Cellular communication between cells occurs via endocrine, paracrine, or direct interactions. Direct cell–cell interaction results in the activation of signaling pathways in which information from the cell’s external contacts is relayed into the cell to regulate gene expression in response to changing external contacts between cells. These interactions are mediated by cell adhesion molecules and tight junction proteins. Other signals are conveyed directly between the cytoplasms of adjacent cells via intercellular channels, referred to as gap junctions, that connect the cytoplasms of adjacent cells [56–58].

Gap junctions are comprised of a family of proteins that includes connexins, innexins, and pannexins [59]. Connexins and innexins are implicated in the formation of gap junctions between adjacent cells that are present in vertebrate and invertebrate species, respectively. Pannexins are also present in vertebrates and share homology with the innexins. However, unlike the innexins, pannexins do not form intercellular channels but rather cellular pores that are implicated in paracellular communication via the release of ATP [59]. Connexins are transmembrane proteins that oligomerize to form hemichannels or connexons [60, 61]. There are 21 different connexins named according to either their molecular mass or sequence homology. The connexons that form the gap junction can be composed of either the same connexin or of different connexins. The composition of the connexon will determine the charge of the pore and the types of molecules which are able to pass through the pore.

In all mammalian species studied to date, four to six different connexins are expressed in the ovary [56, 57]. In all species, Cx43 appears to be the most prevalent connexin and is expressed in granulosa cells (Fig. 7.2), while Cx37 is expressed by the oocyte [56]. Certain studies have suggested that Cx43 is also present between the oocyte and the cumulus/granulosa cell layer [62], although others have failed to observe Cx43 between these cell layers [63, 64]. Veitch [64] claimed that gap junctions between oocytes and granulosa cell layer in fact comprised Cx37. In support of this, Cx37 knockout mice exhibit arrested folliculogenesis at the early antral stage [63] and oocytes fail to become competent [63, 65]. The role of gap junctional communication between the oocyte and the granulosa cells may involve regulation of cAMP transfer. Inhibition of gap junctional communication using heptanol or carbenoxolone results in decreased levels of cAMP in the oocyte [66, 67].

The regulation of Cx43 in rodent granulosa cells is mediated by FSH/LH [68, 69]. FSH appears to play an important role early in follicular development and positively regulates the expression of Cx43 [68, 70]). However, prior to ovulation, when LH levels increase, there is a decrease in Cx43 [71]. This decrease appears related to the LH surge which inhibits Cx43 via the MAPK and PKA signaling pathways [72].

Cx43 knockout mice die shortly after birth [73]. These animals, however, have fewer oocytes at birth [74]. If the oocytes from Cx43 knockout mice are placed under the kidney capsule of a wild-type mouse, or cultured in vitro, folliculogenesis can proceed to the primary follicle stage, but subsequent development is impaired [74]. Similarly, in mice harboring an autosomal dominant point mutation within the Gja1 gene (Gja1 ^Jrt/+ mice) which codes for Cx43, reduced granulosa cell coupling (0–20 % versus wild type) and impaired fertility are observed [75, 76]. This subfertility was due, at least in part, to impaired follicular development, with fewer pre-ovulatory follicles present in ovaries and reduced gonadotropin response. In addition to regulating folliculogenesis, Lin et al. (2003) have shown an inverse correlation between follicular apoptosis and Cx43 levels . Furthermore, the glucocorticoid dexamethasone, known to decrease apoptosis in ovarian follicles, increases levels of Cx43 [77, 78]. Interestingly, as Cx43 levels decrease in apoptosis, there is an increase in Cx32 levels. It has been suggested that Cx32 gap junctions propagate the death signal in the ovary [79, 80]. Thus, toxicants that stimulate a switch from Cx43 to Cx32 in the ovarian follicles may be useful for predicting ovarian apoptosis, thereby highlighting the importance of examining their expression and localization in toxicological studies. Whether or not toxicants like tributyltin (TBT), which induce apoptosis in the ovary during development, mediate changes in Cxs remains to be shown, even though TBT has been shown to decrease Cx43 in developing testes [81].

Toxicology data have also suggested a role of Cx43 in steroidogenesis. Exposure of rats to the environmental pesticide lindane blocks gap junctional communication [82]. In these animals there is also a concurrent decrease in the levels of P450ssc, the cholesterol transporter StAR, and a decrease in progesterone secretion, supporting the notion that Cx43 can influence steroidogenesis.

In vitro culture of neonatal (postnatal day 0–4) rodent ovaries has been used to examine mechanisms underlying both ovarian follicular formation and development and toxicity of xenobiotics causing ovarian follicle loss [83–86]. Methodology involves isolating intact ovaries from young mice, rats, or hamsters and floating them on small pieces of nitrocellulose or other membrane which is in turn floating on the surface of culture medium [87, 88]. The majority of primordial follicles present remain dormant, but a slow steady number of follicles are able to develop normally to the secondary stage (multiple layers of granulosa cells) without added growth factors or hormones. Addition of growth factors or inhibitors can improve or inhibit follicle development or activation [86, 89]. Studies on direct toxic effects of drugs or pollutants have utilized this methodology as well. Follicles can be enumerated and staged by histology, gene expression can be monitored by either microarray or reverse transcription-polymerase chain reaction, and proteins can be localized by immunohistochemistry/immunofluorescence or quantified by Western blotting in spite of the small size of the tissue.

The role of DNA damage in the loss of primordial and primary follicles was examined, using cyclophosphamide as an example [28, 90]. Cyclophosphamide and other alkylating drugs induce DNA adducts, DNA double-strand cross-links, leading to DNA double-strand breaks [91]. In vivo studies determined that a reactive metabolite, phosphoramide mustard (PM), must be produced for follicle loss to occur [92]. Exposing cultured mouse ovaries to various concentrations of different metabolites of cyclophosphamide confirmed that only metabolites that could produce PM induced concentration-dependent loss of primordial and primary follicles [28]. TUNEL and cleaved Caspase 3 staining were used as markers of follicle injury or damage. Increased numbers of follicles with TUNEL staining were observed for 10 μM at 24 h and 3 μM at 48 h after exposure, suggesting that these markers are either not as sensitive as the follicle counts done 8 days after exposure or are time dependent, a more likely scenario. It was then determined whether or not DNA damage is an underlying factor in PM-induced ovarian follicle loss. Immunohistochemistry for phospho-histone H2AFX, a histone which becomes phosphorylated at sites of double-strand breaks [93], was performed to examine whether or not double-strand breaks can be observed in cultured rodent ovaries in response to PM [90]. Phospho-H2AFX staining was detected in a punctate pattern (foci) predominantly in oocytes of both cultured mouse and rat ovaries exposed to PM. Number of foci per oocyte and numbers of follicles with foci were both concentration dependent, peaking at 24–48 h with 3 and 10 μM. In another study, DNA cross-links in granulosa cells were observed and peaked 2 h following a single dose of CPA in immature Sprague–Dawley rats [94]. These results supported DNA damage as a mechanism of ovarian damage. Unfortunately, other immunohistochemical markers of DNA damage and repair remain to be developed for animal models, and more research is needed to further investigate oocyte-specific DNA damage signaling and repair capabilities in oocytes, as a unique target for DNA-damaging agents. In vitro work investigating VCD has examined the role of the KIT/KITL, PI3K, and FOXO3A signaling pathway in VCD-induced follicle loss. Initial evidence that follicle activation is affected by VCD was seen when a single dose of VCD induced an increase of primary follicles in rats [95]. Subsequent to that, global gene expression analyses of cultured rat ovaries and with isolated ovarian follicles identified cKit as a gene altered in both systems [96]. Addition of KITL, but not GDF9 or BMP4, to cultures partially protected against VCD-induced follicle loss. The PI3 kinase inhibitor, LY294002, inhibited follicle activation to primary follicles in cultured rat ovaries, and primordial follicles were not affected by VCD in the presence of LY294022, although numbers of small primary follicles were still reduced by VCD [97]. In cultures of postnatal day 4 rat ovaries, exposure to 30 μM VCD led to decreased immunofluorescence for KIT and FOXO3 protein in oocytes [98]. Decreased staining for phospho-AKT in oocyte nuclei was the earliest observed alteration on day 2 following VCD exposure.

Ovarian follicle counting is a necessary process for evaluating ovarian toxicants. Due to the relatively arduous process of determining follicle numbers during reproductive toxicology studies, automated or semi-automated methods have been sought. Typically for rodent studies, a thorough investigation of follicle numbers involves manually counting and determining the stage of all follicles in every 20th or 40th section of ovaries, following complete sectioning of one ovary per animal that has undergone standard processing and hematoxylin and eosin staining. Oocyte-specific markers were tested to either improve visibility of follicles for manual counting or for computer-assisted enumeration. PCNA has been tested, because oocyte nuclei stain darkly; however, proliferating granulosa cells are also stained and can be used to monitor granulosa cell proliferation and follicle health [99]. In spite of this lack of specificity, Muskhelishvili et al. [100] showed reduced variability in interindividual follicle counts with PCNA immunohistochemical staining. Furthermore, Picut et al. [101] demonstrated improved consistency and higher follicle numbers when PCNA-stained sections were counted manually, with a high correlation between manual counts and semi-automated computer-assisted counting. Cytochrome P450 1B1 immunostaining has also been reported to specifically stain oocyte nuclei and improve visibility of ovarian follicles for counting [102]. We have examined immunofluorescence of zona pellucida proteins (P Devine, Fig. 7.3, unpublished observations), using a specific antibody for this available from ATCC, originally developed by Jurrian Dean [103]. ZP3 are only present at the outer surface of oocytes in follicles of adults, including primordial oocytes although faintly (not shown). In adult mice, ZP3 is also observed in follicle remnants, with the zona pellucida forming sinuous structures within the interstitium (P Devine, Fig. 7.3, unpublished observations). This would represent a potential fluorescent marker that could be used to enumerate and even stage follicles based upon oocyte size if the empty zona pellucidae could be removed from analysis. It could be envisaged that transgenic rodents expressing a fluorescent protein under the regulation of oocyte-specific genes could be developed for specific use in reproductive toxicology studies; however, the cost of such a model and the effort required to produce it have precluded such efforts.

In order to identify healthy versus atretic follicles, TUNEL staining or immunohistochemistry for cleaved Caspase 3 has been typically performed to identify apoptotic granulosa cells in atretic follicles [28, 104]. Conversely, immunohistochemistry for PCNA or BrDU incorporation of proliferating granulosa cells has been examined for healthy follicles with proliferating granulosa cells (e.g., [105]). Immunohistochemistry staining for Anti-Mullerian hormone has also been used to identify non-atretic antral follicles when cisplatin was compared with vehicle only [104]. These markers tend to be qualitative for follicle atresia or health, but they complement, or confirm, histological features such as apoptotic bodies or mitotic figures in secondary or antral follicles.

Ovarian tumors are generally rare in rodent species. We know from numerous studies, with different tissues, that changes in the levels of proteins implicated in cell–cell interactions can undergo dramatic changes in either tumorigenic cells or during the process of carcinogenesis . Intercellular gap junctional communication is known to be consistently downregulated in carcinogenesis, and this can be induced by chemical carcinogens. For example, in the liver of rats treated for 5 days with hexachlorobenzene (HCB), the expression of the gap junction proteins, Cx32 and Cx26, remains significantly lower 50 days after the beginning of the treatment, and GJIC is still decreased 100 days later, but only in female rats [106]. Interestingly, there is a preponderant increase in cancer formation in these female rats when they are subsequently given diethylnitrosamine (DEN), while connexins in the male are not affected by HCB and males develop significantly fewer tumors in response to DEN [106]. In a study using several ovarian cancer cell lines, most of the cell lines displayed reduced expression of connexins [107]. In granulosa cell tumors, it has been reported that Cx43 is decreased or absent. Interestingly, in cells with decreased Cx43 there is an increase in Cx32, which is normally absent in the normal ovary [108]. Thus, the regulation and interplay between these two connexins provide an interesting set of makers which may be useful in understanding chemical-induced carcinogenic transformation in the ovary. In another study, cisplatin treatment of ovarian cancer cells resulted in the formation of both sensitive and resistant cells. Downregulation of p53 in both cell types with siRNA resulted in downregulation of several genes including Cx43. In fact, cells in which Cx43 expression was decreased showed increased drug resistance to cisplatin [109]. Whether or not drugs that induce transporters, such as those of the ABC transporter family, have lesser effects in cells when Cx43 is downregulated remains to be established. However, dual labeling of connexins and certain known transporters may provide important information on potential drug/chemical toxicity in the ovary.

FSH receptor knockout mice (FORKO ) develop ovarian tumors after 12 months of age [110]. These animals are hypergonadotropic but display hypogonadism and have been used as an animal model for menopause. After 12 months of age, FORKO mice develop different types of ovarian tumors, including granulosa cell tumors and serous papillary epithelial adenomas [110]. Ovaries of FORKO mice show increased levels of the tight junction protein claudins (Cldns) 3 and 4, as well as dramatic increases in Cldn11; Cldn1, on the other hand, is repressed [111]. These animals also display steroid hormone imbalance, with high levels of testosterone and low levels of progesterone [111]. These observations are important with regard to identifying effects associated with long-term exposure to steroid-mimicking endocrine-disrupting chemicals, particularly putative androgenic chemicals such as phthalates [112], or other endocrine-disrupting chemicals that alter the production of gonadotropins. Endocrine-disrupting chemicals are molecules that can mimic or inhibit the action of endogenous hormones on target tissues, or alter either hormone secretion or the synthesis of their receptors [113]. Thus, monitoring the expression levels and immunolocalization of ovarian Cldns or connexins represent strong histological biomarkers to evaluate the effects of these chemicals and whose changes may be associated with infertility or rendering the ovary to a precancerous state.

Over the past 30 years, many studies have demonstrated that mammary glands develop earlier in women from many countries, as compared to the 1930s–1940s [114–117]. Estrogens , secreted by the ovaries under the influence of the follicle-stimulating hormone (FSH) and the luteinizing hormone (LH), are essential for mammary gland development during puberty [118, 119]. However, precocious breast development is not associated with an earlier menarche, implying an action of estrogens without the activation of the hypothalamic–pituitary axis [120]. It has been suggested that environmental factors mimicking estrogens actions may be responsible for those perturbations [120–122]. Because endocrine disruptors are widespread in the environment, and because their effects at low concentrations can have major impacts on the tightly orchestrated hormone-dependent development of mammary gland, exposure to endocrine disruptors is particularly disconcerting for women [113, 123, 124]. Mice models are typically used to assess developmental defects induced by chemical exposure or genetic modifications on mammary gland development. While whole mounts have been used to visualize gross developmental defects in mammary glands, the recent development of molecular markers, as well as new in vitro models, allows the assessment of subtle changes at the cellular and molecular levels, particularly changes in cell–cell interactions.