Method of designing molecular recognition peptides based on a structure reconstruction and iterative evolution framework

Abstract

The application discloses a method for designing molecular recognition peptides based on RIE strategy of structural reconstruction and iterative evolution framework, which can be used for reconstructing and optimizing high-affinity MRPs from natural protein-ligand structures. The method identifies non-continuous but functionally critical residues in the natural protein-ligand interface, and reconstructs them into a spatially coordinated binding motif with the help of structural prediction tools such as AlphaFold2; through multiple rounds of iterative optimization, the peptide sequence that minimizes on the basis of retaining binding ability is obtained; the obtained peptide sequence further enters the rational evolution stage, combines the predicted peptide-ligand binding conformation, and realizes the precise regulation of key residues through site-directed mutation and local sequence optimization to enhance the binding affinity, improve the solubility, and stabilize the conformation structure. The strategy effectively copes with many challenges faced in traditional top-down design, and provides an efficient and generalizable solution for high-affinity compact recognition peptide design for small molecule recognition.

Description

Technical Field

[0001] This invention belongs to the technical field, specifically relating to a RIE strategy for designing molecular recognition peptides based on a structural reconstruction and iterative evolution framework. Background Technology

[0002] Small molecule recognition, due to its precise interactions and regulation at the molecular level, is of great significance in life sciences, drug discovery, food safety, and environmental monitoring. Traditional recognition agents such as antibodies, aptamers, and molecularly imprinted polymers (MIPs) have been widely used for small molecule detection, but limitations remain, including complex screening processes, long preparation cycles, and insufficient stability. In recent years, significant progress has been made in the rational design and synthesis of artificial molecules. Among these, molecular recognition peptides (MRPs), as an emerging receptor type, have shown the potential to become a next-generation recognition platform due to the high specificity of natural receptors and the stability and tunability gained through artificial optimization. With the development of technologies such as computational design, directed evolution, and solid-phase synthesis, MRPs are gradually being applied in cutting-edge fields such as precision medicine, intelligent sensing, and bioengineering.

[0003] Currently, MRP design strategies are mainly divided into two categories: top-down and de novo design. De novo design has made progress in recent years, driven by teams such as Baker, but it has high operational barriers and computational costs. The top-down strategy, on the other hand, analyzes the protein-ligand cocrystal structure, extracts key residues, and constructs simplified short peptide sequences. This approach inherits the binding properties of the parent protein while significantly reducing computational and experimental burden. This biomimetic method has become a viable option for researchers across disciplines. With the rapid growth of structure databases (such as PDB and PDB Bind, which contain over 100,000 protein-ligand structures), structure-driven recognition peptide design has a rich data foundation, providing opportunities for the rapid discovery of high-affinity peptide molecules.

[0004] Although a large database of protein-ligand cocrystal structures is available, top-down peptide design remains relatively limited in practical applications. Current computational strategies often only partially utilize natural protein structures, typically generating candidate sequences through direct simulation, domain truncation, or oversimplified structural modeling. For example, Woolfson et al., drawing inspiration from the structure of G protein-coupled receptors (GPCRs), designed peptide-based differential sensors for disease diagnosis from complex mixtures; other researchers extracted key binding motifs from olfactory receptors to construct short peptide probes capable of detecting small gaseous molecules. Similar studies have also utilized natural high-affinity proteins as templates to generate peptide molecules with binding capabilities. These strategies, while preserving key recognition motifs, can increase binding probability at a lower computational cost.

[0005] However, these methods often overlook the contribution of overall protein folding to conformational stability, relying primarily on side-chain interactions for molecular recognition while failing to adequately consider the importance of multi-site collaborative recognition. Furthermore, binding pockets in natural proteins, especially regions used for binding hydrophobic small molecules (such as lipophilic drugs and aromatic compounds), are easily exposed to solution when these hydrophobic residues are extracted into short peptides, leading to poor solubility and insufficient biocompatibility. These issues highlight the challenges of current top-down methods in designing high-affinity MRPs and underscore the urgent need to develop innovative strategies to expand the applicability of this approach and promote the widespread development of small molecule MRPs.

[0006] This study proposes a three-step design framework—the RIE strategy (Resampling-Iteration-Evolution)—for reconstructing and optimizing high-affinity MRPs from native protein-ligand structures. This method identifies discontinuous but functionally critical residues at the native protein-ligand interface and reconstructs them into spatially coordinated binding motifs using structure prediction tools such as AlphaFold2. Through multiple rounds of iterative optimization, peptide sequences that minimize binding capacity are obtained. Subsequently, the resulting peptides undergo a rational evolution phase, combining the predicted peptide-ligand binding conformation with site-directed mutagenesis and local sequence optimization to precisely regulate key residues, thereby enhancing binding affinity, improving solubility, and stabilizing the conformational structure. This strategy effectively addresses many challenges faced in traditional top-down design, including the loss of multi-site recognition, poor solubility due to hydrophobic residue exposure, and conformational instability caused by domain truncation. Overall, the RIE strategy of this invention provides an efficient and scalable solution for designing high-affinity, compact recognition peptides for small molecule recognition. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of the existing technology, the present invention provides a method for designing molecular recognition peptides based on the RIE strategy of structural reconstruction and iterative evolution framework, which provides an efficient and scalable solution for the design of high-affinity compact recognition peptides for small molecule recognition.

[0008] Technical solution: To achieve the above objectives, the technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a molecular recognition peptide, wherein the molecular recognition peptide is any one of the following: a. Molecular recognition peptide I that specifically recognizes the small molecule ligand FA, the amino acid sequence of which is shown in SEQ ID NO.1; b. Molecular recognition peptide II that specifically recognizes the small molecule ligand T3, the amino acid sequence of which is shown in SEQ ID NO.2; c. Molecular recognition peptide III that specifically recognizes the small molecule ligand CHD, with the amino acid sequence shown in SEQ ID NO.3.

[0009] In a second aspect, the present invention provides the application of the aforementioned molecular recognition peptide in the preparation of products for detecting small molecule ligands, wherein the small molecule ligands include FA, T3, or CHD.

[0010] In a third aspect, this invention provides a method for designing molecular recognition peptides based on a structure reconstruction and iterative evolutionary framework (RIE) strategy, comprising the following steps: S1. Resampling: Small molecule ligands and their high-affinity natural receptor protein complexes are obtained from the database. Discontinuous residues, including key binding residues and secondary structure support elements, are extracted and reconstructed into spatially coordinated binding motifs using structure prediction tools. These motifs are denoted as the first-generation molecular recognition peptide MRP1. S2, Iteration: By integrating MD simulation and AlphaFold2 prediction, several rounds of iterative optimization are performed to obtain the molecular recognition peptide MRPn that retains the binding ability. S3. Evolution: Based on the predicted peptide-ligand binding conformation, the minimized molecular recognition peptide MRPn is optimized by site-directed mutagenesis to obtain a target molecular recognition peptide with high affinity.

[0011] Optionally, in some embodiments of the present invention, in step S1, the small molecule ligand includes FA, T3, or CHD; The natural receptor protein complex of FA is the folic acid receptor α with which it is complexed. The natural receptor protein complex of T3 is the thyroid hormone receptor TR, which is complexed with thyroxine. The natural receptor protein complex of CHD is a fatty acid-binding protein that binds to it.

[0012] Optionally, in some embodiments of the present invention, in step S1, the key binding residue is a residue that forms hydrogen bonds and hydrophobic interactions.

[0013] Optionally, in some embodiments of the present invention, in step S1, the reconstructing of the binding motif into a spatially coordinated binding motif using a structure prediction tool specifically includes: constructing an artificial binding pocket using AlphaFold modeling, wherein the artificial binding pocket retains the structural features of the natural receptor binding site in space, denoted as the first-generation molecular recognition peptide MRP1.

[0014] Preferably, the reconstruction specifically includes: swapping or reversing the positions of fragmented sequence segments to preserve the spatial relative positions of the combined pockets, so as to ensure that the new pocket is close to the original pocket. The closeness means that the spatial point set of the spatial structure of the two pockets satisfies RMSD < 3 Å.

[0015] Preferably, the spatially coordinated binding motif specifically refers to the reconstructed sequence folding being able to form a three-dimensional conformation that conforms to the characteristics of a typical peptide structure.

[0016] Optionally, in some embodiments of the present invention, step S2, the iteration specifically includes: The dynamic interactions between MRP1 and the target molecule were captured using all-atom MD simulation, and the binding interface was resampled during the process. Subsequently, a new key pocket was extracted from the dynamic stable state of the MRP1-ligand complex, and the second-generation molecular recognition peptide MRP2 was predicted using AlphaFold2. This process was repeated until the (n+1)th generation molecular recognition peptide MRPn failed to exhibit stable binding in MD simulations exceeding 100 ns, thus obtaining the peptide sequence with the fewest residues, which is the minimized molecular recognition peptide MRPn.

[0017] Preferably, the failure to exhibit stable binding means that the RMSD fluctuation is greater than 0.5, or the binding energy is greater than 0.

[0018] Optionally, in some embodiments of the present invention, in step S3, the site-directed mutation includes: By introducing single-point mutations and analyzing the dynamic binding stability and binding energy of the peptide in aqueous solution using molecular dynamics (MD) analysis, we can evaluate its affinity for screening target molecule recognition peptides (MRPs).

[0019] In a fourth aspect, the present invention provides an apparatus for designing molecular recognition peptides based on a structural reconstruction and iterative evolutionary framework (RIE) strategy, comprising: The acquisition module is used to acquire small molecule ligands and their high-affinity natural receptor protein complexes from the database, extract discontinuous residues including key binding residues and secondary structure support elements, and reconstruct them into spatially coordinated binding motifs using structure prediction tools, denoted as the first-generation molecular recognition peptide MRP1. The iterative module is used to perform several rounds of iterative optimization by integrating MD simulation and AlphaFold2 prediction to obtain the minimum molecular recognition peptide MRPn that retains binding ability; An evolution module is used to combine the predicted peptide-ligand binding conformation with site-directed mutagenesis to optimize the peptide sequence of the minimized molecular recognition peptide MRPn, thereby obtaining a target molecular recognition peptide with high affinity.

[0020] In a fifth aspect, the present invention provides an electronic device comprising: processor; A memory having executable code stored thereon, which, when executed by the processor, causes the processor to perform the method as described in any of the preceding descriptions.

[0021] In a sixth aspect, the present invention provides a computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method described in any of the preceding claims. Beneficial effects: The present invention provides the following advantages compared to the prior art: This invention presents a method for designing molecular recognition peptides based on a Reconstructive Evolution (RIE) strategy within a structural reconstruction and iterative evolution framework. This method effectively addresses many challenges faced by traditional top-down design, including the loss of multi-site recognition capabilities, poor solubility due to exposure of hydrophobic residues, and conformational instability caused by domain truncation. Overall, the RIE strategy of this invention provides an efficient and scalable solution for designing high-affinity, compact recognition peptides for small molecule recognition. Attached Figure Description

[0022] Figure 1 A top-down design and rational co-evolution strategy driven by structural restructuring.

[0023] Figure 2 The first round of design models for each target ligand: (A, C, E) crystal structures of FA, T3, and CHD bound within their respective receptor binding pouches; (B, D, F) superposition of extracted discontinuous binding residues with AlphaFold2-predicted recombinant peptide models.

[0024] Figure 3 Simulations of FA_MRPs generated for resampling and iterative design process: (A, B) Dynamic RMSD of FA_MRP1-5 during a 100 ns simulation; (C) Binding energy of FA_MRP1 and MRP5 calculated by MM / PBSA.

[0025] Figure 4 The changes in RMSF of each residue of FA_MRP1~5 during the MD process.

[0026] Figure 5 Simulations of T3_MRPs generated for resampling and iterative design process: (A, B) Dynamic RMSD of T3_MRP1-4.1 during a 100 ns simulation; (C) Binding energy of T3_MRP1 and MRP4.1 calculated by MM / PBSA.

[0027] Figure 6Simulations of CHD_MRPs generated during resampling and iterative design processes. (A, B, C) Dynamic RMSD of CHD_MRP1-3.3 during a 100 ns simulation; (D, E, F) Binding energies of CHD_MRP1 and MRP3.3 calculated by MM / PBSA.

[0028] Figure 7 For the evolution of MRPs. (A) Heatmap of the binding stability of each residue mutation in FA_MRP5, (C) T3_MRP4.1, and (E) CHD_MRP3.3; (B) Structural comparison of FA_MRP5 and FA_MRP5.4, (D) T3_MRP4.1 and T3_MRP4.1.1, and (F) CHD_MRP3.3 and CHD_MRP3.3.4.

[0029] Figure 8 Classification of (A) MRP residues and (BF) MRP5 mutant.

[0030] Figure 9 MD analysis of the mutated MRP and the target. (A, B) Dynamic changes of RMSD of FA_MRP5.1 to MRP5.5 in a 100 ns simulation; (C) Binding energy of FA_MRP5.1 to MRP5.5 with FA.

[0031] Figure 10 MD analysis of the mutated MRP and the target. (A, B) Dynamic changes of RMSD of T3_MRP4.1.1 to MRP4.1.5 in a 200 ns simulation; (C) Binding energy of T3_MRP4.1.1 to MRP4.1.5 with T3.

[0032] Figure 11 MD analysis of the mutated MRP and the target. (A, B) Dynamic changes of RMSD of T3_MRP4.1.1-MRP4.1.5 in a 200 ns simulation; (C) Binding energy of T3_MRP4.1.1-MRP4.1.5 to T3.

[0033] Figure 12 Mass spectrometry analysis of MRPs and small molecules in a 1:1 mixture. Mass spectra of FA_MRP5.4 (A), CHD_MRP3.3.4 (C), and T3_MRP4.1.1 (F) under ligand-free conditions (200 μM); mass spectra of FA_MRP5.4 (B), CHD_MRP3.3.4 (D), and T3_MRP4.1.1 (E) under ligand-co-existing conditions (all at 200 μM).

[0034] Figure 13To characterize the binding affinity of MRPs to target small molecules using ITC. (A) ITC results of FA_MRP5.4, (B) T3_MRP4.1.1, and (C) CHD_MRP3.3.4; (D) Thermodynamic parameters measured by ITC during the binding of MRPs to small molecules; (E) KL measured by ITC. d .

[0035] Figure 14 Specificity tests for MRPs. (A) Binding behavior of FA_MRP5.4 with FA, MTX and their mixtures; (B) Binding behavior of T3_MRP4.1.1 with T3, T4, LT4, T2 sodium and their mixtures; (C) Binding behavior of CHD_MRP3.3.4 with CHD, DCA, LCA, UDCA and their mixtures. Detailed Implementation

[0036] Although a large database of protein-ligand cocrystal structures is available, top-down peptide design remains relatively limited in practical applications. Current computational strategies often only partially utilize natural protein structures, typically generating candidate sequences through direct simulation, domain truncation, or oversimplified structural modeling. For example, Woolfson et al., drawing inspiration from the structure of G protein-coupled receptors (GPCRs), designed peptide-based differential sensors for disease diagnosis from complex mixtures; other researchers extracted key binding motifs from olfactory receptors to construct short peptide probes capable of detecting small gaseous molecules. Similar studies have also utilized natural high-affinity proteins as templates to generate peptide molecules with binding capabilities. These strategies, while preserving key recognition motifs, can increase binding probability at a lower computational cost.

[0037] However, these methods often overlook the contribution of overall protein folding to conformational stability, relying primarily on side-chain interactions for molecular recognition while failing to adequately consider the importance of multi-site collaborative recognition. Furthermore, binding pockets in natural proteins, especially regions used for binding hydrophobic small molecules (such as lipophilic drugs and aromatic compounds), are easily exposed to solution when these hydrophobic residues are extracted into short peptides, leading to poor solubility and insufficient biocompatibility. These issues highlight the challenges of current top-down methods in designing high-affinity MRPs and underscore the urgent need to develop innovative strategies to expand the applicability of this approach and promote the widespread development of small molecule MRPs.

[0038] This study proposes a three-step design framework—the RIE strategy (Resampling-Iteration-Evolution)—for reconstructing and optimizing high-affinity MRPs from native protein-ligand structures. This method identifies discontinuous but functionally critical residues at the native protein-ligand interface and reconstructs them into spatially coordinated binding motifs using structure prediction tools such as AlphaFold2. Through multiple rounds of iterative optimization, peptide sequences that minimize binding capacity are obtained. Subsequently, the resulting peptides undergo a rational evolution phase, combining the predicted peptide-ligand binding conformation with site-directed mutagenesis and local sequence optimization to precisely regulate key residues, thereby enhancing binding affinity, improving solubility, and stabilizing the conformational structure. This strategy effectively addresses many challenges faced in traditional top-down design, including the loss of multi-site recognition, poor solubility due to hydrophobic residue exposure, and conformational instability caused by domain truncation. Overall, the RIE strategy provides an efficient and scalable solution for designing high-affinity, compact recognition peptides for small molecule recognition.

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments. The present invention can be better understood from the following embodiments. However, those skilled in the art will readily understand that the specific material ratios, process conditions, and results described in the embodiments are for illustrative purposes only and should not, and will not, limit the present invention as described in detail in the claims.

[0040] Example: RIE strategy to construct high-affinity small molecule MRPs from natural receptors The RIE (Resampling-Iteration-Evolution) strategy of this invention is specifically implemented in three steps.

[0041] The first step is resampling. Figure 1 A): Starting with ligands and their high-affinity natural receptor protein complexes, these complexes were obtained from screening results in the RCSB PDB database.

[0042] The abundant protein-ligand binding information in databases provides feasibility for top-down MRP design. Structural mapping of binding sites reveals that the residues necessary for forming the binding pocket are often distributed in the tertiary structure space, rather than existing continuously in the primary sequence. This phenomenon reveals the limitations of relying solely on the native sequence to directly extract the optimal MRP. To address this issue, the three-dimensional structure of the binding pocket was analyzed, and discontinuous residues, including key binding residues and secondary structural support elements, were extracted. An artificial binding pocket was constructed using AlphaFold modeling. This artificial pocket, spatially preserving the structural features of the natural receptor binding site, was named MRP1 and served as the basis for subsequent iterations and evolution.

[0043] The second iteration (Iteration, Figure 1 B) The goal is to obtain the MRP with the fewest residues.

[0044] MRP1 obtained by resampling typically contains more than one hundred residues, making its chemical synthesis challenging. To address this, all-atom molecular dynamics (MD) simulations were used to capture the dynamic interactions between MRP1 and the target molecule, and the binding interface was resampled during this process. Subsequently, new critical pockets (composed of discontinuous residues) were extracted from the dynamic stable state of the MRP1-ligand complex, and the next-generation peptide MRP2 was predicted using AlphaFold2. This iterative process was repeated by integrating MD simulations and AlphaFold2 predictions until MRPn maintained a relatively optimal binding energy while minimizing the number of residues. When stable binding was not observed in MD simulations exceeding 100 ns, MRP(n-1) was redesigned, and the iterative process was repeated until an optimized MRP with desirable properties was obtained.

[0045] The third step of evolution (Evolution) Figure 1 C) Further adjust the properties of the MRP obtained in the previous step, such as affinity and hydrophilicity.

[0046] The high-affinity MRP obtained through the iteration step is close to the optimal short peptide obtained from the pocket, but because it originates from a natural protein, its binding affinity may not yet reach the theoretical maximum. This inherent limitation is difficult to completely eliminate even after recombination modification. Furthermore, the exposure of the hydrophobic pocket in the natural protein structure may lead to poor water solubility, thus limiting its downstream applications. To overcome these shortcomings, single-point mutations were introduced to optimize the MRP sequence, and the impact of the mutations on its stability was evaluated. Several MRP variants were designed based on the mutation heatmap, and their dynamic binding stability and binding energy in aqueous environment were analyzed using MD (Mean Differential Mutation) to evaluate their affinity. Finally, the most promising MRP was selected for subsequent experimental validation.

[0047] The materials and experimental methods used in the following embodiments of the present invention are as follows.

[0048] (1) Reagents and materials The peptides were custom-synthesized by Shanghai Qiangyao Biotechnology Co., Ltd., with a lyophilized purity >95%. The resulting lyophilized powder was dissolved in purified water to prepare a high-concentration stock solution (0.5 mg / mL) and stored at 4°C. Other reagents were at least analytical grade. Folic acid (FA), methotrexate (MTX), triiodothyronine (T3), levothyroxine (T4), levothyroxine sodium (LT4), 3,5-diiodothyronine sodium (T2 Sodium), cholic acid (CHD), deoxycholic acid (DCA), lithocholic acid (LCA), and ursodeoxycholic acid (UDCA) were purchased from Maclean's (Shanghai). The experimental water was from Hangzhou Wahaha Group.

[0049] (2) Mass spectrometry analysis Lyophilized peptides and small molecules were dissolved in a 50% acetonitrile / water (1:1 volume ratio) solution containing 75 mM ammonium acetate. Three peptides and their mixtures with small molecules (1:1 concentration, 200 μM each) were prepared separately, and the concentrations were adjusted according to experimental requirements. All solutions were filtered through a 0.22 μm filter membrane and incubated at room temperature for 30 minutes before analysis. Electrospray ionization (ESI) was used on a QTRAP 4500 mass spectrometer (SCIEX, USA) in positive ion mode, with scan mode Q1, scan rate 200 Da / s, and cluster removal voltage set to 80 V.

[0050] (3) Isothermal titration calorimetry (ITC) determination The thermodynamic binding parameters of MRP and small molecules were determined using a VP-ITC isothermal titration calorimeter (MicroCal Inc., MA, USA).

[29] The three experimental conditions were as follows: FA and FA_MRP5.4 were dissolved in anhydrous DMSO, with final concentrations of 0.1 mM and 1 mM, respectively; T3 and T3_MRP4.1.1 were also dissolved in DMSO, with concentrations of 5 mM (titrant) and 0.25 mM (receiver), respectively; CHD and CHD_MRP3.3.4 were dissolved in a DMSO mixture containing 4% water, with concentrations of 0.1 mM (CHD) and 1 mM (peptide). All solutions were filtered through a 0.22 μm filter before titration.

[0051] The titration process was maintained at 25 °C. The initial injection was 2 μL, followed by 21–26 automated injections of 10 μL each, with a stirring speed of 394 rpm. Each injection lasted 20 seconds, with an interval of 300–420 seconds between injections to ensure system equilibrium. Background titration was performed using DMSO as a blank, and the heat of dilution was subtracted. The final calculated binding parameters included: binding stoichiometry (n), K... d Value, enthalpy change (ΔH), entropy change (ΔS), and free energy change (ΔG).

[0052] (4) Specificity analysis The synthesized RhB-MRP was mixed with NH2-MIL-88(Fe) in 0.2 M acetate-sodium acetate buffer (pH 5.0), with a total volume of 0.5 mL. After standing for a period of time, the target molecule was added and incubated for 15 min, followed by measurement of its fluorescence spectrum. The excitation wavelength of the fluorescence spectrometer was set to 365 nm, the recorded emission spectrum range was 400-650 nm, and the bandwidth between the excitation and emission light was 5.0 nm. The fluorescence intensity (IL) of NH2-MIL-88(Fe) at 450 nm was recorded and compared. 450 The fluorescence intensity of RhB-MRP at 580 nm (I) 580 ).

[0053] (5) Simulation methods and parameters Molecular dynamics simulation parameters: Molecular dynamics simulations of peptide-small molecule interactions were performed using GROMACS software (version 2022.3, http: / / www.gromacs.org / ) with CUDA acceleration. Ligand molecules were constructed in Gaussian09 (C.01) using the DFT / B3LYP method and the 6-311G** basis set, and RESP charges were fitted using the Anttecamber tool. All molecular dynamics (MD) simulations were performed using the AMBER99SB-ILDN force field in conjunction with the TIP3P explicit water model. The system was placed in a cubic simulation box, with the protein-ligand complex at least 1 nm from the box boundary. The remaining space was filled with TIP3P water molecules, and an appropriate amount of Na was added. + and Cl - Ions were used to achieve electroneutrality. Energy minimization was performed using the steepest descent method until the maximum force was less than 100 kJ / mol·nm. The system was then equilibrated for 1 ns under the NVT ensemble and then for 5 ns under the NPT ensemble. The NVT simulation used a V-rescale thermosphere to maintain the temperature at 300 K, while the NPT simulation used Berendsen pressure coupling to maintain the pressure at 1 bar. Both phases used a 2 fs time step. After equilibration, the last frame structure was used as the starting point to impose positional constraints on the ligand and the simulation continued for 1 ns to achieve structural relaxation. Subsequently, a 100 ns free dynamic simulation was performed on the peptide and ligand system (without constraints). The RMSD of the main chain and ligand was calculated using Gromacs by analyzing the trajectory file, and visualization analysis was performed using PyMol (v4.6) and VMD 1.4.7. Graphs were plotted using Xmgrace (Grace-5.1). The binding energy was calculated using the Poisson-Boltzmann surface area (MM / PBSA) method, with the help of the gmx_MMPBSA tool, and with reference to the method of Valdes-Tresanco et al.

[27] The binding energy was assessed based on trajectory data from the 10 ns post-simulation period, taking into account the stability of the complex structure during this timeframe. The analysis included: relative binding energy, van der Waals energy, charge interaction energy, polar solvation energy, and SASA (solvent-accessible surface area) energy.

[0054] Mutation binding stability calculation: The binding stability of each MRP mutant with the small molecule complex was analyzed using the Calculate Mutation Energy (Binding) module in BIOVIA Discovery Studio 2023.

[28] The MRP structure was set as "input molecule," the ligand as "ligand chain," and the mutation mode as "single-point mutation to 20 amino acids." Other parameters were set as follows: dielectric constant 10, solvent dielectric constant 80, maximum mutation number 25, maximum structure number 10, and other parameters remained at their default values. After calculation, the mutation definition for binding energy change was: (a) mutation with stable binding (energy ≤ 0 kcal / mol); (b) mutation with unstable binding (energy > 0 kcal / mol), thus determining the effect of mutations at specific sites on binding stability.

[0055] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.

[0056] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0057] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0058] Example 1: Resampling of binding proteins like Figure 1 The resampling strategy shown in A specifically starts with ligands and their high-affinity native receptor protein complexes, which are derived from the screening results of the RCSB PDB database.

[0059] This study selected three small molecules as recognition targets: FA, T3, and CHD. FA is an essential vitamin that plays a crucial role in cell division and DNA synthesis, especially during pregnancy, where it is vital for fetal development and can effectively prevent neural tube defects. T3 is an important hormone in the human body that regulates energy metabolism and various physiological functions; home monitoring of its free T3 content has been used for the auxiliary diagnosis of thyroid diseases. CHD is one of the main bile acids, and its free state detection is of great value for liver disease screening. Current recognition receptors and corresponding detection methods for FA, T3, and CHD still face challenges such as the inability to distinguish between structural variants (e.g., similar compounds of FA and CHD) or bound and free states (e.g., T3). Therefore, there is an urgent need for recognition elements with higher affinity and specificity to meet the application requirements of rapid detection devices. From a design perspective, T3 and FA are more flexible and polar than commonly used small molecule ligands, which also presents a greater challenge to the design of recognition peptides. Figure 2 A, 2C).

[0060] First, natural protein receptors with high affinity for the target small molecule were obtained from structural databases (specific receptors capable of high affinity binding were selected). For FA, the folate receptor α (PDB ID: 4LRH, resolution 2.80 Å), which is complexed with FA, was selected; this structure was reported by Chen et al.; the receptor for T3 was the thyroid hormone receptor TR (PDB ID: 1XZX, resolution 2.50 Å), which is complexed with thyroxine; and the recognition receptor for CHD was the fatty acid-binding protein (FABP, PDB ID: 2QO6, resolution 1.90 Å) that binds to CHD. These receptors are all involved in the uptake, transport, and regulation of small molecules in vivo and belong to natural functional proteins.

[0061] The FA and its receptor binding region form a three-dimensional, deeply embedded open pocket, and the key residues at the binding site are highly conserved across different subtypes. By extracting its key hydrogen-bonding residues (residues that form hydrogen bonds) and other interacting residues (hydrophobic residues, etc.), and while preserving its helical conformation as much as possible, a spatially coordinated discontinuous sequence fragment was obtained. Figure 2 A, sequence A in Table 1). To ensure that the spatial configuration of the newly constructed sequence is close to the natural binding pocket, the extracted sequence was adjusted (the fragmented sequence fragments were swapped or reversed to preserve the spatial relative position of the binding pocket, so as to ensure that the new pocket is close to the original pocket), resulting in B (Table 1).

[0062] To assess the degree to which the reconstructed pocket preserves the three-dimensional structure of the natural binding pocket, the nearest-neighbor root-mean-square deviation (NN-RMSD) method, based on spatial proximity, was used for quantitative comparison. This method does not rely on residue numbering or sequence order, but rather directly matches based on the distribution characteristics of atoms in three-dimensional space.

[0063] The specific calculation process is as follows: (1) First, extract the key residues involved in ligand binding from the natural structure and select its Cα atom or all heavy atoms as the spatial point set of the original pocket; at the same time, extract the atoms of the corresponding pocket region from the reconstructed structure to construct the spatial point set of the new pocket. (2) Perform overall rigid alignment on the two sets of points to eliminate the effects of overall translation and rotation; (3) For each atom point in the original pocket, search for the atom with the closest spatial distance in the reconstructed pocket as the matching point; (4) Calculate the Euclidean distance between all nearest neighbor pairs and calculate the root mean square deviation (RMSD) as a quantitative indicator of the difference in spatial structure between the two pockets. When it is less than 3 Å, it is considered to have good three-dimensional spatial overlap.

[0064] NN-RMSD analysis based on nearest neighbor matching shows that the NN-RMSD value of the reconstructed pocket predicted by the new sequence obtained by FA combined with pocket extraction is 2.6 Å, and the reconstructed pocket and the naturally bound pocket have good consistency in spatial configuration. Figure 2 B).

[0065] The ligand-binding structure of T3 and its receptor TR is located in the C-terminal ligand-binding domain (LBD). Crystal structures show that T3 is completely enclosed within the binding cavity of the LBD core region. Figure 2 C). The constructed T3_MRP1 sequence (sequence C in Table 1) shows good consistency with the NN-RMSD value of the 1XZX binding pocket after AlphaFold2 prediction, with a spatial configuration of 1.8 Å. Figure 2 D), indicating that the structure may retain bonding functionality.

[0066] For CHD, its hydrophobic regions mainly include steroid ring structures, which form hydrophobic interactions and stable interactions such as π-π stacking with the hydrophobic walls in the FABP binding pocket. Figure 2 E). The AlphaFold2 predicted structure of CHD_MRP1 has a combined pocket NN-RMSD value of 2.2 Å with that in FABP, indicating good consistency in its spatial configuration. Figure 2F), whose sequence is listed in Table 1 (Sequence D).

[0067] Table 1. Specific sequences of peptides

[0068] Example 2: Iterative Optimization Based on Resampling and Structure Reconstruction Further, through interface resampling and iterative optimization of key residues, shorter and more affinity-enhancing recognition peptide variants were constructed. Taking FA as an example, similar to the construction process of FA_MRP1, FA_MRP2 to FA_MRP5 were obtained through iteration, and all-atom explicit solvent MD simulations were performed to analyze their binding behavior in depth. The structure of MRPn was taken from the last frame of MD after the MRP(n-1) system stabilized (RMSD fluctuations were stable) in the previous round of optimization. The stability of each conformation during the dynamic adaptation process was evaluated using RMSD. Figure 3 (AB). The results showed that MRP1~3 and MRP5 maintained a stable binding state with the ligand within 50 ns to 100 ns after MD. Although MRP4 showed some fluctuations in the early stage, it also showed a stable binding characteristic in the last 20 ns.

[0069] Analysis of the binding free energy of the simulated trajectory at the end of the last 10 ns showed that the binding energies of MRP1, MRP2, and MRP3 to FA were -85.104, -92.053, and -90.379 kJ / mol, respectively. Figure 3 The binding energies of MRP4 and MRP5 are similar, possibly related to the larger number of residues and more stable binding pockets. Notably, MRP5, obtained by removing six non-critical terminal residues from MRP4, shows an increase in binding energy from -60.965 kJ / mol to -64.989 kJ / mol, and a significant decrease in RMSD, indicating superior binding stability. A comparison of the RMSD distributions of MRP4 and MRP5 shows that MRP5 has a narrower RMSD distribution range, indicating that its complex is more stable in dynamic simulations. Figure 3 B). This indicates that during the iterative optimization process, unnecessary residues were effectively removed, reducing entropy loss and thus forming a tighter peptide-ligand binding interface. Further root mean square fluctuation (RMSF) analysis showed that the MRP5 complex had lower residue flexibility (B). Figure 4 This indicates that its structure is more rigid, which is beneficial for maintaining stable binding. Therefore, the 18-residue MRP5 was selected as the target FA-binding peptide for further in-depth research (it achieved good stability and affinity even with a residue count of less than 20). The sequences of MRP1 to MRP5 are shown in Table 2.

[0070] Table 2 Sequences of FA_MRP1-5

[0071] In the design of T3-recognition peptides, multiple iterations were conducted, resulting in six candidate peptides (T3_MRP1 to T3_MRP3.1). Among them, the 18-residue MRP3.1 exhibited significant conformational fluctuations in a 200 ns simulation. Figure 5 A), RMSD has a wider distribution range. Binding energy analysis shows that its average and median binding energies are close to 0, indicating that its binding ability is weak or non-existent (A). Figure 5 C). Therefore, MRP3 (i.e., MRP(n-1)) was redesigned to obtain a new peptide sequence named MRP4, and based on this, retaining key binding residues, a new generation of 18-residue peptide MRP4.1 was designed. Both MRP4 and MRP4.1 exhibited a single-peaked RMSD distribution, and MRP4.1 was 3 residues shorter than MRP4, with a narrower RMSD distribution (stable between 0 nm and 0.5 nm), indicating stronger structural stability. Figure 5 B). MM / PBSA analysis results also indicate that MRP4.1 has good binding ability, with a free energy of -48.79 kJ / mol ( Figure 5 C). Structural analysis shows that glycine and leucine in MRP4.1 form hydrogen bonds with the amino, carboxyl, and hydroxyl groups of T3, while adjacent hydrophobic residues interact with T3 through van der Waals forces, forming a cooperative recognition. These interactions together form a spatially selective semi-enclosed configuration, effectively stabilizing T3 and reducing the system's free energy. The sequences of MRP1 to MRP4.1 are shown in Table 3.

[0072] Table 3. T3_MRP1~4.1. Sequences

[0073] For CHD, it binds to the hydrophobic pocket in bile acid-binding protein (BABP) to form a stable complex (K). d =1.2 μM). This protein belongs to the lipid-binding protein family and its main function is bile acid binding and transport. Structurally, it adopts a typical β-barrel fold, which is characteristic of intracellular fatty acid-binding protein domains. Since it mainly relies on the β-barrel structure to form a hydrophobic binding pocket for recognition, rather than charge interaction, maintaining its structural integrity is particularly critical during the design process. After 100 ns simulation of the MRP2.1 obtained in the first round of design, it was found that it could not stably maintain the hydrophobic binding pocket, resulting in a decrease in the contact area with CHD and a significant reduction in binding energy ( ). Figure 6 A, D). To overcome this problem, MRP2.2 was obtained through structural reconstruction, retaining the β-bucket core structure. Subsequently, MRP5 was designed according to the shortened sequence process, but it showed poor stability in binding with CHD in simulations, large fluctuations in RMSD, and a wide range of binding energy variations. Figure 6 B, E). Therefore, MRP3.2 was redesigned and further optimized to obtain 18-residue MRP3.3. Simulation results show that MRP3.3 exhibits excellent performance in both RMSD and predicted binding energy (B, E). Figure 6 C, F), possessing better integration potential. The sequences of multiple MRPs generated during the design process are detailed in Table 4.

[0074] Table 4 Sequences of CHD_MRP1-5

[0075] Example 3: Single-point mutation-assisted MRP evolutionary design The third step of evolution (Evolution) Figure 1 C) Further adjust the properties of the MRP obtained in the previous step, such as affinity and hydrophilicity.

[0076] Due to conformational constraints imposed by the natural receptor protein template during the design process, the initially constructed MRPs may not yet have achieved optimal binding affinity with small molecules. To improve their recognition performance, a computationally assisted single-point mutation strategy was employed to optimize the recognition peptide.

[0077] Using the FA_MRP5 and FA simulated last frame structure as the initial conformation, 20 amino acid single-point mutations were performed at each site, and the binding energy changes of the complex before and after the mutations were calculated to evaluate the effect of each residue on the stability of the complex. Figure 7 A). A mutation energy of 0 kcal / mol indicates a neutral mutation (white), >0 kcal / mol indicates an unstable mutation (red), and <0 kcal / mol indicates a stable mutation (blue). By analyzing the interactions between MRPs and small molecules, the residues in the recognition peptide are classified into: hydrogen bonding, hydrophobic interaction, semi-contact interaction, and non-contact interaction. Figure 8 A). Based on the distribution of mutation energy heatmaps, multiple MRP variants were designed to optimize the binding stability of FA_MRP5. During variant design, all key hydrogen bond interaction sites were preserved, and non-contact, semi-contact, and hydrophobic interaction residues were preferentially mutated.

[0078] First, to improve the water solubility of FA_MRP5, the hydrophobic residues W23, Y26, and C30 were mutated to the hydrophilic residues R23, R26, and H30, constructing variant FA_MRP5.1. Based on this, to enhance its affinity for FA, Q25 was mutated to F25, resulting in FA_MRP5.2.1. Considering that cyclic peptides are generally more stable than linear peptides, to enhance binding strength, cysteine ​​residues were introduced at the N and C ends of MRP5 to form disulfide bonds, constructing the cyclic peptide variant FA_MRP5.3. Furthermore, to simultaneously improve affinity and water solubility, FA_MRP5.4 was designed by mutating W23, Y26, E32, and S34 to R23, R26, F32, and N34, respectively; and FA_MRP5.5 was designed by mutating W23, V24, and Y26 to R23, R24, and R26. Figure 8 BF). The specific sequences for FA_MRP5.1-5.5 are shown in Table 5.

[0079] Molecular dynamics simulations were then performed to analyze the interaction between all MRP mutants and FA. The results showed that the RMSD fluctuated significantly during the simulations of both FA_MRP5.1 and FA_MRP5.5. Figure 9 (A, B) indicates that the binding interface with the ligand is unstable, and ligand detachment occurs in the later stages of the simulation. This instability is also reflected in its positive binding energy ( Figure 9 C) indicates that the two do not have effective affinity.

[0080] In contrast, FA_MRP5.2.1, FA_MRP5.3, and FA_MRP5.4 all exhibited good structural stability in the later stages of the simulation, with overall RMSD remaining at a low level and relatively smooth fluctuations. Figure 9 A and B), whose binding energies are both negative ( Figure 9 C) indicates a strong affinity for FA. Notably, among all stable variants, FA_MRP5.4 exhibits the lowest binding energy to FA (-74.796 kJ / mol), significantly superior to the original FA_MRP5 (-64.989 kJ / mol), suggesting that the binding interface of FA_MRP5.4 is effectively enhanced through appropriate amino acid mutations.

[0081] Comparison of the complex structures of FA_MRP5 and FA_MRP5.4 with FA, for example Figure 7As shown in Figure B, in the FA_MRP5 complex, residues such as V24, S27, I31, and W36 collectively form a semi-closed hydrophobic pocket to enclose the hydrophobic portion of FA, while its hydrophilic end with the carboxyl group faces outwards to interact with surrounding water molecules. In contrast, FA_MRP5.4 forms a larger binding pocket, mainly composed of residues such as P21, L22, V24, I31, and N34, creating a larger binding interface with FA. Notably, in addition to encapsulating the hydrophobic structure, the hydrophilic carboxyl group in the FA molecule is also more widely recognized in MRP5.4. Therefore, FA_MRP5.4 was selected as the optimal peptide for subsequent validation and experiments.

[0082] Table 5 FA_MRP5.1-5.5 Sequences

[0083] Similarly, for T3, using the final frame structure from MRP4.1 molecular dynamics simulations of the T3 complex as the initial conformation, the binding energy changes after single-point mutations of each residue into 20 amino acids were systematically evaluated. Figure 7 C), and potential optimized sites were screened. Based on this, following the design strategy of FA_MRP, four T3 binding variants were constructed, named MRP4.1.1 to MRP4.1.4 (sequences are shown in Table 6). Through 200 ns molecular dynamics simulations of each variant with T3, the RMSD fluctuations and binding free energy distributions were systematically analyzed. Figure 10 The results showed that some variants exhibited significant fluctuations during the simulation, indicating a lack of conformational stability in their binding. In contrast, MRP4.1.1 showed better convergence of its RMSD curve throughout the simulation, with stable ligand relative positions, demonstrating strong conformational retention. Binding energy calculations further validated this trend: compared to the initially designed MRP4.1 (-48.79 kJ / mol), the binding energy of MRP4.1.1 was significantly reduced to -89.43 kJ / mol, demonstrating a substantial increase in binding affinity.

[0084] Table 6. T3_MRP 4.1.1-4.1.5 Sequences

[0085] Combining the RMSD and binding energy results, it can be inferred that MRP4.1.1 formed a more stable non-covalent interaction network during its interaction with T3, including the synergistic contribution of hydrogen bonds and hydrophobic interactions. Figure 7(D) This mode of action not only enhances the affinity between the peptide and the small molecule but also reduces energy fluctuations in the binding pocket during kinetic processes, making the complex more stable. In contrast, the binding energies of the other variants did not show significant improvement, indicating that the contribution of their optimized sites was limited. Therefore, MRP4.1.1 is the best-performing candidate peptide in this round of evolution, with significantly better binding affinity and stability than the original sequence and other variants.

[0086] For CHD, four binding variants were designed, named MRP3.3.1 to MRP3.3.4 (sequences shown in Table 7). In 100 ns molecular dynamics simulations, all four variants exhibited varying degrees of conformational stability. Figure 11 A, B). Binding energy calculations show that the predicted binding energy of MRP3.3.4 is -91.48 kJ / mol, significantly better than the initial design MRP3.3's -40.16 kJ / mol, and the most outstanding among the four variants. Figure 11 C). This result indicates that MRP3.3.4 can form a more stable interaction network with CHD in the simulation system ( Figure 7 F), the non-covalent interactions (such as hydrogen bonds and hydrophobic interactions) during the binding process are more persistent and effective. Notably, the increased binding energy not only signifies enhanced affinity but also suggests that the optimized peptide's binding conformation more closely resembles the spatial characteristics of the natural binding pocket. The convergence of the RMSD distribution further demonstrates that MRP3.3.4 maintains high conformational stability during the kinetic process, thereby reducing the impact of free energy fluctuations on binding. In contrast, the other variants (MRP3.3.1, MRP3.3.2, and MRP3.3.3) have higher binding energies but do not exhibit the ability to form a persistent and tight binding state as shown by MRP3.3.4. MRP3.3.4 demonstrates superior advantages in both predicted binding energy and conformational stability, showcasing its potential as an optimized short peptide and laying the foundation for subsequent experimental validation and the construction of a CHD-based detection platform.

[0087] Table 7 CHD_MRP 3.3.1-3.3.4 Sequences

[0088] Example 4: Mass spectrometry analysis of complexes between MRPs and small molecules To further investigate the binding mechanism of MRPs to small molecule ligands, native mass spectrometry analysis of the recognition peptide-ligand complex was conducted. Analytical techniques capable of accurately determining the binding stoichiometry without disrupting the complex structure remain scarce, especially for systems containing both low-molecular-weight peptides and small molecules. In this study, ESI, a mild ionization method, was used to measure the gas-phase interaction between the recognition peptide and the small molecule, thereby revealing its binding characteristics in solution.

[0089] In the mass spectrum of FA_MRP5.4 (molecular weight 2295.64) ( Figure 12 A) exhibits two characteristic peaks, located at m / z 766.7 and 1148.2, corresponding to its trivalent ion [M+3H]. 3+ and divalent ions [M+2H] 2+ Subsequently, the mass spectra of the mixture of FA_MRP5.4 and FA (molecular weight 442.3) at a 1:1 molar ratio were analyzed. Figure 12 B). New peaks appeared in the mixture, located at m / z 913.0 and 1369.4, corresponding to the [M+3H] of the FA_MRP5.4-FA complex. 3+ and [M+2H] 2+ This indicates that the two formed a stable 1:1 complex. Simultaneously, a new peak at m / z 442.3 was observed in the mixture, corresponding to the [M+H] group of the FA molecule. + The peak (theoretical molecular weight 441.4) indicates the presence of free FA molecules in the solution.

[0090] In the mass spectrum of CHD_MRP3.3.4 ( Figure 12 C), two characteristic peaks appeared (m / z 764 and 1146.5), corresponding to its [M+3H] values ​​respectively. 3+ and [M+2H] 2+ Ion peaks. After mixing with CHD (cholic acid) at a 1:1 ratio, new ion peaks were detected at m / z 900.4 and 1350.3, corresponding to the [M+3H] ion peaks of the CHD_MRP3.3.4-CHD complex. 3+ and [M+2H] 2+ ( Figure 12 D), also showed a 1:1 binding ratio. However, no molecular ion peak of free CHD was detected in this mixed system, indicating that CHD was completely bound in solution.

[0091] Mass spectrum of T3_MRP4.1.1 ( Figure 12E) also shows two ion peaks (m / z 764.0 and 1146.5), corresponding to its trivalent and divalent ion forms, respectively. New peaks appearing in the mixture are located at m / z 907.1 and 1360.6, indicating that MRP4.1.1 and T3 form a 1:1 complex ( Figure 12 F). Furthermore, a peak at m / z 651.40 was also detected in the mixture, corresponding to [M+H] of free T3. + This indicates that some T3 was not bound and still exists in the solution in a free form.

[0092] Since the molecular weight of small molecules is much smaller than that of the recognition peptide, and without considering strong ion competition, it is assumed that ligand binding does not affect the ionization efficiency of the peptide. Therefore, under the first-order approximation, the intensity of the peptide-related peak in the mass spectrometer can approximately represent its concentration in solution. Based on this analysis, the relative abundance of the bound peak and the free peak can be used to estimate the Kc of the three MRPs and their corresponding small molecules. d The values ​​were measured, and the results showed that they were all in the micromolar range.

[0093] Example 5 Affinity determination and intermolecular interaction analysis To comprehensively and accurately determine the binding affinity between MRPs and target small molecules and to deeply elucidate their intermolecular interaction mechanisms, this study employed intermolecular thermodynamics (ITC) analysis. ITC, as a classic thermodynamic research method, can directly obtain the Kb during the binding process without destroying the integrity of the complex. d By combining key thermodynamic parameters such as enthalpy change (ΔH), entropy change (ΔS), and free energy (ΔG), solid experimental evidence is provided to verify the effectiveness and universality of the RIE design strategy. Figure 13 ).

[0094] In the FA and MRP5.4 system, experimental results show that its K d The enthalpy change of 12.7 μM indicates that the peptide can establish a moderately strong binding relationship with FA. Further thermodynamic analysis shows that this binding process is mainly enthalpy-driven. The negative enthalpy change (ΔH = -5189 cal / mol) reveals that this interaction is exothermic and is mainly dominated by hydrogen bonds and electrostatic forces. This result is highly consistent with the hydrogen bond and electrostatic network enhancement predicted by the RIE-based design framework, thus experimentally validating the rationale for RIE. Meanwhile, the strongly negative free energy (ΔG = -6681.98 cal / mol) further demonstrates the spontaneity and thermodynamic advantage of this binding process. In summary, the binding mode of FA-MRP5.4 well reflects the enhancement effect of RIE in optimizing the hydrogen bond and electrostatic network during resampling and iteration.

[0095] In the T3 and MRP4.1.1 system, the K measured by ITC d The value of 8.8 μM indicates that the MRP can strongly recognize T3. Unlike the results of FA, the binding of T3-MRP4.1.1 exhibits an entropy-driven process. The positive enthalpy change (ΔH = 444.2 cal / mol) indicates that the binding process absorbs heat, but due to the strong entropy increase effect (-TΔS = -7330.8 cal / mol), the overall free energy remains highly negative (ΔG = -6886.6 cal / mol), thus ensuring the spontaneity of the interaction. This entropy-dominated binding mode is usually closely related to the rearrangement of solvent molecules and hydrophobic interactions. This result shows that the RIE strategy enables the MRP to adapt to the chemical properties of the target molecule through sequence evolution, that is, to compensate for the weak polar interactions by enhancing hydrophobic interactions, thereby achieving stable binding.

[0096] Regarding the combination of CHD and MRP3.3.4, the K of this system d The value of 8.0 μM indicates that MRP3.3.4 has a good binding ability to CHD. ITC data show that its binding mode is similar to that of T3-MRP4.1.1, both exhibiting entropy-driven processes. The positive enthalpy change (ΔH = 611.2 cal / mol) and the negative entropy term (-TΔS = -7569.2 cal / mol) work together to result in a highly negative free energy (ΔG = -6958 cal / mol), ensuring the spontaneity of the process. Further interfacial analysis reveals that the interactions in this system are dominated by hydrophobic effects, with a limited number of hydrogen bonds. This fully demonstrates that RIE design enables MRP to achieve stable binding to highly hydrophobic bile acid molecules by optimizing hydrophobic interactions.

[0097] Therefore, the ITC results of three representative small molecules—FA, T3, and CHD—all demonstrate that MRPs designed via RIE can achieve high-affinity recognition of target molecules. More importantly, the three exhibit significant differences in thermodynamic characteristics: the binding of FA to MRP5.4 is primarily enthalpy-driven, while the binding of T3 to MRP4.1.1 and CHD to MRP3.3.4 is entropy-driven. This difference not only reveals the crucial role of ligand chemistry in determining the binding mechanism but also further emphasizes the adaptability and universality of the RIE strategy. RIE can optimize hydrogen bonding and electrostatic interactions for polar ligands and drive binding for hydrophobic ligands through hydrophobic effects and solvent recombination, thus providing a universal design framework for peptide recognition of various small molecules.

[0098] Example 6: Specificity evaluation of MRPs To systematically evaluate the specific recognition ability of the designed molecular recognition peptides, this study employed a fluorescence signal response-based detection method to visually reflect the binding behavior and differences between peptide molecules and different targets. By comparing the changes in fluorescence intensity under the interaction of different small molecules, the selectivity of MRPs for target molecules and their potential cross-reactivity can be effectively determined.

[0099] MTX, a drug with a structure similar to FA, was used as a control to evaluate the cross-reactivity of MRP5.4. Figure 14 As shown in Figure A, the experiment shows that MRP5.4 can specifically recognize FA and produce significant fluorescence enhancement, while MTX failed to elicit a similar response, indirectly indicating that the affinity of MRP5.4 for MTX is much lower than its affinity for FA.

[0100] Similarly, such as Figure 14 As shown in Figure B, experiments indicate that MRP4.1.1 exhibits high selectivity for T3, while showing no significant binding to structurally similar analogs such as T4, LT4, and T2 sodium.

[0101] At the same time, such as Figure 14 As shown in Figure C, experiments indicate that MRP3.3.4 exhibits strong specificity for CHD, while showing almost no response to structurally similar bile acids DCA, LCA, and UDCA, suggesting that these analogues cannot form a stable binding structure with MRP3.3.4.

[0102] Overall, the cross-reactivity experiments showed that the MRPs designed by RIE exhibited high specificity in recognizing their respective target molecules, but showed almost no reaction to structure-related small molecules. Figure 14 ).

[0103] These results demonstrate that RIE design can not only effectively preserve the key spatial features and chemical complementarity of the target binding pocket, but also significantly improve the molecular recognition accuracy and selectivity of peptide molecules.

[0104] In summary, this invention proposes and validates the RIE (Resampling-Iteration-Evolution) strategy, a top-down computational design method based on structural reconstruction, for identifying small molecule binding sites in protein-ligand cocrystal structures and thereby achieving rational design of MRPs. Compared with traditional top-down methods that rely on linear truncation, the RIE strategy overcomes multiple limitations of traditional methods by systematically resampling discontinuous binding domains, reconstructing compact binding motifs, and optimizing peptide sequences through rational evolution. Specifically, this strategy effectively avoids the loss of multi-site synergistic recognition, solves the problem of poor solubility caused by exposed hydrophobic residues, and alleviates conformational instability caused by domain truncation.

[0105] The results show that the RIE strategy is not only universal but also exhibits excellent performance in practical applications. Through computational design and experimental verification of three types of small molecules with distinct chemical properties—FA, T3, and CHD—this invention successfully obtained multiple high-affinity MRPs (as shown in SEQ ID NO.1-SEQ ID NO.3, respectively), which demonstrated good binding performance in ITC, mass spectrometry, and specificity detection. This fully demonstrates that the RIE strategy can achieve scalable design and optimization for different types of small molecule targets. This approach not only provides an efficient and systematic new method for MRP design but also proposes a novel concept combining pocket resampling with a large-scale structural database, establishing a universal design framework for small molecule recognition. It expands the design thinking of peptide engineering and lays a solid theoretical and methodological foundation for the practical application of MRPs in food safety, environmental monitoring, and precision medicine.

[0106] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A molecular recognition peptide, characterized in that, The molecular recognition peptide is any one of the following: a. Molecular recognition peptide I that specifically recognizes the small molecule ligand FA, the amino acid sequence of which is shown in SEQ ID NO.1; b. Molecular recognition peptide II that specifically recognizes the small molecule ligand T3, the amino acid sequence of which is shown in SEQ ID NO.2; c. Molecular recognition peptide III that specifically recognizes the small molecule ligand CHD, with the amino acid sequence shown in SEQ ID NO.

3.

2. The application of the molecular recognition peptide according to claim 1 in the preparation of products for detecting small molecule ligands, characterized in that, The small molecule ligands include FA, T3, or CHD.

3. A method for designing molecular recognition peptides based on a structural reconstruction and iterative evolutionary framework (RIE) strategy, characterized in that, Includes the following steps: S1. Resampling: Small molecule ligands and their high-affinity natural receptor protein complexes are obtained from the database. Discontinuous residues, including key binding residues and secondary structure support elements, are extracted and reconstructed into spatially coordinated binding motifs using structure prediction tools. These motifs are denoted as the first-generation molecular recognition peptide MRP1. S2, Iteration: By integrating MD simulation and AlphaFold2 prediction, several rounds of iterative optimization are performed to obtain the molecular recognition peptide MRPn that retains the binding ability. S3. Evolution: Based on the predicted peptide-ligand binding conformation, the minimized molecular recognition peptide MRPn is optimized by site-directed mutagenesis to obtain a target molecular recognition peptide with high affinity.

4. The method according to claim 3, characterized in that, In step S1, the small molecule ligand includes FA, T3, or CHD; The natural receptor protein complex of FA is the folic acid receptor α with which it is complexed. The natural receptor protein complex of T3 is the thyroid hormone receptor TR, which is complexed with thyroxine. The natural receptor protein complex of CHD is a fatty acid-binding protein that binds to it; In step S1, the key binding residues are residues that form hydrogen bonds and hydrophobic interactions.

5. The method according to claim 3, characterized in that, In step S1, the reconstructing of the binding motif into a spatially coordinated binding motif using a structural prediction tool specifically includes: constructing an artificial binding pocket using AlphaFold modeling, wherein the artificial binding pocket retains the structural features of the natural receptor binding site in space, denoted as the first-generation molecular recognition peptide MRP1; Preferably, the reconstruction specifically includes: swapping or reversing the positions of fragmented sequence segments to preserve the spatial relative positions of the combined pockets, so as to ensure that the new pocket is close to the original pocket. The closeness means that the spatial point set of the spatial structure of the two pockets satisfies RMSD < 3 Å. The spatially coordinated binding motif specifically refers to the reconstructed sequence folding being able to form a three-dimensional conformation that conforms to the characteristics of a typical peptide structure.

6. The method according to claim 3, characterized in that, In step S2, the iteration specifically includes: The dynamic interactions between MRP1 and the target molecule were captured using all-atom MD simulations, and the binding interface was resampled during the process. Subsequently, a new key pocket was extracted from the dynamic stable state of the MRP1-ligand complex, and the second-generation molecular recognition peptide MRP2 was predicted by AlphaFold2. This process was repeated until the (n+1)th generation molecular recognition peptide MRPn failed to show stable binding in MD simulations of more than 100 ns, and the peptide sequence with the fewest residues was obtained, which is the minimized molecular recognition peptide MRPn. Preferably, the failure to exhibit stable binding means that the RMSD fluctuation is greater than 0.5, or the binding energy is greater than 0.

7. The method according to claim 3, characterized in that, In step S3, the site-directed mutation includes: By introducing single-point mutations and analyzing the dynamic binding stability and binding energy of the peptide in aqueous solution using molecular dynamics (MD) analysis, we can evaluate its affinity for screening target molecule recognition peptides (MRPs).

8. A device for designing molecular recognition peptides based on a structural reconstruction and iterative evolutionary framework (RIE) strategy, characterized in that, include: The acquisition module retrieves small molecule ligands and their high-affinity natural receptor protein complexes from the database, extracts discontinuous residues including key binding residues and secondary structure support elements, and reconstructs them into spatially coordinated binding motifs using structure prediction tools, denoted as the first-generation molecular recognition peptide MRP1. The iterative module is used to perform several rounds of iterative optimization by integrating MD simulation and AlphaFold2 prediction to obtain the minimum molecular recognition peptide MRPn that retains binding ability; An evolution module is used to combine the predicted peptide-ligand binding conformation with site-directed mutagenesis to optimize the peptide sequence of the minimized molecular recognition peptide MRPn, thereby obtaining a target molecular recognition peptide with high affinity.

9. An electronic device, characterized in that, include: processor; A memory having executable code stored thereon, which, when executed by the processor, causes the processor to perform the method as described in any one of claims 3-7.

10. A computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method as described in any one of claims 37.

Images