Molecular mechanisms of YZG-331 ameliorated insomnia based on the gut microbiota–drug interaction

Abstract

The gut microbiota plays a crucial role in the study of drug efficacy and metabolism. YZG-331 exhibits significant sedative and hypnotic effects. However, the oral bioavailability of YZG-331 in female mice remains low at 27.7%, and it exhibits instability in the gut microbiota, undergoing rapid metabolism, with the intestinal hydrolytic metabolite M2 identified as the characteristic metabolite. Both the underlying mechanisms of YZG-331 therapeutic effects on insomnia and its enzyme-mediated metabolic pathways involving the gut microbiota remain incompletely elucidated. In this study, we found that YZG-331 regulated the tryptophan metabolism and 5-HT production pathways by the gut microbiota. Specifically, YZG-331 increased the abundance of Lactobacillus and Muribaculaceae in the gut, with Lactobacillus reuteri playing a critical role in both the metabolic processes of YZG-331 and the biosynthesis of tryptophan-derived metabolites. YZG-331 enhanced the activation of the tetrahydrobiopterin (BH₄) cofactor system, which subsequently promoted tryptophan hydroxylase (TPH) activity, thereby enhancing 5-HT production. Furthermore, YZG-331 and M2 increased intestinal butyric acid levels, thereby promoting sleep regulation and supporting the restoration of the intestinal barrier. Overall, our study provided the first evidence highlighting the crucial role of the gut microbiota in the therapeutic mechanism and metabolic processing of YZG-331 in the treatment of insomnia.

Graphical abstract

YZG-331 regulates tryptophan metabolism and 5-HT production pathways by the gut microbiota. Lactobacillus reuteri plays a critical role in both the metabolic processes of YZG-331 and the biosynthesis of tryptophan-derived metabolites.

1. Introduction

Insomnia is defined by inadequate sleep duration or notable deterioration in sleep quality, primarily manifesting as difficulties in initiating and maintaining sleep, alongside daytime symptoms such as irritability and fatigue^1,2. The current prevalence of insomnia disorder is estimated to range from 10% to 20%, with a higher incidence observed in women and older people³. The fatigue, mood disturbances, and cognitive deficits linked to insomnia not only severely impair quality of life but also increase the risk of severe health problems, including obesity⁴, depression⁵, cardiovascular diseases⁶, and ovarian cancer⁷. Moreover, previous studies indicated that insomnia could accelerate the onset of inflammatory bowel disease and cause intestinal damage⁸. Clinical studies have revealed that microbiota dysbiosis exists in insomnia patients and probiotic interventions could ameliorate insomnia symptoms^9,10. These findings suggest that the microbiota balance plays a critical role in the progression and treatment of insomnia. In recent years, the significance of the gut microbiota in drug metabolism and pharmacological research has garnered increasing attention, particularly regarding its unique role in brain development and functional regulation^11,12. Dysbiosis of the gut microbiota significantly alters its metabolic activity and resultant metabolites, including short-chain fatty acids (SCFAs), γ-aminobutyric acid (GABA), tryptophan, and bile acids, which are critical mediators of gut–brain axis function. Tryptophan metabolites, such as 5-hydroxytryptophan (5-HTP) and serotonin (5-HT), are among the most studied in host–microbiota interactions, and they play a significant role in regulating sleep and wakefulness¹³. The gut microbiota is involved in the biosynthesis process of 5-HT, thus confirming the “gut–brain axis” hypothesis for the first time with clear chemical evidence¹⁴. Furthermore, the microbial metabolite butyric acid could promote sleep by regulating the activity of orexin neurons in the lateral hypothalamic region of mice¹⁵.

Maintaining gut microbiota balance and correcting dysbiosis could ameliorate insomnia symptoms⁹. Traditional Chinese medicine plays a critical role in correcting gut microbiota dysbiosis and ameliorating sleep disturbances¹⁶. For example, Fructus gardeniae has been shown to promote sleep and enhance immunity by ameliorating gut dysfunction¹⁷. Gastrodia elata Blume is a traditional Chinese medicinal herb widely used to ameliorate and treat insomnia¹⁸. YZG-331, an optimized derivative of N⁶-(1-phenylpropyl) adenosine (NHBA), which is a principal bioactive compound responsible for the calming effects of Gastrodia elata Blume, belongs to the category of natural product-derived compounds. Previous mechanistic studies have revealed that YZG-331 is a multi-target drug, which not only activated adenosine A1 receptors and A2 receptors to ameliorate insomnia¹⁹, but also exerted sedative and hypnotic effects by enhancing glutamic acid decarboxylase activity and elevating GABA levels in the hypothalamus²⁰. YZG-331 exhibits favorable biosafety and has been approved for phase I clinical trial by the National Medical Products Administration (Approval No. 2024LP01750), thereby facilitating related clinical investigations. Pharmacokinetic studies indicated that the oral bioavailability of YZG-331 in female mice was relatively low, at just 27.7%²¹. Only a small amount of YZG-331 could pass through the blood-brain barrier and it was distributed more in the gastrointestinal tract following oral administration²². Furthermore, YZG-331 exhibited instability in the gut microbiota of mice, with over 50% of the parent compound degrading within 120 min²³. Especially, the intestinal hydrolytic metabolite M2 represents a characteristic metabolite of YZG-331 within the gut microbiota²³. However, the potential efficacy of the gut microbiota-derived M2 metabolite of YZG-331 remains unclear. To sum up, the underlying mechanisms through how YZG-331 affects the gut microbiota to ameliorate insomnia, along with the related metabolic pathways and processes within the intestinal tract, remain incompletely elucidated.

In the present study, we aimed to investigate the molecular mechanisms of underlying the pharmacological effects of YZG-331 mediated by the gut microbiota, the interactions between YZG-331 and gut microbial communities. In vivo experiments and metabolomics experiments were conducted to verify that the insomnia efficacy of YZG-331 was associated with gut microbiota. The metabolic process of YZG-331 was identified, and specific enzymes involved in the metabolic process were identified through gene overexpression experiments. Through these investigations, we sought to clarify the bidirectional interactions between YZG-331 and the gut microbiota, thereby providing a theoretical foundation for its application in contemporary clinical research.

2. Materials and methods

2.1. Chemicals and instruments

YZG-331, with a purity exceeding 99%, was synthesized at the Department of Natural Products Chemistry within the Institute of Materia Medica at the Chinese Academy of Medical Sciences in Beijing. P-chlorophenylalanine (PCPA) was procured from Sigma–Aldrich in Shanghai, China. Additionally, 5-HTP, 5-HT, Glipizide hydrochloride, and anaerobic medium (in dry powder form) were obtained from Beijing Solarbio Science & Technology Co., Ltd. (Beijing, China). Phosphoric, propionic, and butyric acids were also acquired from Sigma–Aldrich in St. Louis, MO, USA. TPH ELISA kits for content detection were sourced from the Nanjing Jiancheng Bioengineering Institute (Nanjing, China), BH₄ ELISA kits were obtained from Shanghai Jianglai Biotechnology Co., Ltd. (Shanghai, China). Tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1 beta (IL-1β) were purchased from the Nanjing Jiancheng Bioengineering Institute (Nanjing, China). An LC–MS/MS8060, GCMS-QP2020SE, GC-2014, LC/MSⁿ-IT-TOF (Shimadzu, Japan), constant temperature culture shaker (LYZ-100, Shanghai Longyue Co., Ltd., China), WH-681 vortex mixer was (Jintan Shenglan Co., Ltd., China), small table high-speed centrifuge (Eppendorf, Germany), SMALDIPrep (TransMI, Germany), CM1860 cryostat (Leica Biosystems, Germany) and AP-SMALDI 5 equipped with Q Exactive™ mass spectrometer (Thermo Scientific™ America).

2.2. Animals and insomnia mouse model development

2.2.1. Animals

Female BALB/c mice, weighing 20–25 g, were purchased from Vitalriver (Beijing, China). All animals had free access to food and water. The temperature was maintained at 22–24 °C. The experimental procedures complied with institutional and ethical guidelines, receiving approval from the Laboratories Institutional Animal Care and Use Committee of the Chinese Academy of Medical Sciences and Peking Union Medical College (No. 00002312).

2.2.2. Development of PCPA induced insomnia model and treatments

The PCPA-insomnia model was established using the methodologies outlined in the relevant literature²⁴. In detail, PCPA was dissolved in saline and then BALB/c mice were intraperitoneally injected with PCPA suspension according to the amount of 350 mg/kg for three consecutive days. BALB/c mice (female, 22 ± 3 g) were acclimatized for one week and then randomly assigned to five groups (n = 5 in each group): the control group (oral saline), model group (PCPA, i.p., 350 mg/kg/day), YZG-331 (i.p.) group (PCPA, i.p., 350 mg/kg/day and YZG-331, intraperitoneal injection, 5 mg/kg/day), YZG-331 (i.g.) group (PCPA, i.p., 350 mg/kg/day and YZG-331, oral, 5 mg/kg/day), YZG-331 (i.g.) group (PCPA, i.p., 350 mg/kg/day and YZG-331, oral, 20 mg/kg/day). After insomnia model was established, YZG-331 group (i.g.) was given oral administration of YZG-331 for 7 days. The experimental timeline was summarized in Fig. 1A. Upon completion of the final treatment, the efficacy of YZG-331 was assessed using a pentobarbital-induced sleep test. The feces, serum and brain tissues of mice were used for subsequent non-targeted metabolomics studies and targeted metabolomics analysis. The sample size (n = 5 per group) was determined based on commonly accepted practices in microbiome and metabolomics studies using mice models^25,26. Given the exploratory nature of the 16S rRNA sequencing and non-targeted metabolomics analyses in this study, the selected sample size was intended to identify prominent and consistent trends rather than subtle effect sizes.

Figure 1.

YZG-331 treatment ameliorated PCPA-induced sleep disturbance in mice and metabolomic alterations. (A) Experimental design of the pharmacological effect of YZG-331. (B) Sleep latency time of each group (n = 5). (C) Sleep duration of each group (n = 5). (D) An OPLS-DA model showing the group separation between the model group and the YZG-331 group in feces. (E) Corresponding S-Plot and VIP for feature selection in feces. Each dot represents a feature, and the dots located in the upper right and lower left areas are the selected dots. (F) Volcano plot of fecal metabolomics analysis for YZG-331 vs the model group. Significantly increased and decreased metabolites are indicated by red and blue dots, respectively. (G) Heatmap showing metabolic features among each individual in feces. (H) Relative abundance of tryptophan in feces (n = 5). (I) Relative abundance of 5-HT in feces (n = 5). (J) OPLS-DA model between the model group and the YZG-331 group in serum. (K) Corresponding S-Plot and VIP for feature selection in serum. (L) Volcano plot of metabolomics analysis for YZG-331 vs the model group in serum. (M) Relative abundance of 5-HTP in serum. (N) Spearman correlation analysis between the pharmacodynamic parameters and 5-HTP levels in serum and 5-HT levels in feces. Data are presented as mean ± SD. ∗∗P < 0.01.

2.2.3. Development of chronic restraint-induced insomnia model and treatments

The BALB/c mice were subjected to chronic restraint stress for five consecutive days (from 9:00 am to 3:00 pm, for 6 h each day). Then, PCPA suspension was intraperitoneally injected at a dose of 350 mg/kg for three consecutive days. The pentobarbital sodium-induced sleep test was used to evaluate the chronic restraint-induced insomnia model. Pseudo-germ-free (PGF) mice: we establish a reference method for PGF mice model²⁷. In detail, BABL/c mice were orally treated with cefadroxil (100 mg/kg), terramycin (300 mg/kg), erythromycin (300 mg/kg) twice a day for five consecutive days prior to inducing the chronic restraint-induced insomnia model. Effects of YZG-331 and gut microbiota metabolite (M2) on chronic restraint-induced insomnia model: Thirty female BALB/c mice were randomly allocated into six distinct groups: a control group (n = 5), a model group (n = 5), a M2 group (n = 5), a low-dose YZG-331 (L-YZG-331) group (n = 5), a high-dose YZG-331 (H-YZG-331) group (n = 5), a PGF with high-dose YZG-331 (PGF-H-YZG-331) group (n = 5). The corresponding administration methods for each group are as follows: M2 group (50 mg/kg/day), L-YZG-331 (20 mg/kg/day), H-YZG-331 (50 mg/kg/day), PGF-H-YZG-331 (50 mg/kg/day). All drugs or the metabolite were administered orally for 7 days. Upon completion of the final treatment, the efficacy of M2 and YZG-331 was assessed using a pentobarbital-induced sleep test. The experimental flow chart is shown in Fig. 3A. Fresh feces, serum, and brain tissues were collected to measure 5-HTP and 5-HT levels. As previously described²⁸, quantitative analysis of 5-HTP and 5-HT was conducted using HPLC–MS/MS 8060.

Figure 3.

Spatially metabolomic analysis to localize the alternated metabolic pathway of brain tissue after YZG-331 treatment. (A) Cluster analysis and heatmap of upregulated or downregulated metabolites in each group. (B) Mass spatial imaging analysis of the representative metabolites from Cluster 1 and H&E staining of each group. (C) Mass spatial imaging analysis of the representative metabolites from Cluster 2 and cartoon graphic of each group. (D) The concentration of 5-HTP in mice brain (n = 5). (E) The concentration of 5-HT in mice brain (n = 5). (F) KEGG pathways of differentially abundant metabolites in Cluster 1. (G) KEGG pathways of differentially abundant metabolites in Cluster 2. HP, hippocampus; CB, cerebellum; MD, medulla; PN, pons; MB, midbrain; TH, thalamus; HT, hypothalamus; CTX, cortex; CC, corpus callosum. Data are presented as mean ± SD. ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

2.2.4. The efficacy of M2 on PCPA induced insomnia mice

The PCPA-insomnia model and PGF mice model were established using the methods in sections 2.22, 2.23. BALB/c mice (female, 22 ± 3 g) were acclimatized for one week and then randomly assigned to four groups (n = 5 in each group): the control group, model group, M2 group (50 mg/kg/day) and PGF-M2 group (50 mg/kg/day). M2 was administered orally for 7 days. Upon completion of the final treatment, the efficacy of M2 was assessed using a pentobarbital-induced sleep test.

2.2.5. Humanized gnotobiotic mouse model

To evaluate the translational relevance of the gut microbiota-dependent effects of YZG-331, a humanized gnotobiotic mouse model was employed. In detail, BALB/c mice were orally treated with vancomycin (50 mg/kg), metronidazole (100 mg/kg), neomycin sulfate (100 mg/kg) and ampicillin trihydrate (100 mg/kg) twice a day for seven consecutive days to generate germ-free mice model. Then the germ-free mice were maintained under sterile conditions and colonized with fecal microbiota obtained from patients with insomnia and healthy donors. Stool samples from insomnia patients and healthy donors were collected in a non-interventional clinical trial (Peking Union Medical College Hospital Ethics Committee approval: K26C019). Fecal samples were processed under anaerobic conditions, and mice were orally gavage with fecal suspensions for two weeks. The humanized gnotobiotic mice were randomly assigned to three groups: FMT-Healthy mice group, the FMT-Insomnia mice group and YZG-331 treatment group (50 mg/kg/day). YZG-331 was administered orally for 7 days. Upon completion of the final treatment, the efficacy of YZG-331 was assessed using a pentobarbital-induced sleep test.

2.3. Pentobarbital sodium-induced sleep test

The pentobarbital sodium-induced sleep test in mice is a widely used behavioral assay to evaluate the sedative-hypnotic activity of a drug²⁹. A pentobarbital sodium solution induced sleep test was performed 30 min after the end of administration. The mice were intraperitoneal injection of pentobarbital sodium solution (subthreshold dose: 40 mg/kg) to assess the pharmacodynamic effects. The timing of the injection was precisely recorded. Sleep onset was defined as the point at which the loss of the righting reflex persisted for over 1 min. Additionally, the time until the animals regained their righting reflex was documented. Three distinct time points were recorded, and sleep latency and sleep duration were calculated as pharmacodynamic evaluation parameters. Sleep latency was defined as the time from pentobarbital sodium injection to the time of the righting reflex loss. Sleep duration was considered as the time interval between righting reflex loss and recovery.

2.4. Fecal and serum non-targeted metabolomics methods

A slightly modified sample preparation protocol was utilized for the metabolomics analysis, adapted from a previously established method³⁰. In detail, 100 mg of fecal material was homogenized with 300 μL of deionized water and centrifuged at 9450×g for 10 min at 4 °C. A 100 μL aliquot of the resulting supernatant or 100 μL serum was combined with 300 μL of methanol containing 10 μg/mL of the internal standard succinic acid-d₄. Following additional centrifugation (12,580×g, 10 min, 4 °C), 200 μL of the supernatant was transferred to a new vial and dried under nitrogen (N₂). The dried residue was then reconstituted in a 1:1 (v/v) mixture of acetonitrile and BSTFA (1% trimethylchlorosilane), vortexed for 30 s, and incubated at 70 °C for 60 min to facilitate derivatization. After this process, 1 μL of the supernatant was taken for GC–MS determination. A sample aliquot was injected in split mode at 1:40. The compounds were separated using a DB-624 capillary column (30 m length, 0.25 μm film thickness, 0.25 mm internal diameter, Agilent, USA). The temperature of the column was programmed to initiate at 60 °C for 5 min, followed by an increase of 10 °C/min until reaching 300 °C, where it was maintained for an additional 5 min. Mass spectrometric analysis was performed in SCAN mode, with ion mass data collected within 50 to 600 Da. The inlet temperature was maintained at 280 °C, the ion source at 200 °C, and the quadrupole at 300 °C, with a scan speed of 833 and an event time of 0.2 s. Quality control (QC) samples were regularly injected to ensure the consistency and reliability of the analytical results.

2.5. Targeted metabolomics of tryptophan metabolites analysis

The metabolome of serum and fecal samples was analyzed using a previously established tryptophan-targeted metabolomics approach^26,28. Briefly, 100 μL serum samples were mixed with 300 μL of methanol containing 100 ng/mL of an internal standard for extraction with the organic phase. Approximately 100 mg of fecal sample was accurately weighed, homogenized in 300 μL of deionized water, and centrifuged at 9450×g for 10 min at 4 °C. A 100 μL aliquot of the supernatant was mixed with 300 μL of methanol containing 100 ng/mL of an internal standard. The methanol extracts were then centrifuged at 12,580×g for 10 min at 4 °C, and the resulting supernatant was used for metabolomics analysis. Quantitative analysis of tryptophan (Trp), indole, kynurenine (Kyn), 5-HTP, 5-HT, indole-3-lactic acid (ILA), indole-3-propionic acid (IPA), indole-3-acetic acid (IAA), indole-3-acrylic acid (IArA), and indole-3-carboxaldehyde (IAld) was performed using an HPLC–MS/MS 8060 system.

2.6. Metabolomics data preprocessing and statistical analysis

The statistical analysis methodology employed for metabolomics data was derived from our previously established protocol^31,32. Initially, data were processed utilizing the ‘XCMS' package within RStudio for peak identification and retention time adjustment. The preprocessed data underwent normalization to enhance the clustering of QC samples within the principal component analysis (PCA) score plot. Features exhibiting a relative standard deviation (RSD) of less than 30% in the QC samples were selected for subsequent multivariate analysis. Following log transformation and Pareto scaling, orthogonal partial least squares-discriminant analysis (OPLS-DA) was conducted using SIMCA (v 14.0, MKS Umetrics AB, Sweden). Significant features were determined based on covariance (P[1]) and correlation (P(corr)) values in the S-plot, with criteria set at P[1] > 0.05, P(corr) > 0.50, or P[1] < −0.05, P(corr) < −0.50, and variable importance in projection (VIP) > 1 and volcano plots were generated in R. Metabolite identification was achieved by comparing the fragmentation patterns of the detected metabolites against spectra in the National Institute of Standards and Technology (NIST) database.

2.7. Mass spectrometry imaging analysis of brain tissue

Brain tissues were sectioned sagittally into 12 μm thick slices utilizing a Leica CM1860 cryostat. Upon sectioning, the brain sections were thaw-mounted onto 75 mm × 25 mm microscope slides. Then the slides were dried for 20 min under vacuum for subsequent mass spectrometry imaging (MSI) experiments. The adjacent cryosections were allocated for hematoxylin and eosin (H&E) staining following the standard protocol. 1,5-diaminonaphthalene (1,5-DAN) was dissolved in methanol at a concentration of 10 mmol/L and prepared as a matrix. The matrix was uniformly sprayed on the brain tissue sections using an ultrafine pneumatic sprayer. TransMIT AP-SMALDI 5 system equipped with a Thermo Scientific™ Q Exactive™ mass spectrometer was used for MSI image acquisition. Internal calibration was performed utilizing known 2,5-dihydroxybenzoic acid peaks³³. The mass range was set between 70 and 1050 m/z. For MSI analysis, a lateral resolution was set as 50 μm, and data were collected in negative mode, achieving a mass resolution of up to 70,000 at m/z 200. The acquired raw data were converted into imzML format and visualized using MSiReader. Ion intensity was normalized to total ion current (TIC), the recognized normalization method in MSI analysis. The detected metabolites were annotated using METASPACE, with an m/z tolerance of 10 ppm and FDR <0.1.

2.8. H&E staining

For H&E staining, paraffin-embedded sections were dewaxed in xylene (I and II) for 5 min, followed by treatment with an ethanol gradient. After dewaxing, the tissue sections were stained with hematoxylin for 20 min, rinsed with tap water, differentiated for 30 s, and soaked in water for 15 min. The sections were then stained with eosin for 30 s, followed by another rinse with tap water. The samples were soaked in water for 5 min, dehydrated through an ethanol gradient, cleared in xylene, and finally sealed with neutral resin. The slides were examined and photographed using an optical microscope.

2.9. Determination of short-chain fatty acid in vivo and in vitro

Fecal samples were homogenized in saline at a mass-to-volume ratio 1:9 and subjected to sonication for 30 min. The homogenate was then centrifuged at 9450×g for 15 min. The supernatant was treated with acetone (containing 1% (v/v) phosphoric acid) to extract SCFAs, which were subsequently measured. The gut microbiota cultures containing YZG-331 and M2, each at a final concentration of 50 μg/mL, were incubated in a shaker at 37 °C and 200 rpm for 6, 12, and 24 h, respectively. After each incubation period, acetone (containing 1% (v/v) phosphoric acid) was added for the extraction of short-chain fatty acids (SCFAs), including propionic acid, butyric acid and valeric acid. The mixture was then centrifuged at 9450×g for 10 min, and the supernatant was injected into a gas chromatograph (GC-2020, Shimadzu Corporation) to quantify SCFAs, following previously established protocols³⁴.

2.10. Quantitative real-time PCR (qPCR)

Total RNA was extracted using the Total RNA Isolation Kit (Trizol, Bio Teke, China) and reverse transcribed with HifairTM III 1st Strand cDNA Synthesis SuperMix for Quantitative PCR (qPCR) (gDNA digester plus, Yeasen, China), following the manufacturer's instructions. qPCR was conducted at 50 °C for 2 min and 95 °C for 10 min, followed by 40 cycles of PCR (95 °C for 15 s and 60 °C for 1 min), concluding with a final step at 95 °C for 15 s and 60 °C for 1 min to generate the dissolution curve and terminate the reaction. The primers utilized in this study are detailed in Supporting information Table S1 (Sangon Biotech, China).

2.11. 16S rRNA gene sequencing and analysis

Microbial DNA was extracted from fecal samples utilizing the E.Z.N.A.® Soil DNA Kit (Omega Bio-Tek, USA) following the manufacturer's guidelines. The amplification of the bacterial 16S rRNA gene V3–V4 region was performed using the primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). The resulting PCR products were purified from a 2% agarose gel using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, USA). The purified amplicons were subsequently sequenced using the Illumina MiSeq platform. The 16S rRNA sequencing data were analyzed using the Quantitative Insights Into Microbial Ecology (QIIME) software (version 1.9.1). Operational taxonomic units (OTUs) exhibiting a similarity of over 97% were classified and identified using the Greengenes database (version 13.8).

2.12. Pharmacokinetics of M2 in mice

Ten BALB/c mice were fasted for 12 h but had free access to water before administration. Then five mice were orally administered M2 (50 mg/kg) and five mice were intravenously administered M2 (0.1 mg/kg). 50 μL of blood from the fundus venous plexus was collected at 0, 5, 20, and 30 min, 1, 2, 4, 8, 12 and 24 h after drug administration. Then the whole blood was centrifuged at 500×g for 15 min to prepare plasma. Collected plasma was immediately stored at −80 °C before analysis. For tissue distribution study, the animals were fasted overnight but had free access to water before formal experiment. All mice were orally administered M2 (50 mg/kg). After 5 min, 30 min and 2 h, the heart, liver, spleen, lung, kidney and brain were collected and added with 3-fold volume of PBS before homogenization and stored at −80 °C before analysis.

2.13. Assessment of intestinal barrier permeability by FITC-dextran

Intestinal barrier permeability was assessed using a fluorescein isothiocyanate (FITC)–dextran tracer assay, as previously described with minor modifications. Mice were fasted for 6 h prior to the experiment with free access to water to minimize dietary interference. FITC-dextran (4 kDa) was freshly dissolved in PBS and administered by oral gavage at a dose of 600 mg/kg body weight. Four hours after administration, whole blood samples were collected via the orbital venous plexus and allowed to stand at room temperature for clotting. Samples were then centrifuged at 500×g for 10 min, and the supernatant serum was collected for analysis. Serum samples were diluted with PBS (1:5, v/v) and transferred to black 96-well microplates. Fluorescence intensity was measured using a microplate reader (BioTek, USA) at an excitation wavelength of 485 nm and an emission wavelength of 528 nm. A standard curve was generated using serial dilutions of FITC-dextran prepared in PBS. The serum concentration of FITC-dextran was calculated according to the standard curve and used as an index of intestinal permeability.

2.14. Regulation of tryptophan metabolism in single strains by YZG-331

Lactobacillus reuteri, Lactobacillus rhamnosus, Lactobacillus casei, Lactobacillus acidophilus, and Lactobacillus plantarum (five Lactobacillus strains) were cultured to 10⁶–10⁸ CFU/mL and centrifuged at 4 °C, 9450×g for 10 min to remove the culture medium. Each strain was then resuspended in PBS to eliminate endogenous interference from the medium, and the OD₆₀₀ was measured using a microplate reader to adjust the PBS volume to achieve a uniform bacterial concentration. Each of the five strains was co-cultured in medium with a final tryptophan concentration of 20 μg/mL, supplemented with either YZG-331 at a final concentration of 50 μg/mL or an equivalent volume of PBS. After co-culture at 37 °C for 24 h, the concentrations of tryptophan and 5-HT in the tryptophan metabolic pathway were measured.

2.15. In vitro incubation of YZG-331 with gut microbiota

A medium suitable for anaerobic culture was formulated by dissolving 9.5 g of anaerobic broth containing vital nutrients, inorganic salts, a reducing agent, and an anaerobic indicator in 1 L of purified water. The intestinal contents of BALB/c mice were weighed and combined with the anaerobic culture medium at 1:20 (g/mL) mass-to-volume ratio. Following thorough mixing, the system was purged with N₂ and incubated at 37 °C for 30 min to establish a gut microbiota culture solution. A stock solution of YZG-331 (1 mg/mL) was prepared, and 10 μL of this solution was introduced into a pre-sterilized centrifuge tube, resulting in a final concentration of 10 μg/mL within the system. This solution was mixed with 990 μL of the pre-incubated culture, maintaining anaerobic conditions through nitrogen purging. For the negative control group, a triple-boiled medium was utilized instead. The incubation system was agitated at 37 °C and 200 rpm for intervals of 0, 0.25, 0.5, 1, 2, 4, 6, 12, and 24 h. After each incubation period, the reaction was halted by adding a three-fold methanol solution containing 100 ng/mL of glipizide, followed by vortexing for 30 s and centrifugation at 9450×g for 10 min. A 5 μL aliquot of the supernatant was subsequently injected into an HPLC–MS/MS system for quantitative analysis.

2.16. In vitro incubation of YZG-331 with homogenate of liver

Livers obtained from BALB/c mice were homogenized with three-fold volumes of ice-cold saline to prepare the liver homogenate. The incubation system comprised a total volume of 200 μL, consisting of 2 μL of YZG-331 stock solution (1 mg/mL) and 198 μL of fresh liver homogenate. Incubation was conducted at 37 °C and 800 rpm with an oxygen supply. At intervals of 0, 30, 60, 90, and 120 min, the reaction was terminated by adding three times the volume of the glipizide methanol solution (100 ng/mL), followed by thorough mixing. After centrifugation at 9450×g for 10 min at 4 °C, 5 μL of the supernatant was injected for HPLC–MS/MS analysis.

2.17. Determination of the levels of YZG-331 by LC–MS/MS

The chromatographic conditions were as follows: mobile phase consisted of 0.1% formic acid in water as phase A and methanol as phase B. Chromatographic analysis was performed on a ZORBAX SB-C18 (2.1 mm × 100 mm × 3.5 μm) column at 40 °C, and flow rate was set to 0.4 mL/min, with glipizide serving as the internal standard. The gradient programming of the solvent system was initially at 20% B, 20%–60% B for 0–1 min, 70%–95% B for 2–6.5 min, and 95%–20% B for 6.5–7.5 min. Multiple Reaction Monitoring (MRM) modes were employed for the analysis: 386.85  →  255.15 for YZG-331, 445.90 → 321.05 for glipizide (IS).

2.18. Identification of YZG-331 metabolites by LC/MSⁿ-IT-TOF in vitro and in vivo

The separation of analytes was achieved using an Alltima C18 column (150 mm × 4.6 mm, 5 μm) at a flow rate of 0.4 mL/min, with the column temperature maintained at 40 °C. The mobile phase consisted of 0.5% formic acid in water as phase A and acetonitrile as phase B. The gradient programming of the solvent system was initially at 10% B, 10%–50% B for 0–15 min, 50%–80% B for 15–30 min, and 80%–10% B for 30–35 min. The autosampler was maintained at a temperature of 4 °C. The mass spectrometry conditions were established with a primary data acquisition range of approximately m/z 100–800, operating in positive ion mode. The sample prepared in the “In vitro incubation of YZG-331 with gut microbiota” section was added to acetonitrile, followed by vortexing for 30 s and centrifugation at 9450×g for 10 min in a refrigerated centrifuge at 4 °C. The supernatant was collected and evaporated to dryness under a nitrogen stream. The resulting residue was reconstituted in 200 μL of methanol, and a 10 μL aliquot was injected for analysis via LC/MSⁿ-IT-TOF. YZG-331 was uniformly ground with Tween-80 (less than 1% of the total volume) and prepared as a suspension at 10 mg/mL concentration using 0.5% carboxymethyl cellulose (CMC). An oral dose of YZG-331 was administered at 50 mg/kg. Fecal samples were collected 24 h post-administration. These samples were homogenized in saline at a mass-to-volume ratio 1:9 and subjected to sonication for 30 min. The homogenate was then centrifuged at 9450×g for 10 min. Following this, 200 μL of the supernatant was combined with three volumes of acetonitrile for protein precipitation. The resulting mixture was vortexed for 1 min and centrifuged at 9450×g for 10 min. The supernatant was evaporated to dryness under nitrogen gas and reconstituted in 100 μL of methanol, with a 10 μL aliquot subsequently injected for analysis using LC/MSⁿ-IT-TOF.

2.19. Effect of PGF mice model and on YZG-331 metabolism

The construction method of the PGF mice model was described in “Pseudo-germ-free (PGF) mice” section. YZG-331 was solubilized using Tween-80 and formulated into a 10 mg/mL suspension in 0.5% CMC. This suspension was orally administered to both control and PGF mice at 50 mg/kg. Fecal samples were collected 12 h post-administration and processed following the fecal sample preparation method from the “Identification of YZG-331 metabolites by LC/MSⁿ-IT-TOF in vivo” section. The semi-quantitative ions of YZG-331 and its metabolites are shown in Table 1, Table 2.

Table 1.

Characteristics of YZG-331 metabolites in the gut microbiota incubation system by HPLC/MSⁿ-IT-TOF.

Metabolite	Retention time (min)	Reaction	Fragment characteristic
			[M+H]+	MS/MS
YZG-331	23.4	–	386.21	254.17, 136.08
M1	24.5	Dehydroxylation	370.22	254.14, 136.08
M2	22.8	Hydrolysis	254.17	136.08
M3	16.2	Cleavage	136.12	119.08

Table 2.

Characteristics of YZG-331 metabolites in feces by HPLC/MSⁿ-IT-TOF.

Metabolite	Retention time (min)	Reaction	Fragment characteristics
			[M+H]+	MS/MS
YZG-331	23.4	–	386.21	254.17, 136.08
M1	24.5	Dehydroxylation	370.22	254.14, 136.08
M2	22.8	Hydrolysis	254.17	136.08
M4	19.2	Hydroxylation	270.15	152.06
M5	20.0	Hydroxylation	402.18	286.12, 136.06

2.20. Metabolic study of YZG-331 by 11 individual bacterial strains

Add 0.9 mL of sterilized culture medium to a 1.5 mL conical centrifuge tube, followed by the addition of 100 μL of each bacterial strain (Clostridium perfringens, Staphylococcus epidermidis, Enterococcus Faecium, Lactobacillus casei, Escherichia coli, Enterobacter cloacae, Proteus, Staphylococcus aureus, Lactobacillus reuteri, Enterococcus faecalis, Lactobacillus acidophilus). After flushing with N₂, seal the tubes and incubate at 37 °C for 1 h for pre-warming. Subsequently, add 10 μL of a 5.0 mg/mL methanol solution of YZG-331 and incubate in a shaking incubator at 37 °C. Samples were collected at 0 and 24 h. A 100 μL aliquot of the suspension was mixed with three volumes of methanol containing 100 ng/mL glipizide, vortexed for 30 s, and centrifuged at 9450×g for 10 min. The supernatant was dried under nitrogen gas. The residue was reconstituted in 200 μL of methanol, and 5 μL of the solution was injected for quantification by LC–MS/MS.

2.21. Heterologous expression of Lactobacillus reuteri-derived ribonucleoside hydrolase (RihC) in E. coli BL21(DE3)

The RihC gene (NZ_PUXG01000005.1) from Lactobacillus reuteri was codon-optimized for expression in E. coli BL21(DE3). The gene was amplified by PCR using specific primers, and the resulting product was digested with BamHI and XhoI restriction enzymes. It was then ligated into the pET-28a vector at the corresponding sites to generate the recombinant plasmid pET28a-RihC. The plasmid was transformed into E. coli BL21(DE3) competent cells. The transformed E. coli BL21(DE3) cells harboring the codon-optimized RihC gene were revived from glycerol stocks and cultured at 37 °C with shaking at 220 rpm. Protein expression was induced by adding 0.5 mmol/L isopropyl β-d-1-thiogalactopyranoside (IPTG) when the optical density at 600 nm (OD₆₀₀) reached 0.8–1.0. Induction was carried out overnight at various temperatures. Expressed proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

2.22. Regulation of tetrahydrobiopterin (BH₄), tryptophan hydroxylase (TPH), 5-HT and 5-HTP by YZG-331 in vitro

Intestinal contents were collected from 8-week-old female BALB/c mice and mixed with an anaerobic medium at a weight-to-volume ratio of 1:20 to establish an incubation system for intestinal bacteria. Following thorough mixing, the incubation system was purged with N₂ and maintained at 37 °C for a preincubation period of 30 min. The compound YZG-331, at final concentrations of 20 or 50 μg/mL, was then incubated with the gut microbiota culture for 6 and 12 h at 37 °C. After incubation, the gut microbiota cultures were centrifugated at 9450×g for 10 min, and the supernatant was collected for analysis. The levels of BH₄ and TPH were quantified using BH₄ and TPH ELISA kits at the 6 and 12 h time points. 100 μL of the gut microbiota culture containing YZG-331 at a final concentration of 50 μg/mL was mixed with three times the volume of a methanol solution containing 100 ng/mL of glipizide. This mixture was vortexed for 30 s and then centrifuged at 9450×g for 10 min. A 5 μL aliquot of the resulting supernatant was subsequently injected into a high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS) system to measure levels of 5-HTP and 5-HT.

2.23. Molecular docking

The interactions between YZG-331 and the ribonucleoside hydrolases RihA, RihB, RihC and IAG-NH, which are expressed by gut microbiota, were modeled using Discovery Studio 2023 software (version 23.1.100.23209). The crystal structures of RihA from E. coli were sourced from the Protein Data Bank (PDB) with identifier IYOE. The crystal structures of RihB and RihC from E. coli K-12 was obtained with the identifies 3MKM and P22564. The IAG-NH hydrolase, expressed by S. aureus, was modeled utilizing AlphaFold, referencing UniPort ID A0A380FYY4. Following the optimization of the initial structures of the target compounds and protein crystals, the CDOCKER method was employed to simulate the interactions between the optimized target compounds and the proteins. The proteins were maintained in a rigid conformation during the docking procedure, whereas the ligands were allowed total flexibility. The docking positions were subsequently refined through minimization steps, with all other parameters set to their default values.

2.24. Statistical analysis

Data management and statistical analysis were performed using two-way analysis of variance and Student's t-test via GraphPad Prism Version 8 (GraphPad Software, La Jolla, CA, USA). SIMCA version 14 (MKSUmetrics AB, Sweden) and R version 1.4.1 were also employed for data analysis. We used the ARRIVE reporting guidelines³⁵. Results are presented as means ± standard deviation (SD), with P-values less than 0.05 deemed statistically significant.

3. Results

3.1. YZG-331 ameliorated PCPA-induced sleep disturbance in mice and pointed to a gut tryptophan-5-HT signature

To establish the therapeutic phenotype and identify mechanistic clues, we preliminarily investigated the therapeutic efficacy of YZG-331 using a murine model of insomnia and then profiled fecal and serum metabolomes to search for gut-linked metabolic pathways. The flowchart of the experimental procedure is shown in Fig. 1A. In detail, we established a PCPA insomnia model according to a previously described method²⁴. Following seven days of continuous treatment with YZG-331, we assessed its hypnotic effects using pentobarbital sodium-induced sleep experiment. Pharmacological evaluation focused on the sleep latency and duration across control, model, and YZG-331 treatment groups. As illustrated in Fig. 1B, the model group exhibited significantly prolonged sleep latency compared to both control and YZG-331 treatment groups. Conversely, YZG-331 treatment had extended sleep duration relative to the model group (Fig. 1C). Oral administration of YZG-331 ameliorated PCPA-induced sleep disturbance in mice in a dose-dependent manner. Although intraperitoneal injection of YZG-331 exhibited pharmacological activity, it reduced sleep duration in mice by 23.9% compared to oral administration at the same dose (5 mg/kg). These findings underscore oral administration of YZG-331 represents the primary therapeutic approach for alleviating insomnia-related sleep disturbances. The gut may serve as a key site for the pharmacological activity of YZG-331.

Next, to elucidate the mechanism by which YZG-331 alleviates insomnia via the gut, non-targeted metabolomics analysis of fecal and serum samples from control, model, and YZG-331 group (20 mg/kg i.g.) was performed. As shown in Supporting information Fig. S1A and S1D, QC samples clustered tightly, confirming the reproducibility of the analytical methods. The Hotelling's T2 values didn't exceed the red line, indicating no outliers (Fig. S1B and S1E). Further analysis via OPLS-DA suggested a trend of separation between model and YZG-331 group (R²Y(cum) = 0.99, Q²(cum) = 0.913), suggesting significant differences in metabolic characteristics of feces (Fig. 1D). Permutation tests (n = 200) validated the model's robustness and predictive accuracy (Fig. S1C). Moreover, as depicted in Fig. 1E, covariance P[1] and correlation P(corr) values based on the S-plot were used to discern meaningful features (P[1] > 0.05, P(corr) > 0.50 and P[1] < −0.05, P(corr) < −0.50, VIP >1). Differential volcano plots further highlighted significant metabolites (Fig. 1F), leading to the identification of 26 differentially abundant fecal metabolites. Notably, YZG-331 treatment significantly altered amino acid metabolism, including tryptophan, threonine, glycine, aspartic acid, and serine. Among these metabolites, tryptophan and 5-HT exhibited pronounced changes, with 5-HT upregulated and tryptophan downregulated in the YZG-331 group compared to the model group (Fig. 1G). As shown in Fig. 1H and I, YZG-331 significantly modulated fecal levels of tryptophan and 5-HT, specifically reversing the PCPA-induced elevation of tryptophan and reduction of 5-HT (P < 0.01). Quantitative analysis of fecal samples revealed that YZG-331 significantly reduces tryptophan levels (P < 0.01, Fig. S1G). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was conducted on differential metabolites identified through non-targeted metabolomics. Analysis of the top 25 enriched metabolic pathways revealed perturbation in the tryptophan metabolism pathway (Fig. S1H).

Meanwhile, non-targeted metabolomics analysis was performed on serum samples. The OPLS-DA model suggested a trend of separation in serum metabolic profiles between the model and YZG-331 groups (Fig. 1J), with the OPLS-DA model stability validated through permutation testing (Fig. S1F). Differential metabolites were identified using an S-polt (P[1] > 0.05, P(corr) > 0.50 and P[1] < −0.05, P(corr) < −0.50, VIP >1) (Fig. 1K). Integration of differential volcano polt analysis and semi-quantitative results revealed that YZG-331 significantly increased serum levels of 5-HTP (Fig. 1L and M). In addition, correlation analysis between differential metabolites and insomnia indicators (sleep latency and duration) revealed a strong positive correlation between the 5-HTP in serum and 5-HT in feces, and sleep duration (Fig. 1N). These non-targeted results highlighted tryptophan metabolism and 5-HT metabolites as potential candidates, prompting targeted quantification of downstream tryptophan pathways.

3.2. Targeted tryptophan metabolomics identified 5-HTP/5-HT as the main metabolites to YZG-331

Untargeted metabolomics consistently pointed to tryptophan metabolism, we next performed targeted profiling of major microbial-regulated tryptophan branches to pinpoint the primary responsive axis. Tryptophan is a key metabolite mediating interactions between gut microbiota and the host. In the gut, the three major tryptophan metabolism pathways leading to 5-HT, Kyn, and indole derivatives are direct or indirect control of the microbiota. As shown in Fig. 2A, the result revealed that YZG-331 primarily modulated downstream tryptophan metabolic pathways in fecal samples, with particular emphasis on the indole pathway and the 5-HT biosynthesis pathway. Notably, it exerted a significant regulatory effect on the levels of 5-HTP and 5-HT in fecal samples. In the model group, fecal levels of 5-HT and 5-HTP were significantly reduced. Following YZG-331 intervention, fecal concentration of 5-HT and 5-HTP were significantly elevated (P < 0.01, Fig. 2B and C). Through correlation analysis between the levels of downstream tryptophan metabolites in feces and pharmacodynamic indicators revealed that 5-HT production via the three primary tryptophan metabolic pathways was associated with insomnia-related pharmacodynamic parameters. As shown in Fig. 2D, 5-HTP was negatively correlated with sleep latency and positively correlated with sleep duration. Additionally, metabolomic analysis of serum tryptophan metabolites demonstrated that YZG-331 significantly elevated serum levels of 5-HTP and 5-HT (P < 0.01), and modulated the levels of IPA and IAA within the indole metabolic pathway (Fig. 2E–G). The correlation analysis shown that 5-HT was positively correlated with sleep duration (Fig. 2H). This suggested that YZG-331 could exert its therapeutic effect on insomnia by modulating the conversion of tryptophan to 5-HTP and 5-HT. Notably, as the direct precursor of 5-HT, tryptophan metabolism occupies a central role in the regulation of sleep³⁶. Tryptophan metabolism in the gut is primarily regulated by the gut microbiota. The 5-HT pathway represents a major branch of this process, with approximately 90% of 5-HT synthesized in the gut. Tryptophan is converted into 5-HTP by the enzyme TPH. Subsequently, 5-HTP is decarboxylated by the enzyme aromatic amino acid decarboxylase to from 5-HT³⁷. The 5-HT is a critical neurotransmitter regulating sleep-wake cycles and closely associated with sleep quality³⁸. The prominence of 5-HTP/5-HT changes, together with their association with sleep duration, suggested a gut–brain link, which we next examined in brain tissues.

Figure 2.

5-HT and 5-HTP were the main metabolites in insomnia mice treated with YZG-331. (A) Heatmap depicting the abundance of tryptophan downstream metabolites in feces among each individual. (B) The concentration of 5-HTP in mice feces (n = 5). (C) The concentration of 5-HT in mice feces (n = 5). (D) The correlation between sleep latency and sleep duration with tryptophan downstream metabolites in feces. (E) Heatmap depicting the abundance of tryptophan downstream metabolites in serum among each individual. (F) The concentration of 5-HTP in mice serum (n = 5). (G) The concentration of 5-HT in mice serum (n = 5). (H) The correlation between sleep latency and sleep duration with tryptophan downstream metabolites in serum. Kyn, kynurenine; ILA, indole-3-lactic acid; IPA, indole-3-propionic acid; IAA, indole-3-acetic acid; IArA, indole-3-acrylic acid; IAld, indole-3-carboxaldehyde. ∗P < 0.05, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

3.3. Spatially metabolomic analysis to localize the alternated metabolic pathway of brain tissue after YZG-331 treatment

YZG-331 was extensively distributed within the intestinal tract and demonstrates significant sedative-hypnotic activity²². Through non-targeted metabolomics and targeted tryptophan metabolomics analyses demonstrated significant upregulation of 5-HT and 5-HTP in fecal and serum samples from the YZG-331 (i.g.) treatment group, with a positive correlation observed with sleep duration. 5-HT serves as the biochemical precursor of the sleep-regulating hormone melatonin and modulates sleep by binding to 5-HT receptors, thereby contributing to the treatment of insomnia. We speculated that YZG-331 might exert its pharmacological effects by modulating tryptophan metabolism in the “gut–brain axis”. To localize altered metabolites in brains of insomnia mice and further elucidate the therapeutic effect of YZG-331, the cryosections of mouse brain tissue were analyzed using atmospheric pressure (AP) scanning microprobe matrix-assisted laser desorption/ionization (SMALDI) mass spectrometry imaging (MSI) with a spatial resolution of 50 μm. The detected metabolites were annotated through METASPACE with 10 ppm mass tolerance and with false discovery rate (FDR) less than 20%³⁹.

To investigate insomnia-induced metabolic alterations in brain tissues and assess the regulatory effects of YZG-331, unsupervised clustering analysis was performed on metabolites across control, model and YZG-331 group (20 mg/kg i.g.). This analysis identified four distinct metabolic trends (Clusters 1–4, Fig. 3A). Specifically, metabolites in Cluster 1 of model group exhibited a significant increase at the overall level compared to those in the control and YZG-331 group. In Cluster 2, YZG-331 group significantly elevated metabolites levels relative to model group. To further characterize the metabolic alterations, we visualized the spatial distribution of key metabolites in Clusters 1 and 2. Fig. 3B (left panel) showed H&E stained brain tissue sections, while Fig. 3C (left panel) provided a color-coded schematic of brain regions for precise metabolic location. In the brain tissue of the insomnia-model group analyzed by MSI (Fig. 3B), the levels of 5-hydroxyindole, 5-hydroxyindoleacetaldehyde (5-HIAL), nicotinic acid, N-acetylaspartylglutamic acid, and benzoic acid were elevated. In the brain, 5-HT is initially converted into the intermediate 5-HIAL by monoamine oxidase (MAO), and subsequently oxidized to 5-hydroxyindoleacetic acid (5-HIAA) by aldehyde dehydrogenase (ALDH)⁴⁰. Elevated levels of 5-HIAL in brain indirectly reflect altered 5-HT metabolism, potentially influencing serotonergic activity.

Conversely, metabolites such as aspartic acid, glutamic acid, glutamine, taurine, and 5-hydroxytryptamine (5-HT) were upregulated in YZG-331 group (Fig. 3C). Aspartic acid plays a dual role in modulating sleep quality by influencing N-methyl-d-aspartate (NMDA) receptor activity and promoting the synthesis of gamma-aminobutyric acid (GABA)⁴¹. It can also be converted to oxaloacetate via the tricarboxylic acid (TCA) cycle, which is involved in glutamate synthesis and in the production of GABA⁴². Glutamic acid and glutamine serve as precursors to GABA, an inhibitory neurotransmitter that plays a key role in the pathophysiology of insomnia⁴³. Different regions of the brain are responsible for specific functions. The ventrolateral preoptic area (VLPO) of the hypothalamus is recognized as a central region of sleep promotion, while the hippocampus (HP) plays a crucial role in sleep regulation by influencing GABAergic neurotransmission, thereby influencing sleep quality and the occurrence of insomnia⁴⁴. 5-HT in the hippocampus plays a crucial role in modulating learning, mood and anxiety, primarily through various receptor subtypes like 5-HT1A⁴⁵. Meanwhile, LC–MS/MS analysis of 5-HTP and 5-HT in brain tissue revealed that their levels were elevated following YZG-331 administration (Fig. 3D and E).

KEGG pathway enrichment analysis was conducted separately on metabolites in Cluster 1 and Cluster 2. The results indicated that the metabolic pathways in Cluster 1 were primarily enriched in tryptophan metabolism (Fig. 3F), while those in Cluster 2 were mainly enriched in alanine–aspartate–glutamate metabolism (Fig. 3G). Notably, prior work has demonstrated that YZG-331 regulate the alanine-aspartate-glutamate metabolic pathway, significantly increasing brain GABA levels and producing sedative-hypnotic effects^20,46. Based on the regulatory effects of YZG-331 on the metabolic conversion of tryptophan to 5-HT, it is hypothesized that YZG-331 primarily ameliorated insomnia by modulating 5-HT levels via the “gut–brain axis”. Therefore, subsequent studies will focus on tryptophan metabolism and the 5-HT related mechanisms.

3.4. Gut microbiota was required for YZG-331 efficacy and linked 5-HT1A to orexin A regulation

Our previous pharmacokinetic studies of YZG-331 revealed substantial drug exposure in the gastrointestinal tract following oral administration²². Gut microbiota was identified as the primary site for its metabolic transformation. Specifically, YZG-331 undergoes hydrolysis to form the M2 metabolite (Supporting information Fig. S2A) (m/z 254), a process predominantly mediated by gut microbiota²³. The gut microbial-driven biotransformation underscores its critical role in mediating the therapeutic efficacy of YZG-331 for insomnia. YZG-331 regulated the tryptophan metabolic pathway and enhanced the conversion of tryptophan to 5-HTP and 5-HT, thereby ameliorating insomnia. To investigate whether the regulation of tryptophan metabolism by YZG-331 is mediated by the gut microbiota, and whether its gut microbial metabolite M2 exerts potential pharmacological activity, we conducted pharmacodynamic experiments using a chronic insomnia mouse model. The insomnia model was established by subjecting mice to a 5-day intermittent restraint in transparent tubes to simulate chronic stress-induced insomnia, followed by intraperitoneal injection of 350 mg/kg of PCPA (Fig. 4A). Sleep latency and duration were assessed using a pentobarbital-induced sleep test. On the 8th day of modelling, the insomnia mice exhibited significantly prolonged sleep latency and reduced sleep duration compared to the control group (P < 0.01, Fig. 4B). Serum and fecal samples were collected to quantify 5-HT and 5-HTP levels. As shown in Fig. 4C and D, 5-HT levels were significantly reduced in the insomnia model (P < 0.05). These findings suggest that 5-HT serves as a biomarker strongly associated with insomnia.

Figure 4.

YZG-331and M2 ameliorated insomnia via gut–brain axis. (A) Experimental scheme of the pharmacological effect of YZG-331. (B) The sleep latency time and sleep duration of insomnia model group 8 days after model induction (n = 5). (C) The levels of 5-HTP and 5-HT in the plasma of insomnia model mice (n = 5). (D) The levels of 5-HTP and 5-HT in the feces of insomnia model mice (n = 5). (E) Sleep latency time of the mice after 7 days of YZG-331 treatment (n = 5). (F) Sleep duration of the mice in each group after 7 days of YZG-331 treatment (n = 5). (G) The levels of 5-HTP and 5-HT in the feces of the mice in each group (n = 5). (H) The levels of 5-HTP and 5-HT in the plasma of the mice in each group (n = 5). (I) The levels of 5-HTP and 5-HT in the brains of the mice in each group (n = 5). (J) The mechanism by which YZG-331 enhanced 5-HT signaling to regulate sleep. (K) Relative mRNA expression of 5-HT1A in the brain. (L) Relative mRNA expression of 5-HT7 in the brain. (M) Concentration of Orexin A in the hypothalamus of mice after YZG-331 treatment. (N) Relative mRNA expression of Hcrtr2 in the brain after YZG-331 treatment. (O) Experimental scheme of the pharmacological effect of M2. (P, Q) Sleep latency time and sleep duration of the mice after M2 treatment. (R–T) Concentration of butyric acid in serum, feces, and brain. (U) Concentration of Orexin A in the hypothalamus of mice after M2 treatment. (V) Relative mRNA expression of Hctr2 in the mice after M2 treatment. (W) IBA-1 immunohistochemical assays of the hippocampal tissue (up-scale bar = 50 μm, down-scale bar = 20 μm). Data are presented as mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

Following YZG-331 treatment, the insomnia mice exhibited significantly prolonged sleep latency and reduced sleep duration compared to other groups (P < 0.05, Fig. 4E and F). Specifically, both low-dose and hose-dose YZG-331 (L-YZG-331 and H-YZG-331) treatments significantly improved sleep latency and duration in a dose-dependent manner, highlighting its potent sedative-hypnotic properties. To further verify the role of the gut microbiota in mediating the therapeutic effects of YZG-331 on insomnia, we established PGF mice via oral antibiotic administration⁴⁷. The observation of colony counts on Luria–Bertani (LB) solid media confirmed the successful establishment of the PGF mice model, as shown in Fig. S2B, with a significant reduction in colony numbers compared with the control group. In PGF mice, the sleep duration of YZG-331 was reduced. When compared with mice given the same dose, sleep duration decreased by about 56.2%, and the therapeutic effect was weakened. This shows that gut microbiota was important for the effect of YZG-331.

We next measured 5-HT and 5-HTP levels in serum, brain, and fecal samples to assess the impact of the tryptophan metabolic pathway on insomnia. Fecal analysis revelated that 5-HT and 5-HTP levels were restored to baseline in the L-YZG-331 and H-YZG-331 groups, but remained significantly reduced in the PGF-H-YZG-331 group, indicating impaired tryptophan metabolism likely due to gut microbiota depletion (P < 0.001, Fig. 4G). Corroborating this, serum and brain tissue analysis demonstrated that H-YZG-331 treatment reversed insomnia-associated reduction in 5-HT levels, whereas gut microbiota depletion relieved this effect in PGF-H-YZG-331 (Fig. 4H and I). Collectively, these findings underscored the critical role of the gut microbiota in converting tryptophan to 5-HT/5-HTP and further mediating the hypnotic effects of YZG-331. Based on these phenotypic and metabolomic alterations, we proposed a gut microbiota-mediated mechanistic model by which YZG-331 regulated tryptophan metabolism (Fig. 8A), which was subsequently used as a framework to investigate the molecular mechanism underlying YZG-331 induced enhancement of 5-HTP/5-HT production.

Figure 8.

The mechanism of YZG-331 in regulating tryptophan metabolism. (A) Schematic diagram of tryptophan metabolism. (B) The BH₄ levels in the gut microbiota cultures while cultured with YZG-331 (20 or 50 μg/mL) at 6 and 12 h in vitro were determined via ELISA (n = 5). (C) The TPH levels in the gut microbiota cultures while cultured with YZG-331 (20 or 50 μg/mL) at 6 and 12 h in vitro were determined via ELISA (n = 5) (D) The levels of 5-HTP in the gut microbiota cultures cultured with YZG-331 (50 μg/mL) at 6 and 12 h in vitro were determined by LC–MS/MS (n = 3). (E) The levels of 5-HT in the gut microbiota cultures cultured with YZG-331 (50 μg/mL) at 6 and 12 h in vitro were determined by LC–MS/MS (n = 3). (F) The BH₄ levels in the gut microbiota cultures with PCPA and YZG-331 (n = 5). (G) The BH₄ levels in the PGF-mice gut microbiota cultures with YZG-331 (n = 5). (H) The TPH levels in the gut microbiota cultures with PCPA and YZG-331 (n = 5). (I) The TPH levels in the PGF-mice gut microbiota cultures with YZG-331 (n = 5). (J) Schematic illustration of the hydroxylation-dependent mechanism of YZG-331 in gut microbiota. (K, L) The relative abundance of M4 and M5 (n = 6). (M) The BH₄ levels in the gut microbiota culture with YZG-331, miconazole and DAHP (n = 4). (N) The TPH levels in the gut microbiota culture with YZG-331, miconazole and DAHP (n = 4). (O) The concentration of 5-HTP in the gut microbiota culture with YZG-331, miconazole and DAHP (n = 6). (P) The concentration of 5-HT in the gut microbiota culture with YZG-331, miconazole and DAHP (n = 6). Data are presented as mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

5-HT modulates sleep primarily through interactions with various receptor subtypes. Multiple 5-HT receptor subtypes are expressed in the brain, with the 5-HT1A and 5-HT7 receptors being critical for regulating the sleep-wake cycle. Previous studies have demonstrated that activation of 5-HT1A receptors exerts an inhibitory effect on orexin neurons, thereby promoting sleep and reducing wakefulness (Fig. 4J)^48,49. As shown in Fig. 4K and L, compared to the control group, expression levels of 5-HT1A and 5-HT7 receptors in the brain tissue of model group were significantly reduced (P < 0.01). Following YZG-331 intervention, expression levels of these receptors were significantly elevated, whereas expression levels of these receptors in the PGF group exhibited a downward trend. Meanwhile, the concentration of orexin A in the hypothalamus was significantly elevated in the model group and markedly reduced following H-YZG-331 administration (Fig. 4M). In parallel, the expression of Hcrtr2 was significantly increased in the model group and effectively suppressed by YZG-331 treatment (Fig. 4N). Collectively, these results indicate that modulation of central 5-HT levels was closely associated with the regulation of 5-HT receptor expression, which contributes to downstream orexin signaling involved in sleep-wake regulation.

Furthermore, the M2-treated group exhibited significant improvements in both sleep latency and sleep duration (P < 0.05) as compared with the model group. However, the regulatory effect of M2 on 5-HTP and 5-HT was not significant. These findings indicated that M2 has potential pharmacological activity in the treatment of insomnia via alternative pathways. To further investigate whether the pharmacological activity of M2 was mediated by the gut microbiota, the pharmacokinetic properties and tissue distribution of M2 were initially investigated. The plasma time-concentration curves of M2 were shown in Fig. S2C and S2D. The pharmacokinetic parameters were shown in Supporting information Tables S2 and S3. The elimination half-life of M2 was 0.59 ± 0.17 h after intravenous administration and 2.40 ± 0.99 h after oral administration, respectively. The absolute oral bioavailability of M2 was calculated to be 13.09%. In addition, the tissue distribution of M2 was evaluated following oral administration (Fig. S2E). M2 was predominantly distributed in the liver and kidney, while only a small fraction was detected in the brain. Consistently, the calculated brain/plasma ratio was below 0.3, indicating limited penetration of M2 across the blood–brain barrier. Taken together, these results suggest that M2 is unlikely to exert a direct central effect on sleep regulation.

To further investigate the role of the gut microbiota in the therapeutic effects of M2 on insomnia. PGF mice were established and administered M2, followed by evaluation of its sleep-related efficacy. As shown in Fig. 4O–Q, depletion of the gut microbiota markedly attenuated the pharmacological effects of M2. Specifically, the sleep duration of PGF-M2 group decreased by about 56.2%. These results indicate that the presence of gut microbiota was required for the therapeutic efficacy of M2. Previous studies had shown that the gut microbiota-derived metabolite butyric acid promotes sleep by regulating the activity of orexin neurons in the lateral hypothalamic region, and orexin A plays a critical role in sleep regulation¹⁵.To investigate whether M2 affected butyric acid production, both in vitro and in vivo experiments were performed. M2 significantly promoted butyric acid production after 12 and 24 h in gut microbiota incubation system (Fig. S2F). Consistently, oral administration of M2 markedly increased butyric acid levels in feces, serum and brain tissue in vivo (Fig. 4R–T). In contrast, butyric acid was substantially reduced in PGF-M2 group, indicating that gut microbiota depletion impaired M2-induced butyric acid production. Furthermore, the levels of orexin A in the hypothalamus were examined. M2 group resulted in a significant reduction in hypothalamic orexin A levels and normalized the expression of Hcrtr2 compared with model group (Fig. 4U and V). Taken together, these results suggested that the gut–microbiota-derived metabolite M2 of YZG-331 may regulate sleep by modulating butyric acid levels in the gut microbiota, which was associated with a reduction in hypothalamic orexin A levels.

Moreover, histological assessment elucidated the neuroprotective effects of YZG-331. Chronic insomnia induced severe hippocampal damage in model and PGF-H-YZG-331 groups, characterized by neuronal swelling, vacuolation, and nuclear dissolution (Fig. S2G). H-YZG-331 treatment markedly alleviated these pathological changes. Immunohistochemical analysis of microglial activity showed an increase in intracellular calcium binding protein (Iba-1) immunostaining associated with insomnia (Fig. 4W). H-YZG-331 treatment suppressed microglial activation, thereby conferring neuroprotection. To quantitatively assess neuroinflammatory responses, the protein levels of tumor necrosis factor (TNF-α) and interleukin-1β (IL-1β) were measured in the hippocampus and cortex using ELISA. As shown in Fig. S2J–S2M, insomnia induction resulted in a marked elevation of both cytokines in the hippocampus and cortex compared with the control group. Treatment with H-YZG-331 significantly reduced the levels of TNF-α and IL-1β in both brain regions, with a more pronounced effect observed in the hippocampus. Notably, the anti-inflammatory effects of H-YZG-331 were markedly attenuated in PGF mice. These results demonstrated that YZG-331 alleviated insomnia-associated neuroinflammation at the molecular level in a gut microbiota-dependent manner.

3.5. YZG-331 restored intestinal barrier and inflammation, accompanied by recovery of SCFAs

Emerging evidence highlighted the critical role of the gut–brain axis in the linking intestinal barrier dysfunction to neuroinflammation. Chronic insomnia is known to disrupt intestinal barrier integrity, leading to mucosal injury and inflammatory cell infiltration. To evaluate the effects of YZG-331 on intestinal barrier damage induced by chronic insomnia, histopathological analysis of murine colonic sections was performed using H&E staining (Fig. 5A). In control group, colonic mucosa exhibited structural integrity, with orderly arranged intestinal epithelial cells and no detectable inflammatory cells. In contrast, the insomnia model group displayed severe intestinal epithelial damage, crypts loss, and extensive inflammatory cell infiltration. Similar damage was observed in the PGF-H-YZG-331 group. Notably, H-YZG-331 treatment markedly preserved mucosal structure, characterized by intact crypt architecture and reduced inflammatory cell recruitment. Quantitative analysis revealed a significant reduction in crypt depth in the model group compared with the control group (Fig. 5B), whereas H-YZG-331 treatment alleviated insomnia-induced crypt damage. Additionally, measurement of subepithelial space depth to evaluate mucosal edema (Fig. 5C), the results showed that the subepithelial space depth in the H-YZG-331 group was significantly decreased compared with the model group and PGF-H-YZG-331 group. Consistent with these histopathological findings, chronic insomnia triggered robust intestinal inflammation, as reflected by significantly elevated mRNA expression of TNF-α (Fig. 5D), IL-1β (Fig. 5E), and interleukin-6 (IL-6) (Fig. 5F) in colonic tissues. In contrast, H-YZG-331 treatment markedly suppressed the expression of these pro-inflammatory cytokines. Systemic anti-inflammatory effects were further confirmed by ELISA, showing significantly reduced serum TNF-α and IL-1β levels in H-YZG-331-treated mice (Fig. S2H and S2I).

Figure 5.

YZG-331 improved the intestinal barrier by promoting the expression of tight junction proteins and regulating short-chain fatty acids. (A) H&E staining and histopathology of mouse colon tissue. (B) The depth of the crypt (n = 5). (C) The subepithelial space depth of mucosal oedema (n = 5). (D–F) The mRNA expression of TNF-α, IL-1β, and IL-6 in the mouse colon (n = 3). (G–J) The mRNA expression of Zo1, Occludin, Tjp2, and Cgn in the mouse colon (n = 3). (K) The concentration of FITC-dextran in the plasma of mice. (L) Propionic acid content in the feces of animals in vivo (n = 5). (M) Butyric acid content in the feces of animals in vivo (n = 5). (N) Valeric acid content in the feces of animals in vivo (n = 5). (O) Propionic acid content in the coincubation system of intestinal bacteria and YZG-331 in vitro (n = 3). (P) Butyric acid content in the coincubation system of intestinal bacteria and YZG-331 in vitro (n = 3). (Q) Valeric acid content in the coincubation system of intestinal bacteria and YZG-331 in vitro (n = 3). Data are presented as mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

Tight junction proteins are essential for maintaining intestinal barrier integrity. Accordingly, we quantified the mRNA expression of tight junction–associated genes, including ZO-1, Tjp-2, Occludin, and claudin (Cng), in colonic tissues. H-YZG-331 treatment significantly upregulated the expression of these genes compared with the model group (P < 0.05; Fig. 5G–J), indicating restoration of tight junction integrity. To functionally evaluate intestinal barrier integrity in vivo, an intestinal permeability assay was performed using 4 kDa FITC-dextran. As shown in Fig. 5K, the insomnia model group exhibited a marked increase in serum FITC-dextran levels compared with the control group. H-YZG-331 group significantly reduced circulating FITC-dextran levels, demonstrating a functional restoration of intestinal barrier integrity. Notably PGF-H-YZG-331 group displayed significantly higher serum FITC-dextran levels than the H-YZG-331 group, indicating that depletion of gut microbiota attenuated the barrier-protective effect of YZG-331. These findings provided direct physiological evidence that YZG-331 improved physiological evidence that YZG-331 improved intestinal barrier function.

Given the critical role of SCFAs in intestinal barrier repair and gut–brain communication⁵⁰, SCFA levels were further quantified both in vivo and in vitro. In vivo analysis revealed that chronic insomnia significantly reduced fecal levels of propionic acid, butyric acid, and valeric acid, whereas YZG-331 treatment dose-dependently restored these SCFAs. Marked reductions were observed in the PGF group, confirming the gut microbiota as the primary source of SCFA production (P < 0.001). Importantly, the YZG-331-derived metabolite M2 was consistent with the previous results and could significantly regulate the levels of butyric acid (Fig. 5L–N). These indicated that both the parent compound and its microbiota-derived metabolite contribute to the regulation of butyric acid production. Consistently, in vitro incubation of gut microbiota demonstrated that YZG-331 significantly increased propionic acid, butyric acid, and valeric acid levels after 6, 12, and 24 h (Fig. 5O–Q). Together, these results suggest that YZG-331 can enhance SCFAs production through microbiota-dependent mechanisms, while its metabolite M2 further contributes to this effect.

3.6. YZG-331 attenuated insomnia-associated gut dysbiosis and enriched tryptophan metabolism related microbial

To determine whether the gut microbiota was altered in chronic insomnia and whether such dysbiosis could be modulated by YZG-331 to contribute to its therapeutic effects, we next characterized the gut microbial community in model mice following drug administration. Gut microbiota composition was profiled using 16S rRNA sequencing, with a focus on the V3 and V4 hypervariable regions. Chao1 and Shannon indices were downward trend in the model group compared to control (Fig. 6A and B). This reduction underscores a comprehensive decline in microbial diversity and structural dysbiosis triggered by chronic sleep deprivation. Strikingly, H-YZG-331 group restored both indices to levels comparable to those of the control group, suggesting its potential restorative effect on microbial diversity. Beta diversity analysis evaluates inter-sample bacterial community dissimilarity and similarity. There is pronounced dispersion between model and control groups without clustering, indicating a different bacterial community caused by insomnia (Fig. 6C). Based on the Bray-Curtis distance matrix, permutational multivariate analysis of variance (PERMANOVA) revealed statistically significant differences among groups (pseudo-F = 3.8055, P = 0.001), indicating substantial alterations in microbial community structure. These findings collectively demonstrated that YZG-331 partially ameliorated insomnia-induced gut microbiota dysbiosis. This restoration was further supported by the Venn diagram in Fig. 6D, which illustrates the operational taxonomic unit (OUT) across the control, model, and H-YZG-331 groups. Notably, the model group exhibited a significant reduction in OUT richness compared to both the control and H-YZG-331 groups. In detail, a total of 255 OTUs were shared between the control and H-YZG-331 groups. This finding further underscored the diminution of microbial diversity caused by insomnia and highlighted the restorative capacity of YZG-331.

Figure 6.

YZG-331 attenuated gut dysbiosis in the chronic insomnia mice. (A) Chao1 index of gut microbiota. (B) Shannon index of gut microbiota. (C) β-Diversity of gut microbiota. (D) The Venn diagram of OTU abundance across the control, model, and H-YZG-331 groups. (E) The distribution of intestinal bacteria at the phylum level. (F) The distribution of intestinal bacteria at the class level. (G) Phylogenetic tree and the circular heatmap of relative expression at the genus level. (H) Heatmap of the top 20 intestinal bacteria at the genus level (blue star: genera showing decreasing trends in the YZG-331 group; red stars: genera showing increasing trends in the YZG-331 group). (I) The relative abundance of Lactobacillus. (J) The relative abundance of Muribaculaceae. (K) Growth curve of Lactobacillus reuteri treated with YZG-331 (n = 3). Data are presented as mean ± SD. ∗P < 0.05, ∗∗∗∗P < 0.0001.

To further delineate the specific compositional alteration in the gut microbiota modulated by YZG-331, we performed a hierarchical analysis spanning the phylum, family, and genus levels. At the phylum level (Fig. 6E), the model group displayed a marked reduction in Bacteroidetes abundance and an increase in Firmicutes. This resulted in an elevated Firmicutes/Bacteroidetes (F/B) ratio. The F/B ratio is a well-established pathological indicator of gut dysbiosis, with higher levels frequently associated with inflammation¹². Notably, the model group displayed a significantly increase in F/B ratio than the control group, which is associated with the impaired gut barrier function and inflammatory infiltration. Remarkably, H-YZG-331 group restored the F/B ratio to levels comparable to that of the control group, suggesting that YZG-331 may ameliorate insomnia-induced gut microbiota dysregulation. At the class level, the model group exhibited a significant downregulation of Bacterodia and Bacilli, alongside an upregulation of Clostridia. In contrast, H-YZG-331 group restored the abundance of Bacteroidia, Bacilli, and Clostridia levels (Fig. 6F). Genus level analysis revealed distinct microbial signatures across the control, model, and H-YZG-331 groups. Fig. 6G presents a phylogenetic tree and circular heatmap of relative abundance at the genus level and the heatmap in Fig. 6H highlights the top 20 genera according to relative abundance. Specifically, the model group presented a reduction in Lactobacillus (Fig. 6I) and an upregulation of _NK4A136_group compared to control group. Conversely, H-YZG-331 group exhibited upregulation of Muribaculaceae (Fig. 6J), Bacillus, Enterorhabdus, and Lactobacillus. Using PICRUSt2, we predicted functional changes in the gut microbiota based on marker gene sequences. As shown in Supporting information Fig. S3A, KEGG annotations revealed that YZG-331 significantly regulated pathways related to carbohydrate metabolism, cofactor and vitamin metabolism, and amino acid metabolism. Key pathways influenced by YZG-331 induced in valine, leucine, and isoleucine biosynthesis; alanine, aspartate, and glutamate metabolism; glycine, serine, and threonine metabolism; and tryptophan metabolism (Fig. S3B). Moreover, to explore the interplay between microbial composition and pharmacodynamic outcomes, we performed Spearman correlation analysis among the 20 differentially abundant genera, sleep parameters (latency and duration), 5-HTP/5-HT levels, and inflammatory indicators (Fig. S3C). Sleep duration and brain 5-HT levels were positively correlated with Muribaculaceae, Enterorhabdus, and Lactobacillus. Muribaculaceae, members of the phylum Bacteroidetes, are primarily involved in intestinal carbohydrate fermentation. It contributes to the production of SCFAs, including acetic acid, propionic acid and butyric acid, which modulate intestinal immune responses and inflammatory processes⁵¹.

Previous study has established that Lactobacillus species modulate tryptophan metabolism, a critical pathway linked to neurotransmitter synthesis and gut–brain axis communication^52,53, and YZG-331 significantly increased Lactobacillus abundance. To further investigate the capacity of Lactobacillus species to metabolize tryptophan and synthesize 5-HT, while evaluating the stimulatory effect of YZG-331 on the growth of Lactobacillus strains, this study selected five phylogenetically distinct Lactobacillus strains, including Lactobacillus reuteri, Lactobacillus casei, Lactobacillus acidophilus, Lactobacillus rhamnosus, and Lactobacillus plantarum. Following 24 h of co-incubation with five distinct Lactobacillus strains, YZG-331 enhanced Lactobacillus growth. Notably, the growth-stimulatory effect on Lactobacillus reuteri increased with prolonged incubation time (Fig. 6K and Fig. S3D–S3G). Then, five Lactobacillus strains were co-incubated with tryptophan, YZG-331, for 24 h to evaluate strain-specific metabolic activity. As shown in Fig. S3H and S3I, among the five Lactobacillus species tested, Lactobacillus reuteri was found to accelerate tryptophan metabolism and enhance the production of 5-HT in the presence of YZG-331. These findings demonstrated that YZG-331 enhanced the growth of Lactobacillus reuteri, and under YZG-331 treatment, Lactobacillus reuteri further facilitated the metabolic conversion of tryptophan to 5-HT.

3.7. Gut microbiota mediated YZG-331 biotransformation and identified microbial enzymes contributing to M2 formation

Given that YZG-331 markedly reshaped the gut microbiota and that specific genus were associated with pharmacodynamic outcomes, we next investigated whether the gut microbiota directly participates in the biotransformation of YZG-331 and characterized its metabolic pathways in vitro and in vivo. To clearly elucidate the biotransformation of YZG-331 by gut microbiota, we conducted in vitro incubation assays comparison of murine gut microbiota and liver homogenate. An analytical method for YZG-331 was developed and validated based on established protocols⁵⁴. The molecular structure of YZG-331 is depicted in Fig. 7A, with the fragmentation pathways and ion fragment patterns provided in Supporting information Fig. S4A and S4B. To investigate the primary metabolic organ of YZG-331, it was incubated with murine cecal contents under anaerobic conditions in vitro. A negative control group, utilizing heat-inactivated gut microbiota in anaerobic media, was used to eliminate environmental interference. As shown in Fig. 7B, YZG-331 remained stable in the control group across all timepoint (0–24 h), with relative abundances ranging from 86.4% to 101.3%. In contrast, incubation with gut microbiota resulted in rapid and complete degradation of YZG-331. In detail, its residual levels decreased to 89.1% (0.25 h), 59.9% (0.5 h), 34.7% (1 h), and 8.5% (2 h), culminating in undetectable levels by 4 h. Parallel experiments in murine liver homogenates revealed markedly slower metabolic kinetics (Fig. 7C). YZG-331 exhibited a gradual decline in relative abundance over 2 h, especially with no complete depletion observed within the experimental timeframe. These differential metabolic efficiency highlights the indispensable of gut microbiota in the metabolism of YZG-331.

Figure 7.

Metabolic effects of the gut microbiota on YZG-331 and potential in vivo and in vitro metabolic pathways. (A) The structural formula of YZG-331. (B) The metabolic curve of YZG-331 by intestinal bacteria. (C) The metabolic curve of YZG-331 in liver homogenate. (D) Extracted ion chromatograms of YZG-331 and its metabolites. (E) Potential metabolic pathways of YZG-331 in vivo via the gut microbiota. (F) The relative expression levels of YZG-331 and its metabolites in the feces of PGF mice differed from those in the feces of control mice. (G) The relative expression levels of YZG-331 in 11 intestinal bacterial strains. (H) Relative abundance of YZG-331 in E. coli BL21(DE3) (WT) and E. coli BL21(DE3) (pET28a-RihC). (I) The concentration of M2 in E. coli BL21(DE3) (WT) and E. coli BL21(DE3) (pET28a-RihC). Data are presented as mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

To characterize the gut microbiota-mediated biotransformation of YZG-331, we performed LC/MSⁿ-IT-TOF analysis on in vitro gut microbiota incubations. These mass spectral fragmentation pathways and ion fragment patterns of YZG-331 were presented in Fig. S4A and S4B. In positive ion mode, the YZG-331 parent compound ([M+H]⁺ m/z 386.2150) generated three major metabolites, including M1 (m/z 370.2232), M2 (m/z 254.1693), M3 (m/z 136.1174) (Fig. S4C). MS/MS analysis indicated that M1 arises from dihydroxylation (loss of −OH, Δ−16.0 Da), while M2 corresponds to hydrolytic removal of a deoxyribose moiety (Δ−132.0 Da). M3, a secondary metabolite, likely results from further hydrolysis of M2 Δ−118 Da), consistent with sequential bond cleavage (Fig. S4D). The detailed tandem mass spectrometry information for the parent compounds and metabolites is provided in Table 1. Next, we investigated the in vivo biotransformation of YZG-331. Following the oral administration of YZG-331 to mice, fecal analysis identified four metabolites (M1, M2, M4, and M5) alongside the parent compound (Fig. 7D). M1 (m/z 370.2232) and M2 (m/z 254.1693) mirrored the in vitro dehydroxylation and hydrolytic pathway, respectively. Novel hydroxylation products were observed in vivo: M5 (m/z 402.1856, Δ+16.0 Da) represents a hydroxylation derivative of YZG-331, while M4 (m/z 270.1561, Δ+16.0 Da vs. M2) arises from hydroxylation of M2. Intriguingly, M4 may also originate from M5 (m/z 402.1856) via hydrolytic loss of a deoxyribose unit (−132 Da), suggesting interconnected hydroxylation-hydrolysis cascades (Fig. 7E). The detailed tandem mass spectrometry data provided in Table 2.

PGF mice were further employed to examine the gut microbiota dependency of YZG-331 metabolism. Fecal analysis showed that PGF mice exhibited altered drug metabolite profiles compared to control mice after YZG-331 administration. Specifically, hydrolytic (M2) and hydroxylated (M4 and M5) products were significantly reduced (P < 0.05, Fig. 7F), directly linking their production to the gut microbiota. To delineate microbial contributions to the metabolism of YZG-331, we cultured 11 individual bacterial strains in vitro and examined their capacity to metabolize YZG-331. Universal metabolic capacity was observed across strains, with Lactobacillus reuteri and Enterobacter cloacae achieving near-complete YZG-331 degradation by 24 h (Fig. 7G).

The gut microbiota mediated hydrolytic biotransformation of YZG-331 through targeted ribose cleavage, yielding its principal metabolite M2. Emerging evidence established commensal microbiota as critical modulators of drug bioavailability, with particular bacterial strains expressing ribonucleoside hydrolases (Rih) family of enzymes that govern nucleoside analog metabolism⁵⁵. Nucleoside hydrolases, which hydrolyze ribonucleosides, can be classified on the basis of substrate specificity into pyrimidine-specific (RihA), purine-specific (RihB), and nonspecific types (RihC). Single-bacterium experiments demonstrated that Lactobacillus reuteri exhibits significant metabolic activity towards YZG-331. National Center for Biotechnology Information (NCBI) database analysis revealed the presence of the RihC gene fragment (GeneID:77190328) in the genome of Lactobacillus reuteri. The pET28a-RihC plasmid was first constructed (Supporting information Fig. S5A), and agarose gel electrophoresis confirmed the successful insertion of the RihC gene fragment (921 bp) into the plasmid (6244 bp) (Fig. S5B). The recombinant plasmid was then transformed into E. coli BL21(DE3) competent cells via the heat-shock method (Fig. S5C). PCR amplification and electrophoretic analysis of transformed E. coli BL21(pET28a-RihC) (lanes 1–5), using wild-type E. coli BL21 as a negative control (lane WT) (Fig. S5D), confirmed the successful introduction of the Lactobacillus reuteri RihC gene into E. coli.

Expression of the RihC protein was further confirmed by SDS-PAGE analysis. The theoretical molecular weight of the enzyme was approximately 32 kDa. Wild-type E. coli BL21 (WT) exhibited basal endogenous expression of RihC (lane WT), whereas RihC protein levels were significantly elevated in the E. coli BL21(pET28a-RihC) overexpression strain, confirming the successful construction of the E. coli BL21(pET28a-RihC) engineered strain (Fig. S5E). In the functional verification experiment, E. coli BL21(WT) and E. coli BL21(RihC) strains were co-incubated with YZG-331 to monitor changes in YZG-331 and its metabolite M2. The results demonstrated that E. coli BL21(RihC) strain metabolized YZG-331 significantly faster than the E. coli BL21(WT) (Fig. 7H). The concentration of M2 increased in a time-dependent manner with prolonged incubation. Although E. coli BL21(WT) was also capable of metabolizing YZG-331, the M2 production rate was markedly lower than that of the E. coli BL21(RihC) (Fig. 7I). These findings confirm that the RihC from Lactobacillus reuteri efficiently catalyzes the biotransformation of YZG-331 into M2.

Further molecular docking simulations were conducted to model the binding interaction between the Lactobacillus reuteri RihC protein (UniProt: A0A256VPH7) and YZG-331. The results yielded a notably negative binding energy value (−3.07 kcal/mol), indicating that YZG-331 could form a stable complex with the RihC protein via intermolecular interactions, including hydrogen bonding and hydrophobic forces (Fig. S5F). These findings suggest a strong binding affinity between YZG-331 and RihC protein. Fig. S5G shows the crystal structure of other nucleoside hydrolase. Docking simulations with YZG-331 and nucleoside hydrolase proteins, which were expressed by E. coli and Staphylococcus petrasii. The CDOCKER interaction energy was calculated, as shown in Table 3. Both bacterial nucleoside hydrolases strongly bound to YZG-331. Additionally, 2D docking interaction mapping (Fig. S5H–S5K) identified histidine and tryptophan residues as pivotal for binding sites within the active centers, indicating that these residues are essential for nucleoside hydrolase activity in the metabolism of YZG-331. These results suggest that specific nucleoside hydrolases expressed by the gut microbiota play a crucial role in YZG-331 metabolism. Combined evidence from single-strain experiments and 16S rRNA sequencing analysis demonstrated that Lactobacillus reuteri participates in the hydrolytic metabolism of YZG-331 through the expression of RihC. Concurrently, YZG-331 was found to modulate the abundance of Lactobacillus reuteri and promote its metabolism of tryptophan into 5-HT. These findings reveal a bidirectional interaction between the drug and the gut microbiota: while the microbiota mediates drug metabolism, the drug in turn enhances microbial metabolic capacity by regulating microbial abundance.

Table 3.

The CDOCKER interaction energy of YZG-331 with nucleoside hydrolase proteins.

Organism	Hydrolase	PDB Code/UniPort	CDOCKER_INTERACTION_ENERGY
Escherichia coli	RihA	1YOE	−54.5234
Escherichia coliK-12	RihB	3MKM	−57.0495
Escherichia coliK-12	RihC	P22564	−54.5204
Staphylococcus petrasii	IAG-NH	A0A380FYY4	−52.1333

3.8. The mechanism of 5-HT production promoted by YZG-331 via hydroxylation reaction

Our investigation has revealed that YZG-331 could be metabolized by gut microbiota via two distinct pathways, including hydrolysis to yield M2, and hydroxylation to produce M4 and M5. The hydroxylation pathway often involves proton transfer⁵⁶. Previous study reported that dihydrobiopterin (BH₂) can be reduced to tetrahydrobiopterin (BH₄), an essential cofactor for conversation tryptophan to 5-HTP by proton transfer^28,57. We thereby hypothesized that proton transfer is involved in the hydroxylation of YZG-331 by the gut microbiota. As shown in Fig. 8A, we proposed that BH₂ might accept the protons generated during YZG-331 hydroxylation, thereby facilitating its conversion to BH₄. The generated BH₄ then activates TPH activity. Subsequently, the activated TPH activity leads to the conversion of tryptophan to 5-HTP, and ultimately to 5-HT synthesis. This proposed mechanism positions YZG-331 as a gut microbiota-dependent drug that enhances tryptophan metabolism, potentially underpinning its therapeutic efficacy in insomnia. To further validate this hypothesis, we incubated YZG-331 (20 and 50 μg/mL) with murine gut microbiota and quantified temporal changes in TPH and BH₄ levels. As shown in Fig. 8B, ELISA assays demonstrated a dose-dependent increase in BH₄ levels after 6 h in the YZG-331-treated group compared to controls (P < 0.001). Compared to high concentration YZG-331 (50 μg/mL) at 12 h, BH₄ levels plateaued in 20 μg/mL treatment group, likely due to hydroxylation substrate depletion during prolonged incubation. Concurrently, TPH levels in the 50 μg/mL YZG-331 group were significantly higher than untreated group (Fig. 8C). We also assessed the levels of 5-HTP and 5-HT in the incubation system, as shown in Fig. 8D and E. After 6 and 12 h of incubation, both 5-HTP and 5-HT levels were significantly elevated compared with those in the control group.

To further investigate the regulatory mechanism of YZG-331 on TPH activity, an in vitro incubation experiment was conducted. YZG-331 (50 μg/mL) was co-incubated with the TPH inhibitor (PCPA 100 μmol/L) in the intestinal contents of normal mice. In parallel, YZG-331 was also incubated with intestinal contents from PGF mice. Following a 6 h incubation in all groups, BH₄ and TPH levels were measured. Comparative analysis across treatment groups was used to elucidate whether YZG-331 exerts a direct or microbiota-mediated regulatory effect on TPH activity. As shown in Fig. 8F, PCPA significantly inhibited TPH enzymatic activity. After co-incubation of YZG-331 with intestinal contents from normal mice, TPH level was significantly increased compared to the untreated control group (P < 0.001). YZG-331 effectively reversed the PCPA-induced reduction in TPH activity. Notably, in the incubation system containing intestinal contents from PGF mice (Fig. 8G), YZG-331 failed to regulate TPH level, and the basal TPH level in this model was affected, likely due to homeostasis imbalance resulting from the absence of gut microbiota. Further analysis of BH₄ revealed that YZG-331 significantly increased BH₄ level in the incubation system containing normal gut microbiota (P < 0.01), with no significant difference observed relative to the PCPA treated group (Fig. 8H). In PGF incubation system, YZG-331 exhibited no regulatory effect on BH₄ level (Fig. 8I). Taken together, these findings suggest that YZG-331 enhanced BH₄ activity through gut microbiota-mediated metabolic processes, thereby indirectly improving TPH activity, and promote the synthesis of 5-HTP and 5-HT.

To further validate whether the hydroxylation pathway of YZG-331 was required for the activation of the BH₄-TPH, we employed pharmacological inhibitors targeting hydroxylation and BH₄ synthesis (Fig. 8J). As shown in Fig. 8K and L, inhibition of YZG-331 hydroxylation by miconazole (100 μmol/L) significantly reduced the relative abundance of hydroxylated metabolites M4 and M5 compared with YZG-331 treatment alone (P < 0.01), confirming that miconazole effectively blocked the hydroxylation pathway of YZG-331 in the gut microbiota incubation system. Consistent with the reduction in hydroxylated metabolites, BH₄ levels were significantly increased following YZG-331 treatment, whereas co-incubation with miconazole markedly attenuated this elevation. Further inhibition of BH₄ synthesis using DAHP (2 μmol/L) led to an additional reduction in BH₄ levels, indicating that YZG-331-induced BH₄ elevation was dependent on both hydroxylation and intact BH₄ biosynthesis (Fig. 8M). In parallel, TPH levels exhibited a similar regulatory pattern (Fig. 8N). YZG-331 significantly increased TPH levels in the gut microbiota incubation system, while miconazole partially suppressed this effect and DAHP nearly abolished it (P < 0.001), supporting the notion that activation of TPH by YZG-331 requires hydroxylation-driven BH₄ availability. Downstream metabolites of the tryptophan–5-HT pathway was further examined. As shown in Fig. 8O and P, YZG-331 markedly elevated the concentrations of 5-HTP and 5-HT, whereas inhibition of hydroxylation or BH₄ synthesis significantly attenuated these increases (P < 0.0001). Notably, combined inhibition resulted in the most pronounced suppression of 5-HTP and 5-HT production. To further investigate the relationship among hydroxylation, BH₄ and TPH activity, a rescue experiment was performed using the hydroxylation inhibitor miconazole with and without exogenous BH4 supplementation. As shown in Supporting information Fig. S6A, treatment with YZG-331 significantly increased the levels of BH₄ in the gut microbiota incubation system. However, inhibition of hydroxylation by miconazole markedly reduced BH₄ levels compared with YZG-331 group. Importantly, supplementation with exogenous BH₄ partially restored BH₄ levels under miconazole treatment. Consistently, inhibition of hydroxylation significantly decreased TPH levels, whereas exogenous BH₄ supplementation partially reversed the miconazole-induced reduction (Fig. S6B). Similarly, 5-HT concentrations were markedly reduced following miconazole treatment but were significantly restored upon BH₄ supplementation (Fig. S6C). Collectively, these results demonstrate that the hydroxylation of YZG-331 by gut microbiota is a prerequisite for BH₄ accumulation, which in turn activates TPH and promotes downstream synthesis of 5-HTP and 5-HT. Disruption of either hydroxylation or BH₄ biosynthesis effectively abolishes the regulatory effect of YZG-331 on the tryptophan–5-HT metabolic pathway.

Our results demonstrate that YZG-331 undergoes hydroxylation by gut microbiota and subsequently enhances the conversion of tryptophan into 5-HT. To further elucidate the central mechanism of 5-HT, an in vitro neuronal model was established using HT22 cells to examine the expression of the 5-HT1A receptor under different treatments (Fig. S6D). As shown in Fig. S6E, treatment with 5-HT (10 μmol/L) significantly increased the mRNA expression of 5-HT1A in HT22 cells, while 5-HTP (100 μmol/L) also showed an upregulating trend. In contrast, YZG-331 did not induce a significant change in 5-HT1A expression. These findings indicated that YZG-331 does not directly activate 5-HT1A signaling, but rather exerted its effect indirectly by enhancing 5-HT production. To further link gut microbiota-derived metabolites with neuronal signaling, HT22 cells were treated with the conditioned medium obtained from Lactobacillus reuteri co-incubated with tryptophan. The conditioned medium upregulated 5-HT1A expression in HT22 cells, and this effect was further enhanced when YZG-331 was added to the bacterial culture system (Fig. S6F). Collectively, these results demonstrated that YZG-331 promoted 5-HT production via gut microbiota metabolism, thereby upregulating 5-HT1A receptor expression and ameliorating insomnia.

3.9. YZG-331 retains therapeutic efficacy in humanized microbiota mouse models

To investigate whether the gut microbiota-dependent effects of YZG-331 could be preserved in a human-microbiome context, we established a humanized insomnia mouse model by transplanting fecal microbiota from healthy donors or insomnia patients into germ-free mice (Fig. 9A). After two weeks of colonization, the mice were treated with YZG-331 and subjected to pharmacodynamic and biochemical analyses. 16S rRNA gene sequencing demonstrated robust engraftment of human-derived microbial communities. FMT-Healthy and FMT-Insomnia mice exhibited significantly distinct alpha-diversity profiles, with reduced Chao1 and Shannon index observed in FMT-Insomnia mice (Fig. 9B and C). Beta diversity analysis based on Bray-Curtis distances revealed clear segregation among donor and recipient groups (Fig. 9D), and hierarchical clustering further confirmed that FMT mice clustered closely with their corresponding human donors (Fig. 9E). At the phylum level, FMT-insomnia mice recapitulated the features observed in insomnia donors (Fig. 9F). In addition to 16S rRNA gene sequencing, fecal culturing assays were performed to monitor bacterial colonization dynamics. Representative anaerobic culture plates showed a progressive increase in colony density from Days 0 to 7 after FMT, reaching a stable plateau by Day 14 in both FMT-Healthy and FMT-Insomnia groups (Fig. S6G). Quantification of fecal bacterial loads revealed that CFU were significantly colonized (Day 3), but recovered to levels comparable to control mice by Days 7 and 14 (Fig. S6H and S6I). These results demonstrated that the humanized gnotobiotic mouse model was successfully constructed.

Figure 9.

Humanized gut microbiota mediated the sleep effects of YZG-331. (A) Schematic illustration of the experimental workflow. (B) Chao 1 index of gut microbiota. (C) Shannon index of gut microbiota. (D) PCoA based on Bray-Curtis distances. (E) Relative abundance of phylum level. (F) Hierarchical clustering and taxonomic composition of gut microbiota at genus level. (G) Sleep latency in FMT-Healthy and FMT-Insomnia mice. (H) Sleep duration in FMT-Healthy and FMT-Insomnia mice. (I) The levels of 5-HTP and 5-HT in feces from healthy donors and insomnia patients. (J) The levels of 5-HTP and 5-HT in feces from FMT-Healthy and FMT-Insomnia mice. (K) The levels of 5-HTP and 5-HT in serum from FMT-Healthy and FMT-Insomnia mice. (L) The sleep latency of mice after YZG-331 treatment. (M) The sleep duration of mice after YZG-331 treatment. (N) The levels of 5-HTP and 5-HT in feces after YZG-331 treatment. (O) The levels of 5-HTP and 5-HT in serum after YZG-331 treatment. (P) Chao 1 index of gut microbiota after YZG-331 treatment. (Q) Shannon index of gut microbiota after YZG-331 treatment. (R) PCoA based on Bray-Curtis distances after YZG-331 treatment. (S) Relative abundance of gut microbiota at the phylum level after YZG-331 treatment. (T) Heatmap of the top 15 intestinal bacteria at the genus level. Data are presented as mean ± SD. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

To determine whether the insomnia-associated phenotype could be transferred via human gut microbiota, sleep behavior was evaluated in FMT mice. Compared with FMT-Healthy, FMT-Insomnia mice displayed significantly prolonged sleep latency and reduced sleep duration (Fig. 9G and H). Although the magnitude of change was modest, similar effect sizes have commonly been reported in microbiota-behavior studies⁵⁸, suggesting that insomnia-associated microbial communities may contribute to alterations in sleep-related phenotypes in recipient mice. Consistently, fecal and plasma levels of tryptophan-derived metabolites were significantly altered. Both human insomnia donors and FMT-Insomnia mice showed markedly reduced levels of 5-HTP and 5-HT in feces, indicating that dysregulated tryptophan metabolism is transferable through the human gut microbiota (Fig. 9I–K).

Importantly, oral administration of YZG-331 significantly shortened sleep latency and prolonged sleep duration in FMT-Insomnia mice, restoring these parameters to levels comparable to the FMT-Healthy group (Fig. 9L and M). In parallel, YZG-331 treatment markedly increased fecal and plasma levels of 5-HTP and 5-HT in FMT-Insomnia mice, whereas these metabolites remained low in untreated FMT-Insomnia controls (Fig. 9N and O). Further microbiota profiling revealed that YZG-331 partially normalized the microbial community in FMT-Insomnia mice. Alpha-diversity and beta-diversity analyses showed a shift of the YZG-331 treatment group toward the FMT-Healthy cluster (Fig. 9P–R). At the phylum level, FMT-Insomnia mice displayed a compositional shift characterized by increased Firmicutes relative to FMT-Healthy, whereas YZG-331 treatment partially reversed these alterations (Fig. 9S). At the genus level, YZG-331 treatment group enriched several taxa that were reduced in FMT-Insomnia mice, most notably Lactobacillus, a genus known to participate in tryptophan metabolism and 5-HT biosynthesis (Fig. 9T). Collectively, these results suggest that insomnia-associated gut microbiota may contribute to sleep-related phenotype and associated metabolic alterations in recipient mice, and that the gut microbiota-tryptophan-5-HT axis targeted by YZG-331 remains functionally responsive in a human-microbiome context. These findings support the potential translational relevance of YZG-331 in the treatment of insomnia.

4. Discussion

This study demonstrated that in PCPA-induced mouse models and chronic restraint-induced insomnia models, YZG-331 ameliorated insomnia via gut–brain axis in insomnia mice. In detail, YZG-331 modulated the tryptophan metabolic pathway via the gut microbiota, accelerating the conversion of tryptophan into 5-HTP and 5-HT. Meanwhile, YZG-331 and M2 increased SCFAs both in vivo and in vitro, repaired intestinal barrier dysfunction, alleviated insomnia-associated neuroinflammation and YZG-331 significantly improved insomnia-induced gut dysbiosis by increasing the abundance of Lactobacillus, involved in tryptophan metabolism, and Muribaculaceae, associated with SCFA production. Lactobacillus reuteri playing a critical role in both the metabolic processes of YZG-331 and the biosynthesis of tryptophan-derived metabolites. In vivo and in vitro metabolic analyses revealed that the gut microbiota exert dual metabolic effects on YZG-331: 1) hydrolyzing YZG-331 into its metabolite M2 through the expression of ribonucleoside hydrolase RihC; 2) promoting the conversion of BH₂ to BH₄ through hydroxylation-mediated proton generation, which indirectly activates TPH and enhances tryptophan metabolism. Previous studies have demonstrated that YZG-331 modulates adenosine receptors and the GABAergic system in the central nervous system, both of which play key roles in sleep-wake regulation. The present study identifies a previously unrecognized, gut microbiota dependent pathway through which YZG-331 enhances tryptophan metabolism and SCFA production, thereby indirectly regulating central serotonergic and orexinergic signaling. Through this indirect modulation, the sensitivity or responsiveness of adenosine and GABA-related signaling may be influenced, enabling coordinated central and peripheral actions. Together, these findings suggest that YZG-331 acts through dual and complementary mechanisms: a direct central pathway involving classical neurotransmitter systems, and an indirect peripheral pathway mediated by the gut–brain axis. This multi-target, multi-level mode of action may underlie the robust hypnotic efficacy of YZG-331.

Insomnia is roughly 1.5 times more prevalent in women than in men¹⁵. Adult women not only experience greater difficulty initiating sleep and early-morning awakening, but also rely on prescription sleep medications more frequently than men⁵⁹. In this study, we established the mouse insomnia model using female mice via PCPA (a selective tryptophan hydroxylase (TPH) inhibitor), followed by consecutive 7-day treatment of YZG-331 to evaluate its therapeutic efficacy. YZG-331 ameliorated sleep disorders in a dose-dependent manner, with oral administration demonstrating superior pharmacological activity compared to intraperitoneal administration. Non-targeted metabolomics analysis of fecal samples revealed that YZG-331 modulated multiple amino acid metabolic pathways, among which tryptophan plays a critical role in establishing and maintaining gut–brain axis function. Previous studies have demonstrated that tryptophan is partially converted into 5-HT, a key neurotransmitter involved in the regulation of central nervous system functions, including mood, anxiety, and behavior³⁶. Research involving insomnia patients with inflammatory bowel syndrome (IBS) has identified a correlation between insomnia and altered tryptophan metabolism⁶⁰. Notably, YZG-331 has already received approval for clinical trials, with its primary indication as a sedative-hypnotic agent. Our study demonstrates that YZG-331 significantly modulates intestinal tryptophan metabolism, providing theoretical support for its therapeutic efficacy in clinical trials targeting insomnia. We quantitatively detected downstream metabolites of tryptophan and found that YZG-331 primarily modulates the pathway responsible for 5-HT production from tryptophan. Correlation analysis with relevant pharmacodynamic indices revealed a significant association between 5-HTP and sleep-related parameters. Meanwhile, we employed spatial metabolomics to investigate the regulatory effects of YZG-331 on brain-associated metabolites and identified key metabolites involved in gut–brain axis regulation. Alterations in 5-HT levels in the hippocampus are critical for regulating circadian rhythms and sleep⁶¹. MSI revealed that YZG-331 treatment significantly increased 5-HT levels in the hippocampus. Differential metabolite enrichment analysis of Cluster 1 and Cluster 2 via KEGG showed that Cluster1 was predominantly enriched in the tryptophan metabolism pathway. Previous studies had confirmed that YZG-331 has the potential to regulate brain 5-HT levels, but specific mechanisms underlying its regulation of 5-HT production remain incompletely understood⁴⁶. Given that tryptophan is the biochemical precursor for 5-HT and that its metabolism is closely linked to gut microbiota. Follow-up research will therefore focus on elucidating the interactions between gut microbiota and YZG-331, with particular emphasis on how microbiota–drug interactions mediate 5-HT production through modulation of tryptophan metabolic pathway.

5-HT, a crucial metabolite derived from tryptophan, plays important for regulating circadian rhythms and the sleep-wake cycle, and it is crucial for promoting sleep⁶². The gut microbiota is involved in tryptophan metabolism, exerting a significant influence of 5-HT synthesis. Notably, more than 90% of 5-HT is produced by intestinal cells and regulated by the gut microbiota, which in turn affects mood and behavior⁶³. To simulate the onset and progression of insomnia, this established a chronic insomnia mouse model using chronic restraint stress combined with intraperitoneal injection of PCPA. In addition, a gut microbiota-depleted insomnia model was constructed through antibiotic pretreatment to evaluate the therapeutic effects of YZG-331. The result demonstrated that following. First, depletion of gut microbiota attenuated the effects of YZG-331 on sleep latency and sleep duration. Second, YZG-331 significantly increased the levels of 5-HTP and 5-HT in the tryptophan metabolic pathway. However, following depletion of gut microbiota, the levels of 5-HTP and 5-HT decreased significantly. These suggesting that the gut microbiota plays a crucial role in modulating the tryptophan pathway and mediating the therapeutic effects of YZG-331 in insomnia. Third, YZG-331 exhibited specific neuroprotective effects. Tryptophan is primarily converted into 5-HTP through catalysis by TPH and is subsequently synthesized into 5-HT. During its interaction with the gut microbiota, YZG-331 enhanced TPH enzymatic activity while simultaneously promoting the abundance of specific microbial species involved in tryptophan metabolism.

Chronic insomnia is frequently accompanied by brain inflammation, impairment of intestinal barrier, elevated proinflammatory factor levels, and gut microbiota dysbiosis¹³. SCFAs are essential for reducing inflammation and repairing the intestinal barrier. Our research revealed that YZG-331 can increase the levels of SCFAs such as propionic acid, butyric acid and valeric acid, both in vivo and in vitro models and the gut–microbiota metabolite (M2) increases the levels of butyric acid. Butyric acid, a key SCFAs, can regulate the expression of tight junction proteins and reduce paracellular permeability by activating single-molecule AMP-activated protein kinase (AMPK), which plays a crucial role in maintaining the function and integrity of the intestinal barrier⁶⁴. Moreover, accumulating evidence suggests that butyric acid has potential benefits in improving cognitive function and promoting sleep, further highlighting its significance in the context of insomnia-related pathologies. YZG-331 could enhance butyric acid production through modulation of gut microbiota activity, while the microbiota-derived metabolite M2 further contributed to the increase in butyric acid levels. Both YZG-331 and its microbiota-derived metabolite M2 contribute to the elevation of butyric acid and the activation of the butyrate–orexin signaling pathway. The 16S rRNA results revealed that oral administration of YZG-331 significantly improved the gut microbiota composition in insomnia model mice. In particular, there is an increase in Lactobacillus and Muribaculaceae levels. This increase might reverse the gut microbiota dysbiosis caused by insomnia, suggesting that YZG-331 may restore the balance of gut microbiota an potentially alleviate the associated symptoms of insomnia. Despite the limited sample size of 16S rRNA sequencing, this study detected no statistically significant differences in Lactobacillus abundance between the model and treatment groups. However, this result should be interpreted in the context of multidimensional evidence. Previous studies have established the critical regulatory role of Lactobacillus in tryptophan metabolism⁶⁵. Additionally, in vitro studies further indicated that Lactobacillus reuteri could directly participate in tryptophan metabolism and produce 5-HT. Together, these findings suggest a potential role for Lactobacillus in the YZG-331 mediated regulation of tryptophan metabolism.

We also systematically characterized the metabolism of YZG-331 by the gut microbiota in vitro and in vivo. YZG-331 undergoes dual metabolic pathways of hydrolysis and hydroxylation mediated by gut microbiota. Recent research indicated that the gut microbiota could express a variety of enzymes involved in drug metabolism⁵³. These enzymes include hydrolases, nitroreductases, and azoreductases. Previous studies have shown that after oral administration, YZG-331 mainly accumulates in the intestine, where it undergoes hydrolysis to form M2⁵⁵. Specifically, the gut microbiota metabolizes drugs by expressing specific hydrolases, among which ribonucleoside hydrolase is notable. This enzyme plays a role in the hydrolysis of ribose⁵⁵. YZG-331 can be metabolized by gut bacteria through the hydrolysis of one ribose molecule to produce the M2 metabolite. We also identified Lactobacillus reuteri as having a significant metabolic effect on YZG-331. Through overexpression experiments, we demonstrated that the hydrolytic metabolism of YZG-331 by Lactobacillus reuteri was mediated by RihC. Furthermore, we confirmed that gut microbiota express RihC, enabling the microbial conversion of YZG-331 into its metabolite M2. M2 has potential pharmacological activity in the treatment of insomnia.

In tryptophan metabolism, TPH serve as the primary enzyme, and BH₄ acts as a cofactor for TPH, facilitating the conversion of tryptophan to 5-HTP. This conversion represents the rate-limiting step in 5-HT synthesis. Drugs that undergo hydroxylation reactions in the gut can indirectly generate protons required for the conversion of BH₂ to BH₄. Based on the above-mentioned knowledge, we hypothesize that during the metabolic transformation of YZG-331 by the gut microbiota, many protons are produced. These protons accelerate the synthesis of BH₄, activate TPH, and thereby promote the conversion of tryptophan into 5-HT. This mechanism could potentially elucidate how YZG-331 exerts its therapeutic effects on insomnia. Experimental evidence further supports this hypothesis. When the gut microbiota is depleted, the efficacy of YZG-331 in treating insomnia is significantly diminished, and 5-HT levels are reduced. This clearly indicates that the gut microbiota plays a vital role in the pharmacological activity of YZG-331, highlighting the importance of the gut–brain axis in the action mechanism of this drug. In future studies, direct 5-HT intervention may be employed to further elucidate the underlying mechanisms.

To further evaluate the translational relevance of the gut microbiota-dependent mechanism, we established a humanized insomnia mouse model through FMT from insomnia patients and healthy donors. FMT-Insomnia mice exhibited prolonged sleep latency and moderately reduced sleep duration compared with FMT-Healthy mice, accompanied by decreased levels of tryptophan-derived metabolites such as 5-HTP and 5-HT. Although the magnitude of sleep change was modest, similar effect sizes were reported in microbiota–behavior studies⁵⁸. Importantly, YZG-331 treatment improved sleep parameters and restored tryptophan-derived metabolites in FMT-Insomnia mice, supporting the functional relevance of the microbiota–tryptophan–5-HT axis. Nevertheless, germ-free mice without microbiota transplantation were not included as an additional control, and the moderate behavioral differences observed warrant cautious interpretation. Therefore, this model should be viewed as supportive evidence for a contributory role of insomnia-associated microbiota rather than definitive proof of insomnia transmission.

We acknowledge that the relatively small sample size used for the omics analyses may limit statistical power, particularly for detecting subtle but biologically relevant changes. Therefore, the omics results should be interpreted with caution and are best viewed as exploratory findings that highlight major metabolic and microbial shifts associated with YZG-331 treatment. Future studies with larger cohorts will be necessary to further validate and refine these observations.

5. Conclusions

The gut microbiota plays a crucial role in mediating the therapeutic effects of YZG-331 on insomnia through the regulation of tryptophan metabolism and 5-HT production pathways. Lactobacillus reuteri primarily could directly participate in tryptophan metabolism and produce 5-HT. Specifically, YZG-331 is metabolized by the gut microbiota, which indirectly enhances the activity of the TPH enzyme, thereby activating tryptophan metabolic pathway and promoting the conversion of tryptophan into 5-HT to alleviate insomnia. Furthermore, YZG-331 was transformed into the metabolite M2 via the expression of ribose hydrolase by gut microbiota. YZG-331 increased the levels of SCFAs in the gut both in vivo and in vitro, contributing to regulate sleep, neuroprotection and the restoration of intestinal barrier repair. Meanwhile, both YZG-331 and M2 enhanced the production of butyric acid and activated the butyrate–orexin signaling pathway, thereby regulating sleep. This study provides valuable insights into the clinical application of YZG-331 in precision medicine, emphasizing the critical interplay between the gut microbiota and pharmacological agents. Despite these encouraging findings, this study has several limitations, including the use of only female mice and the preliminary nature of the proposed hydroxylation mechanism, which is supported by functional inhibition rather than direct measurement of proton transfer. Future studies in both sexes and in human cohorts will be required to further validate and refine these conclusions.

Author contributions

Yan Wang, Jiangong Shi, and Jinlyu Sun: conceptualization and methodology. Hengtong Zuo, Yan Wang, and Bin Zhao: writing-original draft preparation, writing-reviewing and editing. Qinglan Guo, Jinyue Lu, Ru Feng, Mengliang Ye, Jingyue Wang, Jianye Song, Jingyu Jin, Jiachun Hu, and Zhao Yi: resources. Hui Xu and Tiefeng Yuan: investigation. Hengtong Zuo, Wanxia Wang, Yongmei Tu, Lingyu Wei, Yifan Zhao: validation. Hengtong Zuo and Yan Wang: formal analysis. Jinlyu Sun, Yan Li, Li Sheng, Jianjun Zhang, Yan Wang: supervision.

Conflicts of interest

The authors declare no conflicts of interest.

Appendix A. Supporting information

The following is the Supporting information to this article.