Computational Studies of Fish Protein Hydrolysates as Human Angiotensin I-Converting Enzyme Inhibitors

Homology modelling and validation of hACE-I

The crystal structure of hACE-I available in the protein data bank (PDB) was routinely used as a receptor (PDB ID: 1O8A) for various docking studies.^[10-12] However, this model lacks 3D structural elements of around 10% sequence and contains several rotamer outliers. Hence, to overcome this issue, the 3D model of hACE-I was built using the homology modelling technique. The target protein, ACE-Homo sapiens, was retrieved from UniProtKB ID: P12821. The 3D hACE-I structure model was built using the SWISS-MODEL online server (https://swissmodel.expasy.org/). Clustal Omega (version 1.2.1) is used to align the amino acid sequence (https://www.ebi.ac.uk/Tools/msa/clustalo/). The PDB ID 6H5W, which has the highest sequence identity with UniProtKB ID: P12821, was used as a template structure for the homology modeling. Global Model Quality Estimation (GMQE) and Quality Model Energy Analysis (QMEAN) values were used to select the most reliable 3D model. In addition, QMEAN (< 4.0) and GMQE (from 0 to 1) were also used to predict a more accurate model. The Ramachandran plot created using PROCHECKER (https://bio.tools/PROCHECK) was used to verify the quality of the 3D structure.

Molecular docking study

The 12 peptides and their 2D chemical diagrams shown in Table 1 were chosen as the ligands for this study based on previous ACE-I inhibitory assay studies.^[6] Chembiodraw (version 2018-18.0) (https://scistore.cambridgesoft.com/chembiodraw/) was used to create the chemical structure of selected peptides. The Ligprep module in Schrodinger 2020-4 was used to optimise the energies of the 3D structures of the ligands.^[13]

Table 1: Structure of fish products with their sequence and molecular weight

Selected Peptide	2D structure of the peptides	Source of isolated peptides	Molecular weight (Daltons)	IC50 value (mg/mL)
Gly-Phe		Salman fish	222.10	0.078
Pro-Pro		Salman fish	212.12	1.912
Gly-Pro-Val		Nile tilapia fish	271.15	2.3
Ala-Arg-Ser		Goby fish	332.18	0.025
Ala-Pro-Gln-Arg		Travally fish	470.26	1.360
Asp-Pro-His-Ile		Leather jacket fish	480.23	0.442
Glu-Leu-Ser-Ala-Pro		Goby fish	515.26	0.24
Arg-Lys-Ser-Ala-Gly		Horse mackerel fish	432.23	0.78
Thr-Phe-Pro-His-Gly-Pro		Pipefish	654.31	0.45
Val-Ser-Gln-Leu-Thr-Arg		Tilapia fish	702.40	1.44
Gly-Pro-Ala-Gly-Pro-Ala-Val		Squid	567.30	2.90
Ser-Pro-Ile-Ala-Pro-Ala-Leu		Flounder fish	667.39	1.27

The 3D hACE-I model built (section 2.1) was used as a receptor. The hACE-I receptor was preprocessed, and final coordinates were prepared using the Schrodinger 2019-3 suite, Maestro 11.7.012. H-bond networks were optimised, and proton states were generated at pH 7.0. The active site of hACE-I was located within a 10 Å radius, and a receptor grid box (14 Å x 14 Å x 14 Å) was built at its centre. The peptides were docked at the active site of the hACE-I using the OPLS3 force field.

The absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of selected peptides were analysed by ADMETSAR (http://lmmd.ecust.edu.cn/admetsar2)^[13] and pkCSM (http://biosig.unimelb.edu.au/pkcsm).^[9]

MD simulations were performed for the hACE-I complexed with ALA-ARG-SER ligand using Desmond (version 2020-3; Schrödinger). The hACE-I: ALA-ARG-SER complex was immersed in a single-point charge water box for 10Å, extending beyond the complex’s atoms. The Na⁺ and Cl^- ions were added to neutralise charges. The salt content was fixed to 0.15 M sodium and chloride ions to simulate physiological conditions. The MD was carried out in the isothermal-isobaric ensemble at 300 K and 1.63 bar pressure for 200 ns at a temperature of 300 K. The OPLS-3e force field was used in the MD simulations. Maestro’s Desmond simulation interaction diagram tool was used to extract the plots and figures.

The MM/GBSA method was used to calculate the binding free energy of protein-peptide complexes. The OPLS3 power field and dissolvable model VSGB of the Schrodinger suite 2019-2 prime model were used to analyse the protein-peptide complex’s free energy.

Targeted template sequence alignment and validation of hACE-I

The ACE bound with cytochrome bo3 ubiquinol oxidase complex (BO3) (PDB ID: 6H5W; resolution: 1.37Å) is the closest template with hACE-I (UniProtKB ID: P12821). The multiple sequence alignment of hACE-I (UniProtKB ID: P12821) with the ACE C-domain (PDB ID: 6H5W) showed a sequence similarity of 0.63 and a sequence identity of 99.26%. [Figure 1a]. The QMEAN score (-3.53) and GMQE (0.82) indicated that the 3D structure of hACE-I is reliable and of good quality. The predicted local similarity graph of hACE-I is shown in Figure 1b. Most of the residue values are close to 1, indicating that the expected model is of good quality. The homology-modeled hACE-I structure of the present study lies within the range of other ACE structures available in PDB, which confirms its consistency [Figure 1c]. Ramachandran plot [Figure 1d] and Ramachandran statistics [Figure 1e] were calculated using Protein Structure Validation Suite (PSVS) to check the quality of the homology model. The Ramachandran plot statistics revealed that 96.2% of amino acid residues of the hACE-I are present in the most favored region, 3.6% in the additional allowed region, and 0.2% in the generally allowed region. And no residue lies in the Ramachandran plot’s disallowed region. The G-Factor explains the stereochemical properties of hACE-I [Figure 1f]. The positive G-Factor value indicates a better score in selected residues (1-540) of the modeled 3D structure of hACE-I [Figure 1g].

Close

Figure 1:

Homology modeling of hACE-I. (a) Multiple sequence alignment of hACE-I (UniProtKB ID: P12821) with ACE C-domain (PDB ID: 6H5W). (b) Local quality estimation of the hACE-I model. (c) The Z-score plot of the hACE-I model. (d) Ramachandran plot of hACE-I model. (e) Ramachandran statistics of hACE-I model. (f) G-factor plot of the hACE-I model. (g) 3D representation of the hACE-I model. hACE-I: Human angiotensin-converting enzyme I, PDB ID: Protein Data Bank ID.

Export to PPT

Molecular docking

The FPH produced, containing antihypertensive components, is processed to isolate ACE inhibitory peptides. In the present study, FPH isolated 12 peptides purified from different fish sources and reported good antihypertensive activity for computational studies [Table 1].

Our current study intended to identify potential peptides as inhibitors for hACE-I. The molecular docking between hACE-I and 12 selected peptides was analyzed using Schrodinger 2020-3 (Maestro 11.7.012). The approved drug captopril was used as a control and docked with hACE-I. The docking results of 12 peptides and captopril are shown in Figure 2. The hACE-I contains three active pockets, S (Ser333, Ser355, Phe490, and Phe512), S1(Gln259, Gln281, Tyr498, Lys489, Lys511, and Tyr520), and S2 (His331, Ala332, His353, Ala354, His361, His383, His491, His513 andTyr501). It also has two sub-active pockets, S1ʹ and S2ʹ (Thr358, His361, Val380, His383, Phe435, Phe457, His491, Tyr501, His513, and Tyr523), and a metal-binding site (Zn2+)^[14], which can be a target for inhibiting the ACE activity.^[10] During hypertension, blocking the catalytic site of hACE-I can be a logical way to inhibit ACE activity.^[13] The molecular docking study reveals the possible interaction between the peptides and human ACE-I. The glide binding score of all peptides is summarized in Table 2A, and the potential interactions are shown in Table 2B. The 2D and 3D interactions for Gly-Phe [Figure 2a], Pro-Pro [Figure 2b], Gly-Pro-Val [Figure 2c], Ala-Arg-Ser [Figure 2d], Ala-Pro-Gln-Arg [Figure 2e], Asp-Pro-His-Ile [Figure 2f], Thr-Phe-Pro-His-Gly-Pro [Figure 2g], Arg-Lys-Ser-Ala-Gly [Figure 2h], Thr-Phe-Pro-His-Gly-Pro [Figure 2i], Val-Ser-Gln-Leu-Thr-Arg [Figure 2j], Gly-Pro-Ala-Gly-Pro-Ala-Val [Figure 2k], and Ser-Pro-Ile-Ala-Pro-Ala-Leu [Figure 2l] with hACE-I are shown in Figure 2.

Close

Figure 2:

Molecular interactions of hACE-I with peptides (a) Gly-Phe: Glycine-Phenylalanine, (b) Pro-Pro: Proline-Proline, (c) Gly-Pro-Val: Glycine-Proline-Valine, and (d) Ala-Arg-Ser: Alanine-Arginine-Serine.

Export to PPT

Close

Figure 2:

(e) Ala-Pro-Gln-Arg: Alanine-Proline-Glutamine-Arginine, (f) Asp-Pro-His-Ile: Aspartic-Proline-Histidine-Isoleucine, (g) Thr-Phe-Pro-His-Gly-Pro: Threonine-Phenylalanine-Proline-Histidine-Glycine-Proline, (h) Arg-Lys- Ser-Ala-Gly: Arginine-Lysine-Serine-Alanine-Glycine.

Export to PPT

Close

Figure 2:

(i) Thr-Phe-Pro-His-Gly-Pro: Threonine-Phenylalanine-Proline-Histidine-Glycine-Proline, (j) Val-Ser-Gln-Leu-Thr-Arg: Valine-Serine-Glutamine-Leucine-Threonine-Arginine, (k) Gly-Pro-Ala-Gly-Pro-Ala- Val: Glycine-Proline-Alanine-Glycine-Proline-Alanine-Valine and (l) Ser-Pro-Ile-Ala-Pro-Ala-Leu: Serine-Proline-Isoleucine-Alanine-Proline-Alanine-Leucine. The hACE-I residues making an H-bond are shown in line with arrow marks. The hACE-I residues, which make hydrophilic interactions, are demonstrated in the surrounding peptides. In the right panel, 3D pose views are shown. The binding site of hACE-I is shown on the surface representation. The peptides are depicted in a ball-and-stick model. The carbon atoms are coloured green. The oxygen, nitrogen, and hydrogen atoms are coloured red, blue, and white. hACE-I: Human angiotensin-converting enzyme I.

Export to PPT

The control molecule captopril shows a G-score of -7.2 kcal/mol, and 12 peptide molecules yielded a G-score from 1.42 to -14.54 kcal/mol, as shown in Table 2. The tripeptide Ala-Arg-Ser produced the highest binding affinity (-14.54 kcal/mol), followed by the Glu-Leu-Ser-Ala-Pro (-10.69 kcal/mol), and Gly-Pro-Val (-9.49 kcal/mol). In addition, the tetra peptides Asp-Pro-His-Ile (-8.12 kcal/mol), Ala-Pro-Gln-Arg (-8.03 kcal/mol), and Penta peptide Arg-Lys-Ser-Ala-Gly (-7.90 kcal/mol) showed higher affinity than the control molecule. Interestingly, the dipeptides Pro-Pro (-8.48 kcal/mol) and Gly-Phe (-8.43 kcal/mol) are better than some of the tri and tetra peptides discussed above. In contrast, the Hexa (Thr-Phe-Pro-His-Gly-Pro and Val-Ser-Gln-Leu-Thr-Arg) and Hepta (Gly-Pro-Ala-Gly-Pro-Ala-Val, and Ser-Pro-Ile-Ala-Pro-Ala-Leu) peptides bindings yielded poor results [Figure 2i and 2j]. These four hexa and hepta peptides are near the binding site but do not fit inside the binding pocket. Thus, out of 12 peptides, eight fit inside the binding pocket of hACE-I, and all showed a higher binding affinity with hACE-I than the control molecule.

The hACE-I active site amino acids His91, Glu162, Gln281, His315, His353, Ala354, Ser355, Ala356, Glu384, Lys511, His513, Tyr520, Tyr523, and Zn707 form hydrogen bonds with the peptides shown in Figure 2. Among these, Gln281 and Tyr523 interact with a maximum of six peptides, while His513 and the Zn atom interact with five peptides. His353 follows, interacting with four peptides.

Glu162 and Ala354 interact with two peptides, while His91, His315, Ser355, and Ala356 each form hydrogen bonds with one peptide. These hydrogen bond interactions highlight the significant roles of charged residues and the aromatic residues Tyr520 and Tyr523 in peptide binding^[15]. Additionally, six of the eight best-binding peptides show strong hydrogen bonding with water molecules within hACE-I. Notably, X-ray crystallographic structures solved by Masuyer et al.^[16] for two natural inhibitory peptides (11-mer and 10-mer) also revealed similar interactions, further supporting these findings. However, the top six peptides with good binding affinities are lower molecular weights ranging from 212 to 515 Daltons.

The antihypertensive potency of fish-derived peptides is expressed as an IC₅₀ value, indicating the half-maximal inhibitory concentration of peptides that can inhibit 50% of ACE activity. In the reported peptides, Ala-Arg-Ser showed high in vitro ACE-I inhibitory activity compared to other peptides, and the IC50 value was calculated as 0.078, 1.912, 0.445, 1.360, 0.02, 0.24, 0.78, 0.45, 1.44, 2.90, 1.27, and 2.3 mg/ml [Table 1].^[3] Therefore, the present analysis showed tripeptides have higher in vitro ACE-I inhibitory activity than other isolated peptides. A similar study was reported by Toopcham et al.^[17], and the study isolated peptides (Ile-Trp, Ile-Tyr, Thr-Val-Tyr, Val-Pro-Trp, and Val-Tyr) from Salmon fish. The Val-Pro-Trp peptide showed higher ACE-I inhibitory activity compared to other peptides.

Molecular dynamics simulation

MD simulations for 200 ns were conducted on the Ala-Arg-Ser complex, which was selected based on its higher Glide dock XP score. The RMSD plot [Figure 3a] showed stabilization after 5 ns with fluctuations between 0.5–1.5Å, indicating minimal conformational changes in the enzyme [Figure 3b]. Atom-wise fluctuations in the ligand were shown in the RMSF plot [Figure 3b], with the hACE-I protein tails fluctuating more than rigid α-helices and β-strands. Interactions of Ala-Arg-Ser with hACE-I included hydrophobic, ionic, and H-bonds [Figure 3c], with residues making more than one interaction represented in dark shades in Figure 3d. Ligand stability was analyzed through six parameters [Supplementary S1], where the RMSD stabilized after 75 ns [Supplementary S1a], and the radius of gyration remained stable post-100 ns [Supplementary S1b]. H-bond interaction with hACE-I stabilized the ligand [Supplementary S1c], and MolSA, PSA, and SASA plots suggested a stable conformation throughout the simulation [Supplementary S1d-S1f]. Figure 4a shows the 2D interaction of Ala-Arg-Ser with residues interacting >30% of the time, while Figure 4b provides torsion profiles around each rotatable bond in the ligand. The probability density and torsion potential indicate the ligand’s minimal conformational strain while maintaining its protein-bound shape. Overall, MD simulations confirmed that Ala-Arg-Ser is a strong binder to the hACE-I active site, making it a potential ACE inhibitor for hypertension.

Close

Figure 3:

Molecular dynamics results of hACE-I with Ala-Arg-Ser peptide. (a) RMSD analysis of hACE-I and Ala-Arg-Ser peptide. (b) RMSF analysis of hACE-I. (c) Molecular interactions of hACE-I with Ala-Arg-Ser peptide. (d) Protein-ligand contact frequency of the entire molecular dynamics simulation. hACE-I: Human angiotensin-converting enzyme I, RMSD: Root mean square deviation, RMSF: Root mean square fluctuation. Ala-Arg-Ser: Alanine-Arginine-Serine.

Export to PPT

Close

Figure 4:

Ala-Arg-Ser peptide interactions and confirmation analysis. (a) 2D schematic details of peptide-Ala-Arg-Ser interactions with the active site of the hACE-I. (b) Torsional analysis of isolated peptide-Ala-Arg-Ser conformations during the 200-ns simulation. Ala-Arg-Ser: Alanine-Arginine-Serine.

Export to PPT

In-vitro ACE-I inhibitory activity

Table 1 shows the antihypertensive activity of the peptides studied in this investigation. The antihypertensive activity depends on the ligand’s length, molecular weight, and its interactions with the receptor. The IC₅₀ value was calculated for each peptide, representing the 50% inhibition of ACE activity. Out of 12 peptides, Ala-Arg-Ser (0.025 mg/ml) showed high in-vitro activity and better ACE-I inhibition than other peptides. We have measured the IC50 value as 0.078, 1.912, 0.445, 1.360, 0.02, 0.24, 0.78, 0.45, 1.44, 2.90, 1.27, and 2.3 mg/ml for the peptides, and the results are summarized in Table 1.^[6] In general, low molecular peptides are absorbed directly in the gut. They tend to escape the degrading effects of gastrointestinal enzymes.^[3,6,9] The permeability of the ACE inhibitory peptides in the gut may be studied using simulated gastrointestinal digestion and absorption in gut cell lines in-vitro studies. The Caco₂ cell line was used to calculate the permeability assay of the four tripeptides isolated from tilapia FPH muscles. According to a report, the permeate IC50 values vary from 0.29 to 1.23 mM.^[17] In another report, high molecular weight peptides showed remarkable antihypertensive activity.^[18] The valine, arginine, tyrosine, and serine showed a strong affinity with ACE.^[19] Thus, low molecular weight peptides are demonstrated to have higher affinity with ACE and are proven to be the inhibitory molecules of ACE. The current computational studies support the previously reported findings, and the Ala-Arg-Ser is a potential candidate to be further validated through in-vitro systems.