Structure-based screening and identification of a novel Aurora-A-targeting peptide with antiproliferative activity against prostate cancer cells

Abstract

Aurora-A is a potential therapeutic target in prostate cancer. In this study, virtual screening identified four Aurora-A-targeting peptides, among which Peptide-1 showed the most favourable profile. Molecular docking and MST assays demonstrated that Peptide-1 had the lowest predicted binding free energy and the strongest binding affinity towards Aurora-A (K_d = 0.72 ± 0.04 μM). MD simulation, MM/PBSA, and free-energy landscape analyses indicated that the Aurora-A–Peptide-1 complex was conformationally stable and mainly driven by electrostatic interactions. MTT assays showed that Peptide-1 inhibited the proliferation of PC3, DU145, and NCI-H660 cells, with weaker activity in RWPE-1 cells. Aurora-A knockdown reduced cellular sensitivity to Peptide-1, supporting its target-dependent activity. qRT-PCR further showed increased p53 and p21 mRNA expression after Peptide-1 treatment in PC3/p53WT cells. These findings suggest that Peptide-1 may act as an Aurora-A-targeting peptide with antiproliferative activity in prostate cancer cells.

Introduction

Prostate cancer represents a major malignancy in the male population and is projected to account for 30% of all newly diagnosed cancers in males in 2025, ranking second among causes of cancer-related mortality in men^1,2. Current clinical management is largely based on the dependence of prostate cancer on androgen receptor (AR) signalling and involves inhibition of AR activity through antiandrogen therapy and androgen deprivation therapy (ADT)³. Although second-generation AR-targeted agents, such as enzalutamide and abiraterone, have provided clinical benefit for some patients^4–6, the development of therapeutic resistance remains a major factor limiting their long-term efficacy and further improvement in patient outcomes⁷. Therefore, the exploration of new therapeutic targets and intervention strategies remains necessary for the treatment of prostate cancer.

Aurora kinase A (Aurora-A) is encoded by the AURKA gene and is a serine/threonine kinase belonging to the Aurora kinase family, which also includes Aurora-B and Aurora-C⁸. It consists of 403 amino acids and has a predicted molecular weight of 45.8 kDa⁹. Structurally, Aurora-A contains a central catalytic kinase domain of 251 amino acids, flanked by a noncatalytic N-terminal region of 132 amino acids and a short C-terminal tail of 20 amino acids, and plays an important role in centrosome maturation, spindle assembly, and regulation of mitotic progression^9–11. Overexpression of Aurora-A can induce genomic instability and neoplastic transformation, suggesting that it may have oncogene-like properties¹². In addition, Aurora-A is frequently overexpressed and amplified in multiple tumour types, including breast, colon, pancreatic, ovarian, prostate, and gastric cancers^13–17, and its upregulation has been associated with increased therapeutic resistance and poorer prognosis^18–20. In prostate cancer, Aurora-A is highly expressed in primary human tumour tissues, murine prostate cancer models, and prostate cancer cell lines, and has been implicated in disease progression by promoting tumour cell growth and migration^21,22. Previous studies have further shown that Aurora-A upregulation is closely associated with malignant phenotypes in prostate cancer, suggesting that it may represent an important therapeutic target in this disease²³. It has also been reported that depletion of Aurora-A can suppress cell growth and induce apoptosis in some models, whereas its overexpression can enhance the expression of AR target genes²⁴. Notably, Aurora-A has also been implicated in multiple non-mitotic functions associated with prostate cancer progression and resistance to first-line therapy, and targeting Aurora-A has shown cytostatic and cytotoxic effects in advanced prostate cancer models²⁵. Moreover, in other tumour models, Aurora-A overexpression has been shown to attenuate paclitaxel-induced cell death¹², whereas interference with its expression can increase the sensitivity of tumour cells to anticancer treatment^26,27, further supporting its potential value as an anticancer target^28–31. Taken together, current evidence suggests that Aurora-A is not only an important regulator of prostate cancer progression but also a factor associated with therapeutic resistance, and may therefore represent a potentially useful intervention target for prostate cancer treatment.

p53 is an important tumour suppressor and transcriptional regulator that also serves as a widely used diagnostic and prognostic biomarker, with key roles in the regulation of cell cycle progression, programmed cell death, cell growth, and genomic stability^32–34. p21 is a canonical downstream target gene of p53 and a key cell cycle inhibitory factor^35,36. Previous studies have shown that loss of p53 function is one of the major events contributing to tumorigenesis and increased resistance to chemotherapeutic agents such as cisplatin and paclitaxel^37,38, and that approximately 50% of human tumours exhibit loss of p53 function³⁹. Aurora-A can promote p53 degradation and suppress its transcriptional activity, thereby weakening p53-mediated transcriptional regulation of its downstream target gene p21 and contributing to tumour cell tolerance to treatments such as cisplatin and ionising radiation^32,40,41. In addition, loss of p53 may lead to aberrant upregulation of Aurora-A and defective mitosis, thereby conferring a growth advantage on cancer cells⁴². Correspondingly, inhibition of Aurora-A expression or activity can upregulate p21 and is associated with restoration or enhancement of the p53-related pathway, thereby inducing cell cycle arrest and suppressing tumour cell proliferation^43–47. Therefore, Aurora-A may promote tumour cell growth and enhance therapeutic resistance, at least in part, through suppression of the p53/p21-related pathway⁴⁸.

Currently reported Aurora-A inhibitors are mainly small molecules, but such agents are generally associated with limitations including relatively high toxicity and potential off-target effects^49,50. For example, although the Aurora-A small-molecule inhibitor alisertib has advanced to clinical evaluation, its further development has been limited by haematologic toxicities such as neutropenia and anemia⁵¹. Another small-molecule inhibitor, XL228, has shown potential for overcoming resistance to single-target therapies and has exhibited broad activity across multiple tumour types; however, its simultaneous inhibition of multiple kinases has also resulted in dose-limiting toxicity⁵². These limitations may be partly related to the fact that many conventional Aurora kinase inhibitors act on the conserved ATP-binding pocket, which can make kinase isoform selectivity challenging and may contribute to off-target effects^53,54. Therefore, targeting regulatory regions outside the ATP-binding pocket may provide an alternative strategy for developing Aurora-A inhibitors with improved functional specificity. Among these regulatory regions, the TPX2-related binding interface is a functionally important surface region of Aurora-A, because TPX2 binding has been shown to regulate Aurora-A localisation, stabilise its active conformation, and promote its activation during mitosis^55,56. In contrast, peptide inhibitors generally exhibit higher target selectivity and may reduce potential off-target effects and systemic toxicity^57–59. Therefore, the development of Aurora-A-targeting peptides may provide a complementary strategy for improving selectivity and addressing some of the limitations associated with small-molecule inhibitors. On this basis, designing Aurora-A-targeting peptides towards the TPX2-related functional interface may represent a rational approach for developing Aurora-A inhibitors with potential mechanistic specificity. In addition, the previously reported Aurora-A small-molecule inhibitor C23, which targets the Aurora-A/TPX2 interaction interface, binds Aurora-A in the low micromolar range and has been shown to reduce Aurora-A activity and induce spindle pole defects in tumour cells⁶⁰. Based on these findings, C23 was selected as the positive control in the present study.

Virtual screening enables the rapid prioritisation of large peptide libraries before experimental evaluation, thereby reducing the scope, time, and cost of subsequent screening; when integrated with molecular docking and related computational approaches, it can also provide preliminary information on peptide–target binding modes to guide subsequent experimental validation^61–63. In the present study, a virtual screening strategy was employed to identify Peptide-1 as an Aurora-A-targeting peptide, which subsequently exhibited antiproliferative activity in prostate cancer cells.

Materials and methods

Chemicals and reagents

Peptides 1–4 were prepared via chemical synthesis by GL Biochem (Shanghai, China). Dimethyl sulfoxide (DMSO) was acquired from Aladdin Reagent (Shanghai, China). Phosphate-buffered saline (PBS), RPMI-1640 medium, foetal bovine serum (FBS) as well as penicillin-streptomycin mixed solution were all procured from Gibco (Grand Island, NY, USA).

Cell lines and culture conditions

Human prostate cancer PC3 cells (cat. no. CRL-1435), human prostate cancer DU145 cells (cat. no. HTB-81), human neuroendocrine prostate cancer NCI-H660 cells (cat. no. CRL-5813), and normal human prostate epithelial RWPE-1 cells (cat. no. CRL-3607) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). All cell lines were maintained in RPMI-1640 medium supplemented with 10% foetal bovine serum (FBS), 80 U/mL penicillin, and 0.08 mg/mL streptomycin, and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. Subculture was performed every 2–3 days to ensure consistent cell status. All subsequent cellular assays were conducted using log-phase cells prepared as single-cell suspensions.

Preparation of the Aurora-A crystal structure

The X-ray crystal structure of Aurora-A bound to a hydrocarbon-stapled proteomimetic of TPX2 was downloaded from the Protein Data Bank (PDB ID: 5LXM; resolution: 2.08 Å). Molecular Operating Environment (MOE; Chemical Computing Group Inc., Montreal, Quebec, Canada) was used to process the structure, and the bound TPX2-derived peptide was removed. The protein structure was then optimised using the “QuickPrep” workflow in MOE under the Amber10:EHT force field. This process included hydrogen atom addition using Protonate 3D, removal of water molecules, and energy minimisation.

Establishment of the peptide library

The peptide library was constructed following procedures described in previous studies⁶⁴. A combinatorial virtual peptide library was constructed using the “QuaSAR-CombiGen” tool in MOE. Briefly, three sets of structural units, each containing 39 tripeptide fragments, were fully enumerated and assembled to generate all possible nonapeptide sequences. During fragment ligation, the terminal oxygen atom at the C-terminus of each fragment was assigned as the “A1” site, whereas the terminal hydrogen atom at the N-terminus was assigned as the “A0” site. Fragment assembly was then performed according to predefined atom-labelling rules through A0-A1 linkage. This procedure generated a virtual library containing 59,319 nonapeptides. Subsequently, a three-dimensional multi-conformational database was generated using the “Conformation Import” tool in MOE. Conformations were generated under the Amber10:EHT force field with a strain limit of 4 kcal/mol, and up to 300 conformations were retained for each peptide. Finally, the resulting conformations were subjected to energy minimisation.

Virtual screening

The prepared Aurora-A structure was used as the receptor, and all native residues in the TPX2-binding pocket were retained. Protonation states were assigned at physiological pH and checked using the Protonate 3D module in MOE. The docking site was defined according to the position of the TPX2-derived peptide in the crystal structure. Accordingly, peptide docking was performed around the TPX2-derived peptide-binding region rather than across the entire protein surface, and the ligand atoms defining this site were used to generate alpha spheres within 5 Å radius to guide the placement of the tested peptides during docking. The constructed virtual peptide library containing 59,319 nonapeptides was then screened. During docking, peptide poses were generated using the “Triangle Matcher” placement method, with up to 100 poses generated for each peptide. The receptor was kept rigid, whereas the peptides were treated as flexible ligands. Initial scoring was performed using the London dG scoring function, followed by rescoring with GBVI/WSA dG. All docking runs were repeated three times under the same parameter settings, and the mean docking score was used for peptide ranking. Based on the docking results, the top four peptides were selected for subsequent experimental evaluation.

Structure–activity relationship (SAR) analysis

To characterise the binding features between Aurora-A and Peptides 1–4, the interaction modes of the peptides within the Aurora-A binding region were analysed using the “Ligand Interactions” tool in MOE, including hydrogen bonds, hydrophobic interactions, and key interacting residues. The results were used to compare the binding patterns of different peptides and to support the interpretation of their activity differences.

Binding affinity measurement by microscale thermophoresis (MST)

The binding affinities of Peptides 1–4 towards Aurora-A were determined using MST. In addition, MST was also employed to assess the binding affinity of Peptide-1 towards Aurora-B and Aurora-C. The MST assay parameters were set according to previously reported experimental protocols with minor modifications⁶⁵. Briefly, Aurora-A protein was fluorescently labelled using the Monolith^TM RED-NHS 2nd Generation Protein Labelling Kit (Cat. No. MO-L011; NanoTemper Technologies GmbH, Munich, Germany). The labelled protein was protected from light and kept on ice until use. Peptides 1–4 were separately dissolved in PBST buffer (PBS containing Tween-20) and serially diluted to generate 12 concentration points. A fixed concentration of fluorescently labelled Aurora-A was then mixed with different concentrations of each peptide and incubated for 5 min in the dark to allow protein–peptide complex formation. After incubation, the samples were loaded into standard capillaries, and thermophoresis-induced fluorescence changes were measured using a Monolith NT.115 instrument (NanoTemper Technologies GmbH). The data were used to fit binding curves and calculate K_d values. Peptide-1 binding to Aurora-B and Aurora-C was also measured using MST following the same procedure.

Molecular dynamics (MD) simulations

The MD simulation parameters were set according to previously reported protocols with minor modifications⁶⁶. Molecular dynamics simulations were performed using GROMACS 2021.5 for the docked Aurora-A–Peptide-1 complex over 50 ns to evaluate its conformational stability. The topology file of the complex was generated based on the AMBER99SB-ILDN force field. The complex was then placed in a cubic periodic box, solvated using the SPC/E water model, and neutralised by adding appropriate counterions (Na^+/Cl^-). Before the production simulation, the system was first energy-minimized for 5000 steps using the steepest descent algorithm, followed by NVT and NPT equilibration with temperature and pressure maintained at 300 K and 1 bar using the V-rescale thermostat and Parrinello–Rahman barostat, respectively. After equilibration, a 50 ns production simulation was conducted. The resulting RMSD.xvg, RMSF.xvg, Rg.xvg, and secondary structure.xvg files were extracted and used to analyse the dynamic stability of the complex. The Aurora-A–C23 complex was also simulated using the same procedure.

Free energy landscape

Based on the RMSD and Rg data of the Aurora-A–Peptide-1 complex, Gibbs free energy landscape data were generated via GROMACS, with 2D and 3D visualisation analyses conducted using Python (version 3.13.3).

MM/PBSA calculations

Molecular mechanics/Poisson–Boltzmann surface area (MM/PBSA) analysis was performed on MD trajectories to estimate the binding free energy of the Aurora-A–Peptide-1 complex and the Aurora-A–C23 complex as follows:

$$\Delta {\mathrm{G}}_{\text{bind}}=\Delta {\mathrm{E}}_{\text{vdW}}+\Delta {\mathrm{E}}_{\text{elec}}+\Delta {\mathrm{G}}_{\text{polar}}+\Delta {\mathrm{G}}_{\text{nonpolar}}$$

where ΔE_vdW and ΔE_elec represent van der Waals and electrostatic interaction energies, respectively, while ΔG_polar and ΔG_nonpolar represent polar and nonpolar solvation free energy contributions. Entropic contributions were not included in the present MM/PBSA estimation.

Serum stability assay

The serum stability of Peptides 1–4 was evaluated according to previously reported methods with minor modifications⁶⁷. Briefly, human serum was diluted to 10% with PBS and centrifuged to remove insoluble components. Each peptide was added to the diluted serum supernatant to obtain a final concentration of 5 μM. The mixtures were incubated at 37 °C, and aliquots of 200 μL were collected at predetermined time points (0, 1, 2, 4, 6, 8, and 12 h). Each aliquot was immediately mixed with 200 μL of acetonitrile to precipitate serum proteins. After centrifugation, the supernatant was collected and analysed by reversed-phase HPLC. The integrated peak area of each peptide was used to calculate the percentage of peptide remaining relative to the initial time point.

Confocal microscopy

Confocal microscopy was performed to examine the cellular uptake and intracellular distribution of Peptide-1, as previously described⁶⁷. Briefly, cells were seeded in 35-mm glass-bottom microwell dishes and cultured overnight. After gently washing twice with PBS, the cells were incubated with 1 μM FITC-labelled Peptide-1 and 1 mg/mL rhodamine-labelled dextran (Dextran^Rho) in complete growth medium at 37 °C for 2 h. After removal of the peptide-containing medium, the cells were washed three times with PBS and then incubated with 10 μg/mL Hoechst 33342 in PBS for 30 min to stain the nuclei. Subsequently, the cells were washed again with PBS and imaged using an LSM 700 laser scanning confocal microscope (Carl Zeiss, Jena, Germany). FITC, Dextran^Rho, and Hoechst 33342 signals were recorded to visualise Peptide-1 uptake, rhodamine-associated intracellular fluorescence, and nuclei, respectively.

MTT assay

Cell viability was evaluated using the MTT assay with appropriate modifications to a previously reported protocol⁶⁸. PC3, DU145, NCI-H660, and RWPE-1 cells were plated in 96-well plates at 5 × 10³ cells/well and cultured overnight before treatment. Peptides were added at two-fold serial concentrations, followed by incubation for 72 h at 37 °C with 5% CO₂. After treatment, 10 μL of MTT reagent was added to each well, and the plates were incubated for another 4 h. The medium was then discarded, and the formazan products were dissolved in 100 μL DMSO. Absorbance was recorded at 570 nm using a BioTek Synergy HT microplate reader, and IC₅₀ values were obtained by fitting dose-response curves.

Transfection of Aurora-A shRNA

The shRNA transduction conditions were set according to previously reported protocols with minor modifications⁶⁹. For shRNA-mediated Aurora-A knockdown, PC3 cells were transduced with lentiviral particles encoding Aurora-A shRNA or non-targeting control shRNA in Opti-MEM medium supplemented with 6 μg/mL polybrene. After 72 h of infection, cells were selected with 5 μg/mL puromycin to establish stable knockdown and control cell populations. Knockdown efficiency was verified before subsequent assays. The selected cells were then seeded into 96-well plates and treated with Peptide-1 at serial concentrations. Cell viability was measured using the MTT assay, and IC₅₀ values were calculated from fitted dose-response curves.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis

To assess downstream transcriptional changes, qRT-PCR was performed to quantify p53 and p21 mRNA levels in control and Peptide-1-treated PC3/p53WT cells (2 and 10 μM). The generation of PC3/p53WT cells was performed with reference to previous studies in which wild-type p53 was introduced into p53-null PC3 cells⁷⁰. Total RNA was extracted from control and the Peptide-1-treated PC3/p53WT cells using TRIzol reagent (Invitrogen, cat. no. 15596026, USA). RNA concentration and purity were measured with a NanoDrop 2000 spectrophotometer. Subsequently, cDNA was synthesised using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific, cat. no. 4374966, Waltham, MA, USA). qRT-PCR was performed on an Applied Biosystems 7500 Real-Time PCR System using SYBR Green qPCR Master Mix. Primer sequences for p53 and p21 were selected based on a previously reported study⁷¹. The primers were p53 forward, 5′-GTCTGGGCTTCTTGCATTCTG-3′, and reverse, 5′-GCTGTGACTGCTTGTAGATGGC-3′; p21 forward, 5′-CCTCAAATCGTCCAGCGACCTT-3′, and reverse, 5′-CATTGTGGGAGGAGCTGTGAAA-3′. GAPDH served as the internal reference gene. Relative mRNA expression of p53 and p21 was calculated using the 2^-ΔΔCt method and expressed as fold changes.

Statistical analysis

Statistical analyses were conducted utilising GraphPad Prism (GraphPad Software, Inc., San Diego, CA, USA). Intergroup comparisons were assessed via one-way analysis of variance (one-way ANOVA). p < 0.001 was considered highly statistically significant. All data are presented as the mean ± standard deviation (SD). Each experimental procedure was replicated independently at least three times.

Results

Virtual screening

Figure 1 illustrated the virtual screening workflow for Aurora-A-targeting peptides. In this study, virtual screening was performed to identify peptides with favourable predicted binding trends towards Aurora-A from a self-constructed peptide library comprising 59,319 peptides. Each peptide in the library was successively docked into the binding pocket of the prepared Aurora-A structure, and the peptides were ranked according to their predicted binding free energies, from the most negative values to the less negative ones. In general, a lower predicted binding free energy suggested a more favourable predicted interaction between the peptide and Aurora-A. Ultimately, the four top-ranked Aurora-A-targeting peptides (Peptides 1–4) were selected for subsequent interaction analysis. As shown in Figure 2, the predicted binding free energies of Peptides 1–4 towards Aurora-A were compared. As shown in Figure 3, the chemical structures of Peptides 1–4 and the positive control C23 were presented. Among them, Peptide-1 exhibited the lowest predicted binding free energy, suggesting that it might have shown the most favourable predicted interaction with Aurora-A in the docking model.

Figure 1.

Virtual screening workflow for the identification of Aurora-A-targeting peptides.

Figure 2.

Predicted binding free energies (kcal/mol) of Peptides 1–4 towards Aurora-A.

Figure 3.

The chemical structures of Peptides 1–4 and C23.

SAR analysis

As shown in Figure 4 and Table 1, Peptides 1–4 occupied the Aurora-A binding pocket and formed hydrogen-bonding and hydrophobic interactions with multiple residues. Arg126, Gln127, Trp128, Glu175, Arg179, Glu183, and Tyr199 consistently participated in hydrogen-bonding interactions with all four peptides, suggesting that these residues likely represent key recognition sites for this peptide series. In addition, all four peptides maintained hydrophobic contact with Leu178, indicating a possible role for this residue in complex stabilisation. Among the four peptides, Peptide-1 exhibited the most extensive interaction network, forming 18 hydrogen bonds, the highest number observed. It formed hydrogen bonds with Arg126, Gln127, Trp128, Asp132, Glu152, Glu170, Glu175, Arg179, Glu183, and Tyr199, with Glu183 and Tyr199 showing the greatest numbers of hydrogen-bond contacts. Although His187 was located near the binding region, no hydrogen-bonding interaction with Peptide-1 was identified in the present docking analysis. Peptide-2 displayed a broadly similar binding mode and formed 17 hydrogen bonds, but lacked interaction with Asp132. By comparison, Peptide-3 and Peptide-4 each formed 12 hydrogen bonds and showed less extensive interaction patterns overall. Peptide-3 retained hydrogen-bonding interactions with Arg126, Gln127, Trp128, Glu170, Glu175, Arg179, Glu183, His187, and Tyr199, whereas Peptide-4 interacted with Arg126, Gln127, Trp128, Glu175, Arg179, Glu183, and Tyr199. Taken together, Peptide-1 showed the most extensive and balanced network of polar and hydrophobic interactions with Aurora-A, followed by Peptide-2, whereas Peptide-3 and Peptide-4 displayed comparatively less elaborate interaction patterns. These interaction features may, at least in part, account for the more favourable predicted binding of Peptide-1.

Figure 4.

Predicted binding modes of Peptides 1–4 in the Aurora-A binding pocket. (A, B) Peptide-1 is coloured yellow; (C, D) Peptide-2, cyan; (E, F) Peptide-3, purple; and (G, H) Peptide-4, brown. In panels A, C, E, and G, Aurora-A is shown in cartoon representation, key residues are displayed as green sticks, and hydrogen bonds are indicated by black dashed lines. In panels B, D, F, and H, the binding pocket is presented in surface representation.

Table 1.

Hydrogen-bonding and hydrophobic interactions between Peptides 1–4 and key residues in the Aurora-A binding pocket.

Aurora-A residue	Interaction type	Peptide-1	Peptide-2	Peptide-3	Peptide-4
Arg126	H-bond	2	2	2	2
Gln127	H-bond	2	2	2	1
Trp128	H-bond	1	1	1	1
Asp132	H-bond	1	0	0	0
Glu152	H-bond	1	2	0	0
Glu170	H-bond	1	1	1	0
Glu175	H-bond	2	2	1	3
Arg179	H-bond	1	1	1	2
Glu183	H-bond	4	3	1	1
His187	H-bond	0	0	1	0
Tyr199	H-bond	3	3	2	2
Leu178	Hydrophobic	Yes	Yes	Yes	Yes

MST assay

To verify the binding tendency predicted by molecular docking, microscale thermophoresis (MST) was applied to quantitatively determine the binding affinities of Peptides 1–4 towards Aurora-A protein, with the well-characterised Aurora-A inhibitor C23 set as the positive control. As listed in Table 2, all tested molecules could specifically bind to Aurora-A, with their dissociation constant (K_d) values falling within the submicromolar to low micromolar range. Peptide-1 possessed the optimal binding affinity among all candidates, showing the lowest K_d value of 0.72 ± 0.04 µM. The positive control C23 exhibited a K_d of 1.22 ± 0.09 µM, which displayed weaker binding than Peptide-1 but remarkably higher affinity than the other three peptides. The binding affinities decreased gradually in the order of Peptide-2 (1.86 ± 0.13 µM), Peptide-3 (4.04 ± 0.25 µM) and Peptide-4 (8.15 ± 0.68 µM). Collectively, MST results confirmed that Peptide-1 exerted the strongest binding capability to Aurora-A among all tested compounds, and the positive control C23 possessed better binding activity than Peptides 2–4. Based on its superior binding affinity, Peptide-1 was selected from the tested peptides for subsequent MD simulation.

Table 2.

Binding affinities of Peptides 1–4 and the positive control C23, and their IC₅₀ values in PC3 cells.

Name	Kd ± SD (μM)	PC3 IC50± SD (µM)
Peptide-1	0.72 ± 0.04	1.15 ± 0.07
Peptide-2	1.86 ± 0.13	3.09 ± 0.20
Peptide-3	4.04 ± 0.25	7.51 ± 0.62
Peptide-4	8.15 ± 0.68	>10
C23	1.22 ± 0.09	3.70 ± 0.18

Binding selectivity of peptide-1 towards Aurora kinase isoforms

To further evaluate the target selectivity of Peptide-1, MST was also performed to determine its binding affinity towards Aurora-B and Aurora-C. As shown in Table S1, Peptide-1 showed much weaker binding affinities towards Aurora-B and Aurora-C, with K_d values greater than 10 μM for both isoforms. Compared with its binding affinity towards Aurora-A (K_d = 0.72 ± 0.04 µM), these results suggest that Peptide-1 may preferentially bind to Aurora-A over Aurora-B and Aurora-C under the present experimental conditions.

MD Simulations of the Aurora-A–Peptide-1 complex

To further evaluate the dynamic stability of the peptide identified from the preceding docking and MST analyses, a 50 ns MD simulation was performed for the Aurora-A–Peptide-1 complex. As shown in Figure 5(A), the RMSD of the complex remained relatively stable after the initial equilibration stage. The RMSF profile of Aurora-A Cα residues in Figure 5(B) showed generally low fluctuations, and the RMSF values of the key amino acid residues involved in peptide recognition were all below 0.26 nm, suggesting that peptide binding did not induce obvious local flexibility disturbances in the protein. Consistently, the major secondary structural elements of Aurora-A were largely maintained throughout the simulation (Figure 5C), while the radius of gyration remained essentially constant at approximately 1.98 nm (Figure 5D), indicating that the overall structural compactness of the protein was well preserved. In addition, the number of intermolecular hydrogen bonds remained at a relatively stable level during the simulation, and approximately 10 hydrogen bonds were still maintained at 50 ns (Figure 5E), suggesting a relatively stable interaction between Peptide-1 and Aurora-A. Taken together, these results indicate that the Aurora-A–Peptide-1 complex maintained overall conformational stability throughout the simulation.

Figure 5.

Molecular dynamics simulation and MM/PBSA analysis of the Aurora-A–Peptide-1 complex. (A) RMSD of the Aurora-A–Peptide-1 complex; (B) RMSF of the Cα residues of Aurora-A; (C) Secondary structure changes of Aurora-A during the simulation; (D) Radius of gyration (Rg) of Aurora-A; (E) Number of hydrogen bonds in the complex (distance <0.35 nm, with a standard bond angle); (F) Contribution of potential hot-spot residues to the binding free energy of Peptide-1; (G) Per-residue energy decomposition of potential hot-spot residues involved in Peptide-1 binding.

MD simulations of the Aurora-A–C23 complex

To validate the reliability of the molecular dynamics (MD) simulation system, a 50 ns MD simulation was performed for the positive-control Aurora-A–C23 complex using the same protocol as that used for the Aurora-A–Peptide-1 complex. As shown in Figure S1A, the RMSD of the Aurora-A–C23 complex increased during the initial stage and then fluctuated around 0.39 nm. Its overall RMSD fluctuation was greater than that of the Aurora-A–Peptide-1 complex, indicating that the Aurora-A–C23 complex exhibited relatively lower conformational stability and was less stable than the Aurora-A–Peptide-1 complex. Nevertheless, RMSF, secondary structure, and Rg analyses showed that Aurora-A still maintained low local residue fluctuations, stable major secondary structural elements, and relatively constant compactness in the C23-bound state, without obvious structural collapse or abnormal conformational changes during the simulation (Figure S1B–D). In addition, the Aurora-A–C23 complex formed fewer intermolecular hydrogen bonds, mainly fluctuating between 0 and 2, whereas the Aurora-A–Peptide-1 complex maintained approximately 10 hydrogen bonds at 50 ns (Figure S1E). These results supported the stability of the Aurora-A protein model and the reliability of the simulation system, while also suggesting that, compared with C23, Peptide-1 might form a more persistent interaction network and exhibit more favourable predicted binding stability towards Aurora-A.

MM/PBSA calculations of the Aurora-A–Peptide-1 complex

The interaction between Aurora-A and Peptide-1 was further analysed using the MM/PBSA method. As shown in Figure 5(F), per-residue energy decomposition indicated that the binding of Peptide-1 to Aurora-A was mainly driven by several key residues. Generally, residues with interaction energies lower than −1 kcal/mol are defined as key hotspot residues for ligand recognition. In the present study, Arg126, Gln127, Asp132, Glu152, Glu170, Glu175, Leu178, Arg179, Glu183, and Tyr199 all showed favourable energy contributions, among which Arg126 and Arg179 showed the most pronounced favourable electrostatic contributions in the decomposition analysis, suggesting that they may represent the major energetic hotspot residues. Further analysis showed that these favourable contributions were derived mainly from electrostatic interactions, whereas polar solvation partially offset this stabilisation to some extent (Figure 5G). In addition, Trp128 and His187 showed slight unfavourable contributions, while the net contributions of Leu178 and Tyr199 were relatively limited. The total binding free energy of the Aurora-A–Peptide-1 complex was calculated to be −64.13 kcal/mol, suggesting favourable binding in the MM/PBSA computational model. Overall, these results suggest that the binding of Peptide-1 to Aurora-A is mainly driven by electrostatic interactions, and that Arg126 and Arg179 may be key residues mediating complex stability.

MM/PBSA calculations of the Aurora-A–C23 complex

The binding characteristics of the Aurora-A–C23 complex were further evaluated using MM/PBSA calculations. The total binding free energy of the Aurora-A–C23 complex was calculated to be −29.77 kcal/mol, which was less favourable than that of the Aurora-A–Peptide-1 complex (−64.13 kcal/mol), suggesting that the Aurora-A–C23 complex may be less stable than the Aurora-A–Peptide-1 complex in the present MM/PBSA model. As shown in Figure S1F, the favourable energy contributions of C23 were mainly concentrated in several residues, including Glu170, Glu175, Leu178, Arg179, Glu183, and Tyr199. Among them, Arg179 showed the most pronounced favourable contribution, suggesting that this residue may play an important role in C23 recognition. Per-residue energy decomposition further showed that the favourable contributions were mainly derived from electrostatic interactions (Figure S1G). Compared with the Aurora-A–Peptide-1 complex, C23 involved fewer favourable hotspot residues. In contrast, Peptide-1 engaged additional residues, including Arg126, Gln127, Asp132, and Glu152, indicating a broader residue-interaction network. These results suggested that Peptide-1 may exhibit a more favourable predicted binding pattern towards Aurora-A than C23 in the present MM/PBSA model.

Free energy landscape

To further evaluate the conformational distribution of the Aurora-A–Peptide-1 complex during the molecular dynamics trajectory, a free energy landscape was constructed based on RMSD and Rg. As shown in Figure 6(A), the low-free-energy region was mainly concentrated within the ranges of RMSD of approximately 0.20–0.24 nm and Rg of approximately 1.97–1.99 nm, forming a relatively concentrated low-energy basin, which suggests that the complex mainly populated a dominant conformational region during the simulation. The three-dimensional free energy surface shown in Figure 6(B) further indicated that the system was characterised by a single major low-energy region, without obviously dispersed multiple independent deep energy basins, suggesting that no significant conformational rearrangement occurred during the simulation. Combined with the preceding molecular dynamics analysis, these results indicate that the Aurora-A–Peptide-1 complex maintained overall conformational stability during the simulation.

Figure 6.

Free energy landscape of the Aurora-A–Peptide-1 complex: (A) two-dimensional contour plot; (B) three-dimensional surface plot.

Serum stability assessment of peptides 1–4

To clarify the stability of Peptides 1–4, we performed a human serum stability assay and calculated the percentage of peptide remaining at different time points. As shown in Figure S2, Peptide-1 showed the highest stability, with approximately 65% remaining at 6 h and approximately 50% remaining at 8 h. Peptides 2–4 were all less stable than Peptide-1. These results indicate that Peptide-1 had relatively better serum stability among the four peptides, supporting its prioritisation for subsequent cellular assays.

Cellular uptake of peptide-1

To evaluate the cellular uptake of Peptide-1, PC3 prostate cancer cells were incubated with FITC-labelled Peptide-1 and observed using confocal laser scanning microscopy. As shown in Figure S3, clear green fluorescence signals were detected inside PC3 cells after treatment, suggesting detectable cellular uptake of Peptide-1. The FITC signal was mainly distributed in the cytoplasm and perinuclear region, with lower signal in the nucleus compared to the cytoplasm, suggesting preferential cytoplasmic localisation of Peptide-1. Hoechst 33342 staining clearly labelled the nuclei, and rhodamine-labelled dextran was used as a fluorescence reference for cellular imaging. Merged images further suggested that Peptide-1 was mainly distributed in the cytoplasm, which was consistent to some extent with the functional localisation of Aurora-A in cytoplasmic mitotic structures, including centrosomes and spindle poles, where Aurora-A participates in centrosome maturation and spindle assembly^60,72,73. Collectively, these results indicated that Peptide-1 possessed detectable cellular uptake capacity and provided support for its potential intracellular biological effects.

MTT assay

To further evaluate the in vitro antiproliferative activities of the Aurora-A-targeting peptides, an MTT assay was performed to examine the effects of Peptides 1–4 on the viability of PC3 cells. As shown in Table 2, all four peptides exhibited antiproliferative activity against PC3 cells, with varying potencies. Among them, Peptide-1 showed the strongest activity, with an IC₅₀ value of 1.15 ± 0.07 µM. Its activity was notably higher than that of Peptide-2 (3.09 ± 0.20 µM), Peptide-3 (7.51 ± 0.62 µM), and Peptide-4 (>10 µM). The positive control C23 showed moderate activity, with an IC₅₀ value of 3.70 ± 0.18 µM, which was weaker than Peptide-1 but stronger than Peptides 3 and 4.

To preliminarily assess its activity spectrum and selectivity, the effects of Peptide-1 were further evaluated in other prostate cancer cell lines. As shown in Table 3, Peptide-1 also exhibited relatively strong inhibitory activity against DU145 and NCI-H660 cells, with IC₅₀ values of 1.34 ± 0.11 μM and 1.82 ± 0.15 μM, respectively, whereas its IC₅₀ value in normal prostate RWPE-1 cells was >10 μM. Collectively, under the in vitro activity evaluation system of this study, Peptide-1 exhibited antiproliferative activity against PC3, DU145, and NCI-H660 cells, while showing minimal effects on normal RWPE-1 cells.

Table 3.

IC₅₀ values of Peptide-1 in prostate cancer cell lines and normal prostate epithelial cells.

Name	IC50 (μM) ± SD
	DU145	NCI-H660	RWPE-1
Peptide-1	1.34 ± 0.11	1.82 ± 0.15	>10

Transfection of Aurora-A shRNA

To preliminarily assess the relationship between the inhibitory activity of Peptide-1 and Aurora-A expression, an Aurora-A-knockdown PC3 cell model was established by shRNA transfection, and the IC₅₀ values of Peptide-1 in shControl and shAurora-A cells were determined. As shown in Table 4, the IC₅₀ value of Peptide-1 in shControl cells was 1.15 ± 0.07 μM, whereas its IC₅₀ value in shAurora-A cells was >10 μM. These results indicated that the sensitivity of PC3 cells to Peptide-1 was markedly reduced after Aurora-A knockdown, suggesting that the antiproliferative activity of Peptide-1 may be associated with the expression level of Aurora-A. Overall, these findings suggest that Aurora-A may be involved in the cellular activity of Peptide-1.

Table 4.

IC₅₀ values of Peptide-1 in shControl and shAurora-A PC3 cells.

Name	IC50 (μM) ± SD
	shControl	shAurora-A
Peptide-1	1.15 ± 0.07	>10

qRT-PCR analysis

Because parental PC3 cells are p53-deficient, PC3/p53WT cells were used to examine p53/p21-related transcriptional responses. To examine whether Peptide-1 affects the p53/p21-related transcriptional response, qRT-PCR was used to analyse the mRNA levels of p53 and p21 in PC3/p53WT cells. As shown in Figure 7, Peptide-1 treatment increased the transcript levels of both genes compared with the control group. At 2 μM and 10 μM, p53 and p21 expression levels were both upregulated, with a more pronounced increase observed in the 10 μM group, suggesting a potential concentration-related trend. Overall, Peptide-1 increased the mRNA expression of p53 and p21 under the present experimental conditions, indicating that its cellular effects may be associated with changes in p53 and p21 mRNA expression.

Figure 7.

Effects of Peptide-1 on p53 and p21 mRNA expression in PC3/p53WT cells. Relative mRNA expression levels of p53 and p21 were measured after treatment with different concentrations of Peptide-1. The control group consisted of untreated cells. Data are presented as mean ± SD. ***p < 0.001 versus the control group.

Conclusion

This study identified four peptides targeting Aurora-A through virtual screening, among which Peptide-1 showed the most favourable overall profile. Peptide-1 exhibited the lowest predicted binding free energy, the strongest binding affinity towards Aurora-A (K_d = 0.72 ± 0.04 μM), and the most extensive interaction pattern within the binding pocket. Molecular dynamics simulation indicated that the Aurora-A–Peptide-1 complex remained conformationally stable, while MM/PBSA analysis suggested that its binding was mainly driven by electrostatic interactions, with Arg126 and Arg179 likely serving as key energetic hotspot residues. Free-energy landscape analysis further showed that the complex mainly populated a dominant low-energy conformational region during the simulation. In cellular assays, Peptide-1 displayed the strongest antiproliferative activity in PC3 cells and also showed inhibitory activity in DU145 and NCI-H660 cells, whereas its activity was weaker in normal RWPE-1 cells. In addition, Aurora-A knockdown markedly reduced the activity of Peptide-1, and Peptide-1 treatment was associated with increased p53 and p21 mRNA expression in PC3/p53WT cells. Collectively, these results suggest that Peptide-1 may act as an Aurora-A-targeting peptide with antiproliferative activity in prostate cancer cells. Future studies will focus on further evaluating the antitumor activity and safety of Peptide-1 using relevant in vivo prostate cancer models. In addition, we will explore combination strategies with existing therapeutic agents and conduct further mechanistic studies to clarify the biological effects of Peptide-1. These future investigations will provide additional evidence for assessing its therapeutic relevance.

Funding Statement

This project was supported by the Fundamental Research Funds for the Central Universities.

Disclosure statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.