Investigation on the characteristics and mechanisms of ACE inhibitory peptides by a thorough analysis of all 8000 tripeptides via binding free energy calculation

Abstract Food‐derived angiotensin I‐converting enzyme (ACE) inhibitory peptides represent a potential source of new antihypertensive. However, their characteristics and binding mechanisms were not well understood. In this study, novel energy calculation and experimentation were combined to elucidate the characteristics and mechanisms of ACE inhibitory tripeptides. ACE inhibitory activity of all 8,000 tripeptides was investigated by in silico experiments. IC50 values of the five top‐rated tripeptides ranged from 5.86 to 21.84 μM. Five hundred top‐ranked tripeptides were chosen for detailed structure–activity analysis, and a significant preference for aromatic amino acids at both C‐ and N‐terminus was found. By binding free energy analysis of nine representative tripeptides via MM/GBSA, electrostatic energy was found to be the leading energy that contributed to the binding of ACE with its high affinity tripeptides. Besides, S355, V380, and V518, three residues positioned around the classical binding pockets of ACE, also played a key role in ACE's binding. Therefore, for tripeptides, their binding pockets in ACE were redefined. In conclusion, the characteristics of ACE inhibitory peptides were more deeply illustrated by the thorough analysis of all tripeptides. The energy analysis allows a better understanding of the binding mechanisms of ACE inhibitory peptides, which could be used to redesign the ACE inhibitors for stronger inhibitory activity.


| INTRODUC TI ON
Cardiovascular disease (CVD) will result in more than 23.6 million deaths each year by 2030 (Deshwal et al., 2017). Among the risk factors for CVD, high blood pressure (BP) is associated with the strongest evidence for causation and it has a high prevalence of exposure. BP is associated with different biochemical pathways. Renin-angiotensin system (RAS) and kallikrein-kinin system (KKS) are pivotal regulators that control BP. Within RAS, ACE catalyzes the conversion of inactive angiotensin I to angiotensin II, which is a powerful vasoconstrictor (Natesh et al., 2003). Within KKS, ACE metabolizes the vasodilator bradykinin (Fu et al., 2017).
Accordingly, blood pressure can be lowered by the inhibition of ACE activity. Currently, ACE inhibitors, including captopril and lisinopril, have achieved widespread usage in clinical treatments of hypertension. However, they can cause undesirable adverse effects, such as hypotension, renal failure, and dysgeusia (Brown & Vaughan, 1998). Thus, ACE inhibitory peptides derived from food have drawn significant attention as they may act as safe alternatives to ACE drugs, as well as additives in functional foods (Deng et al., 2017). ACE inhibitory peptides have been identified from a wide range of food proteins, such as Antarctic krill (Ji et al., 2017), seaweed (Admassu et al., 2018), olive seeds (Esteve et al., 2015), green soybeans (Hanafi et al., 2018), and Chlorella vulgaris (Xie et al., 2018). Among them, short chain peptides like tripeptides are point of focus, as they are predicted to be easily absorbed into the blood circulatory system, and less affected by gastrointestinal digestion (Martin & Deussen, 2017;Yu et al., 2020). Based on structure-activity analysis, several characteristics of the identified ACE inhibitory peptides have been found. For example, a Cterminal Pro is more preferred by ACE inhibitory peptide (Garcia Mora et al., 2014).
The conventional approach for identifying ACE inhibitory peptides involves the generation of peptides encrypted in food proteins by enzymatic hydrolysis, followed by the isolation of ACE inhibitory fractions from hydrolysates via chromatographic techniques and their identification by mass spectrometry (Saadi et al., 2015). However, experiments are time-consuming and labor costing. Furthermore, the active peptides may not be totally harvested due to the incomplete hydrolysis and peptide loss. Limited numbers of ACE inhibitory peptides restrict their structureactivity relationship study. In the last few years, theoretical calculation methods have been employed to study ACE inhibitory peptides, especially in the identification of them. For example, in our previous work, the relationship between docking scores and experimental IC 50 values of 113 ACE inhibitory peptides was investigated, which could be used to predict the IC 50 values of a peptide without doing experiments (Wu et al., 2014). Panyayai et al., (2018 identified several potential ACE inhibitors by virtual screening for the ACE inhibitory activity of all 8,000 tripeptides, which has shown that theoretical calculation method is superior in the ACE-inhibitor identification as screening inhibitors from 8,000 tripeptides can be hardly achieved by experimentations. However, despite an increasing number of ACE inhibitory peptide sequences have been identified, the structural features that govern the ACE inhibitory properties of peptides and the binding mechanisms of ACE with inhibitors are still not well understood. Therefore, in this study, energy analysis was employed to elucidate the characteristics and mechanisms of ACE inhibitory peptides. First, the ACE inhibitory activity of all 8,000 tripeptides was ranked by theoretical calculation methods. Five hundred top-rated peptides were used for structure-activity relationship study. Furthermore, the inhibitory mechanisms of ACE inhibitory peptides were investigated by binding free energy analysis and free energy decomposition. The energy analysis of ACE/tripeptide complexes enhanced the comprehension of the mechanisms of ACE inhibitors.

| Generation of tripeptide library
Python scripts were used to build the tripeptide library sequences, with 8,000 possible combinations generated. Three-dimensional structures of the library were prepared with the Ligprep module in Maestro 2.9 (Schrödinger, LLC) and further minimized with OPLS3 force field parameters.

| ACE preparation
The crystal structure of human ACE (PDB ID: 1O8A) was imported from the Protein Data Bank. Crystallographic water from the structure was removed, whereas hydrogen atoms were added and missing loops were rebuilt. The OPLS_2005 force field was selected for energy minimization. All steps were performed with the protein preparation wizard in Maestro.

| Molecular docking of the tripeptide library with ACE
Docking simulations of the library were performed using Glide module in Maestro. The grid used for docking was created using standard parameters. The grid box coordinates were located at the Zn (II) position. A standard precision protocol was used for the ligand position screening.

| MD simulation process
AMBER18 was used to perform MD simulations. The leap module and ff14SB force field were used for the accomplishment of missing atoms and obtaining of necessary parameters, respectively.
Cationic dummy atom (CaDA) approach was employed to treat zinc (II) in ACE. The TIP3P water box with 10 Å buffer was used as the solvent environment, and sodium ions were added to the system to maintain neutral. Next, the steepest descent and conjugate gradient methods were applied for the energy minimization.
Chemical bonds, including hydrogen atoms, were constrained using the SHAKE algorithm. MD simulation was performed for 50 ns.

| MM/G BSA
The MM/GBSA method can be expressed by the following equation: Where ΔG solv and ΔG gas represent solvation free energy and gas-phase energy, respectively. ΔG gas is calculated by the following equation: Where ΔE ele and ΔE vdW represent electrostatic interaction and van der Waals (vdW) interaction. △ G solv contains polar and nonpolar solvation free energy as follows: The polar solvation free energy ΔG gb is calculated by the GB module, while the nonpolar solvation free energy ΔG np is calculated by: The SASA is the solvent-accessible surface area and calculated by the MSMS program (Sanner et al., 1996). The β and γ are set to 0.00 kcal·mol -1 and 0.005 kcal·(mol·Å 2 ) -1 , respectively.

| Peptide synthesis
Potential ACE inhibitory peptides selected based on the docking results were solid phase synthesized from Bankpeptide Biological Technology CO., Ltd. The purity (>95%) of these peptides was verified by RP-HPLC on an Inertsil ODS-SP column (4.6 mm × 250 mm, 5 μm, Shimadzu Co. Ltd.). Sequences were verified by mass spectrometry (Micromass ZQ, Waters).

| Assay of ACE inhibitory activity
ACE inhibitory activity was measured in vitro, using the method reported by Cushman and Cheung (1971) with modifications. Briefly, 20 μL of inhibitory peptides at different concentrations and 20 μL of 5 mM HHL solution were preincubated at 37°C for 5 min in 90 μL of buffer A (50 mM HEPES, 300 mM NaCl, pH 8.3). Next, 20 μL of ACE solution (0.04 U/mL) was added to trigger the reaction. After incubation at 37°C for 1 hr, the reaction was terminated by adding 50 μL of 1 M HCl. Next, 10 μL of the mixture was separated through RP-HPLC (e2679, Waters) using an XBridge C 18 column (4.6 mm × 150 mm, 5 μm, Waters). The column was eluted by 25% acetonitrile in 0.1% trifluoroacetic acid (v/v) with a flow rate at 1 mL/min. Hippuric acid was monitored at 228 nm. Triplicate tests were performed for each sample. The IC 50 value was defined as the concentration of the peptide required to reduce the hippuric acid peak area by 50%. For comparative purposes, the IC 50 value of captopril was obtained using this methodology.

| Docking results of the tripeptide library
The combinations of 8,000 tripeptides were firstly generated by python scripts, and their 2D structures were converted to 3D structures. Then, 8,000 tripeptides were successfully docked into ACE, and the best pose of each peptide was chosen according to their docking scores. Next, the 8,000 docking scores were ranked. The complete ranking list is given in Table S1. A lower docking score indicates that the peptide in question is a potential ACE inhibitor, as it binds with ACE in a more favorable conformation.
To investigate the validity of the docking results, five top-ranked peptides (WWW, KYY, WRF, WRY, and WQW) and two bottomranked peptides (DGG and GGG) were synthesized, and their IC 50 values were tested in vitro (Table 1). The five top-ranked peptides were potential ACE inhibitors with IC 50 values ranging from 5.86 to 21.84 μM. Though the ACE inhibitory activity of these peptides was weaker than that of captopril (IC 50 = 0.037 ± 0.008 μM), it was reported that ACE inhibitory peptides could be healthier than ACE drugs (Deng et al., 2017). In contrast, the bottom-ranked peptides DGG and GGG showed poor inhibitory activity with IC 50 values higher than 5,000 μM. Therefore, the docking results obtained in this study are reliable and can be used further for ACE inhibitory peptide studies. Figure 1a illustrates the distribution of 8,000 docking scores.
Interestingly, the docking scores of top-ranked and bottom-ranked peptides increased quickly, whereas a huge proportion of peptides with medium scores showed a relatively slow increase in scores.
Therefore, the slope of the curve in Figure 1a was analyzed. As is clearly shown in Figure 1b, the slopes of the peptides ranked be-

| The structure-activity analysis of ACE inhibitory peptides
Five hundred top-ranked peptides were selected to investigate the occurrence frequency of 20 amino acids at the N-, C-, and internal positions. The results are illustrated in Figure 2a-c. High-frequency amino acids are preferential at particular positions.
As shown in Figure 2a, the occurrence frequency of 20 amino acids at the C-terminus polarizes. W, F, and Y were the three most frequently occurring amino acids, with frequencies of 37.0%, 19.0%, and 14.4%, respectively. The total proportion of these three amino acids accounted for 70.4% of all amino acids, indicating that ACE inhibitory peptides strongly prefer C-terminal aromatic amino acids.
Similarly, a virtual screening for ACE inhibitory tripeptides also showed that aromatic amino acids were preferred at C-terminus of the peptide (Panyayai et al., 2018). Several studies have previously isolated ACE inhibitory peptides with C-terminal aromatic amino acids, such as IRW (Majumder & Wu, 2010), VAW (Rao et al., 2012), RGY (Rao et al., 2012), and IVF (Miguel et al., 2007). In addition, QSAR analysis showed that aromatic amino acids at the Cterminus contribute to the ACE inhibitory activities of peptides (Wu et al., 2006). The occurrence frequency of peptides with His located at the C-terminal position was 7%, which ranked fourth. Rodríguez- acids (9.6%, 7.4%, 7.4%, and 6.8%). The frequency of the top-ranked amino acid at the internal position was 1.7 times higher than that of the fifth-ranked amino acid. In contrast, the ratio was 9.7 for the Cterminus and 3.7 for the N-terminus. This analysis revealed that the preference for a particular type of amino acid at the middle position was not as strong as that for the C-and N-positions. The amino acid P had the lowest frequency at the middle position, indicating that an internal P may not be a good choice for ACE inhibitory tripeptides. This conclusion is consistent with a previous study by Miguel et al., (2012) wherein the C-terminal P contributes to the ACE inhibitory activity, but the peptide with an internal position P binds weakly to ACE.
As shown in Figure 2d three peptides with C-terminal W from chia seeds, have been identified as ACE inhibitors (Grancieri et al., 2019). In order to enhance the antihypertensive effects of chia seeds, specific proteases can be used to increase the content of useful peptides. For instance, proteinase K is suitable for the production of peptides with C-terminal W, as they preferentially cleave after W (Saenger, 2013). After the second F I G U R E 2 The occurrence frequency of 20 amino acids at the C-terminus (a), Nterminus (b), and middle position (c) within the 500 top-ranked peptides and the C-terminus (d), N-terminus (e), and middle position (f) within the 500 bottom-ranked peptides digestion by gastrointestinal enzymes, short peptides with high ACE activity are more likely to be obtained. The above methods are economical, safe, and effective antihypertensive strategies which can be applied in food industry.

| Interactions of peptides with ACE
WWW, WQW, and WYW were the three top-ranked peptides with C-terminal W. In order to illustrate the binding mode of ACE inhibitors, the C-terminal W of the three peptides was replaced by C and P, and the interactions between ACE and the nine peptides were investigated (Figure 3). ACE contains three binding pockets.
A354, E384, and Y523 correspond to the S1 pocket, whereas Q281, H353, K511, H513, and Y520 correspond to the S2 pocket. E162 is the only amino acid contained in the S1' pocket (Lan et al., 2015). Hbond is the dominant interaction that stabilizes the ACE-inhibitor complex (Xie et al., 2018;Yuan et al., 2018 (Natesh et al., 2003). A docking simulation by Jimsheena and Gowda (2010) found that the antihypertensive drug lisinopril can coordinate with Zn(II) directly. However, no direct interactions of the nine studied tripeptides with Zn(II) were observed, which might explain their weaker inhibitory activities compared with lisinopril. Among the nine peptides, WWW, WYW, and WQW formed π-π bonds with H383, and the π-π bonds may interfere with the coordination effect of H383 with the Zn ion, which could enhance their inhibitory activities. Similar findings were reported by Abdelhedi et al., (2018) and Yu et al., (2018).
Although the interaction analysis showed certain rules, further energy calculations were carried out to explore the binding mechanisms of ACE inhibitors.

| Binding free energy analysis
The binding free energies (ΔG bind ) of the nine peptides with ACE were calculated by MM/GBSA (Figure 4). The binding free energies of peptides with C-terminal W (WWW, WYW and WQW) were generally lower than those of peptides with C-terminal P, while peptides with C-terminal C (WWC, WYC, and WQC) possessed the highest ΔG bind values among the nine peptides. A lower energy indicated a more stable complex. Therefore, the results for binding energy analysis were in accordance with the docking results, which further proved that C-terminal W contributes to the binding of ACE inhibitors.
The binding free energy can be divided into four parts: van der To further gain insight into the binding mechanisms of peptide-ACE interaction, the contribution of every residue to the binding free energy of ACE/peptide complex was calculated by free energy decomposition and is shown in  Figure 5a. It was assumed that ACE's classical binding pockets may not be completely applicable to tripeptides when ACE and tripeptide were combined. We supposed S355, V380, and V518 should be included in the binding pockets, and the structure of the newly proposed ACE binding pockets was illustrated in Figure 5b and 5c. Furthermore, the interactions between WWW and the three key residues were shown in Figure 5d. S355 formed a H-bond with WWW, which contributed to the stability of the complex, whereas hydrophobic interactions between WWW and S518 and V380 increase its binding affinity.
When designing and modifying ACE inhibitory tripeptides for better inhibition effects, S355, V380, and V518 should also be regarded as key elements.

| CON CLUS ION
In conclusion, experimentation and theoretical calculation methods were combined in this work to elucidate the characteristics and mechanisms of ACE inhibitory peptide by a thorough analysis of all 8,000 tripeptides. The IC 50 values of the five top-rated peptides ranged from 5.86 to 21.84 μM, which showed that they were potential alternatives of ACE drugs. By structure-activity analysis, it was found that aromatic amino acid positioned at whether C-or N-terminus of a tripeptide was linked to high ACE inhibitory activity. This characteristic of ACE inhibitor could guide the processing of antihypertensive functional foods. Additionally, electrostatic energy between ACE and its high affinity tripeptides was more requisite in keeping the stability of the complex than the other energies. S355, V380, and V518, three residues positioned around the classical binding pockets of ACE, also played a key role in ACE's binding. Therefore, for tripeptides, their binding pockets in ACE were redefined. The energy analysis allows a better understanding of the binding mechanisms of the ACE inhibitory peptides, which was useful in the design of ACE inhibitors.
However, one limitation of our work was that whether the top-ranked peptides are C-domain-selective ACE inhibitors was not verified. It has been reported that ACE inhibitors with C-domain specificity have the potential for blood pressure control with fewer adverse effects . Therefore, our future work will focus on the screening for ACE inhibitory peptides with C-domain specificity, as well as the analysis of their characteristics. for providing us computational time.

CO N FLI C T S O F I NTE R E S T
The authors confirm that they have no conflicts of interest with respect to the work described in this manuscript.

E TH I C A L A PPROVA L
This study does not involve any animal or human testing.