| Objective: Polycystic ovary syndrome(PCOS)is the main cause of infertility in reproductive women.The etiology of PCOS is complicated and is related to genetic,environmental,neuroendocrine and other factors.The pathogenesis is still unclear.The clinical manifestations are highly heterogeneous,involving irregular menstruation,infertility,hairy,acne,obesity and metabolic abnormalities.It is the risk factor of metabolic syndrome,cardiovascular disease,type II diabetes and endometrial cancer,affecting reproductive function and physical health of women.Currently,there are no specific diagnostic biomarkers and therapeutic targets of PCOS.Raman spectroscopy has been widely reported in several studies.It has the characteristics of rapid,sensitive and low-cost in detecting samples.In our study,we used Raman spectroscopy to detect both follicular fluid and plasma samples in combination with machine learning models to explore its value in PCOS.The molecular mechanism of follicular dysplasia in PCOS patients are still unclear.Ovarian granulosa cells are important for follicular growth.Follicular fluid can reflect changes in the follicular environment.In this study,quantitative proteomic analysis of follicular fluid from non-PCOS and PCOS women were performed and several differentially expressed proteins were identified.By establishing animal models and conducting in vitro experiments,the expression and regulation of CXCL12 in follicular fluid and granulosa cells of PCOS were studied,laying a foundation for the treatment of disease.Methods: 1.Plasma samples and follicular fluid samples were collected from 50 PCOS and 50 non-PCOS women,respectively.Samples were detected by Raman spectroscopy and the acquired fingerprint spectrum data were analyzed.Machine learning models were established and the prediction effects of models were evaluated.2.Follicular fluid samples were collected from overweight/obese PCOS patients,normal-weight PCOS patients and women with normal ovarian function.Quantitative proteomic study was performed and bioinformatic analyses were conducted.The levels of differentially expressed proteins were detected by ELISA(enzyme linked immunosorbent assay)analysis.Clinical data of patients were compared to further investigate the impact of obesity on PCOS.3.PCOS animal models were constructed by subcutaneous injection of DHEA(dehydroepiandrosterone)to further detect the regulatory effect of CXCL12.The ovarian tissue morphology of the rats was observed by HE staining.Vaginal smears were conducted to monitor the estrous cycle.Expression of CXCL12 and CXCR7 in ovarian tissue and granulosa cells were detected by immunohistochemistry,RT-PCR,western blot and ELISA.BAX,Bcl-2 and PCNA in granulosa cells of rats after different treatments were also detected.KGN cells were treated with different doses and times of DHEA.Western blot,RT-PCR and CCK-8 were used to detect the regulation of CXCR7 and granulosa cell function.Results: 1.The analysis of Raman fingerprint spectral data showed that follicular fluid samples had obvious aggregation phenomenon in non-PCOS and PCOS groups,better than plasma samples.We have established different algorithms based on the data.The performance of Stacking algorithms by using follicular fluid samples was better than that of single model and the sensitivity,specificity and accuracy rate were higher than 80%.2.Quantitative proteomic analysis was conducted to analyze follicular fluid samples from three groups of participants.Several differentially expressed proteins were detected.Bioinformatic results suggested that the identified proteins were related to inflammatory,immunological,metabolic processes and obesity could aggravate these changes.Significantly higher levels of FETUB and C5 were found in follicular fluid of PCOS patients by ELISA.The levels of APOA2 were significantly higher in overweight/obese PCOS group than those in the control group.Levels of CXCL12 in PCOS patients were significantly lower.Obesity was associated with changes in a variety of hormones and metabolic indicators in PCOS patients.3.There were multiple cystic follicles,thinner granulosa cell layers and decreased number of corpus luteum in the DHEA treated rats.Vaginal smears showed irregular estrous cycles of rats.Lower serum CXCL12 levels,lower CXCR7 levels in ovarian tissue,lower CXCL12 and CXCR7 levels were found in granulosa cells of the DHEA treated rats than those in the control group.After treatment with DHEA+CXCL12,the ovarian tissue morphology and estrous cycle of the rats were partly recovered.Moreover,the expression of CXCR7 was increased,BAX was decreased,Bcl-2 and PCNA were increased in granulosa cells.With the increasing concentration or longer time of DHEA treatment,levels of CXCR7 in KGN cells were decreased,which were related to the decrease of proliferation function.Conclusion: 1.Application of Raman spectroscopy combined with machine learning algorithm analyses suggested that follicular fluid samples had higher specificity,sensitivity and accuracy in distinguishing PCOS from non-PCOS data.2.A panel of proteins were found differentially expressed in the follicular fluid of PCOS patients.Changes in the follicular environment might be correlated with inflammatory,immunological,and metabolic abnormalities,which could be aggravated by obesity.3.Supplementation of CXCL12 could ameliorate the effect of androgen on granulosa cells,improve ovarian morphology and function of PCOS rats.CXCL12 might be indicated as a therapeutic target for PCOS. |