Toward the Validation of Maternal Embryonic Leucine Zipper Kinase

Article pubs.acs.org/jmc

Toward the Validation of Maternal Embryonic Leucine Zipper Kinase: Discovery, Optimization of Highly Potent and Selective Inhibitors, and Preliminary Biology Insight B. Barry Touré,*,† John Giraldes, Troy Smith, Elizabeth R. Sprague, Yaping Wang, Simon Mathieu, Zhuoliang Chen, Yuji Mishina, Yun Feng, Yan Yan-Neale, Subarna Shakya, Dongshu Chen, Matthew Meyer, David Puleo,‡ J. Tres Brazell, Christopher Straub, David Sage, Kirk Wright, Yanqiu Yuan,§ Xin Chen, Jose Duca, Sean Kim, Li Tian, Eric Martin, Kristen Hurov, and Wenlin Shao∥ Novartis Institutes for Biomedical Research, 250 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States S Supporting Information *

ABSTRACT: MELK kinase has been implicated in playing an important role in tumorigenesis. Our previous studies suggested that MELK is involved in the regulation of cell cycle and its genetic depletion leads to growth inhibition in a subset of high MELK-expressing basal-like breast cancer cell lines. Herein we describe the discovery and optimization of novel MELK inhibitors 8a and 8b that recapitulate the cellular effects observed by short hairpin ribonucleic acid (shRNA)mediated MELK knockdown in cellular models. We also discovered a novel fluorine-induced hydrophobic collapse that locked the ligand in its bioactive conformation and led to a 20-fold gain in potency. These novel pharmacological inhibitors achieved high exposure in vivo and were well tolerated, which may allow further in vivo evaluation.

■

INTRODUCTION Small molecule inhibition in conjunction with short hairpin ribonucleic acid (shRNA) knockdown can represent a very powerful target validation strategy, the desirable outcome being that the small molecule should match the shRNA-induced phenotype.1,2 Herein, we discuss our reduction to practice of this drug target discovery paradigm to maternal embryonic leucine zipper kinase (MELK). MELK is a serine/threonine kinase and an atypical member of the sucrose nonfermenting 1/ adenosine monophosphate-activated protein kinase (SNF1/ AMPK) family.3,4 Its expression has been found to be elevated in a broad spectrum of cancer types, including breast, ovarian, lung, pancreas, and colorectal cancers.5 In glioma and breast cancer stem cells, overexpression of MELK was shown to be required for cancer cell proliferation.6,7 Correspondingly, MELK has been identified as part of a gene signature representing undifferentiated cancers that are thought to be more aggressive in behavior and linked to poor patient outcome,8 and an increase in MELK expression correlated with more aggressive forms of astrocytoma, breast cancer, melanoma, and glioblastoma.9−12 We have recently described the identification of MELK as a novel oncogenic kinase from an in vivo gain of function tumorigenesis screen using a kinomewide open reading frames library.13 Our studies demonstrated that MELK was highly expressed in the basal-like breast cancer cells and genetic depletion of MELK by shRNA led to growth inhibition in basal-like breast cancer cells but not in luminal breast cancer cells. Our data suggested that MELK could play an essential role in regulating cell mitosis in a subset of cancer © XXXX American Chemical Society

cells; thus it presents an opportunity of a cancer-selective therapeutic target. Several MELK small molecule inhibitors have been reported in the literature. OTSSP167 was an inhibitor described to have subnanomolar potency against MELK in an in vitro biochemical assay and exhibited antiproliferative effects against a number of cancer cell lines.14 The compound also demonstrated in vivo efficacy in tumor xenograft models. However, selectivity of this inhibitor and whether the antitumor effect was solely driven by MELK inhibition were not known. Studies led by a separate group reported the structural characterization of two type I compounds, which inhibited MELK with biochemical IC50 of 27 nM and 12 nM, respectively. When tested against a panel of 50 kinases, the former also inhibited three other kinases (haspin, GSK3β, and CDK2/cylinA) with activities below 1 μM while the latter displayed inhibition against seven further kinases with a potency below 1 μM (AKT1, Flt3, PKCβ, PDK1, RET, PAK4, and PKAα).15 The most potent compound displayed antiproliferative activity against A2789, LNCaP (2 μM) with concomitant increase in p53 or cleaved PARP cellular levels. Finally, Astex in collaboration with Janssen also reported type I and type II MELK inhibitors with low nanomolar potency in biochemical assays.16,17 In the type I series, the most potent compound (IC50 = 37 nM against MELK) showed >50% inhibition at 1 μM against six additional Received: January 15, 2016

A

DOI: 10.1021/acs.jmedchem.6b00052 J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

Article

kinases in a panel of 235 enzymes. Interestingly, the most potent compound in the type II series (19 nM against MELK) was less selective; it showed >50% inhibition at 1 μM for 31/ 243 kinases tested. These compounds were shown to be cell permeable, but no MELK cellular activity or PK data were reported. The advancement in developing pharmacological inhibitors of MELK is encouraging; at the same time, development of well-characterized and selective MELK inhibitors suitable as in vitro and in vivo probes is still needed to further explore the biological functions and therapeutic impact of this target. On the basis of our key findings (discussed above)13 and useful tools we have developed, we initiated medicinal chemistry efforts aimed at identifying highly potent and selective MELK inhibitors guided by structural biology. We then asked whether pharmacological inhibition of MELK could recapitulate the findings from shRNA-based genetic depletion.

■

RESULTS AND DISCUSSION Virtual Screening and Binding Mode of Pyrazole Hit. High-throughput screening (HTS) against a MELK kinase domain biochemical assay with a 250 K Chiron compound library yielded 1123 hits that showed MELK IC50 values less than 10 μM. These hits were then used to build profile-QSAR (pQSAR) and Surrogate AutoShim models.18,19 pQSAR models achieve extremely high accuracy by combining the available MELK IC50 values with all of the historical IC50 data across the entire kinase family. 3D Surrogate AutoShim docking models create a customized MELK scoring function by training pharmacophore “shims” in the active site of an ensemble of 16 diverse kinase crystal structures to best reproduce the MELK IC50 values. These models were subsequently used to screen 1.4 million compound library in silico. From this initial virtual screening campaign (VS1), 1700 compounds were selected for IC50 measurement in the biochemical assay based primarily on sampling the chemical diversity of the Novartis archive not represented in the original data collection. The “novelty/hitrate histogram” for VS1 in Figure 1a shows that virtually all of the selected compounds were very unlike the training set, (Tanimoto coefficient of 0.34, and compounds with experimental activity on EGFR. Each of these bins was clustered and sampled at different prioritized densities to collect the total of 2K compounds. Figure 1b shows that about two-thirds of the compounds selected from VS2 were novel, and the hit-rate for the novel compounds was 40−50%. The combined hits led to several potent medicinal chemistry series with good physical properties and in vitro ADME. Among those, the N-arylpyrazoles (represented by 1a, Chart 1) emerged as a particularly promising series due to their kinase selectivity profile; a prototypical representative of this series

Figure 1. Discovery of selective MELK inhibitors via virtual screening: (a) first iteration of MELK virtual screening, showing hit-rate versus novelty analysis; (b) novelty/hit-rate plot for second iteration of MELK virtual screening. The blue bars represent hit rates from the virtual screen, while the red bars plot the percentage of hits in each bin that share a Bemis and Murcko framework with the frameworks in the training set. The green circles show the hit counts for each similarity bin. For (b), while the majority of the compounds were still chosen for novelty, including novel scaffolds, a significant number were also chosen as SAR sets to further evaluate leads from the first iteration.

Chart 1. Confirmed MELK Virtual Screen Hitsa

Kinase domain activity was measured at a[ATP] = 20 μM and [ATP] = 2 mM.

a

b

(1a) only inhibited three kinases having IC50 < 5 μM (Flt3, PDGFRα, and MAP4K4) besides MELK in a standard panel of 71 kinases. The binding mode of N-arylpyrazole 1a is shown in Figure 2. It is a type I inhibitor, binding MELK with the C-helix and DFG segment in the “in” conformations and utilizing a single hydrogen-bond interaction with the hinge (NH Cys89) combined with numerous van der Waals interactions. The nearly coplanar pyridine−pyrazole rings are sandwiched between Ala38 and Leu86 from the N-lobe and Leu139 and Ile149 from the C-lobe. The pyridine nitrogen forms a watermediated hydrogen bond with Lys40, the conserved lysine in kinases that also forms a salt bridge with Glu57 of the C-helix in the active conformation. In general, MELK ATP-site binders sit higher in the pocket compared to other kinases due to MELK having a bulky Ile (Ile149) at the DFG-1 residue in contrast to the more frequently observed Ala, Gly, or Ser, thus likely increasing the chances of obtaining a MELK-specific inhibitor. The phenyl moiety is positioned perpendicular to Tyr88 forming edge-to-face π-interactions, a unique feature that was described in more detail elsewhere.16 The Leu27, Ile17, Pro90, B



Article

coupling to afford 3-fluoro-4-pyrazolylpyridine 7a or Suzuki− Miyaura cross-couplings to yield 3-methoxy-4-pyrazolylpyridine 7b. The synthesis of unsubstituted 4-pyrazolylpyridine 7c started with the Chan-Lam coupling of 4-(1H-pyrazol-4yl)pyridine with 4-bromophenylboronic acid. The product of this reaction (N-halogenated arylpyrazole 6b) was then carried through the Buchwald N-arylation condition to afford the title compound 7c. 7a was then converted into 3-alkoxy-4pyrazolylpyridines 8a−g by ipso displacement of the fluoride with alkoxides, which were formed in situ for the purpose. To probe the left side of the molecule (N-phenyl replacements), two orthogonal synthetic sequences were deployed. As shown in Scheme 3, analogue alkoxypyridine 12a was prepared in three sequential steps starting with the bromination of commercially available 1-phenylpyrazole (13) to form 4bromopyrazole 14. Suzuki−Miyaura coupling was then used to forge the pyrazole−pyridine bond (14 into 15), and finally, nucleophilic aromatic substitution yielded 12a. The synthesis of related 3-arylpyrazole analogues 12b−d commenced with commercially available building block 9, which was submitted to conventional SNAr reaction conditions to provide 10. Coupling of 10 with 3-boronopyrazole then afforded the key NH-free pyrazole 11; the latter was then N-arylated under Buchwald N-arylation reaction conditions to afford the target compounds 12b−d. MELK SAR, Structural Basis of Potency Improvement, and Rodent PK. Our SAR exploration efforts initially focused on improving the binding efficiency of N-arylpyrazole hit (1a) by targeting amide replacements. The synthesis of a focused set of compounds resulted in the identification of methylsulfone (3a) and N-methylpiperazine (7a) moieties as suitable amide surrogates (Table 1). Using 4-pyrazolylpyridine 3a as a new starting point, we aimed our efforts at identifying potential pyridine replacements. Unlike compound 1b (Chart 1), deletion of the pyridine nitrogen (4-pyridine to phenyl) abrogated the activity of 4-phenylpyrazole 3b in inhibiting MELK (IC50 > 20 μM). Similarly, moving the pyridine nitrogen to the 3-position also had deleterious effect on potency (not shown). We hypothesized that this significant activity cliff20 indicated that the water-mediated H-bonding with Lys40 was both optimal and critical; thus it should be preserved. In agreement, phenol 3c, which is expected to establish a direct interaction with Lys40, was slightly less potent. Overall, these preliminary efforts fell short of our potency goal of achieving submicromolar IC50 values at high ATP concentration (2 mM), which better mimics the cellular concentration. Next, guided by X-ray crystal structure of N-arylpyrazole hit 1a bound to MELK and modeling, we hypothesized that further extending the methoxy group in compound 7b into the ribose pocket would improve potency. More specifically, we targeted piperidine containing 3-alkoxy-4-pyrazolylpyridine 8a, reasoning that the piperidine amine would form a salt bridge with Glu93. The compound was duly synthesized, and gratifyingly this change afforded our first single-digit nanomolar MELK inhibitor (IC50 = 0.002 μM). The X-ray crystal structure of Novartis MELK inhibitor 8a (NVS-MELK8a) bound to MELK (shown in pink in Figure 3) validated the design premise, and in addition, the P-loop was also ordered, which was not seen in initial N-arylpyrazole 1a bound to MELK (vide supra). Compound 8a remained very potent (IC50 = 140 nM) when the ATP concentration in the biochemical assay was shifted from 20 μM to 2 mM. Its potency was well tracked between full-length MELK versus catalytic domain construct (5 nM

Figure 2. Cocrystal structure of 1a bound to MELK. 1a bound in the active site of MELK (PDB code 5IH8). Key interactions with catalytic Lys40, hinge Cys89, Tyr88, and Pro90 are highlighted. The ligand also makes several hydrophobic interactions with the protein. Red spheres represent water molecules with hydrogen bonds indicated by dashed lines.

and Cys89 residues also contribute to the pocket. The remaining amide side chain is solvent exposed, providing a handle to impart better physical chemical properties on the series if needed. The P-loop is disordered as is often observed in kinase inhibitor cocrystal structures. Because this starting scaffold afforded excellent kinases selectivity, we elected to preserve the general binding features throughout the optimization process, specifically the single point interaction with the hinge, and edge-to-face interaction with Tyr88. Guided by the crystal structure, our medicinal chemistry optimization plan focused on decreasing the hydrophobicity in the solvent exposed region, probing the pyridine region and identifying a handle to reach the ribose pocket to further improve potency. Chemistry. The compounds were prepared using three general routes. First, reaction of the commercially available 4bromopyrazole with 1-fluoro-(4-methylsufonyl)benzene afforded the common intermediate sulfone 2, which was coupled with the appropriate boronic acids to form 4-arylpyrazoles 3a− c (Scheme 1). Scheme 2 describes the synthesis of variants that Scheme 1. Synthesis of 4-Arylpyzoles 3a−ca

Reaction conditions: (a) Cs2CO3, 95 °C, 1 h; (b) ArB(OH)2, Pd2(dba)3, P(Cy)3, K3PO4, microwave 130 °C, 20 min. a

replaced the sulfone moiety with N-methylpiperazine. The requisite bromopyrazole 6a was synthesized via a two-step procedure that was initiated by lithiating commercially available bromoarene 4 and in situ trapping of the resulting organometallic intermediate with di-tert-butyl azodicarboxylate (DBAD). The product of this reaction (hydrazide 5) was then deprotected in situ and condensed with 2-bromomalonaldehyde to form pyrazole 6a. The coupling of pyridine derivatives with 6a was achieved using either Negishi crossC



Article

Scheme 2. Synthesis of 7a−c and General Procedure for the Late Stage Introduction of Alkoxidesa

Reaction conditions: (a) n-Bu-Li, −65 °C, 5 min and then DBAD, −60 °C to −10 °C, 30 min; (b) 2-bromomalonaldehyde, TFA, DCM, rt; (c) 2,2,6,6-tetramethylpiperidine, n-BuLi, THF, 0 °C, 20 min, add 3-fluoropyridine, −78 °C, 1 h; (d) ZnBr2, warm to rt, add 6a, Pd(PPh3)4, 60 °C, 1 h, then Pd(amphos)Cl2, 60 °C, 8 h; (e) 3-fluoro-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)pyridine, K3PO4, Pd(amphos)Cl2, dioxane/water, 100 °C, 2 h or 3-methoxy-4-pyridinylboronic acid, PdCl2(PPh3)2, Na2CO3, water, dioxane, microwave 140 °C, 30 min; (f) NaH, alcohols, DMF 25−85 °C; (g) 4-bromophenylboronic acid, Cu(OAc)2, 4 Å molecular sieves, DCM; (h) 1-methylpiperazine, Pd2(dba)3, BINAP, tert-BuONa, dioxane, microwave 130 °C, 20 min. a

Scheme 3. Synthesis of Monocationic Compounds (12a−d)a

Reactions conditions: (a) NaH, tert-butyl 4-(hydroxymethyl)piperidine-1-carboxylate, 70 °C, 5 h; (b) 4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2yl)-1H-pyrazole, Pd(dppf)Cl2·DCM adduct, Na2CO3, dioxane, water, reflux, 16 h. (c) General procedure 3: ArI, Pd2(dba)3, XPhos, tert-BuONa, toluene, 105 °C, 16 h, then deprotection using general procedure 2. (d) AcOH, Br2, 100 °C; (e) 3-fluoro-4-pyridineB(OH)2, Pd(amphos)Cl2, K3PO4, dioxane, 80−95 °C, 2−16 h.

a

equal to the Km. As shown in Table 2, the compound was at least 90-fold more selective in targeting MELK in all cases. These results validated our medicinal chemistry strategy of prioritizing selective hits from the virtual screens at the outset using a small panel of 71 kinases. Compound 8a was fairly soluble (0.22 g/L at pH 6.8) and showed a good permeability in the Caco-2 assay (Papp(A−B) = 14 × 10−6 cm/s). Its extraction ratio in the mouse microsomal assay was moderate to high (ER = 0.75), and no chemical instability issues were detected upon stirring in mouse plasma

versus 2 nM). Furthermore, the dissociation constant (KD) for this compound determined by SPR binding was 2 nM; its binding profile was consistent with moderately slow dissociation kinetics (t1/2 = 60 s) (not shown). To assess its selectivity, the compound was profiled at a concentration of 1 μM against a panel of 456 kinases (KINOMEscan); it only inhibited seven off-target kinases in addition to MELK with >85% inhibition of binding at 1 μM demonstrating great selectivity (Table 2). To verify the panel test result, we determined the IC50 values against four of these off-target kinases at ATP concentrations D



Article

Table 1. SAR Summary

Figure 3. Overlay of 8a and 8f bound in the kinase active site. For 8a (pink sticks with pink ribbon), a salt bridge is observed between the piperidine basic amine and Glu93 as well as additional ordering of the P-loop. For 8f (teal sticks with teal ribbon), asterisks denote the end points for the disordered P-loop. Water molecules are shown as spheres in the same color as the coordinating ligand with hydrogen bonds indicated by dashed lines (PDB code 5IH9 for 8a and PDB code 5IHA for 8f).

Table 2. Kinase Selectivity Profile of 8a

a

Average of at least two separate runs (8a is an average of 26 runs). MELK kinase domain activity was measured at [ATP] = 20 μM. b MELK kinase domain activity was measured at [ATP] = 2 mM. cFull length MELK kinase activity was measured at [ATP] = 20 μM. dSPRderived KD.

Flt3 (ITD) Haspin CSNK2A2 KIT (A829) KIT (D816) MINK PDGFRα PDGFRβ TYK2 MELK

for up to 4 h (not shown). Subcutaneous administration of compound 8a at 30 mg/kg in C57BL/6 mice resulted in good plasma exposure (Table 3). The compound adsorption into the systemic circulation was rapid (Tmax = 0.4 h) and peak plasma concentration reached 6.6 μM. An ascending dose PK study in female athymic nude mice showed that the rate of compound release was maximal at 120 mg/kg and all clearance mechanisms can be saturated at 240 mg/kg (Supporting Information Figure S4). However, when administered orally at 10 mg/kg in C57BL/6 male mice, it showed very poor PK (3.6% oral bioavailability) consistent with very high in vivo clearance (113 mL min−1 kg−1). We hypothesized that the poor

% inhibition at 1 μMa

IC50 (μM)b

96 93 66 79 86 77 79 57 93 86

0.18 0.19 >10

0.42

0.002

a

Screened against 456 kinases in DiscoveRx KINOMEscan panel. IC50 values against off-target kinases were determined using our internal kinase assays. b

oral PK properties were related to two basic amines embedded within the structure, making this compound subject to potential E



Article

Table 3. Pharmacokinetics of 8a and 12a in C57BL/6 Male Mice AUC0−7h (nM·h) clearance (mL min−1 kg−1) Vss (L/kg) Cmax (nM) Tmax (h) F (%)

8a (iv)a

8a (po)a

8a (sc)b

12a (po)c

230 113.8 40.8

83.2

18863

3439

34 0.5 3.6%

6558 0.4 273%

1032 1.7 43%

a

Compound was dosed iv (1 mg/kg) and po (10 mg/kg) in C57BL/6 male mice using 5% ethanol, 10% PG, 25% (20%, v/v) Cremophor EL in PBS buffer as vehicle. bDosed subcutaneously at 30 mg/kg using 10% PG + 5% Solutol. cDosed orally at 5 mg/kg using 10% PG + 25% (20%, v/v) Solutol.

Figure 4. Ortho-fluorine induced hydrophobic collapse in 12b. Binding conformation of 12b was determined by X-ray crystallography (the protein is excluded for clarity) (PDB code 5IHC). Fluorine is within VDW interaction distance of the piperidine (space filling model shown on the right).

lysosomal trapping. Subsequently, we explored decreasing the number of cationic charges at neutral pH by targeting the left side N-methylpiperazine moiety. Potency dropped to 36 nM when this motif was deleted (N-phenylpyrazole 12a); however, as hypothesized, this transformation improved considerably the oral PK profile, F = 43% (Table 3). The PK profile of other similar monocationic compounds was in agreement with this observation (data not shown). After identification of PK improvement handles, further structural changes were then undertaken to understand potency gain drivers in the ribose region. Guided by X-ray insight, cis-cyclohexylamine 8b was designed to interact with both the carboxylic acid of Glu93 and the carbonyl of Glu136. This transformation further increased the potency by 3-fold under both low and high ATP biochemical conditions. The replacement of the piperidine with a pyridine ring (8a to 8c) decreased the potency, and in general, saturated heterocycles were vastly superior to their corresponding aromatic cores. The deletion of the methylene linker had very little impact on the potency (compare compound 8a with 8d), which hinted that the salt bridge interaction with Glu93 was not critical. It should be noted that a similar observation regarding amine-Glu93 interaction was published elsewhere.16 In agreement, compound 8e, featuring a simple bulky cyclohexane group (no amine), was similar to 3alkoxy-4-pyrazolylpyridine 8a in potency. Reduction of the size of the hydrophobic substituent to isobutyl (8f) resulted in 4.5fold decrease in potency, while further truncation to methyl group (7b) resulted in 32-fold potency loss relative to compound 8a. We obtained the X-ray cocrystal structure of MELK and compound isobutyl 8f (see overlay in Figure 3), which confirmed the position of the secondary butyl substituent in the ribose pocket (i.e., no change in binding mode), and in addition, the P-loop was disordered. It follows that the potency gain in the ribose pocket is driven primarily by hydrophobic contacts, and potent monoamine inhibitors can be obtained. However, the replacement of heterocycles in the ribose pocket with hydrophobic substituents resulted in poor physical chemical properties due to increased clogP. Next, using N-phenylpyrazole 12a as starting point, a modest gain in potency was realized upon ortho-fluorination of the phenyl ring (12b, Table 1). The cocrystal structure of this compound bound to MELK showed that the piperidine ring and the fluorine atom were within van der Waals interaction distance (Figure 4). It is worth mentioning that the hydrophobic collapse was not limited to piperidine; other simple hydrophobic groups gave a similar result (not shown). We hypothesized that this hydrophobic collapse21,22 locked

compound 12b in its bioactive conformation, and as such, it can be reliably used for further potency gain. Combining all the above potency SAR learnings (understanding of the role of the piperazine moiety and the fluorine induced hydrophobic collapse) led to the synthesis of morpholine 12c and its subsequent conversion into compound 12d via fluorination of the aromatic ring. This time, the fluorine effect was even more pronounced and afforded 20-fold gain in IC50. As a result of these efforts, we obtained another distinct subclass of potent monocationic MELK inhibitors (12a−d), a representative example of which (12a) demonstrated superior PK profile when compared to the dicationic compound 3-alkoxy-4pyrazolylpyridine 8a, thus validating our design strategy. Cellular Profiling. To determine if compounds are able to inhibit MELK in a cellular context, we first evaluated the effects of compounds 8a and 8b on cell growth in a MELK-dependent cell line, MDA-MB-468, and a MELK-independent cell line, MCF-7. MDA-MB-468 is a triple negative basal breast cancer cell line that expresses MELK at a much higher level in comparison to MCF7, an ER-positive luminal breast cancer cell line (Supporting Information Figure S2). Corresponding to the higher expression of MELK in MDA-MB-468 compared to MCF-7, genetic knockdown of MELK by two different MELKtargeting shRNAs led to growth inhibition of MDA-MD-468, but not MCF-7, as determined by the colony formation assay (Supporting Information Figure S1A). Knockdown efficiency of MELK was monitored by immunoblotting that showed greater than 90% of protein depletion (Supporting Information Figure S1B and Figure S1C). These data are consistent with our previously reported findings.13 Viability following compound treatment was measured using CellTiter-Glo, which monitors the level of ATP present in cells. Consistent with our data using shRNA-mediated depletion of MELK, both compounds 8a and 8b inhibited the growth of MDA-MB-468 cells with an IC50 of approximately 0.06 and 0.1 μM, respectively (Table 4 and Supporting Information Figure S3). The growth inhibitory effects of these compounds on MCF-7 cells were much less, Table 4. Antiproliferative Activity of 8a and 8ba IC50 (μM) cell line

8a

8b

MDA-MB-468 MCF7

0.11 3.68

0.06 1.2

a

Average of at least two experiments in MELK-dependent versus nonMELK dependent cell line.

F



Article

Figure 5. Effect of 8b on cell cycle. (a) Treatment with 8b led to increased apoptosis and polyploidy in MDA-MB-468 in a time-dependent manner. (b) MELK inhibitor treatment did not lead to increased apoptosis and polyploidy in MCF-7.

resulting in IC50 of 1.2 and 3.6 μM, respectively. These data suggest that NVS-MELK8a and NVS-MELK8b exhibit selective activity targeting MELK, leading to more effective proliferation inhibition in MELK-dependent cancer cells compared to MELK-independent cells. Next we examined the effect of compound 8b on the cell cycle of cancer cells. Our previous report demonstrated that MELK knockdown led to G2/M arrest and cytokinesis failure

in basal breast cancer cells.5,13,23 Our cell cycle analysis of MDA-MB-468 in comparison to MCF-7 showed that knockdown of MELK via two different shRNAs led to an increase of cells with 4N DNA, indicating a G2/M accumulation in MDAMB-468 cells (Supporting Information Figure S1D). On the other hand, MELK knockdown showed no effect on the cell cycle profile of MCF-7 cells. We then tested the effect of compound 8b on cell cycle in these two cell lines (Figure 5). G



Article

spectrometer and Thermo CAD detectors with UV detection at 254 nm in both cases. Nuclear magnetic resonance (NMR) spectra were acquired on a Bruker Advance 400 spectrometer (400 MHz) unless stated otherwise. All 1H NMR spectra are reported in parts per million (ppm) and were measured relative to the appropriate reference signals. Data for 1H NMR are described as follows: chemical shift (δ in ppm), multiplicity (s, singlet; d, doublet; t, triplet; q, quartet; quint, quintuplet; m, multiplet; br, broad signal), integration, coupling constant J (Hz). 4-Bromo-1-(4-(methylsulfonyl)phenyl)-1H-pyrazole (2). 1Fluoro-4-(methylsulfonyl)benzene (3.5 g, 20.2 mmol) and 4-bromo1H-pyrazole (3.1 g, 21.1 mmol) with cesium carbonate (8 g, 24.55 mmol) were combined in DMF (30 mL) in a round-bottom flask. The resulting mixture was heated to 95 °C for 1 h, then quenched with water and extracted with ethyl acetate (3 × 150 mL). The combined organic phases was washed with 4% NaCl (aq), dried, filtered, and concentrated. The resulting white solid was triturated from acetonitrile to give the title compound as white crystals (5.91 g, 19.6 mmol). 1H NMR (400 MHz, DMSO-d6) δ: 8.99 (s, 1H), 8.15−8.03 (m, 4H), 8.00 (s, 1H), 3.27 (s, 3H). MS calcd for C10H9BrN2O2S: [M]+ = 302. Found: [M + H]+ = 303. 4-(1-(4-(Methylsulfonyl)phenyl)-1H-pyrazol-4-yl)pyridine (3a). In a 10 mL microwave vial, 4-bromo-1-(4-(methylsulfonyl)phenyl)-1H-pyrazole (2) (235 mg, 0.78 mmol) and 4-pyridineboronic acid (288 mg, 2.34 mmol) were dissolved in dioxane (5 mL). The vial was purged by bubbling N2, then Pd2(dba)3 (21 mg, 0.02 mmol) and tricyclohexylphosphine (15 mg, 0.06 mmol) were added. Aqueous 1 M tripotassium phosphate (1.56 mL, 1.56 mmol) was added, and the reaction mixture was heated to 130 °C (2 bar) for 20 min. The reaction was cooled to ambient temperature, diluted with EtOAc (10 mL), and washed with water followed by brine. The organic layer was decanted and concentrated in vacuo. The crude mixture was purified by silica gel flash chromatography (50−100% EtOAc in heptane) followed by trituration from acetonitrile to obtain the title compound as a white powder (16 mg, 0.05 mmol). 1H NMR (400 MHz, DMSOd6) δ: 9.42 (s, 1H), 8.60 (d, J = 6.06 Hz, 2H), 8.51 (s, 1H), 8.18 (d, J = 8.80 Hz, 2H), 8.10 (d, J = 8.80 Hz, 2H), 7.71−7.76 (m, 2H), 3.28 (s, 3H). HRMS calcd for C15H13N3O2S [M + H]+ = 300.0800. Found [M + H]+ = 300.0804. 1-(4-(Methylsulfonyl)phenyl)-4-phenyl-1H-pyrazole (3b). To a vial capped with a septum were added 4-bromo-1-(4(methylsulfonyl)phenyl)-1H-pyrazole (2) (100 mg, 0.33 mmol), phenylboronic acid (50 mg, 0.41 mmol), bis(tert-butyl(4dimethylaminophenyl)phosphine)dichloropalladium(II) (25 mg, 0.04 mmol), tripotassium phosphate (105 mg, 0.50 mmol), and dioxane (3 mL). The suspension was heated to 95 °C for 2 h under conventional heating. The reaction mixture was allowed to cool, then diluted with EtOAc and filtered through Celite, washing with DCM (2 mL × 3), and MeOH (2 mL × 3). The filtrate was concentrated to a solid, then triturated from acetonitrile to obtain the title compound (85 mg, 86%). 1H NMR (400 MHz, DMSO-d6) δ: 9.20 (s, 1H), 8.35 (s, 1H), 8.21−8.14 (m, 2H), 8.11−8.04 (m, 2H), 7.80−7.71 (m, 2H), 7.48− 7.39 (m, 2H), 7.34−7.24 (m, 1H), 3.27 (s, 3H). HRMS calcd for C16H15N2O2S [M + H]+ = 298.0854. Found [M + H]+ = 299.0847. 4-(1-(4-(Methylsulfonyl)phenyl)-1H-pyrazol-4-yl)phenol (3c). In a vial capped with a septum were added 4-bromo-1-(4(methylsulfonyl)phenyl)-1H-pyrazole (2) (100 mg, 0.33 mmol), 4hydroxybenzene boronic acid (50.4 mg, 0.37 mmol), bis(tert-butyl(4dimethylaminophenyl)phosphine)dichloropalladium(II) (23.51 mg, 0.03 mmol), tripotassium phosphate (106 mg, 0.50 mmol), and dioxane (3 mL). The suspension was heated to 80 °C for 16 h under conventional heating. The reaction mixture was diluted with EtOAc and washed with water (10 mL × 3) followed by brine; the organic portions were combined and dried over Na2SO4, filtered, and concentrated to a dark solid. The crude was triturated from acetonitrile to obtain the title compound (22 mg, 21%). 1H NMR (400 MHz, DMSO-d6) δ: 9.01 (s, 1H), 8.21 (s, 1H), 8.16−8.11 (m, 2H), 8.08− 8.03 (m, 2H), 7.58−7.50 (m, 2H), 6.87−6.77 (m, 2H), 3.27 (s, 3H). HRMS calcd for C16H15N2O3S [M + H]+ = 315.0803. Found [M + H+]+ = 315.0791.

We also included AZD1152, a potent Aurora B inhibitor, in the study for comparison. As shown in Figure 5a, compound 8b treatment in MDA-MB-468 cells led to a time-dependent increase in cells harboring 4N DNA content, similar to what we have observed using MELK shRNAs by FACs analysis. Furthermore, the inhibitor treatment also resulted in an increase of cells with greater than 4N DNA content, suggesting impaired cell mitosis leading to polyploidy. A concomitant increase in cell death was also observed. Similar results were also observed with 3-alkoxy-4-pyrazolylpyridine 8a (not shown). These findings are consistent with the notion that MELK plays an important role in regulating cell mitosis and division; its inhibition could lead to aberrant mitosis and cell death. In contrast to the observations in MDA-MB-468 cells, compound 8b treatment did not show any cell cycle effect in MCF-7 cells (Figure 5b). This is in contrast to the cell cycle effect of AZD1152 in both cell lines. These data support our hypothesis that selective inhibition of MELK leads to specific growth inhibition of cancer cells that are dependent on MELK for proliferation while sparing cells that are independent.

■

CONCLUSIONS In summary, we utilized a virtual screening approach to rapidly discover novel and selective chemical inhibitor series of MELK kinase activity. On the basis of the N-arylpyrazole hit 1a, rational structure-based drug design was utilized to optimize potency, yielding 3-alkoxy-4-pyrazolylpyridine 8a, which potently and selectively inhibited MELK. Both compounds 8a (NVS-MELK8a) and 8b (NVS-MELK8b) retain biochemical potencies against MELK at physiologically relevant high ATP concentration and demonstrate selective antiproliferation and cell cycle effect in MELK-dependent cancer cells. These observations are consistent with those observed using MELK shRNA, suggesting inhibition of MELK using selective pharmacological inhibitors could recapitulate findings by genetic depletion. Moreover, when dosed subcutaneously, 3alkoxy-4-pyrazolylpyridine 8a was well tolerated and achieved very high plasma and tumor exposures (up to 96 μM). On the basis of the results discussed above, we believe that these compounds represent very valuable in vitro and in vivo probes for further studying MELK biology and its unique roles in tumorigenesis.

■

EXPERIMENTAL SECTION

General Procedures. Unless otherwise noted, all materials were obtained from commercial suppliers and used without further purification. All reactions involving air or moisture-sensitive reagents were performed under nitrogen. Chromatographic purification of products was accomplished on a Teledyne Isco with RediSep normalphase silica gel (35−60 μM) and UV detection at 254 nm. Preparative HPLC was performed on a Waters 2545 HPLC system equipped with Waters PDA 2998 and/or Waters 3100 mass spectrometer detection. Method A: Waters Sunfire 30 mm i.d. × 50 mm, 5 μm particle column, eluting with water/acetonitrile with 0.1% TFA modifier, flow rate 75 mL/min, 1.5 mL injection. Method B: Waters X-Bridge 30 mm i.d. × 50 mm, 5 μm particle column, water/acetonitrile with 5 mM NH4OH, flow rate 75 mL/min, 1.5 mL injection volume. All final compounds were purified to >95% (unless otherwise stated) as determined by Waters AcQuity UPLC−UV system equipped with a Waters LCT Premier mass spectrometer with UV detection at 254 nm. Low resolution mass spectra were acquired on LC/MS systems using electrospray ionization methods from a range of instruments of the following configurations: Agilent 1100 HPLC−UV system equipped with Waters ZQ mass spectrometer and Schimadzu ELSD detectors, Waters AcQuity UPLC−UV system equipped with Waters SQ mass H



Article

Di-tert-butyl 1-(4-(4-Methylpiperazin-1-yl)phenyl)hydrazine1,2-dicarboxylate (5). To a stirred solution of 1-(4-bromophenyl)-4methylpiperazine (4) (124.7 g, 490 mmol) in THF (1.9 L) was added 2.5 M n-BuLi solution (234.4 mL) at −78 °C under N2. The internal temperature was kept below −65 °C during the addition. After addition, the mixture was stirred at −78 °C for 5 min. A solution of DBAD (135 g) in THF (0.63 L) was added dropwise. The internal temperature was kept below −60 °C during the addition, then the reaction mixture was warmed to −10 °C in 30 min and quenched by slow addition of water (0.5 L). The resulting mixture was partitioned between EtOAc (2.0 L) and water (1.0 L). The aqueous layer was extracted with EtOAc (0.5 L × 2), and the combined organic layer was washed with brine (0.5 L) and dried (Na2SO4). The organic layer was concentrated under reduced pressure to yield a red oil and purified by silica gel chromatography giving the desired product (162 g, 82%). MS calcd for C21H35N4O4 [M + H]+ = 407.3. Found [M + H]+ = 407.2. 1-(4-(4-Bromo-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (6a). To a stirred solution of di-tert-butyl 1-(4-(4-methylpiperazin-1yl)phenyl)hydrazine-1,2-dicarboxylate (5) (120 g, 296 mmol) in DCM (2.40 L) was added 2-bromomalonaldehyde (53.6 g, 355 mL), followed by TFA (452 mL). After addition, the reaction was stirred at ambient temperature for 2 days. All the volatile solvent was removed under reduced pressure, and the residue was dissolved in EtOAc (2.5 L). Water (0.5 L) was added, followed by portionwise addition of solid NaHCO3 until pH = 8 (∼250 g needed). The aqueous layer was extracted with EtOAc (0.5 L × 2) and DCM (0.5 L). The combined organic layer was dried over Na2SO4 and filtered. The organic layer was concentrated under reduced pressure and the residue was purified by silica gel chromatography, then triturated using ACN (100 mL) and filtered to yield the title compound (40.0 g, 42% yield). 1H NMR (400 MHz, DMSO-d6) δ 8.62 (s, 1H), 7.78 (s, 1H), 7.66−7.58 (m, 2H), 7.08−6.99 (m, 2H), 3.21−3.13 (m, 4H), 2.48−2.42 (m, 4H), 2.22 (s, 3H). MS calcd for C14H18BrN4: [M + H]+ = 321.1. Found [M + H]+ = 321.0. 1-(4-(4-(3-Fluoropyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4methylpiperazine (7a). A solution of 2,2,6,6-tetramethylpiperidine (194 g, 1373 mmol) in THF (2.0 L) was cooled to 0 °C, and n-BuLi (550 mL, 1373 mmol, 2.5 M solution in hexane) was added slowly under N2. The mixture was stirred at 0 °C for 20 min. The mixture was then cooled to −78 °C (dry ice/acetone bath), and a solution of 3fluoropyridine (121.4 g, 1250 mmol) in THF (0.2 L) was added slowly. The mixture was stirred at −78 °C for 1 h. Then a solution of ZnBr2 (310 g, 1375 mmol) in THF (1.25 L) was added to this mixture and slowly warmed up to room temperature (∼1.5 h). To this suspension were added 1-(4-(4-bromo-1H-pyrazol-1-yl)phenyl)-4methylpiperazine (40 g, 125 mmol) and Pd(PPh3)4 (14.0 g, 12.5 mmol). The mixture was heated to 60 °C and stirred for 1 h, then bis(di-tert-butyl(4-dimethylaminophenyl)phosphine)dichloropalladium(II) (Pd(amphos)Cl2) (8.9 g, 12.5 mmol) was added to this mixture, which was then stirred at 60 °C for 18 h. The reaction was allowed to cool to ambient temperature, then quenched with saturated NaHCO3 solution (1.8 L). The mixture was then filtered on a filter funnel. The filtered solid was washed with EtOAc (3 × 2 L). The combined filtrates were transferred to a separatory funnel, and the layers were separated. The organic layer was dried over MgSO4 and filtered. The filtered organics were concentrated under reduced pressure to dryness. The dark residue was chromatographed on silica column (dry loading with silica) to give an orange solid contaminated with 2,2,6,6-tetramethylpiperidine (or its derivative) and other minor impurities. The orange solid was then triturated first with MTBE (500 mL) and then suspended in MeOH (250 mL) and rotated at 50 °C on a rotavap for 30 min. The mixture was diluted with water (750 mL) and rotated at 50 °C on a rotavap for 1 h. The suspension was cooled to room temp and filtered. The tan solid obtained was washed with water (2 × 125 mL), dried to furnish the desired product (17 g, 35%). Alternative Procedure for the Preparation of 1-(4-(4-(3Fluoropyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (7a). A suspension of 1-(4-(4-bromo-1H-pyrazol-1-yl)phenyl)-4methylpiperazine (6a) (200 mg, 0.62 mmol), 3-fluoro-4-(4,4,5,5-

tetramethyl-1,3,2-dioxaborolan-2-yl)pyridine (167 mg, 0.747 mmol), K3PO4 (264 mg, 1.25 mmol), and (Pd(amphos)Cl2) (27 mg, 0.04 mmol) in a mixture of dioxane (5 mL) and water (0.500 mL) was degassed with bubbling nitrogen for 3 min and then heated to 100 °C for 2 h. The resulting mixture was cooled and diluted with EtOAc, then washed with water followed by brine. The organic layer was dried over Na2SO4, filtered, and concentrated in vacuo to an off-white solid. The crude mixture was purified using flash chromatography (12 g RediSep SiO2 column, gradient EtOAc/heptane, 0−100% over 25 min, then 100% for 10 min, then eluted with MeOH/DCM 0−5%) to give the title compound (110 mg, 50%). 1H NMR (400 MHz, DMSO-d6) δ 8.98 (d, J = 1.3 Hz, 1H), 8.61 (d, J = 2.8 Hz, 1H), 8.43 (dd, J = 5.1, 1.2 Hz, 1H), 8.28 (d, J = 1.9 Hz, 1H), 7.88 (dd, J = 6.8, 5.0 Hz, 1H), 7.79−7.66 (m, 2H), 7.15−6.99 (m, 2H), 3.24−3.14 (m, 4H), 2.48− 2.44 (m, 4H), 2.23 (s, 3H). MS calcd for C19H21FN5 [M]+ = 338.2. Found [M + H]+ = 338.2. 1-(4-(4-(3-Methoxypyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4methylpiperazine (7b). To a 5 mL microwave vial containing a mixture of 3-methoxy-4-pyridinylboronic acid (304 mg, 2.0 mmol), 1(4-(4-bromo-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (6a) (213 mg, 0.66 mmol), and 2 M sodium carbonate solution (1.7 mL, 3.3 mmol) in dioxane (3 mL) was added trans-PdCl2(PPh3)2 (46.5 mg, 0.066 mmol, 0.1 equiv). The vial was purged with nitrogen, and the mixture was heated to 140 °C in a microwave reactor for 30 min, then diluted with EtOAc and washed with water and then brine. The crude mixture was concentrated in vacuo, then purified by silica gel flash chromatography (100% EtOAc, then 0−20% MeOH in EtOAc) to obtain the title compound as an off-white powder (81 mg, 35%). 1H NMR (400 MHz, DMSO-d6) δ ppm 8.87 (s, 1H), 8.42 (s, 1H), 8.30 (s, 1H), 8.21 (d, J = 5.05 Hz, 1H), 7.69−7.74 (m, 3H), 7.07 (d, J = 9.35 Hz, 2H), 4.04 (s, 3H), 3.16−3.21 (m, 4H), 2.45−2.48 (m, 4H), 2.23 (s, 3H). HRMS calcd for C20H24N5O [M + H]+ = 350.1981. Found [M + H]+ = 350.1985. 1-Methyl-4-(4-(4-(pyridin-4-yl)-1H-pyrazol-1-yl)phenyl)piperazine (7c). Step 1. A combined suspension of 4-(1H-pyrazol-4yl)pyridine (1.0 g, 13.8 mmol), (4-bromophenyl)boronic acid (2.77 g, 13.8 mmol), copper(II) acetate (3.75 g, 20.7 mmol), and 4 Å molecular sieves (7 g) in dichloromethane (14 mL) was stirred open to air for 2 days. The reaction mixture was vacuum filtered through Celite, then gravity filtered through glass wool. Then, it was purified by flash silica gel chromatography (EtOAc/heptane, 0−30%) to obtain 4(1-(4-bromophenyl)-1H-pyrazol-4-yl)pyridine (6b) (50 mg, 2%). 1H NMR (400 MHz, DMSO-d6) δ 9.28−9.27 (m, 1H), 8.60−8.55 (m, 2H), 8.43 (s, 1H), 7.88 (d, J = 9.0 Hz, 2H), 7.75 (d, J = 9.0 Hz, 2H), 7.73−7.70 (m, 2H). Step 2. A combined mixture of 4-(1-(4-bromophenyl)-1H-pyrazol-4yl)pyridine (50 mg, 0.17 mmol), 1-methylpiperazine (25 mg, 0.25 mmol), BINAP (1 mg, 1.6 μmol), tris(dibenzylideneacetone)dipalladium(0) (1 mg, 1.1 μmol), and sodium tert-butoxide (24 mg, 0.25 mmol) in dioxane (2 mL) was degassed under N2, then capped and heated by microwave to 130 °C for 20 min. The reaction mixture was cooled and diluted with ethyl acetate (20 mL), washed with water (2×) and then brine, concentrated, and purified by flash silica gel chromatography (7 mM ammonia in methanol/dichloromethane, 0− 15%) to give the title compound as a pale yellow powder (18 mg, 34%). 1H NMR (400 MHz, DMSO-d6) δ 9.07 (s, 1H), 8.54 (d, J = 6.1 Hz, 2H), 8.31 (s, 1H), 7.78−7.61 (m, 4H), 7.08 (d, J = 9.2 Hz, 2H), 3.25−3.13 (m, 4H), 2.48−2.44 (m, 4H), 2.23 (s, 3H). HRMS calcd for C19H22N5 [M + H]+ = 320.1875. Found [M + H]+ = 320.1863. General Procedure 1 for SNAr Reactions with Alkoxides. To a solution of alcohol (1.0−3.0 equiv) in DMF (0.02−0.2 M) at 0 °C was added NaH (60% oil immersion) (2−10 equiv) in one portion. The reaction was then warmed up to room temperature for 5−45 min before being cooled back to 0 °C where a solution 1-(4-(4-(3fluoropyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (7a) (1.0 equiv) in DMF was added in a rapid, dropwise manner. The reaction mixture was then heated between 35 and 100 °C depending on alcohol while monitoring by LCMS (0.5−96 h). The resulting reaction mixture was diluted with EtOAc/water (∼10:1), and after effervescence ceased, the layers were partitioned. The organic layer was I



Article

1-Methyl-4-(4-(4-(3-(piperidin-4-yloxy)pyridin-4-yl)-1H-pyrazol-1-yl)phenyl)piperazine (8d). 8d was prepared according to general procedure 1 using 1-(4-(4-(3-fluoropyridin-4-yl)-1H-pyrazol1-yl)phenyl)-4-methylpiperazine (7a) (15 mg, 0.044 mmol), tert-butyl4-hydroxypiperidine (38 mg, 0.18 mmol), and 95% NaH (11 mg, 0.46 mmol) at 50 °C. The title compound was obtained as a TFA salt after Boc-removal using TFA in DCM and HPLC purification (16 mg, 47.3%). 1H NMR (400 MHz, DMSO-d6) δ ppm 9.06 (s, 1H), 8.67 (s, 1H), 8.33−8.44 (m, 2H), 8.04 (d, J = 5.56 Hz, 1H), 7.78 (d, J = 9.09 Hz, 2H), 7.18 (d, J = 9.35 Hz, 2H), 4.92−5.03 (m, 1H), 3.79−4.00 (m, 2H), 3.52−3.61 (m, 2H), 3.27−3.34 (m, 2H), 3.10−3.22 (m, 4H), 2.98−3.06 (m, 2H), 2.89 (s, 3H), 2.19−2.30 (m, 2H), 1.94−2.08 (m, 2H). HRMS calcd for C24H31N6O [M + H]+ = 419.2559. Found [M + H]+ = 419.2542. 1-(4-(4-(3-(Cyclohexyloxy)pyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (8e). 8e was prepared according to general procedure 1 using 1-(4-(4-(3-fluoropyridin-4-yl)-1H-pyrazol1-yl)phenyl)-4-methylpiperazine (7a) (11 mg, 0.03 mmol), cyclohexanol (16 mg, 0.16 mmol), and 95% NaH (8 mg, 0.3 mmol) at 50 °C. Purification by HPLC yielded the title compound (6 mg, 44%). 1H NMR (400 MHz, DMSO-d6) δ 8.85 (s, 1H), 8.44 (s, 1H), 8.30 (s, 1H), 8.16 (d, J = 4.9 Hz, 1H), 7.71 (d, J = 4.9 Hz, 1H), 7.68 (d, J = 9.0 Hz, 2H), 7.08 (d, J = 9.1 Hz, 2H), 4.70−4.62 (m, 1H), 3.22−3.16 (m, 4H), 2.49−2.44 (m, 4H), 2.23 (s, 3H), 2.07−1.98 (m, 2H), 1.77−1.66 (m, 2H), 1.66−1.50 (m, 3H), 1.49−1.26 (m, 3H). HRMS calcd for C25H32N5O [M + H]+ = 418.2607. Found [M + H]+ = 418.2591. 1-(4-(4-(3-Isobutoxypyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4methylpiperazine (8f). 8f was prepared according to general procedure 1 using 1-(4-(4-(3-fluoropyridin-4-yl)-1H-pyrazol-1-yl)phenyl)-4-methylpiperazine (7a) (11 mg, 0.03 mmol), 2-methylpropan-1-ol (12 mg, 0.16 mmol), and 95% NaH (8 mg, 0.3 mmol) at 50 °C. Purification by HPLC yielded the title compound (9 mg, 71%). 1H NMR (400 MHz, DMSO-d6) δ ppm 8.87 (s, 1H), 8.40 (s, 1H), 8.30 (s, 1H), 8.20 (d, J = 5.05 Hz, 1H), 7.73 (d, J = 5.05 Hz, 1H), 7.67 (d, J = 9.09 Hz, 2H), 7.08 (d, J = 9.09 Hz, 2H), 4.04 (d, J = 6.32 Hz, 2H), 3.12−3.23 (m, 4H), 2.42−2.49 (m, 4H), 2.22 (s, 3H), 2.15−2.22 (m, 1H), 1.06 (d, J = 6.82 Hz, 6H). HRMS calcd for C23H30N5O [M + H]+ = 391.2450. Found [M + H]+ = 392.2432. tert-Butyl 4-(((4-Bromopyridin-3-yl)oxy)methyl)piperidine1-carboxylate (10). 10 was prepared using general procedure 1 from tert-butyl 4-(hydroxymethyl)piperidine-1-carboxylate (3.49 g, 16.19 mmol, (60% oil immersion, 0.78 g, 19.43 mmol) and 4-bromo-3fluoropyridine (1.90 g, 10.80 mmol) at 70 °C for 5 h. The crude mixture was purified by silica gel flash chromatography to give an oil (1.8 g, 45%). HRMS calcd for C16H24BrN2O3 [M + H]+ = 371.0970. Found [M + H]+ = 371.0968. tert-Butyl 4-(((4-(1H-Pyrazol-4-yl)pyridin-3-yl)oxy)methyl)piperidine-1-carboxylate (11). A combined suspension of tertbutyl 4-(((4-bromopyridin-3-yl)oxy)methyl)piperidine-1-carboxylate (10) (1.80 g, 4.85 mmol), 4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan2-yl)-1H-pyrazole (1.317 g, 6.79 mmol), dichloro-1,1′-bis(diphenylphosphino)ferrocenepalladium(II) DCM adduct (0.277 g, 0.34 mmol), and sodium carbonate (1.54 g, 14.54 mmol) in a mixture of dioxane (30 mL) and water (6 mL) was degassed for 5 min by a bubbling stream of N2 and then heated at 110 °C for 16 h. The reaction mixture was diluted with EtOAc, washed with saturated sodium bicarbonate, then brine. The organic portion was dried over Na2SO4, filtered, and concentrated in vacuo. The crude mixture was purified by silica gel flash chromatography (80g, 0−2% MeOH/ EtOAc) to give the title compound as an off-white solid (570 mg, 33%). 1H NMR (400 MHz, DMSO-d6) δ 13.15 (s, 1H), 8.36 (s, 1H), 8.30 (s, 1H), 8.14 (d, J = 5.0 Hz, 1H), 8.11 (s, 1H), 7.64 (d, J = 4.9 Hz, 1H), 4.08 (d, J = 6.3 Hz, 2H), 4.01 (dd, J = 13.8, 8.9 Hz, 2H), 2.78 (s, 2H), 2.15−2.01 (m, 1H), 1.79 (dd, J = 13.1, 3.8 Hz, 2H), 1.40 (s, 9H), 1.30−1.12 (m, 2H). MS calcd for C19H25N4O3 [M + H]+ = 358.44. Found [M + H]+ = 359.5. General Procedure 3 for N-Arylation. A sealed vial was charged with tert-butyl 4-(((4-(1H-pyrazol-4-yl)pyridin-3-yl)oxy)methyl)piperidine-1-carboxylate (11) (1.0 equiv), aryl halide (1.1 equiv), XPhos (0.2 equiv), Pd2(dba)3 (0.1 equiv), sodium tert-butoxide (1.6

washed with water, 4% aqueous sodium chloride, brine, the organic layer dried over Na2SO4, filtered, and the organic filtrate concentrated in vacuo. The crude mixture was purified using either flash chromatography (SiO2 and appropriate eluent) or preparative HPLC (acidic or basic methods found in the purification section) with concentration under reduced pressure or lyophilization until pure dried product was obtained. General Procedure 2 for Boc Removal. To a solution of Bocprotected alkoxypyridine, obtained from the reaction according to general procedure 1 (1.0 equiv), in an appropriate solvent (typically DCM but in some cases specified as MeOH), was added a solution of hydrogen chloride (HCl) in dioxane (4 M in dioxane) (5.0−15 equiv), and the resulting mixture was stirred for 2−24 h at room temperature. The reaction mixture was concentrated under vacuum giving a white solid (as the HCl salt) which was either used as is or free-based using NH4OH and further purified by the following methods: (1) silica gel flash chromatography (eluting with 1−5% NH4OH in MeOH/DCM); (2) preparative HPLC followed by lyophilization. 4-((4-(1-(4-(4-Methylpiperazin-1-yl)phenyl)-1H-pyrazol-4-yl)pyridin-3-yloxy)methyl)piperidine (8a). Reaction was performed according to general procedure 1 using 1-(4-(4-(3-fluoropyridin-4-yl)1H-pyrazol-1-yl)phenyl)-4-methylpiperazine 7a (200 mg, 0.59 mmol), tert-butyl 4-(hydroxymethyl)piperidine-1-carboxylate (255 mg, 1.19 mmol), and 60% NaH (95 mg, 2.4 mmol) in DMF (5 mL) at 70 °C for 16 h. The product was purified using silica gel flash chromatography (12 g, 0−5% MeOH/DCM) to obtain the Bocprotected intermediate as a white solid (281 mg, 88%). 1H NMR (400 MHz, chloroform-d) δ 8.39 (d, J = 0.7 Hz, 1H), 8.34 (s, 1H), 8.28 (d, J = 4.9 Hz, 1H), 8.18 (d, J = 0.7 Hz, 1H), 7.63−7.56 (m, 2H), 7.49 (d, J = 4.9 Hz, 1H), 7.08−7.01 (m, 2H), 4.22 (s, 2H), 4.09 (d, J = 6.4 Hz, 2H), 3.30 (s, 4H), 2.81 (s, 2H), 2.63 (s, 4H), 2.40 (s, 3H), 2.13 (d, J = 12.8 Hz, 1H), 1.92 (d, J = 13.2 Hz, 2H), 1.48 (d, J = 3.2 Hz, 9H), 1.38 (d, J = 11.1 Hz, 2H). This intermediate was deprotected using general procedure 2 (280 mg, 0.526 mmol) to yield 300 mg (quantitative) of the de-Boc product as an HCl. An 80 mg portion of this material was purified by HPLC under basic conditions to obtain the title compound as a white solid (32 mg, 14%). 1H NMR (400 MHz, DMSO-d6) δ 8.89 (s, 1H), 8.40 (s, 1H), 8.29 (s, 1H), 8.19 (d, J = 4.9 Hz, 1H), 7.74−7.67 (m, 3H), 7.08 (d, J = 9.1 Hz, 2H), 4.09 (d, J = 6.0 Hz, 2H), 3.24−3.15 (m, 4H), 3.09−3.03 (m, 2H), 2.71−2.64 (m, 2H), 2.48−2.43 (m, 4H), 2.23 (s, 3H), 2.05−1.90 (m, 1H), 1.81−1.68 (m, 2H), 1.29 (qd, J = 12.3, 4.0 Hz, 2H). HRMS calcd for C25H33N6O [M + H]+ = 433.2716. Found [M + H]+ = 433.2721. (1R,4R)-4-(((4-(1-(4-(4-Methylpiperazin-1-yl)phenyl)-1H-pyrazol-4-yl)pyridin-3-yl)oxy)methyl)cyclohexanamine (8b). 8b was prepared according to general procedure 1 using 7a (231 mg, 0.685 mmol), ((1R,4R)-4-aminocyclohexyl)methanol (206 mg, 1.24 mmol), and 60% NaH (120 mg, 3.0 mmol) in DMF (5 mL) at 70 °C overnight. Purification using flash silica gel chromatography (4 g, 5% NH4OH in methanol/DCM) gave crude product as an off-white solid. HPLC purification furnished the title compound (80 mg, 26%). 1H NMR (400 MHz, methanol-d4) δ 9.06 (s, 1H), 8.61 (d, J = 0.8 Hz, 1H), 8.52 (s, 1H), 8.44 (dd, J = 6.0, 0.8 Hz, 1H), 8.34 (d, J = 6.1 Hz, 1H), 7.82−7.75 (m, 2H), 7.26−7.20 (m, 2H), 4.27 (d, J = 5.9 Hz, 2H), 4.03−3.92 (m, 2H), 3.67 (d, J = 12.2 Hz, 2H), 3.23−3.06 (m, 4H), 3.01 (d, J = 5.5 Hz, 3H), 2.26−2.07 (m, 6H), 1.64−1.48 (m, 2H), 1.46−1.34 (m, 2H). HRMS calcd for C26H35N6O [M + H]+ = 447.2872. Found [M + H]+ = 447.2847. 1-Methyl-4-(4-(4-(3-(pyridin-4-ylmethoxy)pyridin-4-yl)-1Hpyrazol-1-yl)phenyl)piperazine (8c). 8c was prepared according to general procedure 1 using 1-(4-(4-(3-fluoropyridin-4-yl)-1H-pyrazol1-yl)phenyl)-4-methylpiperazine (7a) (15 mg, 0.044 mmol), pyridin4-ylmethanol (25 mg, 0.23 mmol), and 95% NaH (11 mg, 0.49 mmol) in DMF (2 mL) at 50 °C. Purification by HPLC yielded the title compound (4 mg, 21.1%). 1H NMR (400 MHz, DMSO-d6) δ ppm 8.89 (s, 1H), 8.63 (d, J = 5.31 Hz, 2H), 8.44 (s, 1H), 8.30 (s, 1H), 8.23 (d, J = 4.80 Hz, 1H), 7.76 (d, J = 5.05 Hz, 1H), 7.65 (d, J = 8.84 Hz, 2H), 7.54 (d, J = 5.31 Hz, 2H), 7.07 (d, J = 8.84 Hz, 2H), 5.48 (s, 2H), 3.12−3.24 (m, 4H), 2.43−2.49 (m, 4H), 2.23 (s, 3H). HRMS calcd for C25H27N6O [M + H]+ = 427.2246. Found [M + H]+ = 427.2228. J



Article

added, and the layers were separated. The aqueous layer was extracted (50 mL × 2) with EtOAc, and the combined organic layers were washed with brine and dried over Na2SO4 and concentrated under reduced pressure to give a crude solid. Purification via flash column chromatography (SiO2, 12g, 10−90% EtOAc/heptane over 25 min) gave the title compound as a white solid (373 mg, 24%). 1H NMR (400 MHz, DMSO-d6) δ 8.80 (s, 1H), 7.88 (s, 1H), 7.86−7.79 (m, 2H), 7.56−7.46 (m, 2H), 7.40−7.29 (m, 1H). MS calcd for C9H8BrN2 [M + H]+ = 223.0. Found [M + H]+ = 223.2. 3-Fluoro-4-(1-phenyl-1H-pyrazol-4-yl)pyridine (15). A combined suspension of 4-bromo-1-phenyl-1H-pyrazole (14) (373 mg, 1.67 mmol), 3-fluoro-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)pyridine (411 mg, 1.84 mmol), tripotassium phosphate (781 mg, 3.68 mmol) and (Pd(amphos)Cl2) (68 mg, 0.096 mmol) in a mixture of dioxane (10 mL) and water (1.0 mL) was degassed under nitrogen, then capped and heated to 95 °C for 4 h. The reaction mixture was cooled and diluted with EtOAc, washed with water (2×), brine, dried over Na2SO4, filtered, and concentrated to an oil. Purification by flash silica gel chromatography (12 g, EtOAc/heptane) gave the title compound as a light yellow solid (173 mg, 43%). 1H NMR (400 MHz, DMSO-d6) δ 9.14 (s, 1H), 8.64 (d, J = 2.8 Hz, 1H), 8.46 (dd, J = 5.0, 1.3 Hz, 1H), 8.37 (d, J = 1.9 Hz, 1H), 7.98−7.88 (m, 3H), 7.61−7.51 (m, 2H), 7.43−7.32 (m, 1H). HRMS calcd for C14H11FN3 [M + H]+ = 240.0937. Found [M + H]+ = 240.0948. Biochemical Activities. MELK kinase activity was determined using MELK protein containing both kinase domain and UBA domain (or full-length MELK, as noted) and KinEASE STK S1 peptide (Cisbio HTRF) as substrate. Phosphorylation of the peptide was measured by homogeneous time-resolved fluorescence technology using XL665 labeled streptavidin and anti-phosphoserine antibody conjugated to Eu3+ cryptate (Cisbio HTRF). For IC50 measurement, the compounds were serially diluted and incubated with MELK protein for 20 min at room temperature before being added to a reaction mixture. The final reaction conditions were 3 pM MELK protein, 300 nM peptide substrate, 20 μM ATP, 50 mM Hepes (pH 7.5), 5 mM MgCl2, 2 mM DTT, and 1% DMSO. The reaction was stopped in 20 min by adding equal volume of streptavidin-XL665/ antibody in detection buffer (CisBio HTRF). Following incubation at room temperature for 60 min, time-resolved fluorescence (Ex, 337; dual Em, 665/620) was measured by Wallac EnVision multilabel reader (PerkinElmer). IC50 values were calculated by nonlinear fourparameter fitting equations. Cell Lines and Culture Conditions. MDA-MB-468 and MCF7 cell lines were obtained from the ATCC and cultured in RPMI and EMEM, respectively, supplemented with 10% FBS and L-glutamine. MDA-MB-468 and MCF7 shRNA stable cell lines were generated following shRNA lentiviral infection and selection. Proliferation Assay. MDA-MB-468 and MCF7 cells were seeded in growth medium into 96-well plates at 1000 and 4000 cells/well, respectively. Sixteen hours after plating, compounds were added to obtain the indicated concentrations, in triplicate, and incubated for 7 days. For each well, ATPLite reagent (PerkinElmer) was added and incubated according to the manufacturer’s instructions. Luminescence was measured on an Envision 2104 multilabel plate reader (PerkinElmer). FACS Cell Cycle Analysis. At harvest time point, growth media, PBS wash, and trypsinized cells from each condition were combined to one conical tube. 0.5 × 106 cells were transferred to 15 mL tube and pelleted at 1100 rpm for 5 min. Cells were washed twice with PBS containing 2% FBS and then resuspended in 0.3 mL of PBS containing 2% FBS. Cells were fixed by adding 3 mL of −20 °C prechilled 70% ethanol dropwise while vortexing and then stored at 4 °C overnight. Before staining, fixed cells were pelleted at 1100 rpm for 5 min at 4 °C and then washed twice with 0.5% BSA in PBS. The pellet was resuspended and stained with 200−500 μL of propidium iodide (50 μg/mL PI Biosure 1021) solution containing 10 μg/mL DNase-free RNase A (Roche 11579681001). After a 30 min incubation at room temperature, the samples were run though a BD FACSCanto analyzer with proper gating. At least 10 000 single cells were collected. The data were analyzed with the FlowJo software.

equiv), and toluene (0.08 M). The mixture was degassed for 2 min with N2 and then heated to 105 °C for 16 h. The reaction mixture was cooled and filtered and then the filtrate concentrated in vacuo. The residue was dissolved in MeOH, a hydrogen chloride solution in dioxane (4 N in dioxane) added, and the reaction was stirred at room temperature until completion. The mixture was concentrated in vacuo, and the crude product was dissolved in 7 N NH3/MeOH, then reconcentrated in vacuo. Methanol was added and the crude solution was purified to afford the desired product. 4-(1-Phenyl-1H-pyrazol-4-yl)-3-(piperidin-4-ylmethoxy)pyridine (12a). 12a was prepared from 3-fluoro-4-(1-phenyl-1Hpyrazol-4-yl)pyridine (15) (100 mg, 0.48 mmol), tert-butyl 4(hydroxymethyl)piperidine-1-carboxylate (157 mg, 0.73 mmol), and 60% NaH (50 mg, 1.25 mmol) in DMF (1.5 mL) heated to 65 °C for 3 h. Reaction mixture was deprotected according to general procedure 2 and purified by HPLC to obtain the title compound (127 mg, 68%) as a TFA salt. 1H NMR (400 MHz, DMSO-d6) δ 9.06 (s, 1H), 8.44 (s, 1H), 8.37 (s, 1H), 8.22 (d, J = 4.9 Hz, 1H), 7.94−7.86 (m, 2H), 7.76 (d, J = 4.9 Hz, 1H), 7.60−7.50 (m, 2H), 7.41−7.31 (m, 1H), 4.12 (d, J = 6.1 Hz, 2H), 3.05 (dt, J = 12.1, 3.2 Hz, 2H), 2.60 (td, J = 12.0, 2.6 Hz, 2H), 2.12−1.97 (m, 1H), 1.79 (dd, J = 13.4, 3.5 Hz, 2H), 1.33 (qd, J = 12.2, 4.0 Hz, 2H). HRMS calcd for C20H23N4O [M + H]+ = 335.1872. Found [M + H]+ = 335.1872. 4-(1-(2-Fluorophenyl)-1H-pyrazol-4-yl)-3-(piperidin-4ylmethoxy)pyridine (12b). 12b was prepared according to general procedure 3 using tert-butyl 4-(((4-(1H-pyrazol-4-yl)pyridin-3-yl)oxy)methyl)piperidine-1-carboxylate (11) (50 mg, 0.139 mmol) and 1-bromo-2-fluorobenzene (30 mg, 0.17 mmol). HPLC purification using acidic method afforded the title compound (26 mg, 41%) as a TFA salt. 1H NMR (400 MHz, DMSO-d6) δ 8.95 (d, J = 2.5 Hz, 1H), 8.66 (bs, 1H), 8.61 (s, 1H), 8.58 (s, 1H), 8.43 (d, J = 5.4 Hz, 1H), 8.46−8.35 (bs, 1H), 8.10 (d, J = 5.5 Hz, 1H), 7.90 (td, J = 7.9, 1.6 Hz, 1H), 7.57−7.48 (m, 2H), 7.47−7.39 (m, 1H), 4.23 (d, J = 6.4 Hz, 2H), 3.41−3.31 (m, 2H), 2.97 (q, J = 11.3 Hz, 2H), 2.36−2.24 (m, 1H), 2.05−1.96 (m, 2H), 1.60−1.45 (m, 2H). HRMS calcd for C20H22FN4O [M + H]+ = 353.1778. Found [M + H]+ = 353.1240. 4-(4-(4-(3-(Piperidin-4-ylmethoxy)pyridin-4-yl)-1H-pyrazol1-yl)phenyl)morpholine (12c). 12c was prepared according to general procedure 3 using tert-butyl 4-(((4-(1H-pyrazol-4-yl)pyridin-3yl)oxy)methyl)piperidine-1-carboxylate (11) (50 mg, 0.139 mmol) and 4-(4-bromophenyl)morpholine (43.9 mg, 0.181 mmol, 1.3 equiv). HPLC purification using basic method afforded the title compound (23 mg, 40%). 1H NMR (400 MHz, DMSO-d6) δ 8.90 (s, 1H), 8.41 (s, 1H), 8.30 (s, 1H), 8.20 (d, J = 5.0 Hz, 1H), 7.75−7.70 (m, 3H), 7.13−7.06 (m, 2H), 4.09 (d, J = 6.0 Hz, 2H), 3.84−3.70 (m, 4H), 3.22−3.14 (m, 4H), 3.00 (dt, J = 12.2, 3.3 Hz, 2H), 2.55 (dd, J = 11.9, 2.6 Hz, 2H), 1.99 (bs, 1H), 1.82−1.68 (m, 2H), 1.29 (qd, J = 12.1, 4.1 Hz, 2H). HRMS calcd for C24H30N5O2 [M + H]+ = 420.2400. Found [M + H]+ = 420.2387. 4-(3-Fluoro-4-(4-(3-(piperidin-4-ylmethoxy)pyridin-4-yl)-1Hpyrazol-1-yl)phenyl)morpholine (12d). 12d was prepared according to general procedure 3 using tert-butyl 4-(((4-(1H-pyrazol-4yl)pyridin-3-yl)oxy)methyl)piperidine-1-carboxylate (11) (100 mg, 0.28 mmol) and 4-(4-bromo-3-fluorophenyl)morpholine (94 mg, 0.36 mmol). HPLC purification using basic method afforded the title compound (6 mg, 5%). 1H NMR (400 MHz, DMSO-d6) δ 8.65 (d, J = 2.4 Hz, 1H), 8.41 (s, 1H), 8.38 (s, 1H), 8.19 (d, J = 4.9 Hz, 1H), 7.74 (d, J = 4.9 Hz, 1H), 7.65 (t, J = 9.1 Hz, 1H), 7.04 (dd, J = 15.2, 2.6 Hz, 1H), 6.93 (dd, J = 9.1, 2.6 Hz, 1H), 4.08 (d, J = 6.3 Hz, 2H), 3.77− 3.71 (m, 4H), 3.24−3.19 (m, 4H), 3.02−2.95 (m, 2H), 2.6−2.4 (m, 2H) (partially overlaps with solvent signal), 2.04−1.91 (m, 1H), 1.81− 1.72 (m, 2H), 1.30−1.17 (m, 2H). HRMS calcd for C24H29FN5O2 [M + H]+ = 438.2305. Found [M + H]+ = 438.2281. 4-Bromo-1-phenyl-1H-pyrazole (14). To 1-phenylpyrazole (1.00 g, 6.94 mmol) in acetic acid (10 mL) was added bromine (0.360 mL, 6.99 mmol) in acetic acid (10.00 mL). This mixture was heated to 100 °C in a thick-walled screw cap reaction vessel for 16 h. The material was cooled to ambient temperature, poured into ice in a 500 mL beaker, and excess saturated aqueous NaHCO3 was added until all the acetic acid had been quenched. EtOAc (100 mL) was K



Article

Crystallography. MELK (3-330) with a carboxy-terminal sixhistidine tag was expressed in E. coli and subsequently purified via Niaffinity resin, anion exchange, and gel filtration. The pure protein was concentrated to approximately 5 mg/mL in 20 mM Tris, pH 8.0, 260 mM NaCl, 2 mM TCEP, and 5% (v/v) glycerol. Apo MELK crystallized from drops containing equal volume of protein and reservoir solution containing 100 mM Hepes, pH 7.6, 0.2 M NaCl, 4.5−16.5% PEG 3350 at 4 °C. For soaks, crystals were harvested and transferred to drops containing 0.25−1 mM compound in 100 mM Hepes, pH 7.6, 0.2 M NaCl, 15−25% PEG 3350 for 12−36 h. For cocrystallization experiments, the protein was incubated with compound for 30 min prior to mixing with the reservoir solution and then crystallized and harvested as described above. Crystals were transferred to cryoprotection solution containing an additional 15− 25% glycerol and flash cooled in liquid nitrogen. Data were collected at 100 K using a Pilatus 6M detector and synchrotron radiation (λ = 1.0000 Å) at the IMCA-CAT beamline 17-ID of the Advanced Photon Source at Argonne National Laboratory or at the X10SA beamline of Swiss Light Source at Paul Scherrer Institut and processed using Autoproc (Global Phasing Ltd.). MELK/ligand cocrystals (obtained from soaking or cocrystallization) are in the space group P212121 with one molecule in the asymmetric unit and diffract to a typical resolution of 1.8−2 Å. Structures were solved by molecular replacement using an internal MELK apo structure originally determined by starting with a MARK3 homology model (PDB code 2QNJ). The model was built using iterative cycles of model building in COOT24 and refined in BUSTER (Global Phasing Ltd.). Individual B-factors were refined using an overall anisotropic B-factor refinement along with a bulk solvent correction. Data collection and refinement statistics are excellent. Structures obtained using cocrystallization and soaking methods with the same compound were identical. The structure factors and coordinates have been deposited in the RCSB PDB with accession codes 5IH8, 5IH9, 5IHA, and 5IHC. Mouse Pharmacokinetic Studies. All animal related procedures were conducted under a Novartis IACUC approved protocol in compliance with Animal Welfare Act regulations and the Guide for the Care and Use of Laboratory Animals. Mice were housed in a temperature- and humidity-controlled animal facility with ad libitum access to food and water and acclimated for at least 3 days before experimental manipulation and with a 12 h light/12 h dark cycle. For pharmacokinetic studies described in Table 3, the intravenous and oral dose was prepared in a solution containing 5% ethanol, 100% PG, 5% CremophorEL, and 80% PBS. The subcutaneous dose was formulated in 10% PG and 25% (20%, v/v) Solutol. Plasma samples were collected at specified time points and stored frozen (−20 °C) until compound analysis. An LC−MS/MS method was used to quantitate MELK8a and 12a drug levels in plasma. All pharmacokinetic (PK) parameters were derived from concentration−time data by noncompartmental analyses using WinNonlin Phoenix, version 6.4 (Certara, St. Louis, MO).

■

the atomic coordinates and experimental data upon article publication.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 617-373-8818. E-mail: [email protected]. Present Addresses †

B.B.T.: Relay Therapeutics, 215 1st Street, 3rd Floor, Cambridge, MA 02142, U.S. ‡ D.P.: Yale University, 333 Cedar Street, New Haven, CT 06519, U.S. § Y.Y.: Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, The Planum Institute, 3B Blegdamsvej 3B, 2200 Copenhagen N, Denmark. ∥ W.S.: AstraZeneca, 35 Gatehouse Drive, Waltham, MA 02451, U.S. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare the following competing financial interest(s): The authors declare financial interests as current or former Novartis employees and shareholders.

■

ACKNOWLEDGMENTS We thank Drs. Dallas Bednarczyk and Sebastien Ronseaux for mechanistic PK studies and Caco-2 permeability data. Use of the IMCA-CAT beamline 17-ID at the Advanced Photon Source was supported by the companies of the Industrial Macromolecular Crystallography Association through a contract with Hauptman-Woodward Medical Research Institute. Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract DE-AC02-06CH11357. We acknowledge the Paul Scherrer Institut, Villigen, Switzerland, for provision of synchrotron radiation beamtime at beamline X10SA of the SLS. The synthetic route development and scaleup of 5 were done at Pharmacore.

■

ABBREVIATIONS USED MELK, maternal embryonic leucine zipper kinase; shRNA, short hairpin ribonucleic acid; SNF1, sucrose nonfermenting 1; AMPK, adenosine monophosphate-activated protein kinase; PARP, poly ADP ribose polymerase; VS, virtual screening; LE, ligand efficiency; DBAD, di-tert-butyl azodicarboxylate; Papp, apparent permeability; ER, extraction ratio; PK, pharmacokinetic; THF, tetrahydrofuran; EtOAc, ethyl acetate; MBTE, methyl tert-butyl ether; MeOH, methanol

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.6b00052. Molecular formula strings (CSV) Genetic knockdown of MELK by two different MELKtargeting shRNAs in MDA-MB-468, MCF-7 (colony formation assay, immunoblotting, quantitative RT-PCR, cell cycle effect), quantitation of MELK expression levels in MDA-MB-468 and MCF-7, IC50 curves for 8a and 8b, mouse maximum-tolerated dose (MTD) and PK studies (PDF)

■

REFERENCES

(1) Weiss, W. A.; Taylor, S. S.; Shokat, K. M. Recognizing and exploiting differences between RNAi and small-molecule inhibitors. Nat. Chem. Biol. 2007, 3, 739−744. (2) Sweis, R. F. Target (In)Validation: A Critical, Sometimes Unheralded, Role of Modern Medicinal Chemistry. ACS Med. Chem. Lett. 2015, 6, 618−621. (3) Lizcano, J. M.; Goransson, O.; Toth, R.; Deak, M.; Morrice, N. A.; Boudeau, J.; Hawley, S. A.; Udd, L.; Makela, T. P.; Hardie, D. G.; Alessi, D. R. LKB1 is a master kinase that activates 13 kinases of the AMPK subfamily, including MARK/PAR-1. EMBO J. 2004, 23, 833− 843.

Accession Codes

PDB codes 5IH8, 5IH9, 5IHA, and 5IHC correspond respectively to 1a, 8a, 8f, and 12b. The authors will release L



Article

(4) Beullens, M.; Vancauwenbergh, S.; Morrice, N.; Derua, R.; Ceulemans, H.; Waelkens, E.; Bollen, M. Substrate specificity and activity regulation of protein kinase MELK. J. Biol. Chem. 2005, 280, 40003−40011. (5) Gray, D.; Jubb, A. M.; Hogue, D.; Dowd, P.; Kljavin, N.; Yi, S.; Bai, W.; Frantz, G.; Zhang, Z.; Koeppen, H.; de Sauvage, F. J.; Davis, D. P. Maternal embryonic leucine zipper kinase/murine protein serinethreonine kinase 38 is a promising therapeutic target for multiple cancers. Cancer Res. 2005, 65, 9751−9761. (6) Gu, C.; Banasavadi-Siddegowda, Y. K.; Joshi, K.; Nakamura, Y.; Kurt, H.; Gupta, S.; Nakano, I. Tumor-specific activation of the CJUN/MELK pathway regulates glioma stem cell growth in a p53dependent manner. Stem Cells 2013, 31, 870−881. (7) Joshi, K.; Banasavadi-Siddegowda, Y.; Mo, X.; Kim, S. H.; Mao, P.; Kig, C.; Nardini, D.; Sobol, R. W.; Chow, L. M.; Kornblum, H. I.; Waclaw, R.; Beullens, M.; Nakano, I. MELK-dependent FOXM1 phosphorylation is essential for proliferation of glioma stem cells. Stem Cells 2013, 31, 1051−1063. (8) Rhodes, D. R.; Yu, J.; Shanker, K.; Deshpande, N.; Varambally, R.; Ghosh, D.; Barrette, T.; Pandey, A.; Chinnaiyan, A. M. Large-scale meta-analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 9309−9314. (9) Pickard, M. R.; Green, A. R.; Ellis, I. O.; Caldas, C.; Hedge, V. L.; Mourtada-Maarabouni, M.; Williams, G. T. Dysregulated expression of Fau and MELK is associated with poor prognosis in breast cancer. Breast Cancer Res. 2009, 11, R60. (10) Liu, G.; Yuan, X.; Zeng, Z.; Tunici, P.; Ng, H.; Abdulkadir, I. R.; Lu, L.; Irvin, D.; Black, K. L.; Yu, J. S. Analysis of gene expression and chemoresistance of CD133+ cancer stem cells in glioblastoma. Mol. Cancer 2006, 5, 67. (11) Ryu, B.; Kim, D. S.; Deluca, A. M.; Alani, R. M. Comprehensive expression profiling of tumor cell lines identifies molecular signatures of melanoma progression. PLoS One 2007, 2, e594. (12) Marie, S. K.; Okamoto, O. K.; Uno, M.; Hasegawa, A. P.; ObaShinjo, S. M.; Cohen, T.; Camargo, A. A.; Kosoy, A.; Carlotti, C. G., Jr.; Toledo, S.; Moreira-Filho, C. A.; Zago, M. A.; Simpson, A. J.; Caballero, O. L. Maternal embryonic leucine zipper kinase transcript abundance correlates with malignancy grade in human astrocytomas. Int. J. Cancer 2008, 122, 807−815. (13) Wang, Y.; Lee, Y. M.; Baitsch, L.; Huang, A.; Xiang, Y.; Tong, H.; Lako, A.; Von, T.; Choi, C.; Lim, E.; Min, J.; Li, L.; Stegmeier, F.; Schlegel, R.; Eck, M. J.; Gray, N. S.; Mitchison, T. J.; Zhao, J. J. MELK is an oncogenic kinase essential for mitotic progression in basal-like breast cancer cells. eLife 2014, 3, e01763. (14) Chung, S.; Suzuki, H.; Miyamoto, T.; Takamatsu, N.; Tatsuguchi, A.; Ueda, K.; Kijima, K.; Nakamura, Y.; Matsuo, Y. Development of an orally-administrative MELK-targeting inhibitor that suppresses the growth of various types of human cancer. Oncotarget. 2012, 3, 1629−1640. (15) Canevari, G.; Re Depaolini, S.; Cucchi, U.; Bertrand, J. A.; Casale, E.; Perrera, C.; Forte, B.; Carpinelli, P.; Felder, E. R. Structural insight into maternal embryonic leucine zipper kinase (MELK) conformation and inhibition toward structure-based drug design. Biochemistry 2013, 52, 6380−6387. (16) Johnson, C. N.; Berdini, V.; Beke, L.; Bonnet, P.; Brehmer, D.; Coyle, J. E.; Day, P. J.; Frederickson, M.; Freyne, E. J.; Gilissen, R. A.; Hamlett, C. C.; Howard, S.; Meerpoel, L.; McMenamin, R.; Patel, S.; Rees, D. C.; Sharff, A.; Sommen, F.; Wu, T.; Linders, J. T. Fragmentbased discovery of type I inhibitors of maternal embryonic leucine zipper kinase. ACS Med. Chem. Lett. 2015, 6, 25−30. (17) Johnson, C. N.; Adelinet, C.; Berdini, V.; Beke, L.; Bonnet, P.; Brehmer, D.; Calo, F.; Coyle, J. E.; Day, P. J.; Frederickson, M.; Freyne, E. J.; Gilissen, R. A.; Hamlett, C. C.; Howard, S.; Meerpoel, L.; Mevellec, L.; McMenamin, R.; Pasquier, E.; Patel, S.; Rees, D. C.; Linders, J. T. Structure-Based Design of Type II Inhibitors Applied to Maternal Embryonic Leucine Zipper Kinase. ACS Med. Chem. Lett. 2015, 6, 31−36.

(18) Martin, E.; Mukherjee, P.; Sullivan, D.; Jansen, J. Profile-QSAR: a novel meta-QSAR method that combines activities across the kinase family to accurately predict affinity, selectivity, and cellular activity. J. Chem. Inf. Model. 2011, 51, 1942−1956. (19) Martin, E. J.; Sullivan, D. C. Surrogate AutoShim: predocking into a universal ensemble kinase receptor for three dimensional activity prediction, very quickly, without a crystal structure. J. Chem. Inf. Model. 2008, 48, 873−881. (20) Furtmann, N.; Hu, Y.; Bajorath, J. Comprehensive analysis of three-dimensional activity cliffs formed by kinase inhibitors with different binding modes and cliff mapping of structural analogues. J. Med. Chem. 2015, 58, 252−264. (21) Desai, M. C.; Vincent, L. A.; Rizzi, J. P. Importance of parallel vectors and "hydrophobic collapse" of the aligned aromatic rings: discovery of a potent substance P antagonist. J. Med. Chem. 1994, 37, 4263−4266. (22) Faller, B.; Spanka, C.; Sergejew, T.; Tschinke, V. Improving the oral bioavailability of the iron chelator HBED by breaking the symmetry of the intramolecular H-bond network. J. Med. Chem. 2000, 43, 1467−1475. (23) Davezac, N.; Baldin, V.; Blot, J.; Ducommun, B.; Tassan, J. P. Human pEg3 kinase associates with and phosphorylates CDC25B phosphatase: a potential role for pEg3 in cell cycle regulation. Oncogene 2002, 21, 7630−7641. (24) Emsley, P.; Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 2126−2132.

M


Toward the Validation of Maternal Embryonic Leucine Zipper Kinase

Recommend Documents