Historically, drugs used in the treatment of cancers also tend to cause damage to healthy cells while affecting cancer cells. Therefore, the identification of novel agents that act specifically against cancer cells remains a high priority in the search for new therapies. In contrast with normal cells, most cancer cells contain multiple centrosomes which are associated with genome instability and tumorigenesis. Cancer cells can avoid multipolar mitosis, which can cause cell death, by clustering the extra centrosomes into two spindle poles, thereby enabling bipolar division. Kinesin-like protein KIFC1 plays a critical role in centrosome clustering in cancer cells, but is not essential for normal cells. Therefore, targeting KIFC1 may provide novel insight into selective killing of cancer cells. In the present study, we identified a small-molecule KIFC1 inhibitor, SR31527, which inhibited microtubule (MT)-stimulated KIFC1 ATPase activity with an IC50 value of 6.6 μM. By using bio layer interferometry technology, we further demonstrated that SR31527 bound directly to KIFC1 with high affinity (Kd=25.4 nM). Our results from computational modelling and saturation-transfer difference (STD)-NMR experiments suggest that SR31527 bound to a novel allosteric site of KIFC1 that appears suitable for developing selective inhibitors of KIFC1. Importantly, SR31527 prevented bipolar clustering of extra centrosomes in triple negative breast cancer (TNBC) cells and significantly reduced TNBC cell colony formation and viability, but was less toxic to normal fibroblasts. Therefore, SR31527 provides a valuable tool for studying the biological function of KIFC1 and serves as a potential lead for the development of novel therapeutic agents for breast cancer treatment.

INTRODUCTION

Inhibition of cell mitosis is a clinically validated strategy for cancer treatment [1]. Traditional anti-mitotic drugs such as taxanes and vinca alkaloids inhibit mitotic spindle assembly by targeting tubulin and disrupting microtubule (MT) dynamics [2]. However, because MTs are also present ubiquitously in normal cells and play very important roles throughout the cell cycle, disruption of MTs results in undesirable side effects, such as peripheral neuropathy, in patients [2,3].

Kinesins are cytoskeletal motor proteins that utilize the energy from ATP hydrolysis to perform mechanical work along MTs and mediate cellular processes such as cargo transport, and spindle and chromosome movement [4]. Mitotic kinesins are required for various aspects of mitosis, including bipolar spindle assembly, chromosome alignment, chromosome segregation and cytokinesis [5]. Since mitotic kinesins function exclusively during mitosis, compounds that specifically inhibit mitotic kinesins could affect only the proliferating cells, and thus are anticipated to be a new type of chemotherapeutic agents with better safety profiles [6]. The development of mitotic kinesin inhibitors has undergone significant progress in the last decade. A number of highly selective inhibitors of several mitotic kinesins, including Eg5 and CENP-E, have been identified [712], and some of those inhibitors have advanced into clinical trials [1316]. It is very encouraging that neuropathy, which is commonly caused by traditional mitotic inhibitors, was not observed on patients treated with mitotic kinesin inhibitors [5]. However, myelosupression is still the primary dose-limiting toxicity, and various haematological side effects were observed on patients treated with those mitotic kinesin inhibitors [5] because they non-differentially block the mitosis of both cancer cells and normal cells.

Kinesin-like protein (KIFC1), a unique mitotic kinesin, has recently emerged as a new anti-cancer drug target [17]. Increased centrosome number, or centrosome amplification, has been reported in a variety of human primary cancers and correlates with aneuploidy, chromosomal instability (CIN) and tumorigenesis [1822]. In mitosis, supernumerary centrosomes can form multipolar spindles which cause aneuploidy and ultimately cell death. However, cancer cells can avoid multipolar mitosis by clustering the extra centrosomes into two spindle poles thereby enabling bipolar division [2325]. Previous studies indicate that KIFC1 is critical for the centrosome clustering in cancer cells with amplified centrosomes [2628]. Knockdown of KIFC1 induced multipolar spindle mitotic defects in cancer cells containing extra centrosomes and caused cancer cell death [28], but had no effects on normal cells [27,28]. Moreover, KIFC1 is required for proper spindle assembly, stable pole-focusing and survival of cancer cells irrespective of normal or supernumerary centrosome number, thus indicating a more general and critical role for KIFC1 in cancer cells [26,27]. Although the mechanism of how KIFC1 regulates the survival of cancer cells is still under investigation, recent studies clearly implicate KIFC1 as an attractive drug target for the development of cancer-cell-selective therapeutics.

KIFC1 expression is significantly elevated in a broad panel of cancer tissues [26,27,29]. Pannu et al. [26] reported that KIFC1 is overexpressed in human breast cancers, and that KIFC1 overexpression correlates with increased aggressiveness and poorer clinical outcomes in breast cancer patients. Recently, we also demonstrated that KIFC1 expression is up-regulated in breast cancer, particularly in triple negative breast cancer (TNBC), and that KIFC1 is highly expressed in human breast cancer cell lines, but is undetectable in primary normal human mammary epithelial cells and weakly expressed in two human fibroblast lines [17]. We further showed that KIFC1 silencing significantly reduced breast cancer cell viability [17]. In the present study, we identified a small-molecule KIFC1 inhibitor, SR31527, through a high-throughput screening (HTS) campaign. Our results indicated that SR31527 inhibited KIFC1 by binding directly to an allosteric site of KIFC1 without involving MTs. Moreover, SR31527 prevented bipolar clustering of extra centrosomes in TNBC cells and significantly reduced TNBC cell colony formation and viability, and was less cytotoxic to normal fibroblasts.

MATERIALS AND METHODS

Preparation of active KIFC1 motor protein

To enable HTS, we established a protocol that allows us to purify a large amount of KIFC1 protein from Escherichia coli. Initially, more than ten constructs encoding human KIFC1 motor domain protein were generated. Among them, two constructs made with pET28 plasmid produced soluble and active KIFC1 proteins in E. coli cells. Protein purification was carried out in two steps with His-tag affinity and gel-filtration chromatography. Specifically, the cell pellet was resuspended in the lysis buffer (75 mM Tris/HCl, pH 8.0, 300 mM sodium chloride, 5% glycerol, 20 mM imidazole, 0.1% Triton X-100, and 0.5 mM TCEP (tris(2-carboxyethyl)phosphine) supplemented with a protease-inhibitor cocktail and 1 μg/ml benzonase nuclease) and was lysed by two passages through a French press. Cell debris was removed by centrifugation and the clear supernatant was passed through an Ni2+-nitrilotriacetate (Ni-NTA) column (HisTrap 5, GE Healthcare) equilibrated with the washing buffer (20 mM Tris/HCl, pH 7.5, 300 mM sodium chloride, 5% glycerol, 20 mM imidazole and 0.5 mM TCEP). The column was then washed extensively with washing buffer to remove non-specific proteins. The bound KIFC1 protein was eluted using a 300 ml linear gradient of 50–400 mM imidazole in an elution buffer (20 mM Tris/HCl, pH 7.5, 300 mM sodium chloride, 5% glycerol, 400 mM imidazole and 0.5 mM TCEP). Eluted fractions were analysed by SDS/PAGE, and fractions containing KIFC1 were pooled together. Further purification was conducted with a gel-filtration column (Superdex 200 26/60, GE Healthcare) and the eluted fractions from the column were analysed by SDS/PAGE. Fractions containing purified KIFC1 were pooled together and concentrated. With this protocol, we were able to obtain approximately 10–15 mg of KIFC1 protein from 1 litre of the E. coli culture.

High-throughput screening

A biochemical assay measuring MT-stimulated ATPase activity of KIFC1 was used for HTS. This assay was based upon the HTS assay of kinesin Eg5 [30], and was run in a 1536-well microtitre plate format. Briefly, 2.5 μl of 15 mM PIPES (pH 7.0) containing 1 mM magnesium chloride, 6 μg/ml MTs, 20 μM paclitaxel, 100 μM ATP, 0.02% Tween 20, 2% DMSO and 35 μg/ml KIFC1 protein were added to each well of 1536-well microtitre plates. The plates were incubated at room temperature for 1 h followed by addition of 1.25 μl of ADP Hunter™ Plus (DiscoverRx) Reagent A and 2.5 μl of Reagent B to detect the ADP production. The plates were further incubated for 0.5 h at room temperature and then bottom read for fluorescence (excitation at 530 nm/emission at 590 nm) on the Envision. A pilot screen of 10000 structurally diverse commercial compounds was performed in a single-dose (10 μM) format in duplicates on two separate days and the percentage inhibition data generated on each day were examined for each compound. The values of Pearson's correlation was 0.93, indicating an excellent correlation. The Z′-value for the campaign was 0.91±0.01; the coefficient of variations (CVs) was 1.36±0.17 for the full reaction and 4.17±1.46 for the background reaction. The hit rate [based on three times the S.D. (3*δ) plus the average percentage inhibition from all screened compounds] was 1.04%. All of these statistical results are indicative of a robust reproducible screen ready for HTS. We then utilized this assay and screened additional 20000 structurally diverse compounds commercially obtained from the Enamine library.

Molecular modelling

Molecular modelling studies were performed on a SGI Altix XE 1200 Linux cluster using the modules of the Schrödinger Suite 2012. Specifically, SiteMap program was used to identify potential binding pocket(s) by mapping the surface of the KIFC1 crystal structure (PDB ID: 2REP). A pocket with a SiteScore value >0.80 is considered suitable for the binding of small-molecule compounds. The 3D models of ligands (SR31527, monastrol) were first generated using LigPrep, then docked into the different binding sites (including ATP-binding site, monastrol-binding site, S1 and S2 sites) using the Glide program following an induced-fit-docking (IFD) protocol [31] which is capable of sampling dramatic side-chain conformational changes as well as minor changes in protein backbone structure. The default docking parameters were first tested by docking monastrol (an Eg5 inhibitor) into its binding site on Eg5. The docked monastrol–Eg5 model excellently reproduced the crystal structure of monastrol–Eg5 complex (PDB ID: 1Q0B). The same IFD protocol as well as parameters were then employed for the docking studies of SR31527 into the different binding sites on KIFC1. Residues within 5 Å (1 Å=0.1 nm) of the docked ligands were allowed to be flexible. The docked results were ranked by the extra-precision (XP) scoring function of Glide.

NMR spectroscopy

The saturation-transfer difference (STD)-NMR data were collected following established protocols [32,33]. Samples containing SR31527 and KIFC1 protein at a concentration ratio of 20:1 were prepared in 2H2O. STD-NMR spectra were recorded with a total of 32000 points, 80 scans and selective saturation of protein resonances at 0, 0.65, 1.67 and 7.61 ppm (−8.18 ppm for the reference spectra), using a series of SEDUCE pulses (1000 points, 50 ms), for a total saturation time of 10 s (SEDUCE-1 pulse is similar to a Gaussian pulse, and has been used by other laboratories [34]). Reference experiments using the free ligands themselves (i.e. without KIFC1 protein) were performed under the same experimental conditions to verify true ligand binding. No STD signals were present in the difference spectra of the free ligand, indicating that the effects observed in the presence of KIFC1 were due to a true saturation transfer from the protein.

Biolayer interferometry

Biolayer interferometry (BLI) was used to study the kinetics of SR31527 binding to KIFC1 on a ForteBIO Octet, according to the manufacturer's instructions and published methods [35,36]. Purified His-tagged KIFC1 protein was first bound to HIS2K biosensors coated with penta-His antibody. The biosensors were then transferred into wells containing SR31527 in serial dilutions from 60 to 0.74 μM concentrations. Protein–ligand binding was measured by monitoring the changes in the interferometric profile of the wavelength of light passing through the sensor. Following a 300-s incubation, the KIFC1-coated sensor tips were transferred to the kinetics buffer to allow dissociation for 900 s. Binding curves were analysed using ForteBIO software, which performs a global fit according to the 1:1 Langmuir model to obtain the kinetic rate constants for each set of interaction conditions.

Cell culture

All breast cancer cell lines and lung fibroblast lines were obtained from the A.T.C.C., and cultured in Dulbecco's modified Eagle medium (DMEM) containing 10% FBS, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin. Cells were grown under standard cell culture conditions at 37°C in a humidified atmosphere with 5% CO2. All cell lines were Mycoplasma-negative.

Confocal microscopy

MDA-MB-231, BT549 and MDA-MB-435s cells were grown on glass coverslips. After 24 h of treatment of SR31527, the cells were fixed with 4% paraformaldehyde in PBS for 20 min, and permeabilized with 0.2% Triton X-100 for 10 min at room temperature. The cells were then incubated with FITC-conjugated anti-α-tubulin antibody (Sigma) and indocarbocyanine (Cy3)-conjugated anti-γ-tubulin (Sigma) for 45 min, and nuclei were stained with NucRed647 for 10 min. The cells were then examined by a laser-scanning confocal microscope (Leica DMI 4000 B). All images were captured with an HCX PL Apo ×63 oil-immersion objective. Images were processed and analysed using Leica's LAS Image Analysis software.

Cell viability assay

Cells were seeded into 96-well tissue culture-treated microtitre plates at a density of 3000–4000 cells/well. After overnight incubation, the cells were treated with SR31527 for 96 h. Cell viability was measured by the CellTiter-Glo Assay (Promega).

Colony formation assay

Cancer cells were seeded at a density of 500 cells/well into six-well plates. At 16 h after the plates were set up, SR31527 was added, and medium was replenished every 3 days. After being incubated for 10–14 days, colonies were fixed with 4% formaldehyde, stained with 0.5 mg/ml Crystal Violet and imaged on a FluorChem HD2 Imager System (Alpha Innotech).

RESULTS

High-throughput screening for KIFC1 inhibitors

To identify small-molecule KIFC1 inhibitors, we established an HTS assay based on an ATPase assay, which detects the MT-stimulated enzymatic activity of KIFC1. A diverse library consisting of 30000 structurally representative compounds was first screened in an HTS format. The primary HTS hits were further confirmed in a 10-point 2-fold dilution concentration–response format. Thirty compounds were found to inhibit KIFC1 activity in a concentration-dependent manner (Supplementary Table S1). Among them, SR31527 displayed the most potent inhibitory effect with an IC50 value of 6.6 μM (Figure 1). The rest of the identified inhibitors were less potent and/or structurally unattractive. For example, although C2 showed a similar IC50 value to that of SR31527 (Supplementary Table S1), it is a phenanthrene derivative with potential non-selective DNA-intercalation properties. In addition, C2 contains a chemically reactive benzoquinone moiety that could potentially react with proteins possessing nucleophilic groups such as thiols. Therefore, we focused our follow-up studies on SR31527.

Structure of SR31527 and its inhibitory effect on KIFC1 function

Figure 1
Structure of SR31527 and its inhibitory effect on KIFC1 function

(A) Chemical structure of SR31527. (B) SR31527 inhibited the MT-stimulated ATPase activity of KIFC1 in a concentration-dependent mode with an IC50 value of 6.6 μM. The kinetic curve was fitted using GraphPad Prism.

Figure 1
Structure of SR31527 and its inhibitory effect on KIFC1 function

(A) Chemical structure of SR31527. (B) SR31527 inhibited the MT-stimulated ATPase activity of KIFC1 in a concentration-dependent mode with an IC50 value of 6.6 μM. The kinetic curve was fitted using GraphPad Prism.

SR31527 binds directly to KIFC1 without involving MT

An important advantage of mitotic kinesin inhibitors over traditional anti-mitotic agents is that they do not interfere with the multi-functional MT, and thus may cause fewer side effects. Since our primary assay detects the MT-stimulated ATPase activities, there is a chance that the identified active compounds may inhibit KIFC1 function through a MT-related mechanism [37]. We therefore utilized BLI to evaluate whether SR31527 can bind directly to KIFC1 without involving MTs. Our BLI results demonstrated direct binding of SR31527 to KIFC1 with a calculated dissociation constant Kd value of 25.4 nM (Figure 2A). We further applied STD-NMR as a secondary binding assay, which is a sensitive and easily applicable technique that not only detects transient binding, but also provides information regarding which part(s) of a ligand interacts directly with a receptor [32,34,38]. The observed STD-NMR spectrum confirmed that SR31527 bound to KIFC1 in the absence of MT, and indicated that the aromatic ring(s) and the amine group of SR31527 interact directly with KIFC1 (Figure 2B).

SR31527 binds directly to KIFC1

Figure 2
SR31527 binds directly to KIFC1

(A) The BLI results confirmed that SR31527 binds KIFC1 (without the existence of MTs) with a calculated dissociate constant Kd value of 25.4 nM. The experiments were performed and analysed using the ForteBIO Octet system. (B) The reference 1D 1H NMR spectra of SR31527 in the presence of KIFC1, at 600 MHz and 298 K (blue-coloured) and the corresponding STD-NMR spectrum (×4) (red-coloured). SR31527 structure is embedded with its aromatic, methyl and amine hydrogen labelled HAr, HM and HN respectively, and their corresponding NMR signals assigned.

Figure 2
SR31527 binds directly to KIFC1

(A) The BLI results confirmed that SR31527 binds KIFC1 (without the existence of MTs) with a calculated dissociate constant Kd value of 25.4 nM. The experiments were performed and analysed using the ForteBIO Octet system. (B) The reference 1D 1H NMR spectra of SR31527 in the presence of KIFC1, at 600 MHz and 298 K (blue-coloured) and the corresponding STD-NMR spectrum (×4) (red-coloured). SR31527 structure is embedded with its aromatic, methyl and amine hydrogen labelled HAr, HM and HN respectively, and their corresponding NMR signals assigned.

SR31527 binds to a novel allosteric site of the KIFC1 motor domain

We conducted structural analysis and docking studies to explore the structural insights of potential interactions between SR31527 and KIFC1. We have previously studied kinesin Eg5 through molecular modelling and simulations, and identified several novel allosteric sites in addition to the well-studied monastrol-binding site (M-site) where most known Eg5 inhibitors bind [39]. By mapping the surface of the KIFC1 crystal structure, we found that two of the previously identified kinesin allosteric pockets, S1 and S2, also exist on KIFC1 with excellent SiteScore of 1.09 and 0.97 respectively (Figures 3A and 3B). A SiteScore value of 0.80 has been found to accurately distinguish between drug-binding and non-drug-binding sites [40], suggesting that both S1 and S2 are potential binding sites for potent KIFC1 inhibitors. On the other hand, due to the relatively short loop-5 of KIFC1 that forms part of the M-site pocket, the SiteScore of the M-site on KIFC1 is only 0.70 compared with 1.02 on Eg5.

Structural illustration of the binding between SR31527 and KIFC1

Figure 3
Structural illustration of the binding between SR31527 and KIFC1

(A) Cartoon presentation of the X-ray structure of the Eg5 motor domain (PDB ID: 1Q0B). The monastrol- and ADP-binding sites, the previously identified allosteric S1 and S2 sites, and the loop-5 are marked. (B) Carton representation of the X-ray structure of KIFC1 motor domain (PDB ID: 2REP) with the different sites and the loop-5 marked. (C) Docked KIFC1–SR31527 model with SR31527 bound at the S2 site of KIFC1 [the model is rotated approximately 180° from the orientation of (A)/(B) for better presentation]. (D) Close-up view of the KIFC1–SR31527 interaction.

Figure 3
Structural illustration of the binding between SR31527 and KIFC1

(A) Cartoon presentation of the X-ray structure of the Eg5 motor domain (PDB ID: 1Q0B). The monastrol- and ADP-binding sites, the previously identified allosteric S1 and S2 sites, and the loop-5 are marked. (B) Carton representation of the X-ray structure of KIFC1 motor domain (PDB ID: 2REP) with the different sites and the loop-5 marked. (C) Docked KIFC1–SR31527 model with SR31527 bound at the S2 site of KIFC1 [the model is rotated approximately 180° from the orientation of (A)/(B) for better presentation]. (D) Close-up view of the KIFC1–SR31527 interaction.

We further performed docking studies to evaluate which site SR31527 may bind to. The SR31527 molecule was docked separately into the different binding pockets of KIFC1, including the M-site, S1, S2 and the ATP-binding site (ATP-site). The docked results indicated that SR31527 fitted best at the S2 site with a docking score of −6.5 kcal/mol, compared with −4.9, −3.1 and −4.8 kcal/mol for the S1, M-site and ATP-site respectively. The S2 site is a cleft-shaped pocket located between helix α4 and α6 of KIFC1. Visual examination of the docked models further confirmed the structural complementarity between SR31527 and the S2 site: the aromatic rings of SR31527 fit well into the hydrophobic pocket of S2 and interact with Tyr100 and Phe347 through π–π stacking, whereas the polar amide group points towards the solvent-accessible area of the pocket (Figures 3C and 3D). Interestingly, in our STD-NMR experiments (Figure 2B), the 1D 1H STD spectrum of the SR31527 methyl group was not observable, which suggested that, unlike the other structural elements of the SR31527 molecule, the methyl group was relatively away from the protein, therefore was not significantly affected by the binding. This is consistent with the docked model where the methyl group pointed towards the outside of the S2 pocket and did not form any specific interactions with the protein. Taken together, our results suggest that SR31527 inhibits KIFC1 by binding directly to the allosteric S2 site of KIFC1.

SR31527 triggers multipolar spindle formation in breast cancer cells

KIFC1 is normally a non-essential kinesin motor protein, but plays a critical role in centrosome clustering in cancer cells [27,28]. Knockdown of KIFC1 induced multipolar spindle mitotic defects in cancer cell lines containing extra centrosomes, and induced cancer cell death [27,28]. Having established that SR31527 binds to KIFC1 and suppresses KIFC1 enzymatic activity, we then examined whether SR31527 induces multipolar spindle formation in breast cancer cells. As shown in Figure 4, SR31527 significantly enhanced the percentage of MDA-MB-231, BT549 and MDA-MB-435s cells containing multipolar spindles, indicating that SR31527 acts as an extra-centrosome-de clustering agent by binding to KIFC1.

SR31527 induced multiple spindle formation in breast cancer cells

Figure 4
SR31527 induced multiple spindle formation in breast cancer cells

(A) BT549 cells were treated with SR31527 at 50 μM. After 24 h of incubation, the cells were fixed, permeabilized and immunolabelled for α- and γ-tubulin for the detection of spindles and centrosomes (green and red fluorescent labelling respectively). DNA was labelled with NucRed647 (blue). (B) MDA-MB-231, BT549 and MDA-MB-435s cells were treated with SR31527 at 50 μM for 24 h. The cells were then fixed and stained as described above, and the percentage of multipolar spindles was calculated in each cell line treated with DMSO control or SR31527. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with corresponding cancer cells treated with DMSO control.

Figure 4
SR31527 induced multiple spindle formation in breast cancer cells

(A) BT549 cells were treated with SR31527 at 50 μM. After 24 h of incubation, the cells were fixed, permeabilized and immunolabelled for α- and γ-tubulin for the detection of spindles and centrosomes (green and red fluorescent labelling respectively). DNA was labelled with NucRed647 (blue). (B) MDA-MB-231, BT549 and MDA-MB-435s cells were treated with SR31527 at 50 μM for 24 h. The cells were then fixed and stained as described above, and the percentage of multipolar spindles was calculated in each cell line treated with DMSO control or SR31527. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with corresponding cancer cells treated with DMSO control.

SR31527 decreases the cell viability and colony formation of breast cancer cells and is less cytotoxic to normal human fibroblasts

Given that SR31527 can bind to KIFC1, suppress KIFC1 enzymatic activity and induce multiple spindle formation in breast cancer cells, we then examined the effect of SR31527 on cell viability of breast cancer cells. As shown in Figure 5, SR31527 inhibited TNBC cell viability in a concentration-dependent manner with IC50 values between 20 and 33 μM in TNBC cell lines MDA-MB-231, BT549 and MDA-MB-435s. To further characterize the anti-cancer activity of SR31527, we performed colony-formation assays in TNBC cells. As shown in Figure 6, SR31527 at 6.25–25 μM significantly suppressed colony formation in TNBC cells.

SR31527 decreased breast cancer cell viability

Figure 5
SR31527 decreased breast cancer cell viability

Breast cancer MDA-MB-231, BT549 and MDA-MB-435s cells in 96-well plates were treated with SR31527 at the indicated concentrations for 96 h. Cell viability was measured by the CellTiter-Glo assay. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with corresponding control value.

Figure 5
SR31527 decreased breast cancer cell viability

Breast cancer MDA-MB-231, BT549 and MDA-MB-435s cells in 96-well plates were treated with SR31527 at the indicated concentrations for 96 h. Cell viability was measured by the CellTiter-Glo assay. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with corresponding control value.

SR31527 inhibited breast cancer colony formation

Figure 6
SR31527 inhibited breast cancer colony formation

Breast cancer MDA-MB-231, BT549 and MDA-MB-435s cells were treated with SR31527 at the indicated concentrations for 10–12 days. The medium was changed every 3 days. Colonies were fixed with formaldehyde and stained with Crystal Violet. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with cells treated with DMSO.

Figure 6
SR31527 inhibited breast cancer colony formation

Breast cancer MDA-MB-231, BT549 and MDA-MB-435s cells were treated with SR31527 at the indicated concentrations for 10–12 days. The medium was changed every 3 days. Colonies were fixed with formaldehyde and stained with Crystal Violet. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with cells treated with DMSO.

A specific KIFC1 inhibitor should not only display its cytotoxicity against malignant cells, but also be less toxic towards normal cells. In our recent study, we reported that KIFC1 is highly expressed in TNBC cell lines, but is almost undetectable in human lung fibroblast line LL47 [17]. We therefore evaluated the effects of SR31527 on LL47 cells. LL47 cells had a moderate growth rate with a doubling time of 41 h in the exponential growth phase (Figure 7A). As shown in Figure 7B, SR31527 at concentrations between 6.25 and 50 μM had almost no effects on LL47 cell viability, whereas SR31527 at 50 μM was able to kill 65–83% of TNBC cells (Figure 5), suggesting that SR31527 is able to selectively kill malignant cells. However, SR31527 at 100 μM killed almost all LL47 cells, indicating that SR31527 displays off-target cytotoxic effects at high concentrations.

Effects of SR31527 on LL47 cell viability

Figure 7
Effects of SR31527 on LL47 cell viability

(A) LL47 cells were plated in 12-well plates (5×104 cells per well), and cells were harvested and counted with Trypan Blue at 48, 72 and 96 h. All values are averages of triplicate determinations with S.D. indicated by error bars. (B) LL47 cells in 96-well plates were treated with SR31527 at the indicated concentrations for 96 h. Cell viability was measured by the CellTiter-Glo assay. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with cells treated with DMSO.

Figure 7
Effects of SR31527 on LL47 cell viability

(A) LL47 cells were plated in 12-well plates (5×104 cells per well), and cells were harvested and counted with Trypan Blue at 48, 72 and 96 h. All values are averages of triplicate determinations with S.D. indicated by error bars. (B) LL47 cells in 96-well plates were treated with SR31527 at the indicated concentrations for 96 h. Cell viability was measured by the CellTiter-Glo assay. All values are averages of triplicate determinations with the S.D. indicated by error bars. **P<0.01 compared with cells treated with DMSO.

DISCUSSION

CIN is a hallmark of many tumours and is frequently caused by extra centrosomes that transiently disrupt the normal bipolar spindle geometry needed for accurate chromosome segregation [18,19,41,42]. It has been demonstrated that KIFC1 is required for the survival of cancer cells with multiple centrosomes, and, more importantly, it is non-essential for normal cells. Therefore, specific inhibition of KIFC1 could be a promising strategy to selectively kill cancer cells without damaging normal healthy cells. In the present study, we identified a small-molecule KIFC1 inhibitor, SR31527, which binds directly to KIFC1 with an IC50 value of 6.6 μM against MT-stimulated KIFC1 ATPase activity. Moreover, SR31527 prevented bipolar clustering of extra centrosomes in TNBC cells and significantly reduced TNBC cell viability with IC50 values between 20 and 33 μM in TNBC cell lines MDA-MB-231, BT549 and MDA-MB-435s. Importantly, SR31527 displayed no cytotoxic effects on LL47 fibroblasts at the concentration up to 50 μM, indicating that SR31527 shows good selectivity against cancer cells.

KIFC1 has emerged as an anti-cancer drug target in recent years. To date, only two groups have reported specific KIFC1 inhibitors in the literature [37,43,44]. Scientists at AstraZeneca identified a small-molecule KIFC1 inhibitor, AZ82, which inhibited MT-stimulated KIFC1 ATPase activity with an IC50 of 0.3 μM and triggered multipolar spindle formation and mitotic catastrophe in cells with amplified centrosomes. However, a non-specific cytotoxic effect of AZ82 was observed at 4 μM, which prevented further studies regarding whether it can selectively kill cancer cells with amplified centrosomes. Additional studies indicated that AZ82 bound specifically to the KIFC1–MT complex, but did not interact directly with KIFC1 or MT when they were not associated [37]. Another KIFC1 inhibitor (CW069) reported by Watts et al. [43] was computationally designed based on the inhibitors of kinesin Eg5. CW069 bound specifically to KIFC1 and inhibited its ATPase activity with an IC50 value of 75 μM. It increased multipolar spindle formation in neuroblastoma N1E-115 cells, but had no effects on bipolar spindle formation in normal human dermal fibroblasts (NHDFs). CW069 was also more potent at decreasing cell viability of N1E-115 cells (IC50=86 μM) than the NHDFs (IC50 187 μM) [43]. In the present study, we demonstrated that SR31527 binds directly to KIFC1 without interacting with MT. Moreover, SR31527 exhibits high selectivity for cytotoxicity towards TNBC cells compared with normal fibroblasts. All together, these findings suggest that by exclusively binding to KIFC1, KIFC1 inhibitors could selectively kill cancer cells over normal cells. However, a previous study demonstrated that KIFC1 is essential for bipolar spindle formation and genomic stability in the primary human fibroblast line IMR-90 [45]. Moreover, KIFC1 is able to promote breast cancer cell proliferation and enhance cell cycle kinetics through G2- and M-phases via centrosome-clustering-independent mechanisms in breast cancer cells [26]. Therefore, future studies are required to address whether there are any other cell cycle abnormalities in normal cells and cancer cells treated with SR31527.

Our computer modelling and NMR results indicated that SR31527 bound to a novel allosteric S2 site on KIFC1. We further attempted mutagenesis studies of the S2 site residues, but failed to obtain active mutant KIFC1 proteins (results not shown), suggesting that residues at the S2 site are important for the structure and function of KIFC1. The allosteric binding of SR31527 is different from AZ82 and CW069, both putatively bind to the M-site of KIFC1. This allosteric S2 site was originally found by us on kinesin Eg5 based on computational analysis of molecular dynamics simulation results [39], and was later confirmed by the crystal structure of an Eg5-inhibitor complex [46]. Since all kinesin proteins contain a structurally similar motor domain, it is not very surprising that such a similar S2 site was also found on KIFC1. The structural features of kinesin proteins are highly conserved, although KIFC1 is the only kinesin which displays a pivotal role in cancer cell centrosome clustering, spindle assembly and survival [27,28]. Therefore, identifying selective inhibitors of KIFC1 is critical for developing cancer-cell-selective agents. One well-known strategy for achieving selectivity is to target a protein's allosteric binding sites [4751]. Although targeting the allosteric M-site for selective Eg5 inhibitors has been successful [11], the M-Site is quite specific for the Eg5 motor due to its unique long loop-5 which forms the lid of the pocket. Therefore, a similar strategy may not work well for other kinesins with short loop-5, such as KIFC1. The S2 site is mainly formed by two structurally conserved helices and has the structural properties that are suitable for the tight binding of small-molecule ligands, therefore it offers an opportunity to identify selective KIFC1 inhibitors. SR31527 thus provides a valuable tool to study the allosteric regulation of KIFC1 and serves as a potential lead for the development of a novel therapeutic agent for cancer treatment. Further structural biology studies, such as through X-ray crystallography, will provide additional structural and biological details of the allosteric KIFC1-inhibitor interactions for drug discovery purposes.

AUTHOR CONTRIBUTION

Wei Zhang coordinated the study, performed the computational modelling and wrote the paper. Yimin Wang and Rongbao Li designed and purified the protein. Ling Zhai, Indira Padmalayam, Rebecca Boohaker, Robert Bostwick, Lucile White and Larry Ross performed the HTS studies. Joseph Maddry, Subramaniam Ananthan, Corinne Augelli-Szafran and Mark Suto contributed their medicinal chemistry expertise to the study and the paper. Rebecca Boohaker and Vandana Gupta conducted the binding assays studies. Wenyan Lu, Bo Xu and Yonghe Li were responsible for the cell-based assays and related writings.

We thank Dr N. Rama Krishna and Dr Ronald Shin of the UAB High Field NMR Facility for their supports on the collection of NMR spectrum.

FUNDING

This work was supported by the Southern Research Internal Fund; the Alabama Innovation Fund; and the National Institutes of Health [grant numbers R01CA124531 and R21CA182056].

Abbreviations

     
  • BLI

    biolayer interferometry

  •  
  • CIN

    chromosomal instability

  •  
  • HTS

    high-throughput screening

  •  
  • IFD

    induced-fit-docking

  •  
  • MT

    microtubule

  •  
  • NHDF

    normal human dermal fibroblast

  •  
  • STD

    saturation-transfer difference

  •  
  • TCEP

    tris(2-carboxyethyl)phosphine

  •  
  • TNBC

    triple negative breast cancer

References

References
1
Nagle
A.
Hur
W.
Gray
N.S.
Antimitotic agents of natural origin
Curr. Drug Targets
2006
, vol. 
7
 (pg. 
305
-
326
)
[PubMed]
2
Hagiwara
H.
Sunada
Y.
Mechanism of taxane neurotoxicity
Breast Cancer
2004
, vol. 
11
 (pg. 
82
-
85
)
[PubMed]
3
Quasthoff
S.
Hartung
H.P.
Chemotherapy-induced peripheral neuropathy
J. Neurol.
2002
, vol. 
249
 (pg. 
9
-
17
)
[PubMed]
4
Miki
H.
Okada
Y.
Hirokawa
N.
Analysis of the kinesin superfamily: insights into structure and function
Trends Cell Biol.
2005
, vol. 
15
 (pg. 
467
-
476
)
[PubMed]
5
Huszar
D.
Theoclitou
M.E.
Skolnik
J.
Herbst
R.
Kinesin motor proteins as targets for cancer therapy
Cancer Metastasis Rev.
2009
, vol. 
28
 (pg. 
197
-
208
)
[PubMed]
6
Sarli
V.
Giannis
A.
Inhibitors of mitotic kinesins: next-generation antimitotics
ChemMedChem
2006
, vol. 
1
 (pg. 
293
-
298
)
[PubMed]
7
Jiang
C.
You
Q.D.
Li
Z.Y.
Guo
Q.L.
Kinesin spindle protein inhibitors as anticancer agents
Expert Opin. Ther. Pat.
2006
, vol. 
16
 (pg. 
1517
-
1532
)
8
Bergnes
G.
Brejc
K.
Belmont
L.
Mitotic kinesins: prospects for antimitotic drug discovery
Curr. Top. Med. Chem.
2005
, vol. 
5
 (pg. 
127
-
145
)
[PubMed]
9
Coleman
P.J.
Fraley
M.E.
Inhibitors of the mitotic kinesin spindle protein
Expert Opin. Ther. Pat.
2004
, vol. 
14
 (pg. 
1659
-
1667
)
10
Wood
K.W.
Lad
L.
Luo
L.
Qian
X.
Knight
S.D.
Nevins
N.
Brejc
K.
Sutton
D.
Gilmartin
A.G.
Chua
P.R.
, et al. 
Antitumor activity of an allosteric inhibitor of centromere-associated protein-E
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
5839
-
5844
)
[PubMed]
11
El-Nassan
H.B.
Advances in the discovery of kinesin spindle protein (Eg5) inhibitors as antitumor agents
Eur. J. Med. Chem.
2013
, vol. 
62
 (pg. 
614
-
631
)
[PubMed]
12
Knight
S.D.
Parrish
C.A.
Recent progress in the identification and clinical evaluation of inhibitors of the mitotic kinesin KSP
Curr. Top. Med. Chem.
2008
, vol. 
8
 (pg. 
888
-
904
)
[PubMed]
13
Gerecitano
J.F.
Stephenson
J.J.
Lewis
N.L.
Osmukhina
A.
Li
J.
Wu
K.
You
Z.
Huszar
D.
Skolnik
J.M.
Schwartz
G.K.
A Phase I trial of the kinesin spindle protein (Eg5) inhibitor AZD4877 in patients with solid and lymphoid malignancies
Invest. New Drugs
2013
, vol. 
31
 (pg. 
355
-
362
)
[PubMed]
14
Holen
K.
DiPaola
R.
Liu
G.
Tan
A.R.
Wilding
G.
Hsu
K.
Agrawal
N.
Chen
C.
Xue
L.
Rosenberg
E.
Stein
M.
A phase I trial of MK-0731, a kinesin spindle protein (KSP) inhibitor, in patients with solid tumors
Invest. New Drugs
2012
, vol. 
30
 (pg. 
1088
-
1095
)
[PubMed]
15
Lee
R.T.
Beekman
K.E.
Hussain
M.
Davis
N.B.
Clark
J.I.
Thomas
S.P.
Nichols
K.F.
Stadler
W.M.
A University of Chicago consortium phase II trial of SB-715992 in advanced renal cell cancer
Clin. Genitourin. Cancer
2008
, vol. 
6
 (pg. 
21
-
24
)
[PubMed]
16
Woessner
R.
Tunquist
B.
Lemieux
C.
Chlipala
E.
Jackinsky
S.
Dewolf
W.
Jr
Voegtli
W.
Cox
A.
Rana
S.
Lee
P.
Walker
D.
ARRY-520, a novel KSP inhibitor with potent activity in hematological and taxane-resistant tumor models
Anticancer Res.
2009
, vol. 
29
 (pg. 
4373
-
4380
)
[PubMed]
17
Li
Y.
Lu
W.
Chen
D.
Boohaker
R.J.
Zhai
L.
Padmalayam
I.
Wennerberg
K.
Xu
B.
Zhang
W.
KIFC1 is a novel potential therapeutic target for breast cancer
Cancer Biol. Ther.
2015
, vol. 
16
 (pg. 
1316
-
1322
)
[PubMed]
18
Lingle
W.L.
Barrett
S.L.
Negron
V.C.
D'Assoro
A.B.
Boeneman
K.
Liu
W.
Whitehead
C.M.
Reynolds
C.
Salisbury
J.L.
Centrosome amplification drives chromosomal instability in breast tumor development
Proc. Natl. Acad. Sci. U.S.A.
2002
, vol. 
99
 (pg. 
1978
-
1983
)
[PubMed]
19
D'Assoro
A.B.
Lingle
W.L.
Salisbury
J.L.
Centrosome amplification and the development of cancer
Oncogene
2002
, vol. 
21
 (pg. 
6146
-
6153
)
[PubMed]
20
Brinkley
B.R.
Managing the centrosome numbers game: from chaos to stability in cancer cell division
Trends Cell Biol
2001
, vol. 
11
 (pg. 
18
-
21
)
[PubMed]
21
Acilan
C.
Saunders
W.S.
A tale of too many centrosomes
Cell
2008
, vol. 
134
 (pg. 
572
-
575
)
[PubMed]
22
Boland
C.R.
Komarova
N.L.
Goel
A.
Chromosomal instability and cancer: not just one CINgle mechanism
Gut
2009
, vol. 
58
 (pg. 
163
-
164
)
[PubMed]
23
Godinho
S.
Kwon
M.
Pellman
D.
Centrosomes and cancer: how cancer cells divide with too many centrosomes
Cancer Metastasis Rev.
2009
, vol. 
28
 (pg. 
85
-
98
)
[PubMed]
24
Gergely
F.
Basto
R.
Multiple centrosomes: together they stand, divided they fall
Genes Dev.
2008
, vol. 
22
 (pg. 
2291
-
2296
)
[PubMed]
25
Quintyne
N.J.
Reing
J.E.
Hoffelder
D.R.
Gollin
S.M.
Saunders
W.S.
Spindle multipolarity is prevented by centrosomal clustering
Science
2005
, vol. 
307
 (pg. 
127
-
129
)
[PubMed]
26
Pannu
V.
Rida
P.C.
Ogden
A.
Turaga
R.C.
Donthamsetty
S.
Bowen
N.J.
Rudd
K.
Gupta
M.V.
Reid
M.D.
Cantuaria
G.
, et al. 
HSET overexpression fuels tumor TT5 progression via centrosome clustering-independent mechanisms in breast cancer patients
Oncotarget
2015
, vol. 
6
 (pg. 
6076
-
6091
)
[PubMed]
27
Kleylein-Sohn
J.
Pollinger
B.
Ohmer
M.
Hofmann
F.
Nigg
E.A.
Hemmings
B.A.
Wartmann
M.
Acentrosomal spindle organization renders cancer cells dependent on the kinesin HSET
J. Cell Sci.
2012
, vol. 
125
 (pg. 
5391
-
5402
)
[PubMed]
28
Kwon
M.
Godinho
S.A.
Chandhok
N.S.
Ganem
N.J.
Azioune
A.
Thery
M.
Pellman
D.
Mechanisms to suppress multipolar divisions in cancer cells with extra centrosomes
Genes Dev.
2008
, vol. 
22
 (pg. 
2189
-
2203
)
[PubMed]
29
Pawar
S.
Donthamsetty
S.
Pannu
V.
Rida
P.
Ogden
A.
Bowen
N.
Osan
R.
Cantuaria
G.
Aneja
R.
KIFCI, a novel putative prognostic biomarker for ovarian adenocarcinomas: delineating protein interaction networks and signaling circuitries
J. Ovarian Res.
2014
, vol. 
7
 pg. 
53
 
[PubMed]
30
Zhang
B.
Senator
D.
Wilson
C.J.
Ng
S.C.
Development of a high-throughput robotic fluorescence-based assay for HsEg5 inhibitor screening
Anal. Biochem.
2005
, vol. 
345
 (pg. 
326
-
335
)
[PubMed]
31
Sherman
W.
Day
T.
Jacobson
M.P.
Friesner
R.A.
Farid
R.
Novel procedure for modeling ligand/receptor induced fit effects
J. Med. Chem.
2006
, vol. 
49
 (pg. 
534
-
553
)
[PubMed]
32
Jayalakshmi
V.
Krishna
N.R.
CORCEMA refinement of the bound ligand conformation within the protein binding pocket in reversibly forming weak complexes using STD-NMR intensities
J. Magn. Reson.
2004
, vol. 
168
 (pg. 
36
-
45
)
[PubMed]
33
Krishna
N.R.
Jayalakshmi
V.
Complete relaxation and conformational exchange matrix analysis of STD-NMR spectra of ligand–receptor complexes
Prog. Nuclear Magn. Reson. Spectrosc.
2006
, vol. 
49
 (pg. 
1
-
25
)
34
Peng
J.W.
Lepre
C.A.
Fejzo
J.
Abdul-Manan
N.
Moore
J.M.
Nuclear magnetic resonance-based approaches for lead generation in drug discovery
Methods Enzymol.
2001
, vol. 
338
 (pg. 
202
-
230
)
[PubMed]
35
Do
T.
Ho
F.
Heidecker
B.
Witte
K.
Chang
L.
Lerner
L.
A rapid method for determining dynamic binding capacity of resins for the purification of proteins
Protein Expr. Purif.
2008
, vol. 
60
 (pg. 
147
-
150
)
[PubMed]
36
Abdiche
Y.
Malashock
D.
Pinkerton
A.
Pons
J.
Determining kinetics and affinities of protein interactions using a parallel real-time label-free biosensor, the Octet
Anal. Biochem.
2008
, vol. 
377
 (pg. 
209
-
217
)
[PubMed]
37
Wu
J.
Mikule
K.
Wang
W.
Su
N.
Petteruti
P.
Gharahdaghi
F.
Code
E.
Zhu
X.
Jacques
K.
Lai
Z.
, et al. 
Discovery and mechanistic study of a small molecule inhibitor for motor protein KIFC1
ACS Chem. Biol.
2013
, vol. 
8
 (pg. 
2201
-
2208
)
[PubMed]
38
Jayalakshmi
V.
Krishna
N.R.
Complete relaxation and conformational exchange matrix (CORCEMA) analysis of intermolecular saturation transfer effects in reversibly forming ligand–receptor complexes
J. Magn. Reson.
2002
, vol. 
155
 (pg. 
106
-
118
)
[PubMed]
39
Zhang
W.
Exploring the intermediate states of ADP–ATP exchange: a simulation study on Eg5
J. Phys. Chem. B
2011
, vol. 
115
 (pg. 
784
-
795
)
[PubMed]
40
Sheth
P.R.
Shipps
G.W.
Seghezzi
W.
Smith
C.K.
Chuang
C.C.
Sanden
D.
Basso
A.D.
Vilenchik
L.
Gray
K.
Annis
D.A.
, et al. 
Novel benzimidazole inhibitors bind to a unique site in the kinesin spindle protein motor domain
Biochemistry
2010
, vol. 
49
 (pg. 
8350
-
8358
)
[PubMed]
41
Ganem
N.J.
Godinho
S.A.
Pellman
D.
A mechanism linking extra centrosomes to chromosomal instability
Nature
2009
, vol. 
460
 (pg. 
278
-
282
)
[PubMed]
42
Sluder
G.
Nordberg
J.J.
The good, the bad and the ugly: the practical consequences of centrosome amplification
Curr. Opin. Cell Biol.
2004
, vol. 
16
 (pg. 
49
-
54
)
[PubMed]
43
Watts
C.A.
Richards
F.M.
Bender
A.
Bond
P.J.
Korb
O.
Kern
O.
Riddick
M.
Owen
P.
Myers
R.M.
Raff
J.
, et al. 
Design, synthesis, and biological evaluation of an allosteric inhibitor of HSET that targets cancer cells with supernumerary centrosomes
Chem. Biol.
2013
, vol. 
20
 (pg. 
1399
-
1410
)
[PubMed]
44
Yang
B.
Lamb
M.L.
Zhang
T.
Hennessy
E.J.
Grewal
G.
Sha
L.
Zambrowski
M.
Block
M.H.
Dowling
J.E.
Su
N.
, et al. 
Discovery of potent KIFC1 inhibitors using a method of integrated high-throughput synthesis and screening
J. Med. Chem.
2014
, vol. 
57
 (pg. 
9958
-
9970
)
[PubMed]
45
Kim
N.
Song
K.
KIFC1 is essential for bipolar spindle formation and genomic stability in the primary human fibroblast IMR-90 cell
Cell Struct. Funct.
2013
, vol. 
38
 (pg. 
21
-
30
)
[PubMed]
46
Ulaganathan
V.
Talapatra
S.K.
Rath
O.
Pannifer
A.
Hackney
D.D.
Kozielski
F.
Structural insights into a unique inhibitor binding pocket in kinesin spindle protein
J. Am. Chem. Soc.
2013
, vol. 
135
 (pg. 
2263
-
2272
)
[PubMed]
47
Changeux
J.P.
50th anniversary of the word “Allosteric”
Protein Sci.
2011
, vol. 
20
 (pg. 
1119
-
1124
)
[PubMed]
48
Ozbabacan
S.E.A.
Gursoy
A.
Keskin
O.
Nussinov
R.
Conformational ensembles, signal transduction and residue hot spots: application to drug discovery
Curr. Opin. Drug Discov. Devel.
2010
, vol. 
13
 (pg. 
527
-
537
)
[PubMed]
49
Eglen
R.M.
Reisine
T.
Human kinome drug discovery and the emerging importance of atypical allosteric inhibitors
Expert Opin. Drug Discov.
2010
, vol. 
5
 (pg. 
277
-
290
)
[PubMed]
50
Fenton
A.W.
Allostery: an illustrated definition for the ‘second secret of life’
Trends Biochem. Sci.
2008
, vol. 
33
 (pg. 
420
-
425
)
[PubMed]
51
Lindsley
J.E.
Rutter
J.
Whence cometh the allosterome?
Proc. Natl. Acad. Sci. U.S.A.
2006
, vol. 
103
 (pg. 
10533
-
10535
)
[PubMed]

Supplementary data