DZF is composed of six Chinese herbal medicines: Coptis chinensis Franch. (Huanglian), Citrus×aurantium L. (Zhishi), Pinellia ternata (Thunb.) Makino (Banxia), Trichosanthes kirilowii Maxim (Gualou), Cinnamomum cassia (L.) J. Presl (Rougui), and red yeast rice (Hongqu). DZF extract (freeze-dried powder) was produced and supplied by Zhejiang Jiu Xu Pharmaceutical Co., LTD. (Batch Number: YC20200501). The extraction method and quality control are detailed in the Chinese patent document titled “A Traditional Chinese Medicine Composition and Its Preparation for Treating Metabolic Syndrome” (Chinese Patent No: ZL201811080557.5) (Xu et al., 2023). High-performance liquid chromatography (HPLC) was used to determine the major active chemical components in DZF, including berberine, naringin, hesperidin, and lovastatin (Zhu et al., 2018). DZF freeze-dried powder was dissolved in ultrapure water with the assistance of ultrasonication to prepare the corresponding concentrations of DZF solution. Metformin hydrochloride tablets (Sino-American Shanghai Squibb Pharmaceuticals Ltd.) were crushed and dissolved in ultrapure water to prepare the corresponding concentrations of metformin solution. Both solutions were freshly prepared and used for animal experiments.

The primary antibodies used included: ZO-1 Antibody (#DF2250, Affinity, United States, 1:500), Occludin Antibody (#DF7540, Affinity, United States, 1:1000), APOB Polyclonal Antibody (20578-1-AP, Proteintech, United States, 1:1000), CD36 Polyclonal Antibody (18836-1-AP, Proteintech, United States, 1:500), FASN Polyclonal Antibody (10624-2-AP, Proteintech, United States, 1:5000), SCD Polyclonal Antibody (28678-1-AP, Proteintech, United States, 1:2000), CPT-1𝛂 Polyclonal Antibody (15184-1-AP, Proteintech, United States, 1:5000), FATP2 Polyclonal Antibody (14048-1-AP, Proteintech, United States, 1:2000), and β-Actin (13E5) (4970S, Cell Signaling, United States, 1:1000). Total cholesterol (TCH/T-CHO) measurement reagent kit (A11-1-1), triglyceride (TG) measurement reagent kit (A110-1-1), and total protein quantification measurement reagent kit (A045-4) were purchased from the Nanjing Jiancheng Bioengineering Research Institute (China). The Mouse Lipopolysaccharides ELISA Kit (CSB-E13066m) was obtained from CUSABIO. Glucose Oxidase Method Kit (E1010) was obtained from Applygen (China).

Experimental mice used in this study included C57BL/Ksj-Lepr db/db mice with leptin receptor deficiency as the test subjects and their littermates, heterozygous db/m mice, as the normal controls. A total of 36 db/db mice and 12 db/m mice, all males and 6 weeks old, were obtained from Changzhou Cavens Laboratory Animal Co., Ltd. (Animal license number: SCXKSu20160010). The mice were housed throughout the experiment in the specific pathogen-free (SPF) animal facility of Guang’anmen Hospital, maintained at a temperature of 25°C ± 1°C, a relative humidity of 55%–65%, and a 12-h light/12-h dark cycle. All mice had free access to standard rodent chow and water. Every effort was made to minimize any potential suffering of the mice. All experiments were approved by the Ethics Committee of IMPLAD (Institute of Medicinal Plant Development), CAMS & PUMC (Beijing, China, Approval NO. SLXD-20170323019), and performed under the guidance of the National Act on use of Experimental Animals (China). After a 2-week acclimation period, the db/db mice were randomly divided into four groups, each consisting of 12 mice. The normal control group (db/m mice) and the model control group mice (db/db + Veh) received ultrapure water (5 mL/kg). The remaining db/db mice were divided into two groups: one received metformin (Met) solution (0.25 g/kg, db/db + Met), and the other received DZF solution (0.40 g/kg, db/db + DZF). All drugs were administered in the form of a solution using ultrapure water as the vehicle. Mouse body weight was measured weekly, and the treatment continued for 12 weeks. At the end of the treatment period, the mice were fasted for 12 h but had access to water. Blood samples were collected under anesthesia, and liver and ileum tissues were isolated. Blood samples were left to stand at room temperature for 30 min, centrifuged at 3,000 rpm for 15 min, and the upper serum was collected and stored at −80°C. Liver and ileum tissues were either stored at −80°C or fixed in 4% paraformaldehyde.

FBG levels were measured at weeks 0, 4, 8, and 12 after drug administration. Each measurement was initiated at the same time, with the mice subjected to a 6-h fasting period without water deprivation. The procedure involved securing the mice, disinfecting the tail tip with 75% alcohol, trimming approximately 1 mm from the tail using ophthalmic scissors, and collecting about 20 μL of blood. A cotton ball was gently applied to the tail tip to stop bleeding. After allowing the blood to stand for 30 min, it was centrifuged at 3,500 rpm for 10 min. Subsequently, 3 μL of the upper serum layer was mixed with 9 μL of ultrapure water to obtain a serum sample diluted fourfold.

A gradient dilution of a 10 mM glucose standard solution was prepared to the following concentrations: 1000, 500, 250, 125, 62.5, 31.25, and 15.625 μM. The working solution was prepared by mixing reagents R1 and R2 in a 4:1 ratio and used immediately. In a 96-well plate, 5 μL of either the serum sample or standard solution was added sequentially, and ultrapure water was used as a blank control well. Each well received 195 μL of the working solution, and the plate was shaken and incubated at room temperature for 30 min. Glucose concentrations were determined by measuring the absorbance at 550 nm using a microplate reader, and a standard curve was generated for glucose concentration calculations.

In the 11th week of drug administration, after a 6-h fasting period, blood glucose measurements were conducted following the same method as described in Section 2.3. After the baseline blood glucose measurement at 0 min, all db/db mice received intraperitoneal insulin injections at a dose of 0.75 U/kg. Blood glucose levels were subsequently measured at 30-, 60-, 90-, and 120-min post-insulin injection. The area under the curve (AUC) was evaluated and calculated using GraphPad Prism Version 9.

Serum samples stored at −80°C were thawed at room temperature. Subsequently, 100 μL of serum was collected and analyzed for triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), aspartate aminotransferase (AST), and alanine aminotransferase (ALT) levels using a fully automated clinical chemistry analyzer.

Liver tissue TG were quantified using the Glycerol-3-Phosphate Oxidase-Peroxidase (GPO-PAP) enzymatic assay kit. Liver tissue samples were retrieved from −80°C storage and 100 mg of tissue was taken. These tissue samples were then added to 900 µL of pre-chilled physiological saline. A handheld electric homogenizer was used to thoroughly homogenize the tissue, followed by centrifugation at 2,500 rpm for 10 min. After centrifugation, three distinct layers were visible: cell debris at the bottom, clear liquid in the middle, and a white lipid layer on top. The clear liquid from the middle layer was collected and used for analysis. In a 96-well plate, 2.5 µL of ultrapure water (blank well), standard solution (standard well), or tissue clear liquid (sample well) were added to separate wells. Each well was then supplemented with 250 µL of working solution, and the plate was incubated at room temperature with agitation for 25 min. Absorbance was measured at a wavelength of 500 nm to determine TG concentrations. TC in the liver was determined using the Cholesterol Oxidase-Peroxidase Method (COD-PAP) reagent kit. The procedure was similar to the one described above, with absorbance measured at a wavelength of 510 nm. The protein concentration corresponding to the liver tissue clear liquid was determined using the Bicinchoninic Acid (BCA) method.

Taking the fixed liver and ileum tissues, paraffin embedding and sectioning were performed. The sections were subjected to Hematoxylin and Eosin (H&E) staining. Optical microscopy analysis revealed cytoplasm in red and nuclei in blue. Frozen sections of liver tissue stored at −80°C were prepared and subsequently stained with Oil Red O after fixation. Optical microscopy analysis showed lipid droplets in shades of orange-red or bright red, with blue-colored nuclei.

On the 12thweek, stool samples were collected and immediately stored at −80°C preparing for DNA extraction. Total microbial genomic DNA from samples of all groups was extracted with CTAB/SDS method, with the purity determined on agarose gels. The V4 region of the 16S rRNA gene was amplified with 515F-806R primers. The PCR products were purified by Agarose gel electrophoresis with GeneJET gel recovery kit (Thermo Fisher, United States). For Library preparation, Ion Plus Fragment Library Kit 48 rxns (Thermo Fisher, United States) was used. For sequencing, Life Ion S5TM or Ion S5TMXL (Thermo Fisher, United States) were used.

The sequencing data were processed to obtain Clean Reads, and sequences with 97% identity were clustered into Operational Taxonomic Units (OTUs) by Uparse software. For species annotation analysis, Mothur method and SSUrRNA database of SILVA were used to obtain different taxonomic classifications. For alpha and beta diversity, Chao1, Shannon, UniFrac distance, UPGMA sample cluster tree was calculated or constructed using Qiime software (1.9.1). R software (2.15.3) was applied to draw the dilution, species accumulation curves, and the PCoA diagrams. The Adonis analysis was performed by R. For Linear Discriminant Analysis (LDA) Effect Size (LEfSe) analysis, LEfse software (1.0) was used with 4 as the filtering value of LDA score by default. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) was adopted to predict metagenome functions of intestinal flora responded to different interventions. PICRUSt is a form of phylogenetic investigation under 16S Bioinformatics software package for metagenomic function prediction by rRNA. Detailed forecasting process to view (http://picrust.github.io/picrust/tutorials/algorithm_description.html). Comparison of PICRUST results between the two groups was performed by the t-test of R.

Frozen serum samples were thawed at room temperature. ELISA kit reagents were equilibrated to room temperature, and a gradient of standard solutions was prepared according to the manufacturer’s instructions to create the necessary working solution. Samples and standard solutions were sequentially added to a 96-well plate for the ELISA reaction. The absorbance was measured at 450 nm using a microplate reader within 5 min after adding the stop solution.

Mouse liver and ileum tissues were lysed in high-efficiency RIPA lysis buffer (Beyotime, China) on ice. The tissue lysates were centrifuged at 12,000 rpm at 4°C for 15 min to obtain the protein-containing supernatant. Protein concentration was determined using the BCA protein quantification assay kit (Applygen, China), and 50 µg of protein was loaded per well for SDS-PAGE gel electrophoresis. SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) gel electrophoresis was performed, and the proteins were then transferred to an activated polyvinylidene difluoride (PVDF) membrane using transfer buffer containing distilled water and methanol. The PVDF membrane was subsequently blocked with 5% skim milk and incubated with the respective primary antibodies overnight at 4°C, with β-actin serving as the internal reference protein. The membrane was then incubated with a secondary antibody conjugated with horseradish peroxidase (HRP) (Goat anti-Rabbit IgG, HRP Conjugated, Cwbio, 1:5000) at room temperature for 1 h. Enhanced chemiluminescence (ECL) plus (Beyotime, China) was applied to the membrane surface, and the membrane was placed in the Molecular Imager Gel Doc XR + System (Bio-Rad, United States) for image acquisition. Image Lab software was used to acquire images, and subsequently, ImageJ was employed for image analysis.

Total RNA from mouse liver and ileum tissues was extracted using the RNAsimple Total RNA Extraction Kit (Tiangen, China). The RNA concentration was determined using a spectrophotometer (Nanodrop 2000c; Thermo Fisher, United States). cDNA synthesis was performed using the FastKing gDNA Dispelling RT SuperMix (Tiangen, China). The qPCR reaction system was prepared using SuperReal PreMix Plus (SYBR Green) (Tiangen, China), and qPCR was conducted in the qTOWER3 Real-time PCR thermal cycler (Analytik Jena, Germany). β-Actin was used as an internal control, and the relative expression of the target genes was calculated using the comparative 2^−ΔΔCT method. The primers corresponding to the target genes are listed in Table 1.

Data were expressed as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism Version 9. Unpaired t-test and One-way ANOVA were used for continuous variables. Differences were considered statistically significant at p < 0.05.

At the initiation of treatment, in the 8th week of age, the body weight of db/db mice was significantly higher than that of db/m group (p < 0.01). Although db/db + Veh group showed a slight decrease in body weight after the 7th week of treatment, there was no statistically significant difference in body weight at the 12th week compared to the peak body weight observed at the 6th week. Following Met intervention, a decrease in body weight in mice was observed from the 7th week onwards. While there was no statistical difference in body weight between Met and Veh groups until the end of the treatment, at the 12th week, the body weight of the Met group was significantly lower than the peak body weight observed at the 6th week (p < 0.05). DZF intervention led to a decrease in body weight in db/db mice from the 5th week of treatment. At the 12th week, the body weight in the DZF group was significantly lower than that at the peak of the 4th week (p < 0.01) and also significantly lower than that of the Veh group at the 12th week (p < 0.01) (Figure 1A).

FBG levels in db/db mice were significantly higher than in db/m mice at the start of the intervention (p < 0.01). Over the 12-week observation period, FBG levels in db/m mice remained stable, while in all three db/db groups, FBG levels showed a continuous upward trend. At each FBG measurement time point during the 12-week treatment, FBG in db/db + Veh group was significantly higher than in db/m group (p < 0.01). At weeks 4, 8, and 12, FBG in db/db + Met group was significantly lower than in db/db + Veh group, with statistical differences (p < 0.05 or p < 0.01). DZF intervention suppressed the rise in FBG levels, with a significant reduction compared to the Veh group at the 12th week (p < 0.05) (Figure 1B).

After 11 weeks of treatment, db/db mice, excluding db/m mice, underwent an ITT test. Met and DZF significantly reduced blood glucose levels at 30, 60, 90, and 120 min after insulin injection (p < 0.05 or p < 0.01) and significantly reduced the area under the curve (p < 0.05 or p < 0.01) (Figures 1C, D).

At 12 weeks of treatment, compared to db/m group, db/db + Veh group showed a significant increase in serum TG, TC, and LDL-C (p < 0.05 or p < 0.01), and a significant decrease in HDL-C (p < 0.05). Compared to db/db + Veh group, the Met group exhibited a significant reduction in TG and LDL-C (p < 0.05 or p < 0.01). The DZF group showed a significant reduction in TG and LDL-C (p < 0.05 or p < 0.01) and a significant increase in HDL-C (p < 0.01) (Figures 1E–H).

As described in the methods, fecal samples were collected from all four groups of mice at week 12, and their microbiota composition was determined. A rarefaction curve and species accumulation model (Specaccum) analysis were employed. The observed species richness and diversity plateaued as sequencing sample size and depth increased, indicating that the current sample size and sequencing depth were sufficient to analyze community structure (Figures 2A, B). At the level of α diversity, db/db + Veh mice exhibited significantly increased richness and diversity compared to db/m mice (p < 0.05 or p < 0.01), and Met-treated db/db mice also displayed higher richness (p < 0.01) (Figures 2C, D). DZF reduced microbial diversity compared to the Met group (p < 0.05) (Figure 2D) but showed no significant difference compared to db/m mice and other db/db mice.

To investigate the overall differences in microbiota structure among groups, principal coordinate analysis (PCoA) based on UniFrac distances was used for β diversity analysis. The differences in PCoA were assessed using Anosim. The unweighted PCoA (Figure 2E) clearly showed a larger difference between gut microbiota of db/m mice and db/db + Veh mice (Anosim, r = 0.9026, p = 0.001). DZF intervention resulted in a significant shift compared to db/db + Veh mice (r = 0.4937, p = 0.001). Relative abundance at the phylum level among groups was compared using unweighted UniFrac distances (Figure 2F), suggesting that the gut microbiota of DZF-treated db/db mice was more similar to that of db/m mice. These results reflect that DZF can alter the gut microbiota composition of db/db mice to make it more similar to that of db/m mice.

Using PICRUSt based on 16S rRNA gene sequencing data and the KEGG database, the impact of DZF on gut microbiota metabolic function in db/db mice was predicted. Figures 2G, H mainly demonstrate differences between groups at KEGG level 2 and level 3 within the KEGG level 1 metabolism category. Differences in gut microbiota metabolic function between db/m and db/db mice primarily involved carbohydrate metabolism, amino acid metabolism, cofactors and vitamins, and lipid metabolism (p < 0.05 or p < 0.01, Figure 2G). DZF intervention could modulate carbohydrate metabolism, amino acid metabolism, lipid metabolism, and energy metabolism in db/db mice (p < 0.05 or p < 0.01, Figure 2H). Specifically, DZF intervention brought the predicted functional capabilities of db/db mouse gut microbiota in closer alignment with those of db/m mouse gut microbiota, including functions related to pyruvate metabolism, C5-Branched dibasic acid metabolism, tryptophan metabolism, and fatty acid biosynthesis (Figures 2G, H).

To gain a deeper understanding of the characteristics of the gut microbiota, LEfSe analysis was employed to identify statistically significant biomarkers differentiating between groups. The LDA bar chart (Figure 3A) presented the key biomarkers for each group with LDA scores exceeding 4, while the accompanying cladogram (Figure 3B) illustrated the taxonomic hierarchy of these biomarkers. These biomarkers were primarily summarized from the phylum down to the family level. In the gut microbiota of db/m mice, biomarkers identified were Firmicutes phylum–clostridia class–clostridiales order–Lachnospiraceae family, and Epsilonproteobacteria class - Campylobacterales order–Helicobacteraceae family. For the gut microbiota of db/db + Veh group, the biomarkers identified were Gammaproteobacteria class–enterobacteriales order–enterobacteriaceae family, and Bacteroidales_S24_7_group family. In db/db + Met group, the identified biomarkers were Rikenellaceae family - Alistipes genus, as well as Ruminococcaceae family. In db/db + DZF group, the biomarkers included Bacteroidetes phylum–Bacteroidia class–Bacteroidales order-Bacteroidaceae family–Bacteroides genus, and at the same taxonomic level, Prevotellaceae family–Alloprevotella genus.

Based on the species annotation results, a bar chart was generated by selecting the top 10 species at the chosen phylum level in each group using a maximum value sorting method (Figure 3C). Bacteroidetes was the most abundant phylum. At the phylum level, the primary comparison focused on Bacteroidetes, Firmicutes, and Proteobacteria. There was no significant difference at the phylum level between db/db + Veh mice and db/m mice (Figures 3D–G). However, interestingly, DZF significantly increased the abundance of Bacteroidetes (p < 0.05, Figure 3D) and decreased Firmicutes (p < 0.05, Figure 3E) in db/db mice, resulting in a significant reduction in the Bacteroidetes/Firmicutes ratio (p < 0.01 vs. db/m, p < 0.05 vs. db/db + Met, Figure 3G). No significant changes were observed in the Met group compared to other groups.

At the family level, among the top 10 high-abundance taxa, Bacteroidaceae had the highest abundance (Figure 3H). Compared to db/m, both db/db + Veh and db/db + Met groups had a significant decrease in Bacteroidaceae (p < 0.01), while DZF intervention increased the abundance of Bacteroidaceae in db/db mice, significantly (p < 0.01) (Figure 3I). Relative to db/m, db/db + Veh group showed a significant increase in Bacteroidales_S24-7_group (p < 0.01), which slightly decreased after DZF and Met interventions but without statistical significance (Figure 3J). Interestingly, Met and DZF treatments resulted in a significant increase in Prevotellaceae compared to db/m (p < 0.01), while no significant difference was observed between db/db + Veh mice and db/m mice (Figure 3K). Rikenellaceae, along with the aforementioned three family-level groups, belongs to the Bacteroidetes phylum and significantly increased in db/db + Veh and db/db + Met groups (p < 0.01 vs. db/m, Figure 3L). Lachnospiraceae, Lactobacillaceae, and Ruminococcaceae belong to the Firmicutes phylum. Following DZF intervention, there was a significant reduction in the abundance of Lachnospiraceae compared to db/m + Veh group (p < 0.05, Figure 3M). In db/db + Met group, there was a tendency to increase the abundance of Ruminococcaceae compared to db/db + DZF (p < 0.01, Figure 3N). No significant intergroup differences were observed in the other microbial families (Figure 3O). Proteobacteria at the family level had Helicobacteraceae and Enterobacteriaceae as higher abundance groups. Compared to db/m, Helicobacteraceae significantly decreased in db/db + Veh (p < 0.01), while Met treatment resulted in a significant increase in Helicobacteraceae (p < 0.05, Figure 3P). Regarding Enterobacteriaceae, a significant increase was observed in db/db mice compared to db/m (p < 0.01), and DZF intervention led to a reduction in Enterobacteriaceae (p < 0.01, Figure 3Q).

At the genus level, Bacteroides was the most abundant genus (Figure 3R). Compared to db/m, both db/db + Veh and db/db + Met groups had a significant reduction in Bacteroides (p < 0.01), while DZF significantly increased its abundance relative to the Veh and Met groups (p < 0.01) (Figure 3S). The abundance of Alistipes was significantly increased in all three db/db mouse groups compared to db/m (p < 0.05 or p < 0.01, Figure 3T). Additionally, DZF treatment significantly increased the abundance of Alloprevotella (p < 0.05), while other group abundances also increased but without statistical significance (Figure 3U). Lachnospiraceae_NK4A136_group, Lactobacillus, and Ruminococcaceae at the genus level showed no significant intergroup differences (Figures 3V, W). The abundance of Helicobacter was significantly decreased in db/db mice compared to db/m (p < 0.01), but increased after Met treatment (p < 0.01) (Figure 3X).

LPS, a unique component of the outer membrane of Gram-negative bacteria, also known as endotoxin, can be absorbed from the intestine into the bloodstream. In this study, it was observed that the serum LPS level in db/db + Veh mice was significantly higher than in db/m mice (p < 0.05). However, after DZF intervention, there was a significant reduction in LPS levels compared to the Veh group (p < 0.05). Met group showed a decreasing trend in LPS levels but without statistical significance (Figure 4A).

Histological examination of small intestine mucosal tissue using H&E staining revealed that in db/m group, the intestinal layers were well-defined, villi were neatly arranged, the intestinal wall was intact, and there was no visible damage. In contrast, in db/db + Veh group, the intestinal villi were shortened and deformed, with a loose and disordered arrangement. The intestinal wall showed signs of damage, and there was a reduction in the number of glands. Met and DZF interventions both led to a partial restoration of the structural integrity of the small intestine. Villi length increased, gaps narrowed, intestinal wall damage reduced, and the number of glands increased (Figure 4B).

Zonula Occludens-1 (ZO-1) and Occludin, known as tight junction proteins, are critical components of the intestinal mechanical barrier. In db/db + Veh group, the expression of ZO-1 and Occludin protein was significantly lower than in db/m group (p < 0.01). The expression of Tjp1 and Ocln mRNA of db/db + Veh group was also significantly lower than db/m group (p < 0.01). Compared to db/db + Veh group, Met significantly upregulated the expression of ZO-1 and Occludin proteins (p < 0.05 or p < 0.01), with no significant difference in Tjp1 and Ocln mRNA levels. Compared to db/db + Veh group, DZF intervention significantly upregulated Occludin protein and Ocln mRNA expression (p < 0.05), while it had no significant impact on ZO-1 expression (Figures 4C–G).

Cluster of differentiation 36 (CD36) is a key protein in intestinal epithelial cells responsible for fatty acid absorption, while apolipoprotein B48 (ApoB48) is a crucial protein for assembling chylomicrons to transport lipids in intestinal cells. In db/db + Veh group, the expression of CD36 and ApoB48 protein was significantly higer than in db/m group (p < 0.01). The expression of Cd36 and Apob mRNA of db/db + Veh group was also significantly higer than db/m group (p < 0.01). This result indicates abnormal elevation in intestinal lipid absorption and transport. Compared to db/db + Veh, DZF intervention significantly downregulated the expression of CD36 protein, ApoB48 protein, Cd36 mRNA and Apob mRNA (p < 0.05 or p < 0.01). Compared to db/db + Veh, Met intervention significantly downregulated ApoB48 protein and Apob mRNA expression (p < 0.05 or p < 0.01) and reduced Cd36 mRNA expression (p < 0.05), with no statistically significant difference in CD36 protein expression (Figures 4C, H–K).

Hepatic tissue morphology was observed through H&E staining. In db/m group, liver tissue displayed intact structure, orderly cell arrangement, well-defined hepatic cords, centrally located cell nuclei, and pink cytoplasm without signs of lipid vacuole degeneration or significant inflammatory cell infiltration. In db/db + Veh group, liver cells were swollen, arranged irregularly, and showed signs of lipid vacuole degeneration, with increased accumulation of lipid droplets, varying in size. Some nuclei were shifted towards the periphery of the cell, and the cytoplasm appeared pale or transparent. In db/db + Met group, there was a moderate reduction in lipid vacuole degeneration, with smaller vacuoles being predominant. The liver cells displayed less swelling and more orderly arrangement, with cytoplasm appearing light pink. In db/db + DZF group, the improvement in lipid deposition was more pronounced, with fewer liver cells exhibiting lipid vacuole degeneration. Some liver cells were arranged in a more regular pattern, with nuclei centrally located (Figure 5A).

Further examination of lipid droplet distribution within liver cells was performed using Oil Red O staining. In db/m group, liver cells were neatly arranged, without noticeable orange-red lipid droplets inside the cells. The nuclei were blue and uniformly sized. In db/db + Veh group, a significant number of irregularly sized orange-red lipid droplets were observed, with liver cells arranged more chaotically. In db/db + Met and db/db + DZF groups, the volume and quantity of orange-red lipid droplets within liver cells were reduced, indicating a decrease in the degree of lipid deposition (Figure 5B).

Hepatic TG and TC levels in db/db + Veh group were significantly higher than those in db/m group (p < 0.01), indicating severe hepatic steatosis in db/db mice. Compared to db/db + Veh, both DZF and Met interventions led to a reduction in hepatic TC and TG to varying degrees (p < 0.05 or p < 0.01), with DZF showing a more significant decrease in TG (Figures 5C, D). Compared to db/m, serum ALT and AST levels were significantly elevated in db/db mice, indicating liver cell damage due to hepatic steatosis. However, after DZF and Met interventions, ALT and AST levels significantly decreased (p < 0.05 or p < 0.01), indicating that DZF and Met could protect liver function (Figures 5E, F).

Fatty acid synthase (FASN) and stearoyl-CoA desaturase 1 (SCD1) are key enzymes in hepatic fatty acid synthesis. In comparison to db/m mice, the expression of FASN protein, SCD1 protein, Fasn mRNA and Scd1 mRNA in the liver of db/db mice were significantly elevated (p < 0.01). Met intervention led to a decrease in protein expression of FASN and SCD1 (p < 0.05 or p < 0.01), with a corresponding decrease in Fasn and Scd1 mRNA expression, although without statistical significance. In contrast, DZF exhibited a more pronounced inhibitory effect on fatty acid synthesis. Compared to db/db + Veh group, db/db + DZF group showed a significant reduction in the expression of FASN protein, SCD1 protein, Fasn mRNA and Scd1 mRNA (p < 0.05 or p < 0.01). Furthermore, the FASN protein and Fasn mRNA expression in db/db + DZF group were significantly lower than that in db/db + Met group (p < 0.05) (Figures 6A, B, D, F, G).

Carnitine palmitoyltransferase-1𝛂 (CPT-1𝛂) is a major enzyme responsible for fatty acid oxidation in the liver. Both db/db and db/m mice showed similar and not significantly different CPT-1𝛂 protein and Cpt1a mRNA expression levels. Following DZF and Met interventions, CPT-1𝛂 protein and Cpt1a mRNA expression was significantly increased (p < 0.05 or p < 0.01), indicating enhanced fatty acid oxidation in liver cells (Figures 6A, C, H).

Fatty acid transport protein 2 (FATP2) mediates the uptake of fatty acids by liver cells. In db/db + Veh group, the expression of FATP2 protein and Slc27a2 mRNA was significantly increased (p < 0.05 or p < 0.01). After DZF intervention, a decrease in the expression of FATP2 protein and Slc27a2 mRNA was observed (p < 0.05 or p < 0.01), indicating that DZF can inhibit the excessive uptake of fatty acids by liver cells. Although there was a decreasing trend with Met intervention, no statistically significant differences were observed (Figures 6A, E, I).