Identification of a peptide inhibitor disrupting the PCSK9-LDLR interaction via pharmacophore-based virtual screening, molecular dynamics simulations and in vitro/vivo evaluation

Abstract

The PCSK9-LDLR interaction, driving elevated LDL-C, is a key driver of ASCVD pathogenesis. Identifying peptides disrupting this interaction offers an alternative ASCVD therapy. Herein, via structure-based virtual screening with Pep2-8 as a control, we identified TPP-4, a high-affinity peptide inhibitor targeting PCSK9. Compared to Pep2-8, TPP-4 showed lower binding free energy (approximately −9.8 kcal/mol) and K_d values (K_d = 0.08 ± 0.01 μM), interacting with PCSK9’s LDLR-binding domain through multiple interactions. CD spectroscopy also provided indirect evidence for these key interactions. Additionally, it stably bound to the LDLR binding domain of PCSK9 during 100 ns MD simulations. It showed good serum stability, negligible HepG2 cytotoxicity, and restored surface LDLR (EC₅₀ = 1.12 ± 0.05 μM). In mice, TPP-4 upregulated hepatic LDLR and reduced plasma total cholesterol levels. In conclusion, these data demonstrate that TPP-4 could be a high-affinity and potent candidate peptide for ASCVD treatment.

Graphical Abstract

Introduction

The global morbidity and mortality of atherosclerotic cardiovascular diseases (ASCVD) continue to rise, making it a significant public health issue¹. In addition, the World Health Organisation reported that approximately 17.9 million people have died each year from ASCVD worldwide². Elevated blood lipid levels are recognised as a key risk factor for ASCVD³. Blood lipids mainly include total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C)⁴. Increased LDL-C represents the most common and clinically significant contributor to ASCVD⁵. Accordingly, lowering LDL-C levels reduces ASCVD events. The 2018 American College of Cardiology/American Heart Association (ACC/AHA) and 2019 European Society of Cardiology/European Atherosclerosis Society (ESC/EAS) recommended LDL-C < 70 mg/dL and < 55 mg/dL for ASCVD patients, respectively⁶.

To date, numerous PCSK9-targeting inhibitors have been developed as therapeutic strategies for hypercholesterolaemia and related cardiovascular diseases. Statins represent the first-line therapy for lowering LDL-C levels and managing ASCVD⁷. However, some patients fail to reach target levels despite maximum doses⁷. Statins reduced cholesterol reduction to exert efficacy, but this reduction upregulated PCSK9 transcription, promoting drug resistance⁸. Ezetimibe, a cholesterol absorption inhibitor, enhanced LDL-C lowering and ASCVD outcomes with statins⁹. PCSK9 inhibitors mainly include monoclonal antibodies, siRNA, small molecule inhibitors, and peptide inhibitors¹⁰. Human monoclonal antibodies (evolocumab, alirocumab) were approved by the Food and Drug Administration (FDA) in 2015 for ASCVD treatment¹¹. Study results demonstrated that most ASCVD patients receiving evolocumab achieved LDL-C levels below 40 mg/dL¹². Alirocumab reduced baseline LDL-C levels by 48.6% in treated patients¹³. Inclisiran, a marketed siRNA drug, reduced baseline LDL-C by approximately 50% with a 300 mg dose administered every 6 months¹⁴. PCSK9-targeting monoclonal antibodies and siRNA drugs are limited by high costs and subcutaneous administration¹⁵. AZD0780, an oral PCSK9 small-molecule inhibitor, is in phase-III clinical trials (NCT07000357) for ASCVD patients, with 30 mg eliciting a 30% decrease in LDL-C levels¹⁶. MK0616 is also undergoing phase-III clinical trials (CTR20242333), with oral administration of 30 mg leading to a 60.9% decrease in LDL-C levels¹⁵. However, the binding interface of PCSK9 with low-density lipoprotein receptor (LDLR) is relatively flat, and PCSK9 lacks well-defined small-molecule binding pockets. Peptide drugs exhibit high cell permeability, strong specificity, and structural mimicry of proteins¹⁷. Thus, identifying peptide inhibitors that disrupt the PCSK9-LDLR interaction represents an alternative therapy for ASCVD.

PCSK9 is primarily synthesised and secreted by the liver¹⁸. It is also a member of the serine protease family¹⁹. PCSK9 binds to the hepatocytes LDLR to form a complex⁷. The complex is endocytosed to lysosomes for degradation, altering LDLR conformation and inhibiting its recycling to cell surface²⁰. The reduction in LDLR number leads to elevated LDL-C levels, increasing the risk of ASCVD²¹. Studies have shown that Heparan Sulphate Proteoglycans (HSPG), Cyclase Associated Protein-1 (CAP-1), and Glucose-Regulated Protein 94 (GRP94) play auxiliary roles in the PCSK9-mediated LDLR degradation pathway²².

Peptide inhibitors of PCSK9 disrupt the interaction between PCSK9 and the EGF-A domain of LDLR, thereby lowering LDL-C levels and conferring therapeutic benefits in patients with ASCVD²³. Pep2-8 is a 13-amino acid linear peptide that mimics the EGF-A domain of LDLR²⁴. It exhibited a certain binding affinity for PCSK9 (dissociation constant (K_d) = 0.76 ± 0.32 μM)²⁴. At a concentration of 50 μM, it restored cell surface LDLR levels by 96.0 ± 9.1% and extracellular LDL-C uptake by 83.3 ± 3.8% in HepG2 cells²⁵. It served as a positive control in this study. However, Pep2-8 exhibited weak affinity for PCSK9, requiring high peptide concentrations to exert therapeutic effects. Thus, there is a need to identify high-affinity peptides that disrupt the PCSK9-LDLR interaction as an alternative.

Structure-based pharmacophore modelling methods enable more accurate and rapid identification of lead compounds from databases, significantly shortening the drug discovery process²⁶. Molecular docking screening takes into account the flexibility of peptide structures²⁷. In previous studies, we successfully identified numerous potential disease-treating peptides through pharmacophore modelling, molecular docking, and efficacy evaluation^28,29. In this study, we successfully discovered a peptide inhibitor (TPP-4) that disrupts the PCSK9-LDLR interaction and exhibits therapeutic potential in ASCVD.

Materials and methods

Materials

All the peptides were purchased from GL Biochem (Shanghai, China). The high-performance liquid chromatography (HPLC) spectra and mass spectrometry (MS) spectra for TPPs 1–5, Pep2-8 and Mutant-TPP-4 are provided in the (Supplementary Materials Figures S1.1–S1.14). The recombinant human PCSK9 protein was purchased from Abcam (Cambridge, MA, USA; catalogue number: ab198471). The Monolith^TM RED-NHS second-generation protein labelling kit was obtained from Nano Temper Technologies GmbH (Munich, Germany). The PBST (Phosphate-Buffered Saline with 0.05% Tween 20) was prepared. Phosphate buffered saline (PBS), Dulbecco’s Modified Eagle’s Medium (high glucose), foetal bovine serum (FBS) and penicillin/streptomycin were obtained from Gibco (Grand Island, NY, USA).

Cell line and culture conditions

The HepG2 cell line (catalogue number: HB-8065) was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). It was cultured in DMEM high glucose with stable L-glutamine supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin. Then, it was placed in a 37 °C incubator with 5% CO₂ in humidified air. For cellular experiments in this study, HepG2 in the log phase was prepared as a single-cell suspension.

Pharmacophore construction

The 1.85 Å-resolution structure of PCSK9 (delta CRD) complex with Pep2-8 (PDB ID: 4NMX) was downloaded from the Protein Data Bank (PDB, https://www.rcsb.org/structure/4NMX).³⁰ The PCSK9-Pep2-8 complex was initially imported into the Molecular Operating Environment (MOE; Chemical Computing Group Inc., Montreal, Canada). Furthermore, the QuickPrep tool was employed for structural preprocessing, including removal of unbound water, addition of polar hydrogen atoms, calculation of partial charges, and energy minimisation (Amber10:EHT force field). Then, the SiteView tool was employed to visualise the binding interface between PCSK9 and Pep2-8. The Ligand Interaction tool was utilised to analyse the interactions at the binding interface. Hydrogen bonding and hydrophobic interactions between PCSK9 and Pep2-8 were the primary considerations. Finally, pharmacophore features were constructed using the pharmacophore query editor based on the interaction patterns, including hydrogen bond donors, hydrogen bond acceptors, and aromatic centres. These pharmacophore features were combined to construct the PCSK9 pharmacophore model for preliminary screening.

Pharmacophore model evaluation

The in-house database comprising 510 peptides was constructed, of which 12 active PCSK9-targeting peptides were all retrieved from published literature^30–32. This dataset was used to evaluate the ability of the constructed PCSK9 pharmacophore model to identify active peptides from the database. The Pharmacophore Search tool was employed for database screening, and the Gunner-Henry (GH) score method was used to quantify the model’s capacity to distinguish active from inactive PCSK9-targeting peptides. The GH score ranges from 0 to 1, representing a null model and an ideal model, respectively.

Construction of a peptide library

The construction method of the peptide database was adopted from previous studies²⁸. The QuaSAR-CombiGen tool was used to generate the combinatorial peptide database from a series of short peptide fragments. Therefore, two tripeptide databases and one tetrapeptide database were enumerated using this tool. The peptide database containing 59,319 peptides was constructed by combining the above three peptide fragments for the virtual screening of PCSK9 inhibitors. Each peptide consisted of ten amino acids. Finally, using the energy minimisation algorithm in MOE, the two-dimensional (2D) structures of these peptides were converted to three-dimensional (3D) structures. The database containing 59,319 peptides was then used for subsequent virtual screening.

Virtual screening

The 59,319 peptides were preprocessed using the QuickPrep tool within MOE. Then, these peptides were subjected to preliminary screening based on the validated pharmacophore model of PCSK9. Root Mean Square Deviation (RMSD) is defined as the square root of the mean squared distance between the query features of the model and the annotated domains of the matched ligand. In the MOE software, a lower RMSD value corresponds to a higher degree of spatial mapping concordance between ligand-annotated domains and query features, and enhanced precision of mapping localisation typically indicates stronger binding affinity of the ligand to the target protein. In this study, only hit compounds with the RMSD value of < 1 Å were selected for subsequent molecular docking assays.

Molecular docking

The selected peptides were computationally docked into the LDLR-binding domain of the PCSK9 to facilitate further screening. Pep2-8 served as the positive control compound, with its docking score used as a reference to screen for candidate peptide inhibitors of PCSK9. For molecular docking between the ligand and PCSK9 protein, a rigid receptor-flexible ligand approach was employed. The grid density of the docking region was set to 0.375 Å. For molecular docking, initial ligand conformations were generated using the Triangle Matcher algorithm and refined via the Rigid Receptor protocol. Conformations were scored in two stages: London dG for initial screening, and GBVI/WSA dG for subsequent high-precision refined scoring. The top 5 scoring conformations were retained for downstream analysis. Given their superior docking scores, the top five peptides were designated TPPs 1–5.

Interaction analysis

The binding modes of the TPPs 1–5 (with optimal conformations) in complex with PCSK9 were exported using MOE. 2D interaction diagrams were generated via the Ligand Interactions tool, while 3D predicted binding mode diagrams were also obtained in MOE. Interaction analysis focused primarily on hydrogen bonds and hydrophobic interactions, where stronger hydrogen bonds and hydrophobic interactions correspond to greater stability of the complex. Additionally, representative conformational snapshots of the binding modes of TPPs 1–5 with PCSK9 complexes were plotted using the LigPlot tool.

MST assay

MST experiments were performed as previously reported³³. Using MST, the Monolith^TM NT.115 instrument (Nano Temper Technologies GmbH) was employed to determine the binding affinities between TPPs 1–5, Mutant-TPP-4, Pep2-8 and PCSK9. The human recombinant PCSK9 was dissolved in PBST at a concentration of 10 μM. After vigorously mixing 7 µL of RED-NHS second-generation dye with 7 µL of NHS labelling buffer, a dye solution with a final concentration of 300 nM was prepared. 90 µL of the 10 μM PCSK9 solution was combined with 10 µL of the dye solution, vortexed briefly, and incubated at 25 °C for 30 min in the dark to minimise dye degradation. The final concentration of RED-NHS-labelled PCSK9 was 20 nM. The ligand was serially diluted 16 times in 2-fold steps using PBST, and equal volumes of each dilution were mixed with the labelled PCSK9. The mixtures were then incubated in the dark for 5 min to allow binding equilibrium. The samples were loaded into the capillary for measurement. The results were analysed using the MO. Affinity Analysis software (Nano Temper Technologies GmbH) to obtain the K_d values, with each peptide tested in triplicate.

CD Spectroscopy

The CD spectra of TPP-4 and PCSK9 both pre- and post-incubation were acquired in accordance with the procedure outlined in the literature^24,34. The samples were dissolved in 0.1 M NaF solution to yield a final concentration of 50 μM. Using a CD cuvette with a 1 mm path length, spectra were acquired at room temperature over the wavelength range of 190–240 nm. Each spectrum was obtained as the average of three scans to improve the signal-to-noise ratio. Baseline correction was achieved by subtracting the background from the sample spectrum.

MD simulation

Molecular dynamics (MD) simulations were performed following protocols adapted from a prior publication^35,36. The 100 ns MD simulations were performed to investigate the dynamic stability of the binding modes in PCSK9-TPP-1 and PCSK9-TPP-4 complexes. The computationally predicted binding modes of PCSK9-TPP-1 and PCSK9-TPP-4 served as the initial structures. The MD simulations were performed using Groningen Machine for Chemical Simulations (GROMACS) software (version: 2021.5). The complexes were parameterised using the AMBER99SB-ILDN force field. The PCSK9–ligand complex was centred in a cubic solvation box with a minimum distance of 10 Å between the complex surface and the box boundaries. The TIP3P model was utilised as the simplified water model for the simulation system. Counterions (Na⁺ and Cl^-) were added at a concentration of 0.15 M to neutralise the net charge of the PCSK9–ligand complex system. Van der Waals and short-range electrostatic interactions were computed with a cut-off distance of 8 Å, whereas long-range electrostatic interactions were handled via the Particle Mesh Ewald (PME) method. After restraining the lengths of hydrogen-containing bonds using the SHAKE algorithm, MD simulations were executed with an integration time step of 2 fs³⁷. The system was energy-minimized via the steepest descent algorithm for 5000 steps. Subsequently, NVT (temperature at 300 K) and NPT (pressure at 1.0 bar) simulations were performed for pre-equilibration. RMSD, RMSF (root mean square fluctuation), protein secondary structure composition, Rg (radius of gyration), and number of hydrogen bonds were calculated to assess the system’s flexibility and stability respectively. The trajectory analysis was performed at an interval of 10 ps.

MM-PBSA calculations

The Molecular Mechanics-Poisson Boltzmann Surface Area (MM-PBSA) method was employed to calculate the binding free energy of the complex system, thereby evaluating its binding stability^38,39. Additionally, the residue decomposition of binding free energy at the PCSK9–ligand interaction interface was computed; residues with the largest contributions to the binding free energy were identified as key amino acids driving the PCSK9–ligand interactions.

Serum stability test

Serum stability experiments for TPP-4 were conducted following the protocols described in the literature²⁸. Human serum was diluted at a ratio of 25% to ensure the accuracy and reliability of the experimental results. The diluted serum was centrifuged at 15,000 rpm for 10 min, and the supernatant was collected for subsequent assays. The TPP-4 stock solution was diluted into the supernatant and incubated at 37 °C. At 0, 30, 60, 120, and 240 min, 200 μL samples were taken and analysed by HPLC. The peak area was integrated to calculate the concentration of TPP-4. The determination of TPP-4 concentration was performed in triplicate.

MTT assay

The MTT assay for TPP-4 was performed following the protocol described in the literature⁴⁰. HepG2 cells were seeded into a 96-well plate at a density of 4,000 cells per well and cultured overnight to permit adherence. The culture medium was aspirated, and the cells were washed with 1 × PBS. 100 μL of serum-free medium containing varying drug concentrations was added to each well, followed by incubation for 48 h. 20 μL of MTT solution (5 mg/mL) was diluted to a final concentration of 1 mg/mL using serum-free medium. 100 μL of the sample solution was added to each well of a 96-well plate and incubated for 3 h. The medium was then aspirated, and 100 μL of DMSO was added to dissolve the formazan crystals. A microplate reader was used to measure the absorbance, which was subsequently used to determine the number of viable cells. The cell survival rate was expressed as a percentage.

In vitro assessment of cell surface LDLR expression

The method for measuring cell surface LDLR in HepG2 cells was performed following literature protocols⁴¹. HepG2 cells were seeded at a density of 1 × 10⁵ cells per well in 48-well plates and incubated overnight under standard culture conditions to allow cell attachment. Subsequently, the medium was replaced with DMEM containing 10% lipoprotein-deficient serum (LPDS) and cultured for 24 h. 15 μg/mL of PCSK9 was mixed with TPPs 1–5, Pep2-8 and evolocumab, respectively, then added to cells and incubated for 4 h. The expression of LDLR on the cell surface was quantified using flow cytometry. Subsequently, HepG2 cells were treated with varying concentrations of TPP-4. Cell surface LDLR was quantified using the aforementioned protocol.

In vivo measurements of hepatic LDLR and plasma cholesterol

Male C57BL/6J mice (10–12 weeks old, 20–25 g) were purchased from Changzhou Cavens Experimental Animal Limited Company (Changzhou, China). The four mice were randomly allocated to each group. A triple-blind design (treatment blinding, outcome assessment blinding, and data analysis blinding) was implemented throughout the in vivo experiments in this study. The in vivo animal evaluation protocol was meticulously and rationally designed,⁴² and all procedures adhered to the ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments) to ensure rigorous and transparent reporting of animal research. Vehicle control group: Solution prepared with 20% propylene glycol, 25% of 20% solutol in water, and 55% PBS. TPP-4 and Evolocumab were administered subcutaneously at 20 mg/kg, twice daily for 3 consecutive days. Mice were maintained under a 12-h light/12-h dark cycle with ad libitum access to food and water. All efforts were made to minimise animal suffering: mice were housed in standard cages to reduce stress, and routine health checks were performed to monitor for signs of discomfort. No analgesics were required for the subcutaneous injections or tail snip procedures, as these are minimally invasive and transient in discomfort; however, if signs of long-term pain are observed, appropriate analgesia should be administered by subcutaneous injection of 0.1 mg/kg buprenorphine. The animal experiments were approved by the Ethics Committee of China Pharmaceutical University.

On day 4, mice were euthanized via CO₂ inhalation using a gradual displacement method (30% chamber volume per minute) with a final CO₂ concentration of 70–80% to ensure rapid, humane euthanasia. Liver tissue samples (25–35 mg) were placed into 2-ml Eppendorf tubes, with subsequent addition of steel beads and liver lysis buffer for homogenisation. The liver homogenates were centrifuged at 16,260 × g and 4 °C for 10 min. The supernatants were used to determine hepatic LDLR protein levels using a mouse LDLR ELISA kit (R&D Systems). Total protein concentration was measured with BCA protein assay kit (Beyotime Biotechnology, Shanghai, China), and hepatic LDLR content was expressed as nanograms per milligram of total protein (ng/mg protein).

Prior to each administration, approximately 80 μL of blood was collected via tail snip into heparinised Eppendorf tubes. Samples were centrifuged at 16,260 × g, 4 °C for 10 min, and supernatants were stored at −80 °C. The WAKO LabAssay^™ Cholesterol quantitative kit was purchased from Shanghai Jinpan Biotech Co., Ltd (Shanghai, China). A 2 μL plasma sample and 300 μL chromogen were added to a 96-well plate. The absorbance of the pale blue pigment at 600 nm was measured to calculate the total cholesterol level (mg/mL) in mouse plasma.

Statistics analysis

Data processing and graphing were performed using GraphPad Prism software (GraphPad Software Inc, San Diego, CA, USA; version 9.5). One-way analysis of variance (ANOVA) was used to assess intergroup mean differences. P values < 0.05 were considered statistically significant. Each experiment was repeated three times. Data are presented as mean ± standard deviation (SD).

Results

Establishment of the PCSK9-based pharmacophore model

The high-resolution X-ray crystal structure of PCSK9 in complex with Pep2-8 was used to derive the computational pharmacophore model for PCSK9. Using the Ligand Interaction tool in MOE, we identified Thr377, Cys378 and Phe379 as key residues involved in hydrogen bonding within the binding pocket. The benzene ring of PCSK9 residue Phe379 forms a π-H interaction with the benzene ring of tyrosine in Pep2-8. Therefore, based on PCSK9-Pep2-8 interactions, four pharmacophore features (F1-F4) were constructed using the MOE pharmacophore editor (Figure 1). Given the π-H interaction between PCSK9 residue Phe379 and Pep2-8, feature F1 (aromatic group, Aro) was constructed. Considering the hydrogen bonding interactions between PCSK9 and Pep2-8, pharmacophore features F2-F4 were constructed. F2 (hydrogen bond acceptor, Acc) corresponds to PCSK9 residues Phe379 and Cys378; F3 (hydrogen bond donor, Don) corresponds to PCSK9 residue Phe379; F4 (Acc) corresponds to PCSK9 residue Thr377. The pharmacophore features encode the critical interaction information of inhibitors bound to PCSK9.

Figure 1.

The pharmacophore model of the LDLR-binding domain in PCSK9 was established using MOE. Pharmacophore features include: F1, an aromatic feature; F2 and F4, hydrogen-bond acceptor features; F3, a hydrogen-bond donor feature.

Pharmacophore model evaluation

The compound database comprising 510 molecules was established to validate the constructed pharmacophore model. Validation data for the PCSK9 pharmacophore model are presented in Table 1. For the PCSK9 pharmacophore model, 10 total hits were identified, nine of which were active compounds. The calculated active yield and active coverage were 90% and 75%, respectively. Additionally, the model exhibited a high enrichment factor of 38, indicative of strong active compound enrichment capacity. Notably, the GH score of the PCSK9 pharmacophore model reached 0.86. These results confirm that the constructed pharmacophore model could effectively distinguish active from inactive compounds in the database, and be thus suitable for subsequent virtual screening of PCSK9 inhibitors.

Table 1.

Pharmacophore model validation using goodness-of-hit score (GH) method.

Serial No.	Parameter	Pharmacophore Model
1	Total molecules in database (D)	510
2	Total number of actives in database (A)	12
3	Total hits (Ht)	10
4	actives hits (Ha)	9
5	% Yield of actives ((Ha/Ht) × 100)	90%
6	% Ratio of actives ((Ha/A) × 100)	75%
7	Enrichment factor (E) ((Ha × D)/(Ht × A))	38
8	False negatives (A − Ha)	3
9	False positives (Ht-Ha)	1
10	Goodness of hit score (GH)a	0.86

^a : GH = (Ha(3A + Ht)/4HtA) × (1 – (Ht – Ha)/(D – A));

GH score of 0.8 ∼ 1 indicates an excellent model.

Table 2.

The binding affinities of Pep2-8, Mutant-TPP-4 and TPPs 1–5 towards PCSK9.

Name	Sequencea	PCSK9 (Kd [µM])
TPP-1	rTRFTSWEEY-NH2	0.17 ± 0.02
TPP-2	kTRFTSWEEY-NH2	0.63 ± 0.05
TPP-3	rTKFTSWEEY-NH2	0.49 ± 0.04
TPP-4	rTRFTSREEF-NH2	0.08 ± 0.01
TPP-5	kTRFTSWEEF-NH2	0.56 ± 0.03
Mutant-TPP-4	aTAFTSAEEF-NH2	>100
Pep2-8	Ac-TVFTSWEEYLDWV-NH2	0.74 ± 0.06

^a Both TPPs 1–5 and Mutant-TPP-4 are subjected to C-terminal amidation. Pep2-8 undergoes N-terminal acetylation and C-terminal amidation. Lowercase letters indicate that the amino acids are in the D-configuration, while uppercase letters indicate that they are in the L-configuration.

Virtual screening

Figure 2 illustrates the multi-step screening workflow for identifying promising PCSK9 peptide inhibitors. The screening library comprises 59,319 peptides. Using the validated pharmacophore model above, 62 peptides were screened from the peptide library. The 62 peptides were docked into the LDLR-binding domain of PCSK9 for further screening. The docking scores of TPPs 1–5 and Pep2-8 are displayed in Figure 3. Following docking score-based ranking, the top five peptides exhibiting the lowest docking scores were advanced to the subsequent in vitro and in vivo evaluations.

Figure 2.

The flowchart of the integrated computational-experimental screening process for peptides targeting PCSK9.

Figure 3.

The binding free energies (kcal/mol) of Pep2-8 and TPPs 1–5.

Interaction analysis

To gain a clear understanding of the binding modes of TPPs 1–5 with PCSK9, we carried out an in-depth investigation. The 2D interaction diagrams of TPPs 1–5 with PCSK9 were generated using the Ligand Interactions tool and presented in Figure S2. Additionally, representative 2D conformational snapshots of the TPPs 1–5-PCSK9 complexes were plotted via LigPlot software and displayed in Figure 4. The hydroxyl hydrogen of Thr377 in PCSK9 acted as a hydrogen bond donor, interacting with the carbonyl oxygen of Thr4 in TPPs 1–5 (2.77 Å). Additionally, the carbonyl oxygen atom of residue Phe3 in TPPs 1–5 formed hydrogen bonds with the hydroxyl hydrogen of residue Phe379 (2.80 Å) and the hydrogen atom on the sulfhydryl group of Cys378 (2.85 Å) in PCSK9, acting as the hydrogen bond acceptor. Arg0 and Arg2 residues of TPP-1 formed three hydrogen bonds with Asp374 of PCSK9, with bond lengths of 2.86 Å, 2.75 Å, and 2.94 Å, respectively. TPP-2 exhibited an identical hydrogen bonding pattern to TPP-1, targeting PCSK9 in the same manner. Arg0 of TPP-3 formed two hydrogen bonds with Asp374 of PCSK9, with bond lengths of 2.75 Å and 2.94 Å. Additionally, TPP-4 formed eight hydrogen bonds with the LDLR-binding domain of PCSK9. Arg0 and Arg2 residues of TPP-5 formed two hydrogen bonds with Asp374 of PCSK9, with bond lengths of 2.86 Å and 2.81 Å, respectively. Finally, the residues Thr1, Ser5, Trp6, and Tyr9 of TPPs 1–5, together with residues Asp238, Ala239, Ile369, Cys378, and Val380 of PCSK9, constituted the hydrophobic region of the interaction interface.

Figure 4.

The 2D interaction plots for TPPs 1–5 and the LDLR-binding domain in PCSK9 using the LigPlot. The hydrogen bonds are shown in green.

3D interaction diagrams of the predicted binding modes of TPPs 1–5 with PCSK9 were generated via MOE software (Figure 5). While visualisation of the TPPs 1–5-PCSK9 interaction interface differed by software, interfacial interaction strength results were consistent. Details of PCSK9 residue interactions at this interface are provided in Table S1.

Figure 5.

The 3D interaction plots for TPPs 1–5 and the LDLR-binding domain in PCSK9. (A, B) TPP-1 are colour-coded by yellow. (C, D) TPP-2, purple; (E, F) TPP-3, orange; (G, H) TPP-4, cyan; (I, J) TPP-5, green. The LDLR-binding domain residues of PCSK9 are displayed as sticks: nitrogen atoms are blue, oxygen atoms are red, and other residues are silver.

MST assay

Using MST assays, the binding affinities of Pep2-8, TPPs 1–5 and Mutant-TPP-4 for human recombinant PCSK9 were quantified, with the results summarised in Table 1. The binding affinity results are presented as K_d values, where a smaller K_d indicates stronger binding affinity⁴³. Pep2-8 was used as the positive control group, with a high K_d value (K_d = 0.74 ± 0.06 µM). Compared to Pep2-8, TPPs 1–5 exhibited lower K_d values, indicating stronger binding affinities. Among TPPs 1–5, TPP-1 (K_d = 0.17 ± 0.02 µM) and TPP-4 (K_d = 0.08 ± 0.01 µM) exhibited significantly stronger binding affinities than the others in the assays. The binding isotherms of TPPs 1–5 and Pep2-8 are presented in Figure S3. Subsequently, domain-directed alanine (A) mutagenesis was performed on the high-affinity TPP-4 peptide sequence to determine its binding affinity with PCSK9. The results demonstrated that compared with TPP-4, the binding affinity between Mutant-TPP-4 and PCSK9 was drastically reduced (K_d > 100 μM).

CD Spectroscopy

Given TPP-4’s high affinity, CD spectra of TPP-4 and PCSK9 both pre- and post-incubation were acquired to assess their interaction. CD spectroscopy is a pivotal tool for analysing the secondary structures of peptides or proteins. In CD spectra, the β-sheet structure is characterised by a negative peak at 218 nm and a positive peak at 195 nm36. The CD spectrum of TPP-4, subsequent to incubation with PCSK9, displayed a prominent β-sheet structural feature (Figure S4). The perturbations in secondary structural features validate the presence of a specific interaction between PCSK9 and TPP-4.

MD simulation

The 100 ns MD simulations were conducted to further evaluate the stability of the binding modes in PCSK9-Pep2-8, PCSK9-TPP-1 and PCSK9-TPP-4 complexes. RMSD, RMSF, secondary structure, Rg and number of hydrogen bonds were the primary metrics taken into account⁴⁴. The RMSD value of the PCSK9-Pep2-8 complex was higher than that of PCSK9 protein alone, with significantly more pronounced fluctuations (Figure 6(A)). TPP-1 reached equilibrium at 40 ns, with the RMSD value remaining stable at 0.4 nm in the PCSK9-TPP-1 complex system (Figure 6(B)). In contrast, TPP-4 rapidly achieved equilibrium and maintained a low RMSD value of 0.2 nm within the PCSK9-TPP-4 complex system (Figure 6(C)). RMSF values of key amino acids at the LDLR binding domain of PCSK9 (Ala220, Ser221, Ser225, Asp238, Ala239, Ile369, Ser372, Asp374, Cys375, Thr377, Cys378, Phe379, Val380, Ser381) were all below 0.28 nm in PCSK9-Pep2-8, PCSK9-TPP-1 and PCSK9-TPP-4 complexes (Figure 6(C–F)). The specific RMSF values for the corresponding residues are presented in Table S2. The multiple hydrogen bonds were gradually formed during this process. The most significant fluctuations occurred at the C/N termini, which exhibit greater structural flexibility. The secondary structure composition of PCSK9-Pep2-8, PCSK9-TPP-1 and PCSK9-TPP-4 complexes showed minimal fluctuations (Figure 6(G–I)). Most residues were composed of Structure, α-Helix, and Coil structures in the complexes. The Rg values of the complexes fluctuated within the range 1.79–1.83 nm (Figure 6(J–L)). In summary, TPP-1 and TPP-4 exhibited significantly greater stability than Pep2-8 throughout the 100 ns MD simulations of their binding to the LDLR binding domain of PCSK9.

Figure 6.

The plots of MD simulation-derived metrics for PCSK9 in complex with Pep2-8, TPP-1 and TPP-4.

MM-PBSA analysis

The binding free energies of PCSK9 with ligands (Pep2-8, TPP-1, and TPP-4) were calculated using the MM-PBSA method, and decomposed into four components: van der Waals, electrostatic, polar solvation, and non-polar solvation. For Pep2-8, hydrogen bond interactions served as the primary driving force for binding. However, the large polar solvation energy exerted an unfavourable effect on the PCSK9-Pep2-8 association (Figure 7(A)). In contrast to Pep2-8, both TPP-1 and TPP-4 mitigated the adverse impact of excessive polar solvation energy, thereby enhancing the stability of their binding to PCSK9 (Figure 7(B,C)). For Pep2-8, residues Asp374 and Thr377 were identified as key residues at the PCSK9-Pep2-8 binding interface (Figure 7(D)). Additionally, Ile369, Asp374, and Phe379 were defined as key residues at the PCSK9-TPP-1 and PCSK9-TPP-4 binding interfaces (Figure 7(E,F)). In summary, Asp374 exhibited strong contributions across all complexes, qualifying it as a “universal key residue” for PCSK9 binding. A greater number of hydrogen bonds at the binding interface correlates with higher binding stability and a lower binding free energy. Compared to Pep2-8, TPP-1 and TPP-4 formed more hydrogen bonds with PCSK9 at their binding interfaces (Figure 7(G–I)). Correspondingly, TPP-1 and TPP-4 exhibited greater binding stability with PCSK9 over the 100 ns MD simulation. The experimental outcomes are in good agreement with the affinity data obtained from MST experiments.

Figure 7.

The energy decomposition, residue contribution, and hydrogen bond dynamics of PCKS9 in complex with Pep2-8, TPP-1 and TPP-4.

PCSK9-D374Y (D374Y gain-of-function mutation) elevates plasma LDL-C levels and increases ASCVD susceptibility in humans⁴⁵. Therefore, binding free energy assays of TPP-4 with PCSK9-D374Y were also performed. The total binding free energies of TPP-4 to wild-type PCSK9 was −39.03 ± 2.36 kcal/mol, while that to PCSK9-D374Y was −41.39 ± 1.81 kcal/mol (Table S2). Compared with wild-type PCSK9, TPP-4 exhibited a lower total binding free energy but stronger binding affinity for PCSK9-D374Y.

Serum stability analysis

Given the strong interaction between PCSK9 and TPP-4, stability assays of TPP-4 were conducted in human serum. After 240 min, the residual rate of TPP-4 in serum remained above 80% (Figure S5). The results indicated that TPP-4 could exist stably in human serum.

In vitro assessment of cell surface LDLR expression

Compared with Pep2-8, all TPPs 1–5 restored surface LDLR levels, with TPP-4 exerting a significant effect. More importantly, TPP-4 and Evolocumab exhibited comparable efficacy in restoring surface LDLR levels in HepG2 cells (Figure 8(A)). Subsequently, LDLR levels were assessed in HepG2 cells treated with varying concentrations of TPP-4. EC₅₀ (median effective concentration), with lower values indicating superior efficacy, was used to assess drug potency. TPP-4 exhibited an EC₅₀ of 1.12 ± 0.05 μM (Figure 8(B)). These observations confirm that TPP-4 significantly upregulates surface LDLR levels in HepG2 cells. Furthermore, the cytotoxicity of TPP-4 in HepG2 cells was evaluated. Data indicated that TPP-4 exerted negligible toxicity towards HepG2 cells, even at a high concentration of 400 μM (Figure S6). Given TPP-4’s high efficacy and low toxicity, subsequent in vivo efficacy evaluations were performed.

Figure 8.

The surface LDLR expression in HepG2 cells. (A) HepG2 cells treated with TPPs 1–5, Pep2-8, and Evolocumab. (B) HepG2 cells treated with TPP-4 at different concentrations. ***P < 0.001 vs Pep2-8.

In vivo measurements of hepatic LDLR and plasma cholesterol

To assess the in vivo transformative potential of TPP-4, hepatic LDLR and plasma total cholesterol levels were quantified in drug-treated mice. Relative to the vehicle control, both TPP-4 and Evolocumab enhanced hepatic LDLR expression levels (Figure 9(A)). Correspondingly, when compared with the vehicle control, both compounds reduced plasma total cholesterol concentrations (Figure 9(B)). Furthermore, TPP-4 demonstrated efficacy comparable to that of Evolocumab in restoring LDLR levels and reducing plasma total cholesterol. Thus, TPP-4 represents a promising peptide that enhances hepatic LDLR levels and reduces plasma cholesterol levels in mice. Furthermore, TPP-4 also holds potential as a therapeutic candidate for ASCVD.

Figure 9.

(A) Liver LDLR levels in mice treated with TPP-4 and Evolocumab. (B) Plasma total cholesterol levels in mice treated with TPP-4 and Evolocumab.

Discussion

Four pharmacophore features were generated based on the interaction of Pep2-8 with PCSK9 at the LDLR binding interface. TPPs 1–5 were identified from the database through screening against this pharmacophore model combined with molecular docking. Although TPPs 1–5 share high sequence homology with Pep2-8 (both harbouring the T-FTS-EE- core motif), these subtle variations impart enhanced potency. Studies by Zhang Yingnan’s group have demonstrated that deletion of the carboxy-terminal L10-V13 sequence in Pep2-8 resulted in undetectable binding affinity to the PCSK9 binding domain³⁰. Nevertheless, the amino-terminal k (lysine)/r (arginine) in TPPs 1–5 forms hydrogen bonds with the Asp374 residue at the LDLR binding domain in PCSK9, a molecular mechanism that enhances their binding affinity. Asp374 is critical for the PCSK9-LDLR interaction, as established by prior research⁴⁶. Despite also lacking the L10-V13 sequence, TPPs 1–5 exhibit high binding affinity.

Among TPPs 1–5 and Pep2-8, TPP-4 exhibited the highest binding affinity for the PCSK9 protein. Compared with TPPs 1–3 and TPP-5, TPP-4 formed the greatest number of hydrogen bonds with the key residues of PCSK9. Consequently, the K_d values of TPP-4 towards recombinant PCSK9 were measured to be 0.08 ± 0.01 μM. Relative to Pep2-8, TPP-4 demonstrated an approximate 9-fold enhancement in binding affinity. Subsequently, domain-directed mutagenesis of TPP-4 using alanine yielded Mutant-TPP-4, with K_d value > 100 μM. These results indicate that the presence of arginine in the TPP-4 sequence is critical for its interaction with the LDLR-binding domain of PCSK9. At 240 min post-incubation, over 80% of TPP-4 persisted in serum. This high stability circumvents the well-recognised limitation of peptide inhibitors, which are inherently susceptible to degradation. Relative to Evolocumab, TPP-4 elicited comparable therapeutic efficacy in both in vitro HepG2 cellular assays and in vivo animal models. In summary, TPP-4 represents a high-affinity and potent candidate peptide that disrupts the PCSK9-LDLR interaction for the treatment of ASCVD.

The EC₅₀ value of Pep2-8 for restoring surface LDLR levels in HepG2 cells was 12.5 μM, whereas that of TPP-4 was approximately 11-fold lower⁴⁷. For the N-methyltetrazole derivative (Melm), the EC₅₀ for restoring surface LDLR levels in HepG2 cells was 6 μM, with TPP-4 showing an approximately 5-fold lower EC₅₀ by comparison⁴⁸. In contrast, peptide 18 exhibited an EC₅₀ of 175 nM for this activity in HepG2 cells³². Furthermore, MK-0616 is a peptidic allosteric inhibitor of PCSK9 that binds to PCSK9, induces conformational changes, and thereby blocks the PCSK9-LDLR interaction⁴⁹. In comparison with MK-0616, TPP-4 exhibits superior targeting specificity and circumvents off-target-associated toxic side effects. In vitro and in vivo experimental evaluations of TPP-4 against PCSK9-D374Y are also planned. Additionally, future studies on TPP-4 could involve the introduction of staples to form a stapled peptide, thereby stabilising its β-sheet structure and further enhancing its efficacy.

Conclusion

In this study, we successfully identified a high-affinity and potent peptide inhibitor (TPP-4) targeting PCSK9 using structure-based virtual screening strategies. Pep2-8 was selected as the positive control. Compared to Pep2-8, TPP-4 exhibited the lowest binding free energy (approximately −9.8 kcal/mol), indicating stronger binding. Interaction analysis revealed that TPPs 1–5 interact with the LDLR-binding domain of PCSK9 via hydrogen bonding and hydrophobic interactions. Furthermore, compared to Pep2-8 (K_d = 0.74 ± 0.06 μM), TPP-4 showed a lower K_d value (K_d = 0.08 ± 0.01 μM). CD spectroscopy also corroborated the key interactions between TPP-4 and the LDLR-binding domain of PCSK9. TPP-4 stably bound to the LDLR-binding domain of PCSK9 during 100 ns MD simulation. Moreover, Serum stability assays demonstrated that TPP-4 exhibited considerable stability in human serum. Finally, TPP-4 showed negligible cytotoxicity against HepG2 cells and efficiently restored surface LDLR levels (EC₅₀ of 1.12 ± 0.05 μM). Notably, in mice, TPP-4 potently upregulated hepatic LDLR expression and lowered plasma total cholesterol. Therefore, data demonstrate that TPP-4 may be a candidate peptide for the future treatment of ASCVD.

Funding Statement

This study was supported by the National Natural Science Foundation of China (82273859).

Disclosure statement

No potential conflict of interest was reported by the authors.

Ethical approval

All experiments using mice were carried out according to an Animal Use Protocol approved by the Ethics Committee of China Pharmaceutical University (Approval number: YSL-202509003).

Data availability statement

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