| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
a Department of Zoology, University of Guelph, Guelph, Ontario, Canada N1G 2W1
ABSTRACT
Apoptosis is a fundamental mechanism in follicular atresia and postovulatory regression in mammals, but its role in teleost ovarian function is currently unknown. This study tested the hypotheses that apoptosis mediates follicular atresia in teleosts and is inducible in vitro by incubation in serum-free conditions. Vitellogenic follicles from rainbow trout (Oncorhynchus mykiss) and goldfish (Carassius auratus) were incubated overnight in serum-free medium and examined for apoptosis by 3'-end-labeling and/or TUNEL analysis. Primary, postovulatory, and oocytectomized vitellogenic trout follicles and atretic goldfish follicles were evaluated in similar fashion. Overall, goldfish follicles had lower levels of DNA fragmentation than trout follicles. The DNA fragmentation in atretic goldfish follicles was similar to that measured in healthy vitellogenic and prematurational follicles; DNA fragmentation did not change after incubation. In the trout, postovulatory and oocytectomized vitellogenic follicles showed significantly greater in vitro susceptibility to apoptosis than intact vitellogenic follicles, whereas primary follicles were least susceptible. The TUNEL analyses revealed that in trout vitellogenic follicles, more thecal/epithelial cells than granulosa cells showed fragmented DNA in vivo, but incubation (24 h) did not result in increased apoptosis in cells of either type. These results indicate that apoptosis is involved in normal ovarian growth and postovulatory regression in teleosts, but that it does not appear to be an early event in teleost follicular atresia.
apoptosis, follicle, granulosa cells, ovary, theca cells
INTRODUCTION
In the mammalian ovary, a vast majority (75%–99.9%) of the finite pool of follicle-enclosed oocytes (follicles) endowed during embryonic development is eliminated before ovulation in a degenerative process known as atresia [1]. Atresia is a highly regulated process within the vertebrate ovary, and it is believed to be essential for the maintenance of ovarian homeostasis [2]. Data from numerous studies conducted during the last decade are strongly suggestive that apoptosis, or programmed cell death, is the underlying molecular mechanism of atresia in mammals [3, 4]; complementary studies using avian models are suggestive that within vertebrates, this cell-death process may represent a conserved mechanism by which somatic and germ cells are selectively removed from the ovary [5].
In the mammalian ovary, apoptotic death of follicular cells during atresia can occur at all stages of follicular development and, in most cases, appears to be initiated within the granulosa cell population [6]. Granulosa cell apoptosis precedes morphological signs of atresia and currently ranks among the earliest indicators of imminent follicle demise [7]. Apoptosis-mediated atresia of ovarian follicles is suggested to be responsible for the elimination of subordinate follicle-oocytes, ultimately leading to the selection of a dominant follicle, or follicle cohort, for ovulation [4, 8]. Numerous in vivo and in vitro studies during the last decade have shown this process to be under the finely co-ordinated influences of a diverse array of gonadotropins, steroid hormones, growth factors, and cytokines [4, 9].
To our knowledge, however, comparatively little effort has been directed toward studying atresia in other vertebrates, although atretic follicles have been morphologically and histologically described in the ovaries of reptilian, amphibian, and piscine species [10]. The role of atresia within the context of normal ovarian development in these organisms remains speculative, although Macer [11] provides evidence that in marine teleosts, atresia may function as a mechanism for optimizing fecundity.
Even less attention has been directed toward identifying the cellular and molecular mechanisms underlying atresia in teleost fish. Notable exceptions are a study by Janz and Van Der Kraak [12], which demonstrated apoptosis in preovulatory ovarian follicles of rainbow trout (Oncorhynchus mykiss) both in vivo and after short-term (24-h), in vitro incubation in serum-free conditions, and a study by Janz et al. [13], which documented increased apoptosis in ovarian follicles of white sucker (Catostomus commersoni) chronically exposed to effluent-contaminated waters. Whereas these studies suggest that apoptosis plays a role in teleost ovarian function, a direct link between apoptosis and atresia in the teleosts has yet to be demonstrated.
The link between apoptosis and follicular atresia is well described in the mammalian ovary [4], but fundamental differences in ovarian structure and function call into question our ability to universally apply insights gained from mammalian studies to the teleost model. For example, teleosts typically display higher fecundity than other vertebrates, recruiting many thousands of oocytes into the ovulatory pool each spawning season. These may be recruited in synchronous, group synchronous, or asynchronous fashion, resulting in one to many spawning episodes per season. Once recruited to the ovulatory pool, each oocyte typically sequesters many times its volume in hepatic-derived lipoproteins (yolk) during the growth phase (vitellogenesis). Consequently, teleost oocytes do not develop an antrum but, instead, retain a monolayer of granulosa cells immediately surrounding the oocyte zona radiata. In addition, teleosts represent one of only two vertebrate groups (the other being amphibians) for which fecundity is potentially indeterminate (see Tyler and Sumpter [14] for a comprehensive review of teleost ovarian development).
We currently have no understanding of how the unique aspects of teleost ovarian structure and function influence patterns of ovarian cell death. The study described herein had two main objectives. First, we wished to investigate the susceptibility of teleost ovarian follicles to apoptosis at different stages of development, both in vivo and during in vitro incubation in serum-free conditions, using both DNA 3'-end-labeling (ladder) and in situ (TUNEL) methodologies. An important aim of this part of the study was to investigate the relationship between morphological observations of atresia and molecular indicators of ovarian follicle apoptosis. To achieve this objective, we selected two species of teleost fish that display differing patterns of ovarian development and differential susceptibilities to atresia: the group synchronous-spawning rainbow trout, a species that displays a low susceptibility to atresia when reared under laboratory conditions [14]; and the asynchronous-spawning goldfish (Carassius auratus), in which the susceptibility to atresia is comparatively higher under laboratory holding conditions.
A second objective was to investigate in situ the relative susceptibilities to apoptosis of the two major follicular cell types, granulosa and theca, both in vivo and after in vitro incubation in serum-free conditions. For this portion of the study, the vitellogenic rainbow trout follicle was selected as the experimental model due to the ease with which distinct cell populations could be identified microscopically. With these investigations, we hope to establish a useful working model that will shed further light on the role of apoptosis in teleost ovarian function.
MATERIALS AND METHODS
Experimental Animals
Virgin, prespawning adult rainbow trout (Table 1) were obtained from the Alma Research Station (University of Guelph). These fish were from a fall-spawning stock and were collected on successive occasions (5, 3, and 1 mo as well as 1 wk) before predicted spawning. Two cohorts of follicles can readily be identified in the ovary of this species. One cohort consists of growing (vitellogenic) follicles that have been recruited for ovulation within the current season; the other consists of dormant (primary) follicles destined for growth and ovulation in the season to follow. The fish were held in the Hagen Aqualab at the University of Guelph in fiberglass tanks (2-m diameter) receiving flow-through well water (10 ± 1°C). The fish were maintained under a photoperiod simulating seasonal conditions and fed a commercial trout diet (Martin Mills, Elmira, ON, Canada) to satiation 3–4 days per week.
|
Common goldfish were purchased from Ozark Fisheries (Stoutland, MO). The ovary of this species develops asynchronously, yielding a heterogeneous population of developing follicles within the ovary of a single fish, including morphologically identifiable [10] atretic follicles. The fish were maintained indoors in fiberglass tanks (1.22-m diameter) with flow-through water at 16 ± 2°C under a constant photoperiod (14L:10D). These fish were fed a commercial trout diet daily to satiation.
All experiments were conducted in accordance with animal use protocols as established by the Animal Care Committee at the University of Guelph.
Chemicals
Bovine serum albumin, glucose, streptomycin sulfate, phenol:chloroform:isoamylalcohol (25:24:1, v/v), chloroform:isoamylalcohol (24:1, v/v), propidium iodide (PI), Proteinase K, and tRNA (from baker's yeast) were obtained from Sigma (St. Louis, MO). Terminal deoxynucleotidyl transferase (TdT; from calf thymus, 25 U/µl), DNase I, RNase (DNase-free, from bovine pancreas), and TUNEL assay reagents (In Situ Cell Death Detection Kit-AP) were obtained from Roche Molecular Biochemicals (Laval, PQ, Canada). Agarose (electrophoresis grade) and 123 base-pair (bp) ladder standard DNA were purchased from Gibco BRL (Burlington, ON, Canada). The 
-32P-dideoxyATP (-32P-ddATP; 3000 Ci/mmole) was obtained from Amersham (Montreal, PQ, Canada).
Preovulatory Follicle Incubations
In vitro incubations were conducted with isolated goldfish and rainbow trout ovarian follicles at different stages of development. Goldfish follicles were characterized by size (vitellogenic [GV], 0.4–0.8 mm in diameter; prematurational [GPM], 0.9–1.1 mm in diameter) as measured under a stereomicroscope equipped with an ocular micrometer; both vitellogenic and prematurational follicles were collected from a single fish for individual experiments. Vitellogenic rainbow trout follicles (TV) were initially characterized according to temporal proximity to predicted spawning (5 mo prespawning [TV5], 3 mo prespawning [TV3], or 1 mo prespawning [TV1]); primary trout follicles (TPr) were readily identifiable by their small size (
1 mm in diameter) and opaque, whitish coloration. For both species, care was taken to ensure that follicles selected for incubation appeared healthy (e.g., uniform shape, color) under a stereomicroscope.
At the time of sampling, fish were killed by cervical transection, and the ovaries were quickly removed, weighed, and placed in ice-cold Cortland's saline (pH 7.6) supplemented with 0.1% (w/v) BSA, 0.1% (w/v) glucose, and 0.01% (w/v) streptomycin sulfate. A subsample of whole ovarian tissue (0 h) was immediately fixed in buffered paraformaldehyde for in situ analysis. Using fine forceps (trout) or gentle aspiration (goldfish), individual follicles were separated from the stromal tissue in a glass Petri dish containing cold Cortland's saline; care was taken to remove residual stromal tissue from follicle surfaces. Duplicate (GV, GPM, and TPr) or triplicate (TV) subsamples of follicles were either snap-frozen in liquid N2 (0 h) or incubated [15] in 24-well polystyrene tissue culture plates (Costar, Corning, NY) containing 1 ml of Cortland's saline for 24 h at 12 ± 1°C (trout) or 18 ± 1°C (goldfish). Incubation temperatures were selected that approximated animal holding conditions. The number of follicles per treatment-group replicate was chosen based on preliminary measurements of total extractable genomic DNA per follicle (Table 1); a yield of 5–10 µg of DNA per pooled sample replicate was considered to be desirable.
Because of DNA extraction difficulties caused by the large volumes of yolk in vitellogenic trout follicles, the oocyte was manually removed from the follicle (oocytectomy) before cryopreservation. This was achieved by puncturing the follicle with fine forceps and then gently peeling away the follicular tissue. The efficacy of this technique was evaluated by performing parallel DNA extractions on the putative "follicle" and the denuded oocytes. Genomic DNA yields of the oocyte portion after oocytectomy were less than the detection limits, indicating that the oocytectomized follicle was relatively intact.
Following incubation, two follicle sample replicates were snap-frozen (after oocytectomy in trout) for 3'-end-labeling analysis; the third replicate of each vitellogenic trout sample group was fixed (intact) in buffered paraformaldehyde for in situ analysis. In situ analyses were not done with isolated goldfish follicles, because preliminary histological investigations showed poor resolution of granulosa and theca cell layers under the light microscope in healthy goldfish follicles.
In a subset of experiments designed to evaluate the putative role of follicle-oocyte communication in regulating follicular apoptosis, serum-free incubations were performed with oocytectomized vitellogenic trout follicles. Conditions for these incubations were exactly as described above for intact follicle-oocytes.
Postovulatory Follicle Incubations
Incubations were performed with rainbow trout postovulatory follicles (TPO) to evaluate the susceptibility of teleost postovulatory follicles to apoptosis in serum-free conditions. Periovulatory (<1 wk prespawning) trout were checked on a daily basis for ovulation (i.e., manually expressible oocytes). Once ovulation had occurred, the fish were immediately killed, and postovulatory follicles were isolated with forceps from the excised ovaries and incubated as described above for vitellogenic follicles. Subsamples of postovulatory follicles were also snap-frozen at the time of removal (0 h) to provide a measure of the in vivo levels of apoptosis.
Atretic Follicles
In a separate experiment designed to evaluate the role of apoptosis during teleost follicular atresia in vivo, morphologically atretic (
-stage) follicles [10] were isolated from the ovaries of five goldfish and immediately snap-frozen for 3'-end-labeling analysis. In addition, fragments (
0.1 g) of ovaries from three goldfish were removed and fixed in buffered paraformaldehyde for in situ analysis.
DNA Extraction and Purification
Total nucleic acids were purified according to methods adapted from those described by Gross-Bellard et al. [16] and by Tilly and Hsueh [17] and modified further as described previously [12]. In addition to the latter modifications, tissue homogenates were incubated with Proteinase K for 60 min at 56°C after initial protein precipitation, followed by phenol:chloroform:isoamylalcohol purification as described previously [12]. Total nucleic acids were ethanol precipitated overnight (-20°C) and collected by centrifugation (14 000 x g for 30 min). The pellet was resuspended in 50 µl of TE buffer (10 mM Tris-HCl and 1 mM EDTA [pH 8.0]), incubated with DNase-free RNase [12], diluted to a 100-µl total volume with TE buffer, phenol-chloroform purified as described previously [12], and then resuspended in 40 µl of double-distilled, deionized, filter-sterilized (0.2-µm mesh) water. Total genomic DNA was quantified by absorbance at 260 nm and stored at -20°C until use.
3'-End-Labeling (Ladder) Analysis
Aliquots (1 µg) of each purified DNA sample were 3'-end-labeled with -32P-ddATP using TdT according to methods adapted from those described by Tilly and Hsueh [17]. Three important modifications were made to these methods. First, during each assay, duplicate (1 µg) samples of 123-bp DNA ladder standard were 3'-end-labeled with -32P-ddATP; these samples served as molecular size standards for estimating the size of oligonucleosomal fragments in sample DNA and as reference measures for total radioactive counts (cpm) per aliquot of labeled DNA during quantification of DNA fragmentation (see Quantification Methods and Data Analysis). Second, after the addition of TdT, samples were incubated for 2 h at 37°C; the extended incubation period was designed to ensure maximal labeling of total genomic DNA. Third, additional duplicate (1 µg) samples of 123-bp DNA ladder standard were incubated as described above, but without TdT. This permitted quantification of nonspecific labeling in each sample and measurement of the efficiency of the removal of excess -32P-ddATP following termination of the reaction (see Quantification Methods and Data Analysis).
Termination of the labeling reactions and removal of excess -32P-ddATP were performed according to the methods described by Tilly and Hsueh [17]. Following overnight resuspension at 4°C in TE buffer, half of each sample aliquot was loaded onto 2% agarose gels and separated electrophoretically for 3.5 h at 65 V (6.5 V/cm). The dried gel was wrapped in clear cellophane and exposed to x-ray film (Kodak X-OMAT; Amersham) for 2–24 h at -80°C to visualize oligonucleosomal fragmentation [17]. The second half of each labeled aliquot was used for quantification of fragmented DNA (see Quantification Methods and Data Analysis).
In Situ TUNEL Analyses
In situ TUNEL analyses were performed with healthy vitellogenic trout follicles, postovulatory trout ovarian fragments, and whole-goldfish ovarian fragments using a commercial assay kit (In Situ Cell Death Detection Kit-AP). Briefly, paraformaldehyde-fixed ovarian samples were dehydrated in a graded ethanol:H2O series, embedded in paraffin, and sectioned (4 µm) onto charged microscope slides (Superfrost* Plus; Fisher Scientific, Nepean, ON, Canada). Slides were deparaffinized for 60 min at 60°C, cleared in a xylene substitute (Hemo-De; Fisher Scientific), rehydrated in a graded ethanol series, and washed in PBS. Slides were permeabilized with Proteinase K (50 µg/ml) for 20 min at 37°C, washed twice with PBS, and then incubated with TUNEL reagents according to manufacturer's protocols. Two slides each were used for negative and positive controls before incubation in each experiment. Negative-control slides were incubated as described above, but without the addition of TdT. Positive-control slides were preincubated in the presence of DNase I (100 ng/ml) for 15 min at 37°C, then washed twice in PBS before incubation with TUNEL reagents.
After the TUNEL reactions were complete, the slides were washed with PBS and incubated with PI and RNase (50 µg/ml each) according to the method described by Brison and Schultz [18]. Slides were then washed in PBS, sealed under coverslips with nail varnish, and examined with a Nikon Optiphot-2 microscope (Nikon, Mississauga, ON, Canada) equipped with a Bio-Rad MRC-600 krypton-argon mixed-gas laser scanning head (Bio-Rad Laboratories, Mississauga, ON, Canada). A series of five to eight randomly selected, nonoverlapping fields per slide were each examined at two emission wavelengths (568 and 488 nm) for respective visualization of PI (all DNA) and fluorescein isothiocyanate (fragmented DNA) labeling. Parallel images from each emission wavelength were digitally collected (CoMOS version 7.1; Bio-Rad Laboratories) and merged using commercial software (Confocal Assistant 4.02; Bio-Rad Laboratories). In the rainbow trout vitellogenic follicle samples, these images were used to quantify the proportion of nuclei with fragmented DNA as a percentage of the total nuclei per field.
Quantification Methods and Data Analysis
Quantification of DNA fragmentation in each treatment group of the 3'-end-labeling experiments was achieved by calculating a fragmentation index (FI), a unitless value representing the ratio of total counts (cpm) per microgram of sample DNA to total counts per microgram of 123-bp ladder standard DNA. Briefly, from the remaining 22.5 µl of -32P-ddATP-labeled DNA in each sample tube, an aliquot (10 µl) was diluted in scintillation cocktail and counted on a liquid scintillation counter. Background counts due to nonspecific binding and residual excess (i.e., unbound) -32P-ddATP were corrected for by subtracting from each sample group the mean counts (cpm/µg) of the duplicate 123-bp DNA ladder samples incubated without TdT. Background counts of these samples were typically less than 2% (1.5% ± 0.01%, n = 8) of the total counts measured in the 123-bp DNA ladder standards incubated in the presence of TdT.
Data are presented as means ± SEM, with each sample (n; Table 1) representing an independent experiment performed using tissues from an individual fish. Mean FI values from all groups were compared by ANOVA; a Tukey's test was employed for post-hoc analyses when necessary. Total percentages of TUNEL-positive nuclei were arcsine transformed and compared using a Student's t-test. Statistically significant differences were accepted when P < 0.05.
RESULTS
3'-End-Labeling
Fragmented DNA, in the ladder pattern characteristic of apoptosis, was evident from autoradiographic analysis of 0-h (snap-frozen) trout follicles at all stages of development. A representative autoradiograph of trout follicular DNA (TV3) is presented in Figure 1.
|
The mean FI for the TPr follicles at 0 h was 0.41 ± 0.13 (n = 4) (Fig. 2A). The FI of the TPr follicles did not change significantly after 24 h of serum-free incubation (0.45 ± 0.13, n = 4).
|
No significant differences in FI were found between the three groups of intact vitellogenic trout follicles (TV5, TV3, and TV1) when compared at either 0 or 24 h (ANOVA, P > 0.05); data from these treatment groups were, therefore, pooled and are henceforth referred to as trout vitellogenic follicles (i.e., TV). The mean FI of the pooled TV follicles at 0 h (0.72 ± 0.14, n = 12) was similar to the mean FI of the TPr follicles; after 24-h incubation in serum-free medium, the mean FI (3.21 ± 0.33, n = 12) increased 4.5-fold over the TV 0-h values (Fig. 2A); the mean FI for incubated TV follicles was significantly greater than the 0-h values for TPr and TV follicles (Tukey's, P < 0.05). When TV follicles were oocytectomized before incubation (TVØ), the mean FI after incubation (5.14 ± 0.34, n = 6) was 6.6-fold higher than the TV 0-h values (Fig. 2A); the mean FI value for TVØ follicles after incubation was significantly greater than the TPr and intact TV follicle values at 0 and 24 h (Tukey's, P < 0.05).
The mean FI for the TPO follicles at 0 h (0.51 ± 0.07, n = 4) was similar to the FI values for TV and TPr follicles; after 24-h incubation, mean quantities of fragmented DNA in TPO follicles (5.95 ± 0.73, n = 4) increased 11.8-fold over the 0-h TPO values (Fig. 2A). The FI of incubated TPO follicles was significantly greater than the 0-h values of the TPr, TV, and TPO follicle groups (P < 0.05), but it was not significantly different from the FI of the incubated oocytectomized follicles.
In goldfish, mean FI values for unincubated vitellogenic (GV; 0.09 ± 0.03, n = 4) and prematurational (GPM; 0.08 ± 0.02, n = 6) follicles at 0 h (Fig. 2B) were significantly lower (P < 0.05) than mean FI values for all trout follicle groups, although the autoradiographic image (not shown) of the goldfish samples clearly indicated apoptotic DNA fragmentation (ladders). The mean FI values of incubated GV (0.07 ± 0.01, n = 4) and GPM (0.04 ± 0.01, n = 6) follicles after 24-h incubation in serum-free conditions did not differ significantly from the respective FI values at 0 h (Fig. 2B), but they were significantly lower (P < 0.05) than the mean FI values for all rainbow trout follicle groups. In preliminary experiments, incubations with goldfish vitellogenic and prematurational follicles were extended for as long as 72 h with no measurable changes in quantities of fragmented DNA (data not shown).
The mean FI for unincubated -stage atretic goldfish follicles (0.12 ± 0.02, n = 5) did not differ significantly from the mean FI values of healthy GV or GPM follicles at 0 h (Fig. 2B). However, it was significantly lower (P < 0.05) than the mean FI values for all trout follicle groups.
In Situ TUNEL Analyses
Specific follicular cell types in vitellogenic follicles of rainbow trout demonstrated differential susceptibilities to apoptosis in vivo (Fig. 3). Approximately 2% (1.8% ± 0.8%, n = 6) of the granulosa cell population was TUNEL-positive at 0 h, whereas approximately 13% (13.0% ± 0.6%, n = 6) of the theca cell population was TUNEL-positive at 0 h. Interestingly, no changes were observed in the percentage of TUNEL-positive cells of either type after 24-h incubation in serum-free conditions (granulosa cells, 1.5% ± 0.7%, n = 6; theca cells, 13.0% ± 2.2, n = 6).
|
In situ analyses of TPO follicles revealed that cells with fragmented DNA were most prevalent in the luteal granulosa cell population (Fig. 4). However, due to the proliferation of follicular cells (theca and granulosa) that is evident after ovulation, we are not currently confident of our ability to accurately quantify the relative proportions of each cell type undergoing apoptosis in the postovulatory follicle.
|
Hematoxylin-eosin-stained sections (4 µm) of goldfish ovarian fragments confirmed the existence of atretic follicles (Fig. 5A) characterized according to defined histological features [10]. In situ (i.e., TUNEL) analysis of these sections revealed that apoptotic cells in atretic follicles were present in distributions similar to those observed in healthy follicles; cells with fragmented nuclei were low in abundance and localized to theca/epithelial cells and extrafollicular regions (stroma) of the ovary (Fig. 5B) despite the presence of clearly identifiable morphological traits associated with atresia in teleosts [10] (Fig. 5A).
|
DISCUSSION
To our knowledge, this is the first study to provide data on the susceptibilities to apoptosis of ovarian follicles in two species of teleost fish both in vivo and during in vitro incubation in serum-free conditions. Our findings are novel, because they provide evidence that whereas apoptosis is involved in normal follicle growth and postovulatory regression in teleost ovaries, it does not appear to be a prominent early event in the onset of follicular atresia in vitellogenic or prematurational teleost ovarian follicles. Supporting evidence for this interpretation is twofold. First, atretic follicles of goldfish, characterized according to established morphological criteria [10], did not display elevated levels of fragmented DNA indicative of apoptotic cell death as compared to healthy follicles. This finding was consistent in both 3'-end-labeling (ladder) and in situ TUNEL analyses. Second, short-term (24-h) incubation in serum-free conditions, an established method for inducing atresia-like changes in mammalian follicles [19], yielded no increase in the onset of spontaneous apoptosis in goldfish follicles and only modest increases in the onset of apoptosis in intact vitellogenic follicles of trout.
These findings are not entirely surprising in light of previous studies on atresia in nonmammalian vertebrates [10]. Histologically, follicular atresia of yolk-bearing oocytes in teleosts and other oviparous vertebrates is characterized by zona radiata breakdown and follicular (i.e., granulosa) hypertrophy (Fig. 5A), followed by invasion and phagocytosis of yolk materials by cells of granulosa origin; only after yolk resorption is complete are the follicular cells observed to degenerate [10, 20]. This chronology of events is believed to facilitate the efficient resorption and recycling of oocyte contents during atresia.
The increases in DNA fragmentation observed with intact trout vitellogenic follicles after incubation in serum-free conditions were similar to values reported in rainbow trout by Janz and Van Der Kraak [12]. Interestingly, however, no changes in the number of TUNEL-positive cells, as a proportion of total cell number, were detected in situ after incubation. These observations led us to draw two conclusions. First, under the given conditions, the DNA fragmentation that we quantified in both nonincubated and incubated teleost ovarian follicles is not necessarily indicative of atresia but, more likely, represents cell death associated with normal, growth-associated cell turnover within the ovarian follicle. Second, a proportion of the increased DNA fragmentation observed after 24-h incubation is likely the result of continued fragmentation of DNA from cells already committed to apoptosis at the time of initial sampling, and it does not necessarily represent de novo apoptotic cell death induced by serum deprivation. The latter conclusion is supported by careful examination of the autoradiographic images of follicular DNA (Fig. 1). Whereas the intensity of signal in low molecular weight (180–540 bp) regions of the autoradiographic DNA is increased after incubation, that in higher molecular weight regions (540–4000 bp) is actually decreased. These findings identify an important caveat to consider when using the ladder assay technique: despite the high sensitivity of the technique, modest increases in DNA fragmentation do not necessarily correspond directly to increases in the number of dying cells.
Our finding that the cohort of dormant, primary follicles in the rainbow trout ovary were less susceptible to apoptosis than healthy, intact vitellogenic follicles is in agreement with our conclusion that growth-associated cell turnover may account for much of the apoptosis measured in the intact, vitellogenic trout follicles. Primary follicles belong to a dormant (i.e., nongrowing) follicle cohort [14]; in contrast to actively growing vitellogenic follicles, less tissue remodeling and, consequently, less apoptosis would be expected.
The relatively high resistance of intact teleost ovarian follicles to apoptosis (as compared to mammals) is similar to observations in the hen ovary, in which hierarchical follicles display an inherent resistance to apoptosis following their recruitment to the pool of rapidly growing follicles destined for ovulation [21]. In the avian ovary, it has been proposed that selection to the ovulatory pool results in altered regulation of genetic elements that suppress or induce apoptosis (e.g., bcl-2 family members) in the follicular cells, resulting in increased resistance to apoptosis. Evidence for this is provided by Johnson et al. [21], who demonstrated that expression of bcl-xlong, a member of the apoptosis-suppressing bcl-2 family of genes, is higher in avian follicles that have been recruited to the ovulatory pool as compared to prehierarchical follicles. Because the intracellular pathways that regulate the onset of apoptosis appear to be highly conserved across diverse taxa, similar molecular mechanisms may be operating in the teleost ovarian follicle.
Direct comparisons between the avian and teleost ovary are difficult, however, because despite structural similarities, marked differences exist in the dynamics of follicular recruitment. In the domestic hen, it has been estimated that less than 1 in 20 prehierarchical follicles are ultimately recruited to the ovulatory pool; the remainder succumb to atresia, mediated by apoptosis [21]. By contrast, teleosts recruit thousands of oocytes to the ovulatory pool, the majority of which are successfully ovulated under normal conditions. Widespread atresia and ovarian recruitment failure may occur under exceptional circumstances [20], but whether massive follicle attrition is a precondition for any stage of teleost ovarian development under normal conditions is unclear at this time.
Resistance to apoptosis may have evolved in teleost follicles in concert with their evolutionary selection for oviparity and high fecundity. For example, prespawning salmonid ovaries may comprise as much as 25% of the total body weight, containing as many as 2000–3000 oocytes per kilogram [22]. Furthermore, each oocyte typically contains large quantities of yolk, a collection of high-energy lipoproteins that fuels embryonic development. The deposition of these yolk reserves represents a substantial portion of the total energy budget of spawning fish.
This contrasts markedly to the mammalian design, in which maternal oocyte investment is typically minimized in favor of substantial embryonic and postnatal investments. Due to the differences in oocyte yolk content between oviparous and viviparous species, the energetic costs associated with follicular atresia are expected to differ, and this may partially explain why massive attrition of follicle-oocytes is not the "norm" in teleost ovaries (as is observed in mammalian ovaries). This is supported by observations in the avian ovary that the onset of yellow-yolk incorporation corresponds with the increased resistance to apoptosis associated with ovulatory selection [21].
Our in vitro experiments with postovulatory follicles and vitellogenic follicles incubated in serum-free conditions after oocytectomy suggest that whereas apoptosis likely represents the ultimate fate for teleost ovarian follicles after ovulation, cell death in vitellogenic follicles may be actively suppressed by the presence of the oocyte. The notion that follicle-oocyte communication may influence the susceptibility of follicular cells to apoptosis has previously been suggested [23, 24]; to our knowledge, the present study presents the first evidence suggesting that a similar mechanism may be operating in the teleost ovary.
At this time, we can only speculate on explanations for the differences in relative susceptibility to apoptosis between the follicles of trout and goldfish. The greater susceptibility of trout follicles was initially surprising, because atretic follicles are typically more abundant in goldfish versus trout ovaries under laboratory holding conditions. In light of the conclusions reached from the data in this study, we suggest that differences in follicular growth rates may explain the between-species differences in susceptibility to apoptosis in vivo. However, we are unaware of any studies that report quantitative comparisons of follicular growth rates between these species. Another possible explanation relates to the structural complexity of the follicle layers in the two species. Histological examinations of trout and goldfish follicles revealed that the thecal and epithelial layers of trout follicles are structurally more complex than those of healthy goldfish follicles. Because the majority of the apoptotic (i.e., TUNEL) signal in trout vitellogenic follicles is localized to theca/epithelial cells, the greater quantities of fragmented DNA in trout follicles may be due to the greater complexity inherent in their tissues. Greater tissue complexity involves more tissue remodeling during follicular growth and development, and apoptosis is well accepted as being fundamental to the process of tissue remodeling [25].
In conclusion, the results of this study suggest that apoptosis is involved in normal development and postovulatory regression in teleost ovaries, but they also challenge the common assertion that apoptosis is the proximate molecular mechanism responsible for follicular atresia in vertebrates. In oviparous species, in which substantial quantities of energy are invested into yolk reserves within the ova, mechanisms have evolved to recycle this energy in the event of atresia. Granulosa cells are intimately involved in oocyte resorption; thus, their demise is likely delayed until oocyte resorption is complete. Future studies will seek to further investigate the role of apoptosis in later stages of atresia and postovulatory regression and to identify biochemical events that precede apoptosis during teleost follicular atresia.
ACKNOWLEDGMENTS
Dr. Melissa Farquhar provided invaluable assistance with confocal microscopy. The comments of three anonymous referees contributed substantially to improvement of the manuscript.
FOOTNOTES
1 Supported by Natural Sciences and Engineering Research Council (NSERC) of Canada grant 0122683 to G.V.D.K. A.W.W. is the recipient of an NSERC post graduate scholarship (PGS-B). 
2 Correspondence. FAX: 519 767 1656; gvanderk{at}uoguelph.ca
Accepted: August 24, 2000.
Received: March 13, 2000.
REFERENCES
plus TGFß on bcl-2 and interleukin-1ß-converting enzyme. J Endocrinol 1998; 157:489–494.[Abstract]. Biol Reprod 1997; 56:1088–1096.[Abstract]This article has been cited by other articles:
|
|
 |
|
  M. J Agulleiro, M. Andre, S. Morais, J. Cerda, and P. J Babin High Transcript Level of Fatty Acid-Binding Protein 11 but Not of Very Low-Density Lipoprotein Receptor Is Correlated to Ovarian Follicle Atresia in a Teleost Fish (Solea senegalensis) Biol Reprod, September 1, 2007; 77(3): 504 - 516. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. V. Andreu-Vieyra, A. G. Buret, and H. R. Habibi Gonadotropin-Releasing Hormone Induction of Apoptosis in the Testes of Goldfish (Carassius auratus) Endocrinology, March 1, 2005; 146(3): 1588 - 1596. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
