For example, the phenolic −OH group donates a hydrogen bond to the backbone aminocarbonyl oxygen of Leu595 in the c-KIT-3 complex, while it plays the roles of hydrogen-bond acceptor and donor with respect to the backbone amidic nitrogen of Cys673 and aminocarbonyl oxygen of Glu671 in the D816V-1 complex, respectively. In the c-KIT-3 complex, the carbonyl moiety on the five-membered ring serves as the hydrogen-bond acceptor for Cys673. Like 1, 3 appears to establish the stronger hydrophobic interactions with the inactive form of the wild type than D816V mutant in the active conformation because the side chain of Phe811 resides in proximity to the ATP-binding site in the former. However, the third hydrogen bond is observed only in the D816V-3 complex between the terminal methoxy group of 3 and the side-chain butyl ammonium ion of Lys623. This indicates that 3 would be capable of raising the inhibitory activity against the D816V mutant by changing the binding mode in such a way as to maximize the hydrogen-bond interactions to compensate for some loss of hydrophobic interactions. Thus, the strengthening of hydrogen bonds in the ATP-binding site may be invoked to elucidate the higher inhibitory activity of 3 for the D816V mutant than for the wild type of c-KIT. With respect to improving the biochemical potency of 3 by chemical modifications, it is noted that 3 is incapable of forming a hydrogen bond with the backbone aminocarbonyl oxygen of Cys673 of both wild type and D816V mutant. This may be one of the reasons for the lower inhibitory activity of 3 than 1. Therefore, the substitution of a hydrogen-bonddonating moiety near the phenolic −OH group of 3 would have the effect of raising the binding affinity by inducing an additional hydrogen bond in the ATP-binding site. It is also worth noting that 3 is accommodated in the ATP-binding site without a significant interaction with the side-chain of Leu644 that appeared to form a van der Waals contact with 1 in both cKIT-1 and D816V-1 complexes. Hence, the enhancement of biochemical potency is anticipated if a nonpolar group is derived at the terminal phenyl ring of 3. The addition of a hydrophobic moiety has merit over a hydrogen-bonding group because the former has little effect on the desolvation cost, whereas the introduction of the latter may reduce the binding affinity due to the increased stabilization in water. Using 3 as the molecular core, we now plan to perform SAR studies to guide the synthesis of the common inhibitors of c-KIT wild type and D816V mutant with low nanomolar activity.

■

⎛A Bij ⎞ ij aq ΔG bind = WvdW ∑ ∑ ⎜⎜ 12 − 6 ⎟⎟ + Whbond rij ⎠ i = 1 j = 1 ⎝ rij ⎛C

∑ ∑ E(t )⎜⎜

EXPERIMENTAL SECTION

i=1 j=1

General Experimental Procedures. Since the structure of the D816V mutant of c-KIT was unavailable in publicly accessible databases, its atomic coordinates were obtained through homology modeling with the latest version of the MODELER program.36 In this model building, we employed an optimization method involving the conjugate gradient method and molecular dynamics simulation to minimize violations of the spatial restraints. The structure of the wild type in the active conformation (PDB entry: 1PKG)37 was used as the template to build that of the D816V mutant. In this active form, JMR and the A-loop of c-KIT stay in open positions and are allowed to approach the active site without spatial restraints. Actually, this active conformation can be made possible in the D816V mutant due to the weakening of electrostatic interactions between JMR and the kinase domain, which has the effect of facilitating the departure of JMR from the autoinhibitory position. In the case of the wild-type c-KIT, the Xray crystal structure in the autoinhibited form (PDB entry: 1T46)38

ij 12 r ⎝ ij

−

qq Dij ⎞ ⎟ + Welec ∑ ∑ i j 10 ⎟ rij ⎠ i = 1 j = 1 ε(rij)rij

+ WtorNtor + Wsol ∑ Si(Occimax − i=1

2

2

∑ Vje−rij /2σ ) j≠i

(1) Here, WvdW, Whbond, Welec, Wtor, and Wsol are the weighting factors of van der Waals, hydrogen bond, electrostatic interactions, torsional motion, and hydration free energy of a putative inhibitor, respectively. rij stands for the interatomic distance, and Aij, Bij, Cij, and Dij are associated with the depths of the potential energy well and with the equilibrium separations between protein and ligand atoms. AMBER force field parameters were used to calculate van der Waals interaction energies as implemented in the original AutoDock program. The hydrogen-bond energy term has an additional weighting factor (E(t)) to reflect the angle-dependent directionality. In calculating the electrostatic interaction energy between c-KIT and a natural product, Mehler et al.’s sigmoidal function was used as the distance-dependent E



Article

dielectric constant (ε(rij)).44 In the entropic term, Ntor denotes the number of rotatable bonds in a natural product. In the solvation free energy term, Si and Vi indicate the atomic solvation parameter and the fragmental atomic volume, respectively, while Occimax represents the maximum occupancy of atom i.45 The atomic parameters derived by Park were used to calculate the hydration free energies of the natural products in the docking library because they showed a good performance in the SAMPL4 blind prediction challenge.44 The substitution of this new solvation free energy term seems to increase the accuracy of the protein−ligand binding free energy function because the binding affinity of an inhibitor with many polar moieties has often been overestimated due to the underestimation of ligand solvation effects in the scoring functions.42 The actual docking simulations began with the construction of the 3D grids of interaction energy. These uniquely defined potential grids for the receptor proteins were then used for docking simulations of all natural products. As the center of the common grids, we used the center of mass coordinates of the ligands that had been removed from the ATP-binding sites of wild-type and D816V mutant c-KIT. These grid maps were of dimension 61 × 61 × 61 points with a spacing of 0.375 Å, yielding a receptor model that included atoms within 22.9 Å of the grid center. Using the protein−ligand binding free energy function shown in eq 1, docking simulations were performed in the ATP-binding sites of wild-type and D816V mutant c-KIT to score and rank the natural products according to the calculated binding affinities. Only the natural products belonging to 100 virtual hits for both target proteins in common were selected for the subsequent biochemical evaluations because the research was focused on the identification of common inhibitors for the wild type and D816V mutant. Enzyme Inhibition Assays. Thirty-seven putative common inhibitors selected from virtual screening were purchased from InterBioScreen Ltd. and tested for having inhibitory activity against the wild type and D816V mutant at a concentration of 100 μM. These enzyme inhibition assays were performed at Ambit Bioscience Corp. (San Diego, CA, USA) using the broad kinase inhibitor staurosporine as the reference. c-KIT-tagged T7 phage strains were grown in parallel in 24-well blocks in an E. coli host derived from the BL21 strain. E. coli were grown to log-phase and infected with T7 phage from a frozen stock (multiplicity of infection = 0.4) and incubated with shaking at 32 °C until lysis (90−150 min). The lysates were centrifuged and filtered (0.2 μm) to remove cell debris. The remaining kinases were produced in HEK-293 cells and subsequently tagged with DNA for qPCR detection. Streptavidin-coated magnetic beads were treated with biotinylated small-molecule ligands for 30 min at room temperature to generate the affinity resins for kinase assays. The liganded beads were blocked with excess biotin and washed with blocking buffer (SeaBlock (Pierce), 1% BSA, 0.05% Tween 20, 1 mM DTT) to remove unbound ligand and to reduce nonspecific phage binding. Binding reactions were assembled by combining kinases, liganded affinity beads, and test compounds in 1× binding buffer (20% SeaBlock, 0.17× PBS, 0.05% Tween 20, 6 mM DTT). Test compounds were prepared as 40× stocks in 100% DMSO and directly diluted (100 μM) into the assay. All reactions were performed in polypropylene 384-well plates in a final volume of 0.04 mL. The assay plates were incubated at room temperature with shaking for 1 h, and the affinity beads were washed with wash buffer (1× PBS, 0.05% Tween 20). The beads were then resuspended in elution buffer (1× PBS, 0.05% Tween 20, 0.5 μM nonbiotinylated affinity ligand) and incubated at room temperature with shaking for 30 min. The kinase concentration in the eluates was measured by qPCR. The natural products with percent of control values lower than 20 for both target proteins were selected to measure the IC50 values. The IC50 determinations were carried out using radiometric kinase assays ([γ-33P]-ATP) at Reaction Biology Corp. (Malvern, PA, USA). The enzymatic activity of c-KIT (GenBank NP_000213; amino acids 544−976, His-tagged, expressed in insect cells; MW = 53.3 kDa, 200 nM in the reaction in the presence of 2 mM MnCl2) or c-KIT (D816V) (GenBank NP_000213.1; amino acids 544−976 with D816V mutation, N-term GST-His-tagged with a 3C cleavage site,

expressed in Sf9 insect cells; MW = 77.4 kDa, 2.5 nM in the reaction) was monitored using 0.2 mg/mL of the substrate (poly[Glu:Tyr] (4:1)) dissolved in the freshly prepared reaction buffer (20 mM HEPES (pH 7.5), 10 mM MgCl2, 1 mM EGTA, 0.02% BRIJ-35, 0.02 mg/mL BSA, 0.1 mM Na3VO4, 2 mM DTT, 1% DMSO). Each putative c-KIT inhibitor was dissolved in 100% DMSO at a specific concentration and diluted in a serial manner with epMotion 5070 in DMSO. After delivering the candidate inhibitor dissolved in DMSO into the kinase reaction mixture by Acoustic Technology (Echo550; nanoliter range), the reaction mixture was incubated for 20 min at room temperature. To initiate the enzymatic reaction, 33P-ATP with a specific activity of 10 μCi/μL was delivered into the reaction mixture to reach the final ATP concentration of 1 μM. Radioactivity was then monitored by the filter-binding method after the incubation of the reaction mixture for 2 h at room temperature. At given concentrations of the inhibitor, biochemical potency was measured by the percent remaining kinase activity with respect to vehicle (dimethyl sulfoxide) reaction. Curve fits and IC50 values were then obtained using the PRISM program (GraphPad Software).

■

AUTHOR INFORMATION

Corresponding Authors

*Tel mail: *Tel mail:

(H. Park): +82-2-3408-3766. Fax: +82-2-3408-4334. [email protected]. (S. Hong): +82-42-350-2811. Fax: +82-42-350-2812. [email protected].

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported by BioNano Health-Guard Research Center funded by the Ministry of Science, ICT & Future Planning (MSIP) of Korea as a Global Frontier Project (H-GUARD_2014M3A6B2060507) and Institute for Basic Science (IBS-R010-G1).

■

REFERENCES

(1) Roskoski, R. Biochem. Biophys. Res. Commun. 2005, 338, 1307− 1315. (2) Hunter, T. Cell 1995, 80, 225−236. (3) Ashman, L. K.; Griffith, R. Expert Opin. Invest. Drugs 2013, 22, 103−115. (4) Lennartsson, J.; Rönnstrand, L. Physiol. Rev. 2012, 92, 1619− 1649. (5) Demetri, G. D.; von Mehren, M.; Blanke, C. D.; van den Abbeele, A. D.; Eisenberg, B.; Roberts, P. J.; Heinrich, M. C.; Tuveson, D. A.; Singer, S.; Janicek, M.; Fletcher, J. A.; Silverman, S. G.; Silberman, S. L.; Capdeville, R.; Kiese, B.; Peng, B.; Dimitrijevic, S.; Druker, B. J.; Corless, C.; Fletcher, C. D.; Joensuu, H. N. Engl. J. Med. 2002, 347, 472−480. (6) Lennartsson, J.; Jelacic, T.; Linnekin, D.; Shivakrupa, R. Stem Cells 2005, 23, 16−43. (7) Mol, C. D.; Lim, K. B.; Sridhar, V.; Zou, H.; Chien, E. Y. T.; Sang, B.-C.; Nowakowski, J.; Kassel, D. B.; Cronin, C. N.; McRee, D. E. J. Biol. Chem. 2003, 278, 31461−31464. (8) Torrent, M.; Rickert, K.; Pan, B. S.; Sepp-Lorenzino, L. J. Mol. Graphics Modell. 2004, 23, 153−165. (9) Chian, R. J.; Young, S.; Danilkovitch-Miagkova, A.; Ronnstrand, L.; Leonard, E.; Ferrao, P.; Ashman, L.; Linnekin, D. Blood 2001, 98, 1365−1373. (10) Bougherara, H.; Subra, F.; Crepin, R.; Tauc, P.; Auclair, C.; Poul, M.-A. Mol. Cancer Res. 2009, 7, 1525−1533. (11) Laine, E.; de Beauchene, I. C.; Perahia, D.; Auclair, C.; Tchertanov, L. PLoS Comput. Biol. 2011, 7, e1002068. (12) Frost, M. J.; Ferrao, P. T.; Hughes, T. P.; Ashman, L. K. Mol. Cancer Ther. 2002, 1, 1115−1124. F



Article

(13) Ma, Y.; Zeng, S.; Metcalfe, D. D.; Akin, C.; Dimitrijevic, S.; Butterfield, J. H.; McMahon, G.; Longley, B. J. Blood 2002, 99, 1741− 1744. (14) Park, H.; Lee, S.; Lee, S.; Hong, S. Org. Biomol. Chem. 2014, 12, 4644−4655. (15) Lee, S.; Lee, H.; Kim, J.; Lee, S.; Kim, S. J.; Choi, B.-S.; Hong, S.-S.; Hong, S. J. Med. Chem. 2014, 57, 6428−6443. (16) Davies, R. J.; Pierce, A. C.; Forster, C.; Grey, R.; Xu, J.; Arnost, M.; Choquette, D.; Galullo, V.; Tian, S.-K.; Henkel, G.; Chen, G.; Heidary, D. K.; Ma, J.; Stuver-Moody, C.; Namchuk, M. J. Med. Chem. 2011, 54, 7184−7192. (17) Kunz, R. K.; Rumfelt, S.; Chen, N.; Zhang, D.; Tasker, A. S.; Bürli, R.; Hungate, R.; Yu, V.; Nguyen, Y.; Whittington, D. A.; Meagher, K. L.; Plant, M.; Tudor, Y.; Schrag, M.; Xu, Y.; Ng, G. Y.; Hu, E. Bioorg. Med. Chem. Lett. 2008, 18, 5115−5117. (18) Duveau, D. Y.; Hu, X.; Walsh, M. J.; Shukla, S.; Skoumbourdis, A. P.; Boxer, M. B.; Ambudkar, S. V.; Shen, M.; Thomas, C. J. Bioorg. Med. Chem. Lett. 2013, 23, 682−686. (19) Richters, A.; Ketzer, J.; Getlik, M.; Grütter, C.; Schneider, R.; Heuckmann, J. M.; Heynck, S.; Sos, M. L.; Gupta, A.; Unger, A.; Schultz-Fademrecht, C.; Thomas, R. K.; Bauer, S.; Rauh, D. J. Med. Chem. 2013, 56, 5757−5772. (20) Vendôme, J.; Letard, S.; Martin, F.; Svinarchuk, F.; Dubreuil, P.; Auclair, C.; Le Bret, M. J. Med. Chem. 2005, 48, 6194−6201. (21) Hu, E.; Tasker, A.; White, R. D.; Kunz, R. K.; Human, J.; Chen, N.; Bürli, R.; Hungate, R.; Novak, P.; Itano, A.; Zhang, X.; Yu, V.; Nguyen, Y.; Tudor, Y.; Plant, M.; Flynn, S.; Xu, Y.; Meagher, K. L.; Whittington, D. A.; Ng, G. Y. J. Med. Chem. 2008, 51, 3065−3068. (22) Chen, N.; Bürli, R. W.; Neira, S.; Hungate, R.; Zhang, D.; Yu, V.; Nguyen, Y.; Tudor, Y.; Plant, M.; Flynn, S.; Meagher, K. L.; Lee, M. R.; Zhang, X.; Itano, A.; Schrag, M.; Xu, Y.; Ng, G. Y.; Hu, E. Bioorg. Med. Chem. Lett. 2008, 18, 4137−4141. (23) Margulies, D.; Opatowsky, Y.; Fletcher, S.; Saraogi, I.; Tsou, L. K.; Saha, S.; Lax, I.; Schlessinger, J.; Hamilton, A. D. ChemBioChem 2009, 10, 1955−1958. (24) Thompson, D.; Miller, C.; McCarthy, F. O. Biochemistry 2008, 47, 10333−10344. (25) Lin, Y.-L.; Roux, B. J. Am. Chem. Soc. 2013, 135, 14741−14753. (26) Kansal, N.; Silakari, O.; Ravikumar, M. Eur. J. Med. Chem. 2010, 45, 393−404. (27) Quinn, R. J.; Carroll, A. R.; Pham, N. B.; Baron, P.; Palframan, M. E.; Suraweera, L.; Pierens, G. K.; Muresan, S. J. Nat. Prod. 2008, 71, 464−468. (28) Gabillot-Carre, M.; Lepelletier, Y.; Humbert, M.; de Sepuvelda, P.; Hamouda, N. B.; Zappulla, J. P.; Liblau, R.; Ribadeau-Dumas, A.; Machavoine, F.; Letard, S.; Baude, C.; Hermant, A.; Yang, Y.; Vargaftig, J.; Bodemer, C.; Morelon, E.; Lortholary, O.; Recher, C.; Laurent, G.; Dy, M.; Arock, M.; Dubreuil, P.; Hermine, O. Blood 2006, 108, 1065−1072. (29) Vendôme, J.; Letard, S.; Martin, F.; Svinarchuk, F.; Dubreuil, P.; Auclair, C.; Le Bret, M. J. Med. Chem. 2005, 48, 6194−6201. (30) Warren, G. L.; Andrews, C. W.; Capelli, A. M.; Clarke, B.; LaLonde, J.; Lambert, M. H.; Lindvall, M.; Nevins, N.; Semus, S. F.; Senger, S.; Tedesco, G.; Wall, I. D.; Woolven, J. M.; Peishoff, C. E.; Head, M. S. J. Med. Chem. 2006, 49, 5912−5931. (31) Park, H.; Eom, J. W.; Kim, Y. H. J. Chem. Inf. Model. 2014, 54, 2139−2146. (32) Park, H.; Shin, Y.; Choe, H.; Hong, S. J. Am. Chem. Soc. 2015, 137, 337−348. (33) Park, H.; Park, S. Y.; Nam, S. W.; Ryu, S. E. J. Biomol. Screening 2014, 19, 1383−1390. (34) Fabian, M. A.; Biggs, W. H., III; Treiber, D. K.; Atteridge, C. E.; Azimioara, M. D.; Bendetti, M. G.; Carter, T. A.; Ciceri, P.; Edeen, P. T.; Floyd, M.; Ford, J. M.; Galvin, M.; Gerlach, J. L.; Grotzfeld, R. M.; Herrgard, S.; Insko, D. E.; Insko, M. A.; Lai, A. G.; Lelias, J.-M.; Mehta, S. A.; Milanov, Z. V.; Velasco, A. M.; Wodicka, L. M.; Patel, H. K.; Zarrinkar, P. P.; Lockhart, D. J. Nat. Biotechnol. 2005, 23, 329−336. (35) Zhang, C.; Ibrahim, P. N.; Zhang, J.; Burton, E. A.; Habets, G.; Zhang, Y.; Powell, B.; West, B. L.; Matusow, B.; Tsang, G.; Shellooe,

R.; Carias, H.; Nguyen, H.; Marimuthu, A.; Zhang, K. Y.; Oh, A.; Bremer, R.; Hurt, C. R.; Artis, D. R.; Wu, G.; Nespi, M.; Spevak, W.; Lin, P.; Nolop, K.; Hirth, P.; Tesch, G. H.; Bollag, G. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 5689−5694. (36) Sali, A.; Blundell, T. L. J. Mol. Biol. 1993, 234, 779−815. (37) Mol, C. D.; Lim, K. B.; Sridhar, V.; Zou, H.; Chien, E. Y.; Sang, B. C.; Nowakowski, J.; Kassel, D. B.; Cronin, C. N.; McRee, D. E. J. Biol. Chem. 2003, 278, 31461−31464. (38) Mol, C. D.; Dougan, D. R.; Schneider, T. R.; Skene, R. J.; Kraus, M. L.; Scheibe, D. N.; Snell, G. P.; Zou, H.; Sang, B. C.; Wilson, K. P. J. Biol. Chem. 2004, 279, 31655−31663. (39) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Adv. Drug Delivery Rev. 1997, 23, 3−25. (40) Gasteiger, J.; Marsili, M. Tetrahedron 1980, 36, 3219−3228. (41) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W. E.; Belew, R. K.; Olson, A. J. J. Comput. Chem. 1998, 19, 1639− 1662. (42) Shoichet, B. K.; Leach, A. R.; Kuntz, I. D. Proteins: Struct., Funct., Genet. 1999, 34, 4−16. (43) Lindström, A.; Edvinsson, L.; Johansson, A.; Andersson, C. D.; Andersson, I. E.; Raubacher, F.; Linusson, A. J. Chem. Inf. Model. 2011, 51, 267−282. (44) Mehler, E. L.; Solmajer, T. Protein Eng., Des. Sel. 1991, 4, 903− 910. (45) Stouten, P. F. W.; Frömmel, C.; Nakamura, H.; Sander, C. Mol. Simul. 1993, 10, 97−120. (46) Park, H. J. Comput.-Aided Mol. Des. 2014, 28, 175−186.

G


Discovery of Dual Inhibitors for Wild Type and D816V Mutant of c-KIT

Recommend Documents