Effects of Bu Shen Hua Zhuo formula on the LPS/TLR4 pathway and gut microbiota in rats with letrozole-induced polycystic ovary syndrome

BSHZF consists of 15-g Tu Si Zi (Cuscutae Semen), 15-g Gou Qi Zi (Lycii Fructus), 8-g Qing Ban Xia (Pinelliae Rhizoma Praeparatum cum Alumine), 15-g Dang Gui (Angelicae Sinensis Radix), 15-g Fu Ling(Poria), 20-g Dan Shen (Salviae Miltiorrhizae Radix et Rhizoma), 10-g Xiang Fu (Cyperi Rhizoma), 10-g Chuan Xiong (Chuanxiong Rhizoma), 15-g Bai Shao (Paeoniae Radix Alba), 6-g Gan Cao (Glycyrrhizae Radix et Rhizoma), 6-g Qing Pi (Citri Reticulatae Pericarpium Viride), 20-g Yi Ren (Coicis Semen), 10-g Che Qian Zi (Plantaginis Semen), 10-g Jie Geng (Platycodonis Radix), 20-g Bai Jiang Cao (Patrinia Scabiosaefolia), and 30-g Dong Gua Pi (Benincasae Exocarpium). Processed pieces of Chinese herbs were purchased from Dongfang Hospital, Beijing University of Chinese Medicine. They were soaked in distilled water at room temperature for 1 h, followed by concentrating a decoction at 100℃, to obtain the extracts of BSHZF containing 5.35-g raw herbs/ml.

A total of 36 female SPF Sprague–Dawley rats with a body mass of (200 ± 10) g were purchased from Weitong Lihua Laboratory Animal Technology Co., Ltd. (Beijing, China; certificate no. SCXK-2016-0006). All rats were raised under a controlled condition (12-h light/12-h dark cycle, temperature 25℃, humidity 60%) in SPF-class housing in a laboratory with free access to rodent feed and sterile water.

Rats were randomly allocated into six groups: control group; model group; Chinese herb high-, medium-, and low-dose group; and metformin group (n=6). After 1 week of adaptive feeding, rats in the control group were administered letrozole (Jiangsu Hengrui Pharmaceutical Co., Ltd., batch no. 190316KG), which was dissolved in 1% carboxymethylcellulose (CMC), 1 mg/kg daily for 21 days by oral gavage to induce PCOS (36). The control group was administered an equal volume and concentration of CMC. From the day 22 on, rats in the control and model groups were treated with CMC and letrozole (1 mg/kg/day), respectively, without interruption. Simultaneously, rats in the Chinese herb high-dose (CHH), medium-dose (CHM), and low-dose (CHL) groups were treated with BSHZF via oral gavage at doses of 4.5, 9, and 18 g/kg per day, respectively (37) combined with letrozole (1 mg/kg/day). In addition, rats in the metformin (MET) group were treated with metformin 0.008 g/kg per day (Sino American Shanghai Shiguibao Pharmaceutical Co., Ltd, batch no. H20023370) and letrozole (1 mg/kg/day). The rats were then treated for 28 days.

All rats were fasted for 12 h after final administration. Rats were randomly allocated into six groups: control group; model group; Chinese herb high-, medium-, low-dose group; and metformin group (n=6). Rats were then weighed, anesthetized with 0.4% pentobarbital (1 ml/100 g) intraperitoneally, and disinfected with 75% ethanol, and their abdominal cavities were opened. Blood samples were collected from the abdominal aorta of each group. The right ovaries were fixed with a 4% polyformaldehyde solution. The left ovaries and feces of the rats were collected and frozen at -80 ℃. Finally, rats were euthanized via cervical dislocation.

Blood samples were centrifuged at 2,000 rpm for 15 min to obtain serum samples. Fasting blood glucose (FBG) levels were examined using a glucose oxidase assay kit (Beijing Bio-Top Institute of Biotechnology). Batch no. DL201901↗, sensitivity: 2.2–22.75 mmol/L, intercoefficient of variation ≤5%, intra-coefficient of variation ≤10%). Fasting insulin (FINS) and testosterone (T) levels were measured using radioimmunoassay kits (Beijing North Institute of Biotechnology; sensitivity: 5–405 μIU/ml and 0.1–20 ng/ml, respectively). All intercoefficients of variation were <10%, and all intracoefficients of variations were <15%. LPS levels were detected using a chromogenic end-point TAL kit (Xiamen Limulus Reagent Experimental Factory Co., LTD, batch no. XMH201903, sensitivity: 0.1-1 EU/ml), and IL-6, IL-18, and TNF-α levels were measured utilizing commercial enzyme-linked immunosorbent assay kits (NanJing JianCheng Bioengineering Institute, Nanjing, China, sensitivity: 2–600 ng/L, 2–600 ng/L, and 5–1,000 ng/L, respectively). All intercoefficients of variation were <10%, and all intracoefficients of variation were <12% for rats.

The right ovaries of the rats were fixed with a 4% paraformaldehyde solution and embedded in paraffin. The ovaries were sectioned into 5-μm slices and stained with hematoxylin and eosin. Histomorphologic examination was performed under a light microscope. The results were reviewed by pathologists.

The entire DNA in fecal samples was extracted using the E.Z.N.A. Soil DNA Kit (Omega Biotek, Norcross, GA, United States) and detected by 1% agarose gel electrophoresis. To analyze the taxonomic composition of the gut microbiota, we selected the V3–V4 hypervariable region of the 16S rRNA gene for subsequent sequencing and that was amplified using a PCR system (ABI GeneAmp 9700, Applied Biosystems, Carlsbad, CA, USA). Amplification was performed in triplicate using the barcoded universal bacterial primer 338F/806R (5′-barcode-ACTCCTACGGGAGGCAGCA-3′/5′-GGACTACHVGGGTWTCTAAT-3′). The products were mingled, confirmed by 2% agarose gel electrophoresis, and purified using the AxyPrep DNA Extraction Kit (AXYGEN, Tewksbury, MA, USA). The MiSeq library was constructed using a TruSeq TM DNA Sample Prep Kit (Illumina, San Diego, CA, USA). Purified products were sequenced on an Illumina MiSeq platform (Illumina, San Diego, CA, USA).

The resulting paired-end reads from MiSeq sequencing were spliced according to the overlap relationship using Flash (version 1.2.11), and the sequence quality was controlled and filtered using QIIME (version 1.9.1). According to the threshold of 97% similarity, operational taxonomic unit (OTU) clustering was conducted for non-repetitive sequences using Usearch (version 7.0), and chimerism was removed. OTU sequences were compared with the Silva (Release119) database. α-Diversity was estimated using the Shannon index in Mothur (version 1.30.2). The composition similarity and overlap at the OTU level are represented by a Venn diagram. The distribution of the compositions at different levels is represented by a community bar diagram. β-Diversity was estimated using principal coordinate analysis (PCoA), with Bray–Curtis dissimilarities calculated according to the levels of altered gut microbiota using R (version 2.15.3). The linear discriminant analysis effect size (LEfSe) was used to distinguish the characteristics of different abundances and associated categories. To predict the microbial function, we used the phylogenetic investigation of communities by the reconstruction of unobserved states (PICRUSt). COG and KEGG databases were used for comparison, and STAMP was used for the difference analysis. Student’s t-test was used to test for significant differences between the groups.

A lysate-containing protease inhibitor was added to the left ovarian tissues of rats, and the cells were lysed in an ice bath with maximum power ultrasound (3×10 s). Subsequently, the cocktail was centrifuged at 4℃, 12,000 rpm for 15 min, and then, the supernatant was collected, and the protein concentration was measured with a BCA assay kit (MD913053↗; MDL). According to the protein quantification results, an equivalent amount of protein was separated by SDS-PAGE and transferred to a PVDF membrane. The membrane was sealed with blocking buffer (MD912056↗, MDL) and incubated with anti-NF-κB p65 (phospho S536) (1:500; Ab76302, Abcam) and anti-TLR4 (1:500; Ab13556; Abcm) antibodies overnight at 4℃. After rinsing three times with TBST, secondary antibodies (1:4,000; goat anti-rabbit IgG; MD912577↗; MDL) were added at room temperature for 60 min. The bolt was then washed with TBST, and proteins were stained with an ECL reagent, processed, and analyzed using a chemiluminescence imaging system (170-8280; Bio-Rad, USA). ImageJ software was used to analyze and calculate the relative gray values of the strips in each experimental group compared with the control group (38). Actin and NF-κB p65 were quantified individually as internal controls. Three independent experiments were conducted.

SPSS (version 22.0) software was used for statistical analyses. The data are expressed as mean ± SEM. The Shapiro–Wilk test was used to verify normality. A two-tailed Student’s t-test was used to confirm statistical significance between the two groups. One-way analysis of variance (ANOVA) followed by least significant difference was used to confirm statistical significance among three or more groups. For non-parametric analysis, the Mann–Whitney U test was used to confirm the statistical significance between two groups, and the Kruskal–Wallis test was used among multiple groups, followed by Tamhane’s T2. P < 0.05 was considered statistically significant. Data were evaluated to exclude any outliers identified by Dixon’s Q test.

In total, two rats died within 10 min after improper oral gavage. One of them was in the model group, and the other was in the MET group. To investigate the impact of BSHZF on PCOS progression, rats were administered letrozole via oral gavage to induce PCOS. Compared to the control group, rats in the model group exhibited obesity, hyperandrogenism, and polycystic ovarian histomorphology, indicating that the PCOS model was successfully established in rats. Importantly, BSHZF significantly prevented weight gain in PCOS rats in a dose-dependent manner. Notably, the degree of weight loss observed in the CHM group was similar to that observed in the MET group, in which PCOS rats were administered metformin (Figure 1A, P<0.01 and P<0.05, respectively). In addition to showing alleviated obesity, rats treated with BSHZF showed significantly lower levels of FBG and FINS (Figures 1B, C). Similarly, rats in the PCOS group showed elevated serum T levels, which were prevented by BSHZF administration (Figure 1D). Dominant follicles and several corpora lutea were observed in the control group. Several cystic follicles were observed in the model group. We believe that morphological changes in the ovaries were reversed by BSHZF administration (Figure 2).

To explore the impact of BSHZF on intestinal microflora in PCOS-model rats, we compared the composition and function of the gut microbiota among each group via 16s rRNA microbial community analysis. After the removal of low-quality sequences, 1,778,160 raw reads were obtained. The rarefaction curve tended to be gentle as the number of reads sampled increased, which indicates that the sequencing amount of each group was close to saturation and covered most microbial species (Figure 3A). As shown in Figure 3B, there were 692, 659,788, 823, 693, and 739 OTUs in the control, model, CHL, CHM, CHH, and MET groups, respectively, with 373 OTUs coexisting in all groups. In addition, the α-diversity in the PCOS model group was significantly lower than that in the control group, as assessed by the Shannon index. However, the CHL and CHM groups showed significantly higher α-diversity than the model group (P<0.05 and P<0.001, respectively; Figure 3C). Notably, the α-diversity was comparable between the CHM and MET groups. As for β-diversity, the PCoA score plot at the OTU level revealed a distinct clustering of the gut microbiota composition in each group, and the model group was significantly separated from the other groups (Figure 3D). A similarity analysis demonstrated that the intergroup distances were significantly greater than the intragroup distances (R = 0.2262. P = 0.011), which suggests that there were significant differences in the gut microbiota composition at the OTU level between the groups. At the phylum, family, and genus levels, the CH and MET groups also significantly differed from the model group (Figures 3E–G).

To analyze the possible reasons for the microbial differences and investigate the influence of BSHZF on the bacterial community, we analyzed the composition of the gut microbiota at different levels among the groups (Figures 3H–J). Taxonomic analysis at the phylum level showed that the fecal microbiota among the groups was composed of five major phyla: Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Epsilonbacteraeota (Figure 3H). In addition, the composition of the microbiota at the family level showed apparent differences among the groups (Figure 3I). At the genus level, the fecal microbiota consisted of Lactobacillus, Lachnospiraceae_NK4A136_group, norank_f:Muribaculaceae, unclassified _f:Lachnospiraceae, Bacteroides, and Prevotellaceae_UCG-001 (Figure 3J). The relative abundances of these genera varied among the groups. To further determine the biomarkers presented as taxa, each group was statistically analyzed using the LEfSe method. A linear discriminant analysis was used to estimate the impact of species richness on the difference effect (Figure 3K). At the phylum level, the Firmicutes/Bacteroidetes ratio significantly differed between the groups (P=0.017). The Firmicutes mean proportion in the control, CHL, CHM, and MET groups significantly decreased (Figure 3L, P=0.010, P=0.032, P=0.000, P=0.007, respectively). However, the Bacteroidetes mean proportion did not significantly differ between the model group and other groups. At the genus level, we found that the relative abundance of Eubacterium_coprostanoligenes_group decreased in the CHH group, whereas that of Lactobacillus evidently increased (Figure 3M, P=0.004 and P=0.014, respectively). Furthermore, the relative abundances of Allobaculum, Bacteroides, Parasutterella, and Ruminococcaceae_UCG-014 in the CHM group were higher than those in the model group (Figure 3N, P=0.049, P=0.031, P=0.469, and P=0.007, respectively).

To predict the potential function of the microbial community among groups, PICRUSt analysis was performed, and the changes in functional pathways between groups were calculated using the STAMP software. Three KEGG pathways showed significant differences between the model and CHH groups (Figure 3O), which suggests that carbohydrate digestion and absorption might have increased in the CHH group. In addition, 32 KEGG pathways showed significant differences between the model and CHM groups (Figure 3P). These pathways are involved in carbohydrate and lipid metabolism, protein digestion, and absorption. In summary, these data indicate that BSHZF normalizes the composition of gut microbiota in PCOS rats.

Finally, we investigated the effect of BSHZF on the LPS/TLR4 signaling pathway. The level of serum LPS was significantly higher in the PCOS group than in the control group, which was reversed by BSHZF administration in a dose-dependent manner. (P<0.01 control versus model, P<0.05 CHH versus model, Figure 4A). Moreover, BSHZF administration significantly decreased TLR4 expression in the ovarian tissues (Figures 4B, C). Consistent with this finding, the level of phospho-p65, the dominant transcription factor downstream of LPS/TLR4 signaling, was increased in the ovarian tissues of PCOS rats. Importantly, BSHZF treatment remarkably reduced the level of phospho-p65 (Figures 4D, E), which suggests that BSHZF inhibited proinflammatory signaling in PCOS ovarian tissue. In line with these data, the serum levels of IL-18 and TNF-α in the CHH group were significantly decreased compared with those in the model group, as evidenced by ELISA (P<0.01 and P<0.05, respectively, Figures 4F, G). Nevertheless, IL-6 levels were comparable between the CH and model groups (Figure 4H). Taken together, BSHZF reduced proinflammatory responses in PCOS ovarian tissue by inhibiting LPS/TLR4 signaling.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/bioproject/PRJNA772159↗, BioProject accession number PRJNA772159