Previtellogenic ovarian fragments from eel, Anguilla australis, were cultured in vitro in a chemically defined medium containing steroids and/or peptide hormones for 18 days in order to investigate their involvement in control of early oocyte growth. 11-Ketotestosterone (11-KT), but not estradiol-17β, induced a significant 10–20% increase in diameters of previtellogenic oocytes and oocyte nuclei in a dose-dependent manner. Effects were greatest for 100 nM 11-KT, a dose that is within the physiological range seen in very early vitellogenic eels in the wild. The effect was not accompanied by obvious ultrastructural changes in the oocytes other than an apparent increase in nuclear size. Similarly, treatment with recombinant human IGF-I resulted in increased oocyte diameters, whereas no such effect was seen after treatment with heterologous insulin, GH, leptin, or human chorionic gonadotropin. Interestingly, lipid supplementation also resulted in an increase in oocyte diameter, and greater radioactivity in ovarian explants following incubation with 14C-triglycerides and 11-KT, but not FSH, suggesting that the androgen may play a role in lipid accumulation into the oocyte. Our results implicate hormones from both the reproductive and the metabolic axes in control of previtellogenic oocyte growth in a teleost fish.
The development of a fertilizable egg in oviparous vertebrates requires a substantial accumulation of nutritive resources for the development of the future embryo. Most of these resources are yolk proteins, which are laid down during vitellogenesis, the major growth phase known to be under gonadotropin and estrogen control (reviewed in Nagahama et al. 1995). Prior to the entry into vitellogenesis, oocytes from a typical teleost fish will have attained a 10- to 100-fold increase in diameter when compared with the oogonial stage. The oogonial stage is characterized by active proliferation of germ cells followed by their entry into the first meiotic division. In Japanese huchen (Hucho perryi), the steroid hormones estradiol-17β (E2) and 17,20β-dihydroxy-4-pregnen-3-one have been implicated in controlling oogonial proliferation and entry into meiosis respectively (Higashino et al. 2003). In contrast, very little information is available on the factors that regulate the previtellogenic or primary growth phase (Tyler et al. 2000, Patiño & Sullivan 2002, Campbell et al. 2006), despite their potential importance in controlling oocyte recruitment, and hence, in governing egg batch size (fecundity) at spawning.
The role of pituitary hormones in previtellogenic oocyte growth is not clear; a recent study identified a peak of follicle-stimulating hormone (FSH) around the time of ovulation in rainbow trout, Oncorhynchus mykiss, and it was suggested that this peak may be related to the stimulation of development of a new batch of oocytes for spawning in the following year (Santos et al. 2001). Studies on coho salmon, Oncorhynchus kisutch, also deduced an important role for FSH in very early stages of oogenesis, well before estrogen-mediated vitellogenesis is initiated (Campbell et al. 2006). In contrast, many early studies on hypophysectomized teleost fish have failed to identify a clear role for pituitary hormones on previtellogenic oocytes in a number of fish species, such as catfish, Heteropneustes fossilis (Sundararaj & Goswami 1968). A study by Khoo (1979) likewise indicated that previtellogenic growth may be pituitary independent in hypophysectomized goldfish, Carassius auratus. In very early vitellogenic silver eel, Anguilla anguilla, diameters of oocytes in hypophysectomized fish 2–3 weeks after surgery were slightly smaller than those of controls, but no other histological differences were apparent (Fontaine et al. 1976). It is possible that the effects of hypophysectomy on early stages of oogenesis were missed in these studies due to the seasonal absence of FSH or the focus on vitellogenic oocytes. Regardless, Khoo identified the apparent stimulatory effects of steroids, especially estrogens, on formation of cortical alveoli (Khoo 1979). More recently, prolonged treatment with 11-ketotestosterone (11-KT) in vivo was found to increase the size of previtellogenic oocytes in shortfinned eel, Anguilla australis (Rohr et al. 2001). However, the authors did not determine whether these androgen effects were direct or indirect (through feedback on the brain and/or pituitary). Organ culture may prove useful to address this issue, but few researchers have taken this approach for the study of oocyte growth. Among those who have a positive effect of testosterone on incubated vitellogenic ovarian tissue from goldfish was seen: the proportion of previtellogenic oocytes was higher after testosterone treatment than that in controls, while survival of vitellogenic oocytes was prolonged (Remacle et al. 1976). In mammals, spontaneous progression of primordial follicles to primary follicles maintained in vitro has been reported, and the presence of an inhibitory substance affecting this progression has been postulated (Fortune et al. 2000). Using in vivo models, androgens have recently been implicated in stimulating growth of primordial ovarian follicles (Vendola et al. 1999a). That these androgen effects are direct was supported by the presence of androgen receptors (ARs) in the ovary and clinical observations. Androgen effects appear to be mediated, at least in part, through insulin-like growth-factor-I (IGF-I), given the increase in both IGF-I and IGF-I receptor mRNA levels that occurred after androgen treatment (Vendola et al. 1999b). Other growth factors, such as growth and differentiation factor-9 (Nilsson & Skinner 2002), kit-ligand/stem-cell factor (Parrott & Skinner 1999) and fibroblast growth factor (Nilsson et al. 2001), and hormones from the growth axis, such as insulin (Kezele et al. 2002) may also play a role throughout the development of mammalian ovarian follicles. Crosstalk between the metabolic/nutritional and reproductive systems is likely to operate also in fishes, as there is ample evidence that nutritional status affects oocyte recruitment and fecundity for a range of species (Tyler & Sumpter 1996). A role for IGF-I in such crosstalk is probable (Campbell et al. 2006).
In vitro approaches have previously proven extremely valuable model systems for assessing hormonal control of spermatogenesis in eels, Anguilla japonica (Miura et al. 1991). Because of the paucity of data on factors controlling previtellogenic growth in teleost fishes and their potential involvement in controlling fecundity, we used a long-term ovarian explant culture system to determine the direct effects of steroids (11-KT, E2) and peptide hormones (gonadotropin, growth hormone, leptin, IGF-I, and insulin) on growth of shortfinned eel previtellogenic ovarian follicles. Trials were repeated several times to ensure that responses were robust and assess whether they were affected by histological protocol. The robust and significant effects of 11-KT prompted follow-up studies aimed at assessing the effects of 11-KT on lipid uptake and AR abundance in ovarian explants in vitro.
Materials and Methods
All eels used for experimentation were previtellogenic female shortfins (400–800 g) in the ‘yellow stage’ mostly obtained from commercial eel-processing companies. Animals were caught by fyke nets and held captive by commercial fishermen for a variable length of time (up to about 3 weeks) until their delivery to these processing companies. Female eels used in Trial IVB, reared to experimental size (500–1000 g) from the glass eels stage and kindly donated by Dr Peter Lee (NIWA, Auckland, New Zealand), were used after 1 week of arrival in the University of Otago facilities. Experiments were conducted in accordance with the guidelines of the University of Otago Animal Ethics Committee (protocols AEC 65-00 and ET 23-02).
Steroids (11-KT, E2) were obtained from Steraloids (Newport, RI, USA) and Leibovitz L-15 medium from Sigma Chemical Co. Human chorionic gonadotropin (hCG; Lot CR127) and ovine growth hormone (oGH; NIH-oGH-15, Lot AFP9220A) were a gift from the National Institute of Health (NIH/NIDDK, Bethesda, MD, USA). Porcine and bovine insulin and recombinant human leptin (rhLEP; Lot 062K1603) were purchased from Sigma Chemical Co., while recombinant human IGF-I (rhIGF-I) was obtained from GroPep (Adelaide, Australia). Medium additives were purchased from Life Technologies (Gibco BRL: proline, aspartate), BDH (Poole, UK: glutamate), or Sigma (lipids, see Trial IV). Labeled triglyceride, [carboxyl-14C]triolein (102 mCi/mmol), was obtained from Perkin–Elmer (Wellesley, MA, USA).
Experiment I: in vitro effects of 11-KT or E2 on eel ovarian explants
Three eels were purchased in August 2001 from Gould Aquafarms (Leeston, New Zealand) and euthanized in 0.3 g/l benzocaine. Eels were bled and briefly submersed in 70% ethanol prior to the removal of ovaries under sterile conditions (Miura et al. 1991). Ovaries were placed in supplemented Leibovitz L-15 medium (suppL15) containing 10 mM HEPES, 1.7 mM l-proline, 0.1 mM l-aspartic acid, and 0.1 mM l-glutamic acid and, with scissors and forceps, cut into small fragments of 1–2 mm in each dimension (Miura et al. 1991).
Using 24-well Costar culture plates, fragments were incubated in 1 ml suppL15 containing 0, 1, 10, 100, or 1000 ng/ml 11-KT (~3, 30, 300, or 3000 nM 11-KT; Steraloids). Porcine insulin (pINS, 1 mg/l) was added to maintain tissue health as recommended for the eel testis explant culture system of Miura et al.(1991). Streptomycin sulfate (0.1 g/l; Sigma Chemical Co.) and penicillin (100 000 IU/l; Sigma Chemical Co.) were added to inhibit bacterial growth. Attempts to float the tissue on elder pith, as described for eel testicular fragments (Miura et al. 1991) were unsuccessful and tissues were therefore submersed instead. Two replicate incubations were used for each treatment. Incubations were maintained for 18 days at 15 °C, the medium being changed every third day. At the end of experimentation, explants were processed using standard histological methods, embedded in paraffin, sectioned at 6 μm, and stained with hematoxylin and eosin. Images were analyzed for oocyte diameters as described in ‘Image analysis’ towards the end of this section.
Eels were obtained in November 2001 from Gould Aquafarms. For this trial, five eels that had served as placebo controls in the preceding week were used. These controls had been injected intraperitoneally (1 ml/kg) with a 50:50 mix of vegetable oil and vegetable shortening only and held in recirculating tanks at 16 ± 0.5 °C for 1 week. They were killed and processed for tissue culture as outlined under Trial IA.
Ovarian explants were exposed to 0, 1, 10, 100, or 1000 nM of 11-KT or E2 in suppL15 medium (340 mOsm) with 1 mg/l pINS and antibiotics added and maintained as described for Trial IA. Two replicate incubations were used for each treatment. Explants were preserved in Baker’s calcium formol and processed for routine methacrylate sectioning at 2 μm (Lokman et al. 1998).
In December 2002, three eels were acquired from a commercial fish exporting company (Ngai Tahu Fisheries, Dunedin, New Zealand). Fish were transferred to the laboratory and maintained in tanks supplied with recirculating water of 15 ± 0.5 °C for 1 week.
Eels were killed and dissected as before (Trials IA & IB), and ovarian tissue incubated in single replicates at 15 °C for 18 days in suppL15 medium containing 1 mg/l pINS, 100 IU/l penicillin, and 1 mg/l streptomycin. 11-KT was tested at doses of 0, 1, 10, 100, or 1000 nM. The zero dose was done in duplicate. Tissue handling was minimized by conducting the experiments in 1.5 ml of suppL15 in 12-well culture plates and changing the medium every sixth, rather than every third day.
At the end of the trial, ovarian fragments were fixed in phosphate-buffered (0.1 M, pH 7.4) 2.5% glutaraldehyde and postfixed in 1% phosphate-buffered OsO4 to prevent shrinkage of oocytes during processing. Tissues were then embedded in epoxy resin, sectioned at 2 μm, and stained with toluidine blue. In addition, ultrathin 80 nm sections were prepared from explants incubated with 0 or 100 nM 11-KT from each of the three eels, and viewed with a Philips CM100 transmission electron microscope to examine ultra-structural changes.
Experiment II: in vitro effects of growth factors and pituitary hormones on eel ovarian fragments
This trial was conducted using ovarian tissue from the same eels as described under Trial IB above. Duplicate explants were incubated in suppL15 containing 0, 1, 30, or 1000 ng/ml oGH, hCG, pINS, or rhIGF-I. The cultures were maintained in the presence of 100 nM 11-KT (see effects of 11-KT in vitro in Trial IA, Results section) and antibiotics for 18 days at 16 °C, and the medium was changed every third day. Tissues were harvested, preserved, and processed as described for Trial IB.
The effects of IGF-I were studied in more detail using a dose-response design with explants prepared from the eels described under Trial IC and using identical culture additives. Recombinant hIGF-I (1, 10, 100, and 1000 ng/ml) dose-response treatments were done without replication, except for the controls (0 ng/ml rhIGF-I), which were done in duplicate. All cultures were supplemented with 100 nM 11-KT. After incubation, explants were fixed as described under Trial IC and prepared for light microscopical observations.
A third trial with rh-IGF-I dose responses (0–1000 ng/ml) was conducted in September 2005. To assess whether IGF-I effects required the presence of androgen, cultures were set up for ovarian tissue from seven previtellogenic eels in suppL15, with or without addition of 100 nM 11-KT. Replication was only between, not within, individual fish. Media, medium changes, and duration of the experiment were as described in Trial IC. Tissues were processed for routine embedding in methacrylate resin and oocyte diameters were determined by microscopy and image analysis (cf. Trial IB).
Experiment III: in vitro effects of leptin on eel ovarian fragments
Effects of rhLEP were investigated in February 2004 using the in vitro system validated above. A total of five previtellogenic eels were used, and ovarian tissue was incubated without replication according to Trial IA with or without leptin (0–1000 ng/ml) in the presence or absence of 100 nM 11-KT and 1 mg/l pINS.
A replicate experiment with leptin was conducted a year later, using the same fish, incubation conditions, and evaluation protocols as described under Trial IIC. 11-KT was again used at 100 nM, whereas leptin was added as 0, 1.25, 12.5, 125, or 1250 ng/ml. All single-run cultures were further supplemented with 1 mg/l bovine insulin.
Experiment IV: effects of lipid supplementation on 11-KT-stimulated oocyte growth in vitro
Trial IVA in vitro accumulation of lipids
Phospholipid (PL; 1,2-diacyl-sn-glycero-3-phosphocholine, Sigma Chemical Co.) or a fatty acid supplement (FAS) containing linoleic acid and oleic acid in BSA (Sigma Chemical Co.) were added to cultures of ovarian explants (same fish as used for Trial IIC) containing suppL15 and 1 mg/l bovine insulin. PL was added at 100 μg/ml, and FAS at 60 ng/ml linoleic acid +30 ng/ml oleic acid. Cultures were run without replication in the presence or absence of 11-KT at 100 nM for a total of 18 days and processed as described under Trial IIC.
Trial IVB in vitro accumulation of 14C-triolein
Ovarian tissue was recovered from five female eels in June 2006 as described for Trial I. For each female, tissue was placed on a piece of soft plastic and finely chopped with razor blades. Chopped tissue was divided into two approximately equal-sized portions. Portions were again equally divided, and so on, until a total of 16 approximately equal portions had been obtained. These were placed in wells of a 24-well plate and incubated in bovine insulin-supplemented L15 with or without 100 ng/ml native coho salmon FSH (kindly donated by Dr Penny Swanson, Northwest Fisheries Science Center, Seattle, WA, USA), and in the presence or absence of 100 nM 11-KT. To all wells, 80 nCi of [carboxyl-14C]triolein was added, and duplicate incubations were accordingly set up for either 1 or 4 days. After incubation, the medium was aspirated off, and the tissue was washed with 1 ml eel Ringer for 30 min at room temperature (ca 18 °C) at 200 rpm. Washing was repeated once thereafter and as much as possible of the tissue was transferred to a plastic centrifuge tube. After any remaining medium was removed, the tissue was frozen at −20 °C and subjected to three freeze–thaw cycles to burst cell membranes. Further cell lysis was achieved using 100 μl proteinase K (10 mg/ml in Tris-EDTA, pH 7.5) for 3 h at 37 °C. Sample volume was increased by addition of 100 μl TE buffer, and lysis completed by adding 100 μl of 3 M NaOH and incubation at 65 °C for 2 h. [Carboxyl-14C]triolein accumulation was then estimated by scintillation counting (Wallac, Turku, Finland).
Experiment V: in vitro effects of 11-KT on ovarian AR gene expression
Ovarian tissue from five female eels was incubated with 0, 1, 10, 100, or 1000 ng/ml 11-KT as described in Trial IC. Tissue was harvested after 18 days, transferred to a plastic centrifuge tube, and flash-frozen in liquid nitrogen. Total RNA was extracted using Trizol Reagent (Invitrogen) and 1 μg was reverse transcribed using random hexamer primers and Superscript III (Invitrogen). cDNAs were stored at −80 °C until analysis for mRNA levels by quantitative real-time PCR (QPCR).
Four micrograms of total RNA, extracted from the testis of a silver-stage male shortfinned eel, was reverse transcribed using oligo-dT15–18 and Superscript III. Shortfinned eel ARα and ARβ cDNA fragments were obtained by PCR, with primer design guided by Genbank Accession Nos AB023960 and AB025361 respectively for the Japanese eel AR cDNAs. Primers used for ARα were forward 5′-TCCATGTTCCTCAACATCCTGGAG-3′ and reverse 5′-GAAGTTCTTCATCCGGATGCAGTG-3′, whereas ARβ cDNA was amplified using forward 5′-TCCACAAATCTTCCTTGTCTGAA-3′ and reverse 5′-GGGTTTCAACTGTCCAATTTTCT-3′. PCR products of 360 and 374 bp in length were obtained after 34 (ARα) or 31 (ARβ) cycles (20 s at 94 °C, 20 s at 58 °C and 30 s at 72 °C) using Roche PCR reagents. PCR products were electrophoresed on agarose gels, excised using the QIAEX Gel Extraction Kit (Qiagen) and ligated into the pGEM T-Easy vector (Promega) according to the manufacturer’s instructions. Following amplification in Escherichia coli XL-1 Blue and recovery of plasmid by Plasmid Midi Kit (Qiagen), cDNAs were sequenced and aligned with the Japanese eel homologs. Accordingly, 99.7 and 98.6% sequence identity was obtained at the nucleotide level for the α- and β-subtype respectively. At the amino acid level, both subtypes showed 100% sequence identity, confirming that the amplified fragments encoded the AR. cDNAs were recovered from the plasmid by restriction enzyme digestion, and standards (100–105 copies/μl) generated for use in QPCR.
Primers for QPCR were designed with primer express as follows: ARα forward, 5′-GTACGTGGATGACCAGATGACAGT-3′; ARα reverse, 5′-CCAGCCCAGAGCGAACAC-3′; ARβ forward, 5′-AACGATTGCACCATTGATAAGCT-3′; ARβ reverse, 5′-CATTCCGGCCTCAAAGCA-3′. Samples were heated for 2 min at 50 °C and 2 min at 95 °C and then cycled 40 times between 95 °C (15 s) and 58 °C (30 s) and 72 °C (30 s), before a final extension step of 1 min at 72 °C using a Stratagene MX-3000P Real Time thermal cycler. PCR products were subsequently subcloned into pGEM T-Easy vector and confirmation sequenced.
QPCR was carried out on duplicate samples of 50 ng cDNA or on 1 μl of standard using Invitrogen’s SYBR Green qPCR Supermix in a reaction volume of 20 μl on a single plate for each target. Primers were used at final concentrations of 400 (ARα reverse) or 800 nM (other primers), determined on the basis of a primer dilution matrix. Amplicons were obtained as described in the previous paragraph. A melting curve was generated at the end of the run to ensure that only a single amplicon was amplified. MX-3000P (Stratagene, La Jolla, CA, USA) software generated standard curves and AR copy number estimates in each sample.
AR copy number was expressed per microgram of total RNA or normalized over copy number of elongation factor-1α (ELF). For the latter approach, we amplified an ELF fragment by PCR using forward primer 5′-ATGGGAAAGGAAAAGATCCACATCA-3′ and reverse primer 5′-TCAAGCTTCTTGCCAGAACGACGGT-3′ at 25 cycles of denaturation (94 °C, 30 s), annealing (52 °C, 30 s), and extension (72 °C, 30 s). Primer design was based on the ELF sequence of zebrafish, Danio rerio (Accession No BC064291). An 1163 bp cDNA, located within the open reading frame, was cloned into the pGEM T-Easy plasmid vector and sequenced, as described for AR above. The sequence had 88.4 and 94.8% sequence identity with the zebrafish cDNA and deduced amino acid sequences respectively.
Plasmids, linearized with Spe I, were agarose gel purified and used as standards in QPCR. QPCR for ELF was conducted using a primer pair (150 nM forward 5′-AATCCTGAAGCCTGGTATGGTGGT-3′ and 100 nM reverse 5′-ACGTTGAAGCCAACATTGTCACCG-3′) nested within the 1163 bp ELF fragment to generate amplicons of 126 bp after 40 cycles of 95 °C (15 s), 62 °C (30 s), and 72 °C (30 s) and melting curve analysis on the Stratagene MX-3000P, as described above. MX-3000P software was again used to generate a standard curve and estimate target gene copy number in each sample.
Microscope images were captured using a Nikon video camera and analyzed through NIH Image software (version 1.62) at low power. Starting with a calibration slide, a scale bar length of 100 μm was selected. During sample image capture, a field of vision was chosen that had maximum occupancy by oocytes; the scale bar was then pasted into the image and the image stored for analysis at a later time. The calibration slide was repeatedly re-scanned during a photographic session to ensure that the magnification did not change.
For analysis, all intact oocytes in an image were first counted. A proportion of oocytes was then used for data collection; thus, only the largest oocytes (20% of total number) were selected to eliminate those oocytes that may have been sectioned off-center. To find these oocytes, the largest 25% of oocytes (by eye) were numbered and oocyte areas were measured by NIH image software. Oocyte diameters were subsequently calculated by taking the square root of oocyte surface area and multiplying this value by 4/π. Data were then sorted by oocyte diameter and the largest 80% (i.e. 20% total oocytes) retained and means calculated to yield a single value for each image, thereby avoiding pseudor-eplication. As much as possible, all images from any one experiment were analyzed by the same operator.
Validation of image analysis
The procedures outlined above were applied independently (1 month apart; no communication on result, etc.) by two different authors (PML, KANG) to the 0 and 30 ng/ml rhIGF-I samples of Trial IV to assess reliability and consistency of oocyte diameter measurements. We extended the validations by comparing the same two treatments after counting not only the largest 20%, but also the largest 10% and largest 40% during image analysis.
Figures were primarily drawn on the basis of the percentage change (mean ± s.e.m.) of response variables to circumvent the large error bars that otherwise result due to fish-to-fish variation, obscuring treatment effects. Accordingly, controls were given arbitrary values of ‘1’. Replicates from the same animal were averaged for graphical presentation, but not for statistical analyses.
Statistically, animal-to-animal variation was controlled for by including animal as a random variable into the model. Thus, effects of hormone treatment on oocyte diameter, nuclear diameter, AR abundance, or lipid accumulation in replicated experiments were tested using a nested univariate model in SPSS 14.0 (SPSS Inc., Chicago, IL, USA) , command UNIANOVA, with FISH and REPLICATE as random variables and using the following design subcommand:/DESIGN HORMONE FISH(HORMONE) REPLICATE(FISH). To compare the effects of 11-KT and E2, a two-way univariate ANOVA was run without the zero dose and without interaction in the model.
Experiments that were run in singles were analyzed using hormone treatment as fixed factor and FISH as random factor in a two- or three-way UNIANOVA/DESIGN HORMONE_1 HORMONE_2 HORMONE_1*-HORMONE_2 FISH. Interaction between hormone treatments was removed from the model if its effect was not significant. Nesting of fish within hormone treatment could not be done due to insufficient degrees of freedom. Normality of data was confirmed by analyzing the spread of residuals, whereas homogeneity of variances was tested on unnested data using Levene’s statistic. Posthoc comparisons of means were done using Tukey. Differences between groups were considered significant for P < 0.05.
Tissues could be maintained in vitro without bacterial or fungal contamination. Histologically, atresia was evident in some oocytes in many of the tissue explants, but the incidence of atresia was not quantified. Some shrinkage of oocytes was noted after processing of explants for histology, especially in paraffin-embedded material, for which no corrections were made. Oocytes from all animals were of comparable size and typically in the range of 50–100 μm. Due to the small size of the explants used for incubation, together with the regular changes of the incubation medium, some tissue could either not be recovered, or in rare circumstances, proved to be connective tissue. Variation in sample size during analyses could therefore not be avoided.
Analyses of images by two independent investigators yielded highly consistent results, as illustrated by regression analysis (Fig. 1). However, a consistent difference between operators in absolute oocyte diameters existed, due to differences in the total number of oocytes counted on a section. These differences were found to reflect inclusion of the smallest oocytes by one, but not the other operator. As a result, the average oocyte diameter of a sample (based on the 20% largest oocytes in a section) decreased with increasing numbers of total oocytes, thus yielding oocytes that on average were 7–8% smaller in diameter. These differences between operators did not affect the statistical outcome, both treatments yielding similar mean diameters that tended to be higher for explants treated with 30 ng/ml rhIGF-I than for controls (Operator I, P = 0.144; Operator II, P = 0.065).
Similarly, basing the average oocyte diameter in a section on counting the largest 10, 20, or 40% in a section did not affect the outcome of the statistical analysis, with rhIGF-I at 30 ng/ml always tending to yield larger oocyte diameters than controls (all P > 0.05; data not shown).
Treatment of ovarian fragments with 11-KT resulted in a significant increase in the diameter of eel oocytes (F4,8 = 6.803; P = 0.011) at all doses, except 10 ng/ml (P = 0.052 for this group; data not shown). No differences in oocyte diameter were found among any of the 11-KT treatments. Similarly, there were no significant differences between individual fish, but replicates did differ significantly for some fish × dose combinations (F3,12 = 7.151; P = 0.005).
Following 18 days of organ culture, oocyte diameters (F1,50 = 21.53; P < 0.001) and nuclear diameters (F1,50 = 20.164; P < 0.001) were significantly greater, by nearly 10%, in cultures containing 11-KT compared with those with E2 added (Fig. 2a and b). Treatment with E2 did not significantly alter oocyte or nuclear diameters. By contrast, 11-KT effects were dose dependent (Fig. 2a and b), and differed significantly from control incubations, except for nuclear diameters of oocytes exposed to 10 nM and for diameters of oocytes exposed to 1000 nM.
Application of protocols for use in transmission electron microscopy (TEM) yielded trends for 11-KT that confirmed those obtained by tissues processed for paraffin (Trial I) or methacrylate resin embedding (Trial II), even with a small sample size of three fish. Thus, treatment of ovarian fragments with 11-KT led to a dose-dependent increase in relative oocyte diameters, which was significant in the physiological range (10 and 100 nM) when compared with control incubations (data not shown). Histologically, ovaries from all females were in the early stages of lipid accumulation and had maximum oocyte diameters ranging between 50 and 70 μm. Effects of 11-KT on oocyte size were readily visible (cf. Fig. 3a and b). However, the increase in diameter brought about by 11-KT were neither reflected in obvious changes in number of lipid droplets nor in mitochondrial number or Golgi apparatus, at least not when comparing explants exposed to 0 and 100 nM 11-KT (Fig. 3c and d). Atresia (Fig. 3e) and oogonia (Fig. 3f) were occasionally found in ultrathin sections.
Growth factors, rather than pituitary hormones, tended to affect oocyte diameters in vitro. Thus, addition of oGH had no effect (F3,12 = 1.374; P = 0.298), whereas only hCG at 1 IU/ml resulted in a significantly smaller oocyte diameter compared with other doses after 18 days of culture (F3,12.3 = 3.562; P = 0.047). In contrast, rhIGF-I induced a highly significant increase in oocyte diameters (F3,12.3 = 6.434; P = 0.007), essentially in a dose-response fashion (Fig. 4a). Similarly, pINS significantly affected oocyte size (F3,12.6 = 7.752; P = 0.003), although a dose response was not apparent. Differences between fish were not significant for any of the treatments.
Responses of ovarian explants to rhIGF-I varied considerably between the three individuals, and a significant treatment effect therefore could not be discerned (P = 0.593). However, there was a tendency for oocyte diameters to increase with increasing dose of IGF-I (data not shown).
Exposure of ovarian explants to IGF-I resulted in a highly significant increase (F4,50 = 6.974; P = 0.011) in oocyte diameters in an approximately dose-dependent fashion (Fig. 4b). Oocyte diameters of ovarian follicles exposed to control media or those containing 1 ng/ml IGF-I averaged 70 μm in diameter, increasing to mean values of up to 92 μm in the highest dose IGF-I groups. Effects of 11-KT were additive and also highly significant (F1,50 = 4.157; P = 0.006), and the steroid increased mean oocyte diameters from 76 to 87 μm across treatment groups.
Experiment III: in vitro effects of leptin on eel ovarian fragments
Leptin did not affect oocyte diameters in either trial (IIIA: F2,29 = 0.527; P = 0.717; IIIA: F4,55 = 0.177; P = 0.949; data not shown). Effects of 11-KT were significant only for trial IIIA (F1,29 = 16.767; P < 0.001; data not shown).
Experiment IV: effects of lipid supplementation on 11-KT-stimulated oocyte growth in vitro
Trial IVA In vitro accumulation of lipids
Addition of lipids to the incubation medium tended to result in increased oocyte diameters (F2,26 = 2.393; P = 0.111). Indeed, for six out of seven females, control incubations yielded the smallest oocyte diameters. Oocyte diameters for the control incubation from the seventh fish were unusual, at nearly twice those seen for some other incubations; exclusion of this single data point changed the probability for an effect of lipid addition on oocyte diameters to P = 0.003 with an average increase of 14 μm to the oocyte diameter in response to adding PL or FAS (Fig. 5). Regardless of whether the outlying data point was included or excluded, a statistically significant additive effect of 11-KT was observed (F1,26 = 9.349; P < 0.001; Fig. 5).
Trial IVB In vitro accumulation of 14C-triolein
Accumulation of 14C-triolein in ovarian fragments was significantly increased (Fig. 6a and b) in the presence of 11-KT (F1,13.1 = 6.587; P = 0.023), but not FSH (F1,13.2 = 1.688; P = 0.216), when analyzing data from both days in the same nested ANOVA. Statistical outcome was regardless of whether the non-significant interaction term (11-KT * FSH) was included in the model. Duration of incubation also significantly increased triolein accumulation in ovarian fragments (F1,52 = 50.242; P < 0.001). When simplifying the statistical model and analyzing data for both days individually, degrees of freedom were insufficient to detect an effect of 11-KT by 1 day (F1,13 = 1.912; P = 0.19) and just insufficient to detect an 11-KT effect by 4 days of incubation (F1,14.3 = 4.464; P = 0.053).
Experiment V: in vitro effects of 11-KT on ovarian AR gene expression
AR transcript abundance was low, with maximum levels of around 7000 and 1700 copies/μg total RNA for ARα and ARβ respectively. QPCR on total RNA without the RT step typically yielded 500–2000 copies of target/μg RNA. There was insufficient RNA to assess contamination of cDNA for all samples; an arbitrary threshold of 2000 copies of target/μg RNA was therefore set as the minimum detectable level, deeming ARβ mRNA levels undetectable in all samples. Samples in which ARα copy number exceeded 2000/μg RNA, genomic contamination copy number was subtracted if a value for that sample could be obtained; in the absence of an estimate of copy number due to genomic contamination, the copy number was set at 1000, midway in the contamination range (500–2000 copies). Estimates for ARα copy number could accordingly be obtained for 11 out of 25 samples. Although insufficient for statistical analysis, data on ARα transcript abundance from only these 11 samples hinted at a reduction in copy number with increasing concentrations of 11-KT in the incubation medium, regardless of whether the data were expressed per microgram total RNA or per 106 copies of ELF (Fig. 7a and b).
Since little is known about the factors controlling previtellogenic growth in teleost fish other than two early reports (Remacle et al. 1976, Khoo 1979), we exposed eel ovarian fragments to a range of hormones invitro. Exposure to E2 did not affect oocyte diameter after 18 days of culture. Remacle et al.(1976) likewise could not detect any effects of estrogens on the goldfish ovary in vitro, irrespective of oocyte stage. In vivo observations, however, seem contradictory; Yu et al.(1979), while feeding estrogen-enriched diets, could not detect any effect on perinucleolar oocytes in ovaries of juvenile coho salmon. In contrast, E2 treatment was correlated with cortical alveolus formation in hypophysectomized goldfish (Khoo 1979). In European silver eels, oocytes increased from an average of around 70 μm in controls (previtellogenic) to around 180 μm (very early vitellogenic) after long-term in vivo E2-treatment (Olivereau & Olivereau 1979). This increase coincided with an increase in accumulated lipids. However, the profound increase in the activity of gonadotropes suggests that the observed gonadal changes are not onlyattributable to E2 but probably also to gonadotropins.
In contrast to E2, in vitro treatment of eel ovarian fragments with 11-KT led to a significant increase in the size of the largest oocytes in all the three trials, irrespective of the method of tissue processing used (paraffin, methacrylate, epoxy resin). The nucleus too, appeared to be larger in oocytes from explants treated with 11-KT, an effect that was confirmed statistically for Trial IB. Ultrastructurally, increased oocyte dimensions were not mirrored by notable changes in cell morphology other than a seemingly larger nucleus (Trial IC); thus, oocytes were recognizably larger when viewed with the electron microscope, but there was no clear evidence for the increased abundance of organelles, such as endoplasmic reticulum or golgi. Similarly, mitochondrial numbers were not clearly altered, although the non-random distribution of mitochondria makes it difficult to ascertain this conclusively. Likewise, there were no histological or ultrastructural indications that the abundance of lipid droplets in the ooplasm had changed. This is not in keeping with the observed lipid accumulation following exposure of eels to 11-KTin vivo (Rohr et al. 2001) or with the uptake of lipids in the form of radiolabeled triglycerides into the ovary that was stimulated by 11-KT (Expt IVb). These observations, together with the finding that the addition of lipid supplements to the incubation medium tended to increase oocyte diameters in the presence of 11-KT (Expt IVa), suggest that lipids were limited in the medium, thus prohibiting ultrastructural change (Trial IC).
Remacle et al.(1976) previously investigated the effect of different hormone preparations on vitellogenic ovarian tissues of goldfish in vitro. Interestingly, the integrity of oocytes up to stage IV (cortical alveolus stage) was better in the presence of testosterone and somatic tissues were less hypertrophied when compared with controls. The authors further suspected the development of nests of spermatogonia within the ovarian explants (Remacle et al. 1976). In our study, the presence of nests of spermatogonia was not apparent.
Androgens have been implicated in oogenesis in mammals, particularly in the growth of small preantral follicles. These follicles resemble previtellogenic follicles of fishes in that growth progresses slowly, occurs both prior to and after puberty (McGee & Hsueh 2000) and appears to be mostly gonadotropin independent (reviewed by Smitz & Cortvrindt 2002). Thus, the number of primordial follicles advancing to the pre- or small antral stage increased in androgen-treated monkeys (Vendola et al. 1998, 1999a). Likewise, women treated with testosterone or suffering from androgen excess (polycystic ovarian syndrome, androgen-producing tumors) showed increased numbers of growing follicles (reviewed in Vendola et al. 1998). These observations contrast with atretogenic actions of androgens on preantral follicles in immature hypophysectomized rats (Hillier 1987). Although some of the stimulatory effects may be indirect, via surrounding stromal tissue (Vendola et al. 1999a), direct androgen effects on ovarian follicles are likely given that AR mRNA abundance in rhesus monkey was the greatest in preantral to small antral follicles (Weil et al. 1998). In the mouse, a role for androgens is also apparent in view of the observations that female AR knock-outs had reduced average litter size (Yeh et al. 2002, Shiina et al. 2006) and an increased incidence of atresia of ovarian follicles (Shiina et al. 2006). The latter observations probably result from impaired folliculogenesis, as evidence has recently been mounting for an important role of androgens in folliculogenesis through modulation of expression or effects of local growth factors (e.g., Hickey et al. 2005, Shiina et al. 2006).
In our previous in vivo work on shortfinned eels (Rohr et al. 2001), 11-KT had a clear effect on the growth of previtellogenic oocytes, but it was not determined whether this was a direct or an indirect effect, mediated via feedback of this steroid on the brain–pituitary axis. The current findings, based around an in vitro bioassay design, comply with the mammalian scenario, and indicate that at least some of the effects seen in vivo are likely to have been direct. The presence of AR mRNA, but not ARβ mRNA, in the eel ovary, albeit at copy numbers some 2–5 fold lower for the α-subtype than those seen in vivo (M Algie and PM Lokman, unpublished observations), or around 20–50 fold lower than those reported in for example, liver or muscle of melengestrol acetate-treated heifers (Pfaffl et al. 2002), support this notion. In other species of fish, such as coho salmon (Fitzpatrick et al. 1994), Atlantic croaker, Micropogonias undulatus (Sperry & Thomas 1999), seabream, Pagrus major (Touhata et al. 1999), goldfish (Wells & Van Der Kraak 2000), and Japanese eel (Ikeuchi et al. 1999), AR transcripts have also been identified in the ovary. However, specific roles of androgens and ARs in the fish ovary have not been described previously, and this is the first study to indicate a possible function, i.e. involvement in control of previtellogenic oocyte growth, at least in eel.
Maximal effects of 11-KT were observed when using doses in the 10–100 nM range (3–30 ng/ml), similar to blood 11-KT levels seen in very early vitellogenic shortfinned eels caught from the wild (20–50 ng/ml; Lokman et al. 1998, 2002). Although not a focal point of other investigations, increased levels of androgens (generally testosterone, rather than 11-KT) during previtellogenesis have been reported in a number of fish species, such as striped bass, Morone saxatilis (Holland et al. 2000) and seahorse, Hippocampus abdominalis (Poortenaar et al. 2004). In females of other species, such as killifish, Fundulus grandis (Greeley et al. 1988), channel catfish, Ictalurus punctatus (Kumar et al. 2000) and sweep, Scorpis lineolatus (Dedual & Pankhurst 1992), elevated androgen levels during previtellogenesis were not obvious. Nonetheless, it is likely that androgens do not only serve as precursors for estrogen synthesis, but also that they have a functional role during previtellogenesis in teleost fish in general.
Completion of the previtellogenic growth phase in fish is followed, at some stage, by initiation of puberty, the activation of the brain–pituitary–gonad axis. In eels, puberty coincides, or is just preceded by, ‘silvering’; silvering encompasses dramatic changes in the physiology and behavior of the eel in preparation for a long-distance oceanic migration and is probably mediated by 11-KT (Rohr et al. 2001, Lokman et al. 2002). Pre-adaptations include increases in eye size, expression of different opsins, jaw remodeling for stream-lining, cardiovascular changes, cessation of feeding, and enhancement of hypoosmoregulatory ability (reviewed in Lokman et al. 2003b). It is believed that silvering only occurs once eels reach a critical condition (fat stores; Larsson et al. 1990). Therefore, it is likely that hormones involved in metabolism and growth modify the activity of the brain–pituitary–gonad axis. However, it is not clear which metabolic hormone is the primary regulator of this axis, and there is no evidence to indicate at which level the axis is most affected. When evaluating the effects of heterologous peptide hormones, namely, gonadotropin, growth hormone, insulin, IGF-I and leptin on the eel ovary in vitro, an effect was noted only for IGF-I. It is possible that heterologous hormones were not recognized by the previtellogenic eel ovary, an issue that is impossible to address without extensive research on receptor–ligand interaction for these peptides. Such data is fortunately available for IGF, a structurally highly conserved protein for which the salmon and human recombinant homologs have comparable activities in salmon embryonic cells in vitro (Upton et al. 1998). Addition of hIGF-I to the incubation medium resulted in a significant increase in eel oocyte diameters, an effect that was additive to that of 11-KT. In the mammalian ovary, IGF-I appears to be important for oogenesis throughout the development of the ovarian follicle. For example, IGF-I can modify ovarian steroidogenesis, and thus appears to play a role in dominant follicle selection (Ginther et al. 2004). Similarly, IGF-I can increase the population of gonadotropin-independent small antral follicles (Gong et al. 1997) and stimulate growth of preantral follicles, possibly through mediating androgen action (Vendola et al. 1999a, 1999b). Wide effects of IGF-I on aspects of oogenesis have been reported in fish also; thus, IGF-I can affect basal or gonadotropin-supported steroidogenesis in vitellogenic or maturing ovarian follicles (Weber & Sullivan 2000, Mendez et al. 2005), and IGFs have been implicated in stimulating germinal vesicle breakdown in a number of fish species (e.g., Weber & Sullivan 2000). Very recently, involvement of IGF-I on early stages of oogenesis in salmon was postulated on the basis of its plasma dynamics during previtellogenesis (Campbell et al. 2006). Our findings on the eel ovary confirm an involvement in regulating previtellogenic ovarian growth, although the mechanisms that cause actual growth remain to be established.
Noticeable animal-to-animal variation in responses to hormones was observed in our experiments, although responses obtained were consistent among trials, and did not depend on the individual investigator, on the tissue processing technique, or on the number of oocytes measured. Therefore, individual variation cannot be attributed to methodology or data analysis. Instead, there probably is a physiological cause for the differences amongst individuals, for instance in relation to differences in IGF-I receptor and AR abundance, issues that require additional study.
In summary, we have shown using a functional in vitro organ culture system that IGF-I and the androgen 11-KT can exert direct effects on the eel ovary, resulting in additive increases in size of previtellogenic oocytes. Moreover, 11-KT, but not FSH, significantly stimulated the accumulation of radiolabeled triglycerides into the ovary. We conclude that growth factors and androgens are involved in the regulation of previtellogenic oocyte growth in a teleost fish.
Received 25 September 2006 First decision 11 December 2006 Revised manuscript received 18 January 2007 Accepted 30 January 2007
We gratefully acknowledge the help from Matthew Downes and Karen Judge, (Department of Zoology) with sample processing, and help from Ken Miller (Department of Zoology) to prepare the figures for publication. Thank you to Jamie McQuillan (Department of Zoology) and Dr Hajime Matsubara (National Institute of Aquaculture, Mie, Japan) for help with real-time PCR and cloning of ELF respectively. We are also grateful to Dr Peter Lee (NIWA) for the gift of captive-reared eels for experimentation and to students Emma Howell, Bryony McNeill, and Sandra Soeder-Hunua for their assistance with image processing. We thank Dr Penny Swanson (Northwest Fisheries Science Center) for the gift of coho salmon FSH, and the National Hormone and Pituitary Program of the NIDDK (NIH, Baltimore, MD, USA) for the gift of oGH. This paper is dedicated to our dear colleague Gerald Stokes, who for many years has been our beacon for histological assistance and advice, but passed away on 19 December, 2003. This study was supported by a grant from the University of Otago Research Committee (PML 2003, 2004). A small part of this study has been previously published in the form of conference proceedings in Fish Physiology & Biochemistry, Volume 28, 283–285 (2003). The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work.
Campbell B Dickey J Beckman B Young G Pierce A Fukada H & Swanson P2006 Previtellogenic oocyte growth in salmon: relationships among body growth plasma insulin-like growth factor-1 estradiol-17β follicle-stimulating hormone and expression of ovarian genes for insulin-like growth factors steroidogenic-acute regulatory protein and receptors for gonadotropins growth hormone and somatolactin. Biology of Reproduction7534–44.
Lokman PM Vermeulen GJ Lambert JGD & Young G1998 Gonad histology and plasma steroid profiles in wild New Zealand freshwater eels (Anguilla dieffenbachii and A. australis) before and at the onset of the natural spawning migration. I. Females. Fish Physiology and Biochemistry19325–338.
Lokman PM Rohr DH Davie PS & Young G2003 The physiology of silvering in anguillid eels - androgens and control of metamorphosis from the yellow to the silver stage. In Advances in Eel Biology pp 331–349. Eds K Aida K Tsukamoto & K Yamauchi. Tokyo: Springer Verlag.
Mendez E Montserrat N & Planas JV2005 Modulation of the steroidogenic activity of luteinizing hormone by insulin and insulin-like growth factor-I through interaction with the cAMP-dependent protein kinase signaling pathway in the trout ovary. Molecular and Cellular Endocrinology22949–56.
Pfaffl MW Daxenberger A Hageleit M & Meyer HHD2002 Effects of synthetic progestagens on the mRNA expression of androgen receptor progesterone receptor oestrogen receptor α and β insulin-like growth factor-I (IGF-I) and IGF-I receptor in heifer tissues. Journal of Veterinary MedicineA4957–64.
Santos EM Rand-Weaver M & Tyler CR2001 Follicle-stimulating hormone and its α and β subunits in rainbow trout (Oncorhynchus mykiss): Purification characterization development of specific radioimmunoassays and their seasonal plasma and pituitary concentrations in females. Biology of Reproduction65288–294.
Tyler CR Santos EM & Prat F2000 Unscrambling the egg – cellular biochemical molecular and endocrine advances in oogenesis. In Proceedings of the 6th International Symposium on the Reproductive Physiology of Fish pp 273–280. Eds B Norberg OS Kjesbu GL Taranger E Andersson & SO Stefansson. Bergen Norway: Institute of Marine Research and University of Bergen.