Small-molecule and mutational analysis of allosteric Eg5 inhibition by monastrol
© Maliga and Mitchison. 2006
Received: 29 June 2005
Accepted: 27 February 2006
Published: 27 February 2006
Skip to main content
© Maliga and Mitchison. 2006
Received: 29 June 2005
Accepted: 27 February 2006
Published: 27 February 2006
A recent crystal structure of monastrol in a ternary complex with the kinesin Eg5 motor domain highlights a novel, induced-fit drug binding site at atomic resolution. Mutational obliteration of the monastrol binding site results in a monastrol-resistant, but otherwise catalytically active Eg5 motor domain. However, considering the conformational changes at this site, it is unclear what specific interactions stabilize the interaction between monastrol and the Eg5 motor domain.
To study the molecular complementarity of the monastrol-Eg5 interaction, we used a combination of synthetic chemistry and targeted mutations in Eg5 to measure the contribution of specific contacts to inhibition of Eg5 in vitro and in cultured cells. Structure-activity data on chemical derivatives, sequence analysis of Eg5 homologs from different species, and the effect of mutations near the drug binding site were consistent with the crystal structure.
The mechanism of monastrol revealed by our data rationalizes its specificity for Eg5 over other kinesins and highlights a potential mechanism of drug resistance for anti-cancer therapy targeting this site in Eg5.
Kinesins are a diverse family of microtubule-based motor proteins important for intracellular transport and cell division in all eukaryotes [1, 2]. Genetic and biochemical dissection of kinesin function implicates specific kinesins in the trafficking of organelles , signaling complexes , and vesicular cargo . During mitosis and cytokinesis, kinesins are essential for microtubule dynamics regulation, assembly and maintenance of bipolar spindles, and accurate chromosome segregation . Assessing the precise contributions of kinesins to highly dynamic processes during both interphase and mitosis is challenging. Genetic tools such as siRNA are general and specific, but lack temporal resolution and reversibility necessary for detailed analysis of dynamic processes. Reversible, small molecule inhibitors of both microtubule- and actin-based motors are proving to be invaluable tools with which to study their functions during cell division [7, 8]. Monastrol, a specific inhibitor of the BimC class kinesin Eg5 (also called kinesin-5 or kinesin spindle protein, KSP) , has permitted more critical analyses of Eg5 function during spindle assembly [9, 10] and as a reversible agent to synchronize cells in metaphase . Furthermore, inhibitors of Eg5 and other mitotic kinesins are plausible anti-cancer drugs now under development and testing [11, 12]. They work by disrupting the mitotic spindle, arresting cancer cells in mitosis, and thus triggering apoptosis .
BimC class kinesins are widely required for bipolar spindle assembly during mitosis and meiosis. They are homotetrameric, plus-end directed kinesins that associate with the spindle during mitosis [14–16] and their inhibition or removal generally results in spindle collapse [7, 17, 18]. Spindles are dynamic, bipolar arrays of microtubules that are maintained in part by a balance of forces between oppositely directed motor proteins [19, 20], and it is likely that Eg5 provides the forces that drives the two spindle poles apart from each other. Monastrol reversibly inhibits microtubule gliding by Eg5 and causes spindle collapse in cells . Eg5 supports anti-parallel sliding of microtubules in vitro  and during poleward flux in Xenopus extract spindles, a process completely inhibited by monastrol . Biochemical analysis of microtubule binding  and gliding by Eg5 in vitro  indicates monastrol induces microtubule release and therefore complete loss of Eg5 function as a microtubule cross-linker and motor.
Guided by the atomic structure of the Eg5-ADP-monastrol ternary complex, we used a combination of synthetic chemistry, targeted mutagenesis, and protein biochemistry to characterize the interaction between the Eg5 motor domain and monastrol at the atomic level. Initially, we confirm that the structure is consistent with species-specificity of monastrol in cells. A structure-activity relationship for monastrol and mutation of target binding residues tests the importance of individual amino acid residues and chemical substituents to the protein-drug interaction.
We crystallized the monastrol-ADP-Eg5 motor complex, and solved its structure, independent of the published work  to 1.8 Å resolution. Our structure confirms the conformational changes in the Eg5 motor domain induced by monastrol binding in protein crystals obtained using distinct precipitant conditions. The details of crystallization and atomic structure determination are available as supplemental information (see Additional files 1 and 2).
All of the point-mutants, with the exception of Y211M, decreased the drug sensitivity of Eg5. Y211M actually increased the monastrol sensitivity of Eg5, perhaps because flexibility of the methionine side chain accommodates a more favorable hydrophobic interaction with the drug. Hydrophobic packing between the thiourea and L214 appears to be crucial for drug binding, illustrated by the dramatic effect of even a subtle L214I mutation on IC50. L214T mutant was resistant to monastrol, as would be expected, but also to derivative 2 (data not shown) with a compensating hydrophilic urea. Hydrophobic contacts from R119 are required for monastrol binding, but electrostatic interactions also appear important (Figure 4B). In summary, our mutational changes near the binding site are consistent with the Eg5-monastrol structure and provide a number of catalytically active mutants of Eg5 that might be useful to examine spindle assembly and drug resistance in Eg5 by expression in cells.
Monastrol is a convenient reagent for mitosis research, but its specificity hitherto has not been established. Experiments designed to probe the role of Eg5 in the mitotic spindle and post-mitotic neurons are predicated on the assumption that monastrol is a specific inhibitor of Eg5 [29, 30]. It is reassuring to note that no kinesin protein contains sequences in loop 5 and α3 that are part of the monastrol binding site. Loop 5 is highly divergent among different kinesin classes , and the sequence of α3 near the monastrol binding site is not conserved. Indeed, most kinesins project bulky amino acid side chains into the putative monastrol binding pocket that would preclude drug binding (Figure 3A). Thus, monastrol should be highly selective among kinesins. However, we note it could easily target unexpected proteins. Its affinity is fairly low by drug standards, and it is often used on cells at 100 μM or more. At such high concentrations, any hydrophobic molecule could bind multiple targets. As a cell biology reagent, monastrol should be replaced with more potent compounds as soon as they are available. One such reagent is S-trityl cysteine, which is inexpensive, more potent than monastrol and appears to target a similar site in Eg5 . In the meantime, pharmacological SAR of different Eg5 inhibitors is a useful approach to implicate a specific role for Eg5 in biological processes .
Monastrol binds a cryptic site that would not have been discovered by virtual screening methods previously employed to find novel small-molecule kinesin inhibitors . It is also notable that this is a flexible drug binding site, which is also the target for the binding of S-trityl cysteine  and 3,5-diaryl-4,5-dihydropyrazoles , compounds that look very dissimilar in terms of overall shape and charge. While this allosteric site is fortuitous for drug discovery, its evolutionary conservation among vertebrates leads us to speculate that it is required for Eg5 function in vivo, either as part of the conformational cycle of the motor domain, or more speculatively by binding an unknown endogenous ligand that could be a protein or a small molecule. Our preliminary small molecule SAR did not yield more potent monastrol derivatives, but highlighted some of the functional groups contributing to Eg5 inhibition. For instance, the specific hydrogen-bonding interaction between Eg5 and the phenol group of monastrol as well as hydrophobic packing of the thiophenol appears crucial for drug binding (Figure 2B), similar to the interactions observed between dihydropyrazoles and Eg5 . A report published during revision of this manuscript described a SAR for monastrol derivatives consistent with our results. However, some inhibitors displaying weak activity in their assays, most notably derivative 2 (Figure 2B) and 4e, were completely inactive in our hands. The discrepancy can be explained by the higher sensitivity of their assay .
Although mutations at this site permit microtubule-stimulated ATP hydrolysis by the motor domain of Eg5, their effect on the kinetic parameters KmATP, K0.5MT, kcat, and IC50 indicate subtle effects on its catalytic cycle. Monastrol decreases the microtubule affinity of Eg5, and one might expect mutations that lower K0.5MT to raise IC50 for monastrol. This may be true for the L214I, a subtle mutation that would effect transitions between ordered and disordered states of α3 . A surprising result of our mutational analysis showed that mutating R119 greatly elevated KmATP (Figure 4B), indicating a weakened affinity for ATP. In light of a proposed model whereby monastrol stabilizes a pre-ATP hydrolysis conformation of the Eg5 motor domain , this merits further investigation using non-equilibrium enzyme kinetics. The mutant L214T, which introduces a hydrophilic residue into a previously hydrophobic environment, lowers kcat, as well as overall protein expression (data not shown), as might be expected. Systematic mutational analysis around the monastrol binding sight might identify mutants with discrete effects on microtubule and ATP binding to help dissect the catalytic cycle of kinesin motor proteins.
The monastrol binding site is a promising target for anti-cancer drug development, but our data showing that relatively subtle mutation confer resistance to drug binding without blocking enzymatic function sound a note of caution; Eg5 has potential for sporadic mutation to result in clinical resistance. Pressure to conserve residues at a substrate binding pocket is one explanation for the selectivity of non-competitive inhibitors among related proteins as compared to substrate-competitive inhibitors . The clinical correlate of this biological phenomenon is that non-competitive inhibitors, exemplified by non-nucleoside reverse transcriptase inhibitors (NNRTIs) to treat HIV infection, display fewer non-specific side effects but clinical resistance is easily acquired by mutations in the target protein . Monastrol is a prototype anti-mitotic agent that displays many of the same benefits and limitations of NNRTIs. One obvious concern, if Eg5 inhibitors developed as anti-mitotic drugs also target this site, is the rapid appearance of resistant clones, especially in the context of DNA damaging agents employed clinically. The importance of the monstrol binding site could be explored by gene replacement experiment, swapping drug resistant point mutants for the wild-type protein. Previously such experiments have been hard in vertebrate cells, but recent progress with SiRNA and bacterial artificial chromosome technologies' makes them feasible, even in human cancer cells .
The published crystal structure of monastrol in a ternary complex with the Eg5 motor domain and ADP , as well as our own structural studies arrived independently, are consistent with the structure-activity relationship of monastrol against purified, recombinant Eg5 motor domain in vitro and in tissue culture cells. Consistent with sequence homology at the drug binding site, we demonstrate the monastrol does not target Klp-61F, the Eg5 homolog in D. melanogaster, but arrests D. rerio cells in mitosis. Finally, we performed mutational analysis at drug-binding residues which high-lights flexibility at the drug binding site, and suggests a mechanism for genetic resistance to monastrol, and possibly more potent Eg5 inhibitors currently being developed as anti-mitotic agents in cancer chemotherapy.
Reagents were obtained from Sigma-Aldrich, unless otherwise indicated. Experimental stocks of taxol (10 mM), PMSF (100 mM), benzimidine hydrochloride (200 mM), 1000 × LPC (10 mg/ml each of leupeptin, pepstatin, and chymostatin), and monastrol derivatives (200 mM) were prepared as DMSO solutions, aliquoted, and stored at -20°C until use. KC25, the standard buffer used in biochemical experiments described below consists of 20 mM PIPES-KOH (pH 6.8), 25 mM KCl. All tissue culture media were supplemented with heat inactivated fetal calf serum (HyClone), penicillin and streptomycin (Cellgro, Mediatech).
We determined the atomic structure of the Eg5-ADP-monastrol complex using multi-wavelength anomalous dispersion to a resolution of 1.80 Å. Although our protein crystals grew under different precipitant conditions than those described by Yan and colleagues, the atomic structure of the Eg5 motor domain were essentially identical to the published results . The atomic coordinates are deposited in the protein database as 1X88. The details of our crystallographic refinement are available as supporting information. Figures in this report were prepared using PyMol  based on the structure reported by Yan, et. al. , (PDB: 1Q0B).
Conditions for growing the following cell lines are as described: D. rerio AB9 (37), D. melanogaster KC167 , and BS-C-1 green monkey kidney cells . For immunofluorescence microscopy, cells were grown on 12 mm circular coverslips to 30 % confluence, treated medium supplemented with drug and 0.2% DMSO, or a DMSO alone, for 6 hours, fixed with formaldehyde, and stained for DNA and microtubules, as described . D. melanogaster KC167 cells were cultured in Schneider's Drosophila Medium (GIBCO) whereas D. rerio AB9 cells and BS-C-1 green monkey kidney cells were cultured at 37°C in Dulbecco's Minimal medium in the presence of 10 % heat inactivated fetal calf serum.
The motor domain from D. melanogaster Klp-61F (amino acid residues 1 to 364) with an engineered C-terminal His6 affinity tag was sub-cloned by PCR from a cDNA library (gift of Christine Field) into pRSETa as described . The sequence agreed with the published sequence for the Klp-61F (NP_476818.1 GI:17136642). Expression and purification of the Klp-61F motor domain was performed, as described .
Monastrol derivatives were synthesized via the Biginelli condensation, purified by silica-gel chromatography, and characterized by 1H-NMR and tandem liquid chromatography/electrospray ionization mass spectrometry, as described . The 1H-NMR resonances in d6-DMSO for each derivative, unless previously reported, are listed below:
Derivative 1: [ethyl 6-methyl-4-(3-hydroxyphenyl)-2-thioxo-1-methyl-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 9.79 (s, 1H), 9.44 (s, 1H), 7.11 (t, J = 8.5 Hz, 1H), 6.6 (m, 3H), 5.13 (d, J = 4 Hz, 1H), 4.11 (q, J = 7.5 Hz, 2H), 3.48 (s, 3H), 2.50 (s, 3H), 1.18 (t, J = 7.5 Hz, 3H).
Derivative 2: [ethyl 6-methyl-4-(3-hydroxyphenyl)-2-oxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 9.35 (s, 1H), 9.14 (s, 1H), 7.67 (s, 1H), 7.09 (t, J = 8 Hz, 1H), 6.67 (d, J = 7 Hz, 1H), 6.66 (s, 1H), 6.62 (d, J = 7.5 Hz, 1H), 5.06 (d, J = 3 Hz, 1H), 4.00 (q, J = 6.5 Hz, 2H), 2.23 (s, 3H), 1.13 (t, J = 7.5 Hz, 3H).
Derivative 4a: [ethyl 6-methyl-4-phenyl-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 10.33 (s, 1H), 9.64 (s, 1H), 7.35 (t, J = 7.5 Hz, 2H), 7.28 (t, J = 7 Hz, 1H), 7.22 (d, J = 7 Hz, 2H), 5.17 (d, J = 3.5 Hz, 1H), 4.01 (q, J - 7 Hz, 2H), 2.29 (s, 3H), 1.11 (t, J = 7 Hz, 3H).
Derivative 4c: [ethyl 6-methyl-4-(3-fluorophenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 10.41 (s, 1H), 9.69 (s, 1H), 7.41 (dd, J = 8.5, 5.5 Hz, 1H), 7.14 (dd, J= 2, 6.5, 1H), 7.07 (d, J = 8.5 Hz, 1H), 6.98 (d, J = 10 Hz, 1H), 5.20 (d, J = 3.5 Hz, 1H), 4.31 (q, J = 7 Hz, 2H), 2.30 (s, 3H), 1.11 (t, J = 7.5 Hz, 3H).
Derivative 4d: [ethyl 6-methyl-4-(3-methoxyphenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 10.33 (s, 1H), 9.63 (s, 1H), 7.27 (t, J = 8 Hz, 1H), 6.85 (d, J = 7 Hz, 1H), 6.78 (d, J = 7.5 Hz, 1H), 6.76 (s, 1H), 5.15 (d, J = 4 Hz, 1H), 4.03 (q, J = 7.5 Hz, 2H), 3.73 (s, 3H), 2.28 (s, 3H), 1.12 (t, J = 7.5 Hz, 3H).
Derivative 4e: [ethyl 6-methyl-4-(3-bromophenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 10.41 (s, 1H), 9.68 (s, 1H), 7.50 (d, J = 8.5 Hz, 1H), 7.38 (s, 1H), 7.34 (t, J = 7.5 Hz, 1H), 7.22 (d, J = 7.5 Hz, 1H), 5.17 (d, J = 3.5 Hz, 1H), 4.03 (q, J = 7.5 Hz, 2H), 2.30 (s, 3H), 1.13 (t, J = 7 Hz, 3H).
Derivative 5a: [methyl 6-methyl-4-(3-hydroxyphenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-ketone]: 10.23 (s, 1H), 9.69 (s, 1H), 9.43 (s, 1H), 7.11 (t, J = 7.5 Hz, 1H), 6.6 (m, 3H), 5.19 (d, J = 3.5 Hz, 1H), 2.31 (s, 3H), 2.16 (s, 3H).
Derivative 5b: [isobutyl 6-methyl-4-(3-hydroxyphenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate]: 10.30 (s, 1H), 9.60 (s, 1H), 9.42 (s, 1H), 7.12 (t, J = 8Hz, 1H), 6.65 (m, 3H), 5.09 (d, J = 4 Hz, 1H), 3.81 (d, J = 7 Hz, 2H), 2.31 (s, 3H), 1.78 (m, J = 7 Hz, 1H), 0.77 (d, J = 6.5 Hz, 6H).
Derivative 5c: [benzyl 6-methyl-4-(3-hydroxyphenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate]: 10.34 (s, 1H), 9.62 (s, 1H), 9.44 (s, 1H), 7.29 (d, J = 5.5, 2H), 7.16 (m, 3H), 7.11 (t, J = 8 Hz, 1H), 6.65 (m, 3H), 5.12 (d, J = 4 Hz, 1H), 5.07 (q, J = 12 Hz, 2H), 2.31 (s, 3H).
Derivative 6: [ethyl 6-isopropyl-4-(3-hydroxyphenyl)-2-thioxo-1,2,3,4-tetrahydropyrimidine-5-carboxylate] : 9.74 (s, 1H), 9.59 (s, 1H), 9.45 (s, 1H), 7.13 (t, J = 7.5 Hz, 1H), 6.65 (m, 3H), 5.08 (d, J = 3.5 Hz, 1H), 4.10 (7t, J = 7 Hz, 1H), 4.01 (q, J = 6.5 Hz, 2H), 1.17 (dd, J = 25, 6.5 Hz, 6H), 1.10 (t, J = 6.5 Hz, 3H).
We used a modified PCR mutagenesis protocol  to introduce mutations into hEg5-367H, a plasmid encoding the C-terminally His6-tagged motor domain (amino acid residues 1–367) of human Eg5. The wild-type and mutant proteins were expressed and purified using Ni-NTA affinity chromatography and concentrated to between 3 and 10 mg/ml protein as described .
ATP hydrolysis by recombinantly expressed kinesin motor domains was measured using an NADH-enzyme coupled assay . ATP hydrolysis was measured in KC25 buffer supplemented with 5 μM taxol, 1 mM ATP-KOH (pH 7.0), the appropriate concentration of motor domain, and the appropriate concentrations of taxol-polymerized microtubules. ATP hydrolysis as a function of microtubule concentration was fit to the Michaelis-Menten equation to determine the maximal rate of microtubule-stimulated ATP hydrolysis (kcat) and the microtubule concentration for half-maximal stimulation (K0.5MT) according to the equation: V = kcat × ([MT]/([MT] + K0.5MT). A similar analysis, V = V1 μM × ([ATP]/([ATP] + KmATP), was used to measure KmATP. Here, V1 μM was the maximal rate of ATP hydrolysis by hEg5-367H or its mutants at 1 μM microtubule concentration. To measure the potency (IC50) of monastrol derivates, we measured ATP hydrolysis in the presence of 1 μM taxol-stabilized microtubules at various drug concentrations and fit the data to the function: V = 1- Vmax ([drug]/([drug] + IC50).
Adenosine 5'- triphosphate
Mixture of leupeptin, pepstatin, and chymostatin
Beta-Nicotinamide adenine dinucleotide, reduced
Non-nucleotide reverse transcriptase inhibitor
Phenyl methyl sulfonyl fluoride
Structure activity relationship
Small inhibitory Ribonucleic Acid
Tris (2-carboxyethyl) phosphine
Members of the Institute for Chemistry and Cell Biology and Mitchison laboratories provided thoughtful review of the manuscript. Dr. Barry H. Paw provided D. rerio AB9 tissue culture cells. Dr. Christine Field generously provided a D. melanogaster cDNA library. Dr. Ulrike Eggert provided cell lines and advice for growing D. melanogaster cells in culture. Professor Stephen C. Harrison (HHMI) and members of his laboratory, in particular Dr. Piotr Sliz and Dr. Lei Jin, were helpful with data collection and structure refinement. This work was supported by NCI program project # CA78048. The Institute for Chemistry and Cell Biology was also supported by NIGMS, program project #GM62566, Merck & Company, Inc., Merck KGaA, and the W. M. Keck Foundation.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.