Effects of Yulin Tong Bu formula on modulating gut microbiota and fecal metabolite interactions in mice with polycystic ovary syndrome

The herb details of YLTB are presented in Table 1. The clinical dose of YLTB was 255g (28), 45 g Dangshen (Codonopsis pilosula (Franch.) Nannf.), 30 g Huangqi (Astragalus mongholicus Bunge), 30 g Baizhu (Atractylodes macrocephala Koidz.), 10 g Shengma (Actaea cimicifuga L.), 10 g Chenpi (Citrus × aurantium L.), 15 g Fuling (Poria Cocos (Schw.) Wolf.), 10 g Banxia (Pinellia ternata (Thunb.) Makino), 30 g Cangzhu (Atractylodes Lancea (Thunb.) Dc.), 30 g Jixueteng (Spatholobus Suberectus Dunn), 10 g Xiangfu (Cyperus rotundus L.), 20 g Tusizi (Cuscuta chinensis Lam.), and 15 g Bajitian (Gynochthodes officinalis (F.C.How) Razafim. & B.Bremer), respectively. The Chinese herbs were purchased from Chongqing Hospital of Traditional Chinese Medicine. They were then suspended in distilled water at room temperature for one hour to soften the raw materials and facilitate the extraction of water-soluble ingredients in subsequent steps. The herbs were boiled for two hours and passed through filter paper. The process lasted two times followed by concentrating a decoction at 100°C, to obtain the extracts of YLTB containing 2.55-g raw herbs/mL. In this paper, the dosages of YLTB used in mice were converted from clinical dosage using the following equation (38, 39):

Where Dm is the administrated dose of YLTB for mouse in the present work, Dh is the clinical dose of YLTB, W is the weight of human body, F is the dose conversion factor of mouse and human. W is set as 60 kg. Considering the dose conversion factor of 9.1 between mouse and human, the high dosage is 38.68 g/kg, the medium dosage is 19.34 g/kg, and the low dosage is 9.67 g/kg.

ACQUITY UPLC HSS T3 (2.1 × 100 mm 1.8 m columns) was used for liquid chromatography–mass spectrometry (LC-MS) (Waters, UPLC; Thermo, Q Exactive). The following chromatographic separation conditions were used: column temperature, 40°C; flow rate, 0.30 mL/min; mobile phase A, water + 0.05 percent formic acid; mobile phase B, acetonitrile; injection volume, 2 μL; automatic injector temperature, 4°C. The MS parameters are as follow: ESI+: spray voltage, 3.0 (ESI+) or 3.2 (ESI-) kV; S-Lens RF level, 30% (ESI+) or 60% (ESI-); heater temperature, 300 °C; sheath gas flow rate, 45 arb; auxiliary gas flow rate, 15 arb; sweep gas flow rate, 1arb; capillary temperature, 350°C. Scan duration is 100 ms, interscan time is 50 ms, and the scan range is 70-1050 m/z. Data was analyzed using Compound Discoverer 3.1 software (Thermo Fisher Scientific,USA), normalized, and converted into a two-dimensional matrix using Excel 2010 software, containing retention time, compound molecular weight, observations (samples), and peak intensity (39, 40).

Female prepuberal C57BL/6 (21-d-old) mice (specific pathogen-free, Animal Qualification Certificate No.2022109) were purchased from the Animal Management Center of Chongqing Medical University. The mice were maintained under standard circumstances (22-24 °C, 50% relative humidity, 12-hour dark/light cycle) with unrestricted access to food and water. The Institutional Animal Care and Use Committee of Chongqing Medical University authorized all mouse experimental procedures (Chongqing, China). Mice were given four days to adjust to their new surroundings before being tested. A pseudo sterile mouse model was established during this four-day acclimatization phase by giving the animals an antibiotic cocktail (ABX) to remove the gut microbiota. Freshly prepared antibiotic cocktail comprised 1 mg mL⁻¹ ampicillin sodium (1146GR005, Biofroxx, Einhausen, Germany), 1 mg mL⁻¹ neomycin sulfate (N814740-5G, MACKUN, Shanghai, China), 1 mg mL⁻¹ metronidazole (M1547-5G, Sigma−Aldrich, MO, USA), and 0.5 mg mL⁻¹ vancomycin hydrochloride (#1161GR1001, Biofroxx, Einhausen, Germany) in drinking water, and the mice were given autoclaved drinking water four days later.

According to the preliminary experiment (Figure S1), 30 mice were randomly divided into six groups (n = 5) to optimize the dosage of YLTB: control group, PCOS group, high dosage YLTB group [YLTB (H), 38.68 g·kg^-1·day^-1], medium dosage YLTB group [YLTB (M), 19.34 g·kg^-1·day^-1], low dosage YLTB group [YLTB (L), 9.67 g kg^-1·day^-1], and metformin group (positive control, 250 mg kg^-1·day^-1, #D150959-5G, Sigma−Aldrich, MO, USA). To establish the PCOS group, mice were fed a high-fat diet (HFD, 60% kcal% fat; #D12492, Research Diets, Inc., NB, USA) and received daily subcutaneous injections of DHEA (#252805-10GM, Millipore, CA, USA; 60 mg/kg, dissolved in 0.1 ml of sesame oil) (41, 42). To determine the most effective YLTB dosage for PCOS, mice were fed an HFD, gavaged with various YLTB dosages, and injected with DHEA on the same day. The control group was fed a normal chow diet (NCD) and injected daily with sesame oil (#8008-74-0, Acros, BE, USA).

In order to verify the effects of high dose YLTB group on glucolipid metabolism, intestinal microbiota and fecal metabolism of PCOS mice, 30 mice were randomly divided into three groups (n = 10), namely, the control group, PCOS group and YLTB group (38.68 g·kg^-1·day^-1). To test the efficacy of the metabolite ferulic acid, 40 mice were randomly divided into four groups (n = 10), namely, the control group, PCOS group, FA1 group (50 mg·kg^-1·day^-1) and FA2 group (100 mg·kg^-1·day^-1) (33, 43). For FA groups, mice were fed an HFD and subcutaneously injected with DHEA daily, as well as administered ferulic acid (purity > 99.00%, #128708, Sigma−Aldrich, MO, USA, dissolved in lukewarm water) by oral gavage every day.

The mice mentioned above were all treated for 20 days. At the end of the experiment, mice were fasted over 12 hours and anesthetized by intraperitoneal injection of 10% chloral hydrate with 0.1 mL/10 g, and then weighed, measured to determine the distance from the tip of the nose to the anus, and tested for fasting blood glucose and oral glucose tolerance tests (OGTTs). The serum was collected and centrifuged for 20 minutes at 3,000 rpm to measure insulin and sex hormones. The ovaries and uteri of mice were collected, weighed, and wax-fixed or snap-frozen and stored at –80 °C for further use.

From the 11th to the 20th day of the gavage treatment, vaginal smears of mice were collected every day. The estrous cycle stage was determined by microscopic analysis of the predominant cell type in vaginal smears.

Enzyme-linked immunosorbent assay (ELISA) kits were used to determine serum testosterone, luteinizing hormone (LH), follicle stimulating hormone (FSH) and serum insulin levels (#JL25196, #JL10432, #JL10329, #JL11459, Shanghai Jianglai Biology, Shanghai, China). A glucometer was used to evaluate blood glucose. The homeostatic model assessment of IR (HOMA-IR) index was calculated using the formula: [FBG (mmol/L)] × [FINS (lU/mL)]/22.5. After an overnight fast, the oral glucose tolerance tests (OGTTs) were administered via oral gavage of 2 g/kg glucose. The blood glucose level of the mice was tested using blood glucose test paper at 0, 15, 30, 60, 90, and 120 minutes. At least 5 mice were used for these experiments.

Every four days, the mice weight was recorded. The length from the tip of the nose to the anus of the mouse was measured after anesthesia, and the Lee’s index and BMI were calculated (Lee’s index = [Body mass (g) × 1000]^1/3/body length (cm), BMI=body weight (kg)/body length (m²)). Biochemical analysis kits were used to assess serum lipid profiles, including TC, TG, HDL-C, and LDL-C (#A111-1-1, #A110-1-1, #A112-1-1, #A113-1-1; Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

Mouse ovaries were harvested, fixed with 4% paraformaldehyde (#BL539A, Biosharp) and embedded in paraffin. Hematoxylin and eosin (H&E) kits were used to stain ovary sections to observe follicle changes. Sections of 5 μm were placed on glass slides, with 40 μm discarded between each section; six sections from each ovary were collected. These tests required at least 5 mice per group (44, 45).

The fecal pellets were collected in sterile cryopreservation tubes immediately after being discharged before sacrifice. Before analysis, the fecal samples were snap-frozen and stored at –80 °C. Fecal samples were sent to Novogene (Beijing, China) for 16S RNA sequencing. DNA was extracted from mouse fecal samples using a Magnetic Soil and Stool DNA Kit (#DP712, Tiangen). DNA concentration and purity were measured on 1% agarose gels. DNA was diluted to 1 ng/μL in sterile water according to the concentration.

The V3–V4 hypervariable region of the 16S rRNA gene was amplified by PCR at 98 °C for 1 minute, followed by 30 cycles of denaturation at 98 °C for 10 seconds, annealing at 50 °C for 30 seconds, elongation at 72 °C for 30 seconds, and final incubation at 72 °C for 5 minutes. The common primer pair (515F5′-CCTAYGGGRBGCASCAG-3’; 806R5′-GGACTACNNGGGTATCTAAT-3′) was used to amplify the bacterial 16S rRNA gene. All PCRs were carried out with 15 µL of Phusion^® High-Fidelity PCR Master Mix (#M0531S, New England Biolabs, MA, USA), 2 μM primers, and 10 ng template DNA. The amplification products were isolated using a 2% agarose gel and purified with a Universal DNA Purification Kit (#DP214, Tiangen, Beijing, China). The NEB Next^® Ultra DNA Library Prep Kit (#E7370L, Illumina, CA, USA) was used for library construction. The established library was detected by an Agilent Bioanalyzer 5400 system and quantified by Q-PCR. Finally, the library was sequenced on an Illumina NovaSeq platform (Illumina Novaseq6000, Illumina, CA, USA), and 250 bp paired-end reads were generated (46).

Individual mouse feces samples (100 mg) were crushed in liquid nitrogen before being resuspended in prechilled 80% methanol using a well vortex (n = 6). The samples were placed on ice for 5 minutes before being centrifuged at 15,000 g and 4 °C for 20 minutes. LC−MS grade water was used to dilute some of the supernatant to a final concentration of 53% methanol. The samples were then transferred to a new tube and centrifuged at 15,000 g and 4 °C for 20 minutes. The supernatant was then injected into the LC−MS/MS apparatus for analysis.

UHPLC−MS/MS analyses were carried out using a Vanquish UHPLC system (ThermoFisher) paired with an Orbitrap Q ExactiveTM HF mass spectrometer (Thermo Fisher) by Novogene. Using a 17-min linear gradient at a flow rate of 0.2 mL/min, samples were injected onto a Hypersil Gold column (100 × 2.1 mm, 1.9 μm). Eluent A (0.1% formic acid in water) and eluent B (methanol) served as the eluents for the positive polarity mode. Eluent A (5 mM ammonium acetate, pH 9.0) and eluent B (methanol) were the eluents for the negative polarity mode. The solvent gradient was set as follows: 2% B, 1.5 min; 2-85% B, 3 min; 100% B, 10 min; 100-2% B, 10.1 min; 2% B, 12 min. The Q ExactiveTM HF mass spectrometer was operated in positive/negative polarity mode, with a spray voltage of 3.5 kV, capillary temperature at 320°C, sheath gas flow rate at 35 arb and aux gas flow rate at 10 arb, S-lens RF level at 60 and aux gas heater temperature at 350°C.

Compound Discoverer 3.1 (CD3.1, ThermoFisher) was used to analyze the raw data files produced by UHPLC−MS/MS to perform peak alignment, peak selection, and quantification for each metabolite. The primary parameters were established: retention time tolerance, 0.2 minutes; actual mass tolerance, 5 ppm; signal intensity tolerance, 30%; signal/noise ratio, 3; and minimum intensity. Peak intensities were then normalized to reflect the total spectral intensity. Based on additive ions, molecular ion peaks, and fragment ions, the normalized data were utilized to predict the molecular formula. To obtain correct qualitative and relative quantitative results, peaks were matched with the MassList, mzCloud (https://www.mzcloud.org/↗), and mzVault databases (46).

Statistical analyses were carried out using the statistical tools R (R version R-3.4.3), Python (Python 2.7.6 version) and CentOS (CentOS release 6.6). The KEGG, HMDB, and LIPIDMaps databases were used to annotate these metabolites. MetaX was used to perform partial least squares discriminant analysis (PLS-DA) and principal components analysis (PCA). Univariate analysis (t test) was used to determine statistical significance (P value). VIP > 1 and P value < 0.05, as well as fold change ≥ 2 or FC ≤ 0.5, were deemed to indicate differential metabolites. The data for clustering heatmaps were standardized using z scores of differential metabolite intensity areas and plotted using the Pheatmap package in R language.

GraphPad Prism version 8.3 was used to analyze all data, which were then displayed as the means ± standard error of the mean (SEM). When P < 0.05, the results were considered statistically significant. Each variable was tested for differences among three or more groups using one-way or two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test to assess statistical significance. The cor.test function from the stats R package was used to perform a Spearman correlation among the levels of fecal metabolites, phenotype, and relative abundance of species. We only performed the correlation for those species (P < 0.05), phenotypes (P < 0.05) and metabolites (P < 0.05 VIP>1) that were statistically significant across groups.

LC-MS analysis data of YLTB was performed using feature extraction and preprocessing with Compound Discoverer 3.1 software, normalized, and edited into a two-dimensional matrix using Excel 2010 software, including retention time, compound molecular weight, observations (samples), peak intensity and library matching (based on mzCloud and mzVault database, the mzVault include OTCML, KEGG and Chemspider database). The aim of this method was to complete the target screening without standard products. Several ingredients in YLTB were identified (Figure 1A, Table 2 and 3). The following seven important compounds were distinguishable, as shown in Figure 1B: (I) Nobiletin, (II) Tangeritin, (III) Berberine, (IV) Quercetin, (V) Catechin, (VI) Chlorogenic acid, and (VII) Liquiritigenin.

Before the effect of YLTB on PCOS was investigated, the PCOS mice were generated by a HFD and DHEA subcutaneous injections (60 mg/kg body weight) for 20 days. The YLTB group mice were additionally gavaged with YLTB (38.68 g·kg^-1·day^-1) at the same time. The usage of YLTB was optimized in our preliminary data, as shown in the Supplement Figure S1. The control group was only daily injected with sesame oil. The mouse intervention process was illustrated in the Figure 2A. To address the effect of YLTB on mice ovary function, multiple parameters were evaluated, including ovarian morphology, mice weight, wet weight of uterus and ovaries, and sex hormone changes.

Ovarian H&E staining were performed to ensure a successful PCOS establishment and to determine YLTB’s effect on ovarian morphology as well as follicle counts (Figure 2B). Many cystic and atretic follicles in the ovaries of the PCOS group were absent in the ovaries of the YLTB group. The PCOS group also showed fewer antral follicles and corpus luteum, while YLTB administration increased them (Figure 2C). Figure 2D illustrated that the mice body weight was significantly decreased in the YLTB group, compared to the PCOS group. There was no significant difference in mice body weight between the control group and YLTB group. The uterus and ovary weights of PCOS mice were also significantly reduced after YLTB therapy (Figure 2E). In addition, YLTB dramatically downregulated serum T and LH levels in PCOS mice. Consistently, the LH/FSH ratio was significantly lower in the YLTB group in comparison to the PCOS group (Figures 2F, G). In summary, the application of YLTB largely restored these ovarian parameters of the PCOS group to the normal levels. Although it could affect sex hormone levels, YLTB did not impact the estrus cycle of the PCOS group (Figure S2 A, B).

To address whether YLTB ameliorates PCOS condition through alternation of glucose and lipid metabolism, the glucose levels, insulin levels, Lee’s index, BMI, lipid levels, and insulin sensitivity as well as the degree of obesity among the mice groups of the control, PCOS and YLTB were compared. Firstly, OGTTs were performed in those mice administered via oral gavage of 2 g/kg of glucose after 12 hours of fasting. The administered mice were sampled for the blood glucose levels using the test paper at 0, 15, 30, 60, 90, and 120 minutes. The glucose levels of the PCOS group were the highest, suggesting that the delayed glucose clearance in the PCOS group was removed in the YLTB group (Figure 3A). Consistently, YLTB significantly attenuated increases in the levels of fasting glucose, serum insulin and HOMA-IR in PCOS mice (Figures 3B–D). Meanwhile, the body size of the PCOS group was larger than the control group, assayed by Lee’s index and BMI, while the values of these two indexes were lowered down in the YLTB group (Figure 3E, F).

The lipid profiles of the three groups were characterized by determination of levels of total TC, TG, LDL-C and HDL-C. Importantly, the TC levels in the YLTB group was lower than the one in the PCOS group (Figure 3G). In contrast, there was no significant changes of TG levels between the PCOS group and the YLTB group (Figure 3H). Notably, although the HDL-C levels in the YLTB group was higher than the one in the PCOS group (Figure 3I), the levels in the PCOS group were similar to in the control group (Figure 3I). On the other hand, LDL-C levels were significantly increased in the PCOS group compared to the control group (Figure 3J). However, YLTB had no significant impact on the elevated LDL-C levels in PCOS mice. This indicates that the mechanism by which YLTB acts on TC to affect PCOS is complicated.

Given that the gut microbiota can regulate many metabolites, we next investigated whether the gut microbiota contributes to YLTB manipulation on sex hormone levels and metabolism indicators in mice with PCOS. Before that, the profile of the gut microbiota, simulated by the fecal microbiota, was described for the control, PCOS, and YLTB groups. The 16S rRNA sequencing analysis was performed with fecal samples separately collected from the groups. The rarefaction curves (Figure 4A) and rank abundance curves (Figure 4B) of the three groups growed in a similar manner, reflecting that the amount of sequencing data and the species richness were qualified for further analysis.

The α diversity indicator Simpson index scored closely for the three groups, showing no difference of species diversity across the three groups (Figure 4C). The β-diversity of the gut microbiota of the three groups were also determined using various approaches such as the UPGMA (unweighted pair group method with arithmetic mean) clustering method, principal coordinate analysis (PCoA), and nonmetric multidimensional scaling (NMDS). The weighted unifrac UPGMA clustering method classified the control and PCOS mice samples into two distinct groups (Figure 4D), implying that the gut microbial composition of the PCOS mice differed from that of the control mice. Although the YLTB group could not be completely separated from the PCOS group, it was significantly different from the PCOS group, demonstrating that YLTB administration alters the gut microbial composition of PCOS mice (Figure 4D). Based on the unweighted unifrac distance, PCoA analysis showed a clear separation among the three groups, with the principal component 1 (PC1) value of 43.61% and PC2 value of 11.03% (Figure 4E). In addition, the degree of variation among the three groups was illustrated using NMDS analysis based on Bray−Curtis distance, when stress is less than 0.2 (stress = 0.158, Figure 4F). Both PCoA and NMDS analyses revealed that the intergroup distances were greater than the intragroup distances, implying that PCOS and YLTB therapy have an effect on the β-diversity of gut microbiota.

In addition to diversity details, the abundance of gut microbiota of the control, PCOS and YLTB groups was analyzed, using the linear discriminant analysis (LDA) effect size (LEfSe) method. In our studies, the cut-off threshold of LDA was set at the score 4.0. Our data showed that the Bacteroidota phylum was enriched in the control mice, and the Firmicutes phylum was enriched in PCOS mice, as well as the Erysipelotrichales order in YLTB mice (Figures 5A, B). This observation demonstrated a great shift of gut microbiota composition from PCOS mice to YLTB mice.

The detailed differences of the taxonomic composition of gut microbiota among the three groups were shown in the Figure 5C–H at the phylum, genus, and species levels. At the phylum levels, Firmicutes and Bacteroidota were dominant in all three groups (Figure 5C). Furthermore, the relative Firmicutes abundance levels were highest in PCOS mice, but the levels of Bacteroidota and Proteobacteria were lower than the ones in other two groups. These changes were attenuated in YLTB mice (Figure 5D). For Bacteroidota, the Bacteroides and Parabacteroides genus abundance levels were lower in the PCOS mice than in the control group (Figure 5E). For Firmicutes, the Blautia and Lactobacillus abundance levels in PCOS mice were restored to the normal levels by YLTB administration. The similar effects were observed with the abundance levels of Parabacteroides (Figure 5E). Besides Blautia and Lactobacillus, Lachnospiraceae UCG-005 and UCG-006 were greatly enriched in the PCOS group, which were also significantly reduced by YLTB (Figure 5F). Bacteroides, Blautia and Lactobacillus genera were further analyzed for the detailed information at the species levels (Figures 5G, H). The data showed that Bacteroides dorei, Bacteroides fragilis, Lactobacillus johnsonii and Lachnospiraceae bacterium 28-4 as well as Lachnospiraceae bacterium 615 were enriched in the PCOS group and could be reverted in the YLTB group (Figure 5H). These results suggested that YLTB can restore PCOS gut microbiota disturbances.

Since we have shown that YLTB administration greatly changed basal metabolism, ovary dysfunction, and the gut microbiota in PCOS mice, we next seek for more association information among fecal metabolites, gut microbiota and PCOS mice phenotypes using biostatistical approaches. Firstly, the fecal metabolite profile from mice was determined by the untargeted metabolomics. Principal component analysis (PCA) and partial least squares discrimination analysis (PLS-DA) showed a distinct separation of those metabolites among the three groups, with different PC1 and PC2 values (Figure 6A, B). The detailed relationship between the metabolites and the three groups were illustrated in Figure 6C. 44 fecal metabolites demonstrated that the abundances of fecal metabolites in PCOS mice were different from those in the control mice, which were significantly manipulated by YLTB. These metabolites mainly consist of hormones, vitamins, and organic acids. For example, hormones include cortisone, hydrocortisone, menaquinone, and etc; vitamins such as ascorbic acid, folic acid and phylloquinone, whereas organic acids include cis-aconitic acid, maleic acid, ferulic acid, and etc (Figure 6C).

Spearman’s correlation analysis was next used to comprehensively analyze the correlations among fecal metabolites, the gut microbiota and host phenotypes (Figure 7). In the relationship between fecal metabolites and the gut microbiota, metabolites (such as ferulic acid, folic acid and methylmalonate) were negatively correlated with 9 bacterial species. Among those metabolites, ferulic acid (FA) was strongly correlated with basal metabolism. The correlations between the gut microbiota and host phenotypes were also shown in the Figure 7. Nine bacterial strains (Alistipes inops, Bacteroides dorei, Bacteroides fragilis, Clostridium sp Culture-54, Faecalitalea sp Marseille-P3755, Lachnospiraceae bacterium 615, Lactobacillus johnsonii, Lactococcus lactis, and Ruminococcaceae bacterium GD7) were positively correlated with the host parameters (including body weight, BMI, TC, T, LH, LH/FSH ratio, FBG, INS and HOMA-IR). Taken together, our findings showed a tight association among fecal metabolites, the gut microbiota and host phenotypes in PCOS mice administered with YLTB.

The Figure 7 showed that FA is a critical metabolite that contribute to the effect of YLTB to the PCOS mice. Therefore, FA was then tested whether it regulates ovarian functions in PCOS mice. The DHEA-treated mice were administered with low-dose FA (FA1 group, 50 mg/kg) or high-dose FA (FA2 group, 100 mg/kg) for 20 days, before they were sampled for further analysis of ovarian function changes (process illustrated as Figure 8A). Next, we tested ovarian morphology, body weight, wet weight of the uterus and ovaries and sex hormone alterations among the control, PCOS, FA1 and FA2 groups. H&E staining revealed that the FA administration reduced the number of cystic and atretic follicles, while promoting the development of antral follicles and corpus lutea in the PCOS ovaries (Figures 8B, C). The body weight, as well as wet uterine and ovary weight, was significantly reduced in both high and low FA-gavaged mice (Figures 8D, E). Consistently, the decreased serum T levels were observed in FA-gavaged mice compared to PCOS mice (Figure 8F). However, FA had no impact on the estrus period (Figure S2C-D). These findings showed that FA alleviated ovarian dysfunction and reduced pathological damage to ovarian tissues in PCOS mice.

FA was next tested for its regulatory effect on disrupted glucose and lipid metabolism in PCOS mice. The Figure 9A illustrated that FA administration largely corrected glucose tolerance impairment and dramatically reduced fasting glucose, serum insulin, and HOMA-IR levels in PCOS mice (Figure 9B–D). Besides, FA significantly lowered the values of BMI, Lee’s index, TC and LDL-C levels (Figure 9E, F, G, J). Consistently, the high FA dose significantly increased the HDL-C levels while decreased the TG levels (Figures 9H, I). Differently, the levels of TG and HDL-C in the low dose FA group were lower than the ones in the PCOS group but without statistical differences (Figures 9H, I). In general, these studies validated the effect of FA on PCOS similar to YLTB.

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: http://www.ncbi.nlm.nih.gov/bioproject/914098↗ (accession number PRJNA914098), http://www.ebi.ac.uk/metabolights/MTBLS6736↗ ( identifier MTBLS6736).