Dietary α-Linolenic Acid-Rich Flaxseed Oil Exerts Beneficial Effects on Polycystic Ovary Syndrome Through Sex Steroid Hormones—Microbiota—Inflammation Axis in Rats

Six-week-old specific-pathogen-free female Sprague–Dawley (SD) rats (body weight, 193 ± 10 g) were purchased from the Laboratory Animal Center of Ningxia Medical University. All animal experiments were approved by the Ethics Committee of Ningxia Medical University (No. 2016-017). The animals were acclimatized for 1 week in polycarbonate cages at a temperature-controlled room (22 ± 2°C, air humidity 40–70%) under a 12-h light and dark cycle.

All animals were fed with a commercial diet of Keaoxieli Feed Co., Ltd., Beijing, China (46.65% crude protein, 20.73% moisture, 0.09% crude fat, 0.13% crude ash, 0.07% crude fiber, microelement calcium, and phosphorus). FO was purchased from Liupanzhen Square Ecological Agricultural Science and Technology Co., Ltd., Ningxia, China, the fatty acid composition of which was identified by gas chromatography–mass spectrometry (GC-MS) (). Letrozole was obtained from Hengrui Pharmaceutical Co., Ltd., Jiangsu, China. Table S1

The schematic time diagram of the experimental design is shown in Figure 1. In brief, after a 1-week period of acclimation with the control diet, 32 female SD rats (6 weeks old) were randomly assigned into four groups (eight rats/group): (a) pair-fed (PF) control (CON) group (PF/CON), (b) FO-fed CON group (FO/CON), (c) PF with letrozole-induced PCOS model (MOD) group (PF/MOD), and (d) PF with FO-fed MOD group (FO/MOD). The MOD groups received letrozole daily for 21 consecutive days at a concentration of 1 mg/kg by gavage, dissolved in 1% aqueous solution of carboxmethlycellulose (CMC) solution, whereas the CON groups were fed daily with 1% CMC, at a concentration of 1 ml/kg, for 21 consecutive days. Vaginal smears were collected daily at 9:00 am and measured with Wright–Giemsa staining for all of the rats. The stage of cyclicity was determined by a microscopic analysis of the predominant cell type. After 21 days of intervention, the rats were under a continuous diestrus phase predominantly exhibiting leukocytes, suggesting that the PCOS model was successful. After modeling, the FO groups received 1 ml/kg of FO daily for 8 weeks by gavage; the PF groups were administrated with 1% CMC (1 ml/kg) daily. The establishment of the PCOS rat model was comparable to that of a previous description (41). Body weight (BW) was measured weekly during 3 weeks of modeling and the subsequent 8 weeks of therapy. The estrous cycle was estimated by collecting vaginal smears daily. At the end of the experiment, all rats were euthanized and associated indications were investigated. Fecal samples and vaginal secretions were snap-frozen and stored at −80°C. The plasma samples were isolated by centrifugation and stored at −80°C.

Vaginal smears were collected and measured with Wright–Giemsa staining for all of the rats daily at 9:00 am. The stage of cyclicity was determined by a microscopic analysis of the predominant cell type. The four phases of the estrus cycle are comprised of different predominant cells: proestrus, predominantly consisting of nucleated epithelial cells; estrus, predominantly consisting of cornified epithelial cells; metestrus, equal mix of leukocytes, nucleated epithelial cells, and cornified epithelial cells; and diestrus, predominantly consisting of leukocytes. The observation period lasted for 77 consecutive days beginning on the first dose of letrozole until the last day of treatment.

After sacrificing the rats, the ovarian tissue samples were immediately fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 4 μm. To evaluate ovary damage, every 10th section (n = 8) was mounted on a glass slide, stained with hematoxylin and eosin (H&E), and analyzed using an Olympus light microscope (Melville, NY, USA) by two persons blinded to the origin of the sections. The numbers of cystic follicles and corpora lutea were counted. According to ovarian H&E staining, the number of cystic follicles and corpus luteum as well as the area of three ovarian tissues was detected in diverse groups. Based on the average value of ovarian tissue area in the negative control group, the number of cystic follicles and corpus luteum in the other three groups was corrected. According to the criteria proposed previously, cystic follicles were defined as those follicles devoid of oocytes, displaying a large antrum cavity, an enlarged thecal cell layer, and a thin granulosa cell compartment containing apparently healthy cells (42).

In order to evaluate the level of sex steroid hormones in rats, plasma was obtained to determine the respective follicle-stimulating hormone (FSH), luteinizing hormone (LH), E2, T, progesterone (PROG), as well as sex hormone binding globulin (SHBG), using enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturers' instructions (Shanghai Jianglai Biotech, Shanghai, China). The sensitivities of the assays were 0.1 IU/L, 0.1 mIU/ml, 0.1 pmol/L, 1.0 pg/ml, 0.1 nmol/L, and 0.1 ng/ml for FSH, LH, E2, T, PROG, and SHBG, respectively. For each hormone, the intra- and inter-assay coefficients of variation were <9 and <11%. Each sample was tested in triplicate.

The biochemical indications of lipid metabolism, including plasma total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C), were respectively determined using AU400 automatic biochemical analyzer (Olympus, Japan).

Insulin resistance was applied to the homeostasis model assessment of insulin resistance (HOMA-IR), which was calculated as [fasting plasma glucose (FPG) (mmol/L) × fasting insulin concentrations (FINS) (mIU/L)]/22.5. After fasting for 8 h, the FPG from one drop of tail blood was measured by using a standard glucometer (One Touch Profile, Johnson & Johnson, Inc. Milpitas, CA, USA). The plasma FINS was measured with a commercial ELISA kit (Fankewei Biology, Shanghai, China).

Plasma and ovarian inflammatory cytokines including IL-1β, IL-6, IL-10, IL-17A, and TNF-α were measured by using ELISA kits according to the manufacturer's instructions (Shanghai Jianglai Biotech, Shanghai, China). The monocyte chemoattractant protein-1 (MCP-1) was measured by using ELISA kits (Fankewei Biology, Shanghai, China). The sensitivities of the assays were 0.1, 1.0, 0.1, 0.1, 1.0, and 0.1 pg/ml for IL-1β, IL-6, IL-10, IL-17A, TNF-α, and MCP-1, respectively. Each sample was tested in triplicate. The plasma LPS levels in each group were examined using a limulus amebocyte lysate kit (Xiamen Bioendo Technology Co., Ltd, Xiamen, China) according to the manufacturer's instruction. Briefly, 50 μl of diluted plasma (1:4 dilutions with endotoxin-free water) was dispensed to each well in a 96-well plate. At the initial time point, 50 μl of the limulus amebocyte lysate reagent was added, respectively. The plate was incubated at 37°C for 30 min. Then, 100 μl of the chromogenic substrate warmed to 37°C was added to each well, and incubation was extended for an additional 6 min at 37°C. The reaction was stopped by adding 100 μl of 25% solution of glacial acetic acid. Optical density at 545 nm was measured with a microplate reader (Thermo Scientific, USA).

After 8 weeks of treatment, five rats from each group were randomly selected and placed in sterilized cages. Fresh feces and vaginal secretions of diverse rats were, respectively, collected and immediately stored at −80°C for subsequent DNA extraction. The rats were put into a clean cage with aseptic filter paper, the end of the tail and rectum of which was squeezed gently. Then, feces were collected with sterile tweezers in a sterile Eppendorf tube after defecation and transferred to −80°C for preservation immediately. Moreover, the vaginal orifice of rat was disinfected with alcohol cotton balls. After drying, we took a new aseptic cotton swab and inserted it into the vagina and rotated it gently. Then, the cotton swab with vaginal secretions was taken out and separated by aseptic scissors as well as stored at −80°C immediately for subsequent DNA extraction. The protocols for sample processing, 16S ribosomal DNA gene amplification, sequencing, and statistical analysis of microbiota data were performed as described previously (43).

Extraction of bacterial DNA by cetyltrimethylammonium ammonium bromide (CTAB) was performed by adding the appropriate amount of lysozyme and sample to 1,000 μl CTAB lysate. The mixture was placed in a 65°C water bath and mixed by inversion several times in order to facilitate the complete lysis of the sample. Next, phenol (pH 8.0), chloroform, and isoamyl alcohol were added to the supernatant after centrifugation so that the ratio of the three was 25:24:1, with mixing by inversion and centrifugation at 12,000 × g for 10 min. In the same way, chloroform and isoamyl alcohol (24:1) were added to the obtained supernatant, followed by centrifugation. The collected supernatant was added with isopropanol. The mixture was precipitated at −20°C after shaking up and down. Then, the mixture was centrifuged again according to the previous centrifugation conditions. The obtained precipitate was washed twice with 1 ml 75% ethanol. Then, the precipitate was blown dry on a clean bench or air-dried at room temperature. The DNA samples were dissolved in ddH₂O. If the sample was difficult to dissolve, it needed to be incubated at 55–60°C for 10 min. Finally, 1 μl of RNase A was added to the dissolved DNA sample, which was allowed to be kept at 37°C for 15 min to obtain bacterial DNA. The extracted DNA was stored at −20°C until application.

The DNA sequences involving the V3 and the V4 regions of the 16S rDNA hypervariable regions were amplified by Phusion® High-Fidelity PCR Master Mix with GC Buffer (New England Biolab, USA) using the following primers (5′ to 3′): 341F-CCTAYGGGRBGCASCAG, 806R-GGACTACNNGGGTATCTAAT. The PCR product was analyzed and separated on 2% agarose gel, which was purified using the GeneJE Gel Recovery Kit (Thermo Scientific, USA). The library was constructed using the TruSeq® DNA PCR-Free Sample Preparation Kit in order to carry out Qubit quantitation and library detection. After passing the test, the library was sequenced using the iIllumina HiSeq 2500 platform by Beijing Novogene Technology Co., Ltd., China.

Standard acetic acid, propionic acid, butyric acid, and valerate acid at a minimum purity of 98% were obtained from Sigma-Aldrich (St. Louis, MO, USA). Phosphoric acid and ether of analytical grade were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). SCFA (acetic acid, propionic acid, butyric acid, and valeric acid) measurements were carried out on a single quadrupole mass spectrometer (5975B-MSD) equipped with 6890N GC (Agilent Technologies, Santa Clara, CA, USA). The samples were mixed with a QL-866 vortex meter (Haimen, Jiangsu, China) and separated from H1850R refrigerated centrifuge (Xiang Yi, Changsha, Hunan, China). A mixed standard stock solution of the four SCFAs was prepared by dissolving an accurately weighed quantity of ether. Working solution series were prepared by appropriate dilutions of mixed standard stock solutions. A 10-point calibration curve was made by adding the working solutions and an equal volume of IS solution covering a range from 0.05 to 250 μg/ml (0.05, 0.1, 0.5, 1, 5, 10, 25, 50, 100, and 250 μg/ml). All these solutions were stored in a freezer at 0°C prior to use. Fecal samples weighing 100 mg were homogenized in 100 μl of 15% phosphoric acid with 100 μl of 250 μg/ml isohexanoic acid solution as IS and 400 μl ether (70 Hz for 1 min). Subsequently, the samples were centrifuged at 4°C for 10 min (12,000 × g), and the supernatants were transferred into a vial prior to the GC-MS analysis. The GC was fitted with a capillary column Agilent HP-INNOWAX (30 m × 0.25 mm i.d. × 0.25 μm) (Agilent Technologies, Santa Clara, CA, USA), and helium was used as the carrier gas at 1 ml/min. Injection was made in split mode at 10:1, with an injection volume of 1 μl and an injector temperature of 250°C. The temperature of the ion source, interface, and quadrupole were 230, 250, and 250°C, respectively. The column temperature was initially 90°C and then increased to 120°C at 10 °C/min, to 150°C at 5°C/min, and finally to 250°C at 25°C/min; and this temperature was kept for 2 min (total run-time of 15 min). The detector was operated in electron impact ionization mode (electron energy 70 eV) using full scan and single ion monitoring (SIM) mode.

Statistical analysis was performed using GraphPad Prism software 6.01 (GraphPad Software Inc., CA, USA) and SPSS 17.0 (IBM Corp., NY, USA). All experimental data were expressed as mean ± standard deviation of at least three independent experiments. Data were determined by one-way analysis of variance to compare the mean values of variables among the groups. Bonferroni's test or Tukey's post hoc test was used to identify the significance of pairwise comparison of mean values among the groups. Moreover, Spearman's correlation analysis was performed to identify the correlations between microbiota and inflammatory indicators. P < 0.05 was considered to be statistically significant.

Wright–Giemsa staining was used to estimate the difference of estrous cycles in diverse groups. Rats in the PF/CON and FO/CON groups exhibited regular estrous cycles of 4–5 days, comprising of proestrus, estrus, metestrus, and diestrus (Figures 2A–F). However, after 18 days of intervention, rats in the PF/MOD group were under a continuous diestrus phase, predominantly exhibiting leukocytes (Figure 2G). Intriguingly, after 20 days of dietary FO intervention, a cyclical variation from nucleated epithelial cells to cornified epithelial cells to a mix of leukocytes, nucleated epithelial cells, and cornified epithelial cells was microscopically observed (Figure 2H), suggesting that dietary FO can improve the disorder of estrous cycle in letrozole-induced PCOS.

Ovary weights in the PF/MOD group were increased with a fairly smooth-white appearance, compared with those in PF/CON group (P = 0.0103, Figures 3A,B). However, dietary FO improved the ovarian morphology and the ovary weights (P = 0.0153, Figure 3B). Moreover, H&E staining was used to determine the alteration of ovarian pathology in diverse groups (Figures 3C–J). Ovaries in the PF/CON and FO/CON groups possessed follicles and fresh corpora lutea (Figures 3C,D,G,H). The granulosa within follicles in the control group showed multiple layers. Compared to the PF/CON group, the number of cystic follicles in the PF/MOD group was increased (P = 0.0027, Figure 3L), but the corpus luteum decreased or disappeared (P = 0.0016, Figure 3K), indicating that the ovulatory function of the letrozole-induced PCOS model was impaired. Importantly, dietary FO administration increased the formation of corpora lutea (P = 0.0025, Figure 3L) as well as decreased the number of cystic follicles (P = 0.024, Figure 3K), demonstrating that dietary FO ameliorated the pathological damage of ovarian tissue and ovulatory function.

To assess the effects of FO on sex steroid hormones in PCOS, we measured six plasma sex hormones involving FSH, LH, E2, T, PROG, and SHBG using ELISA kit. The levels of LH showed no significant change in the diverse groups (P > 0.05, data not shown). However, the levels of LH/FSH and T in the PF/MOD group were significantly higher than those in the PF/CON group (P = 0.0038, P = 0.0054, Figures 4B,D) or the FO/MOD group (P = 0.017, P = 0.0063, Figures 4B,D). FSH, E2, and PROG were markedly reduced in the PF/MOD group compared to those of the PF/CON group (P = 0.0023, P = 0.009, P = 0.0024, Figures 4A,C,E) or the FO/MOD group (P = 0.0322, P = 0.004, P = 0.0013, Figures 4A,C,E). In addition, the levels of SHBG in the PF/MOD group were lower than those of the PF/CON group (P = 0.009, Figure 4F). The above results demonstrated that dietary FO treatment improved the sex steroid hormones in PCOS.

The body weight growth curve of rats in diverse groups were shown in Figure 4G. There was no significant difference in the initial BWs among the four groups. After 11 weeks of intervention, the BWs in the PF/MOD group were dramatically increased compared to those in the PF/CON group (P = 0.0013, Figure 4H). Interestingly, after FO treatment, the final BWs in the FO/MOD group was significantly decreased compared to those in the PF/MOD group (P = 0.027, Figure 4H), demonstrating that dietary FO partially reduced the BWs in the PCOS model. To assess the effects of FO on lipid metabolism in PCOS, the plasma levels of TC, TG, HDL-C, and LDL-C were investigated, respectively. As shown in Figures 4I–K, TC, TG, and LDL-C were higher, while HDL-C was lower in the PF/MOD group than those in the PF/CON group (P = 0.004, P = 0.0052, P = 0.0022, P = 0.0028, Figures 4I–L) or the FO/MOD group (P = 0.0153, P = 0.0274, P = 0.007, P = 0.002, Figures 4I–L). These four lipid indicators showed no significant difference between the PF/CON group and the FO/CON group.

To further understand the characteristic insulin resistance in PCOS after dietary FO treatment, we assessed the levels of FPG, FINS, and HOMA-IR in diverse groups. FPG in the PF/MOD group was increased in comparison with that in the PF/CON group (P = 0.014, Figure 4M). The FO treatment reduced the FPG levels but without a significant difference (Figure 4M). Intriguingly, FINS and HOMA-IR in the PF/MOD group were significantly higher than those in the PF/CON group (P = 0.016, P = 0.0018, Figures 4N,O) or the FO/MOD group (P = 0.0383, P = 0.0055, Figures 4N,O).

After dietary FO administration, we found elevated plasma levels of IL-1β (P = 0.002), IL-6 (P = 0.0197), IL-17A (P = 0.0218), TNF-α (P = 0.0015), and MCP-1 (P = 0.0165) in the PF/MOD group in comparison with those of the PF/CON group (Figures 5A,B,D–F). Moreover, dietary FO reduced the plasma levels of IL-1β (P = 0.0061), IL-6 (P = 0.0231), IL-17A (P = 0.031), TNF-α (P = 0.0013), and MCP-1 (P = 0.0325), respectively, compared to those of the FO/MOD group (Figures 5A,B,D–F). However, plasma IL-10 in PF/MOD was lower than that in the PF/CON group (P = 0.0025, Figure 5C) or the FO/MOD group (P = 0.0315, Figure 5C). Similar results were observed in the ovarian tissues in the diverse groups (PF/CON vs. PF/MOD: IL-1β, P = 0.0121; IL-6, P = 0.0112; IL-10, P = 0.0165; IL-17A, P = 0.0055; TNF-α, P = 0.0074; MCP-1, P = 0.012. PF/MOD vs FO/MOD: IL-1β, P = 0.0135; IL-6, P = 0.0381; IL-10, P = 0.0295; IL-17A, P = 0.0104; TNF-α, P = 0.0072; MCP-1, P = 0.0088, Figures 5A–F). Collectively, dietary FO reduced plasma and ovarian inflammation in PCOS by suppressing pro-inflammatory cytokines and enhancing anti-inflammatory IL-10.

To further assess the effects of FO on gut dysbiosis, permeability, and integrality of gut barrier, plasma-translocated LPS derived from Gram-negative bacteria was detected by the limulus reagent. The plasma LPS concentration in the PF/MOD group was significantly higher than that in the PF/CON group (P = 0.0089, Figure 5G). However, the LPS level in the FO/MOD group was decreased compared to that in the PF/MOD group (P = 0.0286, Figure 5G), suggesting that supplementary FO may inhibit endotoximia in PCOS.

Emerging evidences have demonstrated that the gut microbiota is an essential element in the occurrence and the development of PCOS (23, 44). In order to investigate the effect of dietary FO on the composition and the abundance levels of intestinal microbiota in PCOS, fecal samples were detected by 16S rRNA sequencing. Initially, as a result of α-diversity analysis, the abundance and the diversity of the bacterial community were assessed by observed species index and rarefaction curve. It was observed that the rarefaction curve tended to be flat when the sequence number increased to 20,000, indicating that the amount of sequencing data was reasonable (Figure S1B). The results, including both observed species index and rarefaction curve, showed no difference by Tukey test or Wilcoxon rank sum test (P > 0.05, Figures S1A,B).

Next, we assessed the β diversity of gut microbiota in diverse groups using unweighted principal coordinate analysis (PCoA) and weighted distance matrices (nonmetric multidimensional scaling, NMDS) (Figure 6 and Figure S2). The PCoA analysis showed that the general composition of the gut microbiota in the PF/MOD group was obviously different from that in the PF/CON or FO/MOD groups, respectively (Figure 6). Similar results were obtained from the NMDS analysis (Figure S2).

Finally, we investigated the difference in gut microorganism at the phylum and the genus levels in diverse groups. At the phylum level, ~90% of the gut bacteria in the diverse groups belong to Firmicutes and Bacteroidetes, another 8% comprised Proteobacteria and Actinobacteria, and the rest accounted for a low abundance (Figure 7A). The predominant Firmicutes and Bacteroidetes showed no significant change among the diverse groups. The ratio of Firmicutes to Bacteroidetes (F/B) in the PF/MOD group was higher than that in the PF/CON group (P = 0.0389, Figure 7C), but dietary FO decreased the F/B ratio in PCOS (P = 0.0453, Figure 6C). Similarly, the proportion of Actinobacteria in the PF/MOD group was obviously increased in comparison with that in the PF/CON group (P = 0.0363, Figure 7D). Dietary FO reduced the abnormal abundance of Actinobacteria in PCOS (P = 0.0182, Figure 7D). Collectively, the above results demonstrated that dietary FO supplementation had major effects on the ratio of F/B and Actinobacteria.

At the genus level, the relative abundance of Lactobacillus and Allobaculum in the FO/MOD group was higher than that in the PF/MOD group (P = 0.003, P = 0.0027, Figures 7E,G), whereas Lactobacillus and Allobaculum in the PF/CON group and the PF/MOD group showed no difference (P> 0.05, Figures 7E,G). The relative abundance of Bacteroides in the PF/MOD group was elevated compared to that in the PF/CON group (P = 0.0307, Figure 7F), whereas dietary FO reduced the levels of Bacteroides in the disease (P = 0.0192, Figure 7F). Moreover, the proportions of Butyrivibrio, Desulfovibrio, Ruminiclostridium, and Bifidobacterium in the PF/MOD group were obviously decreased in comparison with those in the PF/CON group (P = 0.0186, P = 0.0063, P = 0.0051, P = 0.038, Figures 7H–K). Intriguingly, dietary FO administration increased the relative abundance of Butyrivibrio, Desulfovibrio, and Bifidobacterium, compared with that in the FO/MOD group (P = 0.0195, P = 0.0026, P = 0.032, Figures 7H,I,K). However, dietary FO reduced the relative abundance of Ruminiclostridium in comparison with that in the PF/CON group (P = 0.0182, Figure 7J). Overall, dietary FO dramatically changed the abnormal proportions of genus components in PCOS by increasing Lactobacillus, Allobaculum, Butyrivibrio, Desulfovibrio, and Bifidobacterium as well as decreasing Bacteroides.

Growing evidences have demonstrated that VMB are intimately linked to female fertility (19, 45). In order to investigate the role of VMB dysbiosis in PCOS, vaginal secretions in the four groups were detected by 16S rRNA sequencing. Observed species index showed that the abundance and the diversity of VMB in the PF/MOD group were lower than those in the PF/CON group (P = 0.0123, Figure S3A) and the FO/MOD group (P = 0.0094, Figure S3A). Moreover, rarefaction curve indicated that the amount of sequencing data was reasonable (Figure S3B). PCoA showed that VMB in the PF/MOD group was obviously different from that in the PF/CON group in terms of species in the vaginal secretions samples (Figure 8A). There was no obvious change in PCoA between the PF/CON group and the FO/CON group as well as the FO/CON group and the FO/MOD group (Figures 8B,D). Interestingly, supplementary FO seemingly altered the vaginal species compared to PF/MOD (Figure 8C). Similar results from the NMDS analysis were obtained (Figure S4).

Next, we investigated the difference of VMB in the phylum and the genus levels in the diverse groups. At the phylum level, we found that Proteobacteria, Bacteroidetes, and Firmicutes were predominant in the diverse groups (Figure 9A). There was no notable difference in the relative abundances of Proteobacteria between the PF/CON group and the PF/MOD group (Figure 9C), whereas dietary FO decreased the proportion of Proteobacteria in the disease (P = 0.017, Figure 9C). Besides that, the level of Firmicutes in the PF/MOD group was lower than that in the PF/CON group (P = 0.0048, Figure 9D). Collectively, the above data indicated that dietary FO possessed major effects on Proteobacteria.

At the genus level (Figure 9B), the relative abundances of Lactobacillus, Faecalibacterium, and Parabacteroides were increased, while Streptococcus was decreased in the FO/MOD group compared to the PF/MOD group (P = 0.0101, P = 0.0198, P = 0.0188, P = 0.0196, Figures 9E–H). Moreover, Lactobacilus in the PF/MOD group was significantly decreased compared to that in the control group (P = 0.0234, Figure 9F). Additionally, Streptococcus abundance in the PF/MOD group was higher than that in the PF/CON group (P = 0.0339, Figure 9G). However, both Faecalibacterium and Parabacteroides in the PF/CON group and the PF/MOD group showed no difference (P > 0.05, Figures 9F,H). Taken together, dietary FO visibly changed the initial proportion of VMB in PCOS by increasing Lactobacillus, Faecalibacterium, and Parabacteroides as well as decreasing Streptococcus.

SCFAs belong to the crucial metabolites of the gut microbiota, mainly including acetic acid, propionic acid, butyric acid, and pentanoic acid. The TIC chromatogram showed that every SCFA can be distinguished clearly and exhibited a good peak shape, indicating that the method and data were reliable (Figure 10A). As a result of detection by liquid chromatography-mass spectrometry, acetic acid (P = 0.0009), propionic acid (P = 0.0011), butyric acid (P = 0.003) as well as pentanoic acid (P = 0.0160) in the PF/MOD group were significantly decreased compared with those in the PF/CON group (Figures 10B–E). Intriguingly, the decreased levels of SCFAs in PCOS were restored by dietary FO administration (acetic acid: P = 0.0145, propionic acid: P = 0.0027, butyric acid: P = 0.001, and pentanoic acid: P = 0.0124, Figures 10B–E).

For the assessment of relationships among hormones, gut microbiota/VMB, and inflammation in PCOS, we performed a correlation analysis (Figure 11). The abundance of anti-inflammatory bacteria including Lactobacillus, Firmicutes, Butyrivibrio, Desulfovibrio, and Bifidobacterium were positively correlated with SCFAs, FSH, E2, PROG, and IL-10, respectively. However, they were negatively associated with LPS, FSH/LH, T, and pro-inflammatory indicators. Reversely, the abundance of “bad bacteria” including Actinobacteria, Bacteroides, and Streptococcus as well as the F/B ratio showed a negative correlation with SCFAs, FSH, E2, PROG, and IL-10, respectively. Moreover, the abovementioned pro-inflammatory bacteria were positively correlated with LPS, FSH/LH, T, and pro-inflammatory indicators, respectively. Additionally, the rest of the different bacteria, including Allobaculum, Ruminiclostridium, Proteobacteria, Faecalibacterium, and Parabacteroides, were only connected with a few parameters, such as PROG. Taken together, there were close interactions and correlations among gut/vaginal bacteria, inflammation, and sex steroid hormones.

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: NCBI, accession number PRJNA624034↗.