Screening method of egg-derived ACE inhibitory peptide and application thereof

Abstract

The application discloses an ACE inhibitory peptide from eggs as well as a screening method and application thereof, and belongs to the technical field of biological medicines. The amino acid sequence of the ACE inhibitory peptide is any one of SEQ ID NO: 1-4. The application constructs an intestinal hydrolysis peptide library through a multi-stage virtual enzymolysis cascade simulation. The intersection of three models, mAHTPred, pLM4ACE (SVM) and pLM4ACE (MLP), is taken to perform activity prediction. Multi-directional immune safety evaluation is performed. Finally, four ACE inhibitory peptides are screened. The application overcomes the defects of high false positive rate, enzyme simulation deviating from reality and lack of immune safety evaluation in the prior art, and has a good application prospect in the field of anti-hypertensive drug development.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology, specifically relating to an egg-derived ACE inhibitory peptide, its screening method, and its application. Background Technology

[0002] Hypertension is a major risk factor for cardiovascular, cerebrovascular, and chronic kidney diseases, posing a significant threat to human health. Within the complex blood pressure regulatory network, angiotensin-converting enzyme (ACE) is a key enzyme regulating the renin-angiotensin system (RAS). ACE plays a central role in blood pressure regulation by catalyzing the conversion of angiotensin I into the potent vasoconstrictor angiotensin II and simultaneously inactivating bradykinin, a vasodilator. Therefore, inhibiting ACE activity is an important strategy for the clinical prevention and treatment of hypertension.

[0003] Currently, commonly used ACE inhibitors in clinical practice are mainly chemically synthesized drugs, such as captopril, enalapril, and lisinopril. Although these drugs have significant blood pressure-lowering effects, long-term use is often accompanied by toxic side effects such as dry cough, headache, hyperkalemia, taste disturbance, and angioedema. Some patients even discontinue treatment due to intolerance of adverse reactions. Against this backdrop, the discovery of dietary ACE inhibitory peptides from natural food proteins as alternatives to synthetic drugs has attracted much attention in the industry due to their high safety, mild action, easy absorption, and lack of obvious side effects.

[0004] Egg white protein is abundant and inexpensive, containing a variety of high-quality proteins such as ovalbumin, ovotransferrin, lysozyme, and ovomucoid, and has been proven to be an important source of potential ACE inhibitory peptides. In recent years, computer-aided screening technology has been introduced into the field of peptide mining, providing a new technical path for high-throughput screening. Among them, Yu Zhipeng et al. (Food Science, 2020, 41(12): 129-135) disclosed a virtual screening method for screening ACE inhibitory peptides of egg white protein. This method uses the ExPASy PeptideCutter program to perform virtual enzymatic digestion (pepsin and trypsin) of egg white protein, predicts biological activity using PeptideRanker, predicts water solubility using Innovagen, and predicts ADMET properties using admetSAR. Combined with pharmacophore model and molecular docking, tripeptides FQK, WGK, and ADW were screened out, and their ACE inhibitory activity was finally verified by in vitro experiments.

[0005] However, existing computer-aided screening techniques still have the following significant shortcomings: First, virtual enzymatic hydrolysis technology lacks differentiated parameter control over the actual processing and digestion environment. Existing technologies use the ExPASy PeptideCutter program for virtual enzymatic hydrolysis, employing only pepsin and trypsin for single-step simulation. This fails to consider the actual enzymatic hydrolysis conditions during industrial pretreatment (such as the types of industrial proteases and potential for missed cleavage), nor does it simulate multi-stage cascade digestion (continuous digestion from the stomach to the intestines). While this simplification reduces computational complexity, it leads to significant discrepancies between the predicted peptide release profile and actual industrial production and in vivo intake scenarios.

[0006] Second, peptide activity prediction relies on a single model or simple rules, resulting in a high false positive rate. Although existing technologies combine PeptideRanker, pharmacophore models, and molecular docking for multi-step screening, their activity prediction essentially still depends on a single scoring system, and the screening scope is limited to tripeptides. Due to the different standards for constructing the underlying training data and the varying dimensions of feature extraction among different prediction tools, the prediction results of a single model or simple combination have significant false positive biases, making it difficult to directly and reliably translate them into real in vitro high-activity data. Especially for longer peptide sequences (such as tetrapeptides and above), the accuracy of existing prediction methods based on simple physicochemical properties or single machine learning models further declines.

[0007] Third, immunogenicity and safety assessments are completely lacking. While existing technologies utilize the admetSAR tool to predict intestinal absorption, blood-brain barrier penetration, and acute oral toxicity of peptides, they do not address any assessments of sensitization, immunogenicity, or the risk of autoimmune reactions. Immunogenicity is a crucial safety indicator determining whether a peptide can be practically used in humans. Especially for known highly allergenic protein sources like egg white, the potential sensitization and immunotoxicity of derived peptides lack systematic evaluation, significantly limiting the practical translation of peptides from laboratory research into pharmaceutical formulations.

[0008] In summary, although egg white protein is an excellent source of ACE inhibitory peptides, existing computer-aided screening methods have significant limitations in terms of activity prediction accuracy, enzymatic digestion simulation reducibility, and immunological safety assessment. Furthermore, the sequences screened by existing methods are relatively short and may face rapid degradation in vivo. Therefore, there is an urgent need to develop an ACE inhibitory peptide screening method that overcomes these shortcomings while maintaining high reducibility in digestion simulation, high prediction accuracy, and effective immunological safety assessment, thus providing technical support for the development of antihypertensive drugs. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide an egg-derived ACE inhibitory peptide, as well as a screening method and application of the peptide.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: An egg-derived ACE inhibitory peptide, selected from one of the following sequences: IPPLVVQK (SEQ ID NO:1), PPGPPIPPLV (SEQ ID NO:2), IPPLVVQQW (SEQ ID NO:3), GVAKPNL (SEQ ID NO:4).

[0011] A pharmaceutical composition comprising the ACE-inhibiting peptide described above and a pharmaceutically acceptable carrier or excipient.

[0012] The application of the ACE inhibitory peptide or pharmaceutical composition in the preparation of antihypertensive drugs.

[0013] A method for screening the protein-derived ACE inhibitory peptide includes the following steps: (1) Multi-stage virtual enzymatic hydrolysis cascade simulation: Using egg white core protein as substrate, industrial hydrolysis simulation, gastric digestion simulation, intestinal primary digestion simulation and intestinal final hydrolysis simulation are carried out in sequence to construct an intestinal hydrolyzed egg white peptide library; (2) Activity prediction based on heterogeneous AI model combination: The peptide sequences in the peptide library obtained in step (1) are input into the mAHTPred meta-predictor, the SVM prediction mode of the pLM4ACE model and the MLP prediction mode of the pLM4ACE model respectively, and the intersection of the three prediction results is taken to obtain high confidence candidate ACE inhibitory peptides. (3) Multi-dimensional immune safety assessment: The candidate peptides obtained in step (2) were subjected to physicochemical sensitization property exclusion, immune epitope homology screening and MHC-II binding affinity assessment to screen out protein-derived peptides with both low immunogenicity and high ACE inhibition potential.

[0014] Preferably, the core protein of the egg white includes ovalbumin, ovotransferrin, ovomucin, lysozyme C, ovomucin α / β, ovoin inhibitor, ovoglycoprotein, avidin, and cystatin.

[0015] Preferably, in the industrial hydrolysis simulation stage described in step (1), four industrial-grade proteases are used, including papain, bromelain, subtilisin and thermophilic protease, and a total of 16 enzymatic hydrolysis combinations are configured, including single-enzyme independent hydrolysis and double-enzyme stepwise hydrolysis, and the missing cleavage site parameter MC=2 is set.

[0016] Preferably, in step (1), the gastric digestion simulation stage calls the pepsin digestion rule and sets the missed cleavage site parameter MC=0; the intestinal primary digestion stage calls the trypsin digestion rule; and the intestinal final hydrolysis stage calls the chymotrypsin digestion rule.

[0017] Preferably, the construction of the intestinal hydrolyzed egg white peptide library in step (1) further includes filtration and standardization: setting the molecular weight of the peptide to be less than 3000 Da, the number of amino acid residues to be ≥2, and performing sequence deduplication.

[0018] Preferably, the pLM4ACE model in step (2) is based on the 320-dimensional high-dimensional embedding features of the polypeptide sequence extracted by the pre-trained protein big language model ESM-2.

[0019] Preferably, the exclusion of physicochemical sensitization attributes in step (3) is performed using the AllerTOP v.2.1 server; the screening of immune epitope homology is performed using the immune epitope database IEDB platform, with the host set to Human; the assessment of MHC-II binding affinity is performed using the MHC-II binding affinity prediction model of IEDB, and the quantification is performed by the median binding percentile ranking of the prediction results of the NetMHCIIpan elution ligand mode.

[0020] The present invention has the following positive effects: This invention constructs a multi-stage cascaded enzymatic hydrolysis simulation strategy involving "industrial-stomach-intestinal" processes, highly replicating the peptide release patterns under real production and in vivo ingestion scenarios, overcoming the limitations of virtual hydrolysis being divorced from reality. This invention cleverly combines high-dimensional embedding features based on the large language model ESM-2 with a meta-predictor integrating physicochemical properties. Through an intersection strategy of SVM and MLP multi-classifiers, it effectively avoids the false-positive prediction problem caused by traditional single machine learning models, significantly improving the prediction accuracy of highly active peptides. This invention conducts in-depth immunological evaluation from three dimensions: physicochemical sensitization properties, T / B cell epitope homology cross-reactivity, and MHC-II molecule affinity, eliminating potential risks of allergies and autoimmune reactions.

[0021] The four peptides that were finally discovered and verified have high activity, are non-toxic, non-allergenic, and have stable physicochemical properties, and have great application prospects in the field of innovative antihypertensive drugs. Attached Figure Description

[0022] Figure 1 This paper presents a molecular docking binding energy distribution of four candidate ACE inhibitory peptides with the angiotensin-converting enzyme (ACE) active site. The vertical axis represents the binding energy (kcal / mol), and the horizontal axis corresponds to the peptide sequence. The results show that the binding energies of the four peptides range from -14.17 kcal / mol to -16.63 kcal / mol, exhibiting good binding affinity.

[0023] Figure 2The curves showing the inhibition rate of ACE activity by four synthetic peptides at different concentrations are presented. The horizontal axis represents the peptide concentration (mg / mL), and the vertical axis represents the ACE inhibition rate (%). The results show a clear dose-dependent effect.

[0024] Figure 3 The ACE inhibitory activity curves and half-maximal inhibitory concentrations (IC50) of four peptides, obtained by fitting a four-parameter logistic regression model, are shown. 50 The fitted graph is used to calculate the IC50 of each peptide. 50 value. Detailed Implementation

[0025] The present invention will be described in detail below with reference to specific embodiments.

[0026] Example 1: Screening of protein-derived ACE inhibitory peptides Step 1: Multi-stage virtual enzymatic hydrolysis cascade simulation and construction of egg white peptide library.

[0027] This step provides a method for constructing an intestinal hydrolyzed egg white peptide library based on multi-stage virtual enzymatic cascade simulation. The specific operation is as follows: 1. Acquisition of core protein sequences of egg white and environmental preparation Configure a Python 3.8 or later programming environment on your computer and install BioPython and the pyteomics bioinformatics library. Use a script to automate access to the UniProt database (https: / / www.uniprot.org / ), retrieve and download the FASTA sequences of 10 core egg white proteins as a substrate library for virtual enzymatic digestion. The 10 core proteins and their corresponding information include: Ovalbumin (P01012), Ovotransferrin (P02789), Ovomucoid (P01005), Lysozyme C (P00698), Ovomucin α / β (Q98UI9 / F1NBL0), Ovoinhibitor (P10184), Ovoglycoprotein (P80276), Avidin (P02701), and Cystatin (P01038).

[0028] 2. Four-stage virtual enzymatic hydrolysis cascade simulation An enzyme digestion rule algorithm was written based on BioPython and pyteomics to perform simulated enzyme digestion in the following four stages on the obtained protein sequence: (1) Industrial hydrolysis simulation stage: Four industrial-grade proteases were selected: papain, bromelain, subtilisin, and thermolysin. A total of 16 hydrolysis combinations were configured in the algorithm, including independent hydrolysis by a single enzyme (4 types) and stepwise hydrolysis by two enzymes (12 types). To accurately reflect the non-ideal hydrolysis state caused by reaction time limitations in real industrial production, the missed cleavages parameter MC=2 was set in the algorithm. The script was run to obtain the first-stage peptide library, and the theoretical degree of hydrolysis (DH) of each hydrolysis combination was calculated. The results showed that the theoretical degree of hydrolysis of different hydrolysis combinations in the industrial hydrolysis simulation stage ranged from 16.51% to 43.14%. Among them, the theoretical degree of hydrolysis of the bromelain and subtilisin combination was the highest, while the theoretical degree of hydrolysis of bromelain alone was the lowest, proving that stepwise hydrolysis by two enzymes can more efficiently cleave egg white protein than single-enzyme hydrolysis.

[0029] (2) Gastric Digestion Simulation Stage: Using the 16 peptide library products generated in the first stage as substrates, virtual cleavage was performed by calling the enzymatic cleavage rules of pepsin (mainly cleaving at the N-terminus or C-terminus of Phe, Tyr, Trp, and Leu residues). To simulate the thorough cleavage conditions under the extremely acidic environment of the human stomach (pH 1.5-2.0), the missed cleavage site parameter was strictly set to MC=0. After cleavage, the second-stage peptide library was constructed.

[0030] (3) Primary digestion stage in the intestine: Using the product of the second stage as a substrate, the trypsin rule is invoked to specifically cleave the carboxyl side of the basic amino acid residues lysine (Lys / K) and arginine (Arg / R) inside the polypeptide chain (excluding the case of proline). The third stage peptide library is constructed.

[0031] (4) Final hydrolysis stage in the intestine: Using the product of the third stage as a substrate, the final fragmentation process is further carried out by calling the chymotrypsin rule.

[0032] Results Analysis: After completing the four-stage cascade enzymatic digestion, the loss and transformation of peptides during the simulated gastrointestinal digestion process were systematically analyzed. During the four-stage cascade simulated gastrointestinal digestion, due to the thorough cleavage by digestive enzymes, some large polypeptides from the first stage were continuously degraded into extremely short, non-functional amino acid sequences. Based on the total number of peptides, 76,916 peptide records were detected in the first stage, decreasing to 37,961 in the fourth stage, a net reduction of 38,955, or a loss rate of 50.65%. Based on the number of unique peptide sequences, 13,393 unique peptides were detected in the first stage, decreasing to 4,226 in the fourth stage, a net loss of 9,167, or a loss rate of 68.45%. Further comparison of the sequence composition between the first and fourth stages revealed that 4,108 unique peptides were retained in the final stage, while 9,285 peptides from the first stage disappeared during subsequent gastrointestinal digestion; simultaneously, 118 new peptide sequences relative to the first stage appeared in the fourth stage. Overall, simulated gastrointestinal digestion significantly reduced the peptide library generated by the initial industrial pretreatment, with a more pronounced decrease in the number of unique peptides, indicating that the digestion process not only reduced the total amount of peptides but also significantly reshaped the peptide sequence composition.

[0033] 3. Peptide library filtering, standardization, and physicochemical characterization analysis To ensure the absorbability of peptides in the human body and the accuracy of subsequent structure-activity relationship analysis, a Python script was used to filter and remove duplicate peptides generated during the final hydrolysis stage in the intestine. The filtering threshold for peptide molecular weight was set to be less than 3000 Da, and single free amino acids were removed from the peptide length (i.e., the number of amino acid residues was limited to ≥2). After removing peptides that did not meet the above parameter requirements and those with repetitive sequences, a protein protein peptide library was established based on the enzymatic combinations of intestinal hydrolysis. Furthermore, the ProtParam module in the BioPython library was used to calculate and extract the total average hydrophilicity index (GRAVY) and terminal residue characteristics of each peptide in the peptide library.

[0034] A panoramic informatics analysis of the finally constructed intestinal hydrolyzed egg white peptide library showed that: Peptide Count: After rigorous filtration and deduplication, the final peptide library contained 4108 unique polypeptide sequences. Among them, the peptides generated by the combined use of Bacillus subtilis protease and thermophilic protease had the largest number, exceeding 3000. Molecular weight and length distribution: The molecular weight of peptides is normally distributed, mainly concentrated between 200-1500 Da; the peptide sequence length is highly concentrated between 3-8 amino acids. This distribution characteristic is consistent with the human intestine's uptake of small, easily absorbed peptides. Overall average hydrophilicity index: The GRAVY index of peptides obtained from different enzymatic hydrolysis pathways generally exhibits an approximately symmetrical distribution, mainly concentrated around 0, with the vast majority of values ​​falling within approximately 0. Within the range of 2.5 to 2.5, the median of each group of violin plots is close to 0, and the distribution of the interquartile range is relatively concentrated, indicating that the peptides produced by different enzymatic hydrolysis systems have little difference in overall hydrophobicity, and are all mainly of moderate hydrophobicity. Terminal residue distribution: Statistical analysis of the C- and N-terminal residues of the peptide library sequence showed that amino acids such as leucine, alanine, valine, and phenylalanine appeared at extremely high frequencies. This distribution characteristic is highly consistent with the known structure-activity relationship of highly active ACE inhibitory peptides, which favor hydrophobic ends and aromatic amino acids (such as Phe, Tyr, and Trp).

[0035] Step 2: Activity prediction and multi-faceted immune safety assessment based on heterogeneous AI model combination.

[0036] This step employs a combination of heterogeneous artificial intelligence prediction tools to predict the high-throughput activity of antihypertensive peptides and conduct multi-dimensional immune safety assessments. The specific operations are as follows: 1. Basic Toxicity Assessment and Initial Screening of Candidate Peptides: To initially eliminate unsafe sequences in the peptide library and reduce the computational load of subsequent complex AI model feature extraction, the egg white peptide library constructed in step one was first subjected to initial toxicity screening. All peptide sequences in the library were input into the ToxinPred 3.0 online server, and the Support Vector Machine (SVM) classification threshold was set to the default 0.0 for toxicity calculation. Sequences predicted as "Toxic" by the model were strictly eliminated. Ultimately, 3245 non-toxic peptides were retained, with sequence lengths mainly between 3 and 10 amino acids. These peptides were selected as safe candidate peptides for subsequent activity screening.

[0037] 2. First prediction output based on mAHTPred meta-predictor: Candidate peptide sequences from the peptide library are input into the mAHTPred prediction model. This model automatically extracts 51-dimensional probabilistic descriptor features based on peptide sequences, covering physicochemical information such as amino acid composition, dipeptide composition, and chain-transfer-distribution, and outputs the first ACE inhibitory activity prediction results of 432 peptides, denoted as output 1.

[0038] 3. Second and Third Prediction Outputs Based on the pLM4ACE Model's Dual Prediction Mode: The non-toxic candidate peptide sequences are input into the pLM4ACE prediction model. The pLM4ACE model is based on a pre-trained protein large language model (ESM-2, specifically the esm2_t6_8M_UR50D model) to extract 320-dimensional high-dimensional embedding features of the peptide sequences, thereby deeply capturing the global spatial structure and evolutionary semantic information of the peptides. Subsequently, the two prediction modes built into the model are called to perform activity evaluation: the SVM-based prediction mode in the pLM4ACE model is called, outputting 325 peptides as the second ACE inhibitory activity prediction result, denoted as output 2; the multilayer perceptron (MLP)-based prediction mode in the pLM4ACE model is called, outputting 316 peptides as the third ACE inhibitory activity prediction result, denoted as output 3.

[0039] 4. The peptide sequences predicted as positive (i.e., possessing potential ACE inhibitory activity) in Outputs 1, 2, and 3 are subjected to intersection processing. Specifically, only sequences that are simultaneously validated as positive by the mAHTPred model, pLM4ACE (SVM mode), and pLM4ACE (MLP mode) are retained. This accurately eliminates false-positive noise caused by a single feature, resulting in a highly confident set of 97 candidate protein peptides. From these, 13 peptides with an activity score > 0.9 are selected as candidate peptides.

[0040] 5. Comprehensive Assessment of Sensitization and Immunotoxicity: To further ensure that the above-mentioned non-toxic and highly active peptides will not induce allergies or autoimmune rejection in the human body, a rigorous multi-dimensional assessment of immunosafety was conducted on the 13 highly active candidate peptides obtained after intersection analysis. (1) Elimination of physicochemical sensitization properties: The candidate peptides are input into the AllerTOP v.2.1 server. The system performs sequence-free comparison of the physicochemical properties of the peptides based on the kNN algorithm and amino acid E-descriptors features, and directly removes the sequences marked as "Allergen" potential allergens; (2) IEDB-based screening for immune epitope homology: For peptides with a predicted activity score >0.9 and extremely high confidence, screening was performed using the IEDB platform. The host was strictly set to Human, and all sequences that showed positive results in T cell, B cell, and MHC ligand experiments were removed to eliminate the potential risk of cross-reactivity with human autoantigens; (3) MHC-II binding affinity depth assessment based on IEDB: The remaining sequences after homology exclusion were input into the MHC-II binding affinity prediction model of IEDB. The official reference set containing 27 high-frequency human leukocyte antigen (HLA) alleles was used for computational analysis. The MHC-II binding affinity was quantified by the median binding percentile rank of the ligand elution pattern prediction results from NetMHCIIpan.

[0041] Final Results: Following the funnel-style screening and evaluation described above, high-risk peptides that could easily induce adverse immune responses were excluded. Six protein-derived candidate peptides with low predicted immunogenicity and high angiotensin-converting enzyme (ACE) inhibition potential were identified: IPPLVVQK, PPGPPIPPLV, IPPLVVQQW, GVAKPNL, LCQGSGGIPPEK, and CQGSGGIPPEK (sequences SEQ ID NO: 1-6, respectively). These were selected as the subjects for subsequent molecular docking and molecular dynamics simulation analysis.

[0042] Example 2: Molecular docking verification of candidate ACE inhibitory peptides To elucidate the molecular-level interaction mode between peptides and ACE, semi-flexible molecular docking simulations were performed using MOE 2024.06 software. (1) Preparation of protein receptors and polypeptide ligands The crystal structure of the ACE protein (PDB ID: 1O86) was obtained from the RCSB PDB database. Dehydration, hydrogenation, charge and energy minimization were performed in MOE software. Zn was retained in the core catalytic region. 2+ Zn ions (Zn 701) and their coordinated water molecules. The active pocket was defined using the "Site Finder" module, locking the docking center in the area containing Zn. 2+ Within the regions of ions and key catalytic residues in the S1 and S2 pockets (Glu384, Ala354, Tyr523, His383, Gln281, etc.), the six previously screened candidate peptides were used to construct three-dimensional structures and minimize their energy, serving as ligands for molecular docking.

[0043] (2) Virtual screening and binding energy analysis Docking results showed that peptides LCQGSGGIPPEK and CQGSGGIPPEK had severe steric hindrance at their binding sites and were therefore eliminated. The remaining four candidate peptides (IPPLVVQK, PPGPPIPPLV, IPPLVVQQW, and GVAKPNL) had no steric conflict and successfully entered the active catalytic cavity of ACE. Statistical analysis showed that the docking binding energies (Scores) of the four peptides were concentrated between -14.17 kcal / mol and -16.63 kcal / mol, all exhibiting good binding affinity. Figure 1 Among them, the binding energies of PPGPPIPPLV and IPPLVVQQW are as low as -16.63 kcal / mol and -16.47 kcal / mol, respectively.

[0044] (3) Interaction mechanism and structure-activity relationship The semi-flexible docking interaction report reveals that all four peptides can undergo metal coordination or ion interaction with the core catalytic ion zinc (Zn7O1) and form a hydrogen bond network with surrounding key residues, competitively blocking substrate-enzyme binding. The specific binding characteristics of each peptide are as follows: Polypeptide IPPLVVQK: Its backbone oxygen atoms and core Zn 2+ Metal coordination and ionic bonds are formed (binding distance 2.00 Å, ionic bond binding energy -16.2 kcal / mol); the polar groups of the main side chain form multiple hydrogen bonds with key residues GLU 384 (3.29 Å), THR 282 (2.80 Å), GLN 281 (2.96 Å) and LYS 454 (2.71 Å) at the S1 / S2 subsites; the carbon atoms of the side chain form H-pi conjugation with HIS 353 (3.48 Å).

[0045] Peptide PPGPPIPPLV: Terminal carboxyl oxygen atom and Zn 2+ Stable metal coordination and ionic bonds are formed (distance 2.07 Å, ionic bond binding energy -14.8 kcal / mol); the N-terminal and main chain atoms form a stable dual interaction of hydrogen and ionic bonds with ASP 453, and form multiple hydrogen bond anchors with MET 450, GLU 376, HIS 513 and ALA 356; the side chain carbon atoms form H-pi interactions with the imidazole ring of HIS383 (3.84 Å).

[0046] Peptide IPPLVVQQW: The C-terminal oxygen atom penetrates deep into the bottom of the pocket, interacting with Zn. 2+It forms short-range metal coordination and ion interactions (distance 1.93 Å, ionic bond binding energy -17.6 kcal / mol); the polar side chain forms double hydrogen bonds with GLN 281 (distance 3.02 Å), and forms a dense network of hydrogen and ionic bonds with LYS 511, HIS 513, ASP 415, and ARG 522; the bulky indole ring of C-terminal tryptophan provides steric hindrance and forms a stable H-pi conjugation with HIS 387 (3.63 Å).

[0047] Peptide GVAKPNL: C-terminal carboxyl oxygen atom and Zn 2+ It forms dual metal coordination and ionic bonds (distances of 2.06 Å and 2.00 Å, with binding energies of -15.0 and -16.3 kcal / mol, respectively); the N-terminal amino group and sequence backbone form multiple hydrogen bonds with MET 450, SER422, and LYS 454; the side-chain carbon and nitrogen atoms form dual H-pi conjugation with the aromatic ring of HIS 353 (distances of 3.84 Å and 3.72 Å, respectively). Molecular docking results confirmed that the four specific sequence peptides screened in this invention can stably occupy the ACE catalytic active center through multidimensional non-covalent interactions such as metal chelation, hydrogen bond networks, and nonpolar conjugation, thus verifying their competitive inhibitory activity against ACE from a structural mechanism perspective.

[0048] Molecular dynamics simulations further confirmed that the binding system of the four peptides screened in this invention with the target protein ACE is extremely stable in simulated physiological water environment. The peptides can form stable low free energy complexes with the target site through continuous hydrogen bonding, which verifies their good targeting binding efficiency from a dynamic microscopic mechanism perspective.

[0049] Example 3: Solid-phase synthesis of novel peptides and verification of their in vitro ACE inhibitory activity The target candidate peptides screened based on the above-mentioned hybrid AI architecture and molecular simulation were artificially synthesized and their in vitro ACE inhibitory activity was verified. The specific operation steps are as follows: 1. Solid-phase synthesis and purity identification of the target peptide To obtain high-purity target peptides for in vitro activity evaluation, four protein-derived peptides (IPPLVVQK, PPGPPIPPLV, IPPLVVQQW, and GVAKPNL) with extremely high binding potential, screened and validated in Example 2, were artificially synthesized using a standard solid-phase synthesis method. The synthesized products were subsequently separated and purified by high-performance liquid chromatography (HPLC), and their molecular weights were precisely verified by HPLC-tandem mass spectrometry (LC-MS). The results showed that the purity of the four peptides obtained was ≥95% (specifically: IPPLVVQK 98.20%, PPGPPIPPLV 97.01%, IPPLVVQQW 96.27%, and GVAKPNL 98.87%). The lyophilized peptide powders were collected and stored at -20°C in the dark for later use.

[0050] 2. In vitro ACE inhibitory activity (IC50) 50 Value) determination To accurately assess the true inhibitory effect of the synthesized peptide on ACE catalytic activity in vitro, the classic hippuryl-L-histyl-L-leucine (HHL) substrate method was used for determination. The specific steps are as follows: (1) Preparation of polypeptide solutions: The above-mentioned lyophilized polypeptide powder was prepared into a series of polypeptide test solutions with mass concentration gradients of 0.05, 0.1, 0.5, 1.0, 1.5 and 3.0 mg / mL by using an appropriate amount of 0.1 M borate-borax buffer (BBS, pH 8.3, containing 0.3 M NaCl).

[0051] (2) Enzymatic reaction procedure: 100 μL (5 mmol / L) of HHL substrate solution and 40 μL of different concentrations of the test peptide solution were added sequentially to centrifuge tubes. After vortexing and mixing, the mixture was preheated in a constant temperature water bath at 37℃ for 5 min. Then, 10 μL of ACE enzyme solution (0.1 U / mL) was accurately added to start the reaction. After mixing, the mixture was precisely incubated in a 37℃ water bath for 60 min. After the reaction was completed, 250 μL (1 mol / L) hydrochloric acid solution was quickly added to denature and inactivate the enzyme with strong acid and terminate the reaction. 1 mL of ethyl acetate was added to each of the above reaction tubes, and the mixture was vigorously shaken in a vortex mixer for 30 s to extract the enzyme completely. After centrifuging the mixture at 4000 rpm for 10 min, 750 μL of the upper organic phase liquid was carefully aspirated and transferred to a clean, heat-resistant glass container. The container was placed in a 60℃ oven and baked until the ethyl acetate was completely evaporated. Add 3 mL of deionized water to the dried test tube and vortex vigorously to completely redissolve the hippuric acid on the tube wall.

[0052] (3) Product detection and inhibition rate calculation: The absorbance of hippuric acid, the enzyme-catalyzed reaction product, was detected at a wavelength of 228 nm using a UV spectrophotometer. The specific formula for calculating the ACE inhibition rate is as follows: Inhibition rate calculation

[0053] A represents the absorbance of the sample group; B represents the absorbance of the control group; and C represents the absorbance of the blank group.

[0054] (4) Calculate IC 50 Value: Based on the ACE inhibition rate of the peptide at different concentrations, the half-maximal inhibitory concentration (IC50) was calculated using a standard four-parameter logistic regression model in R. 50 value).

[0055] 3. Experimental Results and Activity Analysis All four protein-derived peptides exhibited highly significant dose-dependent ACE inhibitory effects. Figure 2 , Figure 3 The coefficients of determination (R²) for all fitted curves. 2 The values ​​are all greater than 0.93, demonstrating extremely high goodness of fit and data reliability. Specific data are as follows: The maximum inhibition rate of the peptide IPPLVVQK (SEQ ID NO:1) reached 83.14 ± 0.90%, IC50 50 The value was 0.541 mg / mL (0.606 mM), R 2 = 0.954; The maximum inhibition rate of the peptide PPGPPIPPLV (SEQ ID NO:2) was 88.37 ± 0.47%, IC50 was 100%. 50 The value was 0.573 mg / mL (0.583 mM), R 2 = 0.985; The maximum inhibition rate of the peptide IPPLVVQQW (SEQ ID NO:3) was 92.09 ± 0.47%, IC50 was 92.09 ± 0.47%. 50 The value was 0.312 mg / mL (0.289 mM), R 2 = 0.932; The maximum inhibition rate of the peptide GVAKPNL (SEQ ID NO:4) reached 94.88 ± 0.47%, IC50 50 The value was 0.103 mg / mL (0.148 mM), R 2 = 0.996.

[0056] In vitro experimental results fully demonstrate that the four peptides exhibit extremely high antihypertensive efficacy at the millimolecular (mM) level, especially the IC50 of the core sequence GVAKPNL. 50 With a concentration as low as approximately 0.148 mM, these novel peptides exhibit extremely strong target inhibitory affinity. They show promising potential for application in the preparation of antihypertensive drugs.

[0057] Appendix 1, Explanation of the Sequence List: SEQ ID NO:1: One of the six polypeptides obtained after preliminary screening in Example 1, namely IPPLVVQK; SEQ ID NO:2: One of the six polypeptides obtained after preliminary screening in Example 1, namely PPGPPIPPLV; SEQ ID NO:3: One of the six polypeptides obtained after preliminary screening in Example 1, namely IPPLVVQQW; SEQ ID NO:4: One of the six polypeptides obtained after preliminary screening in Example 1, namely GVAKPNL; SEQ ID NO:5: One of the six polypeptides obtained after preliminary screening in Example 1, namely LCQGSGGIPPEK; SEQ ID NO:6: One of the six polypeptides obtained after preliminary screening in Example 1, namely CQGSGGIPPEK.

[0058] Appendix 2, Terminology Explanation: 1. ACE inhibitory peptides: These are polypeptides that can inhibit the activity of angiotensin-converting enzyme (ACE). ACE is a key enzyme in regulating blood pressure; inhibiting its activity helps dilate blood vessels and lower blood pressure.

[0059] 2. Virtual enzymatic digestion: Using computer algorithms to simulate the enzymatic digestion process of protein substrates by proteases and predict the possible polypeptide fragments produced, without the need for actual wet experiments.

[0060] 3. Cascade simulation: refers to simulating the enzymatic hydrolysis process under different enzyme or environmental conditions in stages according to the actual physiological or industrial process, such as industrial hydrolysis → gastric digestion → intestinal digestion.

[0061] 4. mAHTPred: A machine learning model that predicts antihypertensive activity based on peptide sequence features (such as amino acid composition, dipeptide composition, etc.), belonging to meta-predictors.

[0062] 5. pLM4ACE: An ACE inhibitory peptide prediction model built on a large protein language model (such as ESM-2), supporting both SVM and MLP prediction modes.

[0063] 6. SVM (Support Vector Machine): A supervised machine learning algorithm used for classification or regression analysis. In this document, it is used to predict whether a peptide has ACE inhibitory activity.

[0064] 7. MLP (Multilayer Perceptron): An artificial neural network structure that includes an input layer, hidden layers, and an output layer, suitable for nonlinear classification problems.

[0065] 8. Immunological safety assessment: A series of methods to assess whether candidate peptides may trigger an immune response (such as allergy or autoimmune response), including sensitization, epitope homology, MHC-II binding affinity, etc.

[0066] 9. MHC-II binding affinity: refers to the strength of binding between a peptide and major histocompatibility complex class II molecules. It is an important indicator for assessing whether a peptide may activate CD4+ T cells and trigger an immune response.

[0067] 10. Missed Cleavages (MC): The number of sites that the protease is allowed to fail to completely cleave during enzyme digestion, reflecting the phenomenon of incomplete cleavage in real enzyme digestion.

[0068] 11. GRAVY (Overall Average Hydrophilicity Index): Used to assess the hydrophilicity or hydrophobicity of peptides or proteins. Negative values ​​indicate hydrophilicity, and positive values ​​indicate hydrophobicity. ACE inhibitory peptides typically exhibit moderate hydrophilicity or hydrophobicity.

[0069] 12. IEDB (Immune Epitope Database): A public database for predicting and querying immune epitope information for T cells, B cells, and MHC ligands.

[0070] 13. AllerTOP v.2.1: An online tool for predicting whether a protein or peptide is an allergen, based on the k-nearest neighbor algorithm and amino acid E-descriptors features.

[0071] 14. NetMHCIIpan: A model for predicting the affinity of peptides for binding to MHC-II molecules, using an eluted ligand model for training and prediction.

[0072] 15. Molecular docking: Predicting the binding mode, binding site, and binding energy between peptides (ligands) and target proteins (such as ACE) through computer simulation.

[0073] 16. IC 50 (Half-maximal inhibitory concentration): refers to the concentration of the polypeptide required to inhibit 50% of enzyme activity. The lower the value, the stronger the inhibitory activity.

[0074] 17. HHL substrate method: A classic method for in vitro determination of ACE activity. The substrate is hippuryl-histyl-leucine (HHL), which is cleaved by ACE to generate hippuric acid. The inhibition rate is calculated by detecting the hippuric acid content.

Claims

1. A protein-derived ACE-inhibiting peptide, characterized in that, The amino acid sequence of the ACE inhibitory peptide is shown in SEQ ID NO:

4.

2. A pharmaceutical composition, characterized in that, The pharmaceutical composition comprises the ACE inhibitory peptide of claim 1 and a pharmaceutically acceptable carrier or excipient.

3. The use of the ACE inhibitory peptide of claim 1 or the pharmaceutical composition of claim 2 in the preparation of antihypertensive drugs.

Images