Design, Synthesis, and Neurotrophic Effect of Arg-Glu-Arg-Met-Ser-(3,5)-Dimethyladamantan-1-Amine In Vitro Evaluations as a Potential NMDAR Antagonist

Methyl-D-aspartate receptor (NMDAR) is an ionotropic glutamate receptor and plays an important role in neuronal degradation of Alzheimer’s disease (AD). According to molecular modeling docking studies, we have designed the compound Arg-Glu-Arg- Met-Ser-(3,5)-dimethyladamantan-1-amine (RERMS-MEM), consisting of an A β PP 5-mer peptide (RERMS) and memantine (MEM). Tis compound could dock into the active sites of N-methyl-D-aspartate receptor type 2B (NMDAR2B) with a − 64.14 kcal/mol CDOCKER interaction energy. Te stability of RERMS-MEM was evaluated through a 50ns molecular dynamics simulation. Te results revealed that the docked ligand-receptor complex was stable. Furthermore, surface plasmon resonance (SPR) revealed that the RERMS-MEM binding afnity to the NMDAR2B fragment exhibited over 15-fold enhancement compared to MEM. Te SH-SY5Y cell assays showed that RERMS-MEM or RERMS at concentrations of 0.1, 1, 10, or 50 μ M could enhance the metabolic rate, and MEM showed no diference compared to the control and indicated cytotoxic efects at 50 μ M. RERMS-MEM at concentrations of 0.01, 0.1, 1, 10, or 50 μ M increased the number of viable cells and reduced the release of lactate dehydrogenase (LDH). RERMS at concentrations of 10 or 50 μ M was similar to RERMS-MEM for increasing viable cells, and MEM showed no diference compared to the control and decreased the number of viable cells at 50 μ M. RERMS-MEM or RERMS at concentrations of 10 or 50 μ M could antagonize A β 25-35 -induced cytotoxicity, and MEM at 50 μ M strengthened the cytotoxicity efects. Te results revealed that RERMS-MEM showed a strong NMDAR-blocking activity as a potential NMDAR antagonist, enhancing the neurotrophic efect and cellular growth in SH-SY5Y cells.


Introduction
Alzheimer's disease (AD) is one of the most common neurodegenerative diseases and often the leading cause of dementia in the elderly. It is caused by the interaction of multiple genetic and environmental risk factors [1,2]. Memory loss correlated with neuronal loss in the hippocampus is the main symptom of AD [3,4], which plays a critical role in AD-induced brain function injury and neurotoxicity. N-methyl-D-aspartate receptors (NMDARs) are ligand-gated cation channels embedded in the cell membrane of neurons and are critically involved in the pathogenesis of AD. Overexcitation of NMDARs leads to excessive infux of Ca 2+ through receptor-associated ion channels, resulting in neuronal cell injury or death. N-methyl-D-aspartate receptor type 2B (NMDAR2B), a member of NMDARs family, plays a critical role in spatial learning and memory [5]. Tus, the blockade of NMDAR2B is benefcial for neuroprotection and the prevention of Aβ-induced neuronal disruption [6]. A new homology-based model of the glutamate binding site of the NMDA receptor NMDAR2B subunit was constructed, satisfactorily explaining the structure-activity relationships among a series of agonists, competitive antagonists, and the glutamate site [7].
According to statistics, there are approximately 47 million people worldwide sufering from dementia, of which 80% are diagnosed with AD [8,9]. However, there is a lack of efective AD drugs available. Te latest drug approved by the USFDA, lecanemab, is a humanized immunoglobulin G1 used for treating patients with mild cognitive impairment or mild dementia. However, due to the absence of efcacy and safety data for AD treatment in the early or late stages, it has not been widely used in countries such as China and Japan [10]. Most AD drugs currently available only target the symptoms of cognitive impairment and cannot prevent disease progression, highlighting the urgent need for the development of safe and efective AD drugs to improve this situation [11].
Molecular docking, molecular dynamics simulation, and ADMETprediction are essential methods in computer-aided drug design. Precise design methods for protein-ligand docking are crucial for identifying novel compounds in drug discovery [12]. Molecular docking enables the efective prediction of binding modes and binding afnities in a protein-ligand complex. Molecular dynamics simulation allows for the analysis of conformational changes in the ligand-receptor complex under physiologically relevant conditions, and the identifcation of key residues is involved in ligand-receptor interactions [13]. ADMET prediction provides insights into drug properties such as absorption, distribution, metabolism, excretion, and toxicity. Tese powerful tools have been extensively employed in various aspects of drug development research.
Accordingly, we aimed to design and synthesize a compound with the structure of RERMS and MEM, namely, structure Arg-Glu-Arg-Met-Ser-(3,5)-dimethyladamantan-1-amine (RERMS-MEM) ( Figure 1) and evaluate the neurotrophic efect as a potential NMDAR antagonist. In the structure of RERMS-MEM, the original framework of memantine is retained, and modifcations on memantine do not afect its blocking efect on NMDA [27]. Te modifcation of RERMS increases the number of binding sites between the compound and NMDAR, enhancing its antagonistic efect on NMDAR.

General Information.
All syntheses and bioassays were environmentally friendly without potential safety or environmental hazards. Protected amino acids with Lconfguration were purchased from Sigma Chemical Co. (MO, USA). Chromatography was conducted on Qingdao silica gel H. Te purity of RERMS-MEM and its intermediates was analyzed by thin-layer chromatography (TLC) (Qingdao silica gel F254, 0.25-mm layer thickness) or high-performance liquid chromatography (HPLC) (SHI-MADZU, JPN), YMC-Pack ODS-A (10 × 250 mm, 5 μm; YMC). 1 H NMR 300 MHz was recorded on a Bruker Avance II-500 spectrometer with DMSO-d 6 as the solvent and tetramethylsilane as an internal standard. ESI/MS was conducted on a mass spectrometer (ZQ 2000; Waters) with a dual ion source of ESI/matrix-assisted laser desorption ionization. Cell count and viability assays were conducted on a Muse ® Cell Analyzer. Statistical analyses of biological data were carried out using the T-test, and p values <0.05 were considered statistically signifcant.
SH-SY5Y cells were purchased from American type culture collection (ATCC, USA) within six months. Te cells were grown in Dulbecco's modifed Eagle's medium (Gibco BRL, New York, USA) containing 10% heat-inactivated foetal bovine serum (Hyclone, Los Angeles, CA, USA), penicillin (100 IU/ml), and streptomycin (100 μg/ml) in T 75 tissue culture fasks under 95% air, 5% CO 2 , and 37°C. 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, cell counting kit-8 (CCK-8), and Aβ 25 [6]. NMDAR was treated as rigid and prepared by AutoDockTools 1.5, merging nonpolar hydrogens and assigning Gasteiger charges and AutoDock elements. Te 10 energy-optimized conformations of the compounds were treated as rigid ligands and prepared by AutoDockTools 1.5, merging nonpolar hydrogens, assigning Gasteiger charges, fnding the root and aromatic carbons, detecting rotatable bonds, and setting torsions. Te grid box dimensions were set to 50 Å × 50 Å × 50 Å with a grid spacing of 0.375 Å. Te Lamarckian genetic algorithm (LGA) was used to fnd the appropriate binding positions, orientations, and conformations of the compounds in the active site of NMDAR. Te global optimization was started with parameters of a population of 300 randomly positioned individuals. Te maximum number of energy evaluations was increased to 2.5 × 10 7 , and the maximum number of generations in the LGA was increased to 2.7 × 10 5 . Te Solis and Wets local search was performed with a maximum number of 3000. During docking experiments, 200 runs were carried out for each compound. Te resulting 200 conformations of each compound were scored by the lowest binding energy and clustered with an RMS tolerance of 2.0 Å.

Molecular Dynamics Simulation Studies.
Te GRO-MACS 2020.3 software was utilized to conduct molecular dynamics (MD) simulations. For parameter and topology generation of proteins and ligands, the amber99sb-ildn force feld and the general amber force feld (GAFF) were employed, respectively. Te simulation box size was optimized to ensure a minimum distance of 1.0 nm between each atom of the protein and the box. Subsequently, the box was flled with water molecules based on a density of 1. To maintain electrical neutrality, Cl− and Na+ions were introduced to replace the water molecules. To minimize energy consumption and eliminate unreasonable contacts or atom overlap, an energy optimization step consisting of 5.0 × 104 iterations using the steepest descent method was performed on the entire system. Following energy minimization, a preliminary equilibration phase was conducted for 100 ps under the NVT ensemble at 300 K to stabilize the system's temperature. Subsequently, a second equilibration phase was simulated under the NPT ensemble at 1 bar and 100 ps. Te primary aim of these simulations was to optimize the interaction among the target protein, solvent, and ions, ensuring a fully pre-equilibrated simulation system. All MD simulations were conducted for 50 ns under an isothermal and isostatic ensemble, maintaining a temperature of 300 K and a pressure of 1 atmosphere. Temperature control was achieved using the V-rescale method, while pressure control employed the Parrinello-Rahman method. Te temperature and pressure coupling constants were set to 0.1 and 0.5 ps, respectively. Te Van der Waals force was calculated using the Lennard-Jones function, with a nonbond truncation distance of 1.4 nm. Te LINCS algorithm was applied to constrain the bond lengths of all atoms. Furthermore, the particle mesh Ewald method with a Fourier spacing of 0.16 nm was utilized to calculate long-range electrostatic interactions.

ADMET Prediction.
Software AutoDock 4 was used to predict the ADMET of RERMS-MEM. Import the small molecule compounds. Open the ADMET Descriptors dialog box. In the "Input Ligands" section, select all the small molecule compounds. In the "ADMET Descriptors" section, choose the default settings, which select all ADMET properties. Run the calculation workfow to initiate the job. Once the job is completed, click on "View Results" to perform result analysis. [37,38] 2.5.1. Preparing Boc-Arg (NO 2 )-Met-OBzl. A solution of 2.5 g (7.85 mmol) of Boc-Arg (NO 2 ), 1.25 g (9.25 mmol) of hydroxybenzotriazole (HOBt), 1.65 g (9.3 mmol) of N, N′dicyclohexylcarbodiimide (DCC) in 30 mL of anhydrous tetrahydrofuran (THF) was stirred at 0°C for 30 min. Ten, a solution of 3.20 g (7.80 mmol) of Tos Met-OBzl in 10 mL of anhydrous THF was added, and the pH was adjusted to pH 9 with N-methyl morpholine (NMM). Te reaction mixture was stirred at room temperature for 12 h, and TLC (CH 2 Cl 2 / CH 3 OH, 30/1) was used to indicate the complete disappearance of Tos Met-OBzl. Te resulting dicyclohexylurea was removed by fltration, and the fltrate was evaporated under reduced pressure. Te residue was dissolved in 50 mL of ethyl acetate and washed successively with aqueous sodium bicarbonate (5%), aqueous citric acid (5%), and saturated aqueous sodium chloride. Te organic layer was separated, dried over anhydrous sodium sulfate, fltered, and evaporated under reduced pressure to yield 3.1 g (74%) of the title compound as a colorless powder.

Preparing Boc-Arg (NO 2 )-Met.
A solution of 2.5 g (4.6 mmol) of Boc-Arg (NO 2 )-Met-OBzl in 30 mL of methanol was stirred at 0°C, to which 10 mL of aqueous NaOH (2 M) was added dropwise. Te reaction mixture was stirred for 5 h, and TLC (CH 2 Cl 2 /CH 3 OH, 30/1) was used to indicate the complete disappearance of Boc-Arg (NO 2 )-Met-OBzl. After fltration, the fltrate was evaporated under reduced pressure. Te residue was dissolved in 30 mL of water and then adjusted to pH 2 with hydrochloric acid (2 M). Te formed precipitates were dissolved in 50 mL of ethyl acetate and washed successively with aqueous sodium bicarbonate (5%), aqueous citric acid (5%), and saturated aqueous sodium chloride. Te organic layer was separated, dried over anhydrous sodium sulfate, fltered, and then evaporated under reduced pressure to obtain 2.04 g (98%) of the title compound as a colorless powder. (1). A solution of 1.95 g (9.5 mmol) of Boc-L-Ser, 1.5 g (11.1 mmol) of HOBt, and 4.22 g (11.1 mmol) of 2-(7azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafuorophosphate (HATU) in 30 mL of anhydrous N, Ndimethylformamide (DMF) was stirred at 0°C for 30 min, to which a solution of 2.16 g (10.0 mmol) of MEM in 10 mL of anhydrous DMF was added and adjusted to pH 9 with NMM. Te reaction mixture was stirred at room temperature for 12 h, and TLC (CH 2 Cl 2 /CH 3 OH, 30/1) was used to indicate the complete disappearance of Boc-L-Ser. Te fltrate was evaporated under reduced pressure. Te residue was dissolved in 50 mL of ethyl acetate and washed successively with aqueous sodium bicarbonate (5%), aqueous citric acid (5%), and saturated aqueous sodium chloride. Te organic layer was separated, dried over anhydrous sodium sulfate, fltered, and evaporated under reduced pressure to yield 2.7 g (78%) of the title compound as a colorless powder.   13 13 C NMR, HPLC of RERMS-MEM were supplied as experiment data (Figure S1-4).

Surface Plasmon Resonance (SPR) Assay.
Following the abovementioned molecular docking results, the NMDAR2B protein fragment was selected for SPR analysis of RERMS-MEM, RERMS, and MEM. Te sequence of the protein fragment was FEYFSPVGYNRCLADGREPGGPSFTIGKA IWLLWGLVFNNSVPVQNPKGTTSKIGSTANLAAFMIQE EYVDQVSGLSDKKFQRPNDFSPPFRFGTVPNGSTERNIR NNYAEMHAYMGKFNQRGVDDALLSLKTGKLDAFIYDA AVLNYMAGRDEGCKLVTIGSGKVFASTGYGIAIQKDSGW KRQVDLAILQLFGDGEMEELEALWLTGICHNEKNEVM SSQLDIDN (containing the active sites of NMDAR). SPR assays were performed by using a Biacore 8 k system (Cytiva) with three steps. Te frst step was protein immobilization. Proteins were diluted in sodium acetate solution (GE Healthcare) and immobilized on a CM5 chip through a 1ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride/N-hydroxy sulfo succinimide (EDC/NHS) reaction. Subsequently, afnity detection was performed according to the operating protocol provided by GE Healthcare. Diluted RERMS-MEM, RERMS, and MEM were added to concentrations of 6.25-200 μM in the running bufer. Te analytes were injected into the system at a fow rate of 30 μL/ min, while the association and dissociation times were 120 s and 400 s, respectively. Te association and dissociation processes were all conducted in the running bufer. In the last step of data processing, the afnity curve ftting was carried out with Biacore insight evaluation software. A steady-state afnity model was used for the curve ftting, and the dissociation equilibrium constant (K D ) was also calculated. Te Rmax was calculated according to the immobilization level. Te SPR assay adopts a double deduction system, and the negative signal is automatically deducted by the instrument.

MTT Assay.
SH-SY5Y cells were seeded onto 96-well plates (∼3 × 10 4 cell/well) and grown until reaching 80% confuence. Cells were either untreated (control) or treated with MEM, RERMS, or RERMS-MEM at concentrations of 0.01, 0.1, 1, 10, or 50 μM for 48 h. Te cells were then exposed to MTT (5 mg/mL) to measure the metabolic rate. Te MTTassay was performed by incubating the cells with MTT solution for 4 h at 37°C. Te formed formazan was dissolved in dimethyl sulfoxide (DMSO). Cell viability was calculated by measuring the absorbance at 570 nm. Statistical analysis was performed using the SPSS 19.0 program, and a signifcance level of p < 0.05 was considered to indicate statistical signifcance.

CCK-8 Assay.
Te CCK-8 assay was performed according to instructions of the CCK-8 assay kit used. SH-SY5Y cells were seeded onto 96-well plates (∼3 × 10 4 cell/well) and grown until reaching 80% confuence. Cells were untreated (control) or exposed to MEM, RERMS, or RERMS-MEM at 0.01, 0.1, 1, 10, or 50 μM for 48 h. Next, 10 μL of CCK-8 solution was added to each well, and the cells were incubated at 37°C for 4 h. Cell viability was calculated by measuring the absorbance at 450 nm. Statistical analysis was performed with SPSS 19.0, with p < 0.05 considered to indicate signifcance.

Cell Count and Viability
Assay. SH-SY5Y cells were seeded onto 24-well plates (∼3 × 10 4 cell/well) and allowed to grow for 4 h. Next, the cells were untreated (control) or exposed to MEM or RERMS at 10 or 50 μM or to RERMS-MEM at 0.01, 0.1, 1, 10, or 50 μM for 48 or 96 h. Te cells were made into a single-cell suspension. Next, 50 μL of cell solution was added to 450 μL of Muse Count & Viability Kit reagent and incubated at room temperature for 5 min.
Te incubated samples were put into Muse ® Cell Analyzer for testing, and the number of total cells including live cells Journal of Chemistry and dead cells were measured. Statistical analysis was performed by using SPSS 19.0, and diferences with p < 0.01 were considered signifcant.

LDH Release
Assay. SH-SY5Y cells were seeded onto 96-well plates (∼3 × 10 4 cell/well) and allowed to grow to 80% confuence. Cells were untreated (control) or exposed to RERMS at 0.01, 0.1, or 1 μM for 24 h. Te supernatant from each well was collected. Te LDH activity assay kit was used to measure the LDH activity according to the manufacturer's instructions. Statistical analysis was performed by using SPSS 19.0, and p < 0.01 was considered signifcant.

RERMS-MEM Antagonized the Aβ 25-35 -Induced
Cytotoxicity. SH-SY5Y cells were seeded onto 96-well plates (∼3 × 10 4 cell/well) and grown until reaching 80% confuence. Cells were treated with Aβ 25-35 of 0.1, 1, 10, 20, or 50 μM as a toxicity screening group, treated with Aβ 25-35 of 20 μM and compounds (including MEM, RERMS, and RERMS-MEM) of 0.1, 1, 10, or 50 μM as drug intervention groups, and untreated as the control group for 24 h, respectively. All the groups of cells were then exposed to MTT (5 mg/mL) to measure the metabolic rate. Te MTT assay was performed by incubating the cells with MTT solution for 4 h at 37°C. Te formed formazan was dissolved in DMSO. Cell viability was calculated by measuring the absorbance at 570 nm. Statistical analysis was performed using the SPSS 19.0 program.

Docking of RERMS-MEM toward the Active Site of NMDAR.
Te molecular docking software AutoDock 4 was used to simulate the binding mode of the designed compound and NMDAR (PDB code: 5UN1). Docking of RERMS-MEM into the active sites of NMDAR was −64.14 kcal/mol of CDOCKER interaction energy. Te docking interaction energy was lower than the standard ligand of NMDAR with −27.52 kcal/mol ( Figure 2). Some interactions were observed between RERMS-MEM and NMDAR (Table 1).

Molecular Dynamics Simulation Studies.
To further investigate the dynamic interactions between the compound and NMDAR and assess the stability of the docked ligandreceptor complex, we conducted molecular dynamics simulations of 50 ns. Te RMSD fuctuated during 1-12 ns of simulation and system reached a converged state for the rest of the course with the root mean square deviation (RMSD) values fuctuated between 0.55 and 0.72 nm (Figure 3(a)). Te motion changes of individual amino acid residues during molecular dynamics simulations can be captured by the root mean square fuctuation (RMSF). Te highest observed fexibilities are related to terminal residues of the protein. Te residues 567 (D subunit), 581 (A subunit), 581 (C subunit), 582 (B subunit), 609 (B subunit), 609 (D subunit), 611 (A subunit), and 617 (C subunit) with higher fuctuations belonged to the loop regions. Te key active site amino acid residues exhibited rigid behavior in all the system indicating the stability of the compounds in the ligandreceptor complex (Figure 3(b)). Te radius of gyration (Rg) is a criterion of system compactness. Te smaller Rg indicates a denser protein structure, while a larger value suggests a looser structure. When the Rg value remains stable, the protein is considered stable throughout the entire simulation process. In the simulations of RERMS-MEM and NMDAR, the Rg value gradually decreased and stabilized after 30 ns, indicating that the system could bind stably (Figure 3(c)). Hydrogen bonding is an important noncovalent force that stabilizes protein structures and serves as a measure of stability for ligand-receptor complex. To assess the stability of the complex, we simulated the number of hydrogen bonds formed between the ligand and protein within a duration of 50 ns (Figure 3(d)).

ADMET Prediction.
Te ADMET parameters were calculated to investigate the drug-like activity of RERMS-MEM. As shown in Table 1, the predicted levels of solubility were 2, indicating its low water solubility. Te BBB-level was 4, indicating a less reliable prediction. Te p value prediction of plasma protein binding, cytochrome P450 2D6 inhibition, and hepatotoxicity parameters was small, indicating a less reliable prediction (Table 2).

Synthesis of RERMS-MEM.
An environmentally friendly synthetic route was designed to obtain RERMS-MEM at sufcient levels of purity and yield (Figure 4). Boc-Ser-MEM and Boc-Arg (NO 2 )-Met-OBzl were synthesized using conventional condensation agents. Boc-Arg (NO 2 )-Met-Ser-MEM was formed (36.7% yield) by coupling Boc-Arg (NO 2 )-Met and Ser-MEM after the removal of OBzl and Boc. Boc-Arg (NO 2 )-Glu (OBzl)-Arg (NO 2 )-Met-Ser-MEM was prepared (11.3% total yield) using the solution method and stepwise synthesis (from the C-terminus to the N-terminus) with Boc-Arg (NO 2 ), Boc-Glu (OBzl), and Arg (NO 2 )-Met-Ser-MEM as materials. Upon removal of all protective groups of Boc-Arg (NO 2 )-Glu (OBzl)-Arg (NO 2 )-Met-Ser-MEM, RERMS-MEM was obtained at 28% yield. Tese data suggested that the reaction conditions were mild and the yield of the individual reaction was acceptable.

SPR Assay.
Following the above molecular docking results, SPR was used to measure the binding afnity between NMDAR2B fragment and RERMS-MEM, RERMS, and MEM. Te result showed a concentration-dependent increase in resonance signals and demonstrated that all three compounds can strongly bind to the NMDAR2B fragment. Biacore insight evaluation software was used to further confrm the K D : RERMS-MEM showed the highest response and best afnity to NMDAR2B fragment, in which the K D values of RERMS-MEM, RERMS, and MEM were 0.601 μM, 2.14 μM, and 9.00 μM, respectively. Te K D value of RERMS-MEM showed a 14.97-fold decrease compared with MEM, which indicated that there is a more powerful afnity between RERMS-MEM and NMDAR2B fragment (Figures 5(a)-5(d)).  (Figure 6(a)). CCK-8 assays showed similar results as the MTT assay ( Figure 6(b)).

RERMS-MEM Increased the Number of Viable Cells.
Te cell viability test was used to further explore the neurotrophic efect of RERMS-MEM in SH-SY5Y cells. Te results showed that no signifcant diference was observed in the treatment of 10 μM MEM for 48 or 96 h compared with the control (Figure 7(c)), but a signifcantly decreased number of viable cells were found at 50 μM MEM compared with the control (Figure 7(d)). Furthermore, a signifcantly     (Figure 8).

RERMS-MEM Decreased the Release of LDH.
Te amount of released LDH is an indicator of cell death; therefore, the LDH assay was used to evaluate the neurotrophic efect of RERMS-MEM in SH-SY5Y cells. Treatment with 0.01, 0.1, 1, 10, or 50 μM RERMS-MEM signifcantly decreased LDH release compared with the control (Figure 9). Furthermore, a signifcantly greater decrease in LDH release

Discussion
AD is a degenerative disease of the central nervous system, in which the loss of neurons is one of the main pathological features. Overexcitation of NMDARs led to neuronal cell injury or death, resulting in neuronal cell loss. MEM, a lowafnity voltage-dependent uncompetitive antagonist of NMDAR, is currently used in combination with other acetylcholinesterase inhibitors in the treatment of AD [14][15][16][17]. Mounting evidence has shown that MEM had no clear positive efects in clinical applications although it showed promising results in the preclinical stages [42]. Previous studies revealed that AβPP 5 and AβPP 5 analogues exert a neurotrophic efect in vitro; thus, MEM was modifed by AβPP 5 aiming to design a safer and neurotrophic antidementia drug that could enhance the binding afnity with NMDAR.
NMDARs have a domain-layered architecture, with the amino-terminal domain and the ligand-or agonistbinding domain residing in the synaptic space and the transmembrane domain spanning the membrane [43,44]. Te docking investigation indicated that RERMS-MEM could dock into the LBD of NMDAR with lower CDOCKER interaction energy compared with the standard ligand, and twenty-nine interactions including van der Waals, conventional hydrogen bond, carbonhydrogen bond, and alkyl were observed between RERMS-MEM and NMDAR. Subsequently, a 50 ns molecular dynamics simulation was carried out on the docked complex. Te system reached a converged state during 12-50 ns with RMSD values fuctuated between 0.55 and 0.72 nm. Te RMSF result refected the stability of the compound in the ligand-receptor complex. Te Rg value gradually decreases and stabilizes after 30 ns, indicating that the system can bind stably. Te number of hydrogen bonds formed by NMDAR and RERMS-MEM was found to be from 1 to 8. Te results of molecular dynamics simulation studies revealed that the ligand-receptor complex was stable. Te ADMET parameters suggested its low water solubility. Te BBB-level, plasma protein binding, cytochrome P450 2D6 inhibition, and hepatotoxicity parameters were less reliable predictions. To further defnite RERMS-MEM as a potential NMDAR antagonist, the SPR assay was used to display the binding afnity among RERMS-MEM, RERMS, MEM, and NMDAR2B fragment. As a result, RERMS-MEM showed the highest response and most powerful afnity to NMDAR2B fragment. What is more, the K D value of RERMS-MEM decreased 14.97-fold compared to MEM. Te results indicated that there is a strong afnity between RERMS-MEM and NMDAR2B fragment, and it should be attributed to the RERMS modifcation, which improved the docking feature and led to more amino acid residues of the active site involved in the interactions between RERMS-MEM and NMDAR2B fragment.
We designed a series of cell assays to evaluate the neurotrophic efect of RERMS-MEM. MTT and CCK-8 assays revealed that RERMS-MEM or RERMS of 0.1, 1, 10, or 50 μM could enhance the metabolic rate, but MEM showed no diference compared with the control and indicated a cytotoxicity efect at 50 μM especially. In addition, the result of RERMS-MEM was similar to AβPP 5 analogues [35], which indicated that MEM modifed by AβPP 5 exerted a neurotrophic efect on cells. With respect to the cell viability and LDH release assay, RERMS-MEM of 0.01, 0.1, 1, 10, or 50 μM increased the number of viable cells and reduced the release of LDH, RERMS of 10 or 50 μM was similar to RERMS-MEM for increasing viable cells, but MEM showed no diference compared with the control and decreased the number of viable cells at 50 μM. In our opinion, the abovementioned two assays indicated that the mechanism of the neurotrophic efect of RERMS-MEM could be described as metabolic rate enhancement and cellular growth-promoting. Furthermore, RERMS-MEM or RERMS of 10 or 50 μM could more strongly antagonize the Aβ 25-35 -induced cytotoxicity, but MEM of 50 μM strengthened the cytotoxicity efect. Te abovementioned result revealed that RERMS-MEM could improve the safety of MEM (maximum clinical dosage is about 93 μmol/d) by the neurotrophic efect. As we know, adherence to medicine is an assignable problem in the history of drug treatment, especially for the elderly with AD, who are sufering from memory loss and cognition hypofunction [45]. As a result, they might have the risk of overdose and aggravation of adverse reactions. Compared with MEM, the modifed compound RERMS-MEM showed no cytotoxicity efect in the same high dose, indicating that it might be safer than MEM. All the cell assays proved that the modifcation design of RERMS-MEM was successful, which enhanced the neurotrophic efect by promoting the metabolic rate and cellular growth in SH-SY5Y cells. Future work will include cellular experiments investigating the binding between RERMS-MEM and NMDAR.

Conclusion
In general, RERMS-MEM, as a potential NMDAR antagonist, enhanced the metabolic rate and promoted cellular growth, showing a neurotrophic efect in SH-SY5Y cells at a low dose. In addition, no cytotoxic efect was observed for RERMS-MEM at a high dose. Considering its promising utilization against AD, this modifed drug is considered worthy of further development. In future studies, our efforts will be focused on further characterization of RERMS-MEM through a series of experiments in animal models of AD.