Null Mutation in Transforming Growth Factor 1 Disrupts Ovarian Function and Causes Oocyte Incompetence and Early Embryo Arrest
http://www.100md.com
《内分泌学杂志》
Research Centre for Reproductive Health, Department of Obstetrics and Gynaecology, University of Adelaide, Adelaide, South Australia 5005, Australia
Abstract
TGF1 is implicated in regulation of ovarian function and the events of early pregnancy. We have investigated the effect of null mutation in the Tgf1 gene on reproductive function in female mice. The reproductive capacity of TGF1 null mutant females was severely impaired, leading to almost complete infertility. Onset of sexual maturity was delayed, after which ovarian function was disrupted, with extended ovarian cycles, irregular ovulation, and a 40% reduction in oocytes ovulated. Serum FSH and estrogen content were normal, but TGF1 null mutant mice failed to display the characteristic proestrus surge in circulating LH. Ovarian hyperstimulation with exogenous gonadotropins elicited normal ovulation rates in TGF1 null mutant mice. After mating with wild-type stud males, serum progesterone content was reduced by 75% associated with altered ovarian expression of mRNAs encoding steroidogenic enzymes 3-hydroxysteroid dehydrogenase-1 and P450 17 -hydroxylase/C17–20-lyase. Embryos recovered from TGF1 null mutant females were developmentally arrested in the morula stage and rarely progressed to blastocysts. Attempts to rescue embryos by exogenous progesterone administration and in vitro culture were unsuccessful, and in vitro fertilization and culture experiments demonstrated that impaired development is unlikely to result from lack of maternal tract TGF1. We conclude that embryo arrest is due to developmental incompetence in oocytes developed in a TGF1-deficient follicular environment. This study demonstrates that TGF1 is a critical determinant of normal ovarian function, operating through regulation of LH activity and generation of oocytes competent for embryonic development and successful initiation of pregnancy.
Introduction
FEMALE REPRODUCTIVE function depends on extensive differentiation, proliferation, and remodeling of the reproductive tract tissues in a cyclic manner. These processes are driven by complex regulatory networks that are governed by gonadotropins and ovarian steroid hormones and mediated at the local level by the paracrine actions of an array of growth factors and cytokines. TGF1 has diverse roles in regulating cell differentiation and proliferation during tissue development, and this cytokine has been implicated in the molecular regulation of several reproductive processes spanning oocyte development and ovulation, early embryogenesis, embryo implantation, and placental morphogenesis (1, 2, 3).
Studies describing in vivo expression of TGF1 and its receptors as well as in vitro cell culture experiments implicate the TGF1 signaling pathway prominently in the events of early pregnancy. In the ovary, TGF1 may act as a local regulator of granulosa cell proliferation and steroid hormone production during follicular development (4, 5, 6), as well as in development and function of the corpus luteum (7, 8). TGF1 is expressed in developing oocytes (9, 10, 11) in which it may participate in oocyte-granulosa cell communication (5, 6). Maternal TGF1 transcripts appear to be sequestered in the oocyte and translated in the developing preimplantation embryo (12), in which TGF1 is attributed an autocrine trophic role (10, 13). At implantation, TGF1 is a potential key regulator of endometrial receptivity (2) and of trophoblast cell differentiation and invasiveness during placental morphogenesis (2, 14). However, the precise role and functional necessity for TGF1 in each of these processes in vivo remains undefined.
An opportunity to study the physiological significance of TGF1 in female reproductive function in vivo is provided by the availability of TGF1 null mutant mice. TGF1 null mice suffer severe multifocal inflammatory lesions and, on a conventional genetic background, do not survive past weaning age (15, 16). Inducing an immunocompromised state in TGF1 null mutant mice by treatment with dexamethasone (17) or anti-CD11 (18), mutation in the 2-microglobulin (19) or major histocompatibility complex class II (20) genes, or introduction of the severe combined immunodeficiency spontaneous mutation (scid) (18) prevents inflammatory pathology and prolongs their lifespan, enabling adult mice to be studied.
It is has been noted that the fertility of TGF1 null mutant mice is severely impaired (21). However, the reasons underpinning reproductive loss in TGF1-deficient mice have not been formally investigated, largely because of difficulties in generating and maintaining sufficient experimental mice to reproductive age. The aim of this study was to examine reproductive function in female TGF1 null mutant mice on the immunocompromised scid background. Here we demonstrate that TGF1 is indeed crucial to female reproductive success, primarily through its effects on gonadotropin regulation of the estrous cycle and ovulation, acquisition of oocyte competence, and early embryo development.
Materials and Methods
Mice and breeding experiments
All animal experiments were approved by the University of Adelaide Animal Ethics Committee and conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (6th edition, 1997). Mice homozygous for a targeted null mutation in the Tgf1 gene and homozygous for the Prkdcscid mutation (TGF1–/–) were bred on a mixed CF1/129/C3H background derived by backcrossing mice bearing a Tgf1 null allele (15) onto Prkdcscid/Prkdcscid mice (18). Mice were propagated using breeding pairs heterozygous for the Tgf1 null mutation and homozygous for the Prkdcscid mutation. The colony was maintained in specific pathogen-free conditions with controlled light (12-h light, 12-h dark cycle) and temperature.
Control mice used in these experiments were TGF1-replete littermates (TGF1+/– or +/+; termed TGF1+/±, unless specified otherwise). B10 male mice (ARC, Perth, Australia) were used as stud males in mating experiments, and CBA F1 (CBA x C57BL/6) males (University of Adelaide, Central Animal House) were used to provide spermatozoa for in vitro fertilization.
The genotype of each mouse in the colony was determined by diagnostic PCR of tail tissue DNA, using the forward primer 5'-GAGAAGAACTGCTGTGTGCG together with the reverse primers (1) 5'-GTGTCCAGGCTCCAAATATAGG to detect the intact Tgf1 gene and (2) 5'-CTCGTCCTGCAGTTCATTCA, which detects the neomycin resistance gene used to disrupt Tgf1. Over the 3-yr duration of the project, a total of 616 female progeny from approximately 250 litters produced by TGF1+/– breeding pairs were genotyped, yielding 222 TGF1+/+, 295 TGF1+/–, and 99 TGF1–/– female mice (ratio of +/+ to +/– to –/–, 1.00:1.32:0.45).
To investigate pregnancy outcome, TGF1–/– and TGF1+/± female mice were caged 1:1 from the age of 8 wk with a stud B10 male of proven fertility and checked daily for the presence of a vaginal plug, indicative of a mating event. After mating, females were separated from males and monitored for visible signs of pregnancy and birth of pups.
Evaluation of onset of puberty and estrous cycle tracking
TGF1–/– and TGF1+/± female mice were observed daily from the time of weaning at 21 d for appearance of a vaginal opening. Once this occurred, daily histological analysis of vaginal smears was undertaken to determine onset of proestrus, defined by the presence of round epithelial cells (22). Mice were killed during the periovulatory period at 1600 h on the day of proestrus, and blood was recovered by heart puncture for serum estrogen, FSH, and LH assay.
Estrous cycles were monitored in another cohort of virgin TGF1–/– and TGF1+/± female mice for 28 d from 6 wk of age by examining vaginal cytology (22). To encourage cycling activity, stud male mice were housed adjacent to females, and bedding from the male cages was added to the female cages. Estrous cycle length was defined as the interval between onset of one estrus event and onset of the next estrus event.
Assessment of early embryo development
To analyze embryo development in vivo, 8-wk-old TGF1–/– and TGF1+/± female mice were caged 1:1 with a stud male B10 mouse of proven fertility and checked daily for the presence of a vaginal plug, nominated d 0.5 postcoitus (pc). Some mice were given progesterone (Sigma, St. Louis, MO) (1 mg in peanut oil, sc) on d 1.5 and 2.5 pc. On d 3.5 pc, between 1100 and 1200 h, blood was collected by cardiac puncture under deep anesthesia with avertin before the animals were killed by cervical dislocation. The reproductive tract was dissected and flushed to recover embryos. The embryos were assessed for developmental stage and classified as one-cell, premorula embryo, morula, blastocyst, or degenerating embryo. The number of corpora lutea on the ovaries were counted as an index of the number of oocytes ovulated, and the ovaries were weighed and then fixed in 10% buffered formalin and paraffin embedded. The uterus was also dissected and fixed for histological analysis.
To investigate the viability of embryos developed in vitro, prepubertal (27–29 d old) TGF1–/– and TGF1+/± females were superovulated by administration of 5 IU Folligon (Intervet, Boxmeer, Holland), followed 50 h later by 5 IU Chorulon (Intervet), in a protocol proven in our hands to yield approximately 30 oocytes from prepubertal CBA x C57BL/6 F1 mice. The oocytes were fertilized either in vitro or in vivo, followed in both cases by in vitro culture to blastocysts. The in vitro fertilization procedure used sperm retrieved from the epididymis of CBA F1 males and capacitated for 1 h in pregassed rS1 media (Vitrolife, Kungsbacka, Sweden), and oocytes were harvested from the ampulla of TGF1–/– and TGF1+/± females 14 h after human chorionic gonadotropin (hCG) injection. This protocol routinely achieved 77% cleavage and 78% blastocyst development when CBA F1 females were used. Sperm were cultured together with the oocytes for 4 h at a concentration of 106 sperm/ml. The oocytes were then washed and transferred to 30 μl drops of rS1 under oil at a density of 10 oocytes per drop. The in vivo fertilization procedure used B10 stud males housed 1:1 with superovulated females. Embryos were retrieved from the oviduct of females on d 0.5 pc and cultured in 30-μl droplets of pregassed rS1 media at a density of 10 oocytes per droplet. All embryos were assessed for cleavage on d 2 of culture and transferred to 30 μl droplets of SQC media (Vitrolife) under oil. Blastocyst formation was assessed on d 5 of culture.
Histological analysis of ovaries and uterus
Ovaries and uterus were dissected from virgin TGF1–/– and TGF1+/± females killed on the day of first proestrus or from mated mice on d 3.5 pc. Serial 7-μm sections of paraffin embedded tissue were cut, mounted on glass slides, and stained with hematoxylin and eosin. Ovarian follicles were classified according to the method of Pedersen and Peters (1968) (23). Briefly, type 1–3 follicles had up to one single layer of granulosa cells, type 4 follicles had two layers of granulosa cells, type 5 follicles had between three and five layers of granulosa cells and no atrum, type 6 follicles comprised a multilayered membrana granulosa with scattered deposits of antral fluid, and type 7 and 8 follicles had a large, well-developed antrum, without (type 7) or with (type 8) a cumulus stalk. Atretic follicles were characterized by the presence of a degenerated oocyte, a high proportion of pyknotic nuclei, and/or a disorganized granulosa structure. The number of follicles at each stage of development or in atresia was counted in every 20th section (approximately every 150 μm) through the entire ovary. The number of follicles per section and the percentage at each stage of development were calculated.
Serum hormone analysis
Serum progesterone and estradiol concentrations were measured using commercial RIA kits DSL-4800 and DSL-3400, respectively (Diagnostic Systems Laboratories, Webster, TX) according to the instructions of the manufacturer, at the National Association of Testing Authorities (Adelaide, Australia) accredited Reproductive Medicine Laboratories (Sydney, Australia). Assays were performed in duplicate, and all samples were measured in a single assay. The lower limits of detection were 8 pM estradiol and 0.3 nM progesterone. The within-assay coefficients of variation were 10.0% in the estradiol assay and 7.9% in the progesterone assay. All data points were above the assay thresholds. The precision of the estradiol assay at low serum estradiol content was confirmed in experiments using serum from mice after ovariectomy and exogenous steroid hormone replacement.
Serum FSH and LH were determined by RIA using reagents kindly provided by Dr. Parlow at National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (Bethesda, MD). Iodinated preparations and antisera used in the FSH and LH RIAs were rFSH-I-9 and anti-rFSH-S-11, and rLH-I-9 and anti-rLH-S-10, respectively. Results are expressed in terms of NIDDK mFSH-RP (AFP-5308D) and NIDDK mLH-RP (AFP-5306A). The iodination preparations were iodinated using Iodogen reagent (Pierce, Rockford, IL). Goat antirabbit IgG (GAR12; Monash Institute of Reproduction and Development, Monash University, Melbourne, Australia) was used as second antibody for the FSH and LH RIAs. Assays were performed in duplicate on 20-μl serum samples, and all samples were measured in a single assay. The lower limits of detection were 0.78 ng/ml FSH and 0.10 ng/ml LH. The within-assay coefficients of variation were 8.1% in the FSH assay and 7.3% in the LH assay. Data points below the assay threshold were assigned the threshold values for data analysis.
Ovarian steroidogenic enzyme mRNA analysis
Quantitative RT-PCR was used to analyze the quantity of steroidogenic acute regulatory protein (StAR), P450 side chain cleavage (P450scc), 3-hydroxysteroid dehydrogenase-1 (HSD31), and P450 17 -hydroxylase/C17–20-lyase (P450c17) mRNA in ovarian tissue from superovulated 27- to 29-d-old female mice killed 14 h after hCG injection. Primers to detect expression of StAR, HSD31, P450scc, and P450c17 mRNAs were designed using PrimerExpress 2.0 software from respective cDNA sequences (24, 25, 26) (GenBank accession no. AF195119). The intron/exon boundaries for StAR were identified using the complete murine sequence (GenBank accession no. AY032730), for P450scc by comparison with the rat sequence (27), and for HSD31 and P450c17 by comparison with the human sequences from Bain et al. (25) and Kagimoto et al. (28), respectively. The primers used were as follows: StAR, forward, 5'CCGGAGCAGAGTGGTGTCAT and reverse, 5'-TGCGATAGGACCTGGTTGATG; HSD31, forward, 5'-GGACAAAGTATTCCGACCAGAAAC and reverse, 5'CAGGCACTGGGCATCCA; P450scc, forward, 5'ACAGACGCATCAAGCAGCAA and reverse, 5'-CACTGCTGATGGACTCAAAGGA; P450c17, forward, 5'-TGGCTTTCCTGGTGCACAA and reverse, 5'-GTGTTCGACTGAAGCCTACATACTG and 18S (Universal 18S primers; Ambion Austin, TX). Ovaries were snap frozen in liquid N2 and stored at –70 C before RNA extraction. Total cellular RNA was extracted according to a modification of the method described by Chomczynski and Sacchi (29). After treatment with RNase-free DNase I (500 IU/ml; 60 min at 37 C) (Roche Molecular Biochemicals, Mannheim, Germany), first-strand cDNA was reverse transcribed from 1 μg RNA using a Superscript II RNase H Reverse Transcriptase kit (90 min at 43 C) (Invitrogen, Carlsbad, CA). The PCR amplification used reagents supplied in a 2x SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), and each reaction volume (20 μl total) contained of 0.5–1 μM 5' and 3' primers and 3 μl cDNA. The negative control included in each reaction consisted of H2O substituted for cDNA. PCR amplification was performed in triplicate in a ABI Prism 5700 Sequence Detection System (Applied Biosystems) according to the instructions of the manufacturer. The absence of alternative PCR products and primer dimers in the PCR product was shown by a single dissociation peak and confirmed by visualization of the PCR product after agarose gel electrophoresis. Serial dilutions of cDNA were analyzed to confirm a linear relationship between cDNA content and quantity of product across the amplification range. The efficiency of each PCR was determined by regression analysis of the fluorescence of the sample at each cycle number. Efficiencies of all PCRs were between 80 and 100%. Data were normalized for 18S mRNA expression using the 2-Ct method (30) and expressed as a ratio of the mean value for TGF1+/± control ovaries.
Statistical analysis
Data were analyzed with SPSS 11.0 for Windows software (SPSS, Chicago, IL), using independent samples t test or one-way ANOVA with post hoc Tukey’s test when appropriate. When data were not normally distributed, nonparametric Mann-Whitney U tests were used. Significance was inferred at P < 0.05. Data are mean ± SEM unless specified otherwise.
Results
Effect of TGF1 null mutation on female fertility
To determine the effect of TGF1 null mutation on the capacity of female mice to initiate and maintain pregnancy, TGF1+/± and TGF1–/– adult females were housed with normal stud B10 males and examined each day for the presence of vaginal plugs and subsequent birth of litters. Only 14 of 34 (41%) TGF1–/– females mated during a 28 d period compared with 35 of 35 (100%) TGF1+/± females. Of the TGF1 null females that did mate, the interval between caging with males and detection of the plug was (mean ± SEM) 12.3 ± 2.5 d compared with 5.2 ± 1.0 d for TGF1+/± females (P = 0.03). When TGF1 null mutant females did mate with B10 stud males, the incidence of establishing successful pregnancy was substantially reduced, with only 3 of 14 mated TGF1–/– mice producing live progeny at term compared with 100% viable pregnancy outcome in mated TGF1+/± females. The litter size in TGF1–/– females was 3.7 ± 1.8 pups (n = 3 litters of 2, 3, and 6 pups per litter) compared with 6.3 ± 1.0 pups (n = 35 litters) in TGF1+/± females (not significant, P > 0.05). The TGF1–/– females that did not give birth to pups did not show visible signs of pregnancy. There was no difference in mating interval or litter size between TGF1+/– and TGF1+/+ females (data not shown), so data from the two genotypes were pooled in this and subsequent experiments.
Effect of TGF1 null mutation on estrous cycling
The failure to mate, and extended interval between placement with a male and mating in TGF1 null mice, suggested disrupted estrous cycling activity. To investigate the effect of TGF1 null mutation on estrous cyclicity, the stage of estrous cycle was tracked by analysis of vaginal cytology for a period of 28 d from 6 wk of age. Cycles were severely perturbed in null mutant mice, with deficiency in TGF1 reducing the frequency of estrus events by 42% (P = 0.001) (Fig. 1, A and B) and increasing mean estrous cycle length by 70% (P = 0.01) (Fig. 1, A and B). Consistent with fewer estrus events was a significant reduction in the percentage of time spent in the proestrus stage of the cycle in TGF1–/– females (Fig. 1C). However, there was no difference in the proportion of time spent in other stages of the estrus cycle, because TGF1–/– females often demonstrated a prolonged estrus event, as indicated by 100% cornified vaginal epithelial cells present for 2 d or more (Fig. 1A).
Effect of TGF1 null mutation on onset of sexual maturity, follicle development, and ovarian hormone synthesis
To further investigate ovarian function, serum and ovarian tissue were recovered for analysis from TGF1+/± and TGF1–/– females during the first proestrus event after onset of sexual maturity. Sexual maturity as measured by age at appearance of vaginal opening and age at first proestrus was delayed in TGF1 null mutants. In TGF1+/± females, 21 of 21 mice achieved vaginal opening by 5 wk of age (30.2 ± 0.5 d after birth; n = 21) compared with TGF1–/– females, in which only 18 of 22 (82%) achieved vaginal opening by 8 wk of age (35.4 ± 1.6 d; n = 18; P = 0.005). TGF1 null mutants were older at the time of first proestrus after vaginal opening [30.8 ± 1.0 d in TGF1+/± females (n = 18) vs. 37.0 ± 1.8 d in TGF1–/– females (n = 12); P = 0.004]. This delay in puberty was associated with a 23% reduction in body weight at 4 wk of age in TGF1–/– females [10.7 ± 0.3 g in TGF1–/– females (n = 8) vs. 13.9 ± 0.5 g in TGF1+/± females (n = 10)].
There was no difference in the absolute weight or weight relative to body size of ovaries from TGF1+/± and TGF1–/– females collected at first proestrus [2.2 ± 0.2 mg in TGF1+/± females (n = 12) vs. 1.7 ± 0.2 mg in TGF1–/– females (n = 9)]. Analysis of histological sections of ovaries showed that follicles in the full range of expected stages of development (Fig. 2, A and C) and mature antral follicles with normal morphology were present regardless of genotype (Fig. 2, B and D). Quantitative analysis of follicle numbers showed that there was no effect of TGF1 status on the relative abundance of primordial/primary follicles (types 1–3), preantral follicles with multiple layers of granulosa cells (types 4, 5), or small antral (type 6) or large antral (types 7, 8) follicles (Fig. 2E). Furthermore, there was no effect on the proportion of atretic follicles (data not shown).
Gonadotropins and estradiol were measured in serum recovered during the periovulatory period. Serum LH content was reduced in TGF1–/– mice to approximately 35% of the mean value detected in TGF1+/± mice (P = 0.008) (Fig. 3A). There was no effect of genotype on serum FSH (Fig. 3B) or serum estradiol (Fig. 3C).
Effect of TGF1 null mutation on ovulation and ovarian steroid synthesis
To investigate the number of oocytes ovulated and their fate after mating in early pregnancy, TGF1+/± and TGF1–/– females were mated with B10 males and killed on d 3.5 pc. The number of oocytes ovulated was measured by counting the total number of fertilized or unfertilized oocytes flushed from the oviduct and uterus and the number of corpora lutea in whole intact ovaries. A total of 42% fewer embryos were flushed from TGF1–/– mice than from TGF1+/± controls (P = 0.0002) (Table 1). Consistent with this, there was a 40% reduction in the number of corpora lutea seen in mice deficient in TGF1 (P = 0.005) (Fig. 4E). Fewer corpora lutea were also clearly evident when the histology of ovarian tissue was examined in sections stained with hematoxylin and eosin (Fig. 4, A and C), although the size and structure of corpora lutea in tissues from TGF1–/– and TGF1+/± females were comparable (Fig. 4, B and D). The reduction in number of corpora lutea was associated with a 50% reduction in absolute ovarian weight in TGF1–/– females compared with TGF1+/± littermates (P < 0.001) (Fig. 4F), corresponding to a 35% reduction when expressed relative to body weight (P < 0.001) (Fig. 4G).
RIA of serum steroid hormone content at d 3.5 pc revealed an 80% reduction in mean progesterone concentration in TGF1–/– females when compared with TGF1+/± controls (P = 0.002) (Fig. 5A). Progesterone output when expressed per corpus luteum was similarly reduced (3.0 ± 1.5 vs. 12.9 ± 1.5 nM per corpus luteum for TGF1–/– and TGF1+/± females, respectively; P = 0.001). Serum estradiol concentration was not affected by genotype (Fig. 5B).
Responsiveness of TGF1 null mutant mice to gonadotropin replacement
To determine whether higher rates of ovulation could be elicited in TGF1 null mice, additional groups of prepubertal TGF1–/– and TGF1+/± females were treated with exogenous gonadotropins. Seven of 10 (70%) immature TGF1–/– female mice were responsive to ovarian hyperstimulation, ovulating 7.1 ± 1.5 oocytes, compared with 10 of 10 (100%) TGF1+/± immature females that ovulated 9.1 ± 1.2 oocytes (not significant, P > 0.05).
To investigate the effect of TGF1 null mutation on ovarian steroidogenesis, ovaries excised from gonadotropin-stimulated prepubertal females approximately 2 h after ovulation were analyzed, and steroidogenic enzyme mRNA expression was measured using quantitative RT-PCR. There was a 50% decrease in expression of mRNA encoding HSD31 and a 4-fold increase in P450c17 mRNA in TGF1–/– compared with TGF1+/± mice (Fig. 6). StAR and P450scc was not significantly altered in TGF1–/– and TGF1+/± ovaries.
Effect of TGF1 null mutation on preimplantation embryo development
The stage of preimplantation development achieved by fertilized oocytes was examined in embryos flushed from the reproductive tract of TGF1–/– and TGF1+/± females killed on d 3.5 pc. A comparable percentage of oocytes from TGF1+/± and TGF1–/– females were fertilized and had initiated cell division (Table 1). However, few embryos from TGF1–/– females were developed to the expected blastocyst stage, with the majority arrested in the morula stage of development (Table 1 and Fig. 7).
To investigate the cause of developmental failure in preimplantation embryos, a series of experiments were devised in an attempt to rescue development of fertilized oocytes from TGF1–/– mice. Administration of exogenous progesterone to TGF1–/– females on d 1.5 and 2.5 of pregnancy did not improve embryo development, as assessed on d 3.5 pc (Table 2). As was seen in untreated TGF1–/– females, a high rate of oocyte fertilization occurred, but most embryos were arrested at morula stage or earlier.
In another approach, designed to discriminate between deficiency in maternal tract vs. absence of embryonic TGF1 in causing embryo demise, embryos were fertilized and cultured ex vivo. The oocytes recovered from prepubertal mice after treatment with gonadotropins were fertilized by in vitro fertilization, or flushed from the reproductive tract after mating, and then cultured in vitro to investigate embryo development in a defined culture media proven to support early embryo development. In vitro fertilization with sperm from wild-type CBA F1 males resulted in comparable rates of cleavage of oocytes from TGF1+/± and TGF1–/– females. However, TGF1+/– embryos from TGF1–/– oocytes failed to develop to blastocyst stage and most frequently did not survive past two-cell stage (Table 2). Similarly, when zygotes were flushed from the oviduct on d 0.5 pc after mating superovulated TGF1–/– and TGF1+/± mice with wild-type B10 males and then cultured in vitro for 4 d, embryos from TGF1–/– oocytes again failed to develop to blastocyst stage (Table 2). In contrast, oocytes from TGF1+/± mice fertilized either in vivo or in vitro developed to blastocyst at the expected high rate.
Effect of TGF1 null mutation on uterine morphology
To examine the effect of TGF1 null mutation on uterine morphology, sections of fixed, paraffin-embedded uteri from d 3.5 pc naturally mated females were stained with hematoxylin and eosin (Fig. 8). There was no overt effect of genotype on the tissue mass or diameter of uterine tissue (data not shown). The uteri from TGF1–/– females were morphologically indistinguishable from uteri dissected from TGF1+/± females in terms of luminal diameter, presence and relative thickness of endometrial and myometrial tissues, and the abundance and size of uterine glands.
Discussion
This study indicates a critical role for TGF1 in the reproductive function of female mice. We have demonstrated three distinct lesions in female fertility caused by TGF1 null mutation. First, ovarian function is perturbed with irregular progression of the estrous cycle associated with a reduction in circulating LH during the characteristic proestrus surge, resulting in less frequent ovulatory events and fewer oocytes ovulated. Second, corpus luteum function as measured by serum progesterone content is compromised. Third, preimplantation embryo development is perturbed in TGF1 null mutant females. Together, these disruptions dramatically reduce the incidence of successful pregnancy and production of live-born progeny to females mice deficient in TGF1.
The major determinant of reduced fecundity in null mutant mice was that ovulation occurred approximately 40% less frequently than in wild-type controls, and, when it did occur, approximately 40% fewer oocytes were ovulated. Synthesis of the gonadotropin LH was severely reduced at the expected time of the LH surge in TGF1 null mutant females. In contrast, administration of exogenous gonadotropins to immature TGF1–/– mice successfully induced ovulation, with a response equal to that seen in wild-type mice in terms of the proportion of mice that ovulated and the number of oocytes released. The effects of TGF1 deficiency on estrous cycling and ovulation can thus largely be attributed to disrupted LH activity, suggesting a requirement for TGF1 in normal hypothalamic or pituitary function. Mice with a mutation in the LH subunit are totally infertile and anovulatory (31), showing the absolute requirement for this hormone in estrous cycling and ovulation. LH was detectable at low levels in most of the TGF1–/– mice, indicating that synthesis must be impaired or dysregulated rather than completely absent. This is consistent with the wide variance in ovarian cycle disturbances manifesting both at a population level and within individual mice. Some null mutant mice cycled more successfully than others, which in the worst cases, showed only one estrus event over the 28-d period of study. More than half of the TGF1 null females failed to mate even once during the 28-d period of caging with males, which might reflect a more severe LH deficiency in these animals.
Although growth trajectory after birth was delayed in TGF1 null mutant mice and their onset of puberty occurred on average 6 d later than in wild-type mice, this was not causally associated with reduced LH at first proestrus, because there was no correlation between body weight and serum LH content in individual mice and, furthermore, the mean body weight of mice at proestrus was similar regardless of genotype (data not shown). Therefore, although TGF1 appears to be required for normal growth rate after birth and this likely explains the delay in puberty in TGF1 null mutant mice, growth impairment is unlikely to account for disrupted LH activity at the time of onset of ovarian cycling.
The observation that approximately 10% of null mutant mice were able to ovulate, mate, and maintain pregnancy successfully to deliver overtly normal pups at term shows that, although TGF1 is clearly essential for normal fecundity, other factors or circumstances must attenuate the severity of effects caused by absence of this cytokine. Because the proportion of null mutant mice able to establish pregnancy after mating was comparable with the proportion in which embryos reached blastocyst stage, it is reasonable to infer that postimplantation loss is not a major contributing factor in reduced fecundity in TGF1-deficient female mice. This suggests that implantation and placental morphogenesis can proceed relatively unimpaired in the absence of maternal TGF1 but does not exclude the possibility that a more rigorous analysis might reveal deficits in the quality of placental structure and function or in fetal development and growth trajectory. A gene dosage effect and rescue by transfer of maternal TGF1 across the placenta is evident when TGF1-deficient fetuses gestate in mothers heterozygous for the TGF1 null mutation (17, 21, 32), but the current data show that maternal cytokine is not an essential requirement for fetal development, at least in the event of heterozygousity for the TGF1 mutation in embryos.
The variability in fertility phenotype seen in the current experiments is likely to reflect differences in the extent to which individual mice carry genes conferring compensation for TGF1 deficiency or other interactions with TGF1 status. It has been noted previously that genetic background has a significant influence on penetrance of embryonic lethality due to defects in pre- and postimplantation development in TGF1–/– embryos (21, 33). On the mixed genetic background CF1/129/C3H used in the current study, breeding pairs heterozygous for the TGF1 null mutation were reasonably fertile, but, in concurrence with previous reports (21, 34), we observed less than the expected number of pups genotyped as homozygous for the null mutation, reflecting embryonic lethality in approximately 50% of homozygous offspring and severely constraining availability of experimental mice. Our attempts to backcross the TGF1 null genotype onto the C57BL/6 background resulted in progressively declining fertility, as experienced previously by others, and can be attributed to a small number of genetic modifiers (21, 33). The number of heterozygous offspring was also approximately 50% less than would be expected, suggesting a Tgf1 gene dosage effect in the developing fetal-placenta unit. Previous studies have reported a selective loss of both TGF1–/– and TGF1+/– conceptuses in TGF1+/– intercrosses (34).
Impaired ovarian function persisted after ovulation, with diminished progesterone production by corpora lutea. Interestingly, estrogen synthesis was unaffected, both during the periovulatory period and in early pregnancy. This suggests that follicular estrogen synthesis is less sensitive to TGF1 deficiency than luteal cell progesterone synthesis and is consistent with the dominant function of LH in initiating luteinization compared with the greater dependence of follicle development on FSH, the circulating level of which was unchanged in the absence of TGF1. Reduced serum progesterone content was seen even when the smaller number of corpora lutea was taken into account and was associated with alterations in expression of mRNAs encoding components of the steroidogenic pathway. HSD31 and P450c17 mRNA expression in ovaries from immature gonadotropin-stimulated TGF1 null mice were decreased and increased, respectively, when compared with synthesis in ovaries from TGF1-replete littermates. HSD31 catalyzes the conversion of pregnenolone to progesterone, and P450c17 catalyzes the conversion of progesterone to androstenedione. The decreased and increased relative expression of these enzymes in TGF1 null mutants is consistent with the observation of reduced progesterone production with normal estradiol production in naturally mated TGF1 null females. Therefore, TGF1 deficiency appears to cause impaired communication in the paracrine or autocrine signals that regulate progesterone synthesis in the corpus luteum.
Although the steroidogenic defect can in part be attributed to reduced LH, altered expression of steroidogenic enzymes was evident after gonadotropin replacement, suggesting that LH-independent mechanisms may also contribute. Inhibitory effects of TGF1 deficiency on progesterone synthesis or secretion might be mediated through direct regulatory effects of this cytokine on luteal cells, because luteal cells express TGF receptors and are responsive to this cytokine in vitro (7, 8). Alternatively, ovarian macrophages potentially act as a mediating cell by virtue of their high responsiveness to TGF1 (35) and specific roles in development and function of the corpus luteum (36, 37, 38). Macrophage-secreted cytokines TNF- and IL-1 are implicated in regulating luteal steroidogenesis (39), and disrupted macrophage regulation leading to overproduction of these antisteroidogenic cytokines in the absence of TGF1 may contribute to the reduced serum progesterone seen in early pregnancy. The ovarian phenotype we describe in TGF1 null mutant mice bears remarkable similarity to that seen in macrophage-deficient Csf1op/Csf1op mice, which also have disrupted ovarian cycles, reduced ovulation rates, and impaired steroidogenesis associated with reduced LH (40). In Csf1op/Csf1op mice, ovarian effects appear to be largely secondary to altered hypothalamic activity caused by absence of macrophages in that tissue (41), so it would be of interest to examine whether hypothalamic macrophage function is similarly altered in TGF1 null mutants.
Despite infrequent estrus and reduction in oocytes ovulated, approximately 40% of TGF1 null females were able to mate. However, in 80% of mice in which mating occurred, pregnancy failed due to impaired preimplantation embryo development. Developmental arrest appeared not to be due to an adverse maternal tract environment secondary to TGF1 or progesterone deficiency but rather was intrinsic to the embryo, because strategies including progesterone replacement or ex vivo fertilization and embryo culture failed to rescue embryonic demise. Furthermore, histological evaluation of the uterus revealed no differences in the structure of the endometrium and glands. This finding does not preclude a role for maternal TGF1 in the reproductive tract but does show that TGF1 deficiency in the oviduct and uterus is an insufficient explanation for the embryo lethality observed.
It is possible that TGF1+/– early embryos had reduced developmental competence by virtue of their depleted capacity to synthesize endogenous TGF1 after embryonic genome activation. Embryos normally synthesize TGF1 (10, 13) and express TGF receptors (10, 11), supporting an autocrine role for TGF1 in preimplantation development (42). A premorula lethality in null embryos has been reported in TGF1 heterozygous breeder pairs and contributes to the less than expected proportion of TGF1–/– offspring observed (21). However, it is unlikely that reduced embryonic TGF1 synthesis is responsible for the developmental failure in embryos from TGF1–/– females mated with TGF1+/+ males described here. In vitro fertilization using sperm from TGF1–/– males and TGF1+/+ oocytes shows that the cleaved embryos (all TGF1+/–) develop to blastocysts at a similar rate to those oocytes fertilized by TGF1+/+ sperm (Ingman, W. V., and S. A. Robertson, manuscript in preparation). Therefore, when the null mutation is paternally inherited, TGF1+/– embryos develop to the blastocyst stage normally. The possibility that maternally imprinted expression of the Tgf1 gene in the developing embryo is required can be disregarded because whether the null gene was inherited maternally vs. paternally was found not to influence the expected frequency of the null gene in offspring (21).
The most compelling explanation for embryonic arrest in null mutant mice is oocyte incompetence, caused by developmental defects occurring during oocyte growth and development in the TGF1 null ovary, such that those follicles that do progress to ovulation yield oocytes less capable of supporting development after fertilization. It is well known that, as oocytes grow and develop in the ovarian follicle, they gradually and sequentially acquire the cellular and molecular machinery required to support early embryogenesis, and that the quality of the follicular milieu directly affects the developmental potential of the ensuing embryo (43, 44). Perturbation can have adverse consequences for structural organization of the embryo, activation of the embryonic genome, and subsequent gene expression patterns several cell cycles into embryo development, often manifesting in arrest during the preimplantation phase (44). Oocytes from TGF1 null mutants appear to be fully meiotically competent (capable of reaching metaphase II), as evidenced by comparable fertilization frequencies with control mice. Impaired cytoplasmic competence of TGF1 null mutant oocytes is more likely, potentially due to absence of oocyte TGF1 protein or mRNA, because TGF1 is known to be a maternally stored transcript (12). Alternatively, oocyte incompetence might reflect a more generalized perturbation of oocyte development in a suboptimal, TGF-deficient follicular environment.
In summary, we conclude that TGF1 deficiency severely impacts female fertility through disrupted ovulation, oocyte development, and corpus luteum function. Absence of maternal TGF1 causes impaired LH activity and also appears to impact the maturing oocyte in such a way as to reduce developmental competence after ovulation and fertilization. The relevance of this work to human follicle development is highlighted by comparable expression patterns for TGF1 and its signaling molecules in the human ovary (45, 46) and the observation that levels of TGF1 in follicular fluid correlate with pregnancy success after in vitro fertilization (47).
Acknowledgments
We thank Dr. Anne O’Connor at Monash Institute for Reproduction and Development (Melbourne, Australia) for performing LH and FSH RIAs, Mr. Alan Gilmore at Reproductive Medicine Laboratories (Adelaide, Australia) for performing ovarian steroid hormone RIAs, Dr. Darryl Russell for assistance with ovarian histology, and Dr. Robert Gilchrist for helpful advice and comments.
Footnotes
This work was supported by the Australian Research Council Discovery Grant DP0211497, a National Health and Medical Research Council of Australia Research Fellowship (to S.A.R.), and a C. J. Martin Career Development Award (to W.V.I.).
W.V.I., R.L.R., and K.W. have nothing to declare. S.A.R. consults for GroPep Ltd. (Adelaide, Australia).
First Published Online November 3, 2005
Abbreviations: hCG, Human chorionic gonadotropin; HSD31, 3-hydroxysteroid dehydrogenase-1; P450c17, P450 17 -hydroxylase/C17–20-lyase; P450scc, P450 side chain cleavage; pc, postcoitus; scid, severe combined immunodeficiency spontaneous mutation; StAR, steroidogenic acute regulatory protein.
Accepted for publication October 21, 2005.
References
Shull MM, Doetschman T 1994 Transforming growth factor-1 in reproduction and development. Mol Reprod Dev 39:239–246
Godkin JD, Dore JJ 1998 Transforming growth factor and the endometrium. Rev Reprod 3:1–6
Ingman WV, Robertson SA 2002 Defining the actions of transforming growth factor in reproduction. BioEssays 24:904–914
Adashi EY, Resnick CE, Hernandez ER, May JV, Purchio AF, Twardzik DR 1989 Ovarian transforming growth factor- (TGF): cellular site(s), and mechanism(s) of action. Mol Cell Endocrinol 61:247–256
Roy SK 1993 Epidermal growth factor and transforming growth factor- modulation of follicle-stimulating hormone-induced deoxyyribonucleic acid synthesis in hamster preantral and early antral follicles. Biol Reprod 48:552–557
Vanderhyden BC, Macdonald EA, Nagyova E, Dhawan A 2003 Evaluation of members of the TGF superfamily as candidates for the oocyte factors that control mouse cumulus expansion and steroidogenesis. Reprod Suppl 61:55–70
Matsuyama S, Shiota K, Takahashi M 1990 Possible role of transforming growth factor- as a mediator of luteotropic action of prolactin in rat luteal cell cultures. Endocrinology 127:1561–1567
Matsubara H, Ikuta K, Ozaki Y, Suzuki Y, Suzuki N, Sato T, Suzumori K 2000 Gonadotropins and cytokines affect luteal function through control of apoptosis in human luteinized granulosa cells. J Clin Endocrinol Metab 85:1620–1626
Roy SK, Ogren C, Roy C, Lu B 1992 Cell-type-specific localization of transforming growth factor-2 and transforming growth factor-1 in the hamster ovary: differential regulation by follicle-stimulating hormone and luteinizing hormone. Biol Reprod 46:595–606
Paria BC, Jones KL, Flanders KC, Dey SK 1992 Localization and binding of transforming growth factor- isoforms in mouse preimplantation embryos and in delayed and activated blastocysts. Dev Biol 151:91–104
Chow JF, Lee KF, Chan ST, Yeung WS 2001 Quantification of transforming growth factor 1 (TGF1) mRNA expression in mouse preimplantation embryos and determination of TGF receptor (type I and type II) expression in mouse embryos and reproductive tract. Mol Hum Reprod 7:1047–1056
Gilchrist RB, Morrissey MP, Ritter LJ, Armstrong DT 2003 Comparison of oocyte factors and transforming growth factor- in the regulation of DNA synthesis in bovine granulosa cells. Mol Cell Endocrinol 201:87–95
Rappolee DA, Brenner CA, Schultz R, Mark D, Werb Z 1988 Developmental expression of PDGF, TGF-, and TGF- genes in preimplantation mouse embryos. Science 241:1823–1825
Morrish DW, Dakour J, Li H 1998 Functional regulation of human trophoblast differentiation. J Reprod Immunol 39:179–195
Shull MM, Ormsby I, Kier AB, Pawlowski S, Diebold RJ, Yin M, Allen R, Sidman C, Proetzel G, Calvin D, Annunziata N, Doetschman, T 1992 Targeted disruption of the mouse transforming growth factor-1 gene results in multifocal inflammatory disease. Nature 359:693–699
Kulkarni AB, Huh CG, Becker D, Geiser A, Lyght M, Flanders KC, Roberts AB, Sporn MB, Ward JM, Karlsson S 1993 Transforming growth factor 1 null mutation in mice causes excessive inflammatory response and early death. Proc Natl Acad Sci USA 90:770–774
Letterio JJ, Geiser AG, Kulkarni AB, Roche NS, Sporn MB, Roberts AB 1994 Maternal rescue of transforming growth factor-1 null mice. Science 264:1936–1938
Diebold RJ, Eis MJ, Yin M, Ormsby I, Boivin GP, Darrow BJ, Saffitz JE, Doetschman T 1995 Early-onset multifocal inflammation in the transforming growth factor 1-null mouse is lymphocyte mediated. Proc Natl Acad Sci USA 92:12215–12219
Kobayashi S, Yoshida K, Ward JM, Letterio JJ, Longenecker G, Yaswen L, Mittleman B, Mozes E, Roberts AB, Karlsson S, Kulkarni AB 1999 2-microglobulin-deficient background ameliorates lethal phenotype of the TGF-1 null mouse. J Immunol 163:4013–4019
Letterio JJ, Geiser AG, Kulkarni AB, Dang H, Kong L, Nakabayashi T, Mackall CL, Gress RE, Roberts AB 1996 Autoimmunity associated with TGF-1-deficiency in mice is dependent on MHC class II antigen expression. J Clin Invest 98:2109–2119
Kallapur S, Ormsby I, Doetschman T 1999 Strain dependency of TGF1 function during embryogenesis. Mol Reprod Dev 52:341–349
Allen E 1921 The oestrous cycle in the mouse. Am J Anat 30:297–348
Pedersen T, Peters H 1968 Proposal for a classification of oocytes and follicles in the mouse ovary. J Reprod Fertil 17:555–557
Clark BJ, Wells J, King SR, Stocco DM 1994 The purification, cloning, and expression of a novel luteinizing hormone-induced mitochondrial protein in MA-10 mouse Leydig tumor cells. Characterization of the steroidogenic acute regulatory protein (StAR). J Biol Chem 269:28314–28322
Bain PA, Yoo M, Clarke T, Hammond SH, Payne AH 1991 Multiple forms of mouse 3-hydroxysteroid dehydrogenase/5-4 isomerase and differential expression in gonads, adrenal glands, liver, and kidneys of both sexes. Proc Natl Acad Sci USA 88:8870–8874
Youngblood GL, Sartorius C, Taylor BA, Payne AH 1991 Isolation, characterization, and chromosomal mapping of mouse P450 17-hydroxylase/C17–20 lyase. Genomics 10:270–275
Oonk RB, Parker KL, Gibson JL, Richards JS 1990 Rat cholesterol side-chain cleavage cytochrome P-450 (P-450scc) gene. Structure and regulation by cAMP in vitro. J Biol Chem 265:22392–22401
Kagimoto M, Winter JS, Kagimoto K, Simpson ER, Waterman MR 1988 Structural characterization of normal and mutant human steroid 17-hydroxylase genes: molecular basis of one example of combined 17-hydroxylase/17,20 lyase deficiency. Mol Endocrinol 2:564–570
Chomczynski P, Sacchi N 1987 Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156–159
Livak KJ, Schmittgen TD 2001 Analysis of relative gene expression data using real-time quantitative PCR and the 2(-C(T)) method. Methods 25:402–408
Ma X, Dong Y, Matzuk MM, Kumar TR 2004 Targeted disruption of luteinizing hormone -subunit leads to hypogonadism, defects in gonadal steroidogenesis, and infertility. Proc Natl Acad Sci USA 101:17294–17299
McLennan IS, Koishi K 2004 Fetal and maternal transforming growth factor-1 may combine to maintain pregnancy in mice. Biol Reprod 70:1614–1618
Bonyadi M, Rusholme SA, Cousins FM, Su HC, Biron CA, Farrall M, Akhurst RJ 1997 Mapping of a major genetic modifier of embryonic lethality in TGF1 knockout mice. Nat Genet 15:207–211
Dickson MC, Martin JS, Cousins FM, Kulkarni AB, Karlsson S, Akhurst RJ 1995 Defective haematopoiesis and vasculogenesis in transforming growth factor-1 knock out mice. Development 121:1845–1854
Bogdan C, Nathan C 1993 Modulation of macrophage function by transforming growth factor , interleukin-4, and interleukin-10. Ann NY Acad Sci 685:713–739
Fukumatsu Y, Katabuchi H, Naito M, Takeya M, Takahashi K, Okamura H 1992 Effect of macrophages on proliferation of granulosa cells in the ovary in rats. J Reprod Fertil 96:241–249
Van der Hoek KH, Maddocks S, Woodhouse CM, van Rooijen N, Robertson SA, Norman RJ 2000 Intrabursal injection of clodronate liposomes causes macrophage depletion and inhibits ovulation in the mouse ovary. Biol Reprod 62:1059–1066
Matsuyama S, Takahashi M 1995 Immunoreactive (ir)-transforming growth factor (TGF)- in rat corpus luteum: ir-TGF is expressed by luteal macrophages. Endocr J 42:203–217
Pate JL, Landis Keyes P 2001 Immune cells in the corpus luteum: friends or foes Reproduction 122:665–676
Cohen PE, Zhu L, Pollard JW 1997 Absence of colony stimulating factor-1 in osteopetrotic (csfmop/csfmop) mice disrupts estrous cycles and ovulation. Biol Reprod 56:110–118
Cohen PE, Zhu L, Nishimura K, Pollard JW 2002 Colony-stimulating factor 1 regulation of neuroendocrine pathways that control gonadal function in mice. Endocrinology 143:1413–1422
Paria BC, Dey SK 1990 Preimplantation embryo development in vitro: cooperative interactions among embryos and role of growth factors. Proc Natl Acad Sci USA 87:4756–4760
Rodriguez KF, Farin CE 2004 Gene transcription and regulation of oocyte maturation. Reprod Fertil Dev 16:55–67
Duranthon V, Renard JP 2001 The developmental competence of mammalian oocytes: a convenient but biologically fuzzy concept. Theriogenology 55:1277–1289
Chegini N, Flanders KC 1992 Presence of transforming growth factor- and their selective cellular localization in human ovarian tissue of various reproductive stages. Endocrinology 130:1707–1715
Osterlund C, Fried G 2000 TGF receptor types I and II and the substrate proteins Smad 2 and 3 are present in human oocytes. Mol Hum Reprod 6:498–503
Fried G, Wramsby H 1998 Increase in transforming growth factor 1 in ovarian follicular fluid following ovarian stimulation and in-vitro fertilization correlates to pregnancy. Hum Reprod 13:656–659(Wendy V. Ingman, Rebecca L. Robker, Kare)
Abstract
TGF1 is implicated in regulation of ovarian function and the events of early pregnancy. We have investigated the effect of null mutation in the Tgf1 gene on reproductive function in female mice. The reproductive capacity of TGF1 null mutant females was severely impaired, leading to almost complete infertility. Onset of sexual maturity was delayed, after which ovarian function was disrupted, with extended ovarian cycles, irregular ovulation, and a 40% reduction in oocytes ovulated. Serum FSH and estrogen content were normal, but TGF1 null mutant mice failed to display the characteristic proestrus surge in circulating LH. Ovarian hyperstimulation with exogenous gonadotropins elicited normal ovulation rates in TGF1 null mutant mice. After mating with wild-type stud males, serum progesterone content was reduced by 75% associated with altered ovarian expression of mRNAs encoding steroidogenic enzymes 3-hydroxysteroid dehydrogenase-1 and P450 17 -hydroxylase/C17–20-lyase. Embryos recovered from TGF1 null mutant females were developmentally arrested in the morula stage and rarely progressed to blastocysts. Attempts to rescue embryos by exogenous progesterone administration and in vitro culture were unsuccessful, and in vitro fertilization and culture experiments demonstrated that impaired development is unlikely to result from lack of maternal tract TGF1. We conclude that embryo arrest is due to developmental incompetence in oocytes developed in a TGF1-deficient follicular environment. This study demonstrates that TGF1 is a critical determinant of normal ovarian function, operating through regulation of LH activity and generation of oocytes competent for embryonic development and successful initiation of pregnancy.
Introduction
FEMALE REPRODUCTIVE function depends on extensive differentiation, proliferation, and remodeling of the reproductive tract tissues in a cyclic manner. These processes are driven by complex regulatory networks that are governed by gonadotropins and ovarian steroid hormones and mediated at the local level by the paracrine actions of an array of growth factors and cytokines. TGF1 has diverse roles in regulating cell differentiation and proliferation during tissue development, and this cytokine has been implicated in the molecular regulation of several reproductive processes spanning oocyte development and ovulation, early embryogenesis, embryo implantation, and placental morphogenesis (1, 2, 3).
Studies describing in vivo expression of TGF1 and its receptors as well as in vitro cell culture experiments implicate the TGF1 signaling pathway prominently in the events of early pregnancy. In the ovary, TGF1 may act as a local regulator of granulosa cell proliferation and steroid hormone production during follicular development (4, 5, 6), as well as in development and function of the corpus luteum (7, 8). TGF1 is expressed in developing oocytes (9, 10, 11) in which it may participate in oocyte-granulosa cell communication (5, 6). Maternal TGF1 transcripts appear to be sequestered in the oocyte and translated in the developing preimplantation embryo (12), in which TGF1 is attributed an autocrine trophic role (10, 13). At implantation, TGF1 is a potential key regulator of endometrial receptivity (2) and of trophoblast cell differentiation and invasiveness during placental morphogenesis (2, 14). However, the precise role and functional necessity for TGF1 in each of these processes in vivo remains undefined.
An opportunity to study the physiological significance of TGF1 in female reproductive function in vivo is provided by the availability of TGF1 null mutant mice. TGF1 null mice suffer severe multifocal inflammatory lesions and, on a conventional genetic background, do not survive past weaning age (15, 16). Inducing an immunocompromised state in TGF1 null mutant mice by treatment with dexamethasone (17) or anti-CD11 (18), mutation in the 2-microglobulin (19) or major histocompatibility complex class II (20) genes, or introduction of the severe combined immunodeficiency spontaneous mutation (scid) (18) prevents inflammatory pathology and prolongs their lifespan, enabling adult mice to be studied.
It is has been noted that the fertility of TGF1 null mutant mice is severely impaired (21). However, the reasons underpinning reproductive loss in TGF1-deficient mice have not been formally investigated, largely because of difficulties in generating and maintaining sufficient experimental mice to reproductive age. The aim of this study was to examine reproductive function in female TGF1 null mutant mice on the immunocompromised scid background. Here we demonstrate that TGF1 is indeed crucial to female reproductive success, primarily through its effects on gonadotropin regulation of the estrous cycle and ovulation, acquisition of oocyte competence, and early embryo development.
Materials and Methods
Mice and breeding experiments
All animal experiments were approved by the University of Adelaide Animal Ethics Committee and conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (6th edition, 1997). Mice homozygous for a targeted null mutation in the Tgf1 gene and homozygous for the Prkdcscid mutation (TGF1–/–) were bred on a mixed CF1/129/C3H background derived by backcrossing mice bearing a Tgf1 null allele (15) onto Prkdcscid/Prkdcscid mice (18). Mice were propagated using breeding pairs heterozygous for the Tgf1 null mutation and homozygous for the Prkdcscid mutation. The colony was maintained in specific pathogen-free conditions with controlled light (12-h light, 12-h dark cycle) and temperature.
Control mice used in these experiments were TGF1-replete littermates (TGF1+/– or +/+; termed TGF1+/±, unless specified otherwise). B10 male mice (ARC, Perth, Australia) were used as stud males in mating experiments, and CBA F1 (CBA x C57BL/6) males (University of Adelaide, Central Animal House) were used to provide spermatozoa for in vitro fertilization.
The genotype of each mouse in the colony was determined by diagnostic PCR of tail tissue DNA, using the forward primer 5'-GAGAAGAACTGCTGTGTGCG together with the reverse primers (1) 5'-GTGTCCAGGCTCCAAATATAGG to detect the intact Tgf1 gene and (2) 5'-CTCGTCCTGCAGTTCATTCA, which detects the neomycin resistance gene used to disrupt Tgf1. Over the 3-yr duration of the project, a total of 616 female progeny from approximately 250 litters produced by TGF1+/– breeding pairs were genotyped, yielding 222 TGF1+/+, 295 TGF1+/–, and 99 TGF1–/– female mice (ratio of +/+ to +/– to –/–, 1.00:1.32:0.45).
To investigate pregnancy outcome, TGF1–/– and TGF1+/± female mice were caged 1:1 from the age of 8 wk with a stud B10 male of proven fertility and checked daily for the presence of a vaginal plug, indicative of a mating event. After mating, females were separated from males and monitored for visible signs of pregnancy and birth of pups.
Evaluation of onset of puberty and estrous cycle tracking
TGF1–/– and TGF1+/± female mice were observed daily from the time of weaning at 21 d for appearance of a vaginal opening. Once this occurred, daily histological analysis of vaginal smears was undertaken to determine onset of proestrus, defined by the presence of round epithelial cells (22). Mice were killed during the periovulatory period at 1600 h on the day of proestrus, and blood was recovered by heart puncture for serum estrogen, FSH, and LH assay.
Estrous cycles were monitored in another cohort of virgin TGF1–/– and TGF1+/± female mice for 28 d from 6 wk of age by examining vaginal cytology (22). To encourage cycling activity, stud male mice were housed adjacent to females, and bedding from the male cages was added to the female cages. Estrous cycle length was defined as the interval between onset of one estrus event and onset of the next estrus event.
Assessment of early embryo development
To analyze embryo development in vivo, 8-wk-old TGF1–/– and TGF1+/± female mice were caged 1:1 with a stud male B10 mouse of proven fertility and checked daily for the presence of a vaginal plug, nominated d 0.5 postcoitus (pc). Some mice were given progesterone (Sigma, St. Louis, MO) (1 mg in peanut oil, sc) on d 1.5 and 2.5 pc. On d 3.5 pc, between 1100 and 1200 h, blood was collected by cardiac puncture under deep anesthesia with avertin before the animals were killed by cervical dislocation. The reproductive tract was dissected and flushed to recover embryos. The embryos were assessed for developmental stage and classified as one-cell, premorula embryo, morula, blastocyst, or degenerating embryo. The number of corpora lutea on the ovaries were counted as an index of the number of oocytes ovulated, and the ovaries were weighed and then fixed in 10% buffered formalin and paraffin embedded. The uterus was also dissected and fixed for histological analysis.
To investigate the viability of embryos developed in vitro, prepubertal (27–29 d old) TGF1–/– and TGF1+/± females were superovulated by administration of 5 IU Folligon (Intervet, Boxmeer, Holland), followed 50 h later by 5 IU Chorulon (Intervet), in a protocol proven in our hands to yield approximately 30 oocytes from prepubertal CBA x C57BL/6 F1 mice. The oocytes were fertilized either in vitro or in vivo, followed in both cases by in vitro culture to blastocysts. The in vitro fertilization procedure used sperm retrieved from the epididymis of CBA F1 males and capacitated for 1 h in pregassed rS1 media (Vitrolife, Kungsbacka, Sweden), and oocytes were harvested from the ampulla of TGF1–/– and TGF1+/± females 14 h after human chorionic gonadotropin (hCG) injection. This protocol routinely achieved 77% cleavage and 78% blastocyst development when CBA F1 females were used. Sperm were cultured together with the oocytes for 4 h at a concentration of 106 sperm/ml. The oocytes were then washed and transferred to 30 μl drops of rS1 under oil at a density of 10 oocytes per drop. The in vivo fertilization procedure used B10 stud males housed 1:1 with superovulated females. Embryos were retrieved from the oviduct of females on d 0.5 pc and cultured in 30-μl droplets of pregassed rS1 media at a density of 10 oocytes per droplet. All embryos were assessed for cleavage on d 2 of culture and transferred to 30 μl droplets of SQC media (Vitrolife) under oil. Blastocyst formation was assessed on d 5 of culture.
Histological analysis of ovaries and uterus
Ovaries and uterus were dissected from virgin TGF1–/– and TGF1+/± females killed on the day of first proestrus or from mated mice on d 3.5 pc. Serial 7-μm sections of paraffin embedded tissue were cut, mounted on glass slides, and stained with hematoxylin and eosin. Ovarian follicles were classified according to the method of Pedersen and Peters (1968) (23). Briefly, type 1–3 follicles had up to one single layer of granulosa cells, type 4 follicles had two layers of granulosa cells, type 5 follicles had between three and five layers of granulosa cells and no atrum, type 6 follicles comprised a multilayered membrana granulosa with scattered deposits of antral fluid, and type 7 and 8 follicles had a large, well-developed antrum, without (type 7) or with (type 8) a cumulus stalk. Atretic follicles were characterized by the presence of a degenerated oocyte, a high proportion of pyknotic nuclei, and/or a disorganized granulosa structure. The number of follicles at each stage of development or in atresia was counted in every 20th section (approximately every 150 μm) through the entire ovary. The number of follicles per section and the percentage at each stage of development were calculated.
Serum hormone analysis
Serum progesterone and estradiol concentrations were measured using commercial RIA kits DSL-4800 and DSL-3400, respectively (Diagnostic Systems Laboratories, Webster, TX) according to the instructions of the manufacturer, at the National Association of Testing Authorities (Adelaide, Australia) accredited Reproductive Medicine Laboratories (Sydney, Australia). Assays were performed in duplicate, and all samples were measured in a single assay. The lower limits of detection were 8 pM estradiol and 0.3 nM progesterone. The within-assay coefficients of variation were 10.0% in the estradiol assay and 7.9% in the progesterone assay. All data points were above the assay thresholds. The precision of the estradiol assay at low serum estradiol content was confirmed in experiments using serum from mice after ovariectomy and exogenous steroid hormone replacement.
Serum FSH and LH were determined by RIA using reagents kindly provided by Dr. Parlow at National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (Bethesda, MD). Iodinated preparations and antisera used in the FSH and LH RIAs were rFSH-I-9 and anti-rFSH-S-11, and rLH-I-9 and anti-rLH-S-10, respectively. Results are expressed in terms of NIDDK mFSH-RP (AFP-5308D) and NIDDK mLH-RP (AFP-5306A). The iodination preparations were iodinated using Iodogen reagent (Pierce, Rockford, IL). Goat antirabbit IgG (GAR12; Monash Institute of Reproduction and Development, Monash University, Melbourne, Australia) was used as second antibody for the FSH and LH RIAs. Assays were performed in duplicate on 20-μl serum samples, and all samples were measured in a single assay. The lower limits of detection were 0.78 ng/ml FSH and 0.10 ng/ml LH. The within-assay coefficients of variation were 8.1% in the FSH assay and 7.3% in the LH assay. Data points below the assay threshold were assigned the threshold values for data analysis.
Ovarian steroidogenic enzyme mRNA analysis
Quantitative RT-PCR was used to analyze the quantity of steroidogenic acute regulatory protein (StAR), P450 side chain cleavage (P450scc), 3-hydroxysteroid dehydrogenase-1 (HSD31), and P450 17 -hydroxylase/C17–20-lyase (P450c17) mRNA in ovarian tissue from superovulated 27- to 29-d-old female mice killed 14 h after hCG injection. Primers to detect expression of StAR, HSD31, P450scc, and P450c17 mRNAs were designed using PrimerExpress 2.0 software from respective cDNA sequences (24, 25, 26) (GenBank accession no. AF195119). The intron/exon boundaries for StAR were identified using the complete murine sequence (GenBank accession no. AY032730), for P450scc by comparison with the rat sequence (27), and for HSD31 and P450c17 by comparison with the human sequences from Bain et al. (25) and Kagimoto et al. (28), respectively. The primers used were as follows: StAR, forward, 5'CCGGAGCAGAGTGGTGTCAT and reverse, 5'-TGCGATAGGACCTGGTTGATG; HSD31, forward, 5'-GGACAAAGTATTCCGACCAGAAAC and reverse, 5'CAGGCACTGGGCATCCA; P450scc, forward, 5'ACAGACGCATCAAGCAGCAA and reverse, 5'-CACTGCTGATGGACTCAAAGGA; P450c17, forward, 5'-TGGCTTTCCTGGTGCACAA and reverse, 5'-GTGTTCGACTGAAGCCTACATACTG and 18S (Universal 18S primers; Ambion Austin, TX). Ovaries were snap frozen in liquid N2 and stored at –70 C before RNA extraction. Total cellular RNA was extracted according to a modification of the method described by Chomczynski and Sacchi (29). After treatment with RNase-free DNase I (500 IU/ml; 60 min at 37 C) (Roche Molecular Biochemicals, Mannheim, Germany), first-strand cDNA was reverse transcribed from 1 μg RNA using a Superscript II RNase H Reverse Transcriptase kit (90 min at 43 C) (Invitrogen, Carlsbad, CA). The PCR amplification used reagents supplied in a 2x SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), and each reaction volume (20 μl total) contained of 0.5–1 μM 5' and 3' primers and 3 μl cDNA. The negative control included in each reaction consisted of H2O substituted for cDNA. PCR amplification was performed in triplicate in a ABI Prism 5700 Sequence Detection System (Applied Biosystems) according to the instructions of the manufacturer. The absence of alternative PCR products and primer dimers in the PCR product was shown by a single dissociation peak and confirmed by visualization of the PCR product after agarose gel electrophoresis. Serial dilutions of cDNA were analyzed to confirm a linear relationship between cDNA content and quantity of product across the amplification range. The efficiency of each PCR was determined by regression analysis of the fluorescence of the sample at each cycle number. Efficiencies of all PCRs were between 80 and 100%. Data were normalized for 18S mRNA expression using the 2-Ct method (30) and expressed as a ratio of the mean value for TGF1+/± control ovaries.
Statistical analysis
Data were analyzed with SPSS 11.0 for Windows software (SPSS, Chicago, IL), using independent samples t test or one-way ANOVA with post hoc Tukey’s test when appropriate. When data were not normally distributed, nonparametric Mann-Whitney U tests were used. Significance was inferred at P < 0.05. Data are mean ± SEM unless specified otherwise.
Results
Effect of TGF1 null mutation on female fertility
To determine the effect of TGF1 null mutation on the capacity of female mice to initiate and maintain pregnancy, TGF1+/± and TGF1–/– adult females were housed with normal stud B10 males and examined each day for the presence of vaginal plugs and subsequent birth of litters. Only 14 of 34 (41%) TGF1–/– females mated during a 28 d period compared with 35 of 35 (100%) TGF1+/± females. Of the TGF1 null females that did mate, the interval between caging with males and detection of the plug was (mean ± SEM) 12.3 ± 2.5 d compared with 5.2 ± 1.0 d for TGF1+/± females (P = 0.03). When TGF1 null mutant females did mate with B10 stud males, the incidence of establishing successful pregnancy was substantially reduced, with only 3 of 14 mated TGF1–/– mice producing live progeny at term compared with 100% viable pregnancy outcome in mated TGF1+/± females. The litter size in TGF1–/– females was 3.7 ± 1.8 pups (n = 3 litters of 2, 3, and 6 pups per litter) compared with 6.3 ± 1.0 pups (n = 35 litters) in TGF1+/± females (not significant, P > 0.05). The TGF1–/– females that did not give birth to pups did not show visible signs of pregnancy. There was no difference in mating interval or litter size between TGF1+/– and TGF1+/+ females (data not shown), so data from the two genotypes were pooled in this and subsequent experiments.
Effect of TGF1 null mutation on estrous cycling
The failure to mate, and extended interval between placement with a male and mating in TGF1 null mice, suggested disrupted estrous cycling activity. To investigate the effect of TGF1 null mutation on estrous cyclicity, the stage of estrous cycle was tracked by analysis of vaginal cytology for a period of 28 d from 6 wk of age. Cycles were severely perturbed in null mutant mice, with deficiency in TGF1 reducing the frequency of estrus events by 42% (P = 0.001) (Fig. 1, A and B) and increasing mean estrous cycle length by 70% (P = 0.01) (Fig. 1, A and B). Consistent with fewer estrus events was a significant reduction in the percentage of time spent in the proestrus stage of the cycle in TGF1–/– females (Fig. 1C). However, there was no difference in the proportion of time spent in other stages of the estrus cycle, because TGF1–/– females often demonstrated a prolonged estrus event, as indicated by 100% cornified vaginal epithelial cells present for 2 d or more (Fig. 1A).
Effect of TGF1 null mutation on onset of sexual maturity, follicle development, and ovarian hormone synthesis
To further investigate ovarian function, serum and ovarian tissue were recovered for analysis from TGF1+/± and TGF1–/– females during the first proestrus event after onset of sexual maturity. Sexual maturity as measured by age at appearance of vaginal opening and age at first proestrus was delayed in TGF1 null mutants. In TGF1+/± females, 21 of 21 mice achieved vaginal opening by 5 wk of age (30.2 ± 0.5 d after birth; n = 21) compared with TGF1–/– females, in which only 18 of 22 (82%) achieved vaginal opening by 8 wk of age (35.4 ± 1.6 d; n = 18; P = 0.005). TGF1 null mutants were older at the time of first proestrus after vaginal opening [30.8 ± 1.0 d in TGF1+/± females (n = 18) vs. 37.0 ± 1.8 d in TGF1–/– females (n = 12); P = 0.004]. This delay in puberty was associated with a 23% reduction in body weight at 4 wk of age in TGF1–/– females [10.7 ± 0.3 g in TGF1–/– females (n = 8) vs. 13.9 ± 0.5 g in TGF1+/± females (n = 10)].
There was no difference in the absolute weight or weight relative to body size of ovaries from TGF1+/± and TGF1–/– females collected at first proestrus [2.2 ± 0.2 mg in TGF1+/± females (n = 12) vs. 1.7 ± 0.2 mg in TGF1–/– females (n = 9)]. Analysis of histological sections of ovaries showed that follicles in the full range of expected stages of development (Fig. 2, A and C) and mature antral follicles with normal morphology were present regardless of genotype (Fig. 2, B and D). Quantitative analysis of follicle numbers showed that there was no effect of TGF1 status on the relative abundance of primordial/primary follicles (types 1–3), preantral follicles with multiple layers of granulosa cells (types 4, 5), or small antral (type 6) or large antral (types 7, 8) follicles (Fig. 2E). Furthermore, there was no effect on the proportion of atretic follicles (data not shown).
Gonadotropins and estradiol were measured in serum recovered during the periovulatory period. Serum LH content was reduced in TGF1–/– mice to approximately 35% of the mean value detected in TGF1+/± mice (P = 0.008) (Fig. 3A). There was no effect of genotype on serum FSH (Fig. 3B) or serum estradiol (Fig. 3C).
Effect of TGF1 null mutation on ovulation and ovarian steroid synthesis
To investigate the number of oocytes ovulated and their fate after mating in early pregnancy, TGF1+/± and TGF1–/– females were mated with B10 males and killed on d 3.5 pc. The number of oocytes ovulated was measured by counting the total number of fertilized or unfertilized oocytes flushed from the oviduct and uterus and the number of corpora lutea in whole intact ovaries. A total of 42% fewer embryos were flushed from TGF1–/– mice than from TGF1+/± controls (P = 0.0002) (Table 1). Consistent with this, there was a 40% reduction in the number of corpora lutea seen in mice deficient in TGF1 (P = 0.005) (Fig. 4E). Fewer corpora lutea were also clearly evident when the histology of ovarian tissue was examined in sections stained with hematoxylin and eosin (Fig. 4, A and C), although the size and structure of corpora lutea in tissues from TGF1–/– and TGF1+/± females were comparable (Fig. 4, B and D). The reduction in number of corpora lutea was associated with a 50% reduction in absolute ovarian weight in TGF1–/– females compared with TGF1+/± littermates (P < 0.001) (Fig. 4F), corresponding to a 35% reduction when expressed relative to body weight (P < 0.001) (Fig. 4G).
RIA of serum steroid hormone content at d 3.5 pc revealed an 80% reduction in mean progesterone concentration in TGF1–/– females when compared with TGF1+/± controls (P = 0.002) (Fig. 5A). Progesterone output when expressed per corpus luteum was similarly reduced (3.0 ± 1.5 vs. 12.9 ± 1.5 nM per corpus luteum for TGF1–/– and TGF1+/± females, respectively; P = 0.001). Serum estradiol concentration was not affected by genotype (Fig. 5B).
Responsiveness of TGF1 null mutant mice to gonadotropin replacement
To determine whether higher rates of ovulation could be elicited in TGF1 null mice, additional groups of prepubertal TGF1–/– and TGF1+/± females were treated with exogenous gonadotropins. Seven of 10 (70%) immature TGF1–/– female mice were responsive to ovarian hyperstimulation, ovulating 7.1 ± 1.5 oocytes, compared with 10 of 10 (100%) TGF1+/± immature females that ovulated 9.1 ± 1.2 oocytes (not significant, P > 0.05).
To investigate the effect of TGF1 null mutation on ovarian steroidogenesis, ovaries excised from gonadotropin-stimulated prepubertal females approximately 2 h after ovulation were analyzed, and steroidogenic enzyme mRNA expression was measured using quantitative RT-PCR. There was a 50% decrease in expression of mRNA encoding HSD31 and a 4-fold increase in P450c17 mRNA in TGF1–/– compared with TGF1+/± mice (Fig. 6). StAR and P450scc was not significantly altered in TGF1–/– and TGF1+/± ovaries.
Effect of TGF1 null mutation on preimplantation embryo development
The stage of preimplantation development achieved by fertilized oocytes was examined in embryos flushed from the reproductive tract of TGF1–/– and TGF1+/± females killed on d 3.5 pc. A comparable percentage of oocytes from TGF1+/± and TGF1–/– females were fertilized and had initiated cell division (Table 1). However, few embryos from TGF1–/– females were developed to the expected blastocyst stage, with the majority arrested in the morula stage of development (Table 1 and Fig. 7).
To investigate the cause of developmental failure in preimplantation embryos, a series of experiments were devised in an attempt to rescue development of fertilized oocytes from TGF1–/– mice. Administration of exogenous progesterone to TGF1–/– females on d 1.5 and 2.5 of pregnancy did not improve embryo development, as assessed on d 3.5 pc (Table 2). As was seen in untreated TGF1–/– females, a high rate of oocyte fertilization occurred, but most embryos were arrested at morula stage or earlier.
In another approach, designed to discriminate between deficiency in maternal tract vs. absence of embryonic TGF1 in causing embryo demise, embryos were fertilized and cultured ex vivo. The oocytes recovered from prepubertal mice after treatment with gonadotropins were fertilized by in vitro fertilization, or flushed from the reproductive tract after mating, and then cultured in vitro to investigate embryo development in a defined culture media proven to support early embryo development. In vitro fertilization with sperm from wild-type CBA F1 males resulted in comparable rates of cleavage of oocytes from TGF1+/± and TGF1–/– females. However, TGF1+/– embryos from TGF1–/– oocytes failed to develop to blastocyst stage and most frequently did not survive past two-cell stage (Table 2). Similarly, when zygotes were flushed from the oviduct on d 0.5 pc after mating superovulated TGF1–/– and TGF1+/± mice with wild-type B10 males and then cultured in vitro for 4 d, embryos from TGF1–/– oocytes again failed to develop to blastocyst stage (Table 2). In contrast, oocytes from TGF1+/± mice fertilized either in vivo or in vitro developed to blastocyst at the expected high rate.
Effect of TGF1 null mutation on uterine morphology
To examine the effect of TGF1 null mutation on uterine morphology, sections of fixed, paraffin-embedded uteri from d 3.5 pc naturally mated females were stained with hematoxylin and eosin (Fig. 8). There was no overt effect of genotype on the tissue mass or diameter of uterine tissue (data not shown). The uteri from TGF1–/– females were morphologically indistinguishable from uteri dissected from TGF1+/± females in terms of luminal diameter, presence and relative thickness of endometrial and myometrial tissues, and the abundance and size of uterine glands.
Discussion
This study indicates a critical role for TGF1 in the reproductive function of female mice. We have demonstrated three distinct lesions in female fertility caused by TGF1 null mutation. First, ovarian function is perturbed with irregular progression of the estrous cycle associated with a reduction in circulating LH during the characteristic proestrus surge, resulting in less frequent ovulatory events and fewer oocytes ovulated. Second, corpus luteum function as measured by serum progesterone content is compromised. Third, preimplantation embryo development is perturbed in TGF1 null mutant females. Together, these disruptions dramatically reduce the incidence of successful pregnancy and production of live-born progeny to females mice deficient in TGF1.
The major determinant of reduced fecundity in null mutant mice was that ovulation occurred approximately 40% less frequently than in wild-type controls, and, when it did occur, approximately 40% fewer oocytes were ovulated. Synthesis of the gonadotropin LH was severely reduced at the expected time of the LH surge in TGF1 null mutant females. In contrast, administration of exogenous gonadotropins to immature TGF1–/– mice successfully induced ovulation, with a response equal to that seen in wild-type mice in terms of the proportion of mice that ovulated and the number of oocytes released. The effects of TGF1 deficiency on estrous cycling and ovulation can thus largely be attributed to disrupted LH activity, suggesting a requirement for TGF1 in normal hypothalamic or pituitary function. Mice with a mutation in the LH subunit are totally infertile and anovulatory (31), showing the absolute requirement for this hormone in estrous cycling and ovulation. LH was detectable at low levels in most of the TGF1–/– mice, indicating that synthesis must be impaired or dysregulated rather than completely absent. This is consistent with the wide variance in ovarian cycle disturbances manifesting both at a population level and within individual mice. Some null mutant mice cycled more successfully than others, which in the worst cases, showed only one estrus event over the 28-d period of study. More than half of the TGF1 null females failed to mate even once during the 28-d period of caging with males, which might reflect a more severe LH deficiency in these animals.
Although growth trajectory after birth was delayed in TGF1 null mutant mice and their onset of puberty occurred on average 6 d later than in wild-type mice, this was not causally associated with reduced LH at first proestrus, because there was no correlation between body weight and serum LH content in individual mice and, furthermore, the mean body weight of mice at proestrus was similar regardless of genotype (data not shown). Therefore, although TGF1 appears to be required for normal growth rate after birth and this likely explains the delay in puberty in TGF1 null mutant mice, growth impairment is unlikely to account for disrupted LH activity at the time of onset of ovarian cycling.
The observation that approximately 10% of null mutant mice were able to ovulate, mate, and maintain pregnancy successfully to deliver overtly normal pups at term shows that, although TGF1 is clearly essential for normal fecundity, other factors or circumstances must attenuate the severity of effects caused by absence of this cytokine. Because the proportion of null mutant mice able to establish pregnancy after mating was comparable with the proportion in which embryos reached blastocyst stage, it is reasonable to infer that postimplantation loss is not a major contributing factor in reduced fecundity in TGF1-deficient female mice. This suggests that implantation and placental morphogenesis can proceed relatively unimpaired in the absence of maternal TGF1 but does not exclude the possibility that a more rigorous analysis might reveal deficits in the quality of placental structure and function or in fetal development and growth trajectory. A gene dosage effect and rescue by transfer of maternal TGF1 across the placenta is evident when TGF1-deficient fetuses gestate in mothers heterozygous for the TGF1 null mutation (17, 21, 32), but the current data show that maternal cytokine is not an essential requirement for fetal development, at least in the event of heterozygousity for the TGF1 mutation in embryos.
The variability in fertility phenotype seen in the current experiments is likely to reflect differences in the extent to which individual mice carry genes conferring compensation for TGF1 deficiency or other interactions with TGF1 status. It has been noted previously that genetic background has a significant influence on penetrance of embryonic lethality due to defects in pre- and postimplantation development in TGF1–/– embryos (21, 33). On the mixed genetic background CF1/129/C3H used in the current study, breeding pairs heterozygous for the TGF1 null mutation were reasonably fertile, but, in concurrence with previous reports (21, 34), we observed less than the expected number of pups genotyped as homozygous for the null mutation, reflecting embryonic lethality in approximately 50% of homozygous offspring and severely constraining availability of experimental mice. Our attempts to backcross the TGF1 null genotype onto the C57BL/6 background resulted in progressively declining fertility, as experienced previously by others, and can be attributed to a small number of genetic modifiers (21, 33). The number of heterozygous offspring was also approximately 50% less than would be expected, suggesting a Tgf1 gene dosage effect in the developing fetal-placenta unit. Previous studies have reported a selective loss of both TGF1–/– and TGF1+/– conceptuses in TGF1+/– intercrosses (34).
Impaired ovarian function persisted after ovulation, with diminished progesterone production by corpora lutea. Interestingly, estrogen synthesis was unaffected, both during the periovulatory period and in early pregnancy. This suggests that follicular estrogen synthesis is less sensitive to TGF1 deficiency than luteal cell progesterone synthesis and is consistent with the dominant function of LH in initiating luteinization compared with the greater dependence of follicle development on FSH, the circulating level of which was unchanged in the absence of TGF1. Reduced serum progesterone content was seen even when the smaller number of corpora lutea was taken into account and was associated with alterations in expression of mRNAs encoding components of the steroidogenic pathway. HSD31 and P450c17 mRNA expression in ovaries from immature gonadotropin-stimulated TGF1 null mice were decreased and increased, respectively, when compared with synthesis in ovaries from TGF1-replete littermates. HSD31 catalyzes the conversion of pregnenolone to progesterone, and P450c17 catalyzes the conversion of progesterone to androstenedione. The decreased and increased relative expression of these enzymes in TGF1 null mutants is consistent with the observation of reduced progesterone production with normal estradiol production in naturally mated TGF1 null females. Therefore, TGF1 deficiency appears to cause impaired communication in the paracrine or autocrine signals that regulate progesterone synthesis in the corpus luteum.
Although the steroidogenic defect can in part be attributed to reduced LH, altered expression of steroidogenic enzymes was evident after gonadotropin replacement, suggesting that LH-independent mechanisms may also contribute. Inhibitory effects of TGF1 deficiency on progesterone synthesis or secretion might be mediated through direct regulatory effects of this cytokine on luteal cells, because luteal cells express TGF receptors and are responsive to this cytokine in vitro (7, 8). Alternatively, ovarian macrophages potentially act as a mediating cell by virtue of their high responsiveness to TGF1 (35) and specific roles in development and function of the corpus luteum (36, 37, 38). Macrophage-secreted cytokines TNF- and IL-1 are implicated in regulating luteal steroidogenesis (39), and disrupted macrophage regulation leading to overproduction of these antisteroidogenic cytokines in the absence of TGF1 may contribute to the reduced serum progesterone seen in early pregnancy. The ovarian phenotype we describe in TGF1 null mutant mice bears remarkable similarity to that seen in macrophage-deficient Csf1op/Csf1op mice, which also have disrupted ovarian cycles, reduced ovulation rates, and impaired steroidogenesis associated with reduced LH (40). In Csf1op/Csf1op mice, ovarian effects appear to be largely secondary to altered hypothalamic activity caused by absence of macrophages in that tissue (41), so it would be of interest to examine whether hypothalamic macrophage function is similarly altered in TGF1 null mutants.
Despite infrequent estrus and reduction in oocytes ovulated, approximately 40% of TGF1 null females were able to mate. However, in 80% of mice in which mating occurred, pregnancy failed due to impaired preimplantation embryo development. Developmental arrest appeared not to be due to an adverse maternal tract environment secondary to TGF1 or progesterone deficiency but rather was intrinsic to the embryo, because strategies including progesterone replacement or ex vivo fertilization and embryo culture failed to rescue embryonic demise. Furthermore, histological evaluation of the uterus revealed no differences in the structure of the endometrium and glands. This finding does not preclude a role for maternal TGF1 in the reproductive tract but does show that TGF1 deficiency in the oviduct and uterus is an insufficient explanation for the embryo lethality observed.
It is possible that TGF1+/– early embryos had reduced developmental competence by virtue of their depleted capacity to synthesize endogenous TGF1 after embryonic genome activation. Embryos normally synthesize TGF1 (10, 13) and express TGF receptors (10, 11), supporting an autocrine role for TGF1 in preimplantation development (42). A premorula lethality in null embryos has been reported in TGF1 heterozygous breeder pairs and contributes to the less than expected proportion of TGF1–/– offspring observed (21). However, it is unlikely that reduced embryonic TGF1 synthesis is responsible for the developmental failure in embryos from TGF1–/– females mated with TGF1+/+ males described here. In vitro fertilization using sperm from TGF1–/– males and TGF1+/+ oocytes shows that the cleaved embryos (all TGF1+/–) develop to blastocysts at a similar rate to those oocytes fertilized by TGF1+/+ sperm (Ingman, W. V., and S. A. Robertson, manuscript in preparation). Therefore, when the null mutation is paternally inherited, TGF1+/– embryos develop to the blastocyst stage normally. The possibility that maternally imprinted expression of the Tgf1 gene in the developing embryo is required can be disregarded because whether the null gene was inherited maternally vs. paternally was found not to influence the expected frequency of the null gene in offspring (21).
The most compelling explanation for embryonic arrest in null mutant mice is oocyte incompetence, caused by developmental defects occurring during oocyte growth and development in the TGF1 null ovary, such that those follicles that do progress to ovulation yield oocytes less capable of supporting development after fertilization. It is well known that, as oocytes grow and develop in the ovarian follicle, they gradually and sequentially acquire the cellular and molecular machinery required to support early embryogenesis, and that the quality of the follicular milieu directly affects the developmental potential of the ensuing embryo (43, 44). Perturbation can have adverse consequences for structural organization of the embryo, activation of the embryonic genome, and subsequent gene expression patterns several cell cycles into embryo development, often manifesting in arrest during the preimplantation phase (44). Oocytes from TGF1 null mutants appear to be fully meiotically competent (capable of reaching metaphase II), as evidenced by comparable fertilization frequencies with control mice. Impaired cytoplasmic competence of TGF1 null mutant oocytes is more likely, potentially due to absence of oocyte TGF1 protein or mRNA, because TGF1 is known to be a maternally stored transcript (12). Alternatively, oocyte incompetence might reflect a more generalized perturbation of oocyte development in a suboptimal, TGF-deficient follicular environment.
In summary, we conclude that TGF1 deficiency severely impacts female fertility through disrupted ovulation, oocyte development, and corpus luteum function. Absence of maternal TGF1 causes impaired LH activity and also appears to impact the maturing oocyte in such a way as to reduce developmental competence after ovulation and fertilization. The relevance of this work to human follicle development is highlighted by comparable expression patterns for TGF1 and its signaling molecules in the human ovary (45, 46) and the observation that levels of TGF1 in follicular fluid correlate with pregnancy success after in vitro fertilization (47).
Acknowledgments
We thank Dr. Anne O’Connor at Monash Institute for Reproduction and Development (Melbourne, Australia) for performing LH and FSH RIAs, Mr. Alan Gilmore at Reproductive Medicine Laboratories (Adelaide, Australia) for performing ovarian steroid hormone RIAs, Dr. Darryl Russell for assistance with ovarian histology, and Dr. Robert Gilchrist for helpful advice and comments.
Footnotes
This work was supported by the Australian Research Council Discovery Grant DP0211497, a National Health and Medical Research Council of Australia Research Fellowship (to S.A.R.), and a C. J. Martin Career Development Award (to W.V.I.).
W.V.I., R.L.R., and K.W. have nothing to declare. S.A.R. consults for GroPep Ltd. (Adelaide, Australia).
First Published Online November 3, 2005
Abbreviations: hCG, Human chorionic gonadotropin; HSD31, 3-hydroxysteroid dehydrogenase-1; P450c17, P450 17 -hydroxylase/C17–20-lyase; P450scc, P450 side chain cleavage; pc, postcoitus; scid, severe combined immunodeficiency spontaneous mutation; StAR, steroidogenic acute regulatory protein.
Accepted for publication October 21, 2005.
References
Shull MM, Doetschman T 1994 Transforming growth factor-1 in reproduction and development. Mol Reprod Dev 39:239–246
Godkin JD, Dore JJ 1998 Transforming growth factor and the endometrium. Rev Reprod 3:1–6
Ingman WV, Robertson SA 2002 Defining the actions of transforming growth factor in reproduction. BioEssays 24:904–914
Adashi EY, Resnick CE, Hernandez ER, May JV, Purchio AF, Twardzik DR 1989 Ovarian transforming growth factor- (TGF): cellular site(s), and mechanism(s) of action. Mol Cell Endocrinol 61:247–256
Roy SK 1993 Epidermal growth factor and transforming growth factor- modulation of follicle-stimulating hormone-induced deoxyyribonucleic acid synthesis in hamster preantral and early antral follicles. Biol Reprod 48:552–557
Vanderhyden BC, Macdonald EA, Nagyova E, Dhawan A 2003 Evaluation of members of the TGF superfamily as candidates for the oocyte factors that control mouse cumulus expansion and steroidogenesis. Reprod Suppl 61:55–70
Matsuyama S, Shiota K, Takahashi M 1990 Possible role of transforming growth factor- as a mediator of luteotropic action of prolactin in rat luteal cell cultures. Endocrinology 127:1561–1567
Matsubara H, Ikuta K, Ozaki Y, Suzuki Y, Suzuki N, Sato T, Suzumori K 2000 Gonadotropins and cytokines affect luteal function through control of apoptosis in human luteinized granulosa cells. J Clin Endocrinol Metab 85:1620–1626
Roy SK, Ogren C, Roy C, Lu B 1992 Cell-type-specific localization of transforming growth factor-2 and transforming growth factor-1 in the hamster ovary: differential regulation by follicle-stimulating hormone and luteinizing hormone. Biol Reprod 46:595–606
Paria BC, Jones KL, Flanders KC, Dey SK 1992 Localization and binding of transforming growth factor- isoforms in mouse preimplantation embryos and in delayed and activated blastocysts. Dev Biol 151:91–104
Chow JF, Lee KF, Chan ST, Yeung WS 2001 Quantification of transforming growth factor 1 (TGF1) mRNA expression in mouse preimplantation embryos and determination of TGF receptor (type I and type II) expression in mouse embryos and reproductive tract. Mol Hum Reprod 7:1047–1056
Gilchrist RB, Morrissey MP, Ritter LJ, Armstrong DT 2003 Comparison of oocyte factors and transforming growth factor- in the regulation of DNA synthesis in bovine granulosa cells. Mol Cell Endocrinol 201:87–95
Rappolee DA, Brenner CA, Schultz R, Mark D, Werb Z 1988 Developmental expression of PDGF, TGF-, and TGF- genes in preimplantation mouse embryos. Science 241:1823–1825
Morrish DW, Dakour J, Li H 1998 Functional regulation of human trophoblast differentiation. J Reprod Immunol 39:179–195
Shull MM, Ormsby I, Kier AB, Pawlowski S, Diebold RJ, Yin M, Allen R, Sidman C, Proetzel G, Calvin D, Annunziata N, Doetschman, T 1992 Targeted disruption of the mouse transforming growth factor-1 gene results in multifocal inflammatory disease. Nature 359:693–699
Kulkarni AB, Huh CG, Becker D, Geiser A, Lyght M, Flanders KC, Roberts AB, Sporn MB, Ward JM, Karlsson S 1993 Transforming growth factor 1 null mutation in mice causes excessive inflammatory response and early death. Proc Natl Acad Sci USA 90:770–774
Letterio JJ, Geiser AG, Kulkarni AB, Roche NS, Sporn MB, Roberts AB 1994 Maternal rescue of transforming growth factor-1 null mice. Science 264:1936–1938
Diebold RJ, Eis MJ, Yin M, Ormsby I, Boivin GP, Darrow BJ, Saffitz JE, Doetschman T 1995 Early-onset multifocal inflammation in the transforming growth factor 1-null mouse is lymphocyte mediated. Proc Natl Acad Sci USA 92:12215–12219
Kobayashi S, Yoshida K, Ward JM, Letterio JJ, Longenecker G, Yaswen L, Mittleman B, Mozes E, Roberts AB, Karlsson S, Kulkarni AB 1999 2-microglobulin-deficient background ameliorates lethal phenotype of the TGF-1 null mouse. J Immunol 163:4013–4019
Letterio JJ, Geiser AG, Kulkarni AB, Dang H, Kong L, Nakabayashi T, Mackall CL, Gress RE, Roberts AB 1996 Autoimmunity associated with TGF-1-deficiency in mice is dependent on MHC class II antigen expression. J Clin Invest 98:2109–2119
Kallapur S, Ormsby I, Doetschman T 1999 Strain dependency of TGF1 function during embryogenesis. Mol Reprod Dev 52:341–349
Allen E 1921 The oestrous cycle in the mouse. Am J Anat 30:297–348
Pedersen T, Peters H 1968 Proposal for a classification of oocytes and follicles in the mouse ovary. J Reprod Fertil 17:555–557
Clark BJ, Wells J, King SR, Stocco DM 1994 The purification, cloning, and expression of a novel luteinizing hormone-induced mitochondrial protein in MA-10 mouse Leydig tumor cells. Characterization of the steroidogenic acute regulatory protein (StAR). J Biol Chem 269:28314–28322
Bain PA, Yoo M, Clarke T, Hammond SH, Payne AH 1991 Multiple forms of mouse 3-hydroxysteroid dehydrogenase/5-4 isomerase and differential expression in gonads, adrenal glands, liver, and kidneys of both sexes. Proc Natl Acad Sci USA 88:8870–8874
Youngblood GL, Sartorius C, Taylor BA, Payne AH 1991 Isolation, characterization, and chromosomal mapping of mouse P450 17-hydroxylase/C17–20 lyase. Genomics 10:270–275
Oonk RB, Parker KL, Gibson JL, Richards JS 1990 Rat cholesterol side-chain cleavage cytochrome P-450 (P-450scc) gene. Structure and regulation by cAMP in vitro. J Biol Chem 265:22392–22401
Kagimoto M, Winter JS, Kagimoto K, Simpson ER, Waterman MR 1988 Structural characterization of normal and mutant human steroid 17-hydroxylase genes: molecular basis of one example of combined 17-hydroxylase/17,20 lyase deficiency. Mol Endocrinol 2:564–570
Chomczynski P, Sacchi N 1987 Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156–159
Livak KJ, Schmittgen TD 2001 Analysis of relative gene expression data using real-time quantitative PCR and the 2(-C(T)) method. Methods 25:402–408
Ma X, Dong Y, Matzuk MM, Kumar TR 2004 Targeted disruption of luteinizing hormone -subunit leads to hypogonadism, defects in gonadal steroidogenesis, and infertility. Proc Natl Acad Sci USA 101:17294–17299
McLennan IS, Koishi K 2004 Fetal and maternal transforming growth factor-1 may combine to maintain pregnancy in mice. Biol Reprod 70:1614–1618
Bonyadi M, Rusholme SA, Cousins FM, Su HC, Biron CA, Farrall M, Akhurst RJ 1997 Mapping of a major genetic modifier of embryonic lethality in TGF1 knockout mice. Nat Genet 15:207–211
Dickson MC, Martin JS, Cousins FM, Kulkarni AB, Karlsson S, Akhurst RJ 1995 Defective haematopoiesis and vasculogenesis in transforming growth factor-1 knock out mice. Development 121:1845–1854
Bogdan C, Nathan C 1993 Modulation of macrophage function by transforming growth factor , interleukin-4, and interleukin-10. Ann NY Acad Sci 685:713–739
Fukumatsu Y, Katabuchi H, Naito M, Takeya M, Takahashi K, Okamura H 1992 Effect of macrophages on proliferation of granulosa cells in the ovary in rats. J Reprod Fertil 96:241–249
Van der Hoek KH, Maddocks S, Woodhouse CM, van Rooijen N, Robertson SA, Norman RJ 2000 Intrabursal injection of clodronate liposomes causes macrophage depletion and inhibits ovulation in the mouse ovary. Biol Reprod 62:1059–1066
Matsuyama S, Takahashi M 1995 Immunoreactive (ir)-transforming growth factor (TGF)- in rat corpus luteum: ir-TGF is expressed by luteal macrophages. Endocr J 42:203–217
Pate JL, Landis Keyes P 2001 Immune cells in the corpus luteum: friends or foes Reproduction 122:665–676
Cohen PE, Zhu L, Pollard JW 1997 Absence of colony stimulating factor-1 in osteopetrotic (csfmop/csfmop) mice disrupts estrous cycles and ovulation. Biol Reprod 56:110–118
Cohen PE, Zhu L, Nishimura K, Pollard JW 2002 Colony-stimulating factor 1 regulation of neuroendocrine pathways that control gonadal function in mice. Endocrinology 143:1413–1422
Paria BC, Dey SK 1990 Preimplantation embryo development in vitro: cooperative interactions among embryos and role of growth factors. Proc Natl Acad Sci USA 87:4756–4760
Rodriguez KF, Farin CE 2004 Gene transcription and regulation of oocyte maturation. Reprod Fertil Dev 16:55–67
Duranthon V, Renard JP 2001 The developmental competence of mammalian oocytes: a convenient but biologically fuzzy concept. Theriogenology 55:1277–1289
Chegini N, Flanders KC 1992 Presence of transforming growth factor- and their selective cellular localization in human ovarian tissue of various reproductive stages. Endocrinology 130:1707–1715
Osterlund C, Fried G 2000 TGF receptor types I and II and the substrate proteins Smad 2 and 3 are present in human oocytes. Mol Hum Reprod 6:498–503
Fried G, Wramsby H 1998 Increase in transforming growth factor 1 in ovarian follicular fluid following ovarian stimulation and in-vitro fertilization correlates to pregnancy. Hum Reprod 13:656–659(Wendy V. Ingman, Rebecca L. Robker, Kare)