Proteomic analysis of human follicular fluid associated with successful in vitro fertilization

Background Human follicular fluid (HFF) provides a key environment for follicle development and oocyte maturation, and contributes to oocyte quality and in vitro fertilization (IVF) outcome. Methods To better understand folliculogenesis in the ovary, a proteomic strategy based on dual reverse phase high performance liquid chromatography (RP-HPLC) coupled to matrix-assisted laser desorption/ionization time-of-flight tandem mass spectrometry (LC-MALDI TOF/TOF MS) was used to investigate the protein profile of HFF from women undergoing successful IVF. Results A total of 219 unique high-confidence (False Discovery Rate (FDR) < 0.01) HFF proteins were identified by searching the reviewed Swiss-Prot human database (20,183 sequences), and MS data were further verified by western blot. PANTHER showed HFF proteins were involved in complement and coagulation cascade, growth factor and hormone, immunity, and transportation, KEGG indicated their pathway, and STRING demonstrated their interaction networks. In comparison, 32% and 50% of proteins have not been reported in previous human follicular fluid and plasma. Conclusions Our HFF proteome research provided a new complementary high-confidence dataset of folliculogenesis and oocyte maturation environment. Those proteins associated with innate immunity, complement cascade, blood coagulation, and angiogenesis might serve as the biomarkers of female infertility and IVF outcome, and their pathways facilitated a complete exhibition of reproductive process. Electronic supplementary material The online version of this article (doi:10.1186/s12958-017-0277-y) contains supplementary material, which is available to authorized users.


Background
In vitro fertilization (IVF) coupled with embryo transfer into uterus has been applied as treatment for infertility several decades. IVF was initially used to assist the reproduction of sub-fertile women caused by tubal factors [1]. With the improvement of IVF techniques, IVF is now a routine treatment for many reproductive diseases. However, the success rate of pregnancy is still a problem in clinical IVF practice, which is only about 50% even if the embryos with normal morphology were used for transfer [2]. In order to select embryos with the best potential good for IVF outcome, morphological assessments of blastocyst and blastocoels have been adopted, but it was still difficult to predict the quality of embryos [3]. Therefore, it was necessary to develop new strategies for embryo quality evaluation. Epidemiologic investigations showed that many intrinsic and extrinsic factors contributed to the quality of embryo. Because oocyte quality directly influences embryo development, HFF (microenvironment of oocyte maturation) became a main factor contributing to the success of IVF treatment [4].
Small antral follicles respond to ovarian stimulation by increasing in size due to rapid accumulation of follicular fluid, as well as granulosa cell divisions, which necessitate follicular basal lamina expansion. The components of HFF had several origins: secretions from granulosa cells, thecal cells, occytes, and blood plasma composition transferred through the thecal capillaries [5]. The major components of HFF were proteins [6], steroid hormones [7], and metabolites [8]. HFF provided a special milieu to facilitate the communications between occyte and follicular cells, the development of follicle and the maturation of occytes. The alteration of HFF proteins reflected disorders of main secretary function of granulosa cells and thecae, and the damage of blood follicular barrier, which was associated with abnormal folliculogenesis [9] and a diminished reproductive potential [10]. In IVF treatment, HFF was easily accessible during the aspiration of oocytes from follicle, and was an ideal source for noninvasive screening of biomakers for oocyte maturation, fertilization success, IVF outcome, pregnancy, and ovarian diseases.
In the postgenomic era, proteomic techniques have been widely used in the field of reproductive medicine. HFF proteome has become a hotspot for research, which not only contributed to discovering proteins related to IVF outcomes, but also improved our comprehensive understanding of physiological process during follicle development and oocyte maturation [11]. Li and co-workers used surface-enhanced laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS) combined with weak cation-exchange protein chip (WCX-2) to search for differentially expressed HFF proteins from mature and antral follicles [12]. Two-dimensional gel electrophoresis (2D-GE) followed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) was also used to identify 8 differentially expressed HFF proteins related to immune and inflammatory responses from controlled ovarian hyperstimulation (COH) and natural ovulatory cycles [13]. Ambekar and co-workers carried out SDS-PAGE, OFF-GEL and SCX-based separation followed by LC-MS/MS analysis to characterize 480 HFF proteins for a better understanding of folliculogenesis physiology [14]. Chen and co-workers explored the HFF biomarkers between successfully fertilized oocytes and unfertilized mature oocytes through nano-scale liquid chromatography coupled to tandem mass spectrometry (nano LC-MS/MS), and found 53 peptides to be potential candidates [15]. Although proteomic researches on HFF deepened our understanding of reproductive process and provided candidates related to oocyte quality, follicle development, IVF outcome and ovarian disorders, it was still essential to fully delineate the HFF networks and pathways involved in the physiology of reproduction and pathophysiology of infertility.
In the present study, we carried out an in-depth proteomic analysis of HFF from women undergoing successful IVF based on dual RP-HPLC coupled to MALDI TOF/TOF MS. The results profiled candidate biomarkers for the prediction of oocyte maturation, fertilization, and pregnancy and provided a new complement for HFF dataset, which will improve the understanding of biological processes and complicated pathways and interaction networks in HFF.

Patients enrollment and sample preparation
The HFF samples were collected from 10 women who underwent IVF treatment and achieved pregnancy. The selected patients met the following criteria: infertility not caused by tubal factor; aged less than 38 years; serum FSH values <12 mIU/mL; undergoing their first fresh egg retrieval cycle; ovulation stimulated with the long protocol. The patients were also without chromosomal abnormalities, polycystic ovary syndrome (PCOS), endometriosis and or endocrine disease. Cause of infertility was simple male factor. The body mass index (BMI) of patients met the normal criteria proposed by WGOC (18.5 ≤ BMI ≤ 23.9 kg/m 2 ) [16][17][18]. Ovarian stimulation and oocyte retrieval were performed as previously described [19]. Briefly, when more than two follicles exceeded 18 mm in diameter, 10,000 IU of HCG (Merck Serono, Swiss) was injected intramuscular. After 36 h, HFF was collected during trans-vaginal ultrasound guided aspiration of oocytes. The resultant HFF samples were macroscopically clear and without contamination of the flushing medium.
The samples were centrifuged at 10,000×g at 4°C for 30 min to produce cell debris-free HFF fraction for further analysis. Concentration of HFF was determined by the Bradford method [20]. This work has been approved by the Ethics Committee of Beijing BaoDao Obstetrics and Gynecology Hospital, and written informed consents were obtained from all participants.

Protein identification
Protein identification was performed with the ProteinPi-lot™ software (version 4.0.1; AB SCIEX). Each MS/MS spectrum was searched against a database (2017_03 released UniProtKB/Swiss-Prot human database, 20,183 entries) and a decoy database for FDR analysis (programmed in the software). The search parameters were as follows: trypsin enzyme; maximum allowed missed cleavages 1; Carbamidomethyl cysteine; biological modifications programmed in the algorithm. Proteins with high-confidence (FDR < 0.01) were considered as positively identified proteins.

Western blot analysis
According to the method described previously [27,28], 50 μg HFF protein were separated by a 12% SDS-PAGE gel and then electronically transferred onto a nitrocellulose membrane. The resultant membrane was blocked with 5% (w/v) skimmed milk for 1 h at 37°C, and then was incubated with the primary antibody (Abcam, Cambridge, USA, diluted 1:2000) at 4°C overnight. After washing with TBST for three times, the membranes were incubated with horse-radish peroxidase-conjugated secondary antibody (diluted 1:5000, Zhong-Shan Biotechnology, Beijing, China) at room temperature for 1 h. The immunoreactive proteins was visualized by enhanced chemiluminescence detection reagents (Pierce, Rockford, IL, USA) (Additional file 1: Table S1).

Results
Identification of high-confidence HFF proteome by dual RP-HPLC coupled with MALDI TOF/TOF mass spectrometry. A peptide sequencing strategy was applied by using two-dimensional chromatography-MALDI TOF/TOF mass spectrometry. We employed high pH (pH 10) reverse phase liquid chromatography to decrease the complexity of the tryptic digest of the HFF proteins, and collected 28 fractions. Then each fraction was further separated by low pH (pH 3) reverse phase liquid chromatography, and spotted on the plate using the Tempo™ LC-MALDI Spotting System. After sequencing by a 5800 MALDI TOF/TOF mass spectrometry, the resultant spectra were analyzed by ProteinPilot™ software by searching the reviewed Swiss-Prot human database (20,183 sequences, 2017_03 released). A total of 219 unique high-confidence (FDR < 0.01) proteins were identified by two replicates (Table 1). Experiment 1 and 2 identified 188 with 2747 unique peptides and 179 proteins with 2800 unique peptides, respectively. 148 common proteins were shared between the two experiments. Figure 1 showed representative MS/MS spectra of peptides from the identified HFF proteins. The m/z of precursor (Fig. 2c) was over 2500, and almost all bions and y-ions were still obtained based on a 5800 MALDI TOF/TOF mass spectrometry.

Bioinformatics analysis of the HFF proteome
The proteins identified by mass spectrometry were broadly placed into several GO categories on the basis of the PANTHER, DAVID and PubMed databases (Fig. 2). Based on molecular function, the majority (31%) of proteins were related to immunity, whereas other involved protein functions were mainly complement and coagulation (17%), protease or inhibitor (14%), and transportation (10%) (Fig. 2a). Based on subcellular localization, the majority (64%) of the identified proteins located in extracellular region. Other main locations were extracellular matix (7%), nuleus (6%), and cytoskeleton (5%) (Fig. 2b). Based on biological process, the majority (28%) of proteins was related to developmental process, and the next prevalence was immunological system process (26%). The other groups were involved into protein metabolic process (12%), reproduction (5%), lipid metabolic process (3%), and transportation (2%) (Fig. 2c).
A protein-protein interaction network was constructed by retrieving the STRING database. 151 proteins were in connection with other proteins, which lead to 738 paired relationships. As an example, 21 of 151 proteins related to basement membrane-specific heparan sulfate proteoglycan core protein (HSPG) was chosen, and 105 paired relationships were connected (Fig. 4).

Comparison of present HFF proteome, the previous reported HFF proteome and human plasma proteome
To disclose the overlap of the HFF proteomes between different labs and to explore the orign of the HFF proteins, the previous reported HFF proteins [14] and the human plasma proteome [29] were selected, whose protein identification criteria were both at a false discovery rate (FDR) of 1%. The results reflected the overlap of our HFF proteins and the previously reported HFF proteins with human plasma proteins (Fig. 5). A total of 49% proteins in our HFF data were common to the previous HFF data. Compared with human plasma proteins, 69% proteins from our HFF data and 64% proteins from previous HFF data were common to human plasma proteins.

Western blotting analysis
To verify the confidence of the proteome data, the expression patterns of 3 HFF proteins (retinol-binding protein 4, vitamin D-binding protein and lactotransferrin) from 10 women undergoing successful IVF were analyzed by western blotting (Fig. 6). Those three proteins could be detected in all 10 HFF samples. Compared with retinol-binding protein 4 and lactotransferrin, the expression of vitamin D-binding protein was relatively constant level in the HFF of ten women.

Discussion
Proteomics has been carried out to discover HFF biomarkers for decades, and liquid chromatography coupled with ion trap MS became widely available with the development of high-throughput sequencing. The identification of HFF proteins from women with and without endometriosis was performed using ESI MS/MS [30]. Nanoflow LC-MS/MS combined with TMT labeling was used to identify HFF biomarkers from women undergoing IVF/ICSI treatment with or without folic acid supplement [31]. Another advance LTQ Orbitrap system coupled with LC was also applied to comparing HFF proteins between fertilized oocytes and non-fertilized oocytes from the same patient [32]. Based on sample pre-fractionation using microscale in-solution isoelectric focusing (IEF), capillary electrophoresis (CE) coupled offline to matrix assisted laser desorption/ionization time of flight tandem mass spectrometry (MALDI TOF MS/MS) identified 73 unique proteins [33]. Hanrieder and coworkers [34] utilized a proteomic strategy of IEF and reversed-phase nano-liquid chromatography coupled to MALDI TOF/TOF mass spectrometry to identify 69 proteins related to controlled ovarian hyper stimulation (COH) during IVF. However, limited proteins were identified which delayed the research of HFF protein networks.
In the present work, a dual RP-HPLC coupled with MALDI TOF/TOF mass spectrometry was performed to identify HFF protein profiles associated with successful IVF, and 219 unique high-confidence (FDR < 0.01) HFF proteins were identified by searching the reviewed Swiss-Prot human database (20,183 sequences). Meanwhile, the new strategy indicated that the effective dual reverse LC pre-fractionation [21] could identify more HFF proteins.
Ambekar and co-workers carried out SDS-PAGE, OFFGEL and SCX-based separation followed by LC-MS/MS (an LTQ-Orbitrap Velos MS) to identify 480 HFF proteins with high confidence (FDR < 0.01) [14]. A comparison with our results and these data showed that more than 50% proteins in present study were not found in previous dataset (Additional file 2: Figure S1), which indicated that the data from different MS platforms were complementary. Retinol-binding protein 4 and vitamin D-binding protein were verified by western blotting, and the results showed they were all expressed in the 10 HFF samples. Lactotransferrin was uniquely included in Ambekar's data, and was also successfully detected by western blotting in our study. This result not only testified the good quality of Ambekar's data, but also facilitated to integrate the data from different MS platform in the future. Interestingly, more than 60% of combined HFF proteins from our data and Ambekar's data were found in the reported human plasma data [29]. HFF was a complex mixture, and the content of HFF mainly originates from the transfer of blood plasma constituents via theca capillaries, and the secretion of granulosa and thecal cells [5]. From the above contrast, we considered the transfer of plasma proteins was the major source of HFF, and the alternative permeability of theca capillaries would change the HFF compositions which inevitably impaired the oocyte quality, and even caused unsuccessful IVF outcome.
Bioinformatics analysis showed that 5% HFF proteins were involved in lipid metabolism and transport process. It has been reported that ageing could decrease apolipoprotein A1 and apolipoprotein CII, while increase apolipoprotein E, which were associated with the decline in production of mature oocytes and the decline in fertility potential [35]. Preconception folic acid supplementation upregulated apolipoprotein A-I and apolipoprotein C-I of the HDL pathway in human follicular fluid, which increased embryo quality and IVF/ICSI treatment outcome [30]. In our HFF data, apolipoprotein A-I, apolipoprotein A-II, apolipoprotein A-IV, apolipoprotein C-I, Fig. 3 Presentative Network of protein HSPG2 in the identified HFF proteome. A total of 21 genes are connected with 105 paired relationships annotated by STRING database. The relationships among proteins were derived from evidence that includes textmining, co-expression, protein homology, gene neighborhood, from curated databases, experimentally determined, gene fusions, and gene co-occurrence (as shown in the legend with different color) apolipoprotein C-II, apolipoprotein C-III, apolipoprotein D, apolipoprotein E, apolipoprotein F, and apolipoprotein M were all found, which indicated that those apolipoproteins were related to cholesterol homeostasis and steroidogenesis and played important roles in the maintenance of oocyte maturation microenvironment.
Pathway analysis showed that complement and coagulation cascades were the most prominent pathways (P_Value = 5.8E-52). Complement cascade promoted coagulation through the inhibition of fibrinolysis, and coagulation cascade in return amplified complement activation. Complement cross_talked with coagulation in a reciprocal way [36]. For example, plasmin, thrombin, elastase and plasma kallikrein could activate C3 [37]. Coagulation activation factor XII could cleave C1 to activate the classical complement pathway [38]. And thrombin could also directly cleave C5 to generate active C5a [39]. Among our HFF proteins, components (F12, KLKB1, PLG, KNG1, F9, F10, SERPINC1, SERPIND1, SERPINA5, F2, PROS1, PROC, SERPINA1, SERPINF2, A2M, CPB2, and FGA) of extrinsic pathway and intrinsic pathway in coagulation cascade and those (FH, FI, FB, C3, C1qrs, SERPING1, C2, C4, C4BP, C5, C6, C7, C8A, C8B, C8G, C9, FGA, FGG, PLG, FGB, F10) of alternative pathway, classical pathway, and lectin pathway in complement cascade were all identified. During follicle development and ovulation, coagulation system in HFF contributed to HFF liquefaction, fibrinolysis and the breakdown of follicle wall [40,41]. Follicle development had been hypothesized as the controlled inflammatory processes in 1994 [42], and inappropriate complement activation was linked to abortion [43]. Inhibition of complement activation improved angiogenesis failure and rescued pregnancies [44]. The paired comparison of HFF with plasma showed C3, C4, C4a, and C9 as well as complement factor H and clusterin might contribute to the inhibition of complement cascade activity for women undergoing controlled ovarian stimulation for IVF [45]. However there were still debates on the role of complement cascade in IVF. Physiologic complement activation protected the host against infection in normal pregnancy [46]. In comparison with those non-fertilized oocytes, C3 was more abundant in HFF from fertilized oocytes [47]. In the course of IVF treatment, the functions of complement and coagulation cascade were very complicated during ovarian hyperstimulation. More works were still deserved in both mechanism research and clinical practice.
Based on the analysis of STRING, we discovered a profound HFF protein-protein interaction networks. 151 of 219 HFF proteins participated in the network with 738 paired relationships. Basement membrane-specific HSPG was found as a node, which was also a potential biomarker for oocyte maturation in HFF. HSPG was widely distributed on the surface of animal cells, and especially strongly expressed in granulosa cells. HSPG played a critical role in controlling inflammation control through binding and activating antithrombin III during folliculogenesis [48]. Women with PCOS showed HSPG defect in follicular development [49], and on the contrary, HSPG was up-regulated in the fertilized-oocyte HFF [32]. In the network, HSPG interacted with 20 of 219 HFF proteins, and constructed 105 paired relationships. We deduced that the loss of HSPG might affect the function of the whole network or more complicated interaction maps, which might cause subsequent failures of oocyte maturation, fertilization, and IVF treatment.

Conclusions
HFF had a natural advantage for the noninvasive prediction of oocyte quality and IVF treatment outcome. The present study would provide a new complementary dataset for better understanding of oocyte maturation, and also delineate a new networks and pathways involved into the folliculogenesis. Furthermore, those novel findings would facilitate to testify the potential biomarkers associated with oocyte quality and IVF outcome. In the future, international laboratory collaboration should be established to standardize and optimize experimental design, patient selection, HFF handling, analysis methods, data standard, and clinical verification, which will greatly promote basic research of reproductive medicine, and ultimately accelerate the clinical transformation.