| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
The Impact of the Human Genome on Endocrinology: Original Articles |
Department of Neurobiology and Physiology (S.A.P., T.K.W.), Northwestern University, Evanston, Illinois 60208; and Departments of Preventative Medicine (A.W.R.), Obstetrics and Gynecology (D.A.F.), and Medicine (T.K.W.), Northwestern University Medical School, Chicago, Illinois 60611
Address all correspondence and requests for reprints to: Teresa K. Woodruff, Ph.D., Department of Neurobiology and Physiology, Northwestern University, 2153 North Campus Drive, Evanston, Illinois 60208.
Abstract
The intraovarian function of gonadally produced activin is unclear, and many in vitro studies have suggested a role for activin in follicle development. To identify the follicular developmental stages at which these ligands may be acting, we have used immunohistochemical localization of the ligand subunits, receptor subtypes, and Smad co-activating proteins within the same follicles. The earliest stages of follicle development (primordial to primary) show no immunoreactivity for the activin subunits or their receptors. Oocytes from these early stages contain immunostaining for Smad2 and Smad4, consistent with signaling by other TGF-ß superfamily members. Immunostaining for the activin type II receptor first appears in oocytes and oocyte-associated cumulus cells at the secondary follicle stage. However, activin is not produced in these follicles, suggesting that either the receptors are inactive at this stage or they are used by another protein. Co-localization of activin and inhibin subunits, receptors, and Smads only occurs in granulosa and theca cells of small antral, aromatase-positive follicles as well as granulosa cells of early atretic follicles. In addition, multivariate statistical analysis reveals that the ligands and their cellular signaling complexes are independently regulated. Together, these data strongly suggest that the intraovarian role of activin is limited to a few developmental stages and that other TGF-ß family members may use this cell autonomous signaling machinery in early follicle development.
OVARIAN FUNCTION IS controlled by a number of endocrine-, paracrine-, and autocrine-acting factors including pituitary-derived gonadotropins and ovarian- produced TGF-ß-related proteins. The TGF-ß superfamily protein activin functions in a wide range of tissues in addition to being produced in the gonads (1). Originally identified as a secretagogue for pituitary FSH, activin also modulates erythroid cell differentiation, neuronal survival, and mesoderm induction (2, 3, 4, 5, 6). In the ovary, activin is thought to be involved in paracrine signaling during folliculogenesis (7). Activin increases follicular atresia in vivo and stimulates granulosa cell steroidogenesis, theca cell proliferation, follicular organization, and oocyte maturation in vitro (8, 9, 10, 11, 12). Dimerization of two ß subunits termed ßA and ßB gives rise to activin A (ßA-ßA), activin B (ßB-ßB), and activin AB (ßA-ßB). When co-expressed with a distantly related 
subunit, the dimeric protein inhibin [-ßA (inhibin A) or -ßB (inhibin B)] is generated. Inhibin and activin exert opposing effects in many cells, including pituitary gonadotrope FSH release and ovarian theca and testicular Leydig cell androgen production (13, 14, 15, 16).
Most members of the TGF-ß superfamily regulate cell function through membrane-bound heteromeric complexes of serine-threonine kinase receptors and intracellular Smad proteins (17). The activin signaling complex includes one of five type II ligand-binding receptors designated Act-RII, Act-RIIB, and three additional isoforms of Act-RIIB. Each receptor isotype has differing affinities for activin and inhibin (18, 19). The substrate for ligand-bound type II receptors is a type I serine-threonine kinase signaling receptor (20, 21, 22, 23). Although the type I receptors ActRIA [activin-like receptor kinase (ALK)-2] and ActRIB (ALK-4) associate with the activin type II receptor, ALK-4 is the predominant activin receptor whereas ALK-2 has been characterized as a bone morphogenetic protein (BMP)-associated receptor and Müllerian inhibiting substance type I receptor (24, 25, 26, 27). It is likely that combinatorial assembly or usage of various receptor subtypes affects activin signal transduction (28). The receptor system for inhibin is less clear. Inhibin may bind multiple receptors including the activin type II receptor ActRII, the TGF-ß type III receptor betaglycan, a pituitary proteoglycan (InhBP), or several unknown cell surface proteins (19, 29, 30, 31). Smad proteins are the cytoplasmic mediators of TGF-ß extracellular signals (32). Receptor-associated Smad2 and Smad3 are phosphorylated by activated activin and TGF-ß type I receptors. These proteins bind to the common mediator Smad, Smad4, and translocate to the nucleus to activate target genes (33, 34, 35).
Activin and inhibin subunit mRNAs are produced abundantly in the ovary. Human dominant follicles express and ßA but not ßB subunits, large preovulatory and small antral follicles express all three, and preantral follicles do not express any of the subunits (36, 37). The human subunit mRNA is localized to granulosa and theca cells of growing follicles whereas in nonhuman primates and rodents the expression of the subunit mRNA is restricted to the granulosa cells, with little to no expression in the theca and interstitial cells (36, 38, 39). In addition, activin receptor types I and II have been identified in the mammalian ovary. In the rat ovary, ActRII is the predominant form detected by RNA blot analysis and in situ hybridization (40, 41). The ActRII receptor localizes to oocytes, corpora lutea, theca, and granulosa cells (41). However, another study identified activin binding to rat ovaries restricted to newly recruited, growing, and Graafian follicles in granulosa, theca, and antral follicular fluid but neither to oocytes nor corpora lutea (42). Activin receptor localization or Smad localization in human ovaries has not been reported. However, RT-PCR analysis has detected mRNA for ActRII, ActRIIB, ActRI, and ActRIB in isolated human granulosa-luteal cells (43, 44), human oocytes, and cumulus cells (45). Also, ActRII, ActRIIB, and Smad2/Smad4 mRNA transcripts have been identified in human granulosa cells, ovarian surface epithelium, and ovarian cancer cell lines and in vitro fertilized oocytes (46, 47, 48).
Co-localization of the proteins involved in the complete activin signal transduction cascade in human follicles has not been reported. Understanding the function of activin requires the identification of overlapping patterns of expression for activin ligands, receptors, and co-activators in both timing and spatial distribution. Moreover, because the nuclear vs. cytoplasmic distribution of Smad proteins predicts Smad activity, immunolocalization of Smads in follicles may provide insight into cells that are fully active. Therefore, to determine the status of the activin signaling components in human follicles, we have co-localized 1) the activin/inhibin , ßA, ßB protein subunits; 2) the activin type II receptors ActRII and ActRIIB; and 3) Smad2 and Smad4 in normal human ovarian follicles. These were compared between follicles determined to be prerecruited, recruited by FSH, or undergoing atresia. Differential localization and expression of the subunits, receptors, and the Smad proteins was observed in these follicular stages. In addition, multivariate statistical analysis was performed to analyze relationships among the ligands, receptors, and signaling proteins in folliculogenesis not only between follicle developmental stages but also within follicle cell types. The results demonstrate significant changes in the activin signal transduction system in granulosa and theca cells during human follicle maturation and atresia.
Materials and Methods
Subject and sample preparation
Ovaries were obtained from women aged 32–43 yr undergoing prophylactic ovariectomy at Northwestern Memorial Hospital (Chicago, IL). The procedures were done in accordance with Institutional Review Board approval after consent of the patient and examination by the pathologist. The subjects were positive for a BRCA mutation but were not reported to have breast cancer at the time of the surgery. The specimens were analyzed for evidence of overt cancer or local dysplasia. No pathology, including epithelial inclusions, could be detected (49). Three to six nonoverlapping parallel sections were taken from each ovary and were fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Blocks were examined for the presence of small antral follicles 1–8 mm in diameter, and ovaries from five of eight women were determined to contain sufficient follicles for analysis. None of the blocks contained a dominant follicle (>11 mm) or a corpus luteum. The latter suggests that the ovaries were taken during the follicular phase of the cycle. Four-micrometer serial sections were taken on a microtome (American Optical Instruments, Buffalo, NY), mounted on Vectabond-coated slides (Vector Laboratories, Inc., Burlingame, CA), and air-dried. Duplicate or triplicate slides were analyzed in separate experiments.
Antibodies
Goat polyclonal antibodies against ActRII, ActRIIB, and ActRIB were purchased from R&D Systems (Minneapolis, MN) and used in immunohistochemical analysis at a final concentration of 2 µg/ml (anti- ActRII), 5 µg/ml (anti-ActRIIB), and 10 µg/ml (anti-ActRIB). Of two lots of ActRIB purchased from R&D Systems, only production lot COB01 was immunoreactive. A subsequent production lot of goat polyclonal antibody (lot COB020081) to ActRIB did not detect ActRIB in human ovaries or in ovarian cancer tissue that previously tested positive (R&D Systems; Jackie Kirk, personal communication). Rabbit polyclonal antibodies against the , ßA, and ßB subunits of inhibin (a gift from Wiley Vale and Joan Vaughn, Salk Institute, La Jolla, CA) were used at a concentration of 2 µg/ml (anti-), 3 µg/ml (anti-ßA), and 4 µg/ml (anti-ßB). Antisera against human placental p450 aromatase was purchased from the Hauptman-Woodward Medical Research Institute (Buffalo, NY) and used at a dilution of 1:600. Goat polyclonal antibodies against Smad2 and Smad4 were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) and used at a final concentration of 20 µg/ml (anti-Smad2) and 10 µg/ml (anti-Smad4). Biotinylated rabbit antigoat and goat antirabbit secondary antibodies were purchased from Vector Laboratories, Inc. (Burlingame, CA) and used at a final concentration of 5 µg/ml.
Immunohistochemistry
Slides were deparaffinized in xylenes and rehydrated through a graded ethanol series. Antigen retrieval was performed in Citra buffer (BioGenex Laboratories, Inc., San Ramon, CA) in a microwave for 2 min at high power and 7 min at low power. After cooling in the antigen retrieval solution, slides were washed in Tris-buffered saline [TBS; 500 mM NaCl and 20 mM Tris (pH 7.6)] and permeabilized in TBS containing 0.1% Tween 20 (TBS-T). Slides were incubated in 3% H2O2 to quench endogenous peroxidase activity, rinsed in TBS, and placed into staining racks (Shandon Industries, Pittsburgh, PA). Endogenous biotin and avidin binding was blocked with the Biotin-Avidin Blocking kit (Vector Laboratories, Inc.). Nonspecific binding was blocked by incubating slides in TBS containing 3% BSA and 10% serum from the host species of the secondary antibody. Serial sections were incubated with the primary antibody at room temperature for 1 h, washed in TBS-T, and incubated with 2.5 µg/ml biotin-labeled secondary antibody. After a 30-min incubation at room temperature, slides were washed in TBS-T and incubated in ABC reagent (Vector Laboratories, Inc.). Horseradish peroxidase was visualized with the diaminobenzidine reagent kit (Vector Laboratories, Inc.), which resulted in a brown precipitate, and then counterstained in hematoxylin (Harris-modified hematoxylin; Sigma-Aldrich Corp., St. Louis, MO). Background staining was determined by replacing the primary antibody with buffer.
Terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling (TUNEL) staining
Cellular apoptosis was detected by labeling DNA fragments using TUNEL with the ApopTag Fluorescein labeling kit (Intergen, Purchase, NY). Control slides were processed in parallel without the addition of the terminal deoxynucleotidyl transferase enzyme. Slides were visualized on a Nikon E300 Epi-Fluorescent microscope, and images taken were acquired using a SpotRT monochrome digital camera (Diagnostic Instruments, Sterling Heights, MI) and Metamorph Imaging Software (version 4.6; Universal Imaging, Downington, PA).
Follicle classification
Follicles were classified according to the following criteria: primordial, less than 25 µm oocyte with few squamous granulosa cells; primary, more than 25 µm oocyte with single layer of rounded granulosa cells; secondary, multiple layers of granulosa cells without formation of an antrum; early antral, less than 1 mm follicle diameter with a visibly forming antrum; healthy small antral, 1–8 mm follicle diameter with an intact granulosa cell layer with no evidence of cellular pyknosis; early atretic, 1–8 mm with some cellular pyknosis and disorganization of the granulosa cell layer; atretic, 1–8 mm with extensive cellular pyknosis and shedding of the granulosa cell layer into the antral cavity. None of the ovarian sections contained a dominant follicle (>10 mm in diameter). Size classification follows Roberts et al. (36).
Statistical analysis
All statistical analyses were carried out using the statistical package NCSS 97 (NCSS Statistical Software, Kaysville, UT). Principal components analysis (PCA) was performed only on follicle categories of "healthy small antral," "early antral," and "atretic" and in the two cell types (granulosa and theca) for each follicle. The granulosa cell and theca cell layers were scored independently for immunoreactivity within each follicle. A total of 62 follicles from 5 women were analyzed. Immunoreactivity was scored on an ordinal scale of 0 (absent) to 3 (intense staining) for each of eight variables (ActRII, ActRIIB, , ßA, ßB, Smad2, Smad4, aromatase). Missing data were given the mean value for the category based on follicle stage and amount of aromatase immunoreactivity. Immunoreactivity data for ActRIB was excluded from the analysis because half of the sections could not be analyzed due to the lack of an immunoreactive antibody. In the PCA, varimax rotation of the components was used. Transformation of the resulting factors by orthogonal rotation resulted in a better visualization of the factors while maintaining their original relationship. Two components with an eigenvalue of greater than 1 were retained. Each of these components was tested in independent ANOVAs for statistical significance in a 3 x 2 (follicle classification vs. cell type) fully crossed matrix.
Results
Healthy antral follicles
Follicles (1–8 mm diameter) were partitioned into those scoring positive for p450 aromatase immunoreactivity and those scoring negative for p450 aromatase immunoreactivity (Fig. 1A
and Fig. 2A). The presence of aromatase in granulosa cells is a hallmark of follicular selection by FSH (50). Representative immunoreactivity of the activin signaling components in nonaromatizing antral follicles is shown in Fig. 1. A summary of all follicles examined in these classes is presented in Table 1. Low levels of the type II receptors, ActRII and ActRIIB, are found in granulosa cells with little to no reactivity in theca cells (Fig. 1, B and C). These follicles have low levels of immunoreactivity for the , ßA, and ßB subunits of inhibin in both granulosa and theca cells and, thus, can make either inhibin A, inhibin B, activin A, activin B, or a combination of those proteins (Fig. 1, D–F). Immunostaining for ßA subunit was uniquely punctate and appears as granules in the cytoplasm of the granulosa cells (Fig. 1E). Both Smad2 and Smad4 immunoreactivity is present in granulosa cells. In the theca cell layer Smad4 is particularly evident. Some nuclear localization of Smad2 and Smad4 can be detected in granulosa cells, but most of the protein is cytoplasmic (Fig. 1, G and H). The theca cell layers show isolated cells with nuclear Smad4 immunostaining (Fig. 1, H and I).
|
|
|
and Table 2). Overall, there is an increase in antibody staining for granulosa cell receptor ActRII (Fig. 2B) and the subunits , ßA, and ßB (Fig. 2, D–F) in the follicle shown in Fig. 2 when directly compared with the aromatase-negative follicle described above and shown in Fig. 1, to which it shares adjacency. Comparing theca cells from p450 aromatase-negative follicles with p450 aromatase-positive follicles, there is an increase in theca cell immunoreactivity in positive follicles for the and ßB subunits and Smad2 and Smad4 (Fig. 2, D and F–H). Nuclear localization of Smad2 and Smad4 is found in granulosa cells, but cytoplasmic localization is maintained in the theca cells. No oocytes from healthy antral follicles were found in any section analyzed; thus, no information about their protein complement is presented here.
|
Many of the 1–8 mm diameter antral follicles were in various stages of atresia. Atresia was measured first by morphological criteria, then verified by TUNEL staining. The follicle depicted in Fig. 3 is undergoing early stage atresia characterized by disorganization of the granulosa cell layer and is characteristic of the majority of early atretic follicles examined. A summary of the immunoreactivity for receptors, subunits, and Smads is given in Table 2. High levels of ActRII staining in granulosa cells as well as ActRIIB staining in granulosa cells is evident (Fig. 3, A and B). Little activin receptor staining is seen in theca cells of the atretic antral follicles. Granulosa and theca cells were immunopositive for the and ßA subunits, and granulosa cells but not theca cells have the ßB subunit (Fig. 3, C–E). Smad2 and Smad4 antibody staining is readily apparent in the granulosa cells, with more cells positive for nuclear localization (Fig. 3, G, H, J, and K) than in healthy antral follicles. Theca cells lack Smad2 and have very little Smad4 antibody reactivity.
|
shows a representative follicle undergoing late stage atresia. Abundant pyknotic granulosa cells are evident and are found within the antral cavity. Summary immunoreactivity for the activin signaling components in these follicles is provided in Table 2. Most pronounced immunostaining is found for the ActRII receptor in the apoptotic granulosa cells. Most of these cells show low levels of ActRIIB and subunit antibody staining while maintaining ßA subunit and ßB subunit immunoreactivity. Smad2 and Smad4 immunoreactivity is also present and nuclear in location. Some nuclear localization of Smad2 and Smad4 can be seen in the theca interna. The granulosa cells in the antral cavity, as well as the theca interna layer bordering the antral cavity, are positive for DNA fragmentation by fluorescent TUNEL assay (Fig. 4, I and L). ActRII immunoreactivity can also be observed in oocytes of atretic follicles as well as the cumulus cells that surround the oocyte (Fig. 5, C and D).
|
|
Preantral follicles from primordial to secondary stage show no consistent staining for the subunit, ßA subunit, ßB subunit, or for ActRIIB in either the granulosa cells, theca cells, or oocytes (Table 2). Granulosa cells from these class of follicles are negative for ActRII; however, oocytes from secondary stage follicles display ActRII immunoreactivity (Fig. 5, A and B). Some cytoplasmic immunostaining can be noted in the granulosa cells immediately adjacent to the oocyte from secondary follicles (Fig. 5B). Most of the observed size classes of preantral follicles contain oocytes and granulosa cells that are immunopositive for Smad2 and Smad4. A sample secondary follicle stained for Smad2 and Smad4 is shown in Fig. 6. More Smad4 immunoreactivity is observed in the oocyte, granulosa, and theca interna than Smad2, in these follicles. In addition, Smad4 localization is restricted to the cytoplasm and not the nucleus of the oocyte (Fig. 6B). Immunoreactivity for Smad2 co-localizes with Smad4 but may be more abundant in the theca interna and externa whereas Smad4 is absent from the theca externa.
|
subunit (Table 1). However, these follicles do not show any immunostaining for either of the type II receptors and little to no immunoreactivity for the Smads (Table 1). PCA
To analyze the interrelationships between the ligand, receptor, and intracellular signaling molecules within cell types and between follicle developmental stages, a PCA was performed. In PCA, variables are measured and transformed into a linear composite of noncorrelated measures called principal components (or factors) that generate new variables. PCA can simplify a system of multiple variables by eliminating redundant variables that carry the same information. In addition, these new factors may uncover structure in the data set and are expected to represent underlying biological phenomena. We used data for receptor variables RII and RIIB; the ligand subunit variables , ßA, and ßB; and the intracellular Smad2 and Smad4. PCA was done on pooled data from all cell types between three categories of follicle development: early antral, healthy antral (p450 aromatase positive and negative), and atretic. Two factors that explain 68% of the total original variance were retained. These factors are displayed graphically in Fig. 7. The cell nonautonomous (inhibin and activin subunits) variables load highly positive on factor 2, and the cell autonomous variables (receptors plus Smads) load highly negative on factor 1. The factors are orthogonal to each other; thus, these data indicate that the ligands and the cellular signaling machinery are largely independent of each other. The designation "cell nonautonomous" will be used here to indicate factor 2, which is represented by the ßA, ßB, and subunits, whereas "cell autonomous" will refer to factor 1, which is represented by ActRII, Smad2, and Smad4.
|
, it can be seen that granulosa cells in healthy antral follicles have moderate amounts of the cell autonomous (factor 1, -0.36) and very high amounts of the cell nonautonomous (factor 2, 0.70) variables, whereas atretic follicles have very high expression of autonomous (factor 1, -0.70) and moderate nonautonomous (factor 2, 0.50) variables. Theca cells in early antral follicles have low values for cell autonomous and nonautonomous variables (factor 1, 0.95; factor 2, -0.68), as do atretic follicles (factor 1, 0.44; factor 2, -0.52), whereas healthy antral follicles have middle values for both factors (factor 1, 0.17; factor 2, 0.08). This suggests that in antral and atretic follicles, both granulosa cells and theca cells are capable of synthesizing and responding to activin. In other follicle types, one or both components (autonomous or nonautonomous) are not expressed or are expressed at low levels.
|
The present study describes the localization of the protein subunits that comprise the secreted ligands, activin and inhibin, as well as the cellular machinery known to transduce the activin signal, and attempts to characterize the biology underlying their regulation in the developing or dying follicle. Previous studies have localized the , ßA, and ßB subunits to normal and diseased human follicles (36, 37, 51). Our data are in general agreement with these studies, with a few minor exceptions. Granulosa cells from 1–8 mm antral follicles contain , ßA, and ßB similar to that described in Roberts et al. (36, 51), although the latter studies did not detect protein immunoreactivity for ßB or subunits in theca cells. A second study by Jaatinen et al. (37) also did not find mRNA for ßB in the theca layer. The immunostaining that we detect for inhibin subunits in atretic follicles is also similar to a previous study (37), but with some differences. Although that study did not detect any subunit in the granulosa cells, it did detect ßA subunit and subunit in theca cells. We show ßA subunit in the granulosa cells, no , but some ßA, and weak ßB in theca cells. These nuances probably reflect either sampling (i.e. degree of atresia or timing of selection) or methodological differences. However, the overall pattern is consistent: granulosa cells of small, aromatizing, antral follicles produce abundant subunits that may dimerize into inhibin and activin; humans differ from rats and primates by the ability of the theca cell layer to produce the subunit (38, 39). Following selection by FSH, there is an up-regulation of all subunits, while during atresia there is an overall decrease in the production of inhibin and activin (37, 52). In addition, we did not detect any consistent staining pattern in oocytes or granulosa cells before antrum formation.
The primary role of inhibin and activin is regulation of pituitary FSH, but it has yet to be proven in vivo whether both proteins have a defined role in folliculogenesis. Genetic disruption of the subunit genes is complicated by the inability to specifically target activin (i.e. disruption of the ß subunit results in deficiencies in both inhibin and activin), potential redundancies in signaling by the A and B isoforms, and confounding effects due to increased activin production in subunit knockout animals (53, 54, 55, 56). In addition, although there are reproductive defects in mice deficient in ActRII, those defects may be related to signaling deficiencies in the pituitary (55). Cell-specific knockout mouse models would dramatically increase our knowledge about the function of these proteins in the ovary. In the meantime, the co-localization of receptors, ligands, and the activated Smad proteins can support a role for these proteins in the ovary and define a stage in folliculogenesis in which they may be acting.
This study is the first to describe the in situ localization of activin receptors and Smads in human follicles. Previously, localization by in situ hybridization failed to detect activin receptors ActRII and ActRIIB in normal or polycystic ovary syndrome ovaries (51). Subsequent to that study, other methodologies have identified both receptor subtypes, and the Smads through which they signal have been identified in human (44, 45, 46) as well as other mammalian ovaries (41, 57, 58). We find that ActRII localizes to oocytes from secondary follicles through larger stages, although not all size classes of oocytes are represented in our sample. Oocytes from secondary follicles differ from those in smaller size classes (i.e. primordial and primary) because they have entered a growth phase as well as acquiring a zona pellucida. In our samples, ActRIIB is mainly restricted to the granulosa cells of small, healthy antral follicles. ActRIIB, Smad2, Smad4, and the broad expression of the inhibin and activin subunits overlap in these follicles, suggesting that if activin has a function in folliculogenesis, this is the relevant stage. Our sample does not include a dominant follicle; therefore, additional studies on the expression of these components are necessary. Additional studies should also include the local expression of follistatin, an activin-binding protein, because production of follistatin abrogates activin signaling and follistatin is expressed in ovarian follicles (45, 59, 60).
The broader distribution of ActRII without the expression of activin and inhibin subunits supports the use of this receptor by other TGF-ß superfamily ligands. Growth differentiation factor (GDF)-5, BMP-7, and Nodal all may act through the ActRII receptor (61). Expression of Nodal in the ovary is unknown. BMP-7 is expressed in theca cells from healthy, Graafian follicles but is undetectable in other follicle stages (62), so it is unlikely that BMP-7 is the ligand for RII in preantral oocytes as well as atretic follicles. GDF-5 is mainly involved in appendicular skeletal and joint development, and mouse mutants of GDF-5 are fertile, suggesting that there are no defects in follicle development (63, 64). Although it seems that oocytes contain Smad2 and Smad4, neither appears to be activated based on the distribution of the signaling proteins throughout the cytoplasm. We have not examined the distribution or localization of other Smad signaling proteins such as the BMP-responsive Smad1/Smad5/Smad8 proteins. Furthermore, the distribution of the ActRI receptors remains to be clarified. We were able to accomplish limited experiments using an antibody against ActRI that showed the majority of immunoreactivity occurs in granulosa cells of small aromatizing, as well as early atretic antral follicles. However, a subsequent lot (COBO2) of antibody was obtained and failed to show any immunoreactivity in our tissue samples or in others that previously tested for positive immunostaining (R&D Systems; Jackie Kirk, personal communication). In other follicle stages, it is certainly possible that human ActRII complexes with the BMP type I receptor BMPR-IB to signal through Smad1/Smad5/Smad8. However, in mouse, BMPR-1B is restricted to oocytes of follicles at the time of antrum formation as well as in oocytes and granulosa cells of antral follicles (65); therefore, it may be another type I receptor that is involved. Conclusions drawn from in vitro studies in rodents and bovids using recombinant activin suggest that activin is a key ligand in preantral follicle development (11, 66, 67, 68, 69). Our data for humans show no activin staining in preantral follicles, suggesting that activin does not play a crucial role at this stage in vivo. However, our data indicate that the receptor system is abundantly present at the preantral stages. This, then, suggests that in vivo another TGF-ß superfamily protein is the authentic ligand and in vitro activin behaves as a surrogate ligand. Whether downstream signaling for bound activin phenocopies the effects of a bound authentic ligand is unknown.
It is interesting that expression of the ActRII receptor in oocytes coincides with the known expression pattern in oocytes for GDF-9 and BMP-15 (70, 71, 72). Mice deficient in GDF-9 exhibit a block in folliculogenesis at the primary to secondary follicle transition (72). GDF-9 mRNA localizes to all oocyte stages, except primordial follicles (71, 72). Also, in situ hybridization detects almost no BMP-15 expression in primary follicles with increase in expression of BMP-15 in oocytes from secondary follicles that contain increasing numbers of granulosa cells (70). The receptor(s) for both BMP-15 and GDF-9 is unknown. Because ActRII is up-regulated in oocytes at the secondary follicle stage, perhaps ActRII plays a role, directly or indirectly, with GDF-9 or BMP-15 in follicle development at this transitional stage.
The signals that lead a follicle down the apoptotic pathway are just as important as the signals that permit survival of the dominant follicle. In granulosa cells undergoing apoptosis, ActRII is up-regulated, the activin and inhibin subunits are down-regulated, and the activation of Smads is high (as determined by nuclear localization). This result suggests that late cellular apoptosis is either driven by another TGF-ß superfamily, or is a consequence of activin signaling earlier in atresia. In the early stages of atresia, the inhibin and activin subunits are not down-regulated and the receptors are expressed, albeit to a lesser extent than in later stages. Activin and TGF-ß are known to inhibit cell cycle progression and induce apoptosis in a variety of cell types, thus it is plausible that these ligands (or other siblings) regulate follicular atresia (73, 74).
We expected to observe correlations between the ligands and the receptors they use during follicle development. By using multivariate statistical analysis on the suite of components that have been proposed to comprise the activin signaling system, we observe the opposite: an uncoupling of the activin cell autonomous and the nonautonomous factors. This uncoupling suggests that each of these components is independently regulated. It does not appear that the uncoupling is due to paracrine signaling (i.e. that one cell type makes the ligands and the other cell type expresses the receptors), unless the signaling is between follicle stages. The uncoupling is likely due to the use of the signaling molecules by other members of the TGF-ß superfamily. The TGF-ß superfamily consists of many more ligands than receptor types, and many ligands are known to bind multiple receptor pairs. The challenge in understanding folliculogenesis is separating out the many variant sensing and response pathways that may change dramatically as a follicle develops. The difficulties of this analysis are that the ligand is a dimer of subunits, the receptor is a complex of two different transmembrane proteins, and the cytoplasmic co-activators are common mediators of several ligands. The analysis here is the first of its kind in the human to evaluate the compartmentalization of these components in follicles of different size classes. This kind of analysis will provide a wealth of detail regarding available activin signaling throughout follicular development.
In conclusion, the proteins that comprise the inhibin and activin ligands, as well as the proteins that constitute the activin cellular signaling machinery, seem to be independently regulated during folliculogenesis. Independent regulation may allow for strict control of cell signaling by activin as well as the broad use of receptors by other TGF-ß superfamily proteins. Additional studies are needed to determine the availability and overlap of these other ligands with the known sets of receptors.
Acknowledgments
We thank Dr. Michael Pins and the Department of Pathology at Northwestern Memorial Hospital (Chicago, IL) for collection, embedding, and sectioning of the human ovaries as well as histological assessment. We are grateful to Dr. Richard Meindl (Kent State University, Kent, OH) for statistical advice. We thank the Vale Laboratory (Salk Institute, La Jolla, CA) for antibodies to the inhibin/activin subunits. We thank Kendall Carlin for additional sectioning. Additionally, we thank Drs. Herman Dierick (The Neuroscience Institute, San Diego, CA) and Robert Holmgren (Northwestern University, Evanston, IL) for helpful discussions and critical reading of the manuscript.
Footnotes
This study was supported by NIH Grants HD035708 and HD037096 (to T.K.W.). S.A.P. is a fellow of the Northwestern University Program in Endocrinology, Diabetes, and Hormone Action (Grant DK07169). T.K.W., A.W.R., and D.A.F. are members of the Robert H. Lurie Comprehensive Cancer Center of Northwestern University.
Abbreviations: ALK, Activin-like receptor kinase; BMP, bone morphogenetic protein; GDF, growth differentiation factor; PCA, principal components analysis; TUNEL, terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling.
Received October 22, 2001.
Accepted January 12, 2002.
References
-hydroxylase expression in human ovarian thecal-like tumor cells. J Endocrinol 148:213–221[Abstract]
, ß A and ß B messenger ribonucleic acids in the normal human ovary and in polycystic ovarian syndrome. J Endocrinol 143:127–137[Abstract]
-Inhibin is a tumour-suppressor gene with gonadal specificity in mice. Nature 360:313–319[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  T.-V. Do, L. A. Kubba, H. Du, C. D. Sturgis, and T. K. Woodruff Transforming Growth Factor-{beta}1, Transforming Growth Factor-{beta}2, and Transforming Growth Factor-{beta}3 Enhance Ovarian Cancer Metastatic Potential by Inducing a Smad3-Dependent Epithelial-to-Mesenchymal Transition Mol. Cancer Res., May 1, 2008; 6(5): 695 - 705. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. E. Telfer, M. McLaughlin, C. Ding, and K. J. Thong A two-step serum-free culture system supports development of human oocytes from primordial follicles in the presence of activin Hum. Reprod., May 1, 2008; 23(5): 1151 - 1158. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Kipp, S. M. Kilen, T. K. Woodruff, and K. E. Mayo Activin Regulates Estrogen Receptor Gene Expression in the Mouse Ovary J. Biol. Chem., December 14, 2007; 282(50): 36755 - 36765. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Q. Li, R. Kumar, K. Underwood, A. E. O'Connor, K. L. Loveland, J. S. Seehra, and M. M. Matzuk Prevention of cachexia-like syndrome development and reduction of tumor progression in inhibin-deficient mice following administration of a chimeric activin receptor type II-murine Fc protein Mol. Hum. Reprod., September 1, 2007; 13(9): 675 - 683. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Kipp, S. M. Kilen, S. Bristol-Gould, T. K. Woodruff, and K. E. Mayo Neonatal Exposure to Estrogens Suppresses Activin Expression and Signaling in the Mouse Ovary Endocrinology, May 1, 2007; 148(5): 1968 - 1976. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. G Knight and C. Glister TGF-{beta} superfamily members and ovarian follicle development. Reproduction, August 1, 2006; 132(2): 191 - 206. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Yang and S. K. Roy Transforming Growth Factor B1 Stimulated DNA Synthesis in the Granulosa Cells of Preantral Follicles: Negative Interaction with Epidermal Growth Factor Biol Reprod, July 1, 2006; 75(1): 140 - 148. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
