Structure-guided discovery of novel ERα-binding peptides with in vitro antiproliferative activity

Abstract

Uterine leiomyoma (UL) is an oestrogen-dependent tumour, and ERα is closely associated with abnormal proliferation of UL cells. This study used structure-guided virtual screening of a 59,319-member peptide library, followed by molecular docking, MST, MD simulation, MM/PBSA analysis, MTT assays, ERα knockdown, and RT-qPCR, to identify ERα-binding peptides. Four peptides showed ERα binding, among which Peptide-4 exhibited the lowest K_d value (9.3 ± 0.5 nM). Docking, MD simulation, and MM/PBSA analyses suggested a relatively stable ERα–Peptide-4 binding mode supported by hydrogen-bonding, hydrophobic, and van der Waals interactions. Peptide-4 inhibited SK-UT-1 cell proliferation (IC₅₀ = 0.77 ± 0.04 μM) with limited activity in HUtSMC cells, and this effect was markedly attenuated after ERα knockdown. Peptide-4 reduced c-Myc and Cyclin D1 mRNA levels, while combined treatment with Anti-miR-149G further increased the antiproliferative effect in SK-UT-1 cells. These findings provide a basis for further evaluation of Peptide-4 in ERα-related UL models.

Introduction

Uterine leiomyoma (UL) represents the most prevalent benign gynecological tumour among women of reproductive age, imposing a substantial health burden on affected patients.¹ Its typical clinical manifestations include menorrhagia, pelvic pain, and in severe cases, even infertility and recurrent miscarriage.^2,3 Currently, clinical therapeutic strategies for UL patients with heterogeneous symptoms commonly involve gonadotropin-releasing hormone (GnRH) agonists, selective progesterone receptor modulators (SPRMs), and oral contraceptives.⁴ Nevertheless, these pharmacological agents are generally associated with adverse effects such as reduced bone mineral density and increased risk of diabetes, which may limit their long-term administration.⁵ Consequently, hysterectomy remains the predominant curative intervention for symptomatic patients to date.⁶ However, hysterectomy is associated with prominent limitations, including high invasiveness and irreversibility, making it unsuitable for patients who wish to preserve fertility or avoid surgery. Thus, there remains a need to develop additional therapeutic agents targeting UL symptoms and associated complications. As an oestrogen-dependent neoplasm, the initiation and progression of UL are influenced by oestrogen-dependent signalling, which exerts its regulatory effects by binding to oestrogen receptors and modulating the physiological functions of target cells.^3,7

Oestrogen receptor alpha (ERα) belongs to the nuclear receptor superfamily and acts as a typical ligand-dependent transcriptional regulator.^8,9 ERα contains a ligand-independent activation function 1 (AF-1) domain at its N-terminus, a DNA-binding domain (DBD) in the central region, and a C-terminal ligand-binding domain (LBD) that additionally integrates the functional activity of ligand-dependent activation function 2 (AF-2).^10,11 As a core structural domain, LBD serves as a key regulatory element in the ligand-dependent signalling pathway, and can directly modulate the signal transduction activity of ERα.¹² Studies have shown that different types of ligands can induce conformational rearrangement of specific helical structures within the LBD, thereby causing the receptor to exhibit transcriptionally active or transcriptionally repressive functional states.^11,13 ERα antagonists or antihormonal agents can competitively bind to the ligand-binding pocket and induce conformational changes in the LBD, whereas the therapeutic efficacy of these agents is often attenuated by cellular adaptive alterations.^9,14 Targeting the ERα–coactivator interaction may provide a strategy for suppressing the gene-regulatory function of ERα.¹⁵ The function of ERα depends on the formation of the ERα-coactivator complex, and the process by which ERα recruits coactivators determines the transcriptional outcome.¹⁶ Inhibiting the protein-protein interaction between agonist-activated ERα and coactivators may attenuate ERα-mediated transcriptional activity.^17,18 Previous studies have identified a stapled peptide (R4K1), which exhibited high binding affinity for ERα and showed antiproliferative activity, which was associated with inhibition of the interaction between ERα and coactivators.¹⁹ This study aimed to identify ERα-binding peptides and compare their binding and cellular activity with those of R4K1.

ERα mediates the transcriptional expression of a variety of downstream genes, thereby regulating key cellular phenotypes such as proliferation, migration, and differentiation.^20,21 Notably, c-Myc and Cyclin D1 represent key ERα-targeted genes involved in cell proliferation.²² Studies have shown that downregulation of c-Myc and Cyclin D1 has been associated with reduced uterine leiomyoma cell proliferation.²³ In addition to canonical ERα-mediated transcriptional regulation, JunB, a member of the AP-1 transcription factor family, may also participate in oestrogen-related proliferative signaling.²⁴ Previous studies have shown that miRNA-149 can promote cell proliferation and suppress apoptosis through a JunB-associated mechanism, suggesting potential crosstalk between miRNA/JunB-related regulation and ERα-mediated expression of proliferation-associated genes.²⁵ Based on this rationale, ERα-targeted functional peptides may provide a strategy for suppressing uterine leiomyoma cell proliferation, whereas co-modulation of the miRNA/JunB-related pathway may influence the expression of proliferation-associated genes such as c-Myc and Cyclin D1. Therefore, we designed Anti-miR-149G and evaluated its combined effect with Peptide-4 in uterine leiomyoma cells.

Computer-aided drug design (CADD) enables the targeted design and optimisation of active molecules by predicting ligand-receptor interaction characteristics, and is characterised by high screening efficiency and low development cost.^26,27 As a core technology of CADD, molecular docking employs computer simulations to predict ligand-binding poses into the binding pockets of target proteins, and further facilitates the virtual screening of candidate compounds by evaluating ligand-receptor binding affinity.^28,29 Unlike molecular docking, which mainly provides static binding poses, molecular dynamics simulations enable time-dependent assessment of the conformational flexibility, structural stability, and interaction patterns of protein–peptide complexes under physiologically relevant simulated conditions.^30,31 In this study, a series of ERα-binding peptides were identified through molecular docking screening. In vitro evaluation results showed that Peptide-4 had a lower Kd value than R4K1 and reduced SK-UT-1 cell proliferation. Combination therapy may provide a strategy to modulate related signalling pathways more effectively than monotherapy.³² Therefore, we designed a nucleotide-based molecule and evaluated its combined effect with Peptide-4.

Materials and methods

Materials

Peptides 1–4, the positive peptide R4K1, scrambled peptide, and unrelated peptide were commercially synthesised by GL Biochem Ltd. (Shanghai, China). Peptide purity was determined by high-performance liquid chromatography (HPLC), with all peptides showing a purity of >98%. Their molecular weights and molecular weights were confirmed by LC–MS (mass spectrum), supporting the expected chemical identities. Recombinant human full-length ERα protein (aa 1–595; ab82606) and recombinant human full-length GPER protein carrying an N-terminal GST tag (aa 1–375; ab152434) were purchased from Abcam (Cambridge, MA, USA). Recombinant human ERβ ligand-binding domain protein (aa 244–530; GST-tagged; A15665) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). The Monolith Protein Labelling Kit RED-NHS 2nd Generation (Cat. No. MO-L011) and Monolith standard treated capillaries (Cat. No. MO-K022) were purchased from NanoTemper Technologies (Munich, Germany). Dimethyl sulfoxide (DMSO; HPLC grade; Cat. No. 34869) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Primers for c-Myc, Cyclin D1, and GAPDH, as well as the anti-miR-149G oligonucleotide, were commercially synthesised by Sangon Biotech (Shanghai, China). All primer and anti-miR-149G sequences were subjected to BLAST analysis to assess their target specificity and minimise potential off-target sequence matching.

Cell culture

The human uterine leiomyosarcoma cell line SK-UT-1 (HTB-114), primary human uterine smooth muscle cells (HuTSMCs; PCS-460–011), human umbilical vein endothelial cells (HUVECs; PCS-100–010), human adult dermal fibroblasts (HDFa; PCS-201–012), and HK-2 cells (CRL-2190) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The liver-derived LO2 cell line (CL-0111) was obtained from Procell Life Science & Technology Co., Ltd. (Wuhan, China). HuTSMCs, HUVECs, HDFa cells, and HK-2 cells were cultured in their respective ATCC-recommended complete growth media, whereas SK-UT-1 and LO2 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% foetal bovine serum and 1% penicillin–streptomycin. High-glucose DMEM (Cat. No. 11965092) and 100× penicillin–streptomycin solution (10,000 U/mL; Cat. No. 15140122) were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Heat-inactivated foetal bovine serum (FBS; Cat. No. 10082147) was obtained from Gibco (Grand Island, NY, USA). All cells were maintained at 37 °C in a humidified incubator containing 5% CO₂. Cells in the logarithmic growth phase were used for subsequent experiments.

Peptide library construction

A three-dimensional virtual heptapeptide library was constructed using the QuaSAR-CombiGen module in Molecular Operating Environment (MOE).³³ Predefined dipeptide and tripeptide fragment libraries were used as building blocks to generate heptapeptide sequences through combinatorial enumeration. N-terminal acetylation and C-terminal amidation were applied as default terminal modifications to mimic terminally capped peptides and reduce terminal charge effects during structure-based screening. After removal of duplicate or invalid entries, three-dimensional peptide structures were generated in MOE and energy-minimized using the Amber10:EHT force field. The final library contained 59,319 peptide sequences and was used for subsequent molecular docking-based virtual screening.

Molecular docking-based screening of the virtual peptide library

The crystal structure of the ERα in complex with stabilised peptide antagonist 6 (PDB ID: 5GTR) was retrieved from the Protein Data Bank (PDB) and used for subsequent molecular docking-based screening. Molecular docking-based virtual screening was performed using MOE software. The ERα structure was imported into MOE and prepared using the QuickPrep module. The docking site was defined according to the original co-crystallized ligand-binding region, and the wall constraint in MOE was applied to restrict the docking search space.³⁴ The virtual peptide library, comprising 59,319 energy-minimized three-dimensional peptide conformations, was subjected to molecular docking into the predefined binding region of ERα using the Dock module in MOE. Docking was performed using the triangular matching placement method and the London dG scoring function. The docking scores were calculated to estimate the predicted binding favorability between each peptide and ERα. Peptides with lower docking scores were considered to have more favourable predicted binding poses and were selected for further analysis. The positive-control peptide R4K1 was docked under the same conditions for comparison.

Microscale thermophoresis assay

The binding affinities of the peptides towards recombinant human ERα protein were determined by microscale thermophoresis (MST).³⁵ The MST assay buffer consisted of potassium dihydrogen phosphate (KH₂PO₄), disodium hydrogen phosphate (Na₂HPO₄), sodium chloride (NaCl), potassium chloride (KCl), and 0.1% (v/v) Tween 20. Before the assay, ERα was fluorescently labelled using a commercial protein labelling kit, and serially diluted peptide solutions were prepared. Each peptide dilution was mixed with the fluorescently labelled ERα solution at a 1:1 volume ratio and incubated at room temperature for 5–10 min to allow ERα–peptide complex formation. The mixtures were then loaded into capillaries and analysed using a Monolith NT.115 instrument (NanoTemper Technologies). MST measurements were performed in temperature-jump mode with the MST power set to 20%. The binding affinities of Peptide-4, unrelated peptide, and scrambled peptide towards ERβ or GPER were determined by MST using the same procedure as described for ERα, except that fluorescently labelled recombinant human ERβ or GPER protein was used as the binding target.

Molecular dynamics simulations

To further evaluate the conformational stability of the ERα–Peptide-4 and ERα–R4K1 complexes under the simulated conditions, molecular dynamics (MD) simulations were performed using GROMACS version 2021.5. The top-ranked binding poses obtained from molecular docking were used as the initial structures. Topology parameters for ERα and each peptide were generated separately and then merged to construct the corresponding ERα–peptide complex systems. Specifically, peptide topology parameters were generated using the GAFF2 force field via the Beikun Cloud platform (https://www.bkunyun.com/v2/console/cloude-dashboard), whereas ERα topology parameters were generated using the AMBER99SB-ILDN force field.^36,37 The ERα–Peptide-4 complex and the ERα–R4K1 peptide complex were simulated under the same conditions. Each complex was solvated in a cubic box using the SPC water model, with a minimum distance of 1.0 nm between the protein and the box edge. Na⁺ and Cl^- ions were added to neutralise the total system charge. Each system was first subjected to 5000 steps of steepest-descent energy minimisation to remove unfavourable contacts. Subsequently, 100 ps NVT and 100 ps NPT equilibration simulations were performed sequentially. During equilibration, the temperature was maintained at 300 K using the V-rescale thermostat, and the pressure was maintained at 1 bar using the Parrinello–Rahman barostat. After equilibration, 100-ns production MD simulations were performed for each system. Simulation trajectories were analysed using built-in GROMACS tools, including the backbone root-mean-square deviation (RMSD) of the ligand, root-mean-square fluctuation (RMSF) of protein residues, radius of gyration (Rg), secondary structure changes, and hydrogen-bond number. Hydrogen-bond occupancy was further calculated using a Python script.

MM/PBSA calculation

MM/PBSA (Molecular Mechanics/Poisson–Boltzmann Surface Area) analysis was performed to evaluate the relative binding free energies and key residue contributions of the ERα–Peptide-4 and ERα–R4K1 complexes under the simulated conditions.³⁸ Based on the production trajectories obtained from the MD simulations described above, structural snapshots were extracted at fixed intervals after the systems reached relative stability and were used for subsequent calculations. The ERα–Peptide-4 and ERα–R4K1 complexes were analysed using the same computational workflow and parameters to ensure comparability between the two systems. The binding free energy was calculated as ΔG_bind = G_complex − G_receptor − G_peptide and further decomposed into van der Waals interaction energy (ΔE_vdW), electrostatic interaction energy (ΔE_ele), polar solvation energy (ΔG_polar), and nonpolar solvation energy (ΔG_nonpolar). The total binding free energy was expressed as ΔG_bind = ΔE_vdW + ΔE_ele + ΔG_polar + ΔG_nonpolar. Residue energy decomposition analysis was then performed for ERα to identify residues with major contributions to Peptide-4 or R4K1 binding. The MM/PBSA results were used to compare the relative binding energy profiles of the ERα–peptide complexes and to characterise the main energetic contributions.

Free energy landscape

Free energy landscape (FEL) analysis was performed to characterise the conformational distributions and relative stable states of the ERα–peptide complexes during MD simulations. After correction for periodic boundary conditions and structural alignment, RMSD and Rg of each complex were calculated and used as collective variables. FELs were generated using the gmx sham tool in GROMACS by projecting the conformational space onto RMSD–Rg coordinates and calculating the relative free energy according to ΔG = −kBT ln P, where kB is the Boltzmann constant, T is the simulation temperature, and P is the conformational probability.³⁹ The resulting FELs were visualised as two-dimensional heatmaps and three-dimensional free-energy surfaces to identify low-energy conformational clusters.

Serum stability assays

The serum stability of Peptide-4 was evaluated by HPLC analysis. Briefly, human serum was diluted to 25% (v/v) with PBS and centrifuged at 15,000 rpm for 10 min. The supernatant was collected and used for the stability assay. Peptide-4 stock solution was added to the serum supernatant to a final concentration of 5 μM and incubated at 37 °C. At predetermined time points (0, 2, 4, 8, 16, and 24 h), 200 μL aliquots were collected and immediately mixed with an equal volume of acetonitrile to precipitate serum proteins and terminate the reaction. After incubation at 4 °C overnight, the samples were centrifuged at 15,000 rpm for 10 min, and the supernatants were subjected to HPLC analysis.⁴⁰ The remaining Peptide-4 content was calculated based on the integrated peak area of Peptide-4 and normalised to the peak area at 0 h, which was defined as 100%.

MTT assays

Cells in the logarithmic growth phase were seeded into 96-well plates at a density of 5 × 10³ cells/well and cultured overnight in an incubator.³⁵ Subsequently, serially diluted peptide solutions at gradient concentrations were separately added to each well, followed by continuous incubation for 48 h at 37 °C in a humidified atmosphere containing 5% CO₂. After the incubation period, 20 μL of 0.5% MTT solution was added to each well, with further incubation for 4 h. The supernatant was then removed, and 150 μL of dimethyl sulfoxide (DMSO) was added to each well, followed by shaking for 10 min to completely dissolve the formazan crystals. Finally, the absorbance of each well at 570 nm was measured using a microplate reader.

Long-term cytotoxicity was evaluated using a 5-day continuous-exposure MTT assay. SK-UT-1, HuTSMCs, HUVECs, LO2, HDFa, and HK-2 cells were used as in vitro cell models. Cells in the logarithmic growth phase were seeded into 96-well plates and, after attachment, treated with Peptide-4, scrambled peptide, or unrelated peptide at serial concentrations up to 10 μM for 5 consecutive days. Fresh peptide-containing medium was replaced daily, and vehicle-treated cells served as controls. At the end of treatment, MTT solution was added to each well and incubated for 4 h. The supernatant was removed, and the formazan crystals were dissolved in DMSO. Absorbance was measured at 570 nm using a microplate reader, and cell viability was calculated relative to the vehicle control group. When cell viability did not decrease below 50% within the tested concentration range, the IC₅₀ value was reported as >10 μM.

The logarithmically growing SK-UT-1 cells were seeded into 96-well plates at a density of 5 × 10³ cells per well, followed by incubation at 37 °C for 24 h to allow cell attachment. Cells were divided into three groups: Peptide-4 group, anti-miR-149G group, and Peptide-4 combined with anti-miR-149G group. An untreated control group was established in parallel, and all other culture and detection conditions were consistent across groups. After treatment, the cells were incubated for an additional 48 h under the same culture conditions. The absorbance was measured using the MTT assay, and the cell proliferation inhibition rate (%) was calculated. The sequence of anti-miR-149G (UUCUUCCCGAGGCCACAUGCUUCUUUAUAUCCCCAUAUGGA UUACUUUGCUAUGGAAUGUAAGGAAGUGUGUGGUUUCGGCAAGGG) is presented in the 5′–3′ orientation.

Lentiviral transduction of ERα shRNA

The lentiviral particles carrying shRNA targeting ERα were purchased from Merck (Sigma-Aldrich, St. Louis, USA). Polybrene was diluted to 6 μg/mL in Opti-MEM medium (Gibco, Grand Island, NY) and added to SK-UT-1 cells together with the lentiviral particles. After 72 h of transduction incubation at 37 °C in a humidified incubator containing 5% CO₂, stable transduced cells were selected in 6-well plates using culture medium containing 5 μg/mL puromycin.^33,41 ERα knockdown efficiency was verified by RT-qPCR before subsequent experiments. The selected stable cell lines were seeded into 96-well plates, and IC₅₀ values were determined using the MTT assay under the same conditions as described above. The shControl group was included as a negative control to account for non-specific effects associated with lentiviral transduction and puromycin selection.

Reverse transcription quantitative polymerase chain reaction

SK-UT-1 cells were seeded in 6-well plates at a density of 5 × 10⁵ cells/well. When cells reached 70% confluence, they were serum-starved for 12 h to reduce basal serum-associated effects. The cells were then divided into three groups: vehicle control, 1 μM Peptide-4, and 5 μM Peptide-4. After 24 h of treatment, cells were collected for total ribonucleic acid (RNA) extraction. To extract total RNA, cells were subjected to lysis using TRIzol reagent (catalog no. 15596026, Invitrogen Corporation, Carlsbad, CA, USA). The purity of total RNA was verified by NanoDrop, with an A260/A280 ratio ranging from 1.8 to 2.0. cDNA was synthesised from 1 μg of total RNA using the Thermo Fisher High-Capacity cDNA Reverse Transcription Kit (Cat. No. 4374966, Waltham, MA, USA). Quantitative real-time PCR (RT-qPCR) was performed using SYBR Green qPCR Master Mix on an Applied Biosystems 7500 Real-Time PCR System to determine the relative mRNA expression levels of c-Myc and Cyclin D1, with GAPDH used as the internal reference gene. The primer sequences are listed in Table S1. The PCR cycling conditions were as follows: initial denaturation at 95 °C for 3 min, followed by 40 cycles of denaturation at 95 °C for 10 s and annealing/extension at 60 °C for 30 s. A melting-curve analysis was performed from 60 °C to 95 °C to verify the specificity of the amplification products.⁴² Relative gene expression was calculated using the 2^−ΔΔCt method.

Statistical analysis

The data processing and visualisation were conducted with GraphPad Prism software (version 9.5, GraphPad Software Inc, San Diego, CA, USA). The number of independent replicates for each experiment or analysis is specified in the corresponding figure legend or table footnote. Data are presented as the mean ± standard deviation (SD). Statistical significance was determined by one-way ANOVA (Analysis of Variance) followed by Dunnett’s multiple-comparisons test for comparisons with the control group. A value of P < 0.05 was considered statistically significant. Statistical significance is indicated as follows: *** P < 0.001.

Results and discussion

Molecular docking-based virtual screening of ERα-binding peptides

In this study, we applied a molecular docking-based virtual screening strategy to identify potential ERα-binding peptides from a database containing 59,319 peptide sequences. The detailed screening workflow is illustrated in Figure 1. First, the crystal structure of ERα (PDB ID: 5GTR) was obtained and subjected to standard preprocessing using MOE. The preprocessed structure was then used for subsequent molecular docking. Peptide sequences in the database were converted into 3D conformations, and docked into the ERα binding site using a standard force field and scoring function to estimate their putative binding potential towards the ERα protein. The reported ERα inhibitor, R4K1, was used as the positive control, and its docking score against ERα was used as a reference threshold for screening. As shown in Figure 2, four top-ranked peptides (Peptides 1–4) based on docking scores were selected. To compare the predicted binding profiles of Peptides 1–4 with that of the positive control R4K1, we analysed their docking-predicted binding free energies and standardised effect sizes relative to R4K1. The mean docking scores of Peptides 1–4 were −8.44 ± 0.34, −9.13 ± 0.61, −9.47 ± 0.72, and −11.89 ± 0.48 kcal/mol, respectively, all of which were lower than that of R4K1 (−6.08 ± 0.40 kcal/mol). After Dunnett’s multiple-comparisons adjustment, these differences were statistically significant (adjusted P < 0.001, Table S2). Among them, Peptide-4 showed the lowest mean docking score and a large standardised effect size (Hedges’ g = −12.12), suggesting a relatively more favourable predicted ERα-binding profile in this computational model. Based on these results, Peptides 1–4 were selected for subsequent experimental validation, and their chemical structures are shown in Figure 3. SciFinder searches of the structures of Peptides 1–4 did not identify any previously reported compounds with identical structures. This result provides preliminary support for the structural novelty of Peptides 1–4 within the searched database.

Figure 1.

Identification workflow of Peptide-4 as an ERα-binding peptide. Molecular docking was used to screen a library of 59,319 peptide sequences, from which four top-ranked peptides were selected for binding-affinity assessment. Structure–binding relationship analysis was then performed based on docking and binding-affinity results, and Peptide-4 was selected for further computational and cell-based evaluation.

Figure 2.

Comparison of docking-predicted binding scores between Peptides 1–4 and the control peptide R4K1. Peptides 1–4 showed significantly lower docking-predicted binding scores than R4K1, with Peptide-4 exhibiting the lowest docking score among the tested peptides. Data are presented as mean ± SD (n = 6). ***P < 0.001 vs R4K1. Corresponding Hedges’ g standardised effect sizes and their 95% confidence intervals are summarised in Table S2.

Figure 3.

Chemical structures of Peptides 1–4 selected for ERα-binding evaluation. The structures are presented to show the peptide backbone and side-chain composition of the four selected peptides.

MST-based evaluation of ERα binding affinity and selectivity

To experimentally assess the binding of the virtually screened peptides to ERα, MST was used to determine the binding affinities of Peptides 1–4, the positive-control peptide R4K1, and sequence-control peptides. Peptides 1–4 showed K_d values of 20.5 ± 1.3, 18.7 ± 1.1, 15.6 ± 0.9, and 9.3 ± 0.5 nM, respectively, all of which were lower than that of R4K1 (K_d = 24.1 ± 1.4 nM) (Table 1). Among them, Peptide-4 exhibited the lowest K_d value, indicating relatively higher ERα-binding affinity among the tested peptides. The scrambled peptide was generated by random rearrangement of the Peptide-4 amino acid sequence to evaluate whether the binding and cellular activity of Peptide-4 were dependent on its specific sequence arrangement. The unrelated peptide was used as a sequence-unrelated control peptide. A UniProt Peptide Search of its amino acid sequence did not reveal any obvious association with ERα, ERβ, GPER, or known ERα-associated coactivator sequences. In contrast, the scrambled and unrelated peptides did not yield reliable binding curves within the tested concentration range and were therefore reported as N.D. (>10 μM), supporting sequence-dependent ERα binding of Peptide-4.

Table 1.

MST-derived binding affinities of Peptides 1–4, the positive-control peptide R4K1, scrambled peptide, and unrelated peptide towards ERα.

Name	Kd (nM) mean ± SD
Peptide-1	20.5 ± 1.3
Peptide-2	18.7 ± 1.1
Peptide-3	15.6 ± 0.9
Peptide-4	9.3 ± 0.5
R4K1a	24.1 ± 1.4
Scrambled peptideb	N.D. (>10 μM)
Unrelated peptideb (Ac-WTGATYL-NH2)	N.D. (>10 μM)

aR4K1 was used as the positive-control peptide.

bN.D., not determined. No reliable binding curve was obtained within the tested concentration range.

To further assess the target selectivity of Peptide-4, we compared its binding to ERα, ERβ, and GPER. MST analysis showed that Peptide-4 bound to ERα with a Kd value of 9.3 ± 0.5 nM, whereas no reliable binding curves were obtained for ERβ or GPER (N.D., >10 μM) (Table 2). Similarly, the scrambled peptide and unrelated peptide did not yield reliable binding curves to ERα, ERβ, or GPER. These results suggest that Peptide-4 showed apparent ERα-binding selectivity under the present MST assay conditions in this assay system. In this study, the scrambled peptide was generated by random rearrangement of the original Peptide-4 sequence to assess the influence of sequence order on ERα binding. The unrelated peptide Ac-WTGATYL-NH₂ was used as a sequence-unrelated control to exclude non-specific peptide binding. Overall, the MST results were consistent with the docking-based screening results and supported the selection of Peptide-4 for subsequent cell-based evaluation.

Table 2.

MST-derived binding selectivity of Peptide-4 towards ERα, ERβ, and GPER.

Name	Kd (nM) ± SD
	ERα	ERβ	GPER
Peptide-4	9.3 ± 0.5	N.D. (>10 μM)	N.D. (>10 μM)
Scrambled peptide	N.D. (>10 μM)	N.D. (>10 μM)	N.D. (>10 μM)
Unrelated peptide	N.D. (>10 μM)	N.D. (>10 μM)	N.D. (>10 μM)

Note. Values are presented as mean ± SD from MST measurements. For samples without a reliable binding curve, binding affinity is reported as N.D. (>10 μM). N.D., not determined. Scrambled peptide and unrelated peptide were used as sequence-control peptides.

Structure-binding relationship analysis of peptides 1–4 towards ERα

To explore the structural features associated with the binding characteristics of Peptides 1–4 to ERα, we analysed their interaction patterns with residues at the ERα binding interface. The distribution of key residues and the spatial conformation of the binding pocket in ERα-Peptides 1–4 complexes are shown in Figure 4, and the interaction types between each peptide and ERα are summarised in Table 3. Peptides 1–4 were predicted to form hydrogen bonds with the Lys362 and Glu542 residues of ERα, suggesting that these residues may contribute to a conserved interaction pattern between the peptides and ERα. In addition, Peptide-3 was predicted to form one hydrogen bond with Asp538, which may further support its interaction with the receptor. Peptide-4 formed an additional hydrogen bond with Asn359, indicating a more extensive hydrogen-bonding pattern than that observed for Peptide-3. Consistent with this interaction profile, Peptide-4 was predicted to form the highest number of hydrogen bonds among Peptides 1–4, which may be associated with its ERα-binding affinity. Among the tested peptides, Peptide-4 showed the lowest K_d value (K_d = 9.3 ± 0.5 nM).

Figure 4.

Predicted interaction modes of Peptides 1–4 within the ERα ligand-binding pocket. (A, C, E, G) Ribbon diagrams showing the docking-derived binding poses of Peptide-1 (yellow, A), Peptide-2 (red, C), Peptide-3 (green, E), and Peptide-4 (cyan, G) in the ERα ligand-binding pocket. Key residues surrounding the peptide-binding region, including Val355, Ile358, Asn359, Lys362, Phe367, Leu372, Val376, Leu379, Asp538, Leu539, Glu542, and Met543, are shown in gray sticks. Hydrogen bonds are indicated by black dashed lines. (B, D, F, H) Corresponding surface electrostatic potential maps of the ERα pocket show the spatial distribution of charged and neutral regions around the docked peptides. Red, green, and purple indicate negative, neutral, and positive electrostatic potential regions, respectively.

Table 3.

Docking-predicted interaction profiles of Peptides 1–4 in the ERα ligand-binding pocket.

Name	Hydrogen bonding interactions		Hydrophobic/contact residuesb
	Interacting residues	No. of hydrogen bondsa
Peptide-1	Lys362, Glu542	3	Phe367, Leu372, Val376, Leu379, Leu539, Val355, Ile358, Met543, Asn359, Asp538
Peptide-2	Lys362, Glu542	3	Phe367, Leu372, Val376, Leu379, Leu539, Val355, Ile358, Met543, Asn359, Asp538
Peptide-3	Lys362, Glu542, Asp538	4	Phe367, Leu372, Val376, Leu379, Leu539, Val355, Ile358, Met543, Asn359
Peptide-4	Lys362, Glu542, Asp538, Asn359	5	Phe367, Leu372, Val376, Leu379, Leu539, Val355, Ile358, Met543

aThe number indicates the predicted hydrogen bonds formed between each peptide and ERα residues in the docking model.

bResidues listed in this column indicate hydrophobic or close-contact residues surrounding the docked peptide. Binding strength should be interpreted together with docking scores, binding-affinity data, and MD simulation results.

Furthermore, Peptides 1–4 showed multiple hydrophobic contacts with hydrophobic residues within the ERα binding pocket, including Phe367, Leu372, Val376, Leu379, Leu539, Val355, Ile358, and Met543. These hydrophobic interactions may contribute to the stabilisation of the peptide–receptor complexes. Compared with Peptide-1, Peptides 2–4 appeared to form more extensive hydrophobic interactions, which may favour predicted spatial complementarity with the ERα binding pocket. Overall, the predicted recognition of Peptides 1–4 by ERα appears to involve the combined contribution of hydrogen bonds and hydrophobic interactions. In particular, the multi-site hydrogen-bonding pattern and hydrophobic contacts of Peptide-4 may be associated with its relatively favourable docking score and MST-derived binding affinity.

Comparative analysis of binding stability and conformational dynamics of peptide-4 and R4K1

To compare the conformational behaviour of the Peptide-4- and R4K1-bound complexes, RMSD, RMSF, Rg, secondary-structure evolution, and hydrogen-bond formation were analysed over the MD trajectories. Overall, the Peptide-4-bound complex showed lower conformational fluctuations and a relatively more stable dynamic profile under the simulated conditions under the present simulation conditions, whereas the R4K1-bound complex formed more hydrogen bonds but exhibited greater global and local structural fluctuations.

RMSD was used to evaluate the conformational deviation of each ligand from its initial binding pose during the MD simulation.³⁰ Peptide-4 rapidly reached a relatively equilibrated state and was mainly maintained at approximately 0.15–0.25 nm, with only a slight increase near the end of the simulation, indicating limited deviation from its initial binding conformation (Figure 5A). In contrast, R4K1 showed a higher RMSD, mainly ranging from approximately 0.40 to 0.55 nm, together with more pronounced periodic fluctuations, suggesting greater conformational rearrangement within the binding pocket (Figure S1A). These results suggest that Peptide-4 maintained a relatively more stable binding pose than R4K1 under the present simulation conditions.

Figure 5.

Dynamic stability and energetic characterisation of the ERα–Peptide-4 complex during molecular dynamics simulations. (A) Peptide-4 RMSD remained within a narrow range (0.15–0.22 nm) after initial equilibration, suggesting that the bound peptide maintained a relatively stable binding-related conformation. (B) RMSF analysis showed limited residue-level fluctuations, indicating limited local flexibility around the binding region. (C) The Rg profile fluctuated narrowly around 1.85–1.88 nm, indicating that the complex maintained a compact conformation without large changes in global compactness. (D) Secondary-structure analysis indicated that the major structural elements were largely preserved throughout the simulation. (E) The hydrogen-bond number between the ERα and Peptide-4 varied during the simulation, indicating dynamic hydrogen-bond interactions at the binding interface. (F) Per-residue MM/PBSA decomposition identified residues with favourable energetic contributions to Peptide-4 binding. (G) Energy-component decomposition showed that van der Waals and electrostatic interactions contributed favourably, whereas polar solvation partially opposed binding.

RMSF analysis was performed to evaluate residue-level flexibility during the MD simulations.⁴³ In the Peptide-4-bound complex, most residues showed RMSF values below 0.20 nm, with only a few local peaks of approximately 0.30–0.40 nm, indicating relatively low flexibility at the binding interface (Figure 5B). In contrast, the R4K1-bound complex exhibited higher RMSF peaks in multiple regions, with several residues approaching 0.40–0.50 nm, suggesting more pronounced local conformational fluctuations upon R4K1 binding (Figure S1B). Together with the RMSD results, these findings suggest that Peptide-4 binding was associated with lower local conformational fluctuations under the present simulation conditions.

Rg analysis was used to assess the overall compactness of the protein during the MD simulations.⁴⁴ Both complexes maintained relatively stable Rg values of approximately 1.85–1.89 nm, with no clear continuous increasing or decreasing trend, suggesting that neither ligand induced substantial changes in global protein compactness over the simulated timescale (Figure 5C and Figure S1C). The Peptide-4-bound complex showed a slightly smoother Rg profile, whereas the R4K1-bound complex exhibited larger RMSD/RMSF fluctuations. These results suggest that the differences between the two systems are more likely related to local conformational dynamics than to global structural changes.

Secondary-structure analysis was performed to assess whether the protein backbone architecture was maintained during the MD simulations.³¹ The results showed that both systems retained relatively stable secondary-structure compositions over the MD simulations. The total number of structured residues was mainly maintained at approximately 190–200, and the number of α-helical residues remained at approximately 145–160. No obvious continuous changes were observed in the numbers of β-sheet, turn, bend, or coil residues. These observations suggest that binding of either ligand was not associated with substantial disruption of the overall protein fold during the simulated timescale (Figure 5D and Figure S1D). Together with the RMSD and RMSF results, the structural variations observed in the R4K1-bound system are more likely related to local conformational adjustments near the binding interface rather than large-scale secondary-structure rearrangements.

Hydrogen-bond number reflects polar interactions between the ligand and protein, but it should not be interpreted as a direct measure of overall complex stability.⁴⁵ The Peptide-4-bound complex formed fewer hydrogen bonds, generally maintaining approximately 1–2 hydrogen bonds during the simulation and occasionally reaching around 3 (Figure 5E). Occupancy analysis identified only two intermolecular hydrogen bonds with occupancies >10%, Peptide-4-Glu380 (13.00%) and Peptide-4-Glu542 (15.87%), suggesting a limited but intermittently maintained hydrogen-bonding pattern. By contrast, R4K1 formed more hydrogen bonds, typically approximately 4–10 (Figure S1E), with several occupancies >10%, including Asn359 (14.57%), Lys362 (11.91%/13.47%), Arg363 (19.16%), Lys467 (22.33%), Ser305 (17.62%), and Val368 (16.05%) (Table 4). Although these results indicate more extensive polar contacts for R4K1, the pronounced temporal fluctuation in hydrogen-bond number suggests that these interactions were highly dynamic. Thus, a larger number of hydrogen bonds does not necessarily imply higher overall complex stability.

Table 4.

Intermolecular hydrogen-bond occupancy of Peptide-4- and R4K1-bound complexes during MD simulations.

Name	Interacting residue	Number of hydrogen bonds	H-bond occupancy (%)
Peptide-4	Glu380	1	13.00
	Glu542	1	15.87
R4K1	Asn359	1	14.57
	Lys362	2	11.91/13.47
	Arg363	1	19.16
	Lys467	1	22.33
	Ser305	1	17.62
	Val368	1	16.05

Note. Hydrogen-bond occupancy was calculated as the percentage of MD trajectory frames in which a given intermolecular hydrogen bond was detected. Only hydrogen bonds with occupancies >10% are listed. For residues forming more than one hydrogen bond, occupancy values are separated by slashes.

Overall, the Peptide-4-bound complex displayed lower RMSD and RMSF fluctuations, a stable Rg profile, and no apparent disruption of secondary-structure composition, suggesting a relatively stable binding conformation under the present simulation conditions. In contrast, although R4K1 formed more hydrogen bonds and showed relatively higher hydrogen-bond occupancies for several residues, its larger conformational fluctuations and more pronounced energetic compensation suggest a more dynamic binding pattern. These findings suggest that Peptide-4 may maintain a relatively more stable and coordinated binding mode under the simulated conditions than R4K1 within the simulated timescale.

MM/PBSA analysis of binding energetics

To further compare the binding characteristics of Peptide-4 and R4K1, MST-derived binding affinities were analysed together with MM/PBSA-calculated binding free energies. The MST-derived dissociation constants were converted into apparent experimental standard binding free energies using the thermodynamic relationship ΔG_exp = RT ln(K_d/1 M), where R = 1.987 × 1 0 –3 kcal/mol/K and T = 298.15 K. Based on the MST results, Peptide-4 showed a K_d of 9.3 ± 0.5 nM, corresponding to the apparent ΔG_exp of approximately −10.96 kcal/mol, whereas R4K1 showed the K_d of 24.1 ± 1.4 nM, corresponding to an apparent ΔG_exp of approximately −10.39 kcal/mol. The lower K_d and slightly more favourable ΔG_exp suggest that Peptide-4 exhibits slightly stronger apparent experimental binding affinity than R4K1.

MM/PBSA analysis further showed that Peptide-4 showed a more favourable MM/PBSA-calculated total binding free energy than R4K1, with ΔG_bind values of −29.09 ± 3.71 and −22.14 ± 1.79 kcal/mol, respectively. For Peptide-4, van der Waals, electrostatic, and nonpolar solvation energies contributed favourably to binding, whereas polar solvation energy showed an unfavourable contribution. In comparison, R4K1 exhibited a much stronger favourable electrostatic contribution, which was largely counterbalanced by a highly unfavourable polar solvation term (Table 5). This pronounced electrostatic–polar solvation compensation suggests that R4K1 binding may involve stronger charged interactions accompanied by a greater desolvation penalty.

Table 5.

MM/PBSA-calculated binding free energy components (kcal/mol) for the binding of Peptide-4 or R4K1 to ERα.

Energy terms (kcal/mol)	Peptide-4	R4K1
Van der waals energy	−39.91 ± 2.64	−37.77 ± 0.99
Electrostatic energy	−65.00 ± 8.71	−457.31 ± 18.83
Polar solvation energy	81.60 ± 2.39	479.08 ± 17.08
Nonpolar solvation energy	−5.78 ± 0.10	−6.14 ± 0.07
Total gas-phase free energy	−104.91 ± 9.10	−495.08 ± 18.86
Total solvation free energy	75.82 ± 2.39	472.94 ± 17.08
Total binding free energy	−29.09 ± 3.71	−22.14 ± 1.79

Residue-level energy decomposition further supported the distinct binding characteristics of the two ligands. In the Peptide-4-bound complex, favourable contributions were mainly distributed among several interfacial hydrophobic residues, including Ile358, Leu372, Val376, Leu379, Leu539, and Met543, suggesting that hydrophobic and van der Waals interactions may help maintain its binding conformation (Figure 5F, G). In contrast, although R4K1 showed favourable contributions from Arg363, Leu372, Glu375, and Leu539, the component decomposition indicated a more pronounced compensation between electrostatic and polar solvation terms, especially for charged-residue-mediated interactions (Figure S1F, G). These results suggest that Peptide-4 may adopt a more balanced interfacial interaction profile, whereas R4K1 binding appears to involve stronger but more highly compensated polar interactions.

It should be noted that the absolute binding free energies calculated by MM/PBSA were substantially more negative than the MST-derived ΔG_exp values. Therefore, the two values should not be considered quantitatively equivalent. This discrepancy may be attributed to incomplete treatment of entropic contributions in MM/PBSA calculations, differences in solvent and ionic environments between the simulated and experimental systems, and limited trajectory sampling that may favour relatively stable conformations.^46,47 Thus, MM/PBSA results are more appropriately used as a semi-quantitative reference for comparing interfacial energy contributions and relative binding trends rather than as an exact reproduction of experimental binding free energies. Taken together, the MST, MM/PBSA, and MD stability analyses suggest that Peptide-4 showed a lower K_d value, more favourable calculated binding energy, and lower conformational fluctuations under the present experimental and simulation conditions than R4K1. In contrast, although R4K1 exhibited stronger electrostatic interactions and more extensive polar contacts, these were accompanied by more pronounced polar solvation compensation and conformational fluctuations, suggesting a more dynamic binding mode. Overall, under the present experimental and simulation conditions, Peptide-4 displayed a more coordinated and relatively favourable binding profile compared with R4K1.

Free-energy landscape analysis of binding conformations

To further compare the energetic distribution of Peptide-4 and R4K1 binding conformations, Gibbs free-energy landscapes were constructed using RMSD and Rg as reaction coordinates. The Peptide-4-bound system displayed a relatively concentrated major low-energy basin, mainly distributed at RMSD values of approximately 0.18–0.23 nm and Rg values of approximately 1.85–1.88 nm, suggesting that Peptide-4 sampled a relatively more compact and less dispersed binding-related conformational ensemble during the simulation (Figure 6). In contrast, the R4K1-bound system showed a broader low-energy region, primarily located at RMSD values of approximately 0.25–0.36 nm and Rg values of approximately 1.84–1.88 nm, indicating greater conformational heterogeneity and dynamic fluctuation (Figure S6). The corresponding three-dimensional free-energy landscapes further supported this trend, with Peptide-4 showing a more defined major energy basin, whereas R4K1 exhibited a broader basin distribution. Together with the RMSD, RMSF, and MM/PBSA analyses, these results suggest that Peptide-4 may maintain a more stable and more focused binding-conformation sampling profile than R4K1 under the present simulation conditions.

Figure 6.

Gibbs free-energy landscape of Peptide-4 bound to the ERα ligand-binding pocket. (A) Two-dimensional free-energy landscape based on Rg and RMSD revealed a dominant low-energy basin centred at approximately RMSD (0.18–0.23 nm) and Rg (1.85–1.88 nm), suggesting a preferred compact binding-related conformation of Peptide-4. (B) The corresponding three-dimensional landscape was consistent with a more defined major energy basin, supporting the relatively stable conformational sampling of the ERα–Peptide-4 complex during the simulations.

Serum stability of peptide-4

To preliminarily assess the serum proteolytic stability of Peptide-4, the percentage of intact peptide remaining was monitored over 360 min. Peptide-4 showed a gradual decrease over time, with approximately 81% of the peptide remaining after 360 min, indicating that it retained a substantial fraction of its intact form under the assay conditions (Figure 7). These findings provide a basis for its subsequent cell-based evaluation.

Figure 7.

Serum stability profile of Peptide-4 over 360 min. Peptide-4 showed a gradual decrease in serum over time, with approximately 81% of the peptide remaining after 360 min, suggesting moderate serum stability under the assay conditions. Data are presented as mean ± SD (n = 3).

In vitro antiproliferative activity and selectivity of peptide-4

To evaluate the in vitro antiproliferative activity of the screened peptides, Peptides 1–4 and control peptides were tested in SK-UT-1 and HUtSMC cells using the MTT assay. Peptides 1–4 inhibited SK-UT-1 cell proliferation to varying degrees, with IC₅₀ values of 8.24 ± 0.56, 6.09 ± 0.37, 2.21 ± 0.18, and 0.77 ± 0.04 μM, respectively, indicating a gradually increased activity profile (Table 6). Among them, Peptide-4 showed the lowest IC₅₀ value and the lowest IC₅₀ value among the tested peptides against SK-UT-1 cells. In contrast, R4K1, scrambled peptide, and unrelated peptide all showed IC₅₀ values >10 μM in SK-UT-1 cells, suggesting that the antiproliferative effect of Peptide-4 may be associated with its specific sequence and binding characteristics. Notably, Peptides 1–4 and all control peptides showed IC₅₀ values >10 μM in HUtSMC cells, indicating limited effects on HUtSMC cell proliferation under the present experimental conditions. Overall, Peptide-4 exhibited the comparatively favourable antiproliferative profile in SK-UT-1 cells while showing no obvious inhibitory effect on HUtSMC cells, supporting its selection for further evaluation.

Table 6.

In vitro antiproliferative activity of Peptides 1–4 and control peptides in SK-UT-1 and HUtSMC cells after 48 h of treatment.

Name	IC50 (μM) Mean ± SD
	SK-UT-1	HUtSMC
Peptide-1	8.24 ± 0.56	> 10
Peptide-2	6.09 ± 0.37	> 10
Peptide-3	2.21 ± 0.18	> 10
Peptide-4	0.77 ± 0.04	> 10
R4K1	> 10	> 10
Scrambled peptide	> 10	> 10
Unrelated peptide	> 10	> 10

Note. Values are presented as mean ± SD from six independent measurements. IC₅₀ values were determined by the MTT assay after 48 h of treatment. Values reported as >10 μM indicate that the IC₅₀ was not reached at the highest tested concentration.

To further evaluate the antiproliferative activity of Peptide-4 over a longer treatment period and its effects on normal human cells, a 5-day continuous MTT assay was performed in SK-UT-1 cells and multiple normal human cell lines. Peptide-4 maintained antiproliferative activity against SK-UT-1 cells, with an IC₅₀ value of 0.35 ± 0.02 μM. In contrast, its IC₅₀ values in HUtSMC, HUVEC, LO2, HDF, and HK-2 cells were all >10 μM, suggesting limited effects on the proliferation of the tested normal cells under the present experimental conditions. Scrambled peptide and unrelated peptide also showed IC₅₀ values >10 μM in all tested cell lines, indicating that the activity of Peptide-4 may be associated with its specific sequence features (Table 7). Overall, Peptide-4 retained evident antiproliferative activity against SK-UT-1 cells after 5 days of continuous treatment, while showing no obvious inhibitory effect on the tested normal human cells, suggesting a degree of in vitro selectivity; however, its in vivo safety remains to be further evaluated.

Table 7.

Five-day in vitro antiproliferative assessment of Peptide-4 and control peptides in SK-UT-1 and tested normal human cell lines.

Name	IC50 (μM) Mean ± SD
	SK-UT-1	HUtSMC	HUVEC	LO2	HDF	HK-2
Peptide-4	0.35 ± 0.02	> 10	> 10	> 10	> 10	> 10
Scrambled peptide	> 10	> 10	> 10	> 10	> 10	> 10
Unrelated peptide	> 10	> 10	> 10	> 10	> 10	> 10

Note. Peptide-4 exhibited a low IC₅₀ value in SK-UT-1 cells, whereas its IC₅₀ values in HUtSMC, HUVEC, LO2, HDF, and HK-2 cells were >10 μM. Scrambled peptide and unrelated peptide did not reach IC₅₀ values in any tested cell line. Data are presented as mean ± SD from MTT assays. Values reported as >10 μM indicate that the IC₅₀ was not reached at the highest tested concentration.

ERα-associated antiproliferative activity of peptide-4 in SK-UT-1 cells

To further assess whether the antiproliferative activity of Peptide-4 is associated with ERα expression, we evaluated the in vitro antiproliferative activities of Peptide-4 and the scrambled control peptide in shControl and ERα-knockdown (shERα) SK-UT-1 cells using the MTT assay. As shown in Table 8, the IC₅₀ value of Peptide-4 in shControl cells was 0.77 ± 0.04 μM, consistent with our previous results. In ERα-knockdown cells, however, the IC₅₀ value increased to >10 μM, indicating a marked attenuation of the antiproliferative activity of Peptide-4. In contrast, the scrambled control peptide showed no obvious inhibitory effect in either cell model, with IC₅₀ values >10 μM in both shControl and shERα cells. These results suggest that the antiproliferative activity of Peptide-4 in SK-UT-1 cells is at least partly associated with ERα expression, further supporting an association with an ERα-related mechanism.

Table 8.

ERα knockdown attenuates the antiproliferative activity of Peptide-4 in SK-UT-1 cells.

Name	IC50 (μM)a
	shControl	shERα
Peptide-4	0.77 ± 0.04	> 10
Scrambled peptide	> 10	> 10

^aIC₅₀ values were determined by the MTT assay after 48 h of treatment and are presented as mean ± SD from six independent measurements. Values reported as >10 μM indicate that the IC₅₀ was not reached at the highest tested concentration. shControl and shERα denote control and ERα-knockdown SK-UT-1 cells, respectively.

Effects of peptide-4 and anti-miR-149G on c-Myc and Cyclin D1 mRNA expression

To further explore molecular changes associated with the antiproliferative effects of Peptide-4, the mRNA expression levels of c-Myc and Cyclin D1 were measured by RT-qPCR in SK-UT-1 cells. Peptide-4 treatment reduced the relative mRNA levels of both c-Myc and Cyclin D1 in a concentration-associated manner, with an approximately 50% decrease at 1 μM and a more pronounced reduction at 5 μM. A similar expression pattern was observed after anti-miR-149G treatment, which also decreased c-Myc and Cyclin D1 mRNA levels at 1 and 5 μM (Figure 8). Because c-Myc and Cyclin D1 are proliferation-associated genes, these results suggest that the antiproliferative activity of Peptide-4 may be associated, at least in part, with the downregulation of proliferation-related gene expression in SK-UT-1 cells.

Figure 8.

Effects of Peptide-4 and anti-miR-149G on c-Myc and Cyclin D1 mRNA expression in SK-UT-1 cells. (A) Peptide-4 treatment decreased the relative mRNA levels of c-Myc and Cyclin D1 at 1 and 5 μM, with a greater reduction observed at 5 μM. (B) Anti-miR-149G treatment produced a similar concentration-associated decrease in c-Myc and Cyclin D1 mRNA expression. mRNA levels were determined by RT-qPCR and normalised to the control group. Data are presented as mean ± SD (n = 3). ***P < 0.001 vs control. Corresponding Hedges’ g standardised effect sizes and their 95% confidence intervals are summarised in Table S3.

Antiproliferative effect of combined peptide-4 and anti-miR-149G treatment

Given the favourable in vitro antiproliferative profile of Peptide-4, anti-miR-149G was included to evaluate whether combined treatment could further affect SK-UT-1 cell proliferation. As shown in Figure 9, treatment with Peptide-4 or anti-miR-149G alone at 1 μM resulted in cell inhibition rates of approximately 55.8% and 46.9%, respectively. Notably, combined treatment with Peptide-4 and anti-miR-149G at the same concentration increased the inhibition rate to 90.4%, which was significantly higher than that observed with either single treatment. These results suggest that Peptide-4 may show increased antiproliferative effects when combined with anti-miR-149G when combined with anti-miR-149G in SK-UT-1 cells.

Figure 9.

Effect of Peptide-4, anti-miR-149G, and their combination on SK-UT-1 cell inhibition. Peptide-4 and anti-miR-149G alone showed moderate inhibitory effects on SK-UT-1 cells, while the combined treatment resulted in a higher inhibition rate than either single treatment. The dashed line indicates the 60% inhibition reference level. Cell inhibition rates were determined by the MTT assay. Data are presented as mean ± SD (n = 3). ***P < 0.001 vs either single-treatment group. Corresponding Hedges’ g standardised effect sizes and their 95% confidence intervals are summarised in Table S4.

Comparison with reported ERα-targeting compounds

Several peptides and small-molecule inhibitors targeting the ERα–coactivator interaction interface have been reported, providing useful benchmarks for contextualising the binding and functional activity of Peptide-4. For example, R4K1, an ERα-targeting stapled peptide, has been reported to exhibit nanomolar binding affinity towards ERα.¹⁹ In the present assay system, R4K1 bound recombinant human ERα with a K_d value of 24.1 nM. TD-stabilised peptide 6 has also been reported to target ERα, with a K_d value of 67 nM.⁴⁸ In addition, the small-molecule ER-AF2 inhibitor VPC-260724 showed ERα binding in the MST assay, with a K_d value of 67.5 μM.⁴⁹ In the present study, Peptide-4 exhibited low-nanomolar apparent binding affinity towards recombinant human ERα (K_d = 9.3 nM) and showed antiproliferative activity in SK-UT-1 uterine leiomyoma cells. The short sequence of Peptide-4, together with its ERα-binding profile and cell-level functional activity, suggests that it may provide a distinct structural reference for future studies on ERα-targeting peptides in uterine leiomyoma.

Limitations and future perspectives

This study has several limitations. First, although computer-aided protein modelling, molecular docking, MD simulations, and MM/PBSA analyses are widely used for hit identification, binding-mode prediction, and interaction-mechanism analysis, these approaches mainly provide theoretical or semi-quantitative information and require experimental validation to further support their biological relevance.^50–53 In the present study, Peptide-4 was preliminarily evaluated using MST, cell-based assays, and ERα knockdown experiments; however, the current evidence is still mainly derived from in vitro binding assays and cell models, and its activity has not yet been assessed in animal models or ex vivo uterine leiomyoma tissues. Second, although Peptide-4 retained a substantial fraction of its intact form in the serum stability assay, this result only represents a preliminary in vitro stability assessment and does not replace pharmacokinetic or delivery-strategy studies. Third, although ERα knockdown markedly attenuated the antiproliferative activity of Peptide-4 and MST assays showed no reliable binding response towards ERβ or GPER, the current data cannot completely exclude the indirect involvement of ERβ, GPER, or non-genomic oestrogen signalling. Finally, the current safety assessment was mainly based on in vitro proliferation assays in selected normal human cell lines and therefore does not fully address potential off-target effects, toxicity, or long-term safety.

Future studies will evaluate the in vivo activity, pharmacokinetic properties, delivery approach, and long-term safety of Peptide-4 in animal models, and will further investigate ERβ/GPER knockdown or pharmacological inhibition and non-genomic oestrogen signalling to further clarify its mechanism and target relevance. In addition, ERα/JunB interaction assays and binding analyses at the promoter or enhancer regions of c-Myc and Cyclin D1 will be performed to further assess the possible contribution of miRNA/JunB-related regulation to the combined effects of Peptide-4 and Anti-miR-149G.

Conclusion

In conclusion, this study identified Peptide-4 as an ERα-binding peptide through a structure-guided virtual screening strategy. Peptide-4 showed apparent binding to ERα under the present MST assay conditions, whereas no reliable binding response was detected for ERβ or GPER under the present MST assay conditions. Computational analyses suggested that Peptide-4 may adopt a relatively stable binding conformation within the ERα binding pocket, supported by hydrogen-bonding, hydrophobic, and van der Waals interactions. In cell-based assays, Peptide-4 inhibited SK-UT-1 cell proliferation, while no obvious inhibitory activity was observed in the tested normal human cell models under the same conditions, suggesting a degree of in vitro cellular selectivity. ERα knockdown markedly attenuated the antiproliferative activity of Peptide-4, supporting an association between its cellular effect and ERα expression. Peptide-4 also reduced the mRNA levels of c-Myc and Cyclin D1, and combined treatment with anti-miR-149G further increased the antiproliferative effect in SK-UT-1 cells. Overall, these findings provide an experimental basis for further investigation of Peptide-4 in ERα-related uterine leiomyoma models.

Funding Statement

This work was supported by the Taizhou Clinical Medical College, Nanjing Medical University 2024 Clinical Research Project (No. TZKY20240206), Scientific Research Funding Project of Taizhou City “311 Project” (2024, 6th Batch).

Disclosure statement

The authors report no conflicts of interest.

Data availability statement

The data presented in the current study are available from the corresponding author upon reasonable request.