Lack of Maternal Glutamate Cysteine Ligase Modifier Subunit (Gclm) Decreases Oocyte Glutathione Concentrations and Disrupts Preimplantation Development in Mice

Abstract

Glutathione (GSH) is the most abundant intracellular thiol and an important regulator of cellular redox status. Mice that lack the modifier subunit of glutamate cysteine ligase (Gclm), the rate-limiting enzyme in GSH synthesis, have decreased GSH synthesis. Nicotinamide nucleotide transhydrogenase, an inner mitochondrial membrane protein, catalyzes the interconversion of reduced nicotinamide adenine dinucleotide and reduced nicotinamide adenine dinucleotide phosphate; reduced nicotinamide adenine dinucleotide phosphate is required for reduction of GSH disulfide. Previous work supports roles for GSH in preimplantation development. We hypothesized that Gclm−/− mice have increased preimplantation embryonic mortality and that this effect is enhanced by absence of a functioning Nnt gene. Gclm−/− females produced significantly fewer pups per litter than Gclm+/+ littermates. Numbers of oocytes ovulated in a natural estrous cycle or upon superovulation did not differ by genotype. Fewer uterine implantation sites were observed in the Gclm−/− females. Prepubertal Gclm−/− and Gclm+/+ females were superovulated, then mated overnight with a Gclm+/+ male. At 0.5 d postcoitum, Gclm−/− females had significantly lower percentages of zygotes with two pronuclei and higher percentages of zygotes with one pronucleus than Gclm+/+ or Gclm+/− females. At 3.5 d postcoitum, a significantly lower percentage of blastocyst stage embryos was recovered from uteri of Gclm−/− females than Gclm+/+ females. Embryonic development to the blastocyst stage, but not the two-cell stage, was significantly decreased after in vitro fertilization of oocytes from Gclm−/− females compared with Gclm+/+ females. The Nnt mutation did not enhance the effects of Gclm genotype on female fertility. These results demonstrate critical roles for maternal GSH in supporting normal preimplantation development.

Reactive oxygen species (ROS) result from the sequential addition of electrons to molecular oxygen, forming superoxide anion radical, hydrogen peroxide, and hydroxyl radical. Superoxide anion radical can react with nitric oxide, forming the reactive nitrogen species peroxynitrite. ROS/reactive nitrogen species can function as intracellular signaling molecules, but because some of them are highly reactive, they can damage important cellular components by reacting with lipids, proteins, and nucleic acids. ROS are generated as byproducts of normal cellular processes, such as oxidative phosphorylation in all tissues and steroidogenesis in the gonads and adrenal glands (1). To prevent the adverse effects of high levels of ROS, the ovary and oviduct possess antioxidant defenses, including vitamins A, C, and E, the antioxidant tripeptide glutathione (GSH), and ROS scavenging enzymes, such as superoxide dismutases, GSH peroxidases, catalase, and GSH-S-transferases (2–8).

GSH is the most abundant intracellular nonprotein thiol and one of the most important intracellular antioxidants. It is present in cells at millimolar concentrations. GSH has numerous intracellular functions, including reduction of hydrogen peroxide and lipid peroxides as a cofactor for GSH peroxidases, detoxification of electrophilic toxicants as a cofactor for GSH-S-transferases, regulation of protein function, and regulation of nucleotide metabolism (9). In the process of reducing peroxides, GSH is oxidized to the disulfide, GSH disulfide (GSSG). GSSG is reduced back to GSH via the action of the enzyme GSH reductase, which requires reduced nicotinamide adenine dinucleotide phosphate (NADPH). C57BL/6J mice have been reported to carry a mutation in the nicotinamide nucleotide transhydrogenase (Nnt) gene (10, 11), which results in the complete absence of mature NNT protein. NNT is an inner mitochondrial membrane protein that catalyzes the interconversion of reduced nicotinamide adenine dinucleotide and NADPH. Under conditions of oxidative stress, the generation of NADPH by NNT is favored. In the nematode Caenorhabdis elegans, Nnt deletion mutants had significantly greater inhibition of growth in response to oxidative stress than nematodes with wild-type Nnt (12).

GSH is synthesized in two ATP-dependent reactions. The first, rate-limiting reaction is catalyzed by glutamate cysteine ligase (GCL), a heterodimer composed of a catalytic subunit [GCL catalytic subunit (GCLC)] and a modifier subunit [GCL modifier subunit (GCLM)] (13, 14). GCLC possesses all the catalytic activity, whereas binding of GCLM decreases the Michaelis constant for the enzyme substrates glutamate and ATP and increases the inhibitory constant for GSH (13, 14).

Ovulated oocytes possess among the highest GSH concentrations of any cell type, on the order of 10 mm (15, 16). Previous studies suggest that oocyte GSH is important for sperm nucleus decondensation and male pronucleus formation after fertilization (16, 17) and for formation of the meiotic spindle during meiosis II (18). The preimplantation embryo has limited ability to up-regulate GSH synthesis before the blastocyst stage and relies on GSH from the oocyte (19–21).

Mice that lack Gclc die during embryonic development (9, 22, 23). Mice that lack Gclm survive and reproduce but have greatly reduced tissue levels of GSH (9, 24–26). Although Gclm−/− mice are able to reproduce, detailed studies of the effects of knocking out Gclm on female fertility have not been conducted. We hypothesized that Gclm−/− females have low GSH concentrations in oocytes, resulting in abnormal development of preimplantation embryos and decreased fertility and that the effect of Gclm genotype is enhanced by the absence of a functioning Nnt gene.

Materials and Methods

Materials

All chemicals and reagents were purchased from Fisher Scientific (Pittsburgh, PA) or Sigma-Aldrich (St. Louis, MO) unless otherwise noted.

Animals

Gclm null mice were generated by disrupting the Gclm gene by replacing exon 1 with a β-galactosidase/neomycin phospho-transferase fusion minigene (26). The mice have been backcrossed eight times onto a C57BL/6J genetic background (B6.129-Gclm^tm1Tjka; hereafter referred to as Gclm−/−). Mice for these experiments were generated by mating Gclm+/− males with Gclm+/− females. Genotyping was carried out using DNA extracted from tail or toe snips as described (24, 26). Our production of over 650 offspring from heterozygous crosses has revealed a slight deviation from the expected Mendelian distribution, with 20% Gclm−/−, 53% Gclm+/−, and 27% Gclm+/+ offspring produced. We have not observed any sex differences in the genotype distributions. The mice were housed in American Association for the Accreditation of Laboratory Animal Care-accredited facilities, with free access to deionized water and irradiated, soy-free laboratory chow (Teklad 2919; Harlan Laboratories, Inc., Indianapolis, IN), on a 14-h light, 10-h dark cycle, with lights on at 0630 h. Temperature was maintained at 20.6–23.9 C. The experimental protocols were carried out in accordance with the Guide for the Care and Use of Laboratory Animals (27) and were approved by the Institutional Animal Care and Use Committee at the University of California Irvine.

Female fertility assessment

Beginning at 2 months of age, Gclm+/+ or Gclm−/− female mice were mated to wild-type, 8-wk-old C57BL/6J male mice (The Jackson Laboratory, Bar Harbor, ME) for a continuous breeding assay (28). Females were checked daily for pups. On the day of birth, live and dead pups were counted and sexed, live pups were weighed, and pups were then humanely euthanized by decapitation after anesthetizing on ice.

Estrous cycling and ovarian histology

Estrous cycling in individually housed, 7-month-old female mice after completion of the fertility study was evaluated every morning for a minimum of 14 d by vaginal cytology (29, 30). Mice were euthanized by CO₂ inhalation on estrus, and ovaries were collected for enumeration of ovulated oocytes and corpora lutea by evaluation of hematoxylin and eosin-stained complete serial sections of the entire ovary by an observer blind to genotype.

Timed breeding studies

For assessment of implantation and of postimplantation mortality, 2-month-old Gclm−/− or Gclm+/+ females were mated with wild-type male proven breeders on the late afternoon of proestrus, as determined by vaginal cytology. The next morning, females were checked for vaginal plugs. The day of vaginal plug was designated gestational day (GD) 1. Females were killed by CO₂ euthanasia on the morning of GD 18. Live and dead fetuses were counted, live fetuses were weighed, and obvious resorption sites were noted. All uteri were then stained for 4.5 h in 10% ammonium sulfide, and a final count of resorption sites was made (31). Females were weighed before mating and on the day of euthanasia, and gestational weight gain was calculated.

Superovulation and collection of preimplantation embryos

Twenty-seven-day-old Gclm−/−, Gclm+/−, and Gclm+/+ females were injected with 5 IU equine chorionic gonadotropin (CG) ip at 1400 h, then injected with 5 IU human CG (hCG) ip 46 h later at 1200 h. Ovulated oocytes were collected from the ampullae of oviducts at 22–24 h after hCG injection.

For collection of embryos, Gclm−/−, Gclm+/−, and Gclm+/+ mice were mated with a wild-type male in the late afternoon after hCG injection. The next morning, females were checked for vaginal plugs. The morning of vaginal plug detection was designated 0.5 d postcoitum (dpc). Zygotes were collected from the ampullae of oviducts at 0.5 dpc, and blastocysts were collected by flushing the uterine horns at 3.5 dpc.

Oocytes and zygotes were collected into flushing and holding medium (FHM) (Specialty Media, Phillipsburg, NJ) containing hyaluronidase (0.3 mg/ml) to release cumulus cells for a maximum of 5 min, then transferred to FHM. Blastocysts were collected into FHM. All embryos were washed by transfer through three drops of FHM under mineral oil. The numbers of fertilized and nonfertilized oocytes and embryos were counted at ×400 using a Zeiss Axiovert 200M microscope (Zeiss, Oberkochen, Germany) equipped with differential interference contrast optics. At the 0.5-dpc time point, oocytes were evaluated for the presence of sperm and/or one or two pronuclei; fragmentation was also noted. For the 3.5-dpc time point, conceptuses were scored for stage of development (unfertilized oocyte, blastocyst, morula, two- to eight-cell embryo, fragmented or degenerating embryo). All evaluations were performed by an observer blinded to genotype.

GSH assays

After treatment with hyaluronidase and initial wash in FHM, oocytes were washed through four drops of collection media [Hanks' balanced salt solution without calcium or magnesium (Invitrogen, Carlsbad, CA), with 5 mg/ml HEPES, and 1 mg/ml BSA] under mineral oil. Oocytes were collected in a small volume of collection medium into tared microcentrifuge tubes. One quarter volume (based on weight of the sample) of 5% sulfosalicylic acid was added, the oocytes were disrupted by homogenization with a handheld mortar and pestle Kontes dounce homogenizer (Kimble-Chase, Vineland, NJ) on ice, incubated on ice for 15 min, centrifuged at 15800 × g at 4 C, and the supernatant was stored frozen at −80 C. Total and oxidized GSH were measured in pooled supernatants from 150 to 231 oocytes per sample (oocytes from two to five mice) using a modification of an enzymatic recycling assay developed by Griffith (6, 32, 33). All samples were assayed in triplicate. Standards of known concentrations of GSSG were prepared in four parts buffer with 1 part 5% sulfosalicylic acid. Total GSH was measured by the addition of reaction mixture containing 0.26 mm NADPH, 0.74 mm 5,5′-dithiobis (2′-dinitrobenzoic acid), and 0.62 mm GSH reductase in sodium phosphate EDTA buffer. The reduction of 5,5′-dithiobis (2′-dinitrobenzoic acid) to thiobis(2-nitrobenzoic acid) was monitored every 10 sec using a microplate spectrophotometer (VersaMax; Molecular Devices, Sunnyvale, CA). GSSG assay was carried using the same procedure, except that reduced GSH was first removed by conjugation with 2-vinylpyridine, followed by chloroform extraction.

Modulation of the reproductive phenotype of Gclm−/− females by mutation in the Nnt gene

C57BL/6J mice have been reported to carry a mutation in the Nnt gene (10, 11), which results in the complete absence of mature NNT protein. NNT is one of several enzymes that can provide NADPH, which is required for the regeneration of GSH by GSSG reductase. Therefore, the decreased fertility and preimplantation embryonic death observed in Gclm−/− females on a C57BL/6J background could be partially due to the presence of the Nnt mutation. To test this, we generated two new lines of mice, which both carry the knockout Gclm allele, one in the presence of the mutant Nnt and one in the presence of wild-type Nnt. These lines were generated by crossing a Gclm−/− male with a C57BL/6JEiJ female mouse. The latter subline separated from the parental C57BL/6J in 1976. It has been maintained as a separately inbreeding subline ever since, and it is known not to carry the Nnt mutation (http://jaxmice.jax.org/strain/000924.html). The Gclm+/−;NntWT/MUT and Gclm+/−;NntWT/MUT offspring of this first cross were mated to generate the Gclm+/−;NntMUT and Gclm+/−;NntWT founders of the two new sublines, which we designate B6.Cg-Gclm^tm1Tjka Nnt^C57BL/6J/Lud (abbreviated Gclm/NntMUT) and B6.Cg-Gclm^tm1Tjka Nnt⁺/Lud (abbreviated Gclm/NntWT).

Gclm−/−/NntWT, Gclm−/−/NntMUT, Gclm+/+/NntWT, and Gclm+/+/NntMUT female mice (n = 6–8 per group) were individually bred with wild-type C57BL/6J males for 16 wk beginning at 3 months of age. Females were checked daily for litters. Male and female offspring were counted, weighed, and euthanized on the day of birth.

Effect of Gclm genotype on in vitro fertilization and embryonic development

Gclm−/−/NntWT, Gclm−/−/NntMUT, Gclm+/+/NntWT, and Gclm+/+/NntMUT female mice were superovulated as described above, and oocytes were collected from the ampullae of the oviducts at 13 h after hCG injection. Oocytes from each mouse were placed in K-RVFE-50 in vitro fertilization medium (Cook Medical, Inc., Bloomington, IN) and were mixed with fresh epididymal sperm from a Gclm+/+ C57BL/6J male. After incubation at 37 C in 5% CO₂, 5% O₂, and 90% N₂ for 4–6 h, the oocytes were washed through four drops of K-RVFE-50 media and incubated for another 20 h. The zygotes were then transferred to potassium simplex optimized medium + ½ amino acids media (Millipore, Bedford, MA) and cultured for four more days. Embryos were scored daily for stage of development by observers blind to genotype.

Statistical analyses

For continuous variables, Levene's test was used to test for homogeneity of variances. The effects of Gclm genotype on various endpoints were analyzed by independent samples t test for equal or unequal variances or by ANOVA, as appropriate. Quantities expressed as fractions (fraction of embryos at a particular developmental stage) were subjected to arcsine square root transformation before analysis (34). Alternatively, if variances were not homogeneous, nonparametric Kruskal-Wallis or Mann-Whitney tests were used. Data are presented as mean ± sem in figures and tables. Statistical analyses were performed using SPSS 17.0 for Macintosh (SPSS, Inc., Chicago, IL).

Results

Decreased oocyte GSH concentrations in Gclm−/− females

Concentrations of total GSH were measured in pools of ovulated oocytes collected from superovulated Gclm−/−, Gclm+/−, and Gclm+/+ mice (150–231 oocytes/pool; n = 2–3 pools/genotype). Total GSH concentrations in oocytes ovulated from Gclm−/− mice were below the limit of detection of the assay, indicating that Gclm−/− oocytes had GSH concentrations that were less than 20% those of oocytes ovulated from Gclm+/+ (1.03 ± 0.11 pmol/oocyte) and Gclm+/− (1.04 ± 0.25 pmol/oocyte) mice. GSH concentrations in oocytes from Gclm+/− mice did not differ significantly from those of Gclm+/+ mice. GSSG was undetectable in all samples.

Gclm−/− females have decreased fertility

To assess the effects of low GSH concentrations on fertility, we conducted a continuous breeding study, in which Gclm−/− and Gclm+/+ females were bred with wild-type males for 20 wk from 2 to 7 months of age. Cumulative litter and pup production of Gclm−/− and Gclm+/+ females are shown in Fig. 1. Gclm−/− and Gclm+/+ females produced equal numbers of litters in 20 wk (5.3 ± 0.3 vs. 5.4 ± 0.4, respectively). However, the Gclm−/− females produced significantly fewer offspring in 20 wk (19.2 ± 2.6 vs. 45.4 ± 4.1; t = −5.63, df = 12, P < 0.001 by t test).

Fig. 1.

Gclm−/− female mice have decreased fertility compared with Gclm+/+ mice. Gclm−/− and Gclm+/+ mice were mated with wild-type males for 20 wk beginning at 8 wk of age. A, Cumulative mean ± sem number of litters produced in 20 wk. B, Cumulative mean ± sem number of pups produced in 20 wk. The effect of genotype on total pup production was statistically significant (P < 0.001 by t test). n = 5–9 per genotype.

Smaller litter size of Gclm−/− females is not due to lower ovulation rate

Ovulated oocytes in the oviduct, fresh corpora lutea, and older corpora lutea were enumerated in serial sections of both ovaries collected on estrous morning. As expected, the numbers of ovulated oocytes and stage 1 (freshly ovulated) (35) corpora lutea were highly correlated, with a correlation coefficient of 0.97, and did not differ by Gclm genotype (t = −0.30, df = 12, P = 0.77; t = −1.09, df = 12, P = 0.30) (Table 1). The number of stage 2 corpora lutea, which is indicative of numbers of ovulations in previous estrous cycles, also did not differ by genotype (t = −1.26, df = 12, P = 0.23) (Table 1). The numbers of oocytes ovulated after a superovulation protocol in prepubertal Gclm+/+, Gclm+/−, and Gclm−/− mice also did not differ by genotype (F = 0.49, df = 34, P = 0.62) (Table 1).

Table 1.

Effect of lack of maternal Gclm on ovulation

	Gclm+/+ (mean±sem)	Gclm−/− (mean±sem)
Oocytes in oviducts on estrusa	8.8 ± 0.7	8.3 ± 1.1
Class 1 corpora lutea on estrusa	9.4 ± 0.5	8.0 ± 0.9
Class 2 corpora lutea on estrusa	9.8 ± 1.7	12.0 ± 0.9

	Gclm+/+	Gclm+/−	Gclm−/−
Oocytes in oviduct after superovulationb	44.6 ± 8.0	45.0 ± 5.6	52.5 ± 5.9

^aGclm+/+, n = 5; Gclm−/−, n = 9.

^bGclm+/+, n = 10; Gclm+/−, n = 13; Gclm−/−, n = 12.

Smaller litter size in Gclm−/− females is not due to postimplantation embryo death

The number of live fetuses per litter on GD 18 was statistically significantly lower in timed-mated Gclm−/− females than in Gclm+/+ females (t = −3.69, df = 18, P = 0.002) (Table 2). The mean maternal gestational weight gain was 7.7 ± 1.3 g in the Gclm−/− females compared with 14.1 ± 1.2 g in the Gclm+/+ females. The difference in weight gain was not statistically significant when adjusted for number of fetuses (F= 0.48, df = 20, P = 0.497). The numbers of dead fetuses and resorption sites did not differ by genotype (t = 1.5, df = 9, P = 0.17 and t = −1.38, df = 15, P = 0.19, respectively). The total number of implants (live plus dead fetuses plus resorptions) was significantly lower in Gclm−/− females than in Gclm+/+ females (t = −3.35, df = 15, P = 0.004). The mean fetal weight per litter did not differ significantly by genotype (data not shown).

Table 2.

Effect of lack of maternal Gclm on postimplantation survival

	Gclm+/+ (mean±sem)	Gclm−/− (mean±sem)
Live fetuses	6.8 ± 0.9	2.4 ± 0.8a
Dead fetuses	0	0.2
Resorption sites	0.9 ± 0.4	1.7 ± 0.4
Total implantations	7.9 ± 0.8	4.3 ± 0.7a

n = 8–10 per group.

^aP < 0.005 by t test.

Smaller litter size in Gclm−/− females is due to embryonic death before the blastocyst stage

Because Gclm−/− females had similar ovulation rates as Gclm+/+ females, but fewer implantations, we reasoned that either oocytes of Gclm−/− females had lower fertilization rates and/or that conceptuses of Gclm−/− females died before implantation. To test this, we superovulated immature females of both genotypes, mated them with wild-type males overnight, and collected oocytes and conceptuses at 0.5 and 3.5 dpc.

Figure 2A shows the effect of Gclm genotype on zygote development at 0.5 dpc. The effect of genotype was statistically significant for the percentage of zygotes with one pronucleus (χ² = 11.1, df = 2. P = 0.004 by Kruskal-Wallis test), the percentage of zygotes with two pronuclei (χ² = 11.2, df = 2, P = 0.004), and the percentage of oocytes with sperm but no pronuclei (χ² = 6.4, df = 2, P = 0.041). Intergroup comparisons showed that a significantly smaller percentage of zygotes of Gclm−/− dams than of Gclm+/+ or Gclm+/− dams had two pronuclei (Z = −2.74, P = 0.006 and Z = −2.89, P = 0.004, respectively, by Mann-Whitney test), and a significantly larger percentage of zygotes of Gclm−/− dams had only one pronucleus (Z = −2.74, P = 0.006 and Z = −2.89, P = 0.004, respectively). The percentage of eggs with a sperm that had penetrated the zona pellucida, but no pronuclei, was higher in Gclm−/− females than in Gclm+/− females (Z = −2.56, P = 0.010 by Mann-Whitney test). In most oocytes with one pronucleus in Gclm−/− females, the pronucleus was located near the second polar body, and there was an intact sperm visible in the egg. The latter observations are consistent with lack of sperm nucleus decondensation resulting in lack of male pronucleus formation. Intergroup comparisons showed no statistically significant differences between Gclm+/− and Gclm+/+ females.

Fig. 2.

Gclm−/− female mice have decreased zygote development and decreased percentages of embryos that reach the blastocyst stage. Superovulated, prepubertal mice of the indicated genotypes were mated with wild-type males, and conceptuses were collected from the oviducts at 0.5 dpc or from the uterus at 3.5 dpc as described in Materials and Methods. A, Mean ± sem fraction of embryos at 0.5 dpc at the indicated stages of development (n = 5–6 per genotype). The effect of genotype was statistically significant by Kruskal-Wallis test for fraction of embryos with one pronucleus (P = 0.004), fraction with two pronuclei (P = 0.004), and oocytes with sperm in the cytoplasm (P = 0.041). B, Mean ± sem fraction of embryos at 3.5 dpc at the indicated stages of development (n = 6–8 per genotype). *, P < 0.05 vs. Gclm+/+ by Mann-Whitney test; †, P < 0.05 vs. Gclm+/− by Mann-Whitney test.

Figure 2B shows the percentages of embryos that attained various developmental stages at 3.5 dpc by maternal genotype. Compared with Gclm+/+ dams, a significantly smaller percentage of embryos of Gclm−/− dams reached the blastocyst stage (Z = −2.74, P = 0.006 by Mann-Whitney test), and larger percentages of embryos of Gclm−/− dams tended to be unfertilized (Z = −1.75, P = 0.081) or to be at the three-cell (Z = −1.94, P = 0.052) and eight-cell (Z = −1.94, P = 0.052) stages at 3.5 dpc. The percentage of degenerating embryos was nonsignificantly increased in Gclm−/− dams. Maternal genotype did not affect the total number of conceptuses recovered from the uteri at 3.5 dpc (19.9 ± 3.3 in Gclm−/− vs. 14.8 ± 1.7 in Gclm+/+).

Nnt mutation does not worsen fertility deficit of Gclm−/− females

The presence of a mutant Nnt gene, which results in absence of mature NNT protein (10, 11), significantly modulated the cumulative number of litters (F = 5.72, df = 28, P = 0.025, effect of Nnt genotype) but did not significantly modulate the cumulative number of offspring (F = 0.993, df = 28, P = 0.329, effect of Nnt genotype) produced by Gclm−/− female mice during a 16-wk breeding study from 3 to 7 months of age (Fig. 3, A and B). The percentage of pups born dead during the breeding study also appeared to be modulated by Nnt genotype (χ² = 14.97, df = 3, P = 0.002, effect of group by Kruskal-Wallis test) (Fig. 3C). Contrary to our hypothesis that the Nnt mutation would exacerbate the effects of lack of Gclm, the Nnt mutation appeared to partially rescue the effects of lack of Gclm on cumulative number of litters produced and percent of pups born dead. Interestingly, the effect of the lack of Gclm on female fertility appeared to be more severe on the mixed C57BL/6J x C57BL/6JEiJ background than on the pure C57BL/6J background in that Gclm−/− females of the mixed background not only produced cumulatively fewer offspring (F = 176.5, df =28, P < 0.001, effect of Gclm genotype) but also produced fewer litters (F = 11.9, df = 28, P = 0.002) than Gclm+/+ littermates, regardless of Nnt genotype (compare Fig. 3, A and B, with Fig. 1, A and B).

Fig. 3.

The presence of a mutation in the Nnt gene modulates the effects of deletion of Gclm on female fertility. Gclm−/−/NntWT, Gclm+/+/NntWT, Gclm−/−/NntMUT, and Gclm+/+/NntMUT females were mated with wild-type males from 3 months of age for 16 wk (n = 6–8 per group). A, Mean ± sem cumulative number of litters produced during the breeding study by genotype. B, Mean ± sem cumulative number of offspring produced during the breeding study by genotype. The effect of Gclm genotype was statistically significant for cumulative litters and offspring (P = 0.002 and P < 0.001, respectively, by two-way ANOVA). The effect of Nnt genotype was statistically significant for cumulative number of litters (P = 0.025) but not for cumulative number of offspring (P = 0.329). The interaction between the genotypes was not statistically significant in either case (P = 0.074 and P = 0.414, respectively). C, Mean ± sem percent of pups born dead to mothers of the indicated genotypes. The effect of genotype was significant overall (P = 0.002 by Kruskal-Wallis test). Bars with different letters differed significantly from one another by Mann-Whitney test, P < 0.04.

Effects of female Gclm knockout and Nnt mutation on in vitro fertilization and embryo development in culture

Oocytes of Gclm−/−/NntWT, Gclm−/−/NntMUT, Gclm+/+/NntWT, and Gclm+/+/NntMUT females fertilized in vitro with sperm from wild-type males progressed to the two-cell stage by 30 h after gamete mixing at similar rates (Table 3). However, by 78 h after gamete mixing, significantly lower percentages of embryos derived from oocytes of Gclm−/− females had progressed to the six-cell through morula stages (F = 9.28, df = 14, P = 0.011, effect of Gclm genotype) (Table 3). By 102 h after gamete mixing, significantly lower percentages of embryos derived from oocytes of Gclm−/− females had progressed at least to the six-cell stage (F = 14.93, df = 14, P = 0.003, effect of Gclm genotype) or had reached the blastocyst stage (F = 9.28, df = 14, P = 0.011, effect of Gclm genotype) (Table 3). Nnt genotype did not significantly affect embryo development in culture (F < 1.5) (Table 3).

Table 3.

Effects of female Gclm knockout and Nnt mutation on in vitro fertilization and embryo development in culture

Genotype of oocyte donor	% two to three cell at 30 h (±sem)	% six-cell to morula at 78 h (±sem)a	% six-cell to blastocyst at 102 h (±sem)b	% blastocysts at 102 h (±sem)a	N
Gclm+/+;NntMUT	55.7 ± 14.6	18.9 ± 6.1	29.1 ± 7.5	15.7 ± 7.6	4
Gclm−/−;NntMUT	34.1 ± 19.4	3.3 ± 3.3	4.2 ± 3.0	0	3
Gclm+/+;NntWT	63.9 ± 10.9	27.8 ± 13.4	34.0 ± 12.4	22.0 ± 18.4	3
Gclm−/−;NntWT	58.1 ± 10.9	7.1 ± 2.7	9.0 ± 3.7	0.4 ± 0.4	4

^aP = 0.011, effect of Gclm genotype by two-way ANOVA on arcsine square root transformed data.

^bP = 0.003, effect of Gclm genotype by two-way ANOVA on arcsine square root transformed data.

Discussion

This study definitively demonstrates a critical role for maternal GSH in preimplantation embryonic development. Specifically, lack of maternal Gclm resulted in dramatically decreased GSH concentrations in ovulated oocytes and in decreased rates of formation of the second pronucleus at 0.5 dpc, decreased development of embryos to the blastocyst stage at 3.5 dpc, decreased uterine implantation rates, and decreased litter sizes. Interestingly, oocytes from Gclm−/− dams that were fertilized in vitro developed to the two-cell stage by 30 h after gamete mixing at similar rates as oocytes from Gclm+/+ dams but failed to progress to the blastocyst stage. Taken together, these results suggest that lack of Gclm and resulting GSH deficiency exerts adverse effects at several stages of early development.

Several genetically modified mouse models of decreased antioxidant capacity have been developed, but few of these have been systematically evaluated for effects on reproductive function. Mice that lack the transcription factor Nrf2 have decreased antioxidant capacity due to decreased basal and inducible transcription of numerous antioxidant defense genes, including the genes of GSH synthesis (36, 37). Nrf2−/− female mice have been reported to have significantly smaller litters than wild-type C57BL/6 mice at 6–20 wk of age, but the cause of smaller litter sizes was not investigated (38). Female mice null for Sod1 are also subfertile, but the cause remains unclear. One group reported that ovaries of Sod1−/− mice had few large antral follicles and corpora lutea, suggesting a defect in the later stages of follicle maturation (39). Another group reported normal ovulation and conception, but increased embryonic lethality in Sod1−/− females (40). Although Sod2−/− mice die within several weeks of birth, transplantation of Sod2−/− ovaries to wild-type hosts resulted in normal follicular development, ovulation, and viable offspring, suggesting that Sod2 is not required for normal ovarian function (39). Normal fertility has been reported in mice that lack Gpx1 (41) or catalase (42) or that bear an inactivating mutation in GSH reductase (43–45), whereas disruption of Gpx4 (46–48) or Gclc (22, 23) results in embryonic lethality. In contrast, γ-glutamyl transpeptidase 1 (Ggt1) null mice display a shortened lifespan, stunted growth, and a severe female reproductive phenotype, with complete infertility, lack of ovarian large antral follicles and corpora lutea, and absent ovulatory response to exogenous gonadotropins (49–51). These mice have decreased ovarian cysteine concentrations, but normal ovarian GSH concentrations, compared with wild-type controls, and the female reproductive phenotype is completely rescued by cysteine replacement (49). Detailed reproductive studies of another Gclm null strain that was generated using a different targeting vector and on a different genetic background have not been published (25), but the females of that strain are reportedly also subfertile (Chen, Y., personal communication). Thus, the limited data to date do not suggest that knockout of different antioxidant genes results in similar female reproductive phenotypes but do suggest that GSH and cysteine are important for female fertility.

We did not observe a significant difference in oocyte GSH concentrations between Gclm+/− and Gclm+/+ females in the present study. Intermediate concentrations of GSH in Gclm+/− mice, which are higher than the levels in Gclm−/− and lower than levels in Gclm+/+, have been reported for some, but not all, tissues and cells tested (25, 26, 52). The slight differences in preimplantation embryo development between Gclm+/− and Gclm+/+ females observed in the present study (Fig. 2A) may be caused by differences in GSH concentrations in other cell types besides the oocytes.

Brief depletion of GSH associated with an oxidant stimulus (diamide or iodoacetamide) in ovulated hamster oocytes before in vitro fertilization disrupted normal hamster oocyte meiotic spindle formation leading to the formation of abnormal female pronuclei (18). However, no spindle abnormalities were reported when bovine oocytes were depleted of GSH using buthionine sulfoximine (BSO), a specific inhibitor of GCL (17). These results suggest that meiotic spindle formation may be sensitive to oxidative stress but not to GSH depletion per se. In our Gclm−/− model, oocyte GSH concentrations are chronically low, but we observed apparently normal formation of the female pronucleus. Biochemical depletion of GSH with BSO during ooctye maturation in vitro (16, 17) or in vivo (15) prevented sperm head decondensation and formation of the male pronucleus after in vitro fertilization. This was believed to be due to the requirement for reduction of protamine disulfide bonds in the sperm nucleus for sperm nuclear reactivation to occur (16, 53). Our finding that very few zygotes of Gclm−/− dams had two pronuclei at 0.5 dpc provide further evidence that high concentrations of GSH in the oocyte are required for normal formation of the male pronucleus. Our in vitro fertilization data suggest that the low GSH concentrations in oocytes of Gclm−/− females may be sufficient to permit eventual decondensation of the sperm nucleus, with similar rates of cleavage stage embryos observed 30 h after gamete mixing for both Gclm−/− and Gclm+/+ females. However, the embryos derived from oocytes of Gclm−/− females failed to develop to the blastocyst stage at 102 h after gamete mixing. The latter results are consistent with a continued requirement for GSH from the oocyte for development beyond the cleavage stage to the blastocyst stage.

Before the blastocyst stage, preimplantation embryos do not express Gclc and are not able to up-regulate GSH synthesis in response to treatment with agents that deplete GSH by conjugation, such as diethyl maleate (19, 20). In contrast, two-cell embryos are able to respond to stimuli that induce oxidative stress by up-regulating Gclc expression and GSH synthesis (19). In vivo treatment with BSO beginning before initiation of a superovulation protocol significantly diminished concentrations of GSH in oocytes and increased the percentage of degenerating embryos retrieved at 1.5 dpc (21). These results are consistent with our observation of decreased in vitro development of embryos of Gclm−/− females to the blastocyst stage compared with embryos of Gclm+/+ females despite similar rates of development to the cleavage stage. In contrast, BSO injections beginning at 0.5 dpc significantly decreased GSH concentrations in oviductal and uterine secretions, but did not decrease zygote GSH concentrations or affect embryonic development at 1.5 or 2.5 dpc (21). Taken together with the results of the present study, these findings support the conclusion that GSH present in the oocyte is critical for normal preimplantation development, whereas GSH in reproductive tract secretions appears to play a lesser, if any, role. Consistent with a role for oocyte GSH in early preimplantation development, zygotes from mice overexpressing GSH synthetase showed better in vitro development than zygotes from wild-type mice (54).

Genetic background is known to modulate the effects of gene deletions. For example, the effects of Dax knockout on gonadal differentiation are much more pronounced on a C57BL/6J background than on a 129/SvlmJ background (55). The effects of lack of Sod2 also vary with background strain, with embryonic lethality occurring on a C57BL/6J background compared with survival to postnatal d 8 on a DBA/2J background (11). Huang et al. (11) showed that this differential sensitivity to Sod2 deletion was associated with a mutation in the Nnt gene in the C57BL/6J strain. NNT is a nuclear-encoded, inner mitochondrial membrane protein that catalyzes the interconversion of reduced nicotinamide adenine dinucleotide and NADPH, with the generation of NADPH predominating. NADPH is required for the reduction of GSSG by GSSG reductase. The Nnt mutation has also been reported to account for the glucose intolerance observed in C57BL/6J mice (10). In the present study, we found that the presence of a mutant Nnt gene modulated the effect of Gclm deletion on female fertility but not in the hypothesized direction. The females that were homozygous null for Gclm and harbored the Nnt mutation (Gclm−/−;NntMUT) produced significantly more litters and lower percentages of pups that were born dead and nonsignificantly more live offspring than Gclm−/−;NntWT females. This suggests a beneficial effect of the Nnt mutation in the setting of low GSH synthesis, which is difficult to explain. One possibility is that the presence of the Nnt mutation leads to up-regulation of other antioxidant genes, partially compensating for the deficiency of Gclm. An alternative explanation is that C57BL/6JEiJ mice may differ from C57BL/6J mice in loci besides Nnt that compensate for the compromised GSH status in Gclm−/− mice. These possibilities should be investigated in future studies.

The results of the present study may be relevant to female reproductive aging. Declines in oocyte quality, such as aneuploidy, decreased mitochondrial fraction, increased dilated smooth endoplasmic reticulum and Golgi fractions, and increased mitochondrial membrane rupture (56–58), are observed with aging in experimental animals and women. Diminished oocyte quality, in turn, is thought to play a role in the decline in litter size (59) and increased preimplantation and postimplantation embryo loss (60, 61) observed in aging mice and the decreased success rates of assisted reproduction in older women (57, 62). Expression of Gclm (56) and concentrations of GSH (63) were reportedly decreased in oocytes from aged mice compared with young adult mice. Our results suggest that the age-related declines in oocyte quality may be caused in part by the decreased capacity of aged oocytes to synthesize GSH. In addition, polymorphisms in Gclm and Gclc (64–67), which affect GSH synthesis, exist in humans, and our results predict that they may affect oocyte quality and fertility in women.

In conclusion, maternal GSH is essential for normal zygote development and normal fertility in female mice. Gclm−/− females have low oocyte GSH concentrations and decreased male pronucleus formation after fertilization, resulting in fewer implantations and smaller litter sizes. This decreased oocyte quality as a result of GSH deficiency may be relevant to female reproductive aging and to fertility in women with polymorphisms in Gclc and Gclm.