Research Resource: Preovulatory LH Surge Effects on Follicular Theca and Granulosa Transcriptomes

Abstract

The molecular mechanisms that regulate the pivotal transformation processes observed in the follicular wall following the preovulatory LH surge, are still not established, particularly for cells of the thecal layer. To elucidate thecal cell (TC) and granulosa cell (GC) type-specific biologic functions and signaling pathways, large dominant bovine follicles were collected before and 21 hours after an exogenous GnRH-induced LH surge. Antral GCs (aGCs; aspirated by follicular puncture) and membrane-associated GCs (mGCs; scraped from the follicular wall) were compared with TC expression profiles determined by mRNA microarrays. Of the approximately 11 000 total genes expressed in the periovulatory follicle, only 2% of thecal vs 25% of the granulosa genes changed in response to the LH surge. The majority of the 203 LH-regulated thecal genes were also LH regulated in GCs, leaving a total of 57 genes as LH-regulated TC-specific genes. Of the 57 thecal-specific LH-regulated genes, 74% were down-regulated including CYP17A1 and NR5A1, whereas most other genes are being identified for the first time within theca. Many of the newly identified up-regulated thecal genes (eg, PTX3, RND3, PPP4R4) were also up-regulated in granulosa. Minimal expression differences were observed between aGCs and mGCs; however, transcripts encoding extracellular proteins (NID2) and matrix modulators (ADAMTS1, SASH1) dominated these differences. We also identified large numbers of unknown LH-regulated GC genes and discuss their putative roles in ovarian function. This Research Resource provides an easy-to-access global evaluation of LH regulation in TCs and GCs that implicates numerous molecular pathways heretofore unknown within the follicle.

In cattle the periovulatory transformation of the 17β-estradiol (E2)-producing dominant follicle during the follicular phase is accompanied by a profound morphologic reorganization of somatic cell layers (ie, granulosa and theca) to produce the ephemeral progesterone (P4)-producing corpus luteum (1, 2). This conversion or luteinization process of the dominant follicle has been well studied with respect to changes in steroid hormone production and the compartmentalization of gonadotropin receptors and steroidogenic enzymes (3, 4) and forms the basis of the “two-cell-two gonadotropin hypothesis of estrogen biosynthesis” (5). Although it is understood that paracrine interactions occur between the theca and granulosa tissue layers and that they are essential for folliculogenesis and likely luteal function, much yet remains to be discovered with respect to these interactions (6). This is particularly true for the role of the LH-responsive thecal cells (TCs) of which very little is known.

Immunocytochemical and in situ hybridization studies in rat have demonstrated that granulosa cells (GCs) can have markedly different levels of LH receptor (LHCGR) expression dependent upon their position within the follicular wall. Higher LHCGR levels are found in mural GCs located close to the basement membrane vs those close to the antrum (7, 8). Indeed, oocyte-derived factors present within the follicular fluid are known to affect the mural GCs, with those residing near the antrum being under a greater influence. An even greater difference in gene expression between GCs intimately associated with the oocyte (ie, corona radiata and cumulus) and the mural granulosa has already been established (8–10).

Expression profiling studies performed over the last couple of years in cattle and other species have shed light on changes that occur during the early and intermediate processes of follicular differentiation such as recruitment, selection, and dominance (11–14). Furthermore, studies aimed at understanding the changes in gene expression following the ovulatory LH surge have been completed in rodent, domestic animals, and primate species. In mice and rats, early events (ie, 1 h after a preovulatory LH/human chorionic gonadotropin [hCG] surge) through 12 hours after the LH surge (ovulation occurs at ∼13–15 h) have been evaluated in GCs (15–17). In primates, a recent analysis of whole periovulatory follicles collected at 0, 12, 24, and 36 hours post-LH in primates was completed (18). In cows and buffalo, suppression-subtractive hybridization (19) and gene-specific microarrays (20, 21) have been used to interrogate granulosa-specific gene expression changes occurring after the preovulatory LH surge. In the paper by Ndiaye et al (19), bovine GCs were evaluated before and 23 hours after hCG administration, and these authors reported only the down-regulated genes (32 in total). Another bovine study, conducted by Gilbert et al. (20), compared GC pre-LH surge gene expression to granulosa collected 6 and 22 hours post-GnRH (using custom arrays). Lastly a recent study, focusing on genes involved in steroid biosynthesis and using RNAseq-technology, was used to comparatively survey expression profiles in cells of the follicular wall of heifers and lactating cows (22). However, to date, no study has provided a comprehensive examination of the effect of the LH surge on all layers of the bovine follicular wall, granulosa, and theca interna, particularly evaluating the differential effects of location of GCs within the follicular wall (ie, antral vs membrane associated).

The present study examines large dominant bovine follicles collected before and 21 hours after a GnRH-induced preovulatory LH surge; follicles were dissected and antral (aspirated) GC (aGC)-, membrane-associated GC (mGC)-, and TC-specific expression profiles were analyzed by mRNA microarray analysis. The bovine model is useful because ultrasonography allows accurate tracking of the developing follicle and because a mono-ovulatory species enhances its relevance to human correlates as well as its general importance to animal husbandry of this economically important species. The results of this study provide the first comprehensive expression profiling analysis of both LH receptor-positive cell types within the periovulatory follicle and cell type-specific regulatory pathways, and functions affected by the preovulatory LH surge are identified.

Materials and Methods

Animals and tissue samples

Follicles were collected from normally cycling cows (n = 6) of the Holstein-Friesian breed after slaughter and immediately processed as described elsewhere (3, 23). Briefly, for collecting large dominant follicles before the preovulatory LH surge, numbers, sizes, and relative locations of follicles (≥5 mm) of a growing cohort were monitored by ultrasonography in normally cycling cows after ovulation and at 3 days and 21 hours before slaughtering using a transrectal 5 MHz linear transducer (Aloka SSD-500, Aloka GmbH, Meerbusch, Germany). In this way, the growth of individual follicles could be tracked. These cows were slaughtered at days 7 (n = 1) and 8 (n = 2) of the estrous cycle (day 0 = day of estrus) during the first follicular wave. In these animals the largest growing follicle (as determined by ultrasound monitoring) was collected, whereas stagnating or regressing follicles were easily identified and avoided. To collect the late preovulatory follicles after the LH surge, cows at day 8 (n = 3) of the cycle (first follicular wave) were ultrasonographically examined and treated with 500 μg PGF_2α (PGF forte, Veyx Pharma GmbH, Schwarzenborn, Germany) to induce luteolysis of the functional corpus luteum. Forty-eight hours later, all animals were scanned and injected with 100 μg of a GnRH analog (Gonavet Veyx, Depherelin, Veyx Pharma GmbH) to induce the LH surge. The animals were slaughtered 23 hours after the GnRH injection, which elicits a preovulatory LH surge about 2 hours after administration, and the largest growing follicle of each animal was dissected away from the remaining ovarian tissue (3, 23).

All procedures involving living animals (injections, ultrasonographic examinations) were according to the German law for animal protection (TierSchG) and in compliance with the European legislation on the care and use of animals.

Isolation of different follicular cell types

Two separate fractions of GCs were successively isolated from each follicle, aGC and membrane-associated granulosa (mGC) previously described as mural granulosa in Refs. (3 and 23). Briefly, the aGC fraction, together with the antral fluid, were collected by puncturing the antral follicle with a 18G needle, and the antral fluid, together with the cumulus cells and free floating or only slightly adherent cells of the GC layer, were aspirated. It should be noted that the cumulus cells and oocyte were not separated from the other aGCs aspirated from the follicles, but they likely make up a very small percentage of the actual transcripts within the aGC preparation, because it is well established that large numbers of mural GCs are removed by aspiration. The fluid was centrifuged (2 min, 400 relative centrifugal force, room temperature), the cell-free supernatant was frozen and stored at −20°C for determination of steroid concentrations, and the sedimented cells were frozen in liquid nitrogen and stored at −80°C for RNA preparation. The follicle was then cut open with scissors, and the follicular wall was peeled away from the remaining ovarian stroma and submersed in Ca²⁺ and Mg²⁺-free PBS buffer (pH 7.4). Membrane-associated GCs (mGCs) were then gently scraped off with a small scalpel blade. Detached cells and surrounding buffer were collected with a pipette, centrifuged (2 min, 400 relative centrifugal force), after which the supernatant was discarded and the sedimented cells were frozen in liquid nitrogen and stored at −80°C until RNA preparation. To isolate the theca cells (TCs), the inner (antral) side of the follicular wall was again carefully scraped with a scalpel blade and washed several times in PBS in order to further eliminate residual GCs attached to the basement membrane. The remaining basement membrane and the TCs attached to its outer side were then transferred to RNAlater solution (Qiagen, Hilden, Germany), stored at 4°C overnight, and then transferred to −20°C for long-term storage until RNA preparation.

Determination of follicular P4 and E2 content

Concentrations of P4 were determined using a competitive single-antibody [³H]RIA (24). Briefly, follicular fluid of each sample (10 μL each) was diluted with assay buffer (90 μL). Ten microliters of the diluted samples were used to measure directly, ie, without extraction, P4 and E2 concentrations. The tracer, [1,2,6,7-³H] progesterone, was purchased from GE Healthcare (Freiburg, Germany). The antibody raised in rabbits was further purified by affinity chromatography on protein A superose (GE Healthcare). The intra- and interassay coefficients of variation (CVs) were 7.4% and 9.8%, respectively. Radioactivity was measured with a β-counter with integrated RIA calculation (TriCarb2900; PerkinElmer, Rodgau, Germany).

Concentrations of E2 were determined with an ultrasensitive [¹²⁵I]RIA (DSL, Sinsheim, Germany). The standard curve was established between 5 and 750 pg/mL. Radioactivity was measured with an automatic γ-counter with integrated RIA calculation (Wizard; PerkinElmer). The detection limit of the method was 3 pg/mL. The intra- and interassay CVs were 8.4% and 10.2%, respectively.

The intrafollicular ratio of E2:P4 was calculated to characterize the endocrinologic status of isolated follicles. Before the preovulatory LH surge only E2-active (E2:P4 > 1) large dominant follicles were processed for cell type-specific transcriptome analysis because it has been demonstrated by others that E2-inactive follicles (E2:P4 < 1) isolated before the LH surge had reduced capacities to specifically bind gonadotropins and display clear histologic signs of atresia (25).

RNA preparation for mRNA microarray analysis

Total RNA was isolated from granulosa and theca with Trizol (Ambion, Forest City, California), per the manufacturer's instruction. RNA quality was assessed with Agilent Bioanalyzer Nano Chips (Agilent Technologies, Santa Clara, California), and all samples had sharp 18S and 28S peaks as well as RNA integrity values over 8.0, indicating high-quality RNA. The RNA was subjected to Gene Chip Eukaryotic Target Labeling Assay (Affymetrix). Biotin-labeled cRNA was fragmented according to Affymetrix protocols. The fragmented cRNA from each sample was hybridized to individual Affymetrix bovine gene array chips using the GeneChip Fluidics Station 400 protocol (Affymetrix), and the hybridized chips were scanned using the Agilent GeneArray Scanner (Affymetrix). Before analysis, all probes that did not show significant probe level hybridization (measured by Affymetrix MAS5.0 algorithm) in any of the 18 samples were filtered out. The raw signal was background corrected, normalized, and gene-level summarized using the Robust Multichip Average (RMA) procedure (26) using default RMA settings. To allow for statistical analysis, 3 preovulatory dominant follicles isolated from 3 different cows (ie, experimental unit) and 3 periovulatory (eg, ∼21-post LH-surge) follicles isolated from 3 different cows were dissected and the aGC, mGC, and TC preparations were collected, such that a total of 18 individual bovine Affymetrix GeneChips were used. The array results have been uploaded in the GEO database (GSE46996).

cDNA synthesis and quantitative RT-PCR

Gene-specific primers for cDNA synthesis and amplification of different transcripts were designed according to published mRNA sequences (Supplemental Table 1 published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org). For cDNA synthesis, 0.25 μg total RNA was reverse transcribed in a 25-μL reaction volume using M-MLV reverse transcriptase, RNase H Minus, Point Mutant (Promega, Mannheim, Germany) with gene-specific, or random and oligo-dT primers. The freshly synthesized cDNA samples were cleaned with the High Pure PCR Product Purification Kit (Roche, Mannheim, Germany) and eluted in 50 μL elution buffer. The identity of products generated with different primer pairs was determined by sequencing.

For qPCR, 0.5 and 0.25 μL of each purified cDNA sample were amplified with the LC 480 SYBR Green I Master Kit (Roche) in 12 μL total reaction volume. Measures from both reactions were averaged, considering the different amounts of cDNA. Amplification and quantification of products were performed in a LightCycler 480 instrument (Roche) under the following cycling conditions: preincubation at 95°C for 5 minutes, followed by 40 cycles denaturation at 95°C for 20 seconds, annealing at 60°C for 15 seconds, extension at 72°C for 15 seconds, and single-point fluorescence acquisition at 83°C for 10 seconds. The melting peaks of all samples were routinely determined by melting curve analysis in order to ascertain that only the expected products had been generated. Additionally, the length of all PCR products was monitored by agarose gel electrophoresis analysis (3% agarose, ethidium bromide stained). Cloned PCR products of the respective transcripts were used to generate external standard curves. Routinely, dilutions of standards covering 5 orders of magnitude (5 × 10⁻¹⁶ to 5 × 10⁻¹² g DNA per reaction) were freshly diluted from stocks of 10⁻⁹ g DNA/μL and coamplified during each run. Abundance of target transcript was calculated relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcripts. In a previous study it has been shown that GAPDH transcript abundance is similar in the aGC, mGC, and TC fractions of large dominant bovine follicles before and after the LH surge (3).

Statistics and bioinformatics

Correlations between mRNA microarray data and levels of transcript abundance (relative to GAPDH transcripts), as determined by quantitative RT-PCR (qRT-PCR), were calculated by Pearson Product Moment analysis; pair-wise comparisons by t test analysis using the SigmaPlot 12.0 software (Jandel Scientific, San Rafael, California).

Gene set enrichment analysis was performed with the Ingenuity Pathway Analysis tool (Ingenuity IPA 8.0, Ingenuity Systems Inc., Redwood City, California) to map differentially expressed transcripts to specific biologic functions and canonical pathways. If the calculated P value was below .05, mapping of the respective transcript to a specific biologic function or canonical pathway was considered significant. For clarity, significant canonical pathways were combined to super ordinate categories (see Results).

Results

Morphologic and physiologic characterization of follicles before and after the LH surge

Ultrasonography of each developing follicle was completed to ensure that only the single largest growing follicles (ie, destined preovulatory follicles) were collected for subsequent analyses. Physiologic (antral E2 and P4 concentrations), morphologic (size, vascularization, color etc.), and molecular markers were used to confirm that the individual follicles collected from the 6 different cows were characteristic of either preovulatory (n = 3) and post-LH (n = 3) follicles before gene expression analysis. Before the LH surge, the mean growing follicle diameter was 13.8 ± 1.2 mm, and the ratio of antral fluid E2:P4 was 4.9 ± 1.5, which was significantly different from follicles of the post-LH group, which were 20.5 ± 0.9 mm and 0.6 ± 0.2, respectively (Supplemental Table 2). Analysis of PTGS2 transcript levels within the GCs of the LH-treated periovulatory follicles were markedly (∼280-fold) elevated in comparison to the preovulatory follicles (Table 1). Thus, the preovulatory follicles were E2 active, which is an important criterion for healthy (nonatretic) large dominant follicles (25), whereas those after LH treatment had a significantly reduced E2:P4 ratio and increased PTGS2 expression, which are both reliable biomarkers for impending ovulation (3, 23, 27–29).

Table 1.

Hybridization Values, Fold Changes, and Correlation Coefficients of Microarray and qRT-PCR Comparison for LH-Responsive TC- and GC-Specific Genes

Gene Symbol	Gene Name	Fold Change qRT-PCRa			Microarray HSI − Mean RMA Hybridization						Correlation of qRT-PCR and HSI r Value
					aGC		mGC		TC
		aGC	mGC	TC	bLH	aLH	bLH	aLH	bLH	aLH
CYP17A1	Cytochrome P450; family XVII, 17α-hydroxylase			−77	47	36	266	26	3452	45	1.00
FSHR	FSH receptor	−4.1			285	70	294	75	29	19	0.91
CYP19A1	Cytochrome P450; family 19; aromatase	−398	−539	−80	6762	17	13603	25	1255	16	0.97
PTGS2	Prostaglandin-endoperoxide synthase 2	274	290	32	16	4287	13	3876	20	643	0.94
ADAMTS1	ADAM metallopeptidase with thrombospondin type 1 motif; 1	29	8.5		45	1272	184	1569	1138	2552	0.89
CCND2	Cyclin D2	−82	−27		5297	64	3993	146	796	296	0.92
CYP11A1	Cytochrome P450; family 11; subfamily A; polypeptide 1				8180	549	9733	974	7146	2852	0,83
HSD3B1	hydr.-δ-5-ster. dehydr.; 3β- and steroid δ-isom. 1				7002	5677	9315	4268	3571	2716	0.89
LHCGR	LH/CG receptor	−7.9	−17	−7.4	280	35	790	46	380	52	0.95
NR5A1	Nuclear receptor subfamily 5; group A; member 1			−2.8	529	352	405	228	642	229	0.73
NR5A2	Nuclear receptor subfamily 5; group A; member 2				8205	5549	9327	5559	1093	1164	0.89
PCNA	Proliferating cell nuclear antigen	−4.2	−2.7		2300	553	2263	841	2443	1942	0.81
PTX3	Pentraxin 3; long	644	711	108	14	9232	15	10921	85	9158	0.93
STAR	Steroidogenic acute regulatory protein				57	287	823	963	9578	4471	0.96
TIMP1	TIMP metallopeptidase inhibitor 1	85	38	10	72	6151	241	9231	1439	14888	0.85

Boldface symbols are the cell-specific and LH-induced (PTGS2) validation genes we evaluated.

^a Only significant LH responses by array are depicted; bLH, before LH surge; aLH, after LH surge; correlation coefficients (r) calculated by Pearson Product Moment analysis including all 18 samples; all P values were <.001.

Microarray analysis, cell-specific markers, and qRT-PCR validation

Principal component analysis of the data revealed that the global gene expression patterns observed in aGCs, mGCs, and TCs from follicles before (n = 3) and after LH surge (n = 3) were tightly clustered within cell types and across treatment groups (before and after LH; Figure 1A). Analysis of the microarray data revealed that of the 24 016 probe sets which represent 11 548 independent annotated bovine genes and 7091 unannotated expressed sequence tags, approximately half were determined as significantly present, based on the significance of the probe level hybridization of their transcripts to the microarray in all 3 biologic replicates (calculated using Affymetrix MAS 5.0 algorithm) in all aGC, mGC, or TC samples either before and after LH, respectively (Figure 1, B and C). The majority of the probe sets, 10 770 and 11 124 were present in all 3 cell types before and after the LH surge, respectively. No striking differences in actual numbers of present genes within a cell type before and after the LH surge were found for the aGC, mGC, or TC samples (comparison of gene numbers across cell types in Figure 1, B and C). In order to estimate the degree of potential contamination of the GC preparations with thecal RNA and vice versa, the microarray hybridization signals of several well-established genes that are commonly associated with one cell type or the other (CYP19A1 and FSHR in GC; and CYP17A1 in TC) were examined (Table 1). Hybridization signal intensity (HSI: RMA normalized fluorescence intensity values surrogate to the transcript hybridization signal) for CYP19A1 in aGCs and mGCs before the LH surge were robust (6762- and 13606-HSI, respectively); however, TCs collected before LH surge also had detectable levels (1255-HSI) of CYP19A1, possibly indicating a small degree of contamination of the TC preparation with this highly expressed GC mRNA. Statistical analysis indicates that the aGC and mGC CYP19A1 levels were significantly (false discovery rate [fdr] < 0.01) greater (5.4- and 10.8-fold, respectively) than those in TCs before the LH surge. After the LH surge, the CYP19A1 HSI dropped to low levels (<25) in all cell types. Another established GC-specific transcript, FSHR, that has a lower initial HSI in aGCs and mGCs (285- and 294-HSI) before the LH surge, was 9.8- and 10.1-fold greater (fdr < 0.02) than the HSI level (<29) observed in TCs, respectively. After the LH surge FSHR expression dropped (70- and 75-HSI) as expected in aGCs and mGCs, respectively, whereas TC values remained low and nearly unchanged (29-HSI vs 19-HSI; Table 1). Examination of CYP17A1 in the 3 cellular preparations indicated that TC levels (3452-HSI) were 73.5- and 13.0-fold greater than those in aGCs (47-HSI) and mGCs (266-HSI). After the LH surge, CYP17A1 levels were low (HSI < 45) and not different across cell types.

Figure 1.

mRNA Microarray Analysis of Present Transcripts within Cells of the aGCs, mGCs, and TC before and after the Preovulatory LH Surge. A, Principal component analysis shown for the 3 cell types, before (bLH) and after (aLH) the LH surge. Each symbol represents an individual replicate and the line connects those from the same cow. B and C, Venn diagrams showing members of probe sets significantly called present, based on the significance of the probe level hybridization to the microarray in all 3 biologic replicates (panel B represents before the LH surge and panel C represents after the LH surge).

Using a panel of 15 genes expressed at varying levels in GCs and TCs we compared the hybridization signals and relative qRT-PCR results for these genes (Table 1). The transcript abundance levels of these 15 selected transcripts, 7 of which were significantly up- or down-regulated by the LH surge, were correlated to qRT-PCR results for all 18 samples. The Pearson product moment correlation test indicated that all 15 transcripts showed a highly significant (P < .001) correlation between qRT-PCR and microarray hybridization values with correlation coefficients between 0.73 and 1.00 (Table 1). Because of this high correlation for genes expressed at divergent levels (low and high expression genes) combined with the tight clustering of biologic replicates observed in the principal component analysis analysis, we postulate that the microarray results are accurate and that they reflect actual relative changes before and after the LH surge for the remaining genes detected in the microarrays.

Cell-specific gene expression markers in GCs and TCs

Comparison of aGCs to mGCs before or after LH yielded few statistically significant differences. Only 1 gene, PLXNB2 (encoding plexin B2) was 2.67-fold greater (fdr < 0.05) in mGCs vs aGCs before LH, whereas 2 genes, HS6ST1 (heparan sulfate 6-O-sulfotransferase 1; 1.52-fold increase) and SSR1 (signal sequence receptor; α −1.88-fold decrease) were different (fdr < 0.05) between mGCs and aGCs after the LH surge. Even increasing the fdr to < 0.1 resulted in no further annotated genes being detected with a fold difference of ≥1.5-fold.

Comparison of TC expression to mGCs or aGCs before LH indicated that 3564 or 3650 genes, respectively, had a ≥1.5-fold (fdr < 0.05) difference in expression across these cell types; of these the majority (2696 genes) were consistently differentially regulated in both GC populations compared with TCs (Supplemental Table 3). After the LH surge 596 or 3833 genes were different (fdr < 0.05) at the ≥1.5-fold level between TCs and mGCs or aGCs, respectively; of these 541 were consistently differentially regulated in both cell populations compared with TCs (Supplemental Table 4). In summary, these data are therefore consistent with the observation that aGCs and mGCs are very similar and quite divergent from TCs.

Using a different approach to identify cell-specific genes (comparison of the significance of the transcript HSI calls, ie, absent vs present calls), we identified 34 genes as specific to the TC preparation: (ie, present in all 6 TC samples and absent [below background hybridization] in all 12 GC samples; with a significance of P < .05; Supplemental Table 5). A large number of these TC-specific transcripts are implicated in endothelial cell function (ie, PECAM1, TMEM100, LMO2, CD300LG, PEAR1). This is likely due to the unavoidable contamination of the TC with the capillary plexus surrounding the highly vascular dominant follicle. The complete lack of these endothelial markers in the avascular GC preparations of all 12 samples is reassuring. Analysis of the GCs identified 15 genes as present in either mGCs or aGCs and absent from all TC samples; 6 of the 15 genes were shown to be LH regulated in the GCs (Supplemental Table 6).

Cell-specific expression of gene transcripts influenced by the LH surge

To determine those genes that are regulated by the LH surge, we included only those probe sets that showed significant probe level hybridization (P ≤ .05, Affymetrix MAS 5.0 algorithm) and that exhibited an absolute fold change that was equal to or exceeded 1.5-fold and had a stringent Benjamini-Hochberg adjusted significance P value (ie, fdr) of 0.01 or less. This resulted in the selection of 2741 and 2417 differently expressed probe sets in aGCs and mGCs, and 226 in TC samples before and after LH, respectively (Figure 2). The relative number of gene transcripts which were down-regulated (aGCs 54%, mGCs 52%, and TCs 56%) was slightly more than numbers of gene transcripts which were up-regulated (Figure 2). Consistent with their similar tissue of origin, aGCs and mGCs shared 1783 transcripts that were similarly and significantly up- or down-regulated (Figure 2). A large number of LH-regulated probe sets (804 and 473) were differentially expressed in only aGCs or mGCs, respectively (Figure 2). However, in most cases these probe sets that were significant for either aGCs or mGCs had a calculated similar fold change (ie, within 2-fold) for the other cell type but failed to reach significance in the other cell type under the highly stringent conditions we established. Comparison of the fold changes for all 3060 probe sets that were LH regulated in GCs (see Supplemental Table 7), showed that only 29 genes exhibited more than a 3-fold difference (P < .05) in response to LH between aGCs and mGCs (Table 2).

Figure 2.

Venn Diagram Showing Numbers of Significantly LH Affected Probe Set Hybridization Signals in GC and TC Preparations. Total numbers of up- (↑) or down-regulated (↓) probe sets are shown in brackets.

Table 2.

LH Regulated Genes That Are Differentially Expressed Between aGCs and mGCs

Gene Name or Chromosomal Location	Gene Symbola	Fold Change Between LH Response of aGCs vs mGCs	P Value	LH Regulated in aGCs fdr < 0.01	LH Regulated in mGCs fdr < 0.01	HSI − Mean RMA Hybridization Value Before LH		HSI − Mean RMA Hybridization Value After LH
						aGC	mGC	aGC	mGC
Nidogen 2	NID2	16.52	.002	Yes	Yes	23	558	3616	5224
Similar to Collagen α-1(XIV) chain precursor	LOC781493	8.91	.001	Yes	No	14	138	371	417
Caveolin 1	CAV1	7.54	.001	Yes	Yes	16	75	6548	6235
Fibronectin 1	FN1	5.10	.002	Yes	No	89	744	1200	1978
Claudin 11	CLDN11	4.91	.011	Yes	Yes	12	57	3518	3409
Dickkopf homolog 3	DKK3	4.77	.001	Yes	Yes	15	92	2273	2952
Collagen, type V, α 2	COL5A2	3.82	.029	Yes	No	15	256	215	956
CART prepropeptide	CARTPT	3.68	.001	No	Yes	45	155	46	43
Chr2:110382-336-110382643(−)	(IGFBP5)	3.67	.033	Yes	No	71	540	573	1181
Dihydropyrimidinase-like 3	DPYSL3	3.56	.001	Yes	YES	21	81	252	277
ρ family GTPase 3	RND3	3.55	.001	Yes	Yes	29	93	4046	3710
ADAM metallopeptidase with thrombospondin type 1 motif; 1	ADAMTS1	3.34	.001	Yes	Yes	45	184	1272	1569
Chromosome 8 open reading frame 4	C27H8orf4	3.32	.025	Yes	Yes	11	37	654	662
Cyclin D2	CCND2	−3.01	.002	Yes	Yes	5297	3993	64	146
SPARC-related modular calcium binding 2	SMOC2	−3.02	.015	Yes	No	942	717	192	442
Hydroxysteroid (17-β) dehydrogenase 1	HSD17B1	−3.05	.010	Yes	Yes	2452	1039	49	62
Heme oxygenase 1	HMOX1	−3.07	.033	Yes	No	367	218	90	164
NADH dehydrogenase (ubiquinone) 1 α subcomplex; 4-like 2	NDUFA4L2	−3.10	.005	Yes	No	62	21	20	22
6-Phosphofructo-2-kinase/fructose-2;6-biphosphatase 3	PFKFB3	−3.28	.011	No	Yes	283	95	504	556
Chromosome 12 open reading frame 32	C5H12orf32	−3.35	.002	Yes	No	226	236	32	113
CD14 molecule	CD14	−3.39	.048	No	Yes	77	20	146	130
Neural precursor cell expressed; developmentally down-regulated 9	NEDD9	−3.41	.032	Yes	No	130	118	21	65
KIAA0101 protein (PCNA-assoc factor)	KIAA0101	−3.45	.018	Yes	Yes	1782	1439	45	126
SAM and SH3 domain containing 1	SASH1	−3.51	.028	Yes	No	105	65	16	35
chr2:88862297–88862784 (−)	(Ankrd44)	−4.04	.009	Yes	No	405	163	70	114
chr11:16969669-16970066 (+)	—	−4.26	.006	Yes	Yes	701	267	27	43
Complement component 1; s subcomponent	C1S	−5.63	.005	No	Yes	20	41	18	210
Guanylate cyclase activator 1A (retina)	GUCA1A	−6.07	.008	Yes	No	485	97	21	25
chr1:73033634-73034079 (+)	—	−8.03	.003	No	Yes	19	200	102	1737

^a Gene symbols enclosed in parentheses are probe sets that recognize transcripts with high homology to a named gene in another species.

Chromosomal locations are based on the Cow October 2011 assembly. Dashes reflect negative or down regulated gene.

Of the 226 probe sets shown to be LH regulated in TCs, only 62 probe sets (representing 57 independent genes) were significantly regulated by LH in TC only. Most (74%) of these TC-specific LH-regulated transcripts (Table 3) were significantly down-regulated after LH-surge, with CYP17A1 encoding steroid 17α-hydroxylase (P450c17), the key enzyme of androgen biosynthesis showing the strongest response. Of the remaining 164 probe sets that were LH regulated in both TCs and GCs all but 2 of these showed the same (ie, either up or down) regulation across the different cell types. The 2 genes CACNA1H (calcium channel; voltage-dependent; T type; α 1H subunit) and SIPA1 (signal-induced proliferation-associated 1) were down-regulated (−2.13-fold, fdr < 0.01; −1.92-fold, fdr < 0.01) in TCs after LH. In contrast, CACNA1H and SIPA1 were up-regulated in aGCs (2.74-fold, fdr < 0.01; 1.46-fold, fdr > 0.5) or mGCs (1.05-fold, fdr > 0.5; 1.69-fold, fdr < 0.01), respectively. The remaining 162 differentially expressed probe sets (representing 136 independent genes), were divided into high, intermediate, and low expression (based on HSI) for depiction in (Table 4). Fold-changes varied widely with the greatest change being the up-regulation of PTX3 (108-, 711-, 643-fold) and the down-regulation of CYP19A1 (−80-, −539-, −397-fold), in TCs, mGCs, and aGCs, respectively. Comparison of these LH-regulated genes to previous studies indicated that a large number of the identified genes have not been previously shown to be LH-regulated in bovine GCs (see Table 4).

Table 3.

LH-Regulated Genes Specific to TCs

Gene Name or Chromosomal Location	Gene Symbola	Fold Change	HSI − Mean RMA Hybridization Value
			Before LH	After LH
Cytochrome P450c17, 17α-hydroxylase/17;20 lyase	CYP17A1	−77.1	3452	45
Ankyrin repeat and sterile α motif domain containing 4B	ANKS4B	−19.1	234	12
Solute carrier family 1; member 3	SLC1A3	−10.5	195	19
chr16:69743779-69744290 (−)	(TRAF5)	−9.5	107	11
Chemokine (C-X-C motif) ligand 14	CXCL14	−9.2	142	15
Tetraspanin 33	TSPAN33	−8.8	270	31
Crystallin; μ	CRYM	−7.8	328	42
Hydroxyprostaglandin dehydrogenase 15-(NAD)	HPGD	−7.1	875	124
Asporin	ASPN	−6.5	709	109
dCTP pyrophosphatase 1	DCTPP1	−5.5	2181	400
Oncoprotein induced transcript 3	OIT3	−5.0	222	45
ChrUn_JH123563:7386-7628 - AffyID Bt.16424.1.S1_at	—	−4.9	609	124
Transmembrane 7 superfamily member 2	TM7SF2	−4.7	624	132
Ryanodine receptor 3	RYR3	−4.5	85	19
Pyridine nucleotide-disulfide oxidoreductase domain 2	PYROXD2	−4.4	216	50
Angiopoietin 4	ANGPT4	−4.0	192	48
BCL2-related protein A1	BCL2A1	−3.9	193	49
Protein phosphatase 2; regulatory subunit B; β	PPP2R2B	−3.9	45	12
Perilipin 5	PLIN5	−3.7	44	12
chr5:17947935-17948135 (−)	—	−3.7	28	8
Guanylate binding protein 4	GBP4	−3.7	106	29
Transient receptor potential cation channel, subfamily M, member 4	TRPM4	−3.3	102	30
Chemokine (C-C motif) ligand 26	CCL26	−3.3	135	41
Small integral membrane protein 10	SMIM10	−3.1	147	48
Cytochrome P450; family 4; subfamily F; polypeptide 3	CYP4F3	−3.0	64	21
Potassium large conductance calcium-activated channel; subfamily M; β member 1	KCNMB1	−3.0	52	18
Nuclear receptor subfamily 5; group A; member 1	NR5A1	−2.8	642	229
Similar to DKFZP586J0619 protein	INTS1	−2.6	347	131
ADP-ribosyltransferase 3	ART3	−2.6	94	36
NLR family; CARD domain containing 5	NLRC5	−2.6	59	23
Gap junction protein; β 5; 31.1 kDa	GJB5	−2.5	243	98
Diacylglycerol O-acyltransferase homolog 2 (mouse)	DGAT2	−2.2	41	18
Carbohydrate (keratan sulfate Gal-6) sulfotransferase 1	CHST1	−2.2	228	102
Mevalonate kinase	MVK	−2.0	571	288
Fibroblast growth factor 1 (acidic)	FGF1	−1.9	30	16
RAN guanine nucleotide release factor	RANGRF	−1.9	115	60
Mitochondrial ribosome recycling factor	MRRF	−1.9	445	233
chr29:48560683-48560989 (+)	(CCND1)	−1.9	23	12
ATP synthase subunit g, mitochondrial-like	ATP5 liter	−1.7	2599	1561
Solute carrier family 25; member 34	SLC25A34	−1.7	33	20
Solute carrier family 25: member 23	SCL25A23	−1.6	67	41
Fragile X mental retardation; autosomal homolog 1	FXR1	−1.6	1542	970
Kruppel-like factor 10	KLF10	2.3	581	1311
ras Homolog gene family; member B	RHOB	2.4	1375	3341
chr19:32921507-32921922 (+)	(Hs3st3b1)	2.6	10	25
chr11:86149219-86149711 (−)	(FAM84A)	2.7	23	60
Chromosome 21 open reading frame 7 ortholog	C1H21ORF7	2.9	18	52
Adenomatosis polyposis coli down-regulated 1	APCDD1	3.0	71	213
Carbonic anhydrase IV	CA4	3.8	45	169
chr8:52926349-52926740 (+)	(Rorb)	3.9	11	42
Coagulation factor V (proaccelerin; labile factor)	F5	4.0	13	53
chr8:26339534-26340004 (−)	(Acer2)	4.3	78	340
Nuclear protein; transcriptional regulator; 1	NUPR1	4.9	63	306
ATP-binding cassette; subfamily G (WHITE); member 1	ABCG1	6.9	14	93
Solute carrier family 7; member 5	SLC7A5	7.7	54	419
Ceruloplasmin (ferroxidase)	CP	8.2	20	166
Lysyl oxidase-like 4	LOXL4	18.8	98	1847

^a Gene symbols enclosed in parentheses are probes sets that recognize transcripts with high homology to a named gene in another species.

Chromosomal locations are based on the Cow October 2011 assembly. Dashes reflect a negative or down regulated gene.

Table 4.

LH-Regulated Genes in GCs and mGCs and TCs

Gene Name or Chromosomal Location	Gene Symbola	aGC Fold Change	mGC Fold Change	TC Fold Change	HSI − Mean RMA Hybridization Value Before LH			HSI − Mean RMA Hybridization Value After LH
					aGC	mGC	TC	aGC	mGC	TC
High Expression Genes
Cytochrome P450; family 19; subfamily A; polypeptide 1	CYP19A1	−397.6	−539.3	−80.1	6762	13603	1255	17	25	16
Carbohydrate sulfotransferase 8	CHST8	−103.8	−89.7	−15.6	2313	2044	270	22	23	17
chr4:81639977-81640459 (+)	—	−63.5	−44.1	−56.9	22293	23339	5402	351	529	95
Hydroxysteroid (17-β) dehydrogenase 1	HSD17B1	−50.5	−16.5	−5.7	2452	1039	197	49	63	34
Brain-expressed X-linked 2	BEX2	−47.6	−52.9	−37.3	15996	16449	7717	336	311	207
Troponin I type 3 (cardiac)	TNNI3	−41.0	−27.9	−10.5	10015	7713	5333	244	276	507
Inhibin; β A	INHBA	−29.5	−29.4	−44.6	24933	27321	9370	846	930	210
chr23:7266183-7266577 (+)	(GCLC)	−24.9	−18.8	−7.9	8260	8319	2630	331	442	335
Solute carrier family 35, member G1	SLC35G1	−20.4	−22.5	−10.8	3422	3945	740	168	175	69
Glutamic-pyruvate transaminase	GPT	−19.2	−26.5	−4.6	1748	1990	276	91	75	60
Janus kinase 3	JAK3	−18.0	−17.7	−12.5	2269	2816	2496	126	159	200
ST3 β-galactoside α-2;3-sialyltransferase 4	ST3GAL4	−15.8	−13.4	−3.0	8672	11352	2212	549	849	726
Arsenic (+3 oxidation state) methyltransferase	AS3MT	−9.0	−9.5	−18.6	347	358	2497	38	38	134
Serpin peptidase inhibitor; clade A; member 3	SERPINA3	−8.7	−7.6	−3.9	3539	2669	3235	408	351	832
Stearoyl-CoA desaturase 5	SCD5	−6.3	−6.6	−2.7	7996	11245	5720	1265	1701	2133
ADP-ribosylation factor GTPase activating protein 3	ARFGAP3	−6.1	−5.1	−2.7	6010	7954	5284	988	1571	1978
chr27:17010375-17010926 (−)	(SORBS2)	−6.1	−4.9	−3.2	3427	2903	4173	565	592	1319
Guanine nucleotide binding protein (G protein); γ 10	GNG10	−5.4	−4.7	−1.8	10398	9768	4266	1911	2068	2332
chr25:38495947-38496497(−)	—	−5.3	−5.9	−2.3	9275	7888	2375	1758	1339	1031
Biliverdin reductase A	BLVRA	−5.0	−4.1	−2.2	4220	3271	1814	848	803	838
chr27:36195193-36195591 (+)	(PLEKHA2)	−4.8	−4.2	−1.9	4214	3869	1608	872	926	835
G protein-coupled receptor 125	GPR125	−4.6	−5.1	−2.4	2078	2619	1215	455	515	499
Enoyl-CoA δ isomerase 1	ECI1	−3.6	−2.1	−2.2	1164	744	684	319	357	304
Transcription elongation factor A (SII)-like 1	TCEAL1	−3.3	−2.5	−2.5	923	858	1418	277	344	562
MAP/microtubule affinity-regulating kinase 1	MARK1	−3.2	−3.6	−3.0	2561	2565	1775	811	703	593
Membrane-spanning 4-domains; subfamily A; member 8B	MS4A8B	−3.1	−6.0	−7.0	1775	3444	4092	564	575	585
Ligand of numb-protein × 2	LNX2	−3.1	−3.0	−3.3	1899	1623	1337	611	544	411
Ubiquitin specific peptidase 48	USP48	−2.5	−2.3	−1.8	1048	1140	640	418	493	347
Transcriptional adaptor 3	TADA3	−2.0	−1.8	−1.9	1489	1180	827	749	648	425
chr21:13524038-13524353 (−)	LOC100848010	−2.0	−1.7	−2.0	2416	2418	2002	1237	1456	1019
chr8:101776156-101776718 (+)	(ZNF462)	−1.8	−2.1	−2.0	845	896	535	461	429	270
ORM1-like 2 (S. cerevisiae)	ORMDL2	−1.6	−1.7	−1.6	965	1010	1103	593	608	696
Progesterone receptor membrane component 1	PGRMC1	−1.3	−1.6	−1.6	1299	2355	2342	1023	1459	1480
Myeloid cell leukemia sequence 1	MCL1	2.3	2.4	1.8	645	574	914	1495	1378	1622
SWI/SNF related; matrix associated; actin dependent regulator of chromatin; subfamily a; member 1	SMARCA1	3.4	3.0	3.1	3258	3635	2445	11008	10934	7674
WD repeat domain 26	WDR26	3.5	2.5	1.8	1323	1515	926	4579	3796	1701
SLAIN motif family; member 1	SLAIN1	3.5	3.9	4.4	631	526	122	2236	2065	534
chr3:8282216-8282683 (−)	(ATF6)	4.2	3.6	2.1	904	1001	681	3828	3587	1428
UDP-GlcNAc:βGal β-1;3-N-acetylglucosaminyltransferase 2	B3GNT2	4.4	4.4	3.5	1779	1735	847	7905	7630	2982
chr1:50742017-50742481 (−)	(CBLB)	4.7	4.6	3.4	237	267	151	1118	1218	513
CDC42 small effector 2	CDC42SE2	5.5	5.5	2.6	696	705	510	3814	3859	1331
TCDD-inducible poly(ADP-ribose) polymerase	TIPARP	5.7	5.5	3.5	265	272	224	1513	1485	774
Solute carrier family 25; member 17	SLC25A17	6.4	4.6	2.2	551	831	646	3551	3829	1428
N-Acylsphingosine amidohydrolase (acid ceramidase) 1	ASAH1	6.5	6.0	3.9	324	342	567	2090	2051	2190
Monoamine oxidase A	MAOA	7.6	6.2	4.2	209	206	174	1583	1270	739
Plasminogen activator; tissue	PLAT	9.3	3.6	4.2	601	1897	801	5588	6754	3336
Translocator protein (18 kDa)	TSPO	9.3	8.3	3.5	156	170	279	1457	1413	977
Retinol binding protein 1; cellular	RBP1	10.2	16.5	7.6	1039	758	1013	10634	12524	7657
chr28:30192805-30192992 (+)	(SAMD8)	10.6	8.6	4.7	251	354	188	2660	3054	890
Metallothionein 2A	MT2A	12.9	14.8	22.6	890	629	306	11508	9286	6917
chr12:67614705-67615018(+)	(GPC6)	13.4	11.9	6.8	244	280	198	3267	3341	1341
Spermidine/spermine N1-acetyltransferase 1	SAT1	15.0	12.7	2.7	1185	1213	4102	17742	15349	11043
Regenerating islet-derived family; member 4	REG4	17.6	10.5	5.2	88	134	360	1540	1413	1873
chr23:35873098-35873230 (−)	(SOX4)	18.9	11.0	3.8	391	558	670	7400	6143	2564
chr26:11594846-11595263 (−)	(SLC16A12)	19.0	12.6	26.4	57	106	14	1079	1337	360
Thrombospondin 2	THBS2	19.6	10.0	5.4	34	92	535	660	911	2863
Nudix (nucleoside diphosphate linked moiety X)-type motif 10	NUDT10	22.0	22.4	12.1	118	130	54	2587	2910	650
Regulator of calcineurin 1	RCAN1	23.1	20.2	3.9	167	181	288	3863	3647	1120
SEMA5A sema domain, 7 thrombospondin repeats, transmembrane domain (TM) and short cytoplasmic domain, (semaphorin) 5A)	SEMA5A	32.9	25.7	4.8	33	49	240	1101	1265	1144
Nudix (nucleoside diphosphate linked moiety X)-type motif 11	NUDT11	34.9	29.7	13.9	176	258	187	6152	7668	2596
FBJ murine osteosarcoma viral oncogene homolog	FOS	40.3	40.2	7.2	122	134	431	4898	5396	3088
Procollagen C-endopeptidase enhancer 2	PCOLCE2	43.1	21.6	7.6	25	55	189	1079	1181	1447
TNF receptor superfamily; member 6b; decoy	TNFRSF6B	51.6	81.1	10.0	28	25	36	1429	2005	361
Neurotensin	NTS	92.8	84.8	9.4	20	21	50	1887	1759	467
TIMP metallopeptidase inhibitor 1	TIMP1	106.8	41.3	9.6	52	251	1563	5600	10386	15001
Chromosome 5 open reading frame, human C12orf75	C5H12ORF75	108.6	76.2	18.9	26	37	156	2791	2803	2941
Vanin 2	VNN2	116.0	140.9	39.6	29	22	30	3364	3056	1170
chr5:5940972-5941399 (−)	(PHLDA1)	116.0	104.7	16.3	23	25	39	2665	2589	641
chr6:22210314-22210785 (+)	(CXXC4)	141.4	83.9	18.6	17	24	19	2432	2032	351
ρ Family GTPase 3	RND3	141.7	39.9	3.4	29	93	644	4046	3710	2217
Prostaglandin-endoperoxide synthase 2	PTGS2	274.0	289.6	32.4	16	13	20	4287	3876	643
Periostin; osteoblast specific factor	POSTN	325.0	194.8	13.3	10	21	114	3121	4067	1514
chr23:19331670-19332170 (+)	(RUNX2)	602.3	276.5	25.3	9	20	61	5282	5645	1536
Pentraxin 3; long	PTX3	643.9	711.4	108.0	14	15	85	9232	10921	9158
Intermediate Expression Gene
chr3:9463181-94632151 (+)	(Dab1)	−19.5	−15.6	−11.0	372	254	229	19	16	21
Solute carrier family 5 (sodium/glucose cotransporter); member 11	SLC5A11	−19.1	−28.5	−13.1	227	369	329	12	13	25
Ethanolamine kinase 2	ETNK2	−11.2	−8.6	−4.1	822	745	451	74	87	110
Transmembrane protein 156	TMEM156	−9.1	−7.3	−3.2	728	468	169	80	65	53
Troponin T type 1 (skeletal; slow)	TNNT1	−8.3	−4.6	−15.4	148	105	898	18	23	58
LH/CG receptor	LHCGR	−7.9	−17.2	−7.4	280	790	380	35	46	52
Breast cancer antiestrogen resistance 3	BCAR3	−7.7	−6.6	−2.6	907	1074	328	118	162	125
chr3:24537369-24537825 (−)	(PDE4DIP)	−4.5	−3.2	−2.7	397	446	865	88	140	320
Aldolase C; fructose-bisphosphate	ALDOC	−4.4	−2.4	−2.5	939	563	582	212	236	232
Protein phosphatase 1; regulatory (inhibitor) subunit 1A	PPP1R1A	−2.9	−3.9	−8.8	124	155	527	43	40	60
unc-119 homolog (C. elegans)	UNC119	−2.8	−2.8	−4.2	238	213	495	84	77	118
Cox2 chaperone homolog	COX20	−2.6	−2.0	−1.7	259	211	191	101	108	115
Phosphofructokinase; muscle	PFKM	−2.5	−2.4	−2.6	327	442	616	132	186	236
Estrogen receptor 1	ESR1	−2.4	−2.7	−1.9	391	426	174	162	160	91
Propionyl CoA carboxylase; β polypeptide	PCCB	−2.2	−2.2	−2.2	362	674	977	165	303	446
Chromosome 19 open reading frame, human C17orf101	C19H17orf101	−1.8	−1.7	−1.8	203	240	243	114	145	136
chr10:70504545-70504958 (+)	(TMEM260)	−1.6	−2.0	−1.8	202	300	258	124	153	140
chr18:53779152-53779618 (+)	(ARHGAP35)	−1.5	−1.7	−1.6	330	357	284	223	205	182
Polyhomeotic homolog 2	PHC2	2.7	2.6	2.0	216	196	201	589	507	401
Filamin A interacting protein 1	FILIP1	4.1	3.7	2.3	103	106	66	419	390	154
Zinc finger; DHHC-type containing 23	ZDHHC23	4.2	2.6	2.4	118	165	56	499	427	133
Retinol binding protein 4; plasma	RBP4	4.7	4.2	3.9	50	83	50	237	349	195
Similar to TPR repeat-containing protein 39	TTC39C	5.7	5.0	5.3	121	127	44	682	633	231
chr8:99285533-99286070 (−)	(ABCA1)	5.8	8.8	5.1	36	32	84	212	280	432
Cylindromatosis	CYLD	5.9	4.4	1.7	156	193	177	914	845	303
A kinase (PRKA) anchor protein 11	AKAP11	6.5	4.9	3.2	135	149	75	875	729	238
Inositol 1;4;5-trisphosphate 3-kinase A	ITPKA	6.7	6.3	2.1	95	96	175	634	606	373
Fibulin 1	FBLN1	6.7	4.4	4.0	97	226	130	652	994	523
Solute carrier family 25; member 12	SLC25A12	9.5	9.7	3.3	95	94	141	905	913	468
chr9:42277341-42277566 (+)	—	15.6	14.0	4.0	31	37	31	490	523	123
TNF receptor superfamily; member 12A	TNFRSF12A	16.5	20.3	6.0	17	30	81	286	617	486
Poly (ADP-ribose) polymerase family; member 8	PARP8	21.6	19.3	6.6	47	44	48	1027	858	316
Regulator of G-protein signaling 2	RGS2	22.1	14.7	4.4	22	26	26	490	376	113
chr20:4115985-4116308 (−)	(Sh3pxd2b)	23.8	18.3	4.6	27	36	119	649	662	550
Protein phosphatase 4; regulatory subunit 4	PPP4R4	23.9	19.6	8.0	39	57	69	930	1113	547
KIAA0226-like	KIAA0226 liter	45.9	35.0	7.7	12	11	13	557	390	97
Glutamate receptor; ionotrophic; AMPA 3	GRIA3	80.5	71.3	8.7	9	10	9	749	680	74
Low Expression Genes
Rh family; B glycoprotein (gene/pseudogene)	RHBG	−11.1	−10.8	−7.0	99	98	89	9	9	13
chr23:30536840-30537310 (−)	(Hist1h4 h)	−8.1	−9.0	−2.9	149	156	91	18	17	31
Collectin subfamily member 11	COLEC11	−5.1	−7.1	−5.1	143	284	173	28	40	34
Cytochrome P450; family 2; subfamily C; polypeptide 87	CYP2C87	−4.5	−4.5	−3.2	108	92	82	24	21	26
Cytochrome P450; family 4; subfamily F; polypeptide 3	CYP4F3	−4.2	−6.4	−16.0	58	64	241	14	10	15
Usher syndrome 1C binding protein 1	USHBP1	−4.1	−4.3	−3.2	140	127	159	34	30	50
chr8:67614674-67615116 (−)	—	−3.1	−3.4	−2.5	184	203	96	60	60	39
Tetraspanin 33	TSPAN33	−2.6	−3.8	−8.8	31	48	186	12	12	21
Chemokine (C-C motif) ligand 25	CCL25	−2.4	−2.5	−2.7	199	167	220	83	66	82
Shisa homolog 3 (Xenopus laevis)	SHISA3	−2.1	−2.9	−3.3	19	29	34	9	10	10
SH3-domain GRB2-like 2	SH3GL2	−1.9	−3.1	−3.6	190	225	114	99	74	32
Sushi domain containing 3	SUSD3	−1.4	−2.0	−3.7	45	58	116	31	29	31
tRNA splicing endonuclease 54 homolog (S. cerevisiae)	TSEN54	−1.3	−1.5	−1.8	95	126	167	71	83	92
Teashirt zinc finger homeobox 2	TSHZ2	−1.3	−1.6	−2.2	55	74	195	41	46	89
LY6/PLAUR domain containing 1	LYPD1	2.8	3.6	3.2	21	17	14	59	63	45
chr10:27864596-27865146 (+)	(KATNBL1)	3.0	2.4	1.9	101	100	49	301	246	90
chr16:68337216-68337707 (−)	(VASH2)	3.2	3.7	2.7	25	33	39	80	122	107
chr2:53798017-53798235 (+)	(ZEB2)	3.8	3.6	6.9	49	44	23	185	157	160
Potassium large conductance calcium-activated channel; subfamily M; β member 4	KCNMB4	5.6	7.6	2.2	25	23	58	139	176	127
Retinoic acid induced 14	RAI14	8.7	8.2	2.7	35	22	67	306	183	181
Carbonic anhydrase XIII	CA13	8.7	11.8	7.1	17	16	14	148	185	100
RAS protein activator like 1 (GAP1 like)	RASAL1	9.4	6.9	3.2	34	37	22	318	257	71
Transmembrane protein 22	TMEM22	11.3	9.6	2.2	11	11	11	126	108	23
chr11:22800828-22801173 (−)	—	13.7	14.0	3.9	14	13	14	197	188	53
Meis homeobox 1	MEIS1	32.3	22.8	2.9	8	11	60	258	246	178

^a Gene symbols enclosed in parentheses are probes sets that recognize transcripts with high homology to a named gene in another species.

Chromosomal locations are based on the Cow October 2011 Assembly. Dashes reflect a negative or down regulated gene.

Gene set enrichment analysis

To uncover biologic functions that are significantly affected by the LH surge, differentially expressed transcripts were analyzed with the Ingenuity Pathway Analysis (IPA) tool. Altogether 1614, 1373, and 146 of the differentially expressed transcripts, which had been identified in aGCs, mGCs, and TCs, respectively, could be significantly assigned to specific IPA biofunctions. Both GC fractions were affected by LH and yielded very similar biofunctional categories, which included “cell cycle, cellular assembly and organization,” and “DNA replication, recombination, and repair,” and “cellular movement.” The “functions annotations” within these broad categories had increased “predicted activation states” that were broadly associated with cellular migration (eg, “organization of cytoplasm/cytoskeleton,” “formation of plasma membrane projections,” “microtubule dynamics;” Supplemental Table 8). In addition, these cells had significantly decreased “predicted activation states” that were associated the “functions annotations” (“segregation of chromosomes,” “proliferation of cells”: Supplemental Table 8).

IPA analysis of the 146 LH-regulated theca genes indicated that the following categories were highlighted: “lipid metabolism,” “small molecule biochemistry,” “endocrine/reproductive system development and function” and “cell death and survival” were highlighted.

The “functions annotations” within these categories all had significant decreased “predicted activation states” associated with “modification of androstenedione” or “apoptosis of epidermal cells” (Supplemental Table 9).

Discussion

This study describes, for the first time, the comparative gene expression profile of TCs cells from preovulatory follicles (before LH surge) to periovulatory follicles (collected ∼21 h after a GnRH-induced LH surge) of any species. Comparison of LH-regulated TC genes with the other LH-responsive cell (ie, granulosa) within the same follicle, also provides additional novel insights into ovarian physiology as well as helps with interpretation of previous “granulosa cell” only studies. LH responsiveness of the mural GCs has long been known to vary with respect to location of GCs with respect to distance from the basement membrane that separates it from the theca. Our results indicate that these differences are mostly quantitative because the actual genes that were expressed in the antral (aspirated) and membrane-associated (scraped) granulosa are not that different, with only a few genes exhibiting any major quantitative differences with respect to position within the follicular wall. Lastly, this study provides a robust and comprehensive accounting of gene expression changes that occur following the LH surge in bovine follicular cells and the identification of a number of novel LH-regulated genes in the bovine follicle, both in GCs and TCs.

Validation of experimental model and cellular preparations

The morphologic, physiologic, and molecular characterization (principal component analysis) of these dominant follicles collected before and after the LH surge, clearly indicate that very little variation was observed across cows, again indicating a very consistent experimental procedure. The increased mean size and the dramatic decrease in the E2:P4 ratio indicates that the conversion from an E2-active to E2-inactive follicles in the wake of the preovulatory LH surge is consistent with previous studies using this same paradigm to collect preovulatory follicles (before LH surge) and periovulatory follicles (collected ∼21 h after a GnRH-induced LH surge) as described earlier (3) as was the induction of PTGS2. The consistency of the array data was further validated by the very high correlations (R = 0.73 to 1.0) for 15 genes when compared with qRT-PCR levels. Because of these highly uniform correlation coefficients, we can be very confident in the relative fold-differences predicted by the arrays.

The purity of GC and TC sample preparations was verified by analysis of CYP17A1 (17α-hydroxylase) or CYP19A1 (aromatase) and FSHR, which have been reported to be restricted to the theca and granulosa layers, respectively (4) by in situ hybridization. Our results are consistent with earlier studies that show both steroidogenic genes and the FSHR gene are largely down-regulated by the preovulatory LH surge (3, 30, 31). The presence of CYP17A1 and CYP19A1 in mGC and TC, respectively, at very low levels, might be explained by slight contamination with the respective other cell type. Following the scraping of the basement membrane, this is probably not avoidable, nor is the incidental sloughing of some TC into the mGC preparation following scraping. Interestingly, 2 of the top LH down-regulated genes as reported by Gilbert et al. (20) in mural bovine GCs, CXCL14 (chemokine [C-X-C motif] ligand 14) and ASPN (asporin), were found in our experiments to be primarily expressed in TCs (142- and 709-HSI) compared with aGCs and mGCs (HSI < 20), respectively. Furthermore, we show that these 2 reportedly GC LH-regulated genes (20) were only LH-regulated in TCs (Table 3), decreasing significantly after the LH surge. This and other previous studies examining GC gene expression in periovulatory follicles did not examine purity of their cellular preparations and based their GC derivation solely on the method of isolation (scraping), which is identical to our procedure for mGCs. Consistent with their distance from the TCs, the aGCs had significantly less CYP17A1 than mGC, yet was still detectable (by hybridization) even though the experimental procedure (needle aspiration) was highly unlikely to allow a significant amount of cross-contamination. This leaves open several alternatives, 1) GCs/TCs may actually express very low levels of “contaminating” mRNA that are sufficient to be detected via the sensitive methods we have used or 2) alternatively, extracellular RNA in the form of exosome encapsulated/associated RNA released from GCs, TCs, or follicular fluid could be associating with the basement membrane or other cell preparations effectively contaminating them. The lack of detection of oocyte-derived transcripts (eg, BMP15 and GDF9 and ZP2, ZP3, and ZP4) in the aGC preparation illustrates that the levels of detectable contaminating RNAs (eg, CYP17A1, CYP19A1, etc.) is of a greater magnitude than that can be detected from a single oocyte; however, it is no reliable indication for the actual absence of oocyte, which might be present in these preparations. The high induction of PTGS2 expression in the GCs following the GnRH-induced LH surge is a conserved mechanism that occurs about 18 hours post-hCG treatment in bovine follicles in vivo and is considered a reliable marker for approaching ovulation (29, 32). Overall these results indicate a very small degree of contamination that may only be evident for those genes with extremely high expression such as CYP17A1 and CYP19A1, because low/intermediate expressing genes (FSHR), although still being detected, were far more cell selective. On the other hand, the uneven distribution of LHCGR transcripts particularly between the aGC and mGC samples indicates that both granulosa preparations, in fact, represent different fractions. As expected from data of others (7), LHCGR transcripts predominated in the GCs close to the basement membrane (mGC [HSI-790] compared with the aspirated cells [aGCs, HSI-280] or TCs [HSI-380] before LH, respectively).

Theca-specific transcriptome is considerably less affected by the preovulatory LH surge than the granulosa-specific transcriptome

Principal component analysis revealed that the LH surge exerted a significant influence on the expression profiles of theca and granulosa samples, leading to the arrangement of the samples into clearly defined clusters, before and after LH. Also theca and granulosa samples could be clearly separated due to their cell type-specific expression profiles. Whereas 25% and 21% of all expressed genes showed a significant reaction to the GnRH-induced LH surge in aGCs and mGCs respectively, only 2% of the genes present in TCs were significantly up- or down-regulated. The vast majority of genes that are LH regulated in granulosa and theca showed the same direction of change (up or down) in both cell types. The strongly varying extent of LH response in theca and granulosa, however, cannot be simply explained by a lower number of LH binding sites (LH receptors) on cells of the theca compared with those of the granulosa. Previous studies, as well as the present study, indicate that both cell types, granulosa and theca, expressed the corresponding LHCGR transcripts at relatively high levels before the onset of the LH surge (see Table 1 and Refs. 3, 4, and 33). One might speculate that the signal transduction pathway downstream of the LH receptor might be different in GCs and TCs during the periovulatory period. This is the first demonstration that the TC transcriptome shows considerably less alteration in the wake of the preovulatory LH surge compared with that of the GCs. This result might be expected if one considers the reduced morphologic and biochemical changes that occur in TCs (loss of androgen biosynthesis, minimal cell hypertrophy) compared with GCs (loss of cell proliferation, loss of aromatase activity, induction of numerous steroidogenic genes, extensive cell hypertrophy).

Novel genes identified in bovine TCs that are regulated by the LH surge

Although critical roles in periovulatory follicular function and subsequent luteal function have been defined for TCs, previous studies examining TC cell gene expression have been limited to individual genes (3, 34, 35) or cultured TC experiments (36). This study is therefore the first to globally identify thecal genes that exhibit changes in mRNA expression in vivo following the LH surge. In total 203 genes (combined results of Tables 3 and 4) exhibited differential expression in the TCs following the LH surge. However, only 57 of these 199 were specifically LH regulated in TCs (Table 3). Of the remaining 141 LH-regulated genes found common to both granulosa and theca, only about 10% or 19 of these LH-regulated genes could be absolutely confirmed to be LH regulated in theca because they had higher (HSI) expression levels than in corresponding GC preparations. A great number of the other 141 genes (Table 4) showed similar levels across the cell types, but without further analysis (ie, in situ hybridization or laser capture dissection followed by qRT-PCR confirmation) cannot be assured as truly LH-regulated genes in TC, and not cross-contamination by granulosa. The 19 additional thecal LH-regulated genes included AS3MT, SORBS2, TCEAL1, MS4A8B, ORMDL2, MCL1, ASAH1, SAT1, THBS2, FOS, PCOLCE2, TIMP1, RND3, TNNT1, PDE4DIP, PPP1R1A, PFKM, PCCB, SIPA1 (encoding arsenic methyltransferase, transcription elongation factor A-like 1, sorbin and SH3 domain containing 2, membrane spanning 4-domains; subfamily A; member 8B, ORM1-like 2, myeloid cell leukemia sequence 1, N-acylsphingosine amidohydrolase 1 [acid ceramidase], spermidine/speriine N1-acetyltransferase 1, thromospondin 2, FOS, procollagen C-endopeptidase enhancer 2, TIMP1, Rho family GTPase 3, troponin T type 1, phosphodiesterase 4D interacting protein [PDE4DIP], transcript variant 8, phosphofructokinase, propionyl CoA carboxylase; β polypeptide, protein phosphatase 1; regulatory subunit 1A, signal-induced proliferation-associated 1, respectively).

Not surprisingly, CYP17A1 mRNA, which encodes the critical TC enzyme involved in androgen biosynthesis had the greatest decrease (1/77 of pre-LH values) following the LH surge. Transcriptional regulation of the CYP17A1 gene has been well studied (31, 37) and steroidogenic factor-1 (SF-1 encoded by NR5A1) one of its primary transcriptional regulators was also decreased to approximately one-third of its pre-GnRH level following the LH surge in TC (Table 3). As an established mediator of steroidogenic activity and known transcriptional activator of CYP17A1, this loss of SF-1 in TCs and lack of LH effect on GCs is consistent with previous observations, which indicate that NR5A2 (also known as LRH-1) is the primary steroidogenic transcription factor in GCs (38). We additionally identified acid ceramidase (ASAH1) as one of 19 genes induced (∼4- to 6-fold) by LH that was common to both TCs and GCs. Recently, ASAH1, which catalyzes the conversion of ceramide to sphingoside (generating sphingoside 1 phosphate) was demonstrated to be induced in adrenocortical cells by ACTH/cAMP, and this was associated with changes in adrenal steroidogenesis, by decreasing the expression of a number of steroidogenic enzymes including CYP17A1 (39, 40). This is the first demonstration that LH may have a similar effect in TCs. Another unique aspect of sphinoside-1 phosphate is its antagonist actions on NR5A1 transcriptional activity (41). Thus, in addition to a decrease in SF-1 expression, acid ceremide-induced sphingosine may further block SF-1 action, ultimately leading to the sharp reduction in CYP17A1 expression we observed. With respect to the remaining LH-regulated thecal genes, a large number are associated with lipid metabolism and regulation of apoptosis (see Supplemental Table 9), both of which would be considered as important in luteinization and formation of the corpus luteum.

To our surprise we found PTX3 (pentraxin 3) as a highly abundant and LH-regulated gene also in TCs. It is well established that this protein plays a role as a modulator of the inflammatory response (42). In the context of the ovarian follicle, however, its expression has only been reported in the context of GCs of periovulatory follicles, where it mediates cumulus cell-oocyte interactions (43). In a recent paper it was also found that it interferes with fibroblast growth factor 2 actions in luteal endothelial cells, thus inhibiting angiogenesis (44). Suggestively, pentraxin 3 might be involved in the regulation of angiogenesis also in the intensely vascularized theca layer. However, the function of its heavy LH-induced up-regulation during the periovulatory period still seems obscure, because this protein is an inhibitor of angiogenesis, whereas the luteinization process is accompanied by intense vascular growth.

The thecal-specific gene with the greatest LH induction we identified was LOXL4 (encoding lysyl oxidase-like 4). This gene's expression increased 18.8-fold (HSI-98 to HSI-1847) following the LH surge, with no concomitant LH induction in aGCs or mGCs (HSI was <148 at all time points in granulosa). Lysyl oxidases are an extracellular copper enzyme that cross-link lysine side chains in mature fibrillar collagen or elastin, regulating the tensile strength and structural integrity of connective tissues (45). LOXL4 shares it amine oxidase motif and the scavenger receptor cysteine-rich protein domain with 4 other family members, lysyl oxidase (LOX) and lysyl oxidase like 1–4 (45). Our study detected expression of LOX, LOXL1, and LOXL2 in TCs and GCs, with TCs having higher levels of LOX (HSI-140) and LOXL1 (HSI-450) than GCs (HSI < 30 for both), whereas equal levels of LOXL2 (all HSI < 62) were found in TCs and GCs. However, no LH regulation was observed for these other lysyl oxidase family members within TC or GC compartments. Conversely, although whole-tissue Northern blot analyses have demonstrated presence of LOXL4 in human ovary as well as numerous other tissues (46), this is the first indication of LH regulation of this gene in any ovarian tissue. Previous studies examining LOX expression and activity have shown it to be down-regulated during early folliculogenesis by FSH (47, 48). In other cells, LOXL4 has been shown to block cell migration by suppressing the synthesis of laminins and α3 integrin and the activity of matrix metallopeptidase 2 (49), and recent studies have shown the full-length protein to be a tumor suppressor gene (50). Thus, the high induction of LOXL4 seen in TCs might be a critical regulatory protein within the extracellular matrix in the highly dynamic periovulatory follicle, and this gene awaits further study. Lastly, most of the newly identified TC genes have not been studied within the context of the periovulatory follicle; therefore this study provides a critical first step in our understanding this largely understudied follicular cell.

Variation in mural GC transcriptome based on location within the periovulatory follicle

Immunohistochemical analyses and in situ hybridization studies have shown that LHCGR levels can vary with distance from the basement membrane (7). The relative expression of LHCGR in our aGC and mGC preparations before the LH surge (Table 1) supports this observation although it did not reach significance. Recent studies examining the gene profile in GC types, mural vs cumulus granulosa, have observed large differences (51–53). Our study did not aim to investigate this difference but instead wished to determine whether differences in the mural granulosa (ie, those near the antrum vs those near the membrane) could be elucidated. Indeed although our aGC samples would likely be contaminated with cumulus cells and the oocyte, we failed to detect oocyte-specific genes or the differences observed in these other papers (51–53). Our study further indicates only minor differences in the 2 different mural GC preparations: mGCs (ie, membrane-associated) and aGCs (aspirated). This inability to detect changes could easily be attributed to a pronounced contamination of the aGC population with mGCs, a result that cannot be prevented using the current procedures. To further evaluate whether a quantitative difference might exist we compared the fold changes in response to LH across the cells (See Supplemental Table 7 for complete list) and identified a small number of genes (29) that were different (Table 2). We observed a preferential loss (one-fourth of its pre-LH levels) of CARTPT transcript in mGCs that was unaffected in aGCs. Decreased follicular fluid levels and expression of CARTPT have been previously tied to bovine follicle selection and increased health status of the dominant follicle in bovine (54). Many of the other genes that were quantitatively differentially expressed (>3-fold difference) between aGCs and mGCs were either extracellular matrix proteins or proteins that interact with or metabolize these proteins. Interestingly, extracellular matrix genes were also enriched in cumulus cells in comparison with mural granulosa in human preovulatory follicles (52). Extracellular matrix proteins separate the granulosa and theca (ie, basement membrane) or are of a specialized type (also known as focimatrix proteins) between GCs (55). Most of these proteins (ie, collagen α-1, claudin 11, a disintegrin-like and metallopeptidase with thrombospondin type 1 motif [ADAMTS1]), had slightly higher expression in the mGCs vs aGCs before the LH surge and fairly equivalent levels after the LH surge. Nidogen 2 and fibronectin 1, both extracellular matrix proteins, had clearly greater expression in the mGC (558- and 744-HSI) compared with almost background levels (23- and 89-HSI) in aGCs before LH, and almost equivalent levels after the LH surge (see Table 2). Two additional genes which modulate extracellular signaling, neural precursor cell expressed; developmentally down-regulated 9 (NEDD9) and SAM and SH3 domain containing 1 (SASH1) were LH-regulated in aGCs, exclusively. These two proteins have been linked to cancer cell migration and invasion (56, 57). Whether these genes may play a role in the ovulation by oocyte migration model as proposed by Akison et al. (58) is unknown. Lastly, several genes involved in cell cycle control, CCND2 (cyclin D2) and KIAA0101 (also known as proliferating cell nuclear antigen-associated factor) were both found to undergo a pronounced decrease in expression after the LH surge, with the decline being of slightly greater magnitude in the aGCs than in the mGCs (see Table 2).

Novel genes identified in bovine GCs that are regulated by the LH surge

Our study identified a large number of heretofore unknown LH-regulated genes in bovine granulosa, even though this is not the first study to characterize the periovulatory GCs following the LH surge. Indeed, a recent study examining late post-LH changes of GC-specific expression profiles in bovine follicles (20) having remarkably different data was reported. Surprisingly, only 121 of LH-regulated genes are common between both studies. In addition, when comparing the transcript abundance data of individual genes, remarkable differences are evident. From the 15 cell marker transcripts shown in Table 1, 14 are represented on both microarray systems (ADAMTS1, CCND2, CYP11A1, CYP19A1, FSHR, LHCGR, NR5A1, NR5A2, PCNA, PTGS2, PTX3, STAR, TIMP1, HSD3B1). From these 14, only CCND2, STAR, and TIMP1 were differentially expressed in the Gilbert et al. (20) data set (see Supplemental Data 1). The fact that PTGS2 was not up-regulated in the study of Gilbert et al. (20), at 22 hours post-LH, is surprising. PTGS2 expression is a conserved mechanism that occurs about 18 hours post-hCG treatment in bovine follicles in vivo (29, 32) and can be considered as a reliable marker for approaching ovulation. A thorough comparison of the 2 studies is presented in Supplemental Data 1. Consistent with our study, of 23 granulosa-derived transcripts that were found down-regulated 23 hours after hCG application using substractive hybridization (19), 13 (ANAPC5, ARFGAP3, CYP19A1, FNDC3B, FSHR, GJA1, IDH3A, INHBA, LRP8, PRC1, RPA2, SCD, TRIB2) were confirmed. In this regard, however, the deviating animal models might be responsible for some of the differences observed between studies. In the study of Gilbert et al. (20) follicles were harvested following an ovarian superovulation regime. In contrast, the monitoring (ultrasonography during estrous cycle, signs of heat, morphologic parameters, antral E2 and P4 concentrations, appearance of PTGS2 transcripts) and hormone treatment regime used during the present study was developed to obtain only a single, well-characterized large dominant preovulatory follicle per animal, thus mimicking the natural cycling situation as close as possible.

We identified 2 members of the Ig-like domain-containing superfamily, IGSF1 and IGSF11, encoding Ig superfamily; members 1 and 11, as LH-regulated genes in GC. These genes have not been identified in GC before and, interestingly, both are regulated by LH in the opposite way: whereas the x-chromosomal IGSF1 is clearly up-regulated in aGCs and mGCs (10.1- and 10.9-fold in aGCs and mGCs, respectively), IGSF11 is significantly down-regulated in both GC populations (−4.9 and −6.7 fold in aGCs and mGCs, respectively). Generally, the Ig domain has evolved to serve diverse biologic functions including growth and development, signaling, adhesion, and protein-carbohydrate interactions (59). Specifically, IGSF11 has been clearly connected to carcinogenesis (60, 61). The role of these specific Igs in luteinizing GC and, in particular, the complementary regulatory characteristic is not yet clear. However considering their function in other tissues, both might play a role during the intense tissue reorganization of the granulosa layer during the formation of the corpus luteum. This might also be applicable to another interesting gene, AMIGO2 (adhesion molecule with Ig-like domain 2), which is up-regulated 6.0- and 3.0-fold in aGCs and mGCs, respectively, by LH. AMIGO2 is a member of the AMIGO/Alvin family, which are transmembrane proteins characterized by 6 leucine-rich repeats and a single Ig-like extracellular domain. AMIGO proteins are mainly found in the nervous system, where they function as novel cell adhesion molecules that facilitate neuronal growth (62). Correspondingly, during the formation of the corpus luteum AMIGO2 that is up-regulated to relatively high levels in aGCs and mGCs by the preovulatory LH surge (1020-HSI and 1117-HSI, respectively; Supplemental Table 7) might also be involved in mediating cell adhesion and regulating migration of GC.

In conclusion, the data of the present study demonstrate that 1) theca cell-specific transcriptome is considerably less affected by the preovulatory LH surge than that of GCs, consistent with the greater changes induced in GCs as they undergo luteinization, 2) aGCs and mGCs show very similar gene expression profiles, thus suggesting a widely equivalent molecular equipment of both GC preparations and 3) numerous novel LH-regulated genes in GCs and TCs were identified. This is particularly true for the TCs where, with the exception of a few known TC marker genes, almost all of the genes were identified as LH regulated for the first time. This research resource comparing the transcriptomic profiles of both thecal and granulosa follicular cells from preovulatory (pre-LH) and post-LH (ie, periovulatory) follicles provides an easily accessible and searchable resource that should greatly benefit the scientific community.