Experimental Study on Anxiolytic and Antidepressant Effects of Zhi-Chi-Jie-Yu Granules in CUMS Mice and Prediction of Underlying Molecular Mechanisms

Abstract

Objective

This study investigated the in‑vivo pharmacological effects of Zhi-Chi-Jie-Yu Granules(ZCK) on CUMS‑induced anxiety‑ and depression‑like behaviors in mice. Post‑hoc exploratory analyses including network pharmacology and molecular docking were further performed to generate potential mechanistic hypotheses for its anti‑depressant action.

Methods

An anxiety‑depression model was established in ICR mice via chronic unpredictable mild stress (CUMS) combined with solitary housing. Mice were divided into six groups (n=10): blank control, model, fluoxetine, and ZCK low‑, medium‑, and high‑dose groups. Behavioral changes were assessed by secondary‑auxiliary (SPT), together with FST, TST, OFT and EPM. Serum inflammatory cytokines and hippocampal neurotransmitters (5‑HT, NE) were measured by ELISA, and hippocampal pathological damage was observed by HE staining. UPLC‑Q‑TOF‑MS was used to characterize the chemical profile of ZCK. Non‑targeted serum metabolomics, network pharmacology, and molecular docking were integrated to predict targets and pathways.

Results

ZCK significantly alleviated CUMS‑induced depressive‑ and anxiety‑like behaviors (P < 0.05 or P < 0.01). It downregulated serum pro‑inflammatory cytokines (IL‑6, TNF‑α), increased hippocampal 5‑HT and NE levels, and attenuated neuronal damage. UPLC‑Q‑TOF‑MS identified 39 core components. Metabolomics revealed 56 differential metabolites, mainly enriched in alanine‑aspartate‑glutamate and arachidonic acid metabolism. Network pharmacology predicted 410 overlapping targets, and the top‑10 core targets contained TP53, AKT1, STAT3, SRC, TNF and others. Considering our experimental findings on neuroinflammation, inflammatory‑related key targets (SRC, TNF, IL-6, NF‑κB and IL-1B) were selected for subsequent molecular‑docking verification. Molecular‑docking simulations suggested potential binding interactions between core components and key targets (binding energies <−5.0 kcal/mol).

Conclusion

ZCK exerts anti‑anxiety and antidepressant effects in CUMS mice, accompanied by reduced peripheral neuroinflammation and improved hippocampal neurotransmitter levels. Multi‑omics and network‑pharmacology analyses further predict potential multi‑dimensional regulatory candidates, including neuroinflammation‑related signaling, neurotransmitter homeostasis and metabolic perturbation. These computational hypotheses require further protein‑level experimental validation. This study provides preliminary in‑vivo evidence for ZCK and an integrated methodological reference for TCM compound research.

Background

Anxiety and depression (AD) are highly prevalent mental disorders in young and middle-aged populations, closely associated with personality, environmental, genetic, and psychological factors.^1,2 The global prevalence of AD ranges from 7.3% to 28.0%, with a heritability rate of 30–50%. In China, the incidence of AD continues to rise and ranks first among mental disorders.³ The pathogenesis is related to neurotransmitter imbalance, endocrine disorders, and immune dysfunction.^4,5 First-line clinical drugs such as serotonin reuptake inhibitors and norepinephrine reuptake inhibitors have limitations including slow onset, high recurrence rate, poor compliance, and adverse reactions such as nausea and vomiting, resulting in unsatisfactory therapeutic effects.⁶ Therefore, exploring highly effective and low-toxic anti-anxiety and antidepressant drugs from TCM has important practical value.

In traditional Chinese medicine, anxiety and depression are categorized as “Depression Syndrome” and “Zangzao”. Its pathogenesis involves dysfunction of the heart, liver, spleen and gallbladder. Treatments focus on soothing the liver, clearing heat and tonifying the heart and spleen.^7,8 Zhi-Chi-Jie-Yu Granules (ZCK) are modified from Zhi Zi Chi Tang in Treatise on Cold Damage by Zhang Zhongjing⁹ and the heat-clearing and fire-purging herbs (Lophatherum gracile, Gardenia jasminoides, and Cyathula officinalis) in Direct Decisions on Pediatric Medicines and Syndromes by Qian Yi. In this formula, Cyathula officinalis is replaced by four medicine–food homologous herbs: Nelumbo nucifera seed heart, Poria cocos, Glycyrrhiza uralensis, and raw Ostreae Concha. ZCK has the effects of clearing heat, detoxifying, relieving restlessness, and resolving depression. In the formula, Gardenia jasminoides clears fire and relieves restlessness;¹⁰ Sojae Semen Preparatum relieves stagnation and restlessness;¹¹ Nelumbo nucifera seed heart clears the heart and calms the mind;¹² Lophatherum gracile clears heat and purges fire;¹³ Poria cocos calms the mind and enhances immunity;¹⁴ Ostreae Concha sedates and calms the spirit;¹⁵ Glycyrrhiza uralensis harmonizes all ingredients, detoxifies, and reduces inflammation.¹⁶ The synergistic action of multiple components fully reflects the advantages of TCM in multi-target and multi-pathway regulation. Modern pharmacological studies further demonstrate that these constituent herbs exert neuroprotective and anti‑inflammatory bioactivities, which lay a material foundation for the anxiolytic‑antidepressant effects of ZCK granules.

Accordingly, we hypothesized that ZCK Granules might alleviate anxiety‑depressive‑like behaviours through potentially modulating neuroinflammation, restoring monoamine neurotransmitters and correcting peripheral metabolic disorders. In this study, a CUMS combined with social isolation mouse model was used to simulate anxiety and depression.^17,18 Behavioral tests, ELISA detection of inflammatory factors and neurotransmitters, HE staining, UPLC-Q-TOF-MS component identification as well as serum metabolomics constituted our primary in‑vivo experimental validation. Subsequent post‑hoc were integrated to explore potential regulatory candidates and generate mechanistic hypotheses for ZCK. This study provides preliminary in‑vivo experimental evidence and computational hypothetical reference for the clinical application and further mechanism exploration of ZCK.

Experimental Materials

Experimental Animals

Seventy-two SPF-grade male ICR mice (4 weeks old, weighing 20–24 g) were purchased from Hunan Slaike Jingda Laboratory Animal Co., Ltd. (Certificate No.: 430727241102430313). Mice were housed in a standardized environment with a temperature of (22 ± 1) °C, relative humidity of (55 ± 5)%, and free access to food and water. After 1 week of acclimatization, the experiment was initiated. All animal procedures were approved by the Laboratory Animal Ethics Committee of Wuhan Hualianke Biotechnology Co., Ltd. (Approval No.: HLK-20240927-001) and performed in accordance with the National Research Council (NRC) Guide for the Care and Use of Laboratory Animals and the 3R (Replacement, Reduction, Refinement) principles for laboratory animal welfare. The reporting of this study complies with the ARRIVE guidelines.

Drugs and Reagents

Zhi-Chi-Jie-Yu Granules (Batch No.: 20240926) were prepared in the laboratory of Hubei University of Chinese Medicine. Fluoxetine Hydrochloride Dispersible Tablets (Batch No.: 21518A) were purchased from Eli Lilly Suzhou Pharmaceutical Co., Ltd. Sucrose, xylene, anhydrous ethanol, eosin Y, embedding paraffin, and neutral gum were purchased from Sinopharm Chemical Reagent Co., Ltd. Universal tissue fixative was purchased from Servicebio. Hematoxylin was purchased from Sigma. ELISA kits for IL-6 (Cat. No. MU30044), IL-1β (Cat. No. MU30369), TNF-α (Cat. No. MU30030), 5-HT (Cat. No. MU30036), and NE (Cat. No. MU30072) were purchased from Wuhan Beiyinlai Biotechnology.

Experimental Equipment

Ice maker (Model FMB50); benchtop low-speed centrifuge (Model JIDI-20D); high-speed refrigerated centrifuge (Model TGL-16M); microplate reader (Model Synergy 2); −80°C freezer (Model DW-86L416G); animal behavior analysis system (KEMAZE); open-field apparatus (Model KW-OPF); tail suspension apparatus (Model KW-XW); elevated plus maze (Model KW-GJ); plantar stimulation apparatus (Model KW-CJ01); strobe light (100 W); 50 mL restraint tubes and vibration frame (homemade); tissue dehydrator (Model Donatello); embedding machine (Model JB-P5); cooling stage (Model JB-P5); rotary microtome (Model RM2016); tissue spreading machine (Model KD-P); oven (Model GFL-230); adhesive slides, embedding frames, and embedding molds; cover slips; biological microscope (Model ECLIPSE Ci-L plus).

Experimental Methods

Preparation of Zhi-Chi-Jie-Yu Granules

Gardenia jasminoides 12 g, Sojae Semen Preparatum 12 g, Nelumbo nucifera seed heart 12 g, Lophatherum gracile 6 g, Poria cocos 6 g, Glycyrrhiza uralensis 3 g, and Raw Ostreae Concha 30 g (calcined before use) were used. Ostreae Concha was refluxed with 10 volumes of water for 1 h, filtered, and mixed with anhydrous ethanol to a final concentration of 60%, then precipitated at 4 °C for 24 h to obtain Solution 1. The other six herbs were extracted twice with 70% ethanol (solid–liquid ratio 1:10, 2 h each time). The filtrates were combined with Solution 1, concentrated into a thick paste, and freeze-dried into dry extract powder. The powder was mixed with excipients (soluble starch: corn dextrin: sucrose = 1:2:2) at a ratio of 1:2, prepared into a soft mass using 70% ethanol as a wetting agent, granulated through a 14-mesh sieve, dried at 60 °C for 30 min, and sized by sieving sequentially through 10-mesh and 80-mesh sieves. Finally, granules between 10 and 80 mesh were collected as the finished Zhi-Chi-Jie-Yu Granules.

CUMS Model Establishment and Grouping

A modified chronic unpredictable mild stress (CUMS) combined with solitary housing model was used. Seventy-two ICR mice were randomly divided into six groups: control group, model group, fluoxetine group (3.03 mg/kg), and ZCK low-dose group (ZCK-L, 1.365 g/kg/d), ZCK medium-dose group (ZCK-M, 2.730 g/kg/d), and ZCK high-dose group (ZCK-H,5.460 g/kg/d). The clinically recommended dose of fluoxetine is 20 mg/60 kg/d, and the equivalent dose for mice converted by body surface area is 3.03 mg/kg. The raw medicinal dosage of the original prescription of Zhichi Jieyu Granule is 81 g, and approximately 18 g granules are obtained after granulation. The equivalent dose for mice calculated based on body surface area is 2.73 g/kg/d. Except for the control group, all mice received one 24 h long‑term stimulus and one short‑term stimulus daily. Long-term stimuli included food deprivation, water deprivation, 45° cage tilt, empty cage, damp bedding, white noise (90 dB), strobe light, and strange odor. Short-term stimuli included 0.7 mA electric shock, cage shaking (2 h), restraint (6 h), ice-water swimming (4 °C, 5 min), heat exposure (55 °C, 10 min), and tail clamping (3 min). The same stimulus was not repeated within 2 days. Modeling was performed for 18 consecutive days, followed by 18 days of modeling combined with intragastric administration. The control and model groups were given an equal volume of distilled water. Body weight was measured every 6 days. The establishment of CUMS depression‑anxiety model was primarily validated by prolonged immobility time in FST and TST, rather than relying on sucrose preference test (SPT). SPT was performed as a secondary auxiliary behavioral index for model evaluation. Detailed procedures for each stimulus are shown in Table 1, and the daily schedule of CUMS modeling is presented in Table 2.

Table 1.

CUMS Modeling Stimuli

Type	Code	Stimulus	Detailed Procedure
Long-term stimulus (24h)	A	Food deprivation	Food was withheld.
	B	Water deprivation	Water was withheld.
	C	Cage tilting	The cage was placed at a 45° angle.
	D	Empty cage	Bedding material was completely removed.
	E	White noise	White noise was played continuously at 90 dB using a loudspeaker.
	F	Stroboscopic light	A 100 W strobe light was used to flash at 25 Hz in a dark environment.
	G	Unpleasant odor	Stimulants with strong odor (camphor balls, peppermint oil, onion, etc) were randomly applied.
	H	Damp bedding	200 mL of water was added into the cage with bedding.
Short‑term stimulus	I	Restraint stress	Several small holes were punched on a 50 mL centrifuge tube. Mice were placed head‑down into the tube with tails exposed, and kept for 6 h.
	J	Ice‑water swimming	Mice were individually placed in ice water (4 °C) and forced to swim for 5 min.
	K	Heat exposure	Mice were placed in an oven at 55 °C for 10 min.
	L	Cage shaking	Mice were placed on a homemade vibration frame for 2 h.
	M	Tail clamping	The tail was clamped 1 cm from the tip with a binder clip for 3 min.
	N	Electric foot shock	Mice were placed in a plantar stimulator and received 0.7 mA electric shocks (5 s per trial, 10 trials with 1 s interval).

Table 2.

CUMS Modeling Protocol

Code	Day 1	Day 2	Day 3	Day 4	Day 5	Day 6
1	A+I	C+K	E+M	F+N	D+L	G+J
2	B+I	D+M	C+J	E+N	F+L	D+M
3	A+I	F+J	H+N	G+M	C+L	D+K
4	F+L	E+J	B+M	C+N	D+K	G+J
5	A+I	E+N	D+K	E+L	C+J	E+M
6	B+I	H+M	G+J	C+N	D+K	E+L

Notes: The codes indicate that the experiment was divided into six segments with body weight measured six times. Modeling alone was performed in the first three segments, while modeling plus drug administration was conducted in the last three segments. A: 24 h food deprivation; B: 24 h water deprivation; C: 45° cage tilt; D: 24 h empty cage; E: 24 h white noise (90 dB); F: 24 h stroboscopic light; G: 24 h unpleasant odor; H: 24 h damp bedding; I: 6 h restraint; J: 5 min ice‑water swimming; K: 10 min heat exposure at 55 °C; L: 2 h cage shaking; M: 2 min tail clamping; N: electric foot shock.

Animals were closely monitored for body weight, fur appearance, locomotor activity and physical injury throughout the experiment. No obvious trauma, thermal injury, severe sickness, or mortality occurred during the experimental procedure.

Sample sizes for different measurements were as follows: 12 mice per group were initially enrolled. Two animals in each group were excluded before behavioural assays based on pre‑defined exclusion criteria (abnormal baseline body weight, sucrose‑preference value beyond ±2 standard deviations of group average), resulting in n = 10 mice per group for behavioural tests and ELISA assays. For untargeted metabolomics analysis, n = 6 representative biological samples per group were selected from the 10 eligible samples for detection.

Body Weight Monitoring

Body weight was measured every 6 days at a fixed time (8:00 AM). Food and water deprivation were avoided 24 h before measurement to eliminate interference.

Behavioral Tests

Sucrose Preference Test (SPT)

Mice were acclimated to 1% sucrose solution and pure water for 48 h, with bottle positions swapped every 12 h to eliminate position preference. After 12 h of food and water deprivation, the bottles were weighed, and consumption was measured over 24 h. Sucrose preference rate was calculated as: sucrose intake/(sucrose intake + water intake) × 100%.¹⁹

Forced Swimming Test (FST)

Mice were placed in a transparent cylinder (15 cm water depth, 23–25 °C), where hind paws and tail could not touch the bottom. After 6 min of adaptation, immobility time (floating without struggling, minimal movements to keep head above water) during the last 5 min was recorded and analyzed using SuperFst software.

Tail Suspension Test (TST)

Mice were suspended 5 cm from the tail tip for 6 min, with the tail inserted into a 5 cm hollow plastic tube and fixed with tape. Immobility time during the last 5 min was recorded and analyzed using SuperFst software.

Open Field Test (OFT)

Mice were acclimated to the test room for 30 min before testing, then placed in the center of a 50 × 50 × 40 cm box (central zone: 25 × 25 cm). After 5 min of free exploration, total moving distance and central zone residence time were recorded via KEMaze software.²⁰

Elevated Plus Maze (EPM)

The EPM consisted of two open arms (5 × 30 cm, no walls), two closed arms (5 × 30 cm, with walls), and a central platform (5 × 5 cm). Mice were placed on the central platform facing an open arm and allowed to explore for 5 min. The percentage of open arm residence time was calculated based on video recordings analyzed with KEMaze software.²¹

Sample Collection

After behavioral tests, mice were fasted for 12 h. After anesthetizing mice with 5% isoflurane (KW-MZJ-1, NJKEWBIO), their blood was collected from the orbital venous plexus and centrifuged (4 ℃, 1000 × g, 15 min), and serum was stored at −80 ℃ for later testing.°C for subsequent detection. The mice were sacrificed by cervical dislocation, and whole brain tissues were rapidly isolated. A portion of brain tissue was fixed in 4% paraformaldehyde for 24 h for pathological examination. The hippocampal tissues were dissected on ice and stored at −80 °C for subsequent biochemical analysis.

HE Staining

Brain tissues were fixed in 4% paraformaldehyde for 24 h, then dehydrated, embedded in paraffin, sectioned, dewaxed, stained with hematoxylin and eosin, dehydrated again, and mounted. Pathological changes in the hippocampal CA1 region were observed under a microscope.

ELISA Analysis

Hippocampal tissues were homogenized in pre-chilled PBS containing protease inhibitors at a ratio of 1 mg:9 mL (tissue: PBS, w/v), homogenized with grinding beads at low temperature, and centrifuged at 10,000×g for 10 min at 4 °C. Supernatants were collected to detect the levels of 5-HT and NE according to the kit instructions. Serum levels of TNF-α, IL-6, and IL-1β were also measured following the kit instructions.

Serum Metabolomics

Sample Preparation

100 μL of serum was pipetted into 1.5 mL EP tubes, and 400 μL of methanol–acetonitrile (1:1, v/v) containing isotope internal standards was added. After vortexing for 30 s, sonication in ice water for 10 min, precipitation at −40 °C for 1 h, and centrifugation at 12,000 rpm for 15 min at 4 °C, supernatants were collected for UPLC-Q-TOF-MS analysis. Quality control (QC) samples were prepared by mixing equal volumes of all supernatants to evaluate the system stability.

UPLC-Q-TOF-MS Conditions

UPLC: Vanquish ultra-high performance liquid chromatograph was used with a Waters ACQUITY UPLC BEH Amide column (2.1 mm × 50 mm, 1.7 μm); mobile phase A: aqueous solution containing 25 mmol/L ammonium acetate + 25 mmol/L ammonia water; mobile phase B: acetonitrile; injection volume: 2 μL; sample tray temperature: 4 °C. MS: Orbitrap Exploris 120 mass spectrometer was used to collect primary and secondary mass spectral data under the control of Xcalibur 4.4 software; parameters: sheath gas 50 Arb, auxiliary gas 15 Arb, capillary temperature 20 °C, resolution 60,000 (MS1) and 15,000 (MS2), collision energy SNCE 20/30/40, spray voltage 3.8 kV (+) and −3.4 kV (-).

Data Analysis

Data were analyzed using SPSS 26.0 and GraphPad Prism 9, and expressed as x±s. Metabolites were identified using HMDB, Lipidmaps v2.3, Metlin, and a self-built database. OPLS-DA models were established, and 7 permutation tests were used to verify model robustness. Differential metabolites were screened with VIP > 1 and P < 0.05. Metabolic pathways were analyzed using MetaboAnalyst 5.0.

Component Identification and Network Pharmacology

UPLC-Q-TOF-MS was used to qualitatively identify the chemical components of ZCK. Compound SMILES structures were acquired from PubChem, and compound‑target prediction was performed based on TCMSP and SwissTargetPrediction databases; anxiety-depression-related targets were retrieved from GeneCards and OMIM databases. Overlapping targets of drugs and diseases were imported into STRING to construct a PPI network. Core targets were screened using CytoHubba plugin in Cytoscape 3.9.0, and a “drug-component-target-pathway” network was constructed. GO and KEGG enrichment analyses were performed using the DAVID database.

TCMSP-derived compounds were filtered by OB ≥30% and DL ≥0.18. Compounds absent from TCMSP were subjected to target prediction via SwissTargetPrediction. Anxiety-depression-related disease targets were collected from GeneCards (relevance-score ≥5). The PPI network was constructed in STRING (Homo sapiens, high-confidence score = 0.700). Hub genes were screened with the MCC algorithm of CytoHubba plugin. GO/KEGG enrichment was conducted using DAVID 6.8 against the whole-human-genome background; KEGG P-values were FDR-corrected for multiple comparisons.

Molecular Docking

Considering the research focus on neuroinflammation-mediated anxiety-depression pathogenesis, we selected five inflammation-associated key targets (SRC, TNF, IL-6, NF-κB and IL-1B) for molecular-docking validation, rather than directly adopting all top-degree nodes from network pharmacology. Although TP53, AKT1 and STAT3 rank high in network degree and participate in universal cellular processes, SRC, TNF, IL6, NF-κB and IL1B are more directly involved in neuroinflammatory cascades relevant to depression pathology, which is the core mechanistic direction of our present study. Protein structures of the top 5 core targets were downloaded from RCSB PDB, and structures of core components were obtained from PubChem. Notably, kaempferol‑3‑O‑rutinoside rather than free kaempferol was positively detected in ZCK by UPLC‑QTOF‑MS. Its aglycone kaempferol was selected for molecular docking, considering that orally administered glycosides are commonly hydrolyzed by intestinal β‑glucosidases into corresponding aglycones as the major bioactive forms in vivo. Each binding pocket was defined by the co‑crystallized ligand in the corresponding PDB structure. The geometric center of the native ligand served as the Grid‑box center, with box size extended to fully cover the active pocket. Native ligands were extracted for redocking and RMSD calculation to validate docking parameters. Proteins and ligands were protonated at pH 7.4, water molecules and non-ligand co-crystallized components were removed (native co-crystallized ligands were retained for redocking), and Gasteiger charges were assigned. Exhaustiveness was set to 8. For IL‑6 and NF‑κB, their structures are protein‑protein complexes lacking co‑crystallized small‑molecule inhibitors; hence positive‑control and decoy‑control benchmarking could not be conducted. All docking‑related metrics and grid‑box parameters are summarized in Table 3. All herbal components and positive‑control ligands shared identical docking settings. Molecular docking was performed using AutoDock Vina 1.1.2, and binding energies were calculated. The docking results were visualized using PyMOL. Notably, AutoDock Vina has a well-documented size-dependent scoring bias; raw scores may overestimate affinities of bulky ligands by ignoring deformation and desolvation effects.²² Ligand-efficiency normalization is recommended for cross-target comparison to mitigate this volume-related artefact. Molecular docking in our study serves as preliminary in‑silico hypothesis screening rather than quantitative prediction of actual experimental binding affinity. Nevertheless, ligand‑efficiency‑based normalization and re‑ranking were not performed in the present work. Notably, parent glycosides such as geniposide have limited blood‑brain barrier (BBB) permeability, but can be biotransformed by gut microbiota into more lipophilic aglycones (eg, genipin) with better BBB penetration. Our docking results indicate potential target‑binding tendency of these active metabolite forms, rather than direct central binding of unmodified parent glycosides. All docking outputs are treated as hypothesis-generating computational evidence rather than definitive proof of physical protein-ligand binding.

Table 3.

Summary of Molecular Docking Validation Parameters and Results

PDB ID	Target	Native Ligand for Redocking	Redocking RMSD (Å)	Grid-Box Centre (x, y, z)	Grid-Box Size (x, y, z)	Positive-Control Ligand	Decoy Ligand	Positive Best Affinity (kcal/mol)	Decoy Best Affinity (kcal/mol)	Validation Performed
1Y57	SRC kinase	MPZ	2.701	−2.972, −0.75, −1.972	88, 78, 70	Dasatinib	Pazopanib	−8.1	−6.9	Yes
2AZ5	TNF-α	SPD-304	1.156	−4.204, −5.453, −67.108	70, 88, 120	Thalidomide	Lenalidomide	−7.4	−7.0	Yes
8C3U	IL-1β	T9C	0.436	−11.49, 15.942, −1.683	44, 40, 46	VX-765	VX-740	−6.9	−6.5	Yes
4CNI	IL-6	–	–	−37.112, 21.851, 12.362	40, 42, 46	–	–	–	–	Not applicable (no co-crystallized small-molecule inhibitor)
7NM1	NF-κB p50-p50	–	–	47.687, 17.206, 38.263	100, 50, 62	–	–	–	–	Not applicable (no co-crystallized small-molecule inhibitor)

Notes: Molecular docking was performed using AutoDock Vina. Exhaustiveness was set to 8. Protein structures were pre‑processed with removal of water molecules and co‑crystallized ligands, protonated at pH 7.4, and assigned Gasteiger charges. All ligand molecules were also protonated at pH 7.4 prior to docking. PDB structures were selected based on resolution quality and the presence of well‑defined small‑molecule binding pockets. For IL‑6 (PDB:4CNI) and NF‑κB p50‑p50 (PDB:7NM1), no co‑crystallized small‑molecule inhibitor is available, therefore redocking, positive‑control and decoy‑control benchmarking were not feasible. All docking results are interpreted as hypothesis‑generating computational clues rather than definitive proof of physical protein‑ligand binding.

Results

ZCK Improves Body Weight in CUMS Mice

As shown in Figure 1, there was no significant difference in body weight among all groups in the early stage of the experiment (P > 0.05, one-way ANOVA). From day 6 of modeling, the body weight gain of the model group was significantly slower than that of the control group, and remained significantly lower than the control group throughout the experiment (P< 0.05, one-way ANOVA). After the start of administration on day 18, the high-dose ZCK group showed the most significant recovery in body weight, while the medium-dose and low-dose ZCK groups also exhibited a tendency of improvement. These results indicate that ZCK can effectively alleviate the body weight growth inhibition induced by CUMS.

Figure 1.

Changes in body weight of mice. n = 10 per group.

ZCK Ameliorates Anxiety- and Depression-Like Behaviors

As shown in Figure 2, compared with the control group, the model group exhibited markedly prolonged immobility time in FST and TST, accompanied by obvious anxiety‑related deficits in OFT and EPM (P<0.01, one‑way ANOVA). A reduction in sucrose preference rate was also observed in model mice, which served as secondary auxiliary observation. After ZCK intervention, the behavioral indexes from FST, TST, OFT and EPM were significantly reversed in a dose‑related manner (P<0.05 or P<0.01, one‑way ANOVA). Of note, the behavioural improvement was not strictly monotonic across dosage groups; the low‑dose and medium‑dose groups exhibited comparable effects in FST and TST, while the high‑dose group generally exerted the most prominent therapeutic benefit. Partial improvement in sucrose preference was also observed after ZCK treatment. These findings confirm that ZCK significantly alleviates CUMS‑induced anxiety‑ and depression‑like behaviors in mice, mainly supported by FST, TST, OFT and EPM results.

Figure 2.

Effects of ZCK on depression‑like behaviors in CUMS depressed mice. n = 10 per group; some data points overlap visually. (A) Sucrose preference rate. (B) Immobility time in the forced swimming test. (C) Immobility time in the tail suspension test. (D) Total moving distance in the open field test. (E) Residence time in the central zone of the open field test. (F) Percentage of open arm residence time in the elevated plus maze test. (G) Movement trajectory in the open field test. (H) Heat map of movement trajectory in the elevated plus maze test (vertical axis: open arms; horizontal axis: closed arms). Statistical significance: *P < 0.05, **P < 0.01 vs Model group; ^#P < 0.05, ^##P < 0.01 vs Control group.

ZCK Alleviates Hippocampal Pathological Damage

As shown in Figure 3, pathological morphological changes in the hippocampal CA3 region were observed under a microscope. In the control group, hippocampal neurons in the CA3 region were abundant, intact, smooth and round in morphology, neatly arranged, densely and uniformly distributed. In the CUMS model group, neuronal arrangement was markedly sparse, accompanied by obvious Nissl body loss, deeply stained and shrunken nuclei, and widened intercellular spaces. Compared with the model group, different degrees of improvement in hippocampal pathological injury were observed in all ZCK treatment groups in a dose‑dependent manner, with the high‑dose ZCK group showing the most obvious amelioration.

Figure 3.

Effects of ZCK on the pathological morphology of hippocampal tissue in CUMS-induced depressed mice (×200). (A) Control group. (B) CUMS model group. (C) Fluoxetine group. (D) ZCK low-dose group. (E) ZCK medium-dose group. (F) ZCK high-dose group.

ZCK Regulates Hippocampal Neurotransmitters

As shown in Figure 4, the levels of 5-HT (Figure 4A) and NE (Figure 4B) in the hippocampus were significantly decreased in the model group compared with the control group (P<0.01). Compared with the model group, medium- and high-dose ZCK markedly upregulated the levels of these neurotransmitters (P<0.01). These results suggest that ZCK can regulate the imbalance of monoamine neurotransmitters in CUMS mice.

Figure 4.

Effects on the contents of neurotransmitters 5-ht and NE in hippocampal tissue. n = 10 per group. (A) The content of 5-ht in hippocampal tissue. (B) The content of NE in hippocampal tissue. Statistical significance: *P < 0.05, **P < 0.01 vs Model group; ^##P < 0.01 vs Control group.

ZCK Inhibits Serum Inflammatory Factors

As shown in Figure 5, serum levels of IL-1β, IL-6 and TNF-α in the model group were significantly higher than those in the control group (P < 0.01) (Figure 5A–C). All ZCK treatment groups reduced the levels of these inflammatory factors in a dose‑dependent manner (P < 0.05 or P < 0.01). These results indicated that ZCK could inhibit CUMS-induced neuroinflammatory responses in mice.

Figure 5.

Effects of ZCK on serum contents of IL-1β, IL-6 and TNF-α in CUMS-induced depressed mice. n = 10 per group. (A) The content of IL-1β in serum. (B) The content of IL-6 in serum. (C) The content of TNF-α in serum. Statistical significance: *P < 0.05, **P < 0.01 vs Model group; ^##P < 0.01 vs Control group.

Serum Metabolomics Analysis

QC samples exhibited good clustering in the two-dimensional PCA score plot and were all distributed within ±2 STD in the PCA‑X one-dimensional distribution, with a correlation coefficient greater than 0.95, indicating good methodological stability and reliable data quality. Multivariate statistical analysis was performed using SIMCA software (Version 16.0.2). PCA results showed obvious separation among the control group, model group, and ZCK high-dose group (Figure 6). OPLS-DA models were established for control vs model and high-dose ZCK vs model, respectively, and 200 permutation tests were applied to verify model validity. For control vs model: R²Y(cum)=0.90, Q² (cum) = 0.72, the Q² intercept of permutation test = −0.86; for high-dose ZCK vs model: R²Y(cum)=0.96, Q² (cum) = 0.81, the Q² intercept of permutation test = −0.57. The negative Q² intercepts from permutation tests indicated no over-fitting of the established OPLS-DA models.

Figure 6.

Multivariate statistical analysis. n = 6 per group. (A) PCA of differential metabolites between the control group and model group. (B) PCA of differential metabolites between the high-dose group and model group. (C) OPLS-DA of the control group and model group. (D) OPLS-DA of the high-dose group and model group. (E) Two hundred permutation tests for the control group vs model group. (F) Two hundred permutation tests for the high-dose group vs model group.

Screening of Differential Metabolites and Metabolic Pathway Enrichment

Differential metabolites were screened according to the criteria of VIP > 1 and P < 0.05, and metabolite identification was performed by combining retention time and secondary mass spectrometry information. A total of 56 differential metabolites were identified, which could serve as potential biomarkers for ZCK intervention in anxiety and depression. Compared with the control group, 16 metabolites were significantly upregulated and 40 were downregulated in the model group. After ZCK treatment, the abnormal levels of these metabolites were markedly reversed (P < 0.05), and the corresponding heatmap is shown in Figure 7A. Metabolic pathway enrichment analysis of differential metabolites was performed using MetaboAnalyst 5.0. With impact > 0 as the threshold, a total of 15 key metabolic pathways were enriched, mainly including alanine-aspartate-glutamate metabolism, glyoxylate and dicarboxylate metabolism, valine/leucine/isoleucine degradation, arachidonic acid metabolism, glycine-serine-threonine metabolism, and glycerophospholipid metabolism. Among them, alanine-aspartate-glutamate metabolism (impact = 0.42068) and phenylalanine metabolism (impact = 0.35714) showed the highest pathway impact. These results suggest that ZCK may exert anxiolytic and antidepressant effects by regulating the above metabolic pathways (Figure 7B).

Figure 7.

Analysis of differential metabolites and metabolic pathways. n = 6 per group. (A) Heatmap of serum differential metabolites in CUMS mice treated with ZCK (n=6). (B) Metabolic pathways involved in the therapeutic effect of ZCK on CUMS. Numbered points correspond to the following enriched pathways, ranked by significance from highest to lowest: [1] alanine, aspartate and glutamate metabolism [2] glyoxylate and dicarboxylate metabolism [3] Valine, leucine and isoleucine biosynthesis [4] arachidonic acid metabolism [5] nicotinate and nicotinamide metabolism [6] arginine biosynthesis [7] glutathione metabolism [8] phenylalanine, tyrosine and tryptophan biosynthesis [9] phenylalanine metabolism [10] citrate cycle (TCA cycle) [11] one carbon pool by folate [12] sphingolipid metabolism [13] glycine, serine and threonine metabolism [14] glycerophospholipid metabolism [15] primary bile acid biosynthesis.

UPLC-Q-TOF-MS Identification of Chemical Components

Based on literature retrieval and mass spectral matching, a total of 39 unique chemical components were identified in ZCK. Some compounds were detected in both positive‑ion and negative‑ion modes; 16 compound entries were acquired under positive‑ion mode and 27 entries under negative‑ion mode after counting both ion forms. These compounds mainly covered flavonoids, iridoid glycosides, phenolic acids, saponins, amino acids and fatty acids. Detailed information is listed in Table 4.

Table 4.

Chemical Component Analysis of ZCK

Identified Compounds	tR (min)	Formula	Observed Mass (m/z)	ms2Adduct	Error (ppm)	MS2 ion
Positive ion mode
L-Arginine	0.54	C6H14N4O2	175.1190	[M+H]+	0.91	70.0655, 60.0551
Trigonelline	0.62	C7H7NO2	138.0552	[M+H]+	2.39	94.0603, 80.0497
L-Proline	0.63	C5H9NO2	116.0703	[M+H]+	−3.45	70.0655, 68.0499
Rutin	7.53	C27H30O16	611.3122	[M+H]+	6.25	303.0499, 465.1088, 177.0551
Neferine	7.83	C38H44N2O6	625.3200	[M+H]+	8.95	206.1181, 160.0757, 391.2266
Genistin	8.16	C21H20O10	433.1143	[M+H]+	3.53	271.0602, 153.0188, 119.0499
Kaempferol-3-O-rutinoside	8.32	C27H30O15	595.1659	[M]+	0.66	287.0550, 449.1139, 153.0188
Narcissoside	8.84	C28H32O16	625.1773	[M+H]+	1.30	317.0655, 479.1037, 179.0602
Naringenin	8.88	C15H12O5	273.0765	[M+H]+	3.44	153.0188, 119.0499, 107.0499
Neodiosmin	9.46	C28H32O15	609.1822	[M+H]+	1.40	301.0705, 463.1289, 179.0602
Ononin	10.15	C22H22O9	431.1340	[M+H]+	0.72	269.0651, 137.0239, 119.0499
Daidzein	10.23	C15H10O4	255.0655	[M+H]+	1.61	199.0568, 137.0239, 119.0499
Genistein	12.09	C15H10O5	271.0603	[M+H]+	2.14	153.0188, 199.0568, 119.0499
Formononetin	14.08	C16H12O4	269.0812	[M+H]+	1.38	223.0655, 137.0239, 119.0499
Glycyrrhizic Acid	15.51	C42H62O16	823.4115	[M+H]+	0.19	451.3189, 355.1086, 191.0501
Erucamide	32.86	C22H43NO	338.3421	[M+H]+	1.06	321.3003, 265.2389, 125.0961
Negative ion mode
Stachyose	0.60	C24H42O21	665.2137	[M+FA-H]−	9.27	365.1095, 203.0505, 161.0399
Citric acid	0.69	C6H8O7	191.0197	[M-H]−	−0.42	173.0087, 129.0192, 87.0087
Neochlorogenic acid	2.11	C16H18O9	353.0879	[M-H]−		191.0501, 179.0345, 135.0451
Chlorogenic acid	5.60	C16H18O9	353.0873	[M-H]−	2.78	191.0501, 173.0499, 135.0451
Geniposide	5.96	C17H24O10	373.1135	[M+FA-H]−	−4.13	211.0613, 183.0501, 121.0401
4-Hydroxycinnamic acid	6.13	C9H8O3	163.0404	[M-H]−	2.21	119.0492, 93.0345, 75.0239
Rutin	7.83	C27H30O16	609.1451	[M-H]−	−2.77	301.0343, 463.0932, 179.0345
Vitexin	7.94	C21H20O10	431.0981	[M-H]−	0.07	285.0445, 153.0192, 135.0451
Hyperoside	8.01	C21H20O12	463.0884	[M-H]−	0.56	301.0343, 179.0345, 151.0087
Cynaroside	8.16	C21H20O11	447.0939	[M-H]−	1.59	285.0445, 153.0192, 135.0451
Genistin	8.18	C21H20O10	431.0977	[M+FA-H]−		270.0524, 152.0110, 118.0414
Kaempferol-3-O-rutinoside	8.37	C27H30O15	593.1505	[M-H]−	−0.35	285.0445, 447.0983, 151.0087
Isochlorogenic acid A	8.68	C25H24O12	515.1201	[M-H]−	2.00	353.0872, 191.0501, 179.0345
Narcissoside	8.87	C28H32O16	623.1616	[M-H]−	−0.63	315.0449, 477.0881, 179.0345
Azelaic acid	9.09	C9H16O4	187.0976	[M-H]−	0.32	169.0866, 125.0760, 83.0655
Rhoifolin	9.16	C27H30O14	577.1572	[M-H2O-H]−	−1.13	285.0445, 431.0983, 151.0087
Isochlorogenic acid C	9.24	C25H24O12	515.1194	[M-H]−	0.08	353.0872, 191.0501, 135.0451
Genistein	9.32	C15H10O5	269.0455	[M-H]−	−1.30	151.0087, 197.0499, 119.0492
Neodiosmin	9.49	C28H32O15	607.1664	[M-H]−	0.68	299.0501, 461.1137, 179.0345
Ononin	10.19	C22H22O9	429.1141	[M-H]−		267.0666, 135.0312, 119.0428
Daidzein	10.26	C15H10O4	253.0507	[M-H]−	0.36	197.0499, 135.0294, 119.0492
Glycitein	10.85	C16H12O5	283.0618	[M-H]−	−0.49	227.0655, 165.0192, 135.0294
Luteolin	10.94	C15H10O6	285.0415	[M-H]−	2.63	153.0192, 135.0451, 119.0492
Formononetin	11.48	C16H12O4	267.0671	[M-H]−	2.25	221.0499, 135.0294, 119.0492
Jaceosidin	12.89	C17H14O7	329.0661	[M-H]−	1.70	285.0526, 253.0345, 177.0120
Isoliquiritigenin	13.64	C15H12O4	255.0669	[M-H]−	2.90	135.0294, 119.0492, 91.0345
Ammonium Glycyrrhizinate	15.56	C42H62O16·NH3	821.3980	[M-H2O-H]−	0.94	449.3033, 353.0930, 191.0501
Cinchonine	17.07	C19H22N2O	293.1770	[M-H]−		176.0805, 134.0599, 108.0445
Oleanic acid	26.85	C30H48O3	455.3556	[M-H]−	4.70	437.3415, 205.1586, 187.1481
Linoleic acid	28.51	C18H32O2	279.2341	[M-H]−	3.40	261.2220, 203.1694, 149.0961
Palmitic acid	29.69	C16H32O2	255.2340	[M-H]−	3.06	237.2111, 197.1798, 125.0961
Oleic acid	30.32	C18H34O2	281.2497	[M-H]−	4.37	263.2376, 205.1850, 151.1117

Network Pharmacology Results

A total of 773 drug-related targets of ZCK and 1981 anxiety‑depression disease-related targets were retrieved from public databases, and 410 overlapping targets were obtained after intersection analysis (Figure 8A). The overlapping targets were imported into the STRING database to construct a PPI network, and topological analysis was performed using the CytoHubba plugin in Cytoscape. The top 10 core targets ranked by degree value were TP53, AKT1, STAT3, SRC, ESR1, JUN, HSP90AA1, EP300, EGFR, and TNF (Figure 8B). Among these targets, inflammation-related factors such as TNF, IL6, and IL1B were consistent with previous pharmacodynamic results, in which ZCK reduced the serum levels of IL-6, TNF-α, and IL-1β. GO functional enrichment analysis (Figure 8C) showed that biological processes (BP) were mainly involved in response to exogenous stimuli, response to hypoxia, response to oxidative stress, response to lipopolysaccharide, and regulation of neuronal apoptosis. Cellular components (CC) were mainly enriched in synaptic membrane, postsynaptic membrane, and membrane raft. Molecular functions (MF) mainly included neurotransmitter receptor activity, GABA receptor activity, and cytokine activity. KEGG pathway enrichment analysis (Figure 8D) predicted that multiple candidate signaling pathways may participate in the anti‑anxiety and anti‑depression effects of ZCK, including lipid and atherosclerosis (hsa05417), AGE‑RAGE‑signaling pathway (hsa04933), TNF‑signaling pathway (hsa04668), HIF‑1‑signaling pathway (hsa04066), PI3K‑Akt‑signaling pathway (hsa04151), and MAPK‑signaling pathway (hsa04010). These pathways are closely involved in neuroinflammation, oxidative stress, neuronal survival, and metabolic regulation (Figure 8E). Notably, due to the inherent promiscuity of the GeneCards database which contains extensive multi-functional disease-related genes, raw network-pharmacology prediction generated multiple cancer- and infection-related targets and enrichment terms. Since cancer‑related pathways are not the research focus of the present study, these targets were excluded from subsequent screening. Only targets closely associated with anxiety‑depression and neuroinflammation were retained for further analysis. Among the top‑10 core targets predicted by network pharmacology, TP53, AKT1 and STAT3 are important regulatory molecules. Combined with our in‑vivo experimental results focusing on neuroinflammation, we further selected inflammation‑relevant targets SRC, TNF, IL6, NF‑κB and IL1B for subsequent molecular docking assessment.

Figure 8.

Network pharmacology analysis. (A) Venn diagram of drug targets and disease targets. (B) Protein-protein interaction (PPI) network of core targets. (C) ZCK component-target-pathway network diagram. (D) Bar chart of GO functional enrichment analysis. (E) Bar chart of KEGG pathway enrichment analysis.

Molecular Docking Verification

The binding energies between core components and key targets were all lower than −5.0 kcal/mol, with the lowest binding energy reaching −7.4 kcal/mol. These simulation results suggest potential intermolecular interactions between active components and target proteins. The heatmap of binding energies is shown in Figure 9A, and the molecular docking visualization results are presented in Figure 9B. Considering the inherent size‑related scoring bias of AutoDock Vina, raw Vina binding energies should be interpreted prudently. Ligand‑efficiency normalization by heavy‑atom count is an available approach to reduce such bias for cross‑ligand comparison, yet this correction was not implemented in our current analysis. The heatmap of binding energies is shown in Figure 9A, and the molecular docking visualization results are presented in Figure 9B. These in‑silico findings provide preliminary computational evidence for component‑target interactions, and further experimental validation is still required.

Figure 9.

Analysis of molecular docking results. (A) Heatmap of molecular docking binding energies between active ingredients in ZCK and core targets. (B) Visualization of molecular docking results.

Discussion

Anxiety and depression have become the most common mental disorders in China. Western medicines have limited efficacy and obvious side effects, while traditional Chinese medicine has unique advantages in holistic regulation, multi-target action and low toxicity.^23,24 ZCK is modified from classical prescriptions. It follows the core pathogenesis of stagnant heat internal disturbance and mental restlessness, and has the effects of clearing heat, relieving restlessness, calming mind and resolving depression.

In this study, a CUMS combined with single-housing model was established in mice. ZCK significantly improved anxiety- and depression-like behaviors, increased sucrose preference and autonomous activity, shortened immobility time, and prolonged open arm residence time. It also relieved the inhibition of body weight gain.

Monoamine neurotransmitter imbalance is a core pathological feature of anxiety and depression. 5-HT and NE play key roles in emotional regulation.^25,26 ELISA results showed that ZCK upregulated hippocampal 5-HT and NE to correct neurotransmitter imbalance. Excessive neuroinflammation and hippocampal neuronal injury are important pathogenic factors.^27,28 ZCK reduced serum levels of IL-1β, IL-6 and TNF-α. Network pharmacology prediction suggested that ZCK may target core inflammatory targets such as TNF and IL6. Consistent with this hypothesis, our in‑vivo experimental results showed that ZCK reduced serum levels of IL‑1β, IL‑6 and TNF‑α, and implying potential modulation of related inflammatory cascades neuroinflammatory cascades. Accumulating evidence indicates that peripheral pro‑inflammatory cytokines modulate central neuroinflammation via the blood‑brain barrier, vagus nerve and HPA axis, so serum cytokine levels indirectly reflect central inflammatory risk under chronic stress. HE staining confirmed that ZCK protected hippocampal neuronal structure and alleviated stress injury.

Serum metabolomics results further confirmed and refined the above regulatory mechanisms. A total of 56 differential metabolites were screened, and three key pathways were enriched. First, the neurotransmitter and amino acid metabolism pathway. Modern pharmacology confirms that metabolic imbalance of glutamate and GABA is a core pathological feature of anxiety and depression.^29,30 Serum metabolomics showed that ZCK significantly reversed abnormal levels of metabolites such as glutamate and phenylalanine. These findings corresponded to the enriched terms of GABA receptor activity and neurotransmitter receptor activity in network pharmacology, jointly verifying that ZCK corrects the imbalance between central excitatory and inhibitory neurotransmitters and improves emotional dysregulation. Second, the inflammation‑related pathway. Serum metabolomics indicated that ZCK inhibits neuroinflammation mediated by arachidonic acid metabolism and downregulates pro‑inflammatory mediators. These results were consistent with inflammatory targets such as TNF, IL6 and IL1β, as well as the TNF signaling pathway, thereby reducing neuronal damage. Meanwhile, network pharmacology enriched the TNF and MAPK inflammatory pathways. The three results were highly consistent, supporting a hypothetical mechanistic model that ZCK alleviates neuronal injury through inhibiting neuroinflammation. Third, the lipid metabolism pathway. Serum metabolomics showed that ZCK also regulates glycerophospholipid metabolism, repairs cell membrane signal transduction and improves emotional dysregulation.

This study used UPLC-QTOF-MS to characterize the chemical composition of ZCK. A total of 39 chemical components were identified, clarifying its material basis. Network pharmacology analysis was performed based on these identified components. We screened 773 drug targets and 1981 disease targets associated with anxiety and depression. The intersection yielded 410 potential therapeutic targets for ZCK. Topological analysis of PPI network identified high‑degree hub targets including SRC and TNF. IL‑6, IL‑1β and NF‑κB were further highlighted as key targets based on our subsequent in‑vivo neuroinflammatory experimental evidence, although they were not ranked within the top‑10 degree nodes in the in‑silico network. Key active ingredients were further identified as geniposide, genistein, kaempferol-3-O-rutinoside (its aglycone kaempferol was used for molecular docking analysis), and daidzein. KEGG enrichment analysis predicted significant enrichment of the candidate TNF, PI3K/AKT, and MAPK pathways, network pharmacology prediction suggested that ZCK may exert therapeutic effects through potential multi‑component, multi‑target, and multi‑pathway interactions. It should be emphasized that network pharmacology in this study serves as a hypothesis‑generating in‑silico approach, and the predicted targets and pathways require further experimental validation.

Based on network‑pharmacology prediction and our neuroinflammation‑related experimental data, we selected several inflammation‑associated targets for molecular‑docking validation. Molecular docking analysis was conducted to verify the interactions between active ingredients and core targets. The results showed that geniposide, genistein, kaempferol-3-O-rutinoside (its aglycone kaempferol was used for molecular docking analysis), and daidzein bound to SRC, TNF, IL6, NFκB, and IL1β with binding energies lower than −5.0 kcal/mol, reaching a minimum of −7.4 kcal/mol. These in‑silico findings suggest potential intermolecular interactions between active ingredients and target proteins, providing preliminary predictive support for the multi‑target characteristics of ZCK.

It should be noted that molecular docking is only an in‑silico predictive tool in this study. Further biophysical experiments such as surface plasmon resonance are required to validate the actual binding affinity between active components and target proteins. Owing to Vina’s size-dependent scoring bias, large ligands like geniposide may obtain artificially favourable raw binding energies.²² Ligand-efficiency normalization is suggested for future work to reduce size-related artefacts during cross-ligand or cross-target comparison. Regarding BBB permeability, most parent glycosides in ZCK poorly cross the BBB due to high polarity. After oral administration, intestinal microbiota mediates deglycosylation to generate corresponding aglycones (genipin, genistein, daidzein, kaempferol), which possess better lipophilicity and are regarded as the more plausible central active forms. Accordingly, docking results should be interpreted as potential interactions between these BBB‑permeable active metabolites and central targets, rather than direct binding of parent glycosides to CNS targets.

Nevertheless, several limitations still exist in the present study. First, regarding experimental design and in vivo detection: ZCK granules contain sucrose-containing excipients, and pre-exposure to sucrose via gavage may exert a mild reward effect. Though intragastric sucrose exerts weaker impact than voluntary drinking, its confounding impact on SPT cannot be completely ruled out. Thus, SPT is treated only as a secondary auxiliary index here. Future studies with matched excipient-only controls are required to further validate ZCK pharmacological effects. Serum-derived IL-1β, IL-6 and TNF-α reflect systemic inflammatory status, and cannot be directly equated with hippocampal central neuroinflammation. Further hippocampal tissue detection is required in future investigations. Direct brain exposure of active compounds was not measured by brain tissue LC-MS/MS in this study, and the actual central active forms of glycoside components require further pharmacokinetic verification. Second, regarding computational analysis and validation: Predicted involvement of TNF, PI3K/AKT and MAPK pathways lacks hippocampal protein-level validation, which we plan to address in follow-up experiments. Several PPI hub genes represent universal highly-connected interactome nodes rather than exclusive ZCK-specific therapeutic targets, and their biological roles will be further dissected in future work. Decoy disease target set benchmarking for enrichment specificity was not performed in this study and will be included in follow-up work. Inconsistency between monoamine-related pathway enrichment results and in-vivo neurotransmitter changes illustrates that computational predictions cannot fully recapitulate complex in-vivo responses.

In conclusion, ZCK produced obvious anti‑anxiety and anti‑depressant effects in CUMS mice, manifested by inhibiting neuroinflammation, regulating neurotransmitters, correcting metabolic disorders and protecting hippocampal neurons. In‑silico analysis predicts that ZCK may target TNF, PI3K/AKT and MAPK signaling, which deserves further experimental validation. This study provides preliminary scientific basis for the clinical application of ZCK and a reference for mechanism research of TCM compounds.

Funding Statement

Hubei Provincial Natural Science Foundation—Joint Fund for Traditional Chinese Medicine Innovation and Development (Hubei Provincial Natural Science Foundation, 2024AFD237); Major Technological Innovation Special Project of Hubei Province (Major Technological Innovation Special Project of Hubei Province, 2023BE011); National Key R&D Program of China: Research and Industrial Demonstration on Key Technologies of Circular Utilization of Traditional Chinese Medicine Resources (National Key R&D Program of China, 2025YFC3509100).

Data Sharing Statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics Approval and Consent to Participate

All animal procedures were approved by the Laboratory Animal Ethics Committee of Wuhan Hualianke Biotechnology Co., Ltd. on September 27, 2024 (Approval No. HLK-20240927-001). The animal study was conducted from October 8 to November 26, 2024. The experimental timeline consisted of a 7-day acclimatization period, followed by 36 days of CUMS modeling and drug administration, and concluded with a 7-day behavioral assessment. All procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals and adhered to the 3R principles.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare no competing interests.