This study has resulted in an increased awareness of Foxl2 function in the ovary. First, through the use of microarray analysis we have added to the growing list of genes that appear to be regulated by Foxl2 and thus may play a role in follicular development in the ovary. Second, we have demonstrated the GnRHR gene promoter is regulated in a positive manner by the transcription factor Foxl2 in the KK1 granulosa cell line.
Affymetrix microarray analysis
In an effort to increase the level of differential gene expression that could be induced by Foxl2 and thereby efficiently detected by microarray analysis, we have used Foxl2 derivatives in this study. This approach appears to have succeeded in that the mouse StAR gene was detected and had been previously demonstrated to be Foxl2 responsive in a human system . Based on our experimental design, we would have predicted that in comparing mock to Foxl2-VP16 transfected cells (NA) all values would be positive due to VP16 transactivation. However, in looking through the genes beginning with the letter "A" in our comprehensive alphabetical listing (Additional file 4), 16 out of 22 in the NA category were negative. The simplest explanations for repression of gene expression by VP16 are provided by the authors of a study that also used a VP16 fusion for microarray analysis and also noted unexpected negative regulation . These investigators speculated that VP16 caused the induction of repressors or squelching of coactivator activity .
The repressor induction mechanism for VP16 repression is a distinct possibility in light of recent studies that have explored the mechanisms involved in the control of Foxl2 transactivation activity. These investigators found that deacetylation of the Foxl2 protein by the SIRT1 deacetylase causes a decrease in Foxl2 transactivation . Sirt1 was also identified as a Foxl2 regulated gene that is activated by Foxl2 . In addition, these investigators demonstrated that the Foxl2 promoter is repressed by Sirt1 expression as part of a feedback mechanism of regulation in response to stress . Therefore, Sirt1 induction could alter the activity of Foxl2-VP16, as well as repress other genes, resulting in down regulation of genes in Foxl2-VP16 transfected cells. A specific example is a study demonstrating that Sirt1 deacetylation of AP1 modulates its function and causes repression of the Cox2 gene .
The comparison of mock to Foxl2-Mad transfected cells (NR) of genes beginning with the letter "A" reveals 10 out of 12 are negatively regulated as expected (Additional file 4). The finding that only 2 out of the 12 were activated by Mad domain repression suggests that Foxl2-Mad functions more reliably as a repressor in comparison to Foxl2-VP16 as an activator in our study. Perhaps this is due to the Mad repression domain interacting specifically with the Sin3 complex deacetylase leading to chromatin remodeling . On the other hand, the VP16 activation domain mechanism of transactivation is the result of interactions with a variety of factors including histone acetylases, basal transcription factors, and the coactivators CBP and Mediator to name a few [20, 21]. Therefore, over-expression of VP16 fusion proteins may lead to repression due to competition for the factors needed for endogenous gene expression . With this in mind, the use of an alternative activation domain with greater specificity in its interactions would have resulted in fewer false repression events. However, we should reiterate that this microarray study was intended to provide a listing of potential Foxl2 target genes and does not have the potential to discern Foxl2 regulatory mechanisms.
This study has compared microarray data from two in vitro studies that utilized the human KGN  and mouse KK1 (this study) cell lines respectively, and an in vivo study that used mouse Foxl2 knockouts . The KK1 cell line was derived from a transgenic female mouse in which SV40 T antigen expression was driven by a 6 Kb inhibin alpha promoter fragment . The mouse developed a large ovarian tumor that was collected after 5 months. The tumor cells had the morphological characteristics of granulosa cells. Subcultures were tested for their cAMP and steroidogenic response to chorionic gonadotropin and the culture with the strongest response (KK1) was characterized further. The KK1 cells were shown to be immortalized luteinizing granulosa cells that expressed LH and FSH receptors, steroidogenic enzymes, and inhibin alpha . The KGN cell line was derived from a 73 year old woman in which granulosa cell carcinoma had recurred . The KGN cells had steroidogenic activities similar to those of normal human granulosa cells and expressed functional FSH receptor . Therefore, in comparisons of these studies, we would assume that the microarray data derived from the KGN and KK1 cell lines is representative of well differentiated granulosa cells while the mouse microarray data (E13, E16, and P0) represents less differentiated granulosa cells from embryonic stages and birth .
As seen in Figure 1, we do find evidence that this assumption is correct when we compare the number of genes shared between KK1 cells and the in vivo mouse data of Garcia-Ortiz et al. . The number of shared genes increases from the embryonic stages (211&225 genes respectively) to 325 genes at birth (P0), indicating that KK1 cells have transcriptional profile more like that of a mature granulosa cell in vivo. Further similarities as well as differences in shared genes among the three comparison groups represented by the different colors in Figure 1 can be found in individual sheets in Additional file 9. Of the five genes that are shared by all groups (Figure 1-white), three have known functions: Pa2g4, Rab28, and Thbs2. Pa2g4 stands for proliferation-associated 2G4, a transcription factor involved in cell growth and signalling . Rab28 is a Ras oncogene family member involved in the regulation of membrane trafficking . Thbs2 is also found in Table 3, and encodes thrombospondin 2, an antiangiogenic protein involved in follicle development . The two genes of unknown function are RIKEN cDNA 5033428C03 which encodes the hypothetical protein LOC74728 (entrez gene ID 74728) and Ta0871 that encodes a hypothetical protein from Thermoplasma acidophilum (entrez gene ID 1456410).
Comparison of our microarray data and comprehensive listing of potential Foxl2 target genes (Additional file 4) to those generated by two other groups of investigators [6, 7] has allowed us to generate a subset of Foxl2 targets that have greater potential of being Foxl2 regulated (Additional file 10). Of particular interest are the genes that are common to all three studies: Bcl11a, Hoxb5, Mrgpre, and Ptpn6 (Tables 1 and 2). The functions of these genes are described below.
Mrgpre, along with Maff and Rspo3, also appear on the qPCR confirmed gene listing of Batista et al. . The Mrgpre gene product is a Mas1 related G protein coupled receptor that may be involved in the sensation or modulation of pain in a subset of sensory neurons . Maff stands for v-maf musculoaponeurotic fibrosarcoma oncogene homolog f (avian). The protein encoded by the gene is a basic-leucine zipper (bZIP) transcription factor that is up-regulated by pro-inflammatory cytokines in myometrial cells . Rspo3 encodes R-spondin 3, a secreted protein that mediates Wnt signaling and is involved in angiogenesis during mouse development .
Finally, the OKdb was utilized to identify genes from this study that were demonstrated to be expressed in the ovary in previous studies. These genes have been divided into three tables based on known functions: 1. gene regulation; 2. signaling; and 3. metabolism, cell adhesion, cytoskeletal, and structural. Our focus now turns to a discussion regarding the functions of genes listed in Tables 1, 2, 3 that are known to be expressed in granulosa cells (OKdb).
In the area of gene regulation (Table 1), Bcl11a and Serpine2 are expressed in granulosa cells although their function in the ovary is unknown. The zinc finger transcription factor Bcl11a was shown to be up-regulated in human granulosa cells treated with FSH . In human erythroid cells where much more is known about the factor, the Bcl11A protein functions as a repressor and is involved in silencing fetal hemoglobin expression in adults . Serpine2 is a serine protease inhibitor that is differentially expressed in large and small follicles in sheep . Serpine2 protein levels are elevated in dominant bovine follicles , whereas the levels of Serpine2 are lower in ovaries of Foxl2 knockout mice suggesting that the gene is induced by Foxl2 . Gabpa and Nr4a2 are two genes in this category that have been characterized to a greater extent with respect to granulosa cell function. Gabpa is an ETS family transcription factor that regulates the Rhox5 homeobox gene in rat granulosa cells . In the regulation of the nicotinic acetylcholine receptor gene, Gabpa recruits the histone acetyl transferase p300 when the promoter is activated, and recruits the histone deacetylase HDAC1 when the promoter is not activated . Nr4a2 was found to be rapidly induced by cAMP in the KGN granulosa cell line . LH was shown to induce Nr4a2 expression in mouse granulosa cells .
Six genes involved in signaling (Table 2) are expressed in granulosa cells. Akt1 is a component of the phosphoinositide 3'-OH kinase (PI3K) pathway and is phosphorylated in response to Igf1 stimulation of bovine granulosa cells . Akt1 has also found in human granulosa cells during follicle development . The human GnRH receptor has been shown to be expressed predominantly in granulosa cells of pre-ovulatory follicles . The role of GnRH in the ovary is diverse as it regulates steroidogenesis, cell proliferation, and apoptosis . Gucy1b3 encodes a guanylate cyclase that is activated by nitric oxide (NO) and is expressed at high levels in granulosa cells of primordial and primary follicles of the rat ovary . NO has been shown to inhibit estrogen production in rat granulosa cells  and steroidogenesis in porcine granulosa cells . Ppp1r1b is a protein phosphatase involved in signal transduction pathways in human granulosa cells in response to dopamine and human chorionic gonadotropin stimulation . Prlr encodes the prolactin receptor which is localized to granulosa cells as well as other cell types in the rat ovary . Prolactin receptor expression in rat granulosa cells is increased by treatment of cultured cells with FSH, LH and hCG . Ptpn6 encodes a protein tyrosine phosphatase that is involved in modulating the signaling cascade activated by PRL in granulosa cells .
The Ctla4, Eda, and Mrgpre genes are in the signaling category but are not found in the OKDB. However, they are worthy of mention due to appearing in this study as well as both the human KGN study  and the mouse knockout study . Ctla4 encodes cytotoxic T-lymphocyte associated protein 4, a receptor/signal transducer that suppresses immune system function and is regulated by the forkhead transcription factor FoxP3 [49–51]. Transcriptional regulation of Ctla4 by Foxl2 in granulosa cells may be the result of similarities in the Forkhead binding sequence elements in the Ctla4 promoter that allow both factors to regulate the gene, with cell type determining the presence of either FoxP3 or Foxl2 in T cells or granulosa cells respectively. The Eda gene encodes the protein ectodysplasin A, a tumor necrosis factor family member with several isoforms, one of which is a transmembrane protein . Mutations in the soluble form of the EDA protein and the EDA receptor are the cause of anhidrotic ectodermal dysplasia, a syndrome that results from impaired development of skin appendages during embryogenesis .
Genes in Table 3 that have been shown to be expressed in granulosa cells include Hspg2, an anticoagulant heparin sulfate proteoglycan involved in follicle development and ovulation in rats. Hspg2 had also been found in human follicular fluid . Odc1 encodes ornithine decarboxylase 1 (ODC1), the rate-limiting enzyme of the polyamine biosynthesis pathway which catalyzes ornithine to putrescine. ODC1 expression is stimulated by LH in granulosa cells and may mediate the effects of LH during the process of follicular development . Thbs2 encodes thrombospondin 2, an antiangiogenic protein involved in follicle development .
Two genes involved in metabolism, StAR and Vldlr, are expressed in granulosa cells and the gene products of both are involved in steroidogenesis. StAR encodes the steroidogenic acute regulatory protein, which transfers cholesterol from the outer to the inner mitochondrial membrane, the rate limiting step in steroidogenesis . Vldlr encodes the very low density lipoprotein receptor, which obtains lipoproteins from plasma, a source of cholesterol for steroidogenesis .