Transcriptomic comparison of ovarian granulosa cells between adult sheep and prepubertal lambs

Background The oocyte development ability of prepubertal animals is significantly lower than that of adult animals. Granulosa cells (GCs) have an important function on regulation of follicular and oocyte development. Therefore, analysis of GC characteristics can be used to explore the developmental mechanism of follicles and oocytes. Results In order to understand the possible reasons for the differences in follicle and oocyte development between lambs and adult sheep, we utilized high-throughput sequencing technique to analyze the transcriptome of GCs from follicle-stimulating hormone (FSH) superstimulated adult ewes and prepubertal lambs. Adult ewes were treated with FSH for 3 days (group A) and lambs were FSH-treated for 2 days (group B) or 3 days (group C). Transcriptome analysis of GCs showed that there were 405 and 159 differentially expressed genes from A vs. B and A vs. C, respectively. The results indicated that prolonging the FSH-treatment of lambs made the GC state of lambs more similar to the adult ewes, but there were still a large number of differentially expressed genes between adult ewes and lambs. Further analysis showed that many differently expressed genes were implicated in cell proliferation and apoptosis, oocyte development and follicular ovulation. Cellular examination demonstrated that fatty acid binding protein 4 (FABP4), which was highly expressed in lamb GCs, had a potential of promoting cell apoptosis. Cytoplasmic phospholipase A2 (PLA2G4A), which was expressed lowly in lamb GCs, may be responsible for reduced synthesis of prostaglandins in cells and impaired follicle/oocyte development. In contrast, glutathione S-transferase β-1 (GSTT2B) and forkhead boxO6 (FOXO6) had no apparent effect on the proliferation and apoptosis of GCs. Conclusions Our study found dramatic transcriptomic differences in GCs between lambs and adult sheep, which may explain the possible reasons for the defects of follicle and oocyte development in lambs compared to adult sheep. Our data provides important information for further understanding the mechanism of follicular development in prepubertal animals and improving their oocyte developmental competence. Supplementary Information The online version contains supplementary material available at 10.1186/s12864-022-08379-x.


Results
In order to understand the possible reasons for the differences in follicles and oocytes development between lambs and adult sheep, we utilized high-throughput sequencing technique to analyze the transcriptome of GCs from follicle-stimulating hormone (FSH) superstimulated adult ewes and prepubertal lambs. Adult ewes were stimulated with FSH for 3 days (group A) and lambs were FSHstimulated for 2 days (group B) or 3 days (group C). Transcriptome analysis of GCs showed that there were 405 and 159 differentially expressed genes from A vs. B and A vs. C, respectively. The results indicated that prolonging the FSH-stimulation of lambs made the GC state of lambs more similar to the adult sheep, but there were still a large number of differentially expressed genes between adult sheep and lambs. Further analysis showed that many differently expressed genes were implicated in cell proliferation and apoptosis, oocyte development and follicular ovulation. Cellular examination demonstrated that fatty acid binding protein 4 (FABP4), which was highly expressed in lamb GCs, had a potential of promoting cell apoptosis. Cytoplasmic phospholipase A2 (PLA2G4A), which was expressed lowly in lamb GCs, may be responsible for reduced synthesis of prostaglandins in cells and impaired follicle/oocyte development. In contrast, glutathione S-transferase β-1 (GSTT2B) and forkhead boxO6 (FOXO6) had no apparent effect on the proliferation and apoptosis of GCs.

Background
The technology of in vitro embryo production with juvenile animals as oocyte donors, so-called juvenile in vitro embryo transfer (JIVET), has a great application prospect in livestock breeding [1]. However, the embryonic development ability of oocytes from prepubertal animals is quite low and thus limits the application of JIVET technology in large scale [2,3]. In sheep, gonadotrophin treatment of prepubertal lambs can stimulate a great number of follicles to grow up to medium sizes (2-4 mm in diameter) at oocyte aspiration, but the de-rived oocytes are always not competent during subsequent development in vitro when used in JIVET program [4,5]. The reasons for poor development of prepubertal oocytes are not fully understood and remains to be investigated.
In vivo, the oocytes grow and develop in follicles and suboptimal follicular environment may compromise the oocyte competence. Within an antral follicle, the oocyte is exposed to follicular uid (FF) and enclosed by surrounding granulosa cells (GCs). GCs exert very important roles on oocyte development through secretion of factors into FF or direct communication with oocytes via gap junctions [6,7]. Thus, the activity of GCs is correlated with oocyte developmental potential.
Recently, transcriptomic and proteomic analysis of GCs found that hundreds of genes or proteins were differentially expressed between adult ewes and juvenile lambs [8,9]. Proteomic pro ling identi ed 243 proteins differentially abundant in lamb FF compared to adult ewes [10]. Transcriptional analysis of whole ovaries revealed that the expression patterns of miRNAs and mRNAs were quite different between adults and lambs [11]. Similarly in bovine, prepubertal heifers and adult cows have distinct properties of follicular environment and many genes that in uence follicular functions and oocyte development are differently expressed in GCs between heifers and cows [12,13]. These studies suggest that the functional insu ciency of prepubertal follicular environment may ex-plain the reduced oocytes quality of prepubertal animals. However, previous studies have not strictly correlated the developmental ability of oocytes with the gene expression of GCs. In addition, the possible functions of identi ed genes need to be tested in cells.
To further understand the mechanism of prepubertal follicle development, in the present study, we compared the transcriptome of GCs between lambs and adults under the same superovulation regime and validated the effects of some differently expressed genes. Based on the data, we indicate that FSHtreatment of lambs for 3 days is more fa-vorable to follicle and oocyte development than treatment for 2 days, but there are still signi cant differences in GC function between lambs and adult ewes.

Animals, hormonal treatment and sample collection
The procedure of all animal experiments was in accordance with the animal care policies of China Agricultural University and was approved by the Animal Ethics Committee at the university. Dorper breed female lambs (4-6 wk old) or adult ewes (3-4 y old) were selected for the experiments and maintained in Inner Mongolia Sino Sheep Tech-nology Co. Ltd, Wulanchabu, Inner Mongolia, China. Each lamb received 180 IU follicle-stimulating hormone (FSH) (Sansheng Pharmaceutical, Ningbo, China) in total, including 4×45 IU (2-day group) or 6×30 IU (3-day group), given at approximately 12 h intervals. At the time of the rst FSH injection, lambs were simultaneously injected with 400 IU equine chorionic gonadotropin (eCG; Sansheng). Adult ewes were implemented an intravaginal CIDR (InterAg, New Zealand) at any days of the cycle and 12 d later they received 6 × 30 IU injection of FSH with an interval of 12 h for each injection. All animals were subjected to one cycle of hormonal treatment and used only once in the experiments.
Oocytes were recovered from lambs and ewes approximately 12 h after the last FSH injection as described previously [14]. The animals were induced to general anesthesia by intramuscular injection of anesthetic and ovaries were exposed by midventral laparotomy. Follicles ≥ 2 mm in diameter on ovarian surface were punctured and aspirated using a 10 G needle disposable syringe. The cumulus-oocyte complexes (COCs) were picked under a stereo microscope and used for in vitro maturation. The remaining GCs were collected and cultured in DMEM/F12 supplemented with 10% fetal bovine serum (FBS) for 2-3 d. When the cell cultures reached 80% of con uence, the cells were harvested and stored at -80°C until RNA extraction.

RNA-seq and identi cation of differentially expressed genes
For Illumina sequencing, total RNAs from GCs were extracted from samples by Trizol Reagent (Tiangen, Beijing, China). The samples were from three groups of animals, including adult ewes (A), lambs with FSH-treatment for 2 days (B) and for 3 days (C), with each group consisting of 3 biological replicates. The quality of total RNAs was assessed using Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). After reverse transcription, cDNA libraries were constructed and sequenced by using Illumina HiSeq™ 2000. After nine sets of raw reads were obtained, the data were deposited by removing the contaminated, datper and low-quality reads, and all reads were uniquely mapped onto the ovine genome (https://www.ncbi.nlm.nih.gov/genome/?term=Ovis%20Aries).
For RNA-seq analysis, we used the Python-based toolkit HTseq to e ciently and accurately mapped reads to genes, and nally used the FPKM (fragment per kilo-bases per million mapped reads) method to calculate the expression level of each gene. Comparisons of FPKM in three groups (A vs. B, A vs. C and B vs. C) were performed, and genes with q-value ≤ 0.01and |log2FC|>1 were considered as signi cant differently expressed genes (DEGs) for subsequent analysis.

Data analysis
We used Gene Ontology (GO) enrichment terms for the functional categories of DEGs. GO has a total of three ontologies, which describe the molecular function of genes, cellular components and biological processes. Then GO enrichment analysis of differentially expressed transcripts was implemented using GO-seq software package. All DEGs were mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg) to further investigate their functions. GO terms and KEGG pathway with corrected p-values less than 0.05 were considered signi cantly enriched. The data sets of proteinprotein interactions (PPIs) were obtained from the online data-base STRING 11.0 (http://www.stringdb.org/) and by a PPI prediction method called domain-domain interaction (DDI) that is based on the conservatism of the domain. After get-ting the PPIs, we made full use of Cytoscape 3.8.2 to show the PPIs and utilized the Clue-GO to cluster the GO terms and KEGG pathways.

Validation by quantitative PCR (qPCR)
Puri ed total RNA of GCs was used as a template for cDNA synthesis by using FastKing gDNA Dispelling RT SuperMix (Tiangen, China). The real-time quantitative PCR was performed using the SuperReal PreMix Plus (Tiangen) in an ABI PRISM 7500 System (Applied Biosystems, USA). The 2 -ΔΔCT method was used to determine the gene expression level. All primers were designed using Primer 5.0 (supplementary Table   S5). All samples were analyzed in triplicates and all experiments were repeated for three times at least.

Overexpression and interference of target genes in GCs
Coding sequences (CDS) of speci c target genes were cloned from GC cDNAs. Accession number of the genes and designed primers are shown in supplementary Table S5. The CDS were constructed into eukaryotic expression vectors, pIRES2-ZsGreen1or pIRES-puro3 (Clontech, USA). The small interfering RNAs (siRNAs) targeting genes and their control siRNA (si-con) were designed and synthesized by Sangon Biotech (Shanghai, China). The overexpression vectors or siRNAs were transfected into cultured GCs using Lipofectamine 3000 (Invitrogen, Carlsbad, CA) according to the manufacturer instructions.

Assessment of cell viability
The CCK-8 assay was used to assess the viability of transfected cells. The cells were seeded in 96-well plates and incubated at 37°C. The indices of cell proliferation were measured using a CCK-8 kit (Beyotime Institute of Biotechnology, China) at 12, 24, 36, 48, and 60 h after transfection. Then 10 µL CCK-8 solution was added into the well and incubated for 1 h at 37°C. The optical density was measured using a microplate analyzer and the reference wavelength was set at 650 nm and the detection wavelength was set at 450 nm.

TUNEL assays
The TUNEL was used for detecting DNA breaks by using One Step TUNEL Apoptosis Assay Kit, red uorescence (Beyotime). Cell slides from different experimental groups were xed with 4% paraformaldehyde at room temperature for 30 min and then treated with 0.3% Triton X-100 in PBS for 5 min. After that, 50 µL TUNEL assay solution was dripped to cell slides, and the slides were incubated at 37°C for 60 min. After cells were counterstained with DAPI, the uorescent images were acquired with a CCD microscope. Six elds were selected randomly from each coverslip for observation and more than 200 cells from each eld were recorded. The intensities of uorescence in positive cells were quanti ed using Image-J software (National Institutes of Health). Apoptosis ratio was calculated as the number of apoptotic cells / total cell numbers.

Detection of arachidonic acid concentration
Sheep arachidonic acid (AA) ELISA detection kit (Beijing Shijichuangxiang Biotechnology, Beijing, China) was used to detect the concentration of AA in the medium of GCs. The cell culture medium was collected by centrifuge at 1000 rpm for 10 min. Antibodies against AA (100 µL) labeled with horseradish peroxidase (HRP) were added to holes of Elisa Plate contained standard sample and samples except the blank, and they were incubated at 37°C for 60 min. Then, the liquid was removed and the substrate of A and B were added to the hole of Elisa Plate and incubated for 15 min in dark. Finally, the terminated solution was added to the hole, and optical density was determined immediately at 450 nm by using a microplate analyzer.

Intracellular ROS measurement
A reactive oxygen species detection kit (Beyotime) was used to detect the levels of ROS in GCs according to the instruction of manufacture. DCFH-DA was added to the cell culture dish and incubated at 37°C for 20 min. Then the cells were harvested and the number of cells was counted by using the cell count plate. The uorescence intensity from 1 ml cell suspension was detected by using the uorescence spectrophotometer, and the uorescence intensity of each 10,000 cells was calculated.

Statistical analysis
Analysis of data were performed using SPSS 18.0 software for one-way analysis of variance and t-test analysis. The graphs, transcriptome map and PPI networks were drawn using GraphPad Prism 5.0 software, R x64 3.6.2 software package and Cytoscape 3.8.2, respectively. Table 1 shows the e ciency of oocyte harvest and embryo production from adult and prepubertal lamb donors. After FSH-treatment, the average number of oocytes recovered in the two lamb groups was much higher than that of adult ewes. After in vitro maturation and fertilization, there was no signi cant difference in cleavage rate between lambs and adults, but the blastocyst rate in the adult group was signi cantly higher than that of both lamb groups. There was no difference in the cleavage rate in the two lamb groups, but the blastocyst rate in the 3-day FSH-treatment group was signi cantly higher than that in the 2-day FSH-treatment group (p < 0.05). Values with different letters (a and b) within the same column are signi cantly different (p < 0.05).

Global gene expression
The transcriptome of ovine GCs was analyzed by global RNA sequencing of each sample. In order to ensure the reliability of data, we ltered out low-quality reads and re-moved the adaptor sequences, and nally obtained 98.36-99.4% clean data in each sample (Supplemental Table S1). As shown in Fig. 1, we calculated the rate of all reads from each sample mapped onto the genome, and showed that 73.30%-77.95% of the reads were uniquely mapped onto the ovine genome (https://www.ncbi.nlm.nih.gov/genome/?term=Ovis%20Aries). These data were used to analysis of gene expression.

Differentially expressed genes
Three groups of samples were analyzed for identifying differential genes between groups, and q-value ≤ 0.01 and |log2FC |> 1 were used as a standard for the DEGs thresh-old. We came up with the unsupervised hierarchical clustering of DEGs ( Fig. 2A). We found 405 DEGs between adult ewes (A) and lambs with 2-day FSH-treatment (B) (referred to as A vs. B), and the number of up-and down-regulated genes was 190 and 215 in B group, respectively. There were 159 DEGs between adult ewes and lambs with 3-day FSH-treatment (C) (referred to as A vs. C), including 72 up-and 87 down-regulated genes in C group, respectively. For B vs. C, 28 DEGs were found, of which 12 were up-regulated and 16 were downregulated in C group (Fig. 2B and supplementary Table S2). Overall, the number of DEGs from A vs. C was decreased by 246 compared to A vs. B.
We further explored the DEGs through pairwise comparative analysis of different groups. There were 101 common DEGs existed in A vs. B & A vs. C ( Fig. 3C and supplementary Table S2). Among them, 53 genes were up-regulated and 48 genes were down-regulated in both lamb groups compared to adult group. These DEGs may repre-sent the intrinsic differences between GCs of lamb and adult sheep, regardless of extending FSH-treatment to 3 days.

Analysis of GO and KEGG pathways
In order to further investigate the biological relevance of all DEGs, we performed GO analysis of DEGs to identify enrichment of biological processes in each group. The DEGs from A vs. B and A vs. C were enriched in many identical GO terms, but the enriched GO terms from A vs. B were more than A vs. C Next, we conducted GO and KEGG enrichment analysis for the common DEGs. A corrected p-value of 0.05 was set as the threshold to identify signi cantly different GO terms and KEGG pathways. As shown in Table 2 and supplementary Table S3, the common DEGs were enriched in 767 GO terms, including 533 biological processes, 91 cell components and 143 molecular functions. A total of 16 GO terms were enriched signi cantly (p < 0.05), including 9 biological processes (such as cell adhesion, extracellular matrix organization and regulation of cell proliferation), 5 cellular components (such as proteinaceous extracellular matrix, cell surface, and extracellular vesicular exosome), and 2 molecular functions (structural constituent of muscle and protein homodimerization activity). KEGG pathway analysis revealed that the common DEGs from A vs. B & A vs. C were enriched in 160 pathways and signi cantly enriched in 38 pathways, including adherens junction, regulation of lipolysis in adipocytes, ECM-receptor interaction, PI3K-Akt pathway, and protein digestion and absorption signaling pathway.  (2) KEGG Pathway 160 (40) The value in parenthesis is the signi cant enrichment number of differentially expressed genes with p-value ≤ 0.05.

Protein-Protein Interaction (PPI) network analysis
Protein-protein interaction maps are promising to reveal many aspects of the complex regulatory network underlying cellular function. The DEGs were analyzed by string database and domain-domain interaction. We found that the PPI network of A vs. B was reconstructed from 123 genes symbols (Fig. 4A  and supplementary Table S4). Biological function analysis on the gene symbols by clueGO revealed that the genes were related to endothelial cell migration, nephron development, carboxylic acid transport, fatty acid transport (BP), contractile ber part, collagen-containing extracellular matrix, actin-based cell projection (CC) and GABA-A receptor activity (MF) (Fig. 4B). The genes were involved in TGF-beta signaling, cholinergic synapse and morphine addiction pathways. The PPI network of A vs. C was reconstructed from the 41 genes symbols (Fig. 4A). The genes were related to complex of collagen trimers (CC), GABA-A receptor activity and extracellular ligand-gated ion channel activity (MF) (Fig. 4C).
The genes were involved in nicotine addiction GABAergic synapse and morphine addiction pathways. The PPI net-work of A vs. B & A vs. C was reconstructed from the 24 genes symbols (Fig. 4A). The genes were related to GABA-A receptor activity (MF) and involved in nicotine addiction, GABAergic synapse and morphine addiction pathways (Fig. 4D).

Validation of expression levels of DEGs
Nineteen genes associated with follicle development, oocyte maturation and ovulation were selected for experimental veri cation, including INHBE, PLA2G4, TNFAIP6, AFP, FBN1, ANGPT1, ANGPT2, SCUBE2, ERBB4, TGFβ1, ID3, ESR2, FST, PTGS2, FOXO6, GSTT2B, FABP4, HTRA3 and SERPINA14. The results suggested that the trend of up-and down-regulation of gene expression was consistent with the RNA-seq, and the gene expression was signi cant different between adult and lamb GCs ( Fig. 5A and B), proving that our data of high-throughput RNA-seq were reliable.

Functional veri cation of some genes
To further investigate the cellular function of genes, four genes including FABP4, PLA2G4A, FOXO6 and GSTT2B, were selected for functional examination in cells. We transfected overexpression vectors of these genes into adult GCs and achieved high expression of the target genes in cells (Fig. 6A-D). While, except for FOXO6, transfection of siRNAs could lead to signi cant decrease of expression of the target genes ( Fig. 6a-d).

The function of FABP4 in GCs
The effect of FABP4 on proliferation and apoptosis of GCs from adult ewes was examined. Compared with the control group, the GC viability began to signi cantly decrease at 24 h after transfection with FABP4 overexpression vector, and the cell activity at 36, 48 and 60 h after transfection was signi cantly lower than that of the control group (Fig. 7A). However, interference of FABP4 had no signi cant effect on the GC activity.
We then examined the expression of apoptosis-related genes (BAX, BCL2 and Caspase3). We found that the relative expression levels of pro-apoptotic gene BAX and Caspase3 were signi cantly increased in FABP4 overexpressed GCs from adult ewes, while no signi cant changes were observed in the expression of anti-apoptotic gene BCL2. However, there were no signi cant differences in expression of BCL2, Caspase3 and BAX between the interference and control groups (Fig. 7B). TUNEL assay showed that the proportion of TUNEL-positive GCs was increased in the overexpression group, while the apoptotic ratio was decreased in the interference group (Fig. 7C).
The function of PLA2G4A in GCs PLA2G4A is a cytosolic phospholipase that can promote the production of arachidonic acid (AA), a substrate of prostaglandin synthesis. We measured the level of arachidonic acid in culture medium of GCs, and found that the AA concentration in medium of lamb GCs was signi cantly lower than that of adult GCs (Fig. 8A), suggesting that the capacity of secreting AA from lamb GCs was inferior to adult GCs.
Overexpression of PLA2G4A in adult GCs signi cantly increased the content of AA in the medium and interference of PLA2G4A decreased the AA level. In lamb GCs (Fig. 8B), overexpression of PLA2G4A also resulted in a signi cant increase of AA levels (Fig. 8C).

The function of GSTT2B in GCs
The effects of GSTT2B on antioxidation, proliferation and apoptosis of GCs were investigated. The results showed that the expression levels of antioxidant genes SOD1, SOD2, CAT and GPX1 did not change signi cantly after GSTT2B overexpression in either adult (Fig. 9C) or lamb GCs (Fig. 9c). The expression of GPX1 was signi cantly decreased in the interference group (Fig. 9C), while no obvious changes were observed in the expression of SOD1, SOD2 and CAT ( Fig. 9C and c). We further found that the over-expression or interference of GSTT2B had no signi cant effect on the ROS levels ( Fig. 9B and b), cell viability ( Fig. 9A and a) and apoptosis ( Fig. 9D and d) in both adult and lamb GCs.

The function of FOXO6 in GCs
FOXO6 is related to regulation of cell apoptosis and the expression of FOXO6 was signi cantly lower in lamb compared to adult GCs. We found that, however, overexpression of FOXO6 in lamb GCs had no signi cant in uence on cell viability, cell apoptosis and the relative expression levels of BAX, BCL2 and Caspase3 in cells (Fig. 10A-C). As the siRNAs tested were not e cient for knockdown of FOXO6 expression, the interference experiment was not done in the study.

Discussion
Low in vitro developmental ability of prepubertal oocytes suggests that the oocytes undergo insu cient development during growth in vivo. In the sheep JIVET program, the lambs are usually treated with injection of FSH for 2 days prior to oocyte recovery [5,15]. In this study, we showed that FSH-treatment for 3 days signi cantly enhanced the development of lamb oocytes when compared to FSH-treatment for 2 days, although it was still not comparable to adult oocytes from the same treatment protocol. Thus, increasing FSH-treatment and extending the in vivo development duration of prepubertal oocytes would be favorable to the improvement of oocyte competence.
It is well known that there is a bidirectional communication between GCs and oocytes during development in follicles [6,7]. GCs can coordinate the nuclear and cytoplasmic maturation of oocytes [16] and affect the growth and development of follicles [17]. Therefore, analysis of GC characteristics can be used to evaluate the quality of follicles and oocytes.
In this study, we used transcriptomic analysis to investigate the possible functions of GCs in FSH-treated lambs and found dramatic differences in gene expression patterns between lamb and adult GCs. Especially, a total of 405 DEGs were found in lambs with 2-day FSH-treatment vs. adult ewes, which is similar to a previous report claiming 311 DEGs between lamb and adult GCs [9]. However, the number of DEGs decreased to 159 in lambs with 3-day FSH-treatment vs. adult ewes. The numbers of signi cantly enriched (q-value ≤ 0.05) GO terms and pathways from A vs. C were signi cantly less than that of A vs. B, which indicated that lengthening FSH treatment time probably made the follicular pattern of lambs more similar to adult ewes. Thus, compared to FSH treatment for 2 days, lambs with FSH-treatment for 3 days had been probably improved on some biological functions, such as cellular response to transforming growth factor beta stimulus, regulation of cell proliferation and extracellular vesicular exosome, etc.
We further investigated the DEGs by pairwise comparative analysis of the three groups and found 101 common DEGs shared with both 2-day and 3-day FSH-treatment of lambs vs. adults. Analysis of KEGG pathway combined with GO terms indicated that these common DEGs were enriched in biological processes such as cell adhesion, extra-cellular matrix organization, regulation of cell proliferation and so on. These DEGs were signi cantly enriched in some pathways including adherens junction, ECM-receptor interaction, cell adhesion molecules and PI3K-Akt pathway and so on. Among them, adherens junction (AJs) is the most common type of intercellular adhesion, which is pivotal for maintaining tissue structure and cell polarity, and can restrict the movement and proliferation of cells [18]. Cell adhesion molecule (CAMs) is a glycoprotein expressed on the cell surface, which plays an important role in a series of biological processes, including hemostasis, immune response, in ammation, embryonic development and neuronal tissue development [19,20]. In ECM-receptor interaction signaling pathway, extracellular matrix is composed of complex mixtures of structural and functional macromolecules, including glycoproteins, pro-teoglycans, aminoglycans and so on. It plays a major role in the morphogenesis of tissues and organs and in maintaining the structure and function of cells and tissues [21]. The interaction between cells and extracellular matrix leads to direct or indirect control of cellular activities, such as adhesion, migration, differentiation, proliferation and apoptosis [22]. In addition, extracellular matrix is also involved in the process of cumulus expansion and plays a role in oocyte maturation [23]. PI3K-Akt signal transduction pathway is activated by Akt phosphorylation and partici-pated in the regulation of basic cellular functions, such as the transcription, translation, proliferation and growth [24].
In order to further clarify the function mechanism of the difference of genes between lambs and adult ewes. We drew the PPI networks, which not only came from the string database, but also utilized a PPI prediction method called DDI. At the same time, enrichment of function analysis was studied by using clueGO. The results showed that CD83 was the core gene of three PPI networks. The expression of CD83 was up-regulated in lamb GCs. Previous studies have shown that CD83 is a member of the immunoglobulin (Ig) superfamily, which is elevated in the serum of patients with autoimmune disease and hematological malignancies and has an immune suppressive function [25]. In addition, Liu et al proposed that ovulation involved the expression and function of molecules that exert potent roles in innate immune responses, and granulosa cells appeared to play immuno-protective-like functions for the ovulated oocytes [26]. Another report found that the possible initiation of early follicular atresia in small antral follicles during the follicular phase had an interaction with the presence of immune cells [27]. Our results suggest that up-regulation of CD83 in lamb GCs may have an adverse effect on the immune system and development of oocytes.

Enrichment of function analysis revealed same and different enrichment of GO terms between A vs. B
and A vs. C. The candidate genes from PPI network of A vs. B were enriched in endothelial cell migration, fatty acid transport, Rho GTPase binding and collagen-containing extracellular matrix, and so on. Rho GTPase binding is participated indirectly in oocyte maturation, blastocyst formation and stem cell development by regulated Rho-associated coiled-coil-containing protein (ROCKs) [28]. Endothelial cell migration is essential for angiogenesis [29]. Meanwhile, angiogenesis is participated in the follicular development and ovulation [30], and thus there is a certain connection between the two. Extracellular matrix (ECM) constitutes the follicular basal lamina and is responsible for aggregation of granulosa cells, and it is vital for supporting ovarian follicle growth and maintaining its proper function [31]. In this study, we found some DEGs encoding ECM-related proteins in collagen-containing extracellular matrix (A vs. B) and complex of collagen trimmers (A vs. C), such as collagens COL15A1, FBN1 (A vs. B) and COL4A5, COL4A6, COL5A2 (A vs. C) [32]. Therefore, the enrichment of GO terms was related to the differences in follicular development and ovulation between lamb and adult ewes.
We veri ed the expression of 19 DEGs that are implicated in follicle and oocyte development by qRT-PCR.

Inhibin B (INHBE, also known as activin E) is expressed in GCs of early antral follicles and dominant
follicles, which promotes the development of single dominant follicle by inhibiting the secretion of pituitary FSH in follicular phase [33]. Follistatin (FST) is a binding protein of activins. The activinfollistatin system controls the growth and differentiation of antral follicles by affecting the production of gonadotropin receptors and steroids [34]. As a follicle develops to a dominant follicle, activin secretion is decreased, while inhibin and follistatin production are increased [35]. In our results, INHBE and FST were expressed lowly in lambs GCs, which may hinder the selection of dominant follicles and permit a large number of lamb antral follicles to grow in response to FSH stimulation.
ERBB4 is one of the four receptors in mammalian EGFR system [36]. Studies have demonstrated that EGF-like factors and EGFR signaling pathway played vital roles in the regulation of oocyte meiosis resumption, the expansion of cumulus cells and ovulation [37,38]. Prevot et al found that there was a signi cant delay in puberty in ERBB4 mutant mice, and their reproductive function was also impaired in adulthood [39]. In view of this, ERBB4 may play a certain role in the reproductive process, and its low expression in lamb GCs may not facilitate the activation of the EGFR signaling pathway, thus re ecting some developmental defects of lamb oocytes.
In agreement to a previous report [9], we found that HTRA3 was differently expressed between adult and lamb GCs. HTRA3 is a member of the HtrA family of serine proteases and plays an important role in ovarian development, GC differentiation and luteinization [40]. It was reported that the expression level of HTRA3 was signi cantly increased in luteinized GCs [41]. In this study, we found that the expression of HTRA3 in lamb GCs is signi cantly higher than that in adult GCs. This implies that lamb GCs may undergo luteinization, which is not favorable to the full development of follicles and oocytes.
ESR2, TGFβ1, TNFAIP6, PTGS2 and AFP are related to ovulation. As a receptor of estrogen, ESR2 is critical for full development and maturation of follicles and oocytes [42]. TGFβ1 can induce PTGS2 expression and prostaglandin E2 (PGE2) production in GCs through Smad signaling pathway, which further promotes oocyte maturation and ovulation [43]. TNFAIP6 speci cally expresses in mucilaginous liquefaction cumulus-oocyte complexes (COCs) [44], which is indispensable for the formation of extracellular matrix in COCs during ovulation [45]. Prostaglandin is involved in ovulation, and PTGS2 plays a key role in synthesis of prostaglandins [46]. AFP is a serum glycoprotein. Female mice with AFP knockout were unable to ovulate due to hypothalamic dysfunction [47]. The expression of these genes was signi cantly lower in lamb GCs than adult GCs, which may be the reasons that lamb follicles rarely ovulate.
FBN1 is a glycoprotein and is involved in regulating the activity of TGFβ. When FBN1 is silenced, the proliferation of cumulus cells is signi cantly increased and the apoptosis of cumulus cells is prevented [48]. The high expression of FBN1 in lamb GCs may inhibit the proliferation of GCs and affect the normal secretory and metabolic process of GCs.
SCUBE2 is a new type of vascular growth factor receptor 2 co-receptor and can enhance the signal transduction induced by vascular endothelial growth factor (VEGF) during angiogenesis, thus promoting angiogenesis [49]. ANGPT1 and ANGPT2 are participated in maturation and stabilization of blood vessels via tyrosine kinase receptors [50,51]. ID3 is involved in the control of cell cycle and cell fate, and may also play an important role in angiogenesis [52]. Fraser et al suggested that angiogenesis plays a critical role in follicular development, and interference with the process of angiogenesis can inhibit follicular development or prevent ovulation [53]. Low expression of ANGPT1, ANGPT2 and SCUBE2 in lamb GCs may hinder follicular maturation and ovulation by inhibiting follicular angiogenesis.
In this study, we examined the functions of four genes, FABP4, PLA2G4A, GSTT2B and FOXO6, in ovine GCs. FABP4 is expressed mainly in adipocytes and has been widely studied in obesity metabolic syndrome and cardiovascular disease [54]. Van et al observed that the ultrastructural characteristics of apoptotic GCs in bovine ovary was similar to those of FABP4 positive cells [55]. High expression of FABP4 was observed in the GCs of atretic follicles in mice [56]. It was found that FABP4 is closely related to the occurrence of polycystic ovary syndrome [57]. In this study, we found that FABP4 was highly expressed in GCs of lambs and demonstrated that overexpression of FABP4 promoted the apoptosis of GCs. FABP4 may play its role through PPARγ signal pathway [56]. It is also suggested that the promotion of apoptosis by FABP4 may be related to the upregulation of TNFAIP6 and down-regulation of TGF-α [58]. These may be the reasons that FABP4 promotes apoptosis, but other mechanisms may also exist and need to be further investigated.
PLA2G4A is a cytosolic phospholipase that can promote the production of arachidonic acid (AA), a substrate of prostaglandin synthesis [59]. Boruszewska et al suggested that PGE2 could facilitate successful oocyte maturation and oocyte survival in the cow [60]. In this study, the expression of PLA2G4A in GCs of adult ewes and 3-day FSH-treated lambs were higher than that of 2-day FSH-treated lambs. Overexpression and RNAi experiments demonstrated that PLA2G4A had an important effect on the ability of GCs to secrete AA. AA is a substrate for the synthesis of prostaglandin E2 that regulates the cumulus cells expansion [61]. Thus, the defective development of oocytes and follicles in lambs may be partially due to the low concentration of prostaglandin E2, and prolonged FSH treatment of lambs would be bene cial to the PLA2G4A expression and prostaglandin E2 synthesis, thereby pro-moting the oocyte maturation and development. In addition, Sapieha et al believed that AA, as an unsaturated fatty acid, promotes angiogenesis [62]. Therefore, the low concentration of prostaglandin E2 in lamb follicles may reduce the formation of mature blood vessels on follicles, and thereby affect the oocyte maturation and follicular ovulation.
GSTs are involved in glutathione-dependent oxidation resistance [63]. GSTT2B is a member of the GSTs family, but its function is little known. It may act as an antioxidant, promoting detoxi cation of reactive oxygen species (ROS). Our results suggested that GSTT2B was expressed lowly in lamb GCs. Therefore, we speculated that lamb GCs may have a lower antioxidant capacity related to GSTT2B de ciency. However, we found that overexpression of or interference with this gene in lamb and adult GCs, had no signi cant roles in antioxidation and apoptosis in cells.
FOXO factors are an important regulator of cell cycle, apoptosis, DNA repair, antioxidant stress and cell lifespan [64,65]. FOXO6 is a member of the FOXO family and is a vital regulator of insulin or liver glucose metabolism [66]. Knockdown of FOXO6 in colorectal cancer cells inhibited the cell proliferation, migration, invasion and glycolysis [67]. So far, there is no report of FOXO6 in ovine GCs. Our study showed that the expression of FOXO6 was signi cantly decreased in GCs of lamb treated with FSH for 2 and 3 days. We thought that FOXO6 might be a critical gene in regard to the difference in GC functions between lambs and adult ewes, and it may be related to cell proliferation. However, FOXO6 had no effect on cell proliferation and apoptosis in our assays. The role of FOXO6 in GCs needs further investigation.

Conclusions
In this study, we show that there were notable differences in transcriptional patterns of GCs between lambs and adult ewes, and a number of differently expressed gene were identi ed between the two groups of females, which partially explains the possible rea-sons for the developmental defects in lamb follicles and oocytes. Although the real actions of these differently expressed genes need to be further  The rate of reads mapped onto the genome. Each loop represents one alignment type and the radian of loop represents the rate of alignment. The total reads represent total clean reads of samples, and the ratio is 1. A_1-A_3: adult ewe group; B_1-B_3: group of lambs with 2-day FSH-treatment; C_1-C_3: group of lambs with 3-day FSH-treatment.    The relative expression level of FABP4 (A, a), GSTT2B (B, b), FOXO6 (C, c) and PLA2G4A (D, d) in transfected adult GCs. "Control" represents transfection with the empty plasmid; "Overexpress" represents transfection with overexpression vectors (transfected with 2.5 or 5.0 μg of plasmids); "siRNA" represents transfection with siRNAs (the e ciency of 3 different siRNA were examined). The results are expressed as mean ±SEM. * p < 0.05, ** p < 0.01, *** p < 0.001,**** p< 0.0001. Scale bar, 20 μm. The histogram is the quanti cation of TUNEL staining. The results are expressed as mean ±SEM. "Control" represents transfection with the empty plasmid; "Overexpress" represents transfection with the FABP4 overexpression vector; "siFABP4" represents transfection with siRNA; "NC" stands for interference negative control group.* p < 0.05, ** p < 0.01, *** p < 0.001,**** p< 0.0001.  The results are expressed as mean ± SEM. "control" represents the transfection with the empty plasmid, pIRES-puro3. "Overexpress" represents the transfection with GSTT2B overexpression vector. "NC" stands for interference negative control group; "siGSTT2B" represents the transfection with siRNA.  Supplementary Files