Exploring the Potential of Peptides Derived from the Coelomic Fluid of Echinometra lucunter Targeting Inflammatory Cytokines in Placental Syndromes: In Silico Study

Ligand preparation

A total of 12 peptide sequences identified from the coelomic fluid of Echinometra lucunter were retrieved from a previously conducted peptidomics profiling study.¹⁶ The three-dimensional structures of the peptides were modelled using UCSF Chimera version 1.17.1 ( Table 1). Native co-crystallised ligands from each protein target were extracted using BIOVIA Discovery Studio 2024 to serve as control ligands. All ligands underwent refinements using the Molecular Operating Environment (MOE) application (version 2022.02), ensuring an optimised geometry with a Root Mean Square (RMS) gradient convergence threshold set to 0.001 kcal/mol/Ų.¹⁷

Protein preparation

The three-dimensional crystallographic structures of six key protein targets—c-Met, Interleukin-1β (IL-1β), Interleukin-10 (IL-10), Macrophage Migration Inhibitory Factor (MIF), Platelet-Derived Growth Factor (PDGF), and TNF-Related Apoptosis-Inducing Ligand (TRAIL)—were retrieved from the Protein Data Bank (https://www.rcsb.org) ( Table 2). These proteins were selected based on their well-documented involvement in the pathogenesis of placental syndromes, further substantiated by Mendelian randomization findings from a previous study.¹⁸ All target proteins were prepared, neutralized, and refined to a Root Mean Square (RMS) gradient of 0.001 kcal/mol/Ų using the MOE software (version 2022.02).¹⁷

Absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction

Absorption, distribution, metabolism, excretion, and toxicity (ADMET) were conducted for all E. lucunter peptides. Canonical SMILES notations of each peptide were submitted to the SwissADME webserver (http://www.swissadme.ch/index.php; accessed April 21, 2025) to predict ADME properties. This evaluation aimed to predict the potential impacts of the peptides on human health by assessing key pharmacokinetic parameters. Specifically, the analysis included predictions of human intestinal absorption, Caco-2 permeability, blood-brain barrier (BBB) permeability, cytochrome P450 enzyme 2D6 interactions (CYP2D6 substrate and inhibitor), and water solubility (Log S (ESOL)).¹⁹

Toxicity predictions were carried out using the ProTox-III webserver (https://tox.charite.de/protox3/; accessed April 21, 2025). The ProTox classified the peptides into six toxicity categories based on predicted LD₅₀ values (mg/kg BW): Class I (LD₅₀ ≤ 5 mg/kg BW; fatal if swallowed), Class II (5 < LD₅₀ ≤ 50 mg/kg BW; fatal if swallowed), Class III (50 < LD₅₀ ≤ 300 mg/kg BW; toxic if swallowed), Class IV (300 < LD₅₀ ≤ 2000 mg/kg BW; harmful if swallowed), Class V (2000 < LD₅₀ ≤ 5000 mg/kg BW; may be harmful if swallowed), and Class VI (LD₅₀ > 5000 mg/kg BW; non-toxic).²⁰

Molecular docking

Molecular docking was performed using the MOE application (version 2022.02). The binding site of the target protein was identified using the Site Finder tool from the MOE application. Ligand-protein interactions were analysed and evaluated based on binding affinity (kcal/mol) and Root Mean Square Deviation (RMSD) values. The peptide was selected based on the RMSD values below 2.5 Å and lower binding affinity (more negative values) compared to control ligands. Structural visualisation in both 2D and 3D formats further validated the spatial proximity and interaction patterns between peptides and their protein targets, supporting their candidacy as selective peptides.¹⁷

Molecular dynamics simulation

Molecular dynamics (MD) simulations were conducted using YASARA Dynamics version 4.3.13. Selected protein–peptide complexes were imported into the application by selecting the “Set Target” and “Macro & Movie” options from the Options menu. The simulation environment was configured under physiological conditions, with a temperature of 310 K, atmospheric pressure of 1 atm, pH 7.4, and a 0.9% NaCl concentration. The AMBER14 force field was applied, and the system was enclosed within a cubic periodic boundary cell extended by 10 Å. The total simulation time was set to 50,000 picoseconds (50 ns). Post-simulation, the Root Mean Square Deviation (RMSD) of the Cα atoms was calculated using the MD_analysis macro to evaluate structural stability over time. The results of the MD simulations were visualised using ORIGINPro 2024 (OriginLab, Massachusetts, United States).²¹

Peptide membrane permeability prediction

The membrane permeability of the selected peptides was predicted using the PerMM (Permeability of Molecules across Membranes) web server.²² Simulations were performed under physiological conditions (pH 7.4, 298 K temperature) using the “drag” optimization algorithm. The output included peptide transfer energy profiles as a function of their position relative to the membrane centre, along with membrane binding affinity (ΔG, kcal/mol) and membrane permeability coefficients (logPerm).²²

Absorption, distribution, metabolism, excretion, toxicity (ADMET) prediction

The in silico ADME analysis revealed that most of the evaluated peptides (Peptides A–L) exhibited favourable pharmacokinetic profiles. In terms of absorption, seven peptides (Peptides A, E, F, I, J, K, and L) showed high predicted human intestinal absorption (HIA > 0.80) ( Table 3). The remaining peptides demonstrated moderate absorption values, which are still acceptable for further development. Despite favourable absorption, all peptides were predicted to have low Caco-2 permeability, which may present a limitation in intestinal permeability; however, it does not exclude their bioavailability potential, particularly when alternative delivery systems are considered.

In terms of distribution, only Peptide F was predicted to cross the blood–brain barrier (BBB), indicating potential utility in targeting the central nervous system (CNS). The other peptides were non-permeant to the BBB, which could be advantageous for avoiding undesired CNS-related side effects in peripheral applications. None of the peptides were identified as substrates or inhibitors of the cytochrome P450 2D6 (CYP2D6) enzyme, indicating a low risk of CYP-mediated metabolism or drug–drug interactions—an important consideration for safety and consistency in pharmacokinetics. Similarly, none were predicted to inhibit the organic cation transporter 2 (OCT2), suggesting that these peptides are unlikely to interfere with renal excretion pathways. Regarding solubility, all peptides displayed good aqueous solubility as predicted by the ESOL model. Peptides C and H were classified as highly soluble, Peptides B, D, E, F, G, and L as very soluble, and Peptides A, I, J, and K as soluble. This favourable solubility profile supports their potential for oral and parenteral formulation development.

The toxicological assessment of peptides derived from E. lucunter exhibited a range of toxicity classifications from Class III (toxic) to Class VI (non-toxic) ( Table 4). Among them, peptide L was categorised as non-toxic (Class VI) with an LD₅₀ of 6,000 mg/kg BW, whereas peptide G was classified as toxic (Class III) with an LD₅₀ of 210 mg/kg BW. Five compounds—peptide B (2,000 mg/kg BW), peptide C (1,190 mg/kg BW), peptide D (836 mg/kg BW), peptide H (2,000 mg/kg BW), and peptide J (2,000 mg/kg BW)—were categorised to Class IV (harmful). Another five compounds—peptide A (2,400 mg/kg BW), peptide E (5,000 mg/kg BW), peptide F (2,500 mg/kg BW), peptide I (5,000 mg/kg BW), and peptide K—were categorised under Class V (may be harmful). Similarity scores ranged from 65.77% to 100%, with prediction accuracy above 68% for all compounds.

Organ toxicity assessment revealed that the majority of peptides were classified as inactive across five organ toxicity parameters: hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, and cytotoxicity. Specifically, all peptides, except peptide C, were predicted to be inactive across all parameters ( Table 5). In contrast, peptide C was predicted to be active in terms of hepatotoxicity and immunotoxicity.

Molecular docking

TRAIL

Based on the molecular docking result, all peptides, except peptide H, showed lower binding affinity than the control ligand (-6.96 kcal/mol) and RMSD values below 2.5 Å. Although peptide H (REGSPDLR) exhibited the lowest binding affinity (-10.72 kcal/mol) among the other, peptide H has RMSD values more than 2.5 Å (2.68 Å) ( Table 5). Therefore, peptide A (FLMLVDGH) was selected based on its favourable binding affinity (-10.02 kcal/mol) and stable conformation (RMSD = 2.39 Å). In addition to docking parameters, peptide A also exhibited a low toxicity profile (toxicity class V, LD50 = 2,400 mg/kg BW) and a favourable ADME profile. The 2D visualisation of molecular docking results revealed that peptide A formed more interactions with amino acid residues of TRAIL than the control ligand. Peptide A formed six interactions with the amino acid residues of TRAIL, consisting of one acidic hydrophilic (Glu140), one basic hydrophilic (Lys204), three polar hydrophilic (Ser121, Ser133, Gln193), and one greasy hydrophobic (Ile266) interaction ( Figure 1). In comparison, the control ligand formed one polar hydrophilic interaction (Gln193) with the amino acid residues of TRAIL. A similar interaction at the active site of TRAIL (Gln193) was observed between peptide A and the control ligand.

Figure 1. Two- and three-dimensional visualizations of the TRAIL complex with A) Peptide A and B) control ligand.

Molecular docking results revealed that peptides A, B, C, G, H, I, and J have lower binding affinity than the control ligand and an RMSD value lower than 2.5 Å ( Table 5). Among the favourable peptides, peptide A has the lowest binding affinity (-10.44 kcal/mol). Based on 2D interaction analysis, peptide A formed four interactions with the PDGF, consisting of two polar interactions with Asn134 and Ser136, and a greasy hydrophobic interaction with Ile38 ( Figure 2). Meanwhile, the control ligand formed one greasy hydrophobic interaction at Ile38. Both peptide A and the control ligand shared a similar interaction at Ile38 at the active site of PDGF.

Figure 2. Two- and three-dimensional visualizations of the PDGF complex with A) Peptide A and B) control ligand.

MIF

All peptides of E. lucunter showed lower binding affinity than the control ligand (5.95 kcal/mol). In terms of RMSD values, all peptides, except peptide H, met the criteria of an RMSD value below 2.5 Å ( Table 5). Peptide A was selected due to its lowest binding affinity (-9.32 kcal/mol) compared to the other favourable peptides. Based on 2D interaction analysis, peptide A formed three interactions with the MIF active site: one basic hydrophilic interaction (Lys32) and three greasy hydrophobic interactions (Phe113, Ile64, and Pro1). On the other side, the control ligand created one basic interaction (Lys32) and two greasy hydrophobic interactions (Ile64 and Phe113). Notably, both peptide A and the control ligand shared common interaction sites at Lys32, Ile64, and Phe113, indicating a similar binding orientation within the MIF active site ( Figure 3).

Figure 3. Two- and three-dimensional visualizations of the MIF complex with A) Peptide A and B) control ligand.

IL-10

In the context of IL-10 protein targeting, all candidate peptides demonstrated RMSD values below 2.5 Å and lower binding affinities compared to the control ligand (−4.44 kcal/mol). Among them, peptide J exhibited the lowest binding affinity (−10.54 kcal/mol); however, peptide J also presented a relatively high predicted toxicity (toxicity class IV; LD₅₀ = 2,000 mg/kg BW). Due to safety considerations, peptide A was selected instead, as it exhibited a lower predicted toxicity (toxicity class V; LD₅₀ = 2,400 mg/kg BW), while maintaining a lower binding affinity (−10.26 kcal/mol) and an RMSD value of 2.41 Å ( Table 5). Two-dimensional interaction analysis revealed that peptide A formed four hydrophilic interactions with IL-10, involving two acidic residues (Asp41 and Glu50) and two basic residues (Arg27 and Lys40). In contrast, the control ligand interacted with four residues, including Asp41 (acidic), Lys40 (basic), Gln42 (polar), and Phe30 (greasy). Notably, both peptide A and the control ligand shared interaction sites at Asp41 and Lys40, suggesting potential overlap in binding specificity ( Figure 4).

Figure 4. Two- and three-dimensional visualizations of the IL-10 complex with A) Peptide A and B) control ligand.

IL-1β

Molecular docking results indicated that all of the peptides have lower binding affinity than the control ligand (-6.23 kcal/mol) and RMSD below 2.5 Å ( Table 5). Peptide A was selected for further analysis due to its potential as a multi-target inhibitor, targeting TRAIL, PDGF, MIF, and IL-10. Peptide A has a binding affinity value of -8.29 kcal/mol and RMSD value of 1.94 A. Peptide A engages IL-1β through five key interactions: one acidic hydrophilic interaction (Glu25), two basic hydrophilic interactions (Lys74 and Lys77), one polar interaction (Tyr24), and one hydrophobic interaction (Leu80). In contrast, the control ligand formed fewer interactions, including a polar hydrophilic interaction (Thr79) and two hydrophobic interactions (Leu26 and Leu80) ( Figure 5). Importantly, both peptide A and the control ligand shared a similar hydrophobic interaction with Leu80.

Figure 5. Two- and three-dimensional visualizations of the IL-1β complex with A) Peptide A and B) control ligand.

c-Met

Molecular docking analysis revealed that ten out of twelve peptides exhibited root mean square deviation (RMSD) values lower than 2.5 Å and lower binding affinities compared to the control ligand (−6.87 kcal/mol). Among the favourable peptides, peptide A demonstrated the most promising interaction, with a binding affinity of −9.27 kcal/mol and an RMSD of 1.78 Å ( Table 5). Peptide A was therefore selected for further analysis due to its potential as a multi-target inhibitor across five additional protein targets. Two-dimensional interaction analysis showed that peptide A established five molecular interactions with the target protein: Glu1120, Glu1127, Arg1203, Lys1259, and Phe1260. These interactions included two acidic hydrophilic (Glu1120 and Glu1127), two basic hydrophilic (Arg1203 and Lys1259), and one hydrophobic (Phe1260) contact. In contrast, the native c-Met inhibitor formed only three interactions: Glu1127, Asp1204, and Val1201, comprising two acidic hydrophilic and one hydrophobic interaction. Notably, both peptide A and the c-Met inhibitor shared a similar interaction at residue Glu1127. The three-dimensional conformation of the c-Met-peptide A complex is illustrated in Figure 6.

Figure 6. Two- and three-dimensional visualizations of the c-Met complex with A) Peptide A and B) control ligand.

Permeability of Molecules through Membranes (PerMM) Analysis

PerMM (Permeability of Molecules through Membranes) analysis of Peptide A (FLMLVDGHN) showed that this peptide exhibited the strongest membrane-binding affinity among the tested candidates, with a calculated free energy of interaction (ΔG) of -6.76 kcal/mol with a DOPC lipid bilayer. Despite this favourable membrane anchoring, Peptide A demonstrated extremely low permeability, indicated by a log P_BLM value of -19.82. Additionally, the energy barrier at the bilayer centre reached a peak of +24.96 kcal/mol, further confirming its inability to translocate across the membrane passively ( Figure 7).

Figure 7. The membrane permeability results of Peptide A.

Molecular dynamics simulations

Molecular dynamics (MD) simulations of six receptor–ligand complexes revealed that Peptide A forms stable interactions with both IL-10 and IL-1β. This stability was evidenced by Cα backbone root mean square deviation (RMSD) values that consistently remained below 2 Å throughout the 50-nanosecond simulation trajectories. The simulation timeframe falls within the widely accepted 50 ns window for evaluating ligand–receptor complex stabilization,²³ and the resulting RMSD values fall well within the established threshold of <3 Å, indicating productive and stable interactions.²⁴ Among the targets assessed, IL-10 and IL-1β displayed the highest binding stability, suggesting their strong potential for functional modulation by Peptide A ( Figure 8).

Figure 8. Molecular dynamics simulation results of Peptide A with MIF, TRAIL, c-Met/HGF, IL-1β, IL-10, and PDGF.