Design and synthesis of (E)-3-benzylideneindolin-2-one derivatives as potential allosteric inhibitors of Aurora A kinase

Abstract

The mitotic kinase Aurora A, a pivotal regulator of the cell cycle, is overexpressed in various cancers and has emerged as one of the most promising targets for anticancer drug discovery. However, the lack of specificity and potential toxicity have impeded clinical trials involving orthosteric inhibitors. In this study, allosteric sites of Aurora A were predicted using the AlloReverse web server. Based on the non-ATP competitive inhibitor Tripolin A and molecular docking information targeting the desired allosteric site 3 of Aurora A, a series of (E)-3-benzylideneindolin-2-one derivatives were designed and synthesized. Compared to Tripolin A, our compounds AK09, AK34 and AK35 have stronger inhibitory effects and can be further investigated as potential allosteric inhibitors. Moreover, the compound AK34 with the strongest inhibitory activity (IC₅₀ = 1.68 μM) has a high affinity for Aurora A (K_D = 216 nM). According to the analysis of the structure–activity relationship of the compounds and the results of their molecular docking models, these compounds tend to act on the allosteric site 3 of Aurora A.

The compound AK34, derived from Tripolin A, shows a pronounced inhibitory effect on Aurora A kinase with high affinity (IC₅₀ = 1.68 μM, K_D = 216 nM). SAR and molecular docking indicate that these (E)-3-benzylideneindolin-2-one derivatives tend to act on an allosteric Site 3 predicted by AlloReverse.

Introduction

Cancer poses a substantial threat to human health, causing approximately ten million deaths worldwide annually. Extensive research on cancer has revealed the overexpression of Aurora kinases in various types of cancer cells, including those associated with breast, liver, and colon cancers.^1–8 Three isoforms of Aurora kinases have been identified in mammals and named as Aurora A (AURKA), Aurora B (AURKB), and Aurora C (AURKC) respectively.^9,10 These kinases belong to the serine/threonine kinase family¹¹ and play crucial roles in regulating mitotic spindle function, centrosome maturation, chromosome alignment and segregation, as well as cytoplasmic fission.^12–15 Notably, their catalytic structural domains exhibit a high degree of similarity,^1,16,17 posing challenges in developing selective inhibitors.

In recent decades, small molecules such as VX680, AZD1152 and MLN8054 have been developed as ATP-competitive inhibitors for Aurora kinases, but the lack of specificity and potential toxicity have hindered their approval for cancer treatment in clinical trials.^18–20 Moreover, traditional orthosteric inhibitors may lead to target resistance emergence. Fortunately, allosteric inhibitors targeting less conserved sites offer a unique advantage in terms of selectivity and safety,^21–23 and coadministration of allosteric and orthosteric drugs provides a revolutionary strategy to overcome drug resistance.²⁴ The obstacles of orthosteric inhibitors of Aurora kinases and the scarcity of allosteric inhibitors necessitate the initiation of allosteric inhibitor development, which has the potential to cure cancers caused by Aurora kinase dysfunction.

AURKA regulates a greater number of mechanisms of tumor promotion than the other two kinases, thus positioning it as a key kinase for targeted drug development.^25–28 AURKA is activated through Thr288 autophosphorylation and regulated by protein binding partners such as TPX2 (targeting protein for Xenopus kinesin-like protein 2) and TACC3 (transforming acidic coiled-coil containing protein 3).^29–35 TPX2 not only serves as a substrate for AURKA but also acts as the most extensively studied activator, enhancing AURKA's catalytic activity by at least sevenfold and promoting autophosphorylation through its interaction with the three allosteric pockets of AURKA.^31,32 The residues Tyr8, Tyr10, Phe16, and Trp34 within TPX2 are essential for forming a stable complex³⁶ (Fig. 1), indicating that compounds that block the interaction between AURKA and TPX2 would disrupt both the localization and activity of AURKA. The compound Tripolin A functions as a non-ATP competitive inhibitor of AURKA, which can reduce the localization of AURKA on spindle microtubules (MTs) and affect centrosome integrity, spindle formation and length, as well as interphase MT dynamics.³⁷ Here, we demonstrate that the structure of Tripolin A has an E-configuration (Fig. 2) and combine the structural information of Tripolin A with that of AURKA to develop allosteric inhibitors that can effectively inhibit the catalytic activity of AURKA.

Fig. 1. Potential allosteric sites of AURKA predicted by AlloReverse and pockets bound by TPX2.

Fig. 2. The chemical structure of Tripolin A and its docking model with the allosteric site 3 of AURKA (PDB ID: 1MQ4).

Results and discussion

Allosteric sites predicted and molecular docking of the compound Tripolin A on AURKA

The use of computational assistance has proven to be invaluable in the exploration of the mechanism of allosteric regulation. For instance, molecular dynamics (MD) simulations have elucidated the previously elusive allosteric inhibitory mechanism of protein tyrosine phosphatase 1B (PTP1B)³⁸ and allosteric activation mechanism of SIRT6.³⁹ The Allosteric Database (ASD), freely available to all users at http://mdl.shsmu.edu.cn/ASD/, has been systematically developed since 2009 to provide comprehensive information on allosteric regulation and aim to offer a one-stop platform for allosteric drug design.^40–44 Among these resources, AlloReverse presents a multiscale understanding of hierarchical allosteric regulations based on the theory of “reversed allosteric communication”, which indicates that allosteric sites are also regulated by orthosteric sites.⁴⁵ Thus, AlloReverse was applied to predict that AURKA possesses five potential allosteric sites, accurately predicting three reported allosteric pockets (sites 1, 2 and 3) as well (Fig. 1). Additionally, site 1 and site 3 exhibit coupling since they share three residues on regulation pathways, whereas site 2 demonstrates a distinct regulatory direction (Fig. S1†). The outcome implies that future study could focus on both site 1 and site 3 for the design of allosteric molecules. Furthermore, molecular docking analysis reveals that Tripolin A shows a favorable binding pose towards site 3, with particular emphasis on the interaction between its hydroxyl group and the Glu183 residue of AURKA (Fig. 2).

Design, synthesis and structure–activity relationship of Aurora kinase A inhibitors

To investigate the role of hydroxyl groups, we retained the core structure of Tripolin A to study the position and number of ones. The compounds (AK01–11, 13, 22–35) were synthesized by a Knoevenagel reaction between the substituted aldehyde and oxindole. The condensation was carried out in ethanol with piperidine as the base catalyst in a microwave reactor (Scheme 1).

Scheme 1. Synthetic pathway for compounds (AK01–11, 13, 22–35). Substituted benzaldehydes replaced by indolealdehyde for AK12. Reagents and conditions: (a) piperidine (cat.), ethanol, microwave, 110 °C, 30 min.

In the case of compounds containing zero or one hydroxyl group, the inhibitory effect on AURKA is weak (compounds AK01 to AK04, Table 1). Then, we increased the number of hydroxyl groups on the A ring and obtained compounds AK05 to AK10. Among these compounds, both AK06 and AK10 demonstrated excellent inhibitory activity. The hydroxyl groups of these two compounds are located at the meta and para positions of ring A respectively, exhibiting a consistent structure–activity relationship with that of a single hydroxyl group. Docking results between these three molecules (AK06, AK09 and AK10) and the desired site 3 revealed that both hydroxyl groups in AK06 can interact with Glu183 (Fig. S2A†), and a similar interaction was observed for compound AK09, which was obtained by introducing a hydroxyl group at the 2-position of the A-ring (Fig. S2B†), while one hydroxyl group in AK10 remains unengaged (Fig. S2C†). It is noteworthy that substitution of this idle hydroxyl group in compound AK10 with a methoxy group resulted in decreased inhibitory potency for the resulting compound AK11 (Table 1).

Table 1. Structure–activity relationship of the first batch of modified Tripolin A derivatives^a.

Compound	R	Inhibition% (10 μM)	Inhibition% (100 μM)	IC50 (μM)
Tripolin A	2-OH, 5-OH	23.42 ± 1.72	35.30 ± 2.42	ND
AK01	H	5.65 ± 1.64	13.92 ± 6.04	ND
AK02	2-OH	7.26 ± 2.14	1.22 ± 3.15	ND
AK03	3-OH	11.94 ± 3.41	30.20 ± 1.25	ND
AK04	4-OH	18.77 ± 7.51	31.24 ± 4.34	ND
AK05	2-OH, 3-OH	22.05 ± 2.24	74.06 ± 0.52	ND
AK06	3-OH, 4-OH	66.34 ± 1.09	93.22 ± 4.78	4.89 ± 0.32
AK07	2-OH, 4-OH	12.45 ± 8.25	28.22 ± 4.92	ND
AK08	3-OH, 5-OH	19.83 ± 3.18	75.55 ± 5.13	ND
AK09	2-OH, 4-OH, 5-OH	64.57 ± 0.18	98.31 ± 2.45	8.46 ± 0.89
AK10	3-OH, 4-OH, 5-OH	82.72 ± 0.57	99.83 ± 0.84	4.43 ± 1.20
AK11	3-OCH3, 4-OH, 5-OH	57.55 ± 7.58	95.94 ± 1.25	8.83 ± 0.49

^aValues represent the mean ± standard deviation (SD) of three separate experiments; ND indicates not determined. Some compounds have both Z- and E-configurations, with the E-configuration being the predominant one and the ratios are provided in the ESI.†

Given the difficulty in acquiring title compounds, our primary focus was on further exploration of compound AK06 rather than compound AK10. We attempted to simultaneously replace its two hydroxyl groups and obtained compounds AK12 and AK13 with reduced inhibitory effect. To investigate the impact of compound configuration on kinase activity, we reduced the double bond of AK06, resulting in compound AK14 which showed a marked loss of inhibitory power. Similarly, Tripolin A demonstrated comparable results after the same modification (compound AK15). These findings underscore the non-negligible role that compound configuration plays in modulating kinase activity. Meanwhile, we substituted the secondary amine of indolone with N-methyl (compound AK16) and expanded the 5-membered ring of 2-indolone to a 6-membered ring (compounds AK17, 18, and 19). These structural modifications resulted in a reduced ability to inhibit kinase activity. Interestingly, the desired compound 19, which contains a 4-dihydrochromogen fragment, has a stronger inhibitory effect on AURKA than compound AK17 and AK18. Removal of either hydroxyl group led to a decrease in inhibitory effect (compounds AK20 and AK21), which mirrored the structure–activity relationship for 2-indolone (Table 2). Unfortunately, the limited chemical reactivity of the 4-dihydrochromogen moiety has impeded further exploration. The N-methylated 3-benzylidieneindolin-2-one AK16 was obtained using a similar procedure as described in Scheme 1, starting from N-methyloxindole. Also, compounds AK17–21 were synthesized from dihydroquinolin-4-one, dihydrothiochromen-4-one, and chroman-4-one instead of indolin-2-one as shown in Scheme 1.

Table 2. Structure–activity relationship of the second batch of modified compounds^a.

Compound	Structure	Inhibition% (10 μM)	Inhibition% (100 μM)
AK12		49.43 ± 6.18	45.21 ± 7.65
AK13		16.90 ± 1.86	10.02 ± 8.63
AK14		12.87 ± 2.20	7.52 ± 0.81
AK15		9.32 ± 1.64	6.75 ± 2.52
AK16		22.98 ± 3.46	82.65 ± 2.37
AK17		15.24 ± 3.23	46.38 ± 4.76
AK18		22.74 ± 3.41	40.32 ± 2.63
AK19		29.20 ± 3.41	63.45 ± 8.92
AK20		28.37 ± 5.28	33.89 ± 2.58
AK21		36.29 ± 2.18	51.41 ± 1.74

^aValues represent the mean ± standard deviation (SD) of three separate experiments. Some compounds have both Z- and E-configurations, with the E-configuration being the predominant one and the ratios are provided in ESI.†

In consideration of the favourable inhibitory effect conferred by the hydroxyl group in the para-position, the para-position hydroxyl group of the compound AK06 was unaltered while changing the meta-position hydroxyl group to a methoxy (compound AK22) or a methyl group (compound AK23), halogen atoms (compounds AK24 to AK27), and an amidogen (compound AK28). Although compounds containing chlorine or bromine show considerable inhibitory effect, they are still inferior to the compound AK06 with hydroxyl groups (Table 3).

Table 3. Structure–activity relationship of the third batch of (E)-3-benzylideneindolin-2-one derivatives^a.

Compound	R1	Inhibition% (10 μM)	Inhibition% (100 μM)	IC50 (μM)
AK22	–OCH3	21.72 ± 7.91	30.25 ± 8.75	ND
AK23	–CH3	16.49 ± 1.07	28.24 ± 0.17	ND
AK24	–F	47.05 ± 3.74	62.25 ± 1.20	ND
AK25	–Cl	48.36 ± 3.56	84.79 ± 0.56	10.54 ± 1.26
AK26	–Br	47.53 ± 0.43	85.73 ± 1.32	11.52 ± 0.64
AK27	–I	33.20 ± 0.71	70.79 ± 2.28	ND
AK28	–NH2	41.58 ± 0.95	85.50 ± 6.09	14.00 ± 2.45

^aValues represent the mean ± standard deviation (SD) of three separate experiments; ND indicates not determined. The compounds have both Z- and E-configurations, with the E-configuration being the predominant one and the ratios are provided in ESI.†

After conducting the aforementioned investigation, it was determined that the hydroxyl groups in the A ring should not be altered. Instead, the B ring was modified by the introduction of a methyl group and a halogen group at position B-7. The introduction of the halogens fluorine and iodine to the target location is a challenging process, and our research scope is limited to chlorine (AK29) and bromine (AK30). Amongst these three compounds, AK31 with a methyl group exhibited superior inhibitory effects on AURKA. In light of these findings, we proceeded to vary the position of the methyl group on the B ring (compounds AK32 to AK34). It was found that the compound AK34, which had the methyl group at the B-4 position, displayed an increase in inhibitory potency (IC₅₀ = 1.68 μM) in comparison to AK06 (IC₅₀ = 4.89 μM). After applying the same modification of compound AK10, the obtained compound AK35 was also observed to exhibit enhanced inhibitory effects (Table 4). Upon docking the two compounds AK34 and AK35 to site 3, it was observed that there was a change in the binding pose, accompanied by alterations in their interaction with amino acid residues (Fig. S2D and E†). It can be concluded that one hydroxyl group of the compound AK35 does not enter the pocket at all and does not appear to enhance the inhibitory potency of the molecule. This result is consistent with the outcomes of the kinase activity test of comparing AK34 and AK35.

Table 4. Structure–activity relationship of the fourth batch of (E)-3-benzylideneindolin-2-one derivatives^a.

Compound	R′	R1	Inhibition% (10 μM)	Inhibition% (100 μM)	IC50 (μM)
AK29	7-Cl	H	16.71 ± 1.12	36.14 ± 1.53	ND
AK30	7-Br	H	18.26 ± 3.98	31.96 ± 2.43	ND
AK31	7-CH3	H	10.44 ± 4.71	56.62 ± 0.69	ND
AK32	6-CH3	H	48.73 ± 3.21	89.12 ± 2.52	10.51 ± 1.61
AK33	5-CH3	H	45.23 ± 1.32	94.71 ± 1.62	12.74 ± 2.16
AK34	4-CH3	H	94.81 ± 1.82	97.01 ± 1.12	1.68 ± 0.46
AK35	4-CH3	–OH	92.82 ± 1.45	101.47 ± 0.88	1.68 ± 0.55

^aValues represent the mean ± standard deviation (SD) of three separate experiments; ND indicates not determined. The compounds have both Z- and E-configurations, with the E-configuration being the predominant one and the ratios are provided in ESI.†

Although our compounds are derived from Tripolin A, the inhibitory effect of Tripolin A on AURKA activity observed in our kinase activity test differs from that reported in the literature (IC₅₀ = 1.5 μM).³⁷ The NMR data of Tripolin A synthesised by us matches that reported in the literature,³⁷ indicating that the two are the same compound. In addition, the inhibitory effect of MLN8237, an orthosteric inhibitor currently undergoing phase III clinical trials, on the enzymatic activity of AURKA was evaluated. The result of our test (IC₅₀ = 5.0 nM) is in close agreement with the previously published data (IC₅₀ = 1.2 nM),^46,47 which confirms the reliability and accuracy of our kinase inhibition assay.

ATP competition, affinity and cell viability inhibition

Following an investigation into the structure–activity relationship of (E)-3-benzylideneindolin-2-one derivatives, five highly potent inhibitors targeting AURKA were selected for further study. The impact of varying ATP concentrations on their inhibition of AURKA enzyme activity was investigated using an in vitro kinase assay. The results revealed that compounds AK06 and AK10 displayed a remarkable increase in IC₅₀ values as ATP concentration increased, which is consistent with the trend shown by reversine,⁴⁸ an ATP-competitive inhibitor of AURKA. In contrast, compounds AK09, AK34, and AK35 are less affected by the concentration of ATP, suggesting their potential as allosteric inhibitors (Fig. 3A). Surprisingly, compound AK34 had a strong affinity for AURKA (K_D = 216 nM, Fig. 3B) as determined by isothermal titration calorimetry (ITC). After mutations in each of the three amino acid residues that interact with compound AK34, a decrease in the affinity of the compound for AURKA was observed (Fig. S4†). This finding lends support to the accuracy of the predicted binding sites.

Fig. 3. (A) Graph showing IC₅₀ values (in μM, mean, n = 3) of compounds in the presence of different ATP concentrations, using an in vitro kinase assay. (B) ITC binding curves demonstrate the binding affinity of compound AK34 and AURKA. (C) Cell survival of MDA-MB-231 cells treated with compounds for 48 h. (D) Cell survival of HeLa cells treated with compounds for 48 h.

The effects of the five compounds and Tripolin A were evaluated on MDA-MB-231 cells proliferation by CCK8 assay. Notably, compound AK06 (IC₅₀ = 58.12 μM) showed the most potent inhibitory effect (Fig. S3A, Table S1†), while the remaining compounds only demonstrated high cell viability inhibition only at a concentration of 200 μM (Fig. 3C). Fortunately, the compounds AK06 (IC₅₀ = 11.93 μM) and AK34 (IC₅₀ = 18.15 μM) had considerable inhibitory effects on the growth of HeLa cells (Fig. 3D and S3B, Table S2†).

Experimental

Chemistry

¹H and ¹³C NMR spectra were measured on a Bruker AVANCE II (¹H at 400 MHz, ¹³C at 100 MHz) magnetic resonance spectrometer. ¹H chemical shifts were reported in ppm using residual DMSO (δ 2.50) as internal standards. Coupling constants (J) were reported in Hertz (Hz). Proton decoupled ¹³C NMR spectra were reported in ppm (δ) relative to residual DMSO (δ 39.52). The NMR information and spectra of these compounds are provided in the ESI.† The melting point was determined using a MP50 melting point system (Mettler Toledo).

General procedure for the synthesis of the compounds

Equimolar amounts (1 mmol) of the oxindole and aldehyde were dissolved in ethanol (6 ml), 2 to 3 drops of piperidine was added and the mixture was heated in a sealed vessel. Then, the vessel was heated to 110 °C for 30 min in a microwave synthesizer (CEM Discover 2.0). The reaction mixture was cooled, and the precipitated solid was filtered and rinsed carefully with cold ethanol.

Due to the presence of exocyclic double bond, the compounds we have synthesised (except AK14, AK15) may exist as E or Z isomers. According to the literature, it is challenging to generate the Z isomers of AK19, AK20, and AK21 under conventional conditions. Therefore, it can be assumed that they are all E isomers.⁴⁹AK17 and AK18 share a similar structure with these compounds and have similar hydrogen spectra, from which it is assumed that their conformations are the E-isomer and Z-isomer, respectively.

In the literature, whether the compound is an E isomer or a Z isomer can be judged by the chemical shift of hydrogen protons on the A ring (H-2, H-6). Chemical shifts of 7.45–7.84 ppm is assigned to the E isomer, and 7.85–8.53 ppm to the Z isomer.^50,51 By analysing the NMR data of our synthesised compounds, it has been determined that they are E isomers. Although some of the compounds undergo different degrees of conformational transformations in DMSO-d₆ solvents, making our spectra show two isomers, the E isomers are still the main components of the compounds. It has to be mentioned that Tripolin A is the E isomer, not the Z isomer reported in the literature,³⁷ because we failed to find any hydrogen proton signals at the chemical shift 7.85–8.53 ppm. Moreover, if Tripolin A is present as the Z isomer, NOESY analysis would reveal interactions between the vinylic proton and H-4 on the indolinone ring, but the latter interaction can't be observed in the spectrum (Fig. S5†), and it is concluded that Tripolin A is present as the E isomer.

Allosteric sites predicted and molecular docking

The potential allosteric sites of AURKA were predicted by AlloReverse available at https://mdl.shsmu.edu.cn/AlloReverse/ or http://www.allostery.net/AlloReverse/. To begin, a job name and the PDB ID (1MQ4) were inputted or a PDB file for AURKA was uploaded. Next, the relevant chains and orthosteric ligand ADP were selected. Finally, the job was ran and the results were downloaded. Molecular docking was performed using the glide docking module of the Maestro software with standard precision.

Protein expression and purification

Aurora A protein truncate (122–403) was transformed and expressed in BL21 (DE3) receptor cells and obtained by multiple purification processes including an AKTA pure protein purification system, SDS-PAGE electrophoresis, and a Superdex 75 Increase 10/300 GL gel filtration chromatography column. Three amino acid residues (Lys166, Glu175, or His201) of AURKA were all mutated to alanine, resulting in three mutants of AUKRA. The mutants of the AURKA proteins were obtained through the same methodology previously outlined.

Kinase activity assays

Kinase activity was measured in a ProxiPlate-384Plus white shallow plate (PerkinElmer) using a ADP-Glo kinase detection kit (Promega). The reaction buffer is composed of 20 mM HEPES (pH 7.4), 20 mM NaCl, 1 mM EGTA, 10 mM MgCl₂, 0.02% Tween-20, 0.1 mg mL⁻¹ BSA, and 50 μM DTT.

The compounds were dissolved in DMSO and prepared into a 50 mM master batch, then diluted with buffer to form a working solution of 5% DMSO. 1 μL was taken and added to a 384-well plate, then 2 μL of AURKA protein working solution was added, and the reaction was placed in a wet box and incubated at room temperature for 30 min, then 2 μL of the mixture of TPX2, ATP and kemptide working solution was added and reacted for 1 h at room temperature in a wet box (AURKA : ATP : TPX2 : kemptide = 1 : 5000 : 1 : 5000). After the reaction was completed, 5 μL of ADP-Glo reagent was added for 40 min at room temperature, then 10 μL of kinase detection reagent was added for 30 min at room temperature, and luminescence was detected by Synergy Neo zymography. The assay was performed as a top readout with an integration time of 1 s and an assay height of 9.25 mm. A DMSO control (negative control, 0% inhibition), a control without AURKA protein (background value), and a positive control of MLN8237 (10 nM) accompanied the assay.

The competitive inhibition experiment of ATP was carried out according to the above kinase activity determination method, and the other components in the system were kept unchanged. The inhibitory ability of compounds on AURKA enzyme activity at the final concentration of 5 μM, 50 μM and 500 μM of ATP was investigated, and the IC₅₀ was compared.

Cell viability assays

MDA-MB-231 and HeLa cell lines (from Cell Bank/Stem Cell Bank, Chinese Academy of Sciences) were cultured in a humidified air incubator at 37 °C and 5% CO₂. Cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin.

Cells were seeded at 5 × 10³ cells per well in 96-well plates for 24 hours, then the cells were treated with different doses of compounds (0.78 μM to 200 μM) or DMSO for negative control groups (0.2%). After 48 hours, DMEM was discarded in the culture plate, and CCK8 reagent at 10% DMEM was added for 2 hours. The absorbance was then measured at 450 nm and background corrected at 650 nm using Synergy Neo zymography (Biotek, USA). The percentage of viable cells was calculated by setting the negative control group cells to 100%.

Isothermal titration calorimetry

The buffer was added to the sample protein AURKA, and the reference material and compound AK34 were added to the syringe. The sample protein AURKA and syringe are put into the ITC instrument, and the temperature is controlled at a constant temperature. Then the reference material is injected into the sample AURKA, and the baseline value of the thermal effect signal is recorded. Finally, the material to be tested is added to the sample protein, the change value of the thermal effect signal is recorded, and the thermodynamic parameters are calculated according to the measured results.

Conclusions

In conclusion, AlloReverse was employed to predict five allosteric sites of AURKA, and site 1, site 2, and site 3 have great potential for drug design. Molecular docking results suggested that Tripolin A could bind to site 3. Leveraging the molecular structure of Tripolin A, several derivatives with stronger inhibitory effect on AURKA than Tripolin A were synthesized. Compounds AK06, AK09, AK10, AK34 and AK35 have the strongest inhibitory effect on AURKA, but only compounds AK09, AK34 and AK35 are less affected by the concentration of ATP, which can be further studied as potential allosteric inhibitors. Besides, compound AK34 inhibits HeLa cell growth well and has a high affinity for AURKA. According to the analysis of the structure–activity relationship of the compounds and the results of their molecular docking model, these three compounds are all very likely to act on the allosteric site 3 of AURKA. Nevertheless, more research is necessary to ascertain the precise binding site.

Data availability

The data supporting the results of this study are provided in the article or ESI.†

Author contributions

Y. L. Jiao, M. Z. Zhao, J. Zhang, T. G. Liang and J. F. Xu designed the experimental protocol and drafted the manuscript. Y. L. Jiao synthesized all the compounds in this paper. J. Zhong and S. B. Ning completed the kinase activity test. Y. L. Jiao and M. Z. Zhao carried out the cell experiment. ITC experiment done by M. Z. Zhao. All authors read and approved the final manuscript.

Conflicts of interest

There are no conflicts to declare.