Heterogeneity of Ovarian Theca and Interstitial Gland Cells in Mice

It has been established that two developmentally and functionally distinct cell types emerge within the mammalian testis and adrenal gland throughout life. Fetal and adult types of steroidogenic cells (i.e., testicular Leydig cells and adrenocortical cells) develop in the prenatal and postnatal period, respectively. Although the ovary synthesizes steroids postnatally, the presence of fetal-type steroidogenic cells has not been described. We had previously established transgenic mouse lines in which fetal Leydig cells were labeled with an EGFP reporter gene by the FLE (fetal Leydig enhancer) of the Ad4BP/SF-1 (Nr5a1) gene. In the present study, we examined the reporter gene expression in females and found that the reporter gene is turned on in postnatal ovaries. A comparison of the expressions of the EGFP and marker genes revealed that EGFP is expressed in not all but rather a proportion of steroidogenic theca and in interstitial gland cells in the ovary. This finding was further supported by experiments using BAC transgenic mice in which reporter gene expression recapitulated endogenous Ad4BP/SF-1 gene expression. In conclusion, our observations from this study strongly suggest that ovarian theca and interstitial gland cells in mice consist of at least two cell types.


Introduction
Similar to the testis, the ovary exerts two predominant functions. One is the production of mature germ cells, while the other is the synthesis of sex steroids, which are thereafter delivered to the whole body and thereby induce female-specific characteristics. Estradiol, a potent estrogen, is synthesized in the ovary by successive reactions mediated by five enzymes: cholesterol sidechain cleavage cytochrome P450scc (CYP11A), 3β-hydroxysteroid dehydrogenase (3β-HSD), 17alpha-hydroxylase/17, 20-lyase cytochrome P450 (CYP17), 17β-hydroxysteroid dehydrogenase (17β-HSD), and aromatase P450 (CYP19) [1,2]. Interestingly, the former three enzymes are expressed within theca cells surrounding follicles while the latter two enzymes are within the granulosa cells inside follicles, indicating that estradiol-the final product of the sex steroid synthetic pathway in the ovary-is produced by cooperative actions between the two distinct cell types. In addition to theca and granulosa cells, the interstitial gland cells in ovaries of rodent species are steroidogenic as well, possessing functions similar to those of theca cells [3].
Ad4BP/SF-1 (NR5A1), a member of a nuclear receptor superfamily, was originally identified as a factor that regulates steroidogenic CYP gene expression in the adrenal cortex [4][5][6]. Evidence accumulated in subsequent studies has demonstrated that Ad4BP/SF-1 targets all steroidogenic genes required for the syntheses of gonadal steroids as well as of corticosteroids [7,8]. This functional relevance to steroidogenesis in the gonads is supported by the distribution of Ad4BP/SF-1, which is enriched in steroidogenic cells, such as Leydig cells in the testis and theca and interstitial gland cells in the ovary. In addition to its involvement to steroidogenesis, the functions of Ad4BP/SF-1 have been discussed from a developmental aspect, since mice in which its gene expression is disrupted develop neither adrenal glands nor gonads [9][10][11]. As a possible reason for this tissue agenesis, a recent study revealed that Ad4BP/SF-1 is involved in a variety of biological processes, [12] such as energy metabolism, through regulating glycolytic genes [13]. These findings raised a possibility that disruption of the Ad4BP/SF-1 gene leads to aberrant cellular functions and thereby results in this striking phenotype.
Developmentally, the male and female gonads in mammals start to functionally differentiate in the fetal and postnatal ages, respectively. Sertoli and Leydig cells emerge in the fetal testis, whereas ovarian theca and granulosa cells differentiate after birth in mice. The differences between the Leydig cells that emerge in the fetal testis and the Leydig cells in the adult testis have been discussed in terms of their morphological and functional features [14]. Studies that successfully distinguished these cell types based on differential gene expressions support this notion [15][16][17][18]. Moreover, a recent study clearly demonstrated that fetal Leydig cells (FLCs) and adult Leydig cells (ALCs) can be distinguished by how their enhancers of the Ad4BP/SF-1 gene are utilized [19]. A DNA fragment that can induce gene expression in FLCs but not ALCs was isolated from the gene. In subsequent transgenic mouse experiments involving a construct carrying an EGFP (enhanced green fluorescence protein) reporter gene under the control of the fetal Leydig enhancer (FLE), EGFP was expressed only in FLCs, but not in ALCs.
In the present study, we discovered that the FLE is activated in a population of steroidogenic theca and interstitial gland cells in postnatal mouse ovaries. These EGFP-positive cells were all positive for Ad4BP/SF-1, whereas only a subpopulation of Ad4BP/SF-1-positive cells was positive for EGFP. These observations provide evidence for the first time that the ovary, similar to the testis, contains two cell types in theca and interstitial gland cells.

Materials and Methods
DNA construction and generation of transgenic mice mFLE-EGFP (mutant FLE-EGFP; referred to as SmAc-1.8-Ad4BP(LBmut)-EGFP in our previous study) transgenic mice have been described previously [20]. mFLE-mCherry was constructed by replacing EGFP with mCherry. The resulting construct was injected into the pronuclei of fertilized eggs to generate transgenic mice as described previously [21,22]. A bacterial artificial chromosome (BAC) containing approximately 106-kb and 100-kb flanking regions at the 5' and 3' ends of the Ad4BP/SF-1 gene, respectively, was purchased from BACPAC Resources, Children's Hospital Oakland Research Institute (clone ID RP23-354G20; Oakland, CA, USA), and subjected to Red/ET system-based recombineering (Gene Bridges Gmbh, Heidelberg, Germany) [23][24][25]. A modification vector for the BAC recombineering was constructed as follows (Fig 1). A 479-bp fragment upstream from the first ATG in the 2nd exon of the Ad4BP/SF-1 gene was used as the 5' homologous arm. The 5' arm-EGFP-polyA fragment was amplified from mFLE-EGFP by PCR using the primer set 5'-GGGGTACCCGAATCTC TCCCAATGTCGT-3' and 5'-GGAATTCATATTAACGCTTACAATTTACGCGT-3', which carry KpnI and EcoRI sites at their respective 5' ends. A 499-bp fragment in intron 4 was used as the 3' homologous arm. This fragment was amplified from the BAC DNA by PCR with the primer set 5'-CGGGATCCAGCAACAGGAAGAACTTTCGAG-3' and 5'-ATAAGAATGCGGC CGCTAGAAAGGGCCAGTGAGAAAAG-3', which carry BamHI and NotI sites at their respective 5' ends. The 5' arm-EGFP-polyA fragment and 3' arm fragment were digested and inserted into KpnI/EcoRI sites and BamHI/NotI sites, respectively, of a PL451 FRT (flippase recombination target)-loxP-neo BAC modification vector (gifted from Dr. Neal G. Copeland, Houston Methodist Hospital, TX, USA). The constructed vector was sequenced, and an EGFP-Neo cassette was introduced into RP23-354G20 by Red/ET system-based homologous recombination. Finally, the Neo cassette was deleted by arabinose-induced flippase (flp)-mediated recombination at FRT sites. The BAC construct (Ad4BP-BAC-EGFP) was purified and injected into fertilized eggs, as previously described [26]. The presence or absence of the transgenes in the animals was determined by PCR, using the primers for EGFP [20] or mCherry (5'-CGCCGAC ATCCCCGACTACTT-3' and 5'-CAGCCCATGGTCTTCTTCTGC-3'). All protocols followed for animal experiments were approved by the Animal Care and Use Committee of Kyushu University (Permit Number: A26-001). Mice were sacrificed after sevoflurane anesthesia, and all efforts were made to minimize suffering.

EGFP expression driven by the FLE of Ad4BP/SF-1 gene in the ovary
In our previous study, we established transgenic mouse lines in which an EGFP reporter gene is expressed under the control of the FLE together with the proximal upstream region of Ad4BP/SF-1 (SmAc-1.8-Ad4BP(LBmut)-EGFP [20]. In this reporter gene construct, an Lhx9 binding site localized at the proximal upstream region was mutated (mFLE) in order to abolish the weak EGFP expression in interstitial cells other than FLCs of the fetal testis, as well as that in somatic cells of sexually indifferent gonads of both sexes. This construct will henceforth be referred to as mFLE-EGFP. Since the EGFP-positive cells in the fetal testis were mostly positive for 3β-HSD, this mutation seemed not to affect the expression of EGFP in FLCs.
Although steroidogenic FLCs are differentiated in fetal testes, cells such as fetal theca cells have not been thought to differentiate in the fetal ovaries. Therefore, EGFP reporter gene expression was expected to be undetected in the ovary. In fact, fluorescent microscopic observation showed a clear difference of the EGFP signals between the fetal testis and ovary at E15.5 (Fig 2A and 2B). Careful observation detected a small number of cells displaying very weak EGFP signals in the fetal ovary, although it is unknown whether the expression was induced by the FLE, the basal promoter, or both. Thereafter, clear expression became detectable in postnatal ovaries; EGFP signals could be seen clearly in the ovary at early postnatal day ( Fig 2C) and the signal became stronger gradually at P14, at P21 and the adult stage (Fig 2D-2F). With respect to the EGFP expression in the testis, the signal could be detected even in the postnatal period, as previously described [19], indicating that FLCs persist in the postnatal testes (Fig 2G-2J).

Expression of EGFP in a subpopulation of Ad4BP/SF-1 immunoreactive cells in the ovary
The expression of EGFP in the developing ovaries was examined using immunofluorescence. EGFP-positive cells were localized sporadically surrounding early-stage follicles at P7 (Fig 3A). The same pattern was seen surrounding the developing follicles at P14 and P21. In addition, there were EGFP-positive cells in the stromal region (Fig 3B and 3C). In the cycling ovary at the adult stage, the distribution of EGFP-positive cells surrounding follicles was essentially similar to that at the former stage, and EGFP-positive cells were assembled in the stromal region ( Fig 3D). The stromal region of mammalian ovaries such as rodents has been discovered to contain steroidogenic cells, and thus the region has been referred to as an interstitial gland. In summary, our observations strongly suggested that FLE is activated in not all, but rather a proportion of steroidogenic theca and interstitial gland cells in the ovary. 16.4±1.6% of Ad4BP/SF-1-positive cells were positive for EGFP in the adult ovary.
Ad4BP/SF-1 and 3β-HSD (3β-hydroxysteroid dehydrogenase) were used as markers for steroidogenic cells in the gonads. Indeed, Ad4BP/SF-1 and 3β-HSD were detected immunohistochemically in the interstitial cells of the fetal testes, and the signals were observed to overlap completely. The expression of Ad4BP/SF-1 but not 3β-HSD was detected weakly in Sertoli cells (Fig 3E). Similar to in the testis, the strong signals of Ad4BP/SF-1 and 3β-HSD overlapped in the theca and interstitial gland cells in the adult ovary ( Fig 3F). Additionally, the expression of 3β-HSD was detected weakly in the granulosa cells, while that of Ad4BP/SF-1 was undetectable in the follicles. Regulation of Ad4BP/SF-1 gene expression in developing follicular granulosa cells is still controversial. Its expression appears to be detectable in some follicles but undetectable in others. This difference is unlikely due to developmental stage of the follicle because similar size between follicles does not necessarily mean similar levels of expression of Ad4BP/SF-1, even though they may appear healthy but not atretic. Considering that Hsd3b1 gene expression is regulated by Ad4BP/SF-1, the absence of Ad4BP/SF-1 from the follicle where 3β-HSD is expressed suggests that a transcription factor other than Ad4BP/SF-1 would be involved in Hsd3b1 gene regulation. In this regard, it is interesting to note that LRH-1 (NR5A2), which potentially binds to the same sequence as Ad4BP/SF-1 does, regulates Cyp19 gene expression in ovarian granulosa cells [30,31]. Likewise, LRH-1 in place of Ad4BP/SF-1 may regulate Hsd3b1 in such follicles, as shown in Fig 3F. The expression of EGFP in the adult ovaries of mFLE-EGFP transgenic mice was compared with those of endogenous Ad4BP/SF-1 and 3β-HSD. As shown in Fig 3G-3I, the staining for EGFP overlapped with those for Ad4BP/SF-1 in the ovarian theca and interstitial gland cells. However, interestingly, Ad4BP/SF-1-positive cells were not always positive for EGFP, and EGFP single positive cells were not observed. When the expression of 3β-HSD was examined, a partially overlapped distribution between EGFP and 3β-HSD was observed again (Fig 3M- 3O). These results suggested that theca and interstitial gland cells consist of at least two different cell types: FLE is activated within one of them but not activated within the other.
As previously described, the expression of EGFP could be observed even in postnatal testes of mFLE-EGFP mice [19]. This persistent expression of EGFP was compared with those of Ad4BP/SF-1 and 3β-HSD in adult testes. As expected, both Ad4BP/SF-1 and 3β-HSD are expressed in all interstitial Leydig cells, while EGFP was expressed in a population of Ad4BP/ SF-1 (Fig 3J-3L) and 3β-HSD positive cells (Fig 3P-3R), indicating that FLCs are present in the adult testis. In conclusion, the gonads of the two sexes seems similar from the aspect that the steroidogenic cells consist of two cell types in terms of their enhancer usage of Ad4BP/ SF-1 gene. Comparison of reporter gene expressions driven by FLE and the whole genomic region of the Ad4BP/SF-1 gene Subsequently, we attempted to compare the reporter gene expression driven by FLE with that by the BAC clone containing the Ad4BP/SF-1 gene locus. Reporter gene expression driven by the BAC construct (Ad4BP-BAC-EGFP described in Materials and Methods) was expected to recapitulate endogenous Ad4BP/SF-1 gene expression, and indeed, as expected, EGFP expression was observed in the ovary, testis, adrenal gland, pituitary, and ventromedial hypothalamus (VMH) by fluorescence microscopy (Fig 4A-4E). These results are consistent with those previously obtained using a yeast artificial chromosome as an expression vector [32]. Immunohistochemical experiments for Ad4BP/SF-1 demonstrated the EGFP expression overlapped with endogenous Ad4BP/SF-1 in the testis (Fig 4F-4H), adrenal cortex (Fig 4I-4K), pituitary ( Fig  4L-4N), and VMH (Fig 4O-4Q). Although a clear signal was not detected by fluorescent microscopic in the spleen, expression in the endothelial cells of the splenic sinus was identified by immunofluorescence [33] (Fig 4R-4T).
The ovarian expression of EGFP in the BAC transgenic mice was investigated immunohistochemically. Strong EGFP expression was observed in the theca cells surrounding well-developed antral follicles (Fig 5A), preantral follicles (Fig 5B), and corpora lutea (Fig 5C). Likewise, interstitial gland cells were strongly stained with EGFP ( Fig 5B). In addition to these relatively strong expression signals, weaker expression was observed in granulosa cells and luteal cells (Fig 5A-5C). The distribution and strength of these expressions were consistent with the endogenous expression of Ad4BP/SF-1 (Fig 5D-5I). The expression of 3β-HSD was detected in theca cells, interstitial gland cells, and granulosa cells (with the exception of cumulus cells) of well-developed follicles, and luteal cells (Fig 5M-5O). The expressions of 3β-HSD in theca and interstitial gland cells overlapped well with those of EGFP, while the expressions of 3β-HSD in granulosa and luteal cells were stronger than those of EGFP (Fig 5J-5L and 5P-5R).
To compare the cells labeled with the BAC and mFLE constructs, we additionally established transgenic mouse line using mCherry as the reporter gene (mFLE-mCherry). Ad4BP-BAC-EGFP and mFLE-mCherry mice were crossed, and the resulting ovaries carrying the two transgenes were subsequently examined. The ovaries displayed fluorescence for EGFP and mCherry ( Fig  6A-6C). To localize precisely the mutual distribution of the transgenes, the adult ovaries were subjected to immunohistochemical analyses for EGFP and mCherry. As shown in Fig 6D-6I, all mCherry-positive theca and interstitial gland cells were positive for EGFP recapitulating endogenous Ad4BP/SF-1 expression, while the EGFP-positive cells were not always positive for mCherry. mCherry single positive cells were not observed.

Discussion
It has been established that mammals develop fetal and adult types of steroidogenic cells in the adrenal gland (the fetal and adult adrenocortical cells) [22,34,35] and testis (fetal and adult Leydig cells) [14,19,36]. In rodents, the fetal adrenal cortex starts to disappear around birth, and by puberty onset has completely disappeared in males, whereas it disappears completely during the first pregnancy in females [37]. Likewise, it has been suggested that fetal Leydig cells (FLCs) are replaced by adult Leydig cells (ALCs) in the early postnatal period, as discussed below. However, the ovaries are different from the male gonad, in that steroidogenic cells are not yet differentiated during the fetal period; indeed, no fetal theca and interstitial gland cells with an appearance resembling FLCs have thus far been reported. We previously identified FLE in the Ad4BP/SF-1 gene, and thereby established transgenic mice in which FLCs are labeled with EGFP (mFLE-EGFP) [20]. Tracing the labeled cells unexpectedly revealed that EGFP-positive FLCs are present until the adult period. This persistent presence of FLCs in the adult testis of rodents has been suggested by morphological studies to date [38,39]. Different from the testis, the ovary was believed to not harbor any fetal type of steroidogenic cells. Nevertheless, the present study demonstrated that EGFP-positive cells, in which FLE is activated, emerge in the postnatal ovaries. Immunohistochemical experiments suggested that these cells are a population of theca and interstitial gland cells. This finding was supported by comparing the expression of a reporter gene driven by an Ad4BP/SF-1 BAC clone (Ad4BP-BAC-EGFP). Based on the observations, it is highly likely that theca and interstitial gland cells consist of at least two cell populations, in one of which the FLE of Ad4BP/SF-1 is activated, while in the other a yet-unknown enhancer is. This differential enhancer usage between the two cell types bears strong resemblance to that between FLCs and ALCs, although the enhancer for ALCs has yet to be characterized.
Although it has not been reported that the ovarian theca and interstitial gland cells consist of multiple cell types in mammals, the presence of two types of theca cells has been described in the ovaries of medaka fish: one type expresses cyp17 while the other expresses cyp19 [40]. Considering the differential expression of the steroidogenic genes, the former cells seem to synthesize either androstenedione or testosterone as the final product, and the latter seem to synthesize estradiol. However, there has not been any report on mammals showing the Cyp19 gene to be expressed in theca and interstitial gland cells. Therefore, it is unlikely that the two types of theca cells found in this study correspond to those found in medaka fish. Nevertheless, it is still interesting to note that two types of theca cells have been found in these two evolutionarily distant animals.
Because we currently do not have any marker gene to distinguish these two cell types, it remains unclear whether other mammalian species develop them. Additionally, it is still unclear whether they are functionally differentiated or not. Addressing these issues would further our understanding of the functional significance of the two theca and interstitial gland cell types in the ovary.

Author Contributions
Conceived and designed the experiments: K. Miyabayashi K. Morohashi. Performed the experiments: K. Miyabayashi KT HO TB YS. Analyzed the data: K. Miyabayashi KT. Wrote the paper: K. Miyabayashi K. Morohashi.