Catechins in insomnia-Alzheimer’s disease comorbidity: A network pharmacology and molecular docking study

Abstract

The comorbidity of insomnia and Alzheimer’s disease (AD) is strongly driven by the interplay between circadian rhythm disruption and immune dysfunction. Catechins are multi-target polyphenols capable of modulating both processes, yet their precise mechanism of action remains elusive. Insomnia and AD related differentialy expressed target genes were specifically sourced from circadian rhythm and immune phenotypes. Two machine learning algorithms were then applied to refine and identify the pivotal targets from this phenotype-driven target pool. Functional enrichment analyses, including kyoto encyclopedia of genes and genomes and gene set enrichment analysis, were performed to elucidate the involved signaling pathways. The compound-comorbidity differentialy expressed target genes related to circadian and immunity network identified catechin (C) and epicatechin as core components. PED4D, HMOX1, and IGF1 were robustly identified as the pivotal targets. PDE4D was found to be centrally involved in the complement pathway. The complement pathway and the phosphatidylinositol 3-kinase/Akt signaling pathway were significantly enriched. This dual pathway regulation converges to govern microglia-mediated synaptic pruning, a process integral to both sleep and neurodegeneration. This study unveils a mechanistic link from core catechins to the regulation of synaptic pruning via circadian-immune crosstalk, offering a novel therapeutic perspective for the insomnia-AD comorbidity.

1. Introduction

Insomnia is a sleep continuity disorder (difficulty falling asleep or maintaining sleep), 10% to 15% of people suffered from it.^[1] Many studies have shown that sleep disorders are risk factors for Alzheimer’s disease (AD).^[2,3] AD is one of the most prevalent neurodegenerative diseases affecting millions of people globally,^[4] characterized by progressive cognitive decline, intracellular hyperphosphorylated tau protein and extracellular aggregates of Aβ peptide.^[4] Disruption of circadian rhythm and immune dysfunction are important pathological mechanisms of insomnia and AD,^[5-8] and circadian rhythm dysregulation promotes pro-inflammatory responses of the immune system.^[9] Insomnia is the most common disruption of the sleep-wake cycle. Circadian rhythm abnormalities precede the onset of AD.^[10,11] Furthermore, insomnia is associated with increased inflammatory dysregulation,^[7] and immune processes may critically impacts the etiology and pathogenesis of AD.^[9] Insomnia and AD have posed heavy burdens on families and society,^[12] Despite significant advances in treatment, we are still lacking safe and effective medications.^[13] And polyphenols, which mainly derived from nature, have emerged as the most promising therapeutic drugs.^[14-16]

Catechins is the major polyphenols, it has become research hotspot in food and nutrition science due to its remarkable antioxidant, anti-inflammatory and preventive effects against chronic diseases.^[17] It exhibited functions to enhance insomnia and AD.^[18-20] Morever, the mechanisms has achieved certain achievements. For instance, they improved sleep via enhancing Cl(−) influx,^[21] regulating NO/ADMA/DDAH pathway,^[22] improved AD through enhancing antioxidant system via Keap1/Nrf2 pathway,^[20] and ROS/TXNIP/NLRP3 pathway, and improved AD through abating the hyperphosphorylation of tau protein and downregulated Aβ1-42 expression.^[20] However, the existing research mostly focuses on the effects of a single major component of catechins (such as EGCG) or crude extracts on a single disease. There is currently a lack of systematic research on the different structural types of catechins (such as monomers, dimers, and phenolic acids) in the intervention of insomnia-AD comorbidity. To systemly explore the functional laws of the relationship among catechins structure-effect-mechanism in insomnia-AD management, this study selected 5 key components with representative structures and functions as the basic monomer skeletons: (+)-catechin (C), (−)-epicatechin (EC); (−)-gallocatechin gallate (GCG) as the representative of esterified catechins, gallic acid (GA) as the repesentative of ester-type catechins; and procyanidin B2 (PB2) as the classic dimer model.^[23,24]

Therefore, this study was the first to employ network pharmacology, molecular docking, and molecular dynamics (MD) methods to systematically study the multi-target action mechanisms of 5 core catechin components – C, EC, GA, GCG, and PB2 in the intervention of insomnia-AD comorbidity. Figure 1 presented the workflow of this sduty.

Figure 1.

Workflow of this study. AD = Alzheimer’s disease, DEG = differentially expressed gene, EC = epicatechin.

2. Materials and methods

2.1. Prediction of potential targets

The structural diagrams of catechins – C, EC, GA, GCG, and PB2 were obtained via Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform database (https://www.tcmsp-e.com/). Subsequently, these structure diagrams were uploaded into PharmMapper database (https://lilab-ecust.cn/pharmmapper/index.html) to collect the targets of catechins.

2.2. Screening of DEG of insomnia regulated by catechins

Through Gene Expression Omnibus database (GEO) (https://www.ncbi.nlm.nih.gov/geo/), we collected data related to insomnia: GSE208668,^[25] GSE39445^[26], and GSE56931,^[27] which include 17, 216, and 126 insomnia samples, 25, 222, and 123 control samples, respectively. Differentially expressed genes targets (DEG) of insomnia (insomnia-DEG) were identified through GEO2R, applying the following screening criteria: P.adjust < .05, |log₂fold change (FC) |> 0.5. And then, we overlapped insomnia-DEG with the targets of catechins to get DEG1.

2.3. Screening of DEG of AD regulated by catechins

Through GEO database, we obtained data related to AD: GSE122063,^[28] GSE140829^[29], and GSE44770,^[30] which include 56, 129, and 129 AD samples, 247, 104, and 109 control samples, respectively. DEG of AD (AD-DEG) was identified through GEO2R, applying the following screening criteria: P.adjust < .05, |log₂fold change (FC) |> 0.5. And then, we overlapped AD-DEG with catechins targets to get DEG2.

2.4. Screening shared DEG of catechins anti-insomnia-AD related to circadian rhythm and immunity

We get circadian rhythm associated genes and immunity related genes from GeneCards database (https://www.genecards.org). And then, we employed a Venn diagram to identify the DEG related to insomnia, AD, circadian rhythm, immunity, and catechins targets, and the resulting genes was then designated as DEG3.

2.5. Screening key components

The catechins-DEG3 network was formed utilizing Cytoscape 3.7.1 software, and the degree values of each component were calculated using NetworkAnalyzer plugin. Subsequently, the components with the top 2 highest degree values were identified as key components.

2.6. ADME and toxicity predictions

ADMET properties include absorption, distribution, metabolism, excretion, and toxicity. The ADME and toxicity evaluation of the 2 key componets were predicted using ADMETlab (https://admet.scbdd.com/).

2.7. Enrichment analyses for GO and KEGG pathways enrichment

Gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) pathways were conducted utilizing Metascape database (https://metascape.org/gp/index.html),^[31] with a threshold for statistical significance P < .05. The main GO and KEGG pathways enrichment were visualized utilizing an online visualization platform (https://bioinformatics.com.cn/).

2.8. Identification of key candidate genes

Two machine learning methods, least absolute shrinkage and selection operator (LASSO), and random forest (RF) were applied to screen out key candidate genes. Two R packages “randomForest”^[32] and “glmnet”^[33] were utilized. And then, a Venn diagram and principal component analysis (PCA) were performed to screen out the key gene.

2.9. GSEA analysis

To assess the pathway activity of the key gene, Reactome pathway analysis was performed using gene set enrichment analysis (GSEA) software,^[34,35] adjusted P-value (P.adj) of <.05 and NES > 1 were set as the significance threshold.

2.10. Validation of ROC curve

To evaluated the association of key gene with insomnia-AD, the disease-control validation model was constructed grounded on GSE208668,^[25] GSE56931,^[27] GSE9442,^[36] GSE122063,^[28] GSE36980^[37] sets. GSE208668, GSE56931, and GSE9442 included 17, 126, 35 insomnia samples and 25, 123, and 36 control samples, respectively, while GSE122063 and GSE36980 include 56, 33 AD samples and 247, 47 control samples, respectively. The receiver operating characteristic (ROC) curve were charted, the area under the curve (AUC) was calculated with pROC package.^[38] The differential expression levels of key gene were illustrated through scatter plots using ggplot2 package.

2.11. Molecular docking

Three-dimensional structural models of the key candidate genes (PDE4D, HMOX1, and IGF1) were accessed from Protein Data Bank (PDB) database (https://www.rcsb.org/search), and then, they were removed small molecules and water, converted into PDBQT file using MGLTools (version 1.5.7). Two key components (C and EC) SDF format were accessed from PubChem database (https://pubchem.ncbi.nlm.nih.gov/), subsequently, they were processed into PDBQT files using MGLTools. Molecular dockings of PDE4D, HMOX1, and IGF1 with C and EC were conducted employing AutoDock Vina (version 1.1.2).

2.12. MD simulations

MD simulations were conducted using Gromacs 2022 program. The AMBER14SB force field, and TIP3P water model, were used in this study. At 1 bar, berendsen method was employed to regulate the pressure. At 298 K, 100 ps of NVT, and NPT equilibrium, simulations were performed, followed by 100 ns of MD simulation. The trajectory of the simulation was examined using VMD and PyMOL (version 2.6), the analysis of binding free energy was performed utilizing g_mmpbsa software.

3. Results

3.1. Identification of potential targets for catechins anti-insomnia-AD

In this study, we discovered 418 targets related to the 5 catechins utilizing PharmMapper database. Furthermore, we gathered targets linked to insomnia-AD from GEO database, yielding 10,712 insomnia-DEG and 7552 AD-DEG. A total of 2708 circadian rhythm-related genes and 50,000 immunity-related genes were collected from GeneCard database. We identified 133 DEG1 and 80 DEG2, respectively (Fig. 2A, B). Figure 2D, E showed the hotmaps of DEG1 and DEG2, respectively. Subsequently, we uncovered 17 shared genes (DEG3; Fig. 2C).

Figure 2.

Analysis of DEG of insomnia-Alzheimer’s disease (AD) regulated by catechins. (A)Venn diagrams of catechins genes and insomnia-DEG. (B) Venn diagrams of catechins genes and AD-DEG. (C)Venn diagrams of catechins genes, DEG1, DEG2, circadian rhythm genes, and immunity genes. (D) Hotmap of DEG1. (E) Hotmap of DEG2. AD = Alzheimer’s disease, DEG = differentially expressed gene.

3.2. Screen out key components

We developed catechins-DEG3 network (Fig. 3A). Remarkably, all catechins showed interactions with various targets. The degree value represented the number of connections and significance. Among the catechins, C registered the highest degree value (15), succeeded by EC (14), GCG (10), PB2 (7), and GA (6) having the lowest. Ultimately, our findings suggest that C and EC may be pivotal catechins.

Figure 3.

Catechins–DEG3 network, gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis of DEG3. (A) Catechins–DEG3 network diagram. (B) GO analysis of DEG3. (C) KEGG pathway analysis of DEG3. DEG = differentially expressed gene, GO = gene ontology, KEGG = kyoto encyclopedia of genes and genomes.

3.3. Analysis of ADME and toxicity

The absorption, distribution, metabolism, excretion (ADME), and toxicity properties of C and EC were assessed utilizing ADMETlab (https://admet.scbdd.com/). The results are presented in Table 1.

Table 1.

The pharmacokinetic parameters of C and EC.

	C	EC
LogPapp (Caco-2 permeability)	−6.652	−6.25
Pgp-inhibitor	Yes	No
Pgp-substrate	No	No
HIA (human intestinal absorption)	No	No
F (20% bioavailability)	Yes	No
F (30% bioavailability)	No	No
PPB (plasma protein binding)	87.686	93.86
BBB (blood–brain barrier)	Yes	Yes
CYP450 1A2 inhibitor	No	No
CYP450 1A2-substrate	No	No
CYP450 3A4 inhibitor	No	No
CYP450 3A4 substrate	No	No
CYP450 2C9 inhibitor	No	No
CYP450 2C9 substrate	No	No
CYP450 2C19 inhibitor	No	No
CYP450 2C19 substrate	No	No
CYP450 2D6 inhibitor	No	No
CYP450 2D6 substrate	No	Yes
T 1/2 (half life)	1.677	0.72
CL (clearance)	1.137	1.914
hERG (hERG blockers)	Yes	No
H-HT (human hepatotoxicity)	No	No
Ames (ames mutagenicity)	Yes	No
SkinSen (skin sensitization, [r]LLNA)	No	No

BBB = blood–brain barrier, C = catechin, CL = clearance, EC = epicatechin, HIA = human intestinal absorption, H-HT = human hepatotoxicity, PPB = plasma protein binding.

3.4. GO and KEGG pathway enrichment

To comprehensively investigate catechins’ systemic impact on insomnia-AD, functional annotation was conducted for DEG3. GO enrichment indicated these genes mainly participant in nuclear receptor-mediated signaling pathway, cellular response to lipid, regulation of phosphatidylinositol 3-kinase (PI3K)/protein kinase B signal transduction and response to steroid hormone (Fig. 3B). KEGG enrichment analysis highlighted endocrine resistance, MAPK signaling pathway, PI3K-Akt signaling pathway and Chemical carcinogenesis - receptor activation (Fig. 3C).

3.5. Machine learning-driven identification of 3 key candidate genes

To identify key candidate genes, we employed LASSO and RF algorithms on DEG3. Based on insomnia dataset GSE208668, 6 candidate genes were identified utilizng LASSO algorithm (Fig. 4A), 16 candidate genes were screened out via RF algorithm (Fig. 4B, C). Similarly, in AD dataset GSE122063, 8 candidate genes were screened out utilizng LASSO algorithm (Fig. 4D), 15 candidate genes were identified via RF algorithm (Fig. 4E, F). Within these 4 cohorts, a Venn diagram illustrated 3 key candidate genes: PDE4D, HMOX1, and IGF1 (Fig. 4G).

Figure 4.

Screen out 3 key candidate genes. (A, D) Least absolute shrinkage and selection operator (LASSO) analysis screen out key candidate genes of catechins in insomnia and AD treatment, respectively. (B, C and E, F) Random forest (RF) analysis screened out key candidate genes of catechins in insomnia and AD treatment, respectively. (G) Venn diagram illustrate the intersection of candidate genes. AD = Alzheimer’s disease, LASSO = least absolute shrinkage and selection operator, RF = random forest.

3.6. Validation of ROC

In order to further confirm the crucial role of PDE4D, HMOX1, IGF1 within insomnia-AD, ROC analysis were employed. The AUC value within 0.5–1, indicates the gene had good predictive ability. As Figure 5A, C, and E showed that these 3 key candidate genes possesse a good diagnostic ability for insomnia-AD (AUC > 0.5). Furthermore, analysis of expression levels suggested that these 3 key candidate genes play a significant role in both insomnia and AD (Figure 5B, D, and F).

Figure 5.

ROC and expression level analysis. (A) Receiver operating characteristic (ROC) curve validation of PDE4D in insomnia-AD. (B) Expression level of PDE4D in insomnia-AD. (C) ROC curve validation of HMOX1 in insomnia-AD. (D) Expression level of HMOX1 in insomnia-AD. (E) ROC curve validation of IGF1 in insomnia-AD. (F) Expression level of IGF1 in insomnia-AD. AD = Alzheimer’s disease, ROC = receiver operating characteristic.

3.7. PED4D was identified as key gene and GSEA analysis

To identify the key gene, we applied PCA statistical technique. GSE9442^[36] (insomnia) and GSE36980 (AD)^[37] datasets were used for validation. GSE9442 included 35 insomnia samples and 36 control samples, while GSE36980 comprising 33 AD samples and 47 control samples. The cumulative variance contribution rates of principal component 1 (PC1) were 38.1% and 51.9% in insomnia and AD dataset, respectively (Fig. 6A, B). Notably, in insomnia dataset, HMOX1 exhibited the highest weight (0.7094) , PDE4D was the second (0.6667). In AD dataset, PDE4D was the highest (0.6757). We identified PDE4D as key gene due to its consistent upregualted expression trend in both datasets. GSEA analysis revealed that complement cascade is positively enriched in both insomnia and AD (Fig. 6C, D).

Figure 6.

The key gene was identified via PCA and its GSEA analysis. (A) Principal component analysis (PCA) analysis of 3 key candidate genes in insomnia dataset. (B) PCA analysis analysis of 3 key candidate genes in AD dataset. (C, D) Complement cascade in insomnia and AD dataset. AD = Alzheimer’s disease, PCA = principal component analysis.

3.8. Validation of molecular docking

Molecular docking was used to evaluate affinity constants. The docking scores of C and EC in relation to PDE4D, HMOX1 and IGF1 ranged from −5.4 to −8.3 kcal/mol (Fig. 7A–F). Values below −5.0 kcal/mol indicate possible binding, while those falling under −7.0 kcal/mol suggest robust binding. Therefore, C and EC showcased a stronger binding affinity towards PDE4D, HMOX1, and IGF1. For PDE4D, C forms hydrogen bonds with ASP318, ASN321, THR333, and GLN369, establishes Pi-Sulfur, Pi-Pi Stacked, and Pi-Alkyl hydrophobic interactions with MET273, PHE372, and ILE336, respectively, constructs van der Waals interactions with MET357, LEU319, and TRP332 (Fig. 7A). EC forms hydrogen bonds with GLN369, ASP318, TYR159, ASN321, and THR333, establishes Pi-Alkyl hydrophobic interactions with ILE336 and LEU319, constructs van der Waals interactions with PHE340, TYR329, and HIE164 (Fig. 7B). For HMOX1, C forms hydrogen bonds with ASP140, GLN38, THR135, establishes Pi-Sulfur, Pi-Pi Stacked, Pi-Alkyl, and Pi-Cation hydrophobic interactions with MET34, PHE207, ILE147, and ARG136, constructs van der Waals interactions with GLY143, VAL146, and GLY139 (Fig. 7C). EC forms hydrogen bonds with ARG136 and ASP140, constructs van der Waals interactions with MET34, GLY139, and LEU147(Fig. 7D). For IGF1, C forms hydrogen bonds with GLY7, PHE25, GLN40, and VAL44 establishes Pi-Sigma, constructs van der Waals interactions with VAL11, ASN26, GLY42, ILE43, ASP45, THR421, and LEU10 (Fig. 7E). EC forms hydrogen bonds with TYR24, MET59, and CYS61, constructs van der Waals interactions with LYS27, PRO28, ARG55, and ARG21 (Fig. 7F).

Figure 7.

Molecular docking analysis. (A) Catechin (C)-PDE4D. (B) Epicatechin (EC)-PDE4D. (C) C-HMOX1. (D) EC-HMOX1. (E) C-IGF1. (F) EC-IGF1. C = catechin, EC = epicatechin.

3.9. Validation of MD

To assess the complex stability of PDE4D with C and EC, MD simulations were performed. The root mean square deviation (RMSD) and root mean square fluctuation (RMSF) are primarily used to assess the qualitative stability. As illustrated in Figure 8A, the RMSD values of PDE4D-C and PDE4D-EC complexes rapidly stabilized within 1 nm, indicating the relative stability. RMSF values also indicated the stable binding pocket (Fig. 8B) . The tightness of complexes was evaluated using radius of gyration (Rg). As Figure 8C showed the tightness of these complexes remained consistent. The solvent accessible surface area (SASA) serves as a key indicator for conformational stability. As shown in Figure 8D, the SASA values averaged 7.5 nm², exhibiting conformational stability. The distances between C, EC and the center of PDE4D gradually stabilized, indicating progressively stable binding (Fig. 8E). Analysis of the simulation trajectory revealed C and EC consistently bind to PDE4D (Fig. 8F). Figure 8G showed the hydrogen bonds number, demonstrating the binding stability. Furthermore, the binding free energy (ΔG_binding) was determined through MM/GBSA method. As illustrated in Figure 8H, the ΔG_binding for the 2 complexes were −13.117 ± 4.979 kJ/mol (PDE4D-C), and −51.691 ± 1.022 kcal/mol (PDE4D-EC), respectively, suggesting the strong binding, which aligns with the molecular docking results. Overall, the complexes of C and EC with PDE4D exhibited favorable kinetic properties.

Figure 8.

Molecular dynamics (MD) simulations of PDE4D with C and EC. (A) Root mean square deviation (RMSD) results. (B) root mean square fluctuation (RMSF) results. (C) Rg results. (D) Results of SASA. (E) The distances between C, EC, and the center of PDE4D. (F) Analysis of the simulation trajectory. (G) The hydrogen bond count. (H) The binding free energy (ΔG_binding). C = catechin, EC = epicatechin, MD = molecular dynamics, RMSD = root mean square deviation, RMSF = root mean square fluctuation, SASA = solvent accessible surface area.

4. Discussion

Insomnia, particularly in conjunction with AD, is gradually acknowledged as a significant public health problem globally.^[12,39] Hence, the development of an effective treatment for patients with insomnia and AD comorbidity is extremely important. Catechins, a polyphenol existing by nature, has demonstrated capability as a multi-purpose therapeutic agent,^[14-16] However, there is scant research investigating catechins on insomnia-AD comorbidity management. To bridge the gap, the present study employed tripartite computational strategy integrating network pharmacology, molecular docking, and MD simulations to elucidate the mechanisms responsible for the therapeutic effects of catechins on insomnia-AD.

C and EC were identified as pivotal compounds with the most targets in our research, highlighting the fundamental role of the basic monomer skeletons in insomnia-AD comorbidity. While GCG, GA, and PB2 represent distinct structures, their topological positions suggest complementary, rather than overarching, mechanisms. This underscores the necessity of our comparative approach, revealing a potential collaborative strategy within the catechin family. And study showed that C effectively restore circadian rhythm,^[40] regulate inflammatory response through activation of nuclear factor-kappa B,^[41] reduce pro-inflammatory cytokines production,^[42] Additionally, C can prevent spatial learning and memory deficits by reducing amyloid beta oligomers and boosting synaptic plasticity mediators.^[41] EC modulates circadian rhythms, alters mRNA and protein levels of clock genes,^[43] enhances immunity,^[44] offsets the negative effects of sleep deprivation in women,^[45] inhibits Aβ aggregation.^[46] These evidences suggest that C and EC, or their better derivatives are promising polyphenolic candidates for insomnia-AD comorbidity. Although the ADMET prediction of this study indicated that C and EC face challenges of low oral bioavailability – a commonality shared with most natural polyphenolic compounds – this finding, rather than diminishing their value, precisely charts a course for future formulation research. The clear pharmacological mechanisms we predicted provide a compelling rationale for investing in advanced delivery systems (e.g., nanoformulations) to overcome this limitation. Thus, combining the multi-target activities revealed by computation with innovative formulation techniques emerges as a viable and effective strategy for translating these natural compounds into clinical therapies.

PDE4D has been identified as key target via machine learning and PCA. PDE4D is 1 subtype of Phosphodiesterase4 (PDE4) enzymes, the isoform enriched in the central nervous system, with predominant localization within hippocampus, it regulates immune-inflammatory responses and circadian rhythm by modulating cyclic adenosine monophosphate/cyclic guanosine monophosphate signaling.^[47,48] There is report showed PDE4D has relation with sleepiness,^[49] its role in insomnia is not very clear. PDE4D has been regarded as a target in the treatment of AD. Morever, there is limited research about the relationship of PDE4D in insomnia-AD. Even so, it was admitted that PDE4D is the mediator of sleep deprivation induced memory impairments,^[50] insufficient sleep compromises the stability of long-term memory and overall cognitive performance through upregulating PDE4D expression.^[51] Most research reports showed PDE4D inhibitors are effective in enhancing the cognition and memory of AD,^[52] it significantly ameliorate cognitive dysfunction induced by sleep deprivation.^[51] Theses reports highlighted PDE4D’s potential application in treating insomnia-AD comorbidity. In our current research, PDE4D was upgulated in both insomnia and AD patients, based on the previous research reports, it seems that C and EC might be PDE4D inhibitor. HMOX1, also known as heme oxygenase 1(HO-1), catalyzes the conversion of free heme to biliverdin and iron, with the concomitant release of carbon monoxide. It is the oxidative markers, related to chronic sleep deprivation,^[53] and the pathological characteristics of AD,^[54] furthermore, there is limited research about the relationship between HMOX1 and circadian rhythm in insomnia-AD. IGF1 is a protein similar to insulin in function and structure and is a member of a family of proteins involved in mediating growth and development. It binds to its receptor (IGF1R), and then to transmits signals to PI3K/Akt pathway, which is the shared pathway of insomnia-AD.^[55,56]

And then, we detect the binding affinity of C and EC with PDE4D utilizing computational simulations. The findings from molecular docking suggested that the docking scores of C and EC concerning PDE4D consistently remained below −7 kcal/mol, which indicates a strong binding ability. Subsequently, MD simulations were performed to examine the temporal dynamic behavior and stability of PDE4D-C and PDE4D-EC complexes over time. RMSD, RMSF, Rg, and SASA, along with the number of hydrogen bonds, suggested that the 2 complexes exhibited robust structural integrity and maintained compactness. Furthermore, ΔG_binding of the 2 complexes demonstrating a strong binding affinity that corroborated the molecular docking results. Thus, we confirmed C and EC may exert their anti-insomnia-AD effects through interactions with PDE4D, and downregulate its expression.

To understand PDE4D function in insomnia and AD, GSEA analysis was performed. Reactome pathway analysis indicated that the complement cascade is significantly enriched. The complement system serves as a critical safeguard against invading microorganisms by identifying and eliminating pathogens, apoptotic cells, and cellular debris. complement system was triggered via 3 principal pathways: the classical, alternative, and lectin pathway.^[57] Complement component 1q (C1q) is the initiator in the classical pathway, C3 is the central effector molecule of the 3 pathways.^[58] Complement system acts as both a trigger and a contributor to synaptic loss. Mice lacking C1q, C4, or C3 exhibit deficits in synaptic pruning, resulting in synapses excess.^[59,60] Through the classic complement cascade, microglia phagocytos synapses to remove inappropriate connections.^[61] It was reported that sleep restriction triggered complement C3-C3aR axis-mediated microglial phagocytosis of synapses, and microglia-mediated impaired synaptic elimination contributes to cognitive deficits, while the sedative dexmedetomidine reversed the phenomenon.^[61] A recent study showed that PDE4 inhibitor suppresses reactive microglia by inhibiting HMGB1/C1q/C3 pathway and reduces synaptic pruning by microglia.^[58] Thus, we speculated that C and EC may enhance synaptic pruning via interacting at PDE4D and downregulate its expression, which was involved in complement cascade.

5. Conclusions

In this current research, with network pharmacology and computer simulation techniques, we proposes a novel mechanistic model wherein C and EC alleviate insomnia-AD comorbidity, C and EC exhibited the broadest target spectrum, establishing them as foundational modulators. C and EC ameliorate comorbid pathologies by co-regulating complement cascade, and PI3K/AKT pathway, via HMOX1, IGF1, and PDE4D. MD simulations confirmed C and EC stable and high-affinity binding to the pivotal target PDE4D. MD simulations highlights the potential of these catechins as promising lead compounds. These findings deepen our comprehension of catechins. Furthermore, the identified ADMET profile, while indicating a common challenge of limited bioavailability for natural polyphenols, precisely charts a course for future formulation research to enhance their therapeutic potential. Further experimental researches, particularly those integrating innovative delivery systems, are essential to validate these mechanisms.

Author contributions

Conceptualization: Yuxuan Liu, Shijun Wang.

Formal analysis: Yuxuan Liu, Jiaxiu Han.

Investigation: Yuxuan Liu, Jiaxiu Han.

Software: Yuxuan Liu.

Writing – original draft: Yuxuan Liu.

Writing – review & editing: Yuxuan Liu, Shijun Wang.