Changes in the transcriptome of bovine ovarian cortex during follicle activation in vitro

Abstract

The signals that regulate activation, a key transition in ovarian follicular development, are still not well understood, especially in nonrodent species. To gain insight into the regulation of this transition in cattle, we combined a microarray approach with an in vitro system in which ovarian cortical pieces cultured in control medium are enriched for primordial follicles, whereas pieces cultured with insulin are enriched for primary follicles. Total RNA was extracted from cultured cortical pieces, and then transcripts were identified and analyzed using the Affymetrix Bovine Genome GeneChip array. Around 65% of the transcripts in the bovine GeneChip were detected in cultured cortical pieces. Comparison between pieces cultured with or without insulin generated 158 differentially expressed transcripts. Compared with controls, 90 transcripts were upregulated and 68 were downregulated by insulin. These transcripts are involved in many biological processes and functions, but most are associated with cellular growth or cell cycle/cell death. The transcript encoding ubiquitin-conjugating enzyme E2C (UBE2C) was significantly upregulated during follicle activation, and Ingenuity Pathways Analysis revealed that UBE2C can interact with the tumor suppressor phosphatase and tensin homolog (PTEN). Both PTEN mRNA and protein were lower in cortical pieces cultured with insulin than in controls. In addition, FOXO3a, a downstream effector of PTEN signaling, underwent nuclear-cytoplasmic shuttling during primordial to primary follicle development in bovine fetal ovaries, further suggesting the involvement of the PTEN pathway in follicle activation in cattle. Genes and pathways identified in this study provide interesting candidates for further investigation of mechanisms underlying follicle activation.

the ovarian follicle is a key functional unit of the ovary. In most domestic animals and primates, follicle formation occurs during fetal life, whereas in rodents follicles form around the time of birth. After formation, primordial follicles remain arrested (up to decades in long-lived mammals, like humans) until they are activated to begin growth. Follicle activation, the transition of primordial follicles from quiescence into the growth phase, is characterized by a change in shape of the single layer of granulosa cells from flattened to cuboidal and the initiation of oocyte growth. Follicle activation is an irreversible process, and therefore, the rate at which follicles become activated is critical in regulating the size of the resting primordial follicle pool, which affects a female's reproductive life span and fertility. So far, mechanisms controlling primordial follicle activation are still poorly understood, especially in species of practical interest like domestic animals and humans.

Although how follicle activation is regulated in vivo is largely unknown, activation of primordial follicles in vitro has been achieved for several species, including rodents (10), cattle (2, 43), baboons (42), and humans (16). In cultures of whole rodent ovaries, kit ligand (33), leukemia inhibitory factor (30), basic fibroblast growth factor (28), insulin (23), and BMP-4 (31) can promote the primordial to primary follicle transition, whereas anti-Müllerian hormone (29), stromal derived factor-1 (15), progesterone, and estradiol (6, 21) inhibit the primordial to primary follicle transition. Moreover, results for mutant mouse models suggest that the phosphatase and tensin homolog (PTEN)/phosphoinositide 3-kinase (PI3K) signaling pathway controls follicle activation through the forkhead transcription factor 3a (FOXO3a). Oocyte-specific ablation of either Pten or Foxo3a in mice caused global primordial follicle activation (4, 20, 35).

Compared with rodents, very little is known about the regulation of follicle activation in domestic animals and humans, which are species of practical interest. In cattle, insulin and kit ligand promote, whereas anti-Müllerian hormone and steroids (progesterone and estradiol) inhibit, follicle activation in vitro (7, 13, 32, 44). Although these studies have begun to elucidate factors controlling follicle activation in cattle, most of the previous studies were based on testing individual “candidate factors” that appear to be important in rodents, to determine if they promote or inhibit the initiation of bovine follicle growth. Progress has been made using this approach, but the progress has been slow. More importantly, there may be additional factors that play important roles during follicle activation but are not yet identified. Microarray can measure the levels of RNA transcripts derived from thousands of genes simultaneously. In the present study, a gene discovery approach was used to identify new factors and genes that potentially regulate follicle activation in cattle by determining differences in global gene expression profiles between ovarian tissue enriched for resting primordial follicles or growing primary follicles. Cattle provide an excellent experimental model for studying follicle formation and activation, not only because they are an important food source, but also because the timing and the process of early folliculogenesis are remarkably similar in cattle and humans.

In cattle, the length of gestation is around 279 days, and formation of primordial follicles begins during the second-trimester of pregnancy. Therefore, most follicles in bovine fetal ovaries at 5–8 mo of gestation are primordial follicles, and this makes bovine fetal ovaries a good experimental model for studying follicle activation (43). Primordial follicles reside in the outer layer of the ovary, the cortex, and our lab developed a culture system that supports activation in vitro of primordial follicles in ovarian cortical pieces cultured in medium containing ITS+ [insulin-transferrin-selenious acid + BSA and linoleic acid; (43)]. Most primordial follicles in pieces of ovarian cortex obtained from fetal calves during the last trimester of pregnancy activate within 2 days of culture with ITS+ (12, 43). When cortical pieces are cultured with TS+ (identical to ITS+ but without insulin), cortical tissue remains healthy, but there is no increase in primary follicles after 2 days in culture (13). This shows that the insulin in ITS+ is responsible for activation and has provided an experimental model for comparing ovarian cortical pieces under conditions that do not or do promote activation in vitro. Using the experimental model, we identified candidate transcripts, potentially involved in regulation of follicle activation in cattle, in the present study by comparing the transcripts in bovine cortical pieces cultured with TS+ (containing predominantly primordial follicles) to cultures with ITS+ (containing predominantly primary follicles).

MATERIALS AND METHODS

Collection of bovine fetal ovaries.

Bovine female fetuses [5–8 mo gestation, estimated by crown-rump length (11)] were obtained from a local slaughterhouse (Cargill Regional Beef, Wyalusing, PA). Ovaries were collected and transported to the laboratory in Leibovitz L-15 medium (Life Technologies, Grand Island, NY) supplemented with 1% fetal bovine serum, 50 IU/ml penicillin, and 50 μg/ml streptomycin (Life Technologies) at ambient temperature (20–22°C), as previously described (43).

Culture of ovarian cortical pieces.

The cortex of fetal ovaries (n = 3 fetuses) was dissected from the medullary tissue and cut into ∼0.5 to 1 mm³ pieces. Freshly isolated cortical pieces from each fetus were fixed for histological analysis of numbers of primordial and primary follicles as previously described (43), as day 0 controls. The other cortical pieces were placed on uncoated culture well inserts (2 pieces/well; Millicell-CM, 0.4 μm pore size; Millipore, Bedford, MA) in the wells of 24-well Costar culture plates (Corning, Corning, NY) with 300 μl Waymouth's MB 752/1 medium (Life Technologies) supplemented with 25 mg/l pyruvic acid (Sigma Chemical, St. Louis, MO), antibiotics (50 IU/ml penicillin, 50 μg/ml streptomycin; Life Technologies), and ITS+ (6.25 μg insulin, 6.25 μg transferrin, 6.25 ng selenious acid, 1.25 mg BSA, 5.35 μg linoleic acid per milliliter; Collaborative Biomedical Products, Becton Dickinson Labware, Bedford, MA) or TS+ (identical to ITS+, but without insulin). Cortical pieces (36–40 pieces/treatment/fetus) were cultured at 38.5°C in a humidified incubator gassed with 5% CO₂/95% air for 2 days. At the end of culture, four pieces per treatment per fetus were fixed for histological analysis of follicle activation, and the remaining pieces were snap-frozen for later extraction of RNA for microarray analysis.

Histological analysis.

Follicle activation in cortical pieces was assessed by histological morphometry as previously described (43). In brief, cortical pieces were fixed for 1 h in 2.5% glutaraldehyde, 2.5% formaldehyde in 0.075 M cacodylate buffer. The pieces then were embedded in LR white plastic and serially sectioned with an ultra-microtome at 2 μm. Sections were stained with toluidine blue. Follicles in every 20th plastic section were staged and counted to assess follicle activation. This sampling protocol ensures that no follicle is counted twice.

Microarray analysis.

Total RNA in cultured cortical pieces (n = 3 fetuses) was first extracted using Trizol reagent (Life Technologies), followed by a second purification using the RNeasy Mini Kit (QIAGEN, Valencia, CA). The quality of RNA was examined using an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA). Double-stranded cDNA was synthesized from mRNA using cDNA synthesis kit according to the manufacturer's instructions (Affymetrix, Santa Clara, CA) and then served as a template for subsequent in vitro transcription (IVT) reactions to produce biotin-labeled complementary RNA (cRNA) using the GeneChip IVT labeling kit (Affymetrix). Biotin-labeled cRNA was then hybridized to the GeneChip Bovine Genome Array (Affymetrix; n = 6, 1 per treatment per fetus) and scanned by GeneArray 3000 scanner (Affymetrix).

Gene ontology, canonical pathway, and functional network analyses.

Gene ontology, canonical pathway, and functional network analyses were generated through the use of Ingenuity Pathways Analysis (IPA; Ingenuity Systems, http://www.ingenuity.com). The analyses were performed to explore, understand, and discover molecular interaction networks in gene expression data that contribute to the regulation of follicle activation.

Real-time PCR analysis.

Real-time PCR was used to 1) validate the results for four transcripts that were differentially expressed in the microarray analysis and 2) examine levels of mRNA for PTEN because the IPA results suggested its involvement in insulin-induced follicle activation in cattle. In brief, total RNA in freshly isolated cortical pieces and pieces of ovarian cortex cultured with TS+ or ITS+ for 2 days (different pieces from those used for the microarray analysis; n = 3 fetuses) was isolated and then converted into cDNA using random primers and Superscript II reverse transcriptase (Life Technologies). Using the same amount of cDNA, we performed real-time PCR in duplicate with an ABI 7300 series real-time PCR machine (Applied Biosystems, Foster City, CA) using SYBR Green Master Mix (Eurogenetec, Seraing, Belgium). Primers were designed using Oligoperfect Designer (Life Technologies), and sequences for bovine cell division cycle protein 20 (CDC20) were as follows: forward, 5′-GTC TGA CCA TGA GCC CAG AA-3′; reverse, 5′-GGT GGA TGA GGC TGC TTT T-3′. Sequences for bovine hydroxysteroid (17-beta) dehydrogenase 11 (HSD17β11) were: forward, 5′-AAG AGA AAA TCA GTC ACC GGA G-3′; reverse, 5′-TGG CTG TTT CCT CAA GTC C-3′. Sequences for bovine ubiquitin-conjugating enzyme E2C (UBE2C) were: forward, 5′- ACG GTG AAG TTC CTC ACA CC-3′; reverse, 5′-AGA ACA CAG GGA GAG CTG GA-3′. Sequences for bovine junctional adhesion molecule 1 (JAM1) were: forward, 5′-CCA AGC TGT CCT GCT CCT AC-3′; reverse, 5′-GAA TGG AAG GTG ATG CCA GT-3′. Sequences for bovine phosphatase and tensin homolog (PTEN) were: forward, 5′-CAT AAC GAT GGC TGT GGT TG-3′; reverse, 5′-CCC CCA CTT TAG TGC ACA GT-3′. Levels of bovine RNA polymerase II mRNA were used as the reference mRNA for normalization, because it was reported to have the most consistent expression in different tissues (34). Radonic et al. (34) compared the mRNA transcription profiles of 13 putative reference genes, including glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and β-actin, in 16 different tissues and in cultured cells and showed that “classical” reference genes are unsuitable, whereas RNA polymerase II was the mRNA with the most constant expression. Sequences for bovine RNA polymerase II were: forward, 5′-CTT CCA ACA AGG CTT TCG AG-3′; reverse, 5′-GCT CAG CAC ATC TTT CAC CA-3′. Amplification reactions were performed in duplicate for CDC20, HSD17β11, UBE2C, JAM1, PTEN, and RNA polymerase II for every cDNA sample (n = 3 fetuses) for 40 cycles in 96-well real-time PCR plates. Relative mRNA levels were quantified, and we calculated fold changes by comparing the mRNA expression in ovarian cortical tissues cultured with ITS+ vs. those with TS+.

Western blot analysis.

Total protein from pieces of ovarian cortex (24–30 pieces/treatment/fetus; n = 3 fetuses) cultured with TS+ or ITS+ was extracted with RIPA buffer, and concentrations were determined by Bradford protein assay. Equal amounts of protein (∼25 μg) were loaded on a 10% sodium dodecyl sulfate-polyacrylamide gel, separated electrophoretically, and transferred to a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk in TBS-T buffer (0.1% Tween-20 in Tris-buffered saline) for 1 h at room temperature to reduce the nonspecific binding. Then the membrane was incubated with rabbit anti-human PTEN and rabbit anti-human β-actin polyclonal antibody (Cell Signaling, Danvers, MA), followed by a horseradish peroxidase-conjugated anti-rabbit secondary antibody (Santa Cruz Biotechnology). Immunoreactivity was detected by enhanced chemiluminescence (Santa Cruz Biotechnology). The membrane was exposed to X-ray film; the film was scanned, and band density was determined with Kodak ID image analysis software (Kodak). Levels of β-actin protein were used as the loading control for normalization.

Immunohistochemistry.

Ovaries from bovine female fetuses (5–8 mo gestation, n = 4) were fixed in Bouin's solution and then processed for embedding in paraffin. Sections (8 μm) were cut, deparaffinized in xylene, hydrated in an ethanol series, and then microwaved for 15 min to enhance antigen retrieval. Endogenous peroxidases were blocked by incubating the sections with 3% H₂O₂ for 10 min. Rabbit anti-human FOXO3a polyclonal antibody (Novus Biologicals) was used for immunohistochemical detection of FOXO3a. Each section was incubated with 10 μg/ml primary antibody overnight at 4°C. The biotinylated secondary antibody was then applied for 1 h at room temperature. The antigen-antibody-enzyme complex was detected and visualized with the Histostain-SP kit (Zymed Laboratories). Negative controls were sections adjacent to the sections being tested, but with the primary antibody replaced with nonimmune serum, and were included in each run.

Statistical analysis.

Mean numbers of primordial and primary follicles per section were calculated for freshly isolated and cultured cortical pieces. Means were compared by two-way ANOVA (with treatment and animal as the factors). When a significant P value was obtained with ANOVA, differences among individual means were tested by Duncan's multiple range test. A total of six microarray chips (1 per treatment per fetus, n = 3 fetuses) were analyzed. For microarray analysis, Affymetrix GeneChip Operating Software was used to analyze raw array data of pairs of TS+ vs. ITS+ samples to obtain signal, signal to log ratio values, change call, and associated P values, using the TS+-treated sample in each pair as baseline. Paired t-tests were performed, and an arbitrary P value of 0.05 was chosen to generate an initial set of differentially expressed genes. This set was further analyzed and prioritized using a combination of a more stringent P value (0.001) and fold change. Real-time PCR and Western blot data were analyzed by one-way ANOVA or t-test, respectively.

RESULTS

Morphological analysis of cortical pieces cultured with TS+ and ITS+.

In freshly isolated cortical pieces (day 0 controls), most follicles were at the primordial stage as reported previously (43). After 2 days in culture with ITS+, there was a significant increase in the number of primary follicles and a concomitant decrease in the number of primordial follicles compared with day 0 controls; the majority of follicles were at the primary stage with some at the primordial stage, as expected (Fig. 1A). In contrast, in pieces cultured for 2 days with TS+, which is identical to ITS+ but without insulin, most follicles remained at the primordial stage and there was no increase in the number of primary follicles (Fig. 1A). The cultured cortical pieces looked healthy with or without insulin (Fig. 1B). There was no difference in the percentage of healthy follicles (generally >90%), indicating that culturing bovine cortical pieces without insulin (TS+) maintained follicular health as expected (13). Therefore, the morphometric analysis confirmed the expected enrichment for primordial follicles in freshly isolated cortical pieces and pieces cultured with TS+ and the enrichment for primary follicles in cortical pieces cultured with ITS+.

Fig. 1.

A: numbers of primordial and primary follicles (± SE) in fetal bovine ovarian cortical pieces after 0 or 2 days in culture with medium supplemented with ITS+ or TS+. ITS+: insulin-transferrin-selenious acid + BSA and linoleic acid; TS+: identical to ITS+ but without insulin. Means with no common letters (a, b; x, y) within follicle type across treatments are different (*P < 0.05, n = 6 cultures, 2 from each of 3 fetuses). B: representative photomicrographs showing that freshly isolated (day 0) and cultured cortical pieces contain predominantly one class of follicles. Day 0, predominantly primordial follicles (arrow) with an oocyte surrounded by a single layer of flattened granulosa cells; after 2 days of culture with TS+, predominantly primordial follicles (arrow) and after 2 days of culture with ITS+, predominantly primary follicles (arrow) with an oocyte surrounded by a single layer of cuboidal granulosa cells.

Affymetrix GeneChip analysis.

Messenger RNAs isolated from cultured cortical pieces were analyzed using Affymetrix Bovine Genome GeneChips to investigate their global gene expression profile. The Affymetrix Bovine Genome GeneChip contains 24,027 probe sets and can monitor expression of ∼23,000 bovine transcripts, according to the manufacturer (Affymetrix). In the present study, 15,665 or 65% of the total probe sets were detected in cortical pieces cultured with TS+ or ITS+. The microarray data are publicly available through the Gene Expression Omnibus website (http://www.ncbi.nlm.nih.gov/geo/) with the accession number GSE69094.

The global gene expression profiles of cortical pieces cultured with TS+ or ITS+ were compared. Differentially expressed genes were identified in each paired comparison using an initial criterion of a t-test P value < 0.05. Further analysis of that set with a more stringent P value of 0.001 identified the most highly differentiated transcripts, which were then prioritized in combination with the fold change for each transcript. With this approach, comparison between pieces treated with TS+ and ITS+ generated 158 differentially expressed transcripts. Among 158 significantly altered transcripts, relative to TS+, 90 transcripts were upregulated in cortical pieces treated with ITS+ (Table 1). The upregulated transcripts include insulin-induced gene 1 (INSG1) and galectin 1 (LGALS1), known to be important for cell proliferation and growth, stearoyl-coenzyme A desaturase 1 (SCD1), which is involved in metabolism, and cell division cycle protein 20 (CDC20) and mitotic spindle-associated protein P126, which are involved in regulating the cell cycle. The mRNA for ATP synthase (ATP5G1), which is involved in ATP metabolic processes, was also upregulated in association with insulin-induced follicle activation. Relative to TS+, 68 transcripts were downregulated in pieces treated with ITS+ (Table 2), including transcripts like B-cell lymphoma 6 (BCL-6) and decay-accelerating factor (DAF), which are involved in cell survival/death and negative regulation of cell proliferation. Transcripts encoding high mobility group box transcription factor 1 (HBP1) and frizzled homolog 1 (FZD1) were also downregulated.

Table 1.

Transcripts upregulated during bovine follicle activation in vitro (ITS+ vs. TS+)

Affymetrix ID	Fold (log2)	P Value	Gene Name	Gene Symbol
Bt.4798.1.S2_at	2.02	0.000012	stearoyl-coenzyme A desaturase	>SCD
Bt.4733.1.S1_at	1.86	0.000513	phosphoglycerate dehydrogenase	PHGDH
Bt.17179.1.S1_at	1.53	0.000641	similar to NP_766093.1 hypothetical protein LOC212427
Bt.27910.1.S1_at	1.45	0.000036	chromosome 20 (human) open reading frame 54	C20orf54
Bt.13588.1.A1_at	1.44	0.000369	similar to phosphoserine aminotransferase isoform 1	LOC533044
Bt.10648.1.S1_at	1.42	0.000818	similar to Ribonucleoside-diphosphate reductase M2 chain	LOC508167
Bt.2725.1.S1_at	1.32	0.000364	similar to Ubiquitin-conjugating enzyme E2 C (Ubiquitin-protein ligase C)	MGC134111
Bt.2112.1.S1_at	1.32	0.000094	Transcribed locus
Bt.11587.3.A1_a_at	1.30	0.000270	similar to Mitotic spindle associated protein p126	LOC504585
Bt.21523.1.S1_at	1.29	0.000587	centromere protein-A	CENP-A
Bt.18776.1.S1_at	1.26	0.000088	KIAA0101 protein	KIAA0101
Bt.16065.1.S1_at	1.23	0.000985	similar to cytoskeleton-associated protein 2	LOC515249
Bt.24614.1.S1_at	1.18	0.000321	similar to insulin-induced gene 1 isoform 1	LOC511899
Bt.13588.2.S1_at	1.17	0.000863	similar to phosphoserine aminotransferase isoform 2	LOC617679
Bt.21087.1.A1_at	1.09	0.000741	similar to transcription factor 19	LOC514216
Bt.9767.1.S1_a_at	1.08	0.000031	similar to XP_861944.1 squalene monooxygenase isoform 2
Bt.1181.1.S1_at	1.07	0.000960	similar to extra spindle poles like 1	MGC137029
Bt.17939.1.S1_at	1.05	0.000846	similar to XP_531892.2 PREDICTED: similar to Lamin B1
Bt.3764.1.S1_at	1.05	0.000404	similar to isopentenyl-diphosphate delta isomerase	MGC139111
Bt.24997.1.S1_at	1.03	0.000206	similar to XP_127466.2 PREDICTED: hypothetical protein
Bt.22763.2.S1_a_at	1.01	0.000078	3-hydroxy-3-methylglutaryl-coenzyme A synthase 1	HMGCS1
Bt.639.1.S1_at	0.98	0.000175	similar to Cell division cycle protein 20 homolog	LOC515376
Bt.20617.1.S1_at	0.97	0.000459	similar to nerve growth factor receptor (TNFRSF16) associated protein 1	LOC516056
Bt.22854.2.S1_at	0.96	0.000155	fatty acid synthase	FASN
Bt.3786.1.S1_a_at	0.94	0.000759	similar to 3-beta-hydroxysteroid-delta	MGC128402
Bt.4585.1.S1_at	0.93	0.000630	acetyl-Coenzyme A acetyltransferase 2	ACAT2
Bt.26700.1.S1_at	0.90	0.000894	histamine N-methyltransferase	HNMT
Bt.8964.2.S1_a_at	0.87	0.000635	CDNA clone MGC:128551 IMAGE:7985190
Bt.4198.1.S1_at	0.87	0.000092	similar to deoxythymidylate kinase (thymidylate kinase)	LOC506946
Bt.7331.1.S2_at	0.86	0.000659	aurora kinase B	AURKB
Bt.2712.1.S1_at	0.84	0.000606	serine (or cysteine) proteinase inhibitor	SERPINA5
Bt.20893.2.S1_at	0.81	0.000147	pyridoxine 5′-phosphate oxidase	PNPO
Bt.1242.1.S1_at	0.81	0.000282	ATP synthase, H+-transporting, isoform 1	ATP5G1
Bt.8003.1.A1_at	0.79	0.000103	hypothetical LOC513406	MGC128333
Bt.621.1.S1_at	0.79	0.000737	cytochrome P450, family 51, subfamily A, polypeptide 1	CYP51
Bt.5412.1.S1_at	0.78	0.000068	branched chain keto acid dehydrogenase E1	BCKDHB
Bt.27599.1.A1_at	0.77	0.000223	similar to XP_854777.1 Probable ergosterol biosynthetic protein 28
Bt.16094.1.S1_at	0.77	0.000586	similar to Enhancer of zeste homolog 2 (ENX-1)	LOC509106
Bt.27248.1.A1_at	0.75	0.000574	Transcribed locus
Bt.5159.1.A1_s_at	0.74	0.000244	mevalonate kinase (mevalonic aciduria)	MVK
Bt.25837.1.S1_at	0.74	0.000264	similar to CG15792-PA, isoform A	LOC515158
Bt.21724.2.S1_a_at	0.74	0.000543	farnesyl-diphosphate farnesyltransferase 1	FDFT1
Bt.21724.1.S1_at	0.73	0.000673	farnesyl-diphosphate farnesyltransferase 1	FDFT1
Bt.1701.2.A1_a_at	0.73	0.000387	similar to Coatomer zeta-2 subunit (Zeta-2 coat protein) (Zeta-2 COP)	LOC616222
Bt.5472.1.S1_at	0.72	0.000098	lectin, galactoside-binding, soluble, 1 (galectin 1)	LGALS1
Bt.21833.1.S1_at	0.70	0.000558	similar to AYP1 protein	MGC137048
Bt.3778.1.S1_at	0.69	0.000867	similar to IFITM5	LOC526461
Bt.26777.1.S1_at	0.69	0.000372	Transcribed locus
Bt.28502.1.S1_at	0.68	0.000666	similar to Protein C20orf129	LOC508561
Bt.5723.1.S1_at	0.67	0.000719	mitochondrial NADH:ubiquinone oxidoreductase B14.7	b14.7
Bt.10166.1.A1_at	0.67	0.000287	Transcribed locus
Bt.20163.3.S1_a_at	0.67	0.000412	peptidylprolyl isomerase-like 1	PPIL1
Bt.23393.1.S1_at	0.66	0.000521	similar to Sodium/potassium-transporting ATPase alpha-2 chain precursor	LOC515161
Bt.6162.1.S1_at	0.64	0.000756	similar to putative c-Myc-responsive	LOC613560
Bt.21406.1.A1_at	0.62	0.000519	similar to XP_853222.1 cytochrome b5 reductase b5R.2 isoform 1
Bt.29655.1.S1_a_at	0.62	0.000446	similar to UPF0287 protein DC13	MGC137594
Bt.12465.1.A1_at	0.60	0.000393	similar to ribonucleotide reductase M1	LOC505537
Bt.4661.1.S1_at	0.56	0.000776	similar to Cell division protein kinase 4 (Cyclin-dependent kinase 4)	MGC133903
Bt.10503.1.S1_at	0.55	0.000574	flap structure-specific endonuclease 1	FEN1
Bt.2195.1.A1_at	0.54	0.00001	similar to Guanine nucleotide-binding protein G(I)/G(S)/G(O) gamma-10 subunit	MGC128212
Bt.1199.1.S1_at	0.50	0.00050	similar to ER lumen protein retaining receptor 3	LOC509687
Bt.22297.1.S1_at	0.47	0.000196	Transcribed locus
Bt.2150.1.S1_at	0.46	0.000007	similar to G-protein signaling modulator 2	LOC513654
Bt.3898.1.S1_at	0.45	0.000660	isocitrate dehydrogenase 3 (NAD+) alpha	IDH3A
Bt.8351.1.S1_at	0.41	0.000729	similar to CG13926-PA	LOC504867
Bt.15628.1.S1_at	0.40	0.000918	Transcribed locus
Bt.2594.1.S1_at	0.39	0.000120	splicing factor, arginine/serine-rich 2	SFRS2
Bt.20823.1.A1_at	0.37	0.000141	Transcribed locus
Bt.17778.1.A1_at	0.36	0.000185	similar to 39S ribosomal protein L40	LOC515650
Bt.2255.1.S1_at	0.35	0.000278	IMP (inosine monophosphate) dehydrogenase 1	IMPDH1
Bt.21073.1.A1_at	0.35	0.000747	Transcribed locus
Bt.1479.1.S1_at	0.34	0.000982	similar to cytochrome b, ascorbate dependent 3	LOC615893
Bt.265.1.S1_at	0.34	0.000512	ubiquinol-cytochrome c reductase, Rieske iron-sulfur polypeptide 1	UQCRFS1
Bt.3960.2.S1_at	0.33	0.000385	voltage-dependent anion channel 1	VDAC1
Bt.5963.1.S1_at	0.30	0.000986	hypothetical protein LOC614320	MGC128806
Bt.27018.1.S1_at	0.30	0.000399	similar to Serine/threonine-protein kinase 16	MGC137500
Bt.20823.2.S1_at	0.30	0.000011	Transcribed locus
Bt.20441.1.A1_at	0.28	0.000141	Transcribed locus
Bt.22.1.S1_at	0.26	0.000148	protein phosphatase 1	PPP1R8
Bt.21997.1.S1_at	0.26	0.000548	PPM1K protein	PPM1K
Bt.18863.1.S1_at	0.25	0.000673	similar to Probable dolichyl pyrophosphate Glc1Man9GlcNAc2 alpha-1,3-glucosyltransferase	LOC538731
Bt.4253.3.S1_a_at	0.24	0.000862	similar to polymerase (RNA) II (DNA directed) polypeptide D isoform 1	LOC617422
Bt.16111.1.S1_at	0.21	0.000227	Transcribed locus
Bt.26687.1.S1_at	0.20	0.000013	similar to achalasia, adrenocortical insufficiency, alacrimia (Allgrove, triple-A)	LOC506561
Bt.4637.1.A1_a_at	0.18	0.000313	cytochrome c oxidase subunit VIII, heart	COX8
Bt.11219.1.A1_at	0.15	0.000267	similar to Bifunctional 3-phosphoadenosine 5-phosphosulfate synthethase 1	MGC127389
Bt.4503.1.S2_at	0.08	0.000460	mitochondrial carrier homolog 2	Mtch2

ITS+, insulin-transferrin-selenious acid + BSA and linoleic acid; TS+, identical to ITS+ but without insulin.

Table 2.

Transcripts downregulated during bovine follicle activation in vitro (ITS+ vs. TS+)

Affymetrix ID	Fold (log2)	P Value	Gene Name	Gene Symbol
Bt.20397.1.S1_at	−2.04	0.000027	similar to small inducible cytokine B14 precursor	MGC127163
Bt.7594.1.S1_at	−1.27	0.000433	similar to Spermidine/spermine N(1)-acetyltransferase 1 (SSAT) (SSAT-1)	MGC127572
Bt.13289.2.S1_at	−1.25	0.000272	Transcribed locus
Bt.11993.1.S1_at	−1.20	0.000998	similar to 17-beta-hydroxysteroid dehydrogenase 11 (17-beta-HSD 11)	MGC137132
Bt.16583.1.S1_at	−1.19	0.000242	similar to Voltage-dependent calcium channel gamma-like subunit	MGC126930
Bt.22083.2.S1_at	−1.14	0.000023	Ribosomal protein L5	RPL5
Bt.24417.2.S1_at	−1.12	0.000430	Transcribed locus, decay accelerating factor 1 (Mus musculus)
Bt.13798.1.S1_at	−1.09	0.000026	strongly similar to NP_001007646.1 hypothetical protein
Bt.21938.1.S1_at	−1.03	0.000549	Transcribed locus
Bt.7670.1.A1_at	−1.02	0.000662	similar to B-cell lymphoma 6 protein	LOC539020
Bt.7307.2.A1_a_at	−0.98	0.000996	Transcribed locus
Bt.22117.1.S1_at	−0.97	0.000188	similar to NP_570134.1 alpha 1 type XVII collagen
Bt.3253.1.A1_at	−0.96	0.000324	hypothetical protein LOC617871	LOC617871
Bt.4768.1.S2_at	−0.90	0.000539	junctional adhesion molecule 1	JAM1
Bt.20431.2.S1_at	−0.89	0.000882	similar to Max interacting protein 1	LOC614509
Bt.12412.1.S1_at	−0.87	0.000130	similar to high mobility group box transcription factor 1	MGC128984
Bt.16058.1.A1_at	−0.87	0.000336	Transcribed locus
Bt.16436.3.A1_at	−0.87	0.000440	hypothetical protein LOC615685	LOC615685
Bt.28393.1.S1_at	−0.86	0.000979	decay-accelerating factor 1	DAF
Bt.3191.1.A1_at	−0.85	0.000781	similar to DRE1 protein	LOC533510
Bt.29945.1.S1_at	−0.81	0.000180	Transcribed locus
Bt.28642.1.A1_at	−0.81	0.000305	Transcribed locus
Bt.13224.1.A1_at	−0.81	0.000820	similar to cytochrome P450, family 4, subfamily v	MGC127339
Bt.26415.1.A1_at	−0.81	0.000081	Transcribed locus
Bt.24258.1.S1_at	−0.78	0.000885	similar to XP_533481.2 mannosidase, alpha, class 1A
Bt.16058.2.S1_at	−0.77	0.000665	Transcribed locus
Bt.12472.1.S1_at	−0.76	0.000133	glutamate-ammonia ligase (glutamine synthase)	GLUL
Bt.20072.1.S1_at	−0.74	0.000732	Transcribed locus
Bt.26635.2.S1_at	−0.73	0.000161	frizzled homolog 1 (Drosophila)	FZD1
Bt.19107.1.S1_at	−0.70	0.000553	Transcribed locus
Bt.8691.1.S1_at	−0.70	0.000620	Similar to calcium-regulated heat-stable protein (24kD)	MGC137010
Bt.21096.1.S1_at	−0.69	0.000928	similar to XP_548991.2 GRIP1-associated protein 1 isoform 1
Bt.6987.1.S1_at	−0.59	0.000526	Transcribed locus
Bt.19107.2.A1_at	−0.57	0.000376	Transcribed locus
Bt.1373.1.A1_at	−0.56	0.000238	similar to Y55F3AM.9	MGC128936
Bt.9238.1.S1_at	−0.55	0.000672	solute carrier family 38, member 2	SLC38A2
Bt.28617.1.S1_at	−0.53	0.000049	similar to Erythrocyte band 7 integral membrane protein	LOC617026
Bt.13370.1.A1_at	−0.51	0.000280	Transcribed locus
Bt.1202.1.S1_at	−0.51	0.000008	similar to MAF1 protein	MGC127274
Bt.25345.1.A1_at	−0.49	0.000069	derivatives expressed transcript 2
Bt.9268.1.S1_at	−0.49	0.000713	similar to MOCO sulphurase C-terminal domain containing 1	LOC537028
Bt.23893.1.A1_s_at	−0.48	0.000113	Transcribed locus
Bt.16211.2.S1_at	−0.47	0.000720	Transcribed locus
Bt.9176.1.A1_at	−0.45	0.000235	Ribosomal protein L5	RPL5
Bt.413.1.S1_at	−0.45	0.000565	alpha-galactosyltransferase 1 (glycoprotein)	GGTA1
Bt.24307.1.A1_at	−0.44	0.000889	Transcribed locus
Bt.4489.1.S1_at	−0.44	0.000265	kelch domain containing 3	KLHDC3
Bt.22665.1.S1_at	−0.42	0.000698	Transcribed locus
Bt.911.1.S1_at	−0.42	0.000694	similar to XP_533192.2
Bt.23857.2.S1_at	−0.42	0.000285	Transcribed locus
Bt.23859.1.S1_at	−0.41	0.000593	Transcribed locus
Bt.20064.2.S1_at	−0.41	0.000538	similar to serologically defined colon cancer antigen 33	LOC516106
Bt.22000.1.A1_at	−0.41	0.000756	cysteine-rich, angiogenic inducer, 61	CYR61
Bt.1690.1.S1_at	−0.34	0.000389	similar to ring finger protein 130	LOC525735
Bt.12249.1.S1_at	−0.33	0.00003	similar to jumonji domain containing 1A	LOC536073
Bt.9936.1.S1_at	−0.32	0.000125	similar to NP_056579.1 Greb1 protein
Bt.12811.1.S1_a_at	−0.32	0.000293	uveal autoantigen with coiled-coil domains and ankyrin repeats	UACA
Bt.4574.1.S1_at	−0.30	0.000369	CDNA clone MGC:128349 IMAGE:7947341
Bt.20602.1.S1_at	−0.29	0.000472	ubiquitin-conjugating enzyme E2B (RAD6 homolog)	UBE2B
Bt.11326.1.S1_at	−0.29	0.000637	similar to CG8580-PA, isoform A	LOC614292
Bt.6771.1.S1_at	−0.27	0.000264	similar to EH domain binding protein 1	LOC533149
Bt.11714.1.S1_at	−0.26	0.000781	Solute carrier family 25 member 5	SLC25A5
Bt.24579.1.S1_at	−0.25	0.000288	similar to Mediator of RNA polymerase II transcription subunit 31	LOC532868
Bt.934.1.A1_at	−0.19	0.000795	similar to PHD finger protein 20	LOC529591

Identification of biological pathways and functions associated with follicle activation.

To further understand the biological and molecular functions represented by those 158 transcripts significantly altered during insulin-induced follicle activation, we performed an IPA functional analysis. The functional analysis identified the biological functions that were most significant in the data set. There were 59 molecular functions identified by this analysis as significantly associated with the altered transcripts. The most significant function associated with these transcripts is cellular growth and proliferation (Fig. 2). Other top functions associated with insulin-induced follicle activation include cell death, connective tissue disorder, cellular function and maintenance, cell cycle, cell-to-cell signaling and interaction, cellular assembly and organization, and reproductive system development and function. The top 17 functions are shown in Fig. 2.

Fig. 2.

Biological functions identified by Ingenuity Pathways Analysis that are associated with the differentially expressed transcripts identified during insulin-induced follicle activation in vitro. Numbers to the right of the bars indicate the number of transcripts in that category. The vertical line crossing all the bars indicates the threshold of significance (P = 0.05), bars to the right of this line have a P value < 0.001.

To further refine analysis of the 158 significantly altered transcripts associated with insulin-induced follicle activation, we next investigated how these transcripts interact biologically by mapping them to genetic networks available in the Ingenuity database. These networks identified functional relationships among gene products based on known biological interactions in the literature. Analysis resulted in a total of seven networks (Fig. 3). In agreement with gene function analysis, the most significant network (network 1) identified for follicle activation includes transcripts involved in cellular growth and proliferation. The second most significant network (network 2) includes genes involved in cell cycle/apoptosis. Other networks include genes involved in cancer/cell morphology, cell-to-cell signaling and interaction, cellular growth and proliferation, cellular assembly and organization, and carbohydrate metabolism/energy production (Fig. 3).

Fig. 3.

Ingenuity Pathways Analysis (IPA) implicated 7 biological networks associated with differentially expressed transcripts identified during insulin-induced follicle activation in vitro. The sizes of the pieces of pie represent the score for each network, which was computed by IPA, according to the fit of that network to the set of differentially expressed transcripts (i.e., the bigger the piece, the better the fit).

One benefit of IPA is that the application focuses not only on user-identified genes (in the present study significantly altered transcripts associated with insulin-induced follicle activation), but also on many additional genes that may be biologically relevant to user-defined genes. Thus, this analysis allows researchers to identify relationships between seemingly disparate changes in gene expression. Messenger RNA for UBE2C was significantly upregulated during insulin-induced follicle activation (Fig. 4), and network 1 from the IPA suggests UBE2C interacts with the tumor suppressor PTEN, and PTEN can act on upregulated ATP5G1 (Fig. 4). In addition, LGALS1 mRNA was upregulated during follicle activation, and it is predicted to interact with proteins encoded by spindle genes (SPN) in network 1. In network 5 (cellular development, hematological system development/function), Kelch-like 24 mRNA (KLHL24), which was downregulated during insulin-induced follicle activation, is suggested to be negatively regulated by kit ligand (KITLG), which can stimulate follicle activation in rodents and cattle (13, 33).

Fig. 4.

The most prominent biological network associated with insulin-induced follicle activation in vitro, the network of cell proliferation and growth, which was generated by IPA. This network includes upregulation of the UBE2C transcript and, thus, implicates the PTEN/FOXO3a pathway (large oval circle) in insulin-induced follicle activation in vitro. Various shapes represent the functional class of the mRNA products. For example, squares represent cytokines/growth factors, horizontal ovals represent transcription factors, and diamonds represent enzymes. The intensity of shading indicates the degree of expression. A solid line indicates a direct interaction and a dashed line indicates an indirect interaction.

Expression of CDC20, HSD17β11, UBE2C, and JAM1 mRNA in bovine ovarian cortex during follicle activation.

Real-time RT-PCR was used to validate some of the findings from the microarray. Abundance of mRNA for CDC20 and UBE2C, cell cycle regulators, was more than 9 or 11 times higher, respectively, in cortical pieces treated with ITS+ than those with TS+ (Fig. 5). This is in accordance with microarray results showing that transcripts encoding CDC20 and UBE2C were upregulated during insulin-induced follicle activation (Fig. 5 and Table 1). Abundance of mRNA for HSD17β11 and JAM1 was about three or six times lower, respectively, in pieces treated with ITS+, which is consistent with microarray data showing that HSD17β11 and JAM1 mRNA was downregulated during insulin-induced follicle activation (Fig. 5 and Table 1).

Fig. 5.

Comparison between fold change based on microarray (black bars) and real-time RT-PCR (white bars) analyses of levels of mRNA for CDC-20, HSD17β11, UBE2C, and JAM1 in fetal ovarian cortical pieces cultured for 2 days with control medium (TS+) or with insulin (ITS+), which induces activation of follicles (n = 3 fetuses). Values are expressed as fold change (± SE) relative to cortical pieces cultured with TS+. For all transcripts examined, the 2 methods showed a similar direction of the change in the presence of insulin.

Expression of PTEN mRNA and protein in bovine ovarian cortex during follicle activation.

Based on the microarray data, IPA predicted that the PTEN pathway is involved in insulin-induced follicle activation (Fig. 4). Therefore, real-time PCR and Western blot were used to detect PTEN mRNA and protein, respectively. PCR analysis showed that the expression of PTEN mRNA was around 3.8- and 3.3-fold greater in freshly isolated cortical pieces (day 0) and pieces cultured for 2 days with control medium (TS+), respectively, compared with pieces cultured with insulin (ITS+, P < 0.05, Fig. 6A). Consistent with the PCR results, PTEN protein was detected by Western blot in bovine ovarian cortical pieces, and levels of PTEN were 20-fold higher in control cortical pieces cultured with TS+ than those with ITS+ (P < 0.05, Fig. 6B).

Fig. 6.

Results of real-time PCR analyses of PTEN mRNA (A) and Western blot analyses of PTEN protein (B) in bovine ovarian cortical pieces cultured for 2 days in vitro without (TS+) or with insulin (ITS+), which induces activation of follicles (n = 3 fetuses). Values are expressed as fold-change (± SE) from TS+ (set at 1). Messenger RNA from freshly isolated cortical pieces (day 0 of culture) is also included in A. Significant differences (P < 0.05) are indicated by no common letters (a, b) in A and by an asterisk in B.

Immunolocalization of FOXO3a protein in bovine fetal ovaries.

Immunolocalization of FOXO3a, a downstream effector of PTEN signaling, was performed on paraffin sections from fresh bovine fetal ovaries (5–8 mo gestation) of the same age as the fetal ovaries used for microarray. Areas with primarily either primordial or primary follicles in ovarian sections were chosen to show the differences in FOXO3a immunostaining. Staining for FOXO3a was observed in the cytoplasm of all oogonia and oocytes (Fig. 7A). Some granulosa cells of primordial and primary follicles also showed immunoreactivity for FOXO3a. Interestingly, there was strong staining for FOXO3a in the nuclei of most oogonia and many oocytes of primordial follicles (Fig. 7A). However, no nuclear staining was detected in oocytes of primary follicles (Fig. 7B). This suggests that nuclear-cytoplasmic shuttling and degradation of FOXO3a protein occurs during follicle activation in cattle.

Fig. 7.

Immunolocalization of FOXO3a in bovine fetal ovaries (5–8 mo gestation). A: strong brown staining for FOXO3a was detected in the nuclei of most of oogonia (arrowhead) and many oocytes of primordial follicles (arrows). B: no nuclear staining for FOXO3a was present in oocytes of primary follicles (arrow). Some granulosa cells of primordial and primary follicles show FOXO3a immunoreactivity. Bars = 40 μm. Results are typical of 4 fetuses.

DISCUSSION

The regulation of follicle activation is still largely a mystery. The addition of particular hormones/growth factors can stimulate or inhibit activation in vitro, but the molecular changes that mediate the initiation of growth and development of both the oocyte and surrounding granulosa cells are not understood. Techniques like microarray offer the possibility of examining levels of thousands of mRNA transcripts simultaneously. In mammalian species of practical interest, such as humans and domestic animals, follicle formation and acquisition of the capacity to activate occur in the female fetus over a period of weeks. This makes the application of molecular techniques especially challenging. Our experimental system involved the enrichment of fetal ovarian cortical pieces for primordial or primary follicles by culture with control medium or with insulin, one of the factors that can stimulate bovine follicle activation. Activation occurs within 48 h, and microarray analysis of tissue retrieved at that time revealed a host of differences in transcripts in ovarian cortex enriched for primary vs. primordial follicles. The transcript encoding UBE2C was upregulated in cortical pieces enriched for primary follicles and was linked by IPA with PTEN, an upstream regulator of FOXO3a. These latter two factors were identified as inhibitors of activation by knockout studies with mice (4, 35), and our follow-up studies showed that a dramatic decrease in PTEN and in nuclear FOXO3a accompanied bovine follicle activation. This suggests the intriguing hypothesis that upregulation of UBE2C, and thus the ubiquitin proteasome system, leads to degradation of PTEN and less nuclear accumulation of FOXO3a and, hence, to follicle activation in cattle.

A total of 158 differentially expressed transcripts were associated with the transition from resting primordial to primary follicles. Several upregulated transcripts encode metabolic enzymes, including SCD1, a key regulator of energy metabolism. This is consistent with the greater metabolic activity of primary follicles vs. resting primordial follicles. Scd1 was also significantly upregulated during follicle activation in microarray studies in rats (22), but not in mice (46).

A number of transcripts encoding proteins promoting cell proliferation and growth, like INSIG1, galectin 1, and nerve growth factor receptor associated protein 1, were upregulated in association with the primordial to primary follicle transition. A knockout study indicated that nerve growth factor is involved in follicle activation and the primary-to-secondary follicle transition in mice (9). Also, transcripts encoding proteins that control the cell cycle, such as CDC20 and MAP16, were upregulated in bovine cortical tissue enriched for activated follicles. Consistent with those findings, functional analysis by IPA showed that top functions associated with transcripts differentially expressed during follicle activation are cellular growth and proliferation, cell cycle, cell-to-cell signaling and interaction, cellular movement, and cellular assembly and organization. These biological functions or processes are anticipated as the primordial follicles leave developmental arrest to initiate growth. Morphologically, follicle activation is characterized by the growth of the oocyte and a change in the shape of granulosa cells from flattened to cuboidal, and initiation of granulosa cell proliferation. These changes may involve numerous mechanisms including cell-cell signaling and interaction, cell cycle progression, and cellular assembly. It is possible that the activation stimulator we used, insulin, may have effects on ovarian cortical tissue that are not related to activation. However, microarray studies in mice (8, 46), rats (22), primates (1), and humans (26) similarly revealed differential expression of genes involved in the cell cycle and cell proliferation and growth. The transcripts that were upregulated are not exactly the same among studies, suggesting that species differences may exist.

Interestingly, transcripts encoding high mobility group box transcription factor 1 (HBP1) were downregulated in insulin-treated tissues, enriched for activated follicles. HBP1 is a high mobility group domain transcriptional repressor that regulates proliferation in differentiated tissues. Expression of the HBP1 gene leads to cell cycle arrest in vitro and in vivo (37, 38). Little is known about HBP1 during ovarian development, but Arraztoa et al. (1) reported expression of the HBP1 gene in primate primordial oocytes. Together with our findings, these data suggest that decreased HBP1 may induce the release of primordial follicles from cell cycle arrest and their entry into the growth phase to become primary follicles. In addition, HBP1 suppresses WNT/FZD1 signaling to prevent cell proliferation (36). WNT ligands and frizzled G protein-coupled receptors control cell fate, including embryonic development of the ovary in mice (17). In the present study, the transcript encoding frizzled homolog 1 (FZD1) was downregulated in ovarian cortex with many activated follicles. Wnt-4 is expressed in ovaries of neonatal mice between birth and day 5, when primordial follicles are being formed and some begin to develop into primary follicles (17). Together, these results suggest the involvement of WNT and FZD1, which may be regulated by HBP1, in bovine follicle activation. However, the exact roles of HBP1 and the WNT/FZD1 pathway in initiation of follicular growth remain to be investigated. The present study also showed that HSD17β11 mRNA was downregulated during follicle activation. HSD17β11 has been suggested to play a role in androgen metabolism during steroidogenesis (5), and thus, these data are consistent with previous studies showing androgens stimulate early follicular development (41, 45). Surprisingly, transcripts encoding BCL-6 and DAF, which are involved in cell survival/death, were also downregulated during insulin-induced follicle activation. However, our morphological study showed that cultured cortical pieces looked healthy with or without insulin, and there was no difference in the percentage of healthy follicles. Roles of these transcripts remain to be elucidated.

Kit ligand (KITLG) and its receptor c-kit (KIT) appear to be important for initiation of follicle growth in rodents and cattle (19, 27). Surprisingly, no significant changes in transcripts encoding KIT or KITLG were detected during follicle activation in the present study. Similarly, a microarray study in rats also showed no differences in levels of transcripts encoding KITLG in neonatal ovaries enriched for either primordial follicles or primary follicles (22). In contrast, an array study in mice demonstrated that Kit mRNA was upregulated in postnatal day 4 ovaries, which contain primordial and primary follicles, compared with day 2 ovaries, which have only primordial follicles (8). Reasons for the discrepancy between the microarray results and the experimental studies implicating KIT/KITLG in follicle activation in cattle and rats are not clear. One possibility might be the low level of expression of KIT and KITLG in bovine ovarian cortical pieces and whole rat neonatal ovaries. That hypothesis could be tested by examining their transcriptome profiling by the RNA-seq technique. RNA-seq is a newly developed technology and has some advantages over microarray, especially in detecting low-abundance transcripts and discovering new genes (47). Moreover, in the current study mRNA for KLHL24 was downregulated during bovine follicle activation, and our IPA analysis predicts that KITLG could act on KLHL24. The function of KLHL24 is not clear so far, so the physiological relevance of regulation of KLHL24 by KITLG remains to be elucidated.

We further analyzed differentially expressed transcripts associated with follicle activation in vitro by mapping them to IPA genetic networks to investigate how these transcripts interact biologically and to identify functional relationships between gene products, based on known biological interactions. Seven networks were identified. The top two networks with high significance scores are 1) cell proliferation and growth and 2) cell cycle and cell death. Interestingly, the network of cell proliferation and growth predicts that protein encoded by UBE2C mRNA, which was upregulated during follicle activation in the present study, can interact with PTEN. The modification of proteins with ubiquitin is an important cellular mechanism for targeting abnormal or short-lived proteins for degradation. The UBE2C gene encodes a member of the E2 ubiquitin-conjugating enzyme family, which is required for the destruction of mitotic cyclins and for cell cycle progression. PTEN is a member of the tyrosine phosphatase family and functions as a major negative regulator of PI3K action by removing the phosphate in the D3-phosphate group of phosphoinositide-3,4,5-triphosphate (24). The PI3K pathway is a fundamental signaling pathway for the regulation of cell proliferation, survival, migration, and cell cycle entry (3). Reddy et al. (35) reported that oocyte-specific deletion of Pten in mice caused massive activation of primordial follicles, demonstrating a critical inhibitory role of the PTEN/PI3K signaling cascade in the primordial to primary follicle transition in mice. Similarly, we showed that there was a wholesale activation of primordial follicles in bovine cortical pieces cultured with ITS+ and a concomitant dramatic decrease in PTEN mRNA and protein levels, suggesting that insulin induces follicle activation by decreasing PTEN in cattle. So far, several studies have shown that PTEN functions are regulated by ubiquitination (25, 39), but ovarian tissue was not examined. The transcript encoding UBE2C was upregulated in cortical pieces cultured with ITS+; therefore, we hypothesize that insulin induces upregulation of UBE2C, which then leads to degradation of PTEN and, hence, to follicle activation. Besides the PTEN pathway, the network of cell proliferation and growth also predicts that protein encoded by LGALS1 mRNA, whose expression was upregulated during follicle activation in the current study, can interact with proteins encoded by spindle (SPN) genes. SPN genes have been shown to have important functions in oogenesis in Drosophila (14), suggesting a potential role of LAGALS1/SPN protein interaction in early follicular development in larger mammals.

We also provided immunocytochemical evidence for nuclear to cytoplasmic translocation of FOXO3a during follicle activation in cattle. The forkhead box O (FOXO) transcription factors (FOXO1, FOXO3a, FOXO4, and FOXO6) are an important family of proteins because they regulate the expression of genes involved in apoptosis, cell cycle, cell differentiation, glucose metabolism, and other cellular functions (18). Studies from Castrillon's lab (4, 20) showed that Foxo3a−/− mice exhibit a distinct global activation of primordial follicles, and FOXO3a, lying downstream of PTEN, undergoes nuclear-cytoplasmic shuttling during primordial to primary follicle development. Moreover, differential expression of mRNA and protein for FOXO3a in mouse ovaries suggests that posttranscriptional regulation of FOXO3a protein stability is very important (20). Furthermore, FOXO protein stability is regulated by the ubiquitin proteasome system in other tissue types (40). How FOXO is regulated in the ovary is not clear. In the present study, our IPA analysis suggests that increased UBE2C may be responsible for the decreased level of FOXO3a, probably through the ubiquitin-proteasome system. The decrease in FOXO3a would then initiate follicle activation. In addition, since activation of ubiquitin is a process requiring ATP as an energy source, upregulation of the transcript encoding ATP5G1, identified by microarray in the present study, further supports the involvement of the ubiquitin-proteasome system in insulin-induced follicle activation.

Microarray technologies have also been used for gene expression profiling during early folliculogenesis in mice (8, 46), rats (22), primates (1), and humans (26). In all these studies, the transition from primordial to primary follicles is associated with changes in the expression of many cell cycle, metabolism, and growth-regulating genes. However, the specific mRNA transcripts that changed in each category in studies on mice, rats, cattle, primates, and humans are not exactly the same. Also, changes in the expression of some mRNAs, for example kit ligand and its receptor c-kit that appear to be important for initiation of follicle growth, were detected in one species but not in others (8, 22). Moreover, mRNAs regulating cell death were abundantly expressed during follicle activation in the current study and studies in mice (8, 46) and primates (1), but not in rats (22) and humans (26). There are several explanations for the differences. First, there may be species differences in the regulation of follicle activation. Second, differences in the changes in mRNA expression associated with follicle activation may be due to the use of different types of samples (whole ovaries vs. cortical pieces vs. isolated oocytes) and thus reflect the contributions of different proportions of ovarian cell types. Third, follicle activation was achieved differently. In our study, follicle activation was induced in vitro by insulin, which may also cause changes that are not involved in activation, whereas neonatal ovaries or oocytes isolated from primordial and primary follicles by laser capture microdissection were used in mouse and human studies, respectively (1, 8, 26, 46). Finally, the exact composition of gene chips between species/among studies was not the same.

In summary, our in vitro experimental model allowed the identification of 158 transcripts that were differentially expressed in tissues during insulin-induced bovine follicle activation in vitro by microarray. The differentially expressed transcripts code for proteins mainly involved in the processes of cellular growth and proliferation, cell cycle regulation, and metabolism. Several genes identified in the present study, especially HBP1, SCD1, and UBE2C, and pathways like WNT/FED1 and PTEN/FOXO3a provide interesting candidates for further investigation of mechanisms underlying follicle activation in cattle and other large mammals. Elucidation of theses mechanisms will advance our knowledge of this key transition in follicular development, especially in mammals of practical interest, like domestic animals and humans.

GRANTS

The research was supported by National Institute of Child Health and Human Development Grant HD-35168.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: M.Y.Y. and J.E.F. conception and design of research; M.Y.Y. performed experiments; M.Y.Y. analyzed data; M.Y.Y. and J.E.F. interpreted results of experiments; M.Y.Y. prepared figures; M.Y.Y. drafted manuscript; M.Y.Y. and J.E.F. edited and revised manuscript; M.Y.Y. and J.E.F. approved final version of manuscript.