Autophagy Is Essential for Preimplantation Development of Mouse Embryos

See allHide authors and affiliations

Science  04 Jul 2008:
Vol. 321, Issue 5885, pp. 117-120
DOI: 10.1126/science.1154822


After fertilization, maternal proteins in oocytes are degraded and new proteins encoded by the zygotic genome are synthesized. We found that autophagy, a process for the degradation of cytoplasmic constituents in the lysosome, plays a critical role during this period. Autophagy was triggered by fertilization and up-regulated in early mouse embryos. Autophagy-defective oocytes derived from oocyte-specific Atg5 (autophagy-related 5) knockout mice failed to develop beyond the four- and eight-cell stages if they were fertilized by Atg5-null sperm, but could develop if they were fertilized by wild-type sperm. Protein synthesis rates were reduced in the autophagy-null embryos. Thus, autophagic degradation within early embryos is essential for preimplantation development in mammals.

During the transition from oocyte to embryo, maternal proteins and RNAs are rapidly degraded and replaced by zygotic proteins and RNAs. Degradation of maternal RNAs is thought to be mediated by binding of regulatory proteins to the 3′ untranslated regions of target RNAs or by microRNAs (1, 2). In contrast, the mechanisms by which maternal proteins are degraded remain poorly understood. In mammals, protein degradation accelerates shortly after fertilization and is apparent by the early two-cell stage (3). Early embryogenesis may rely on the maternal protein stores as nutrients. Several maternal proteins are degraded by the ubiquitin-proteasome system (4, 5), but it is unknown whether macroautophagy (referred to as autophagy hereafter), another major degradation system, plays an important role during this period. During autophagy, a portion of cytoplasm is sequestered into an autophagosome; this then fuses with the lysosome, and the cytoplasm-derived materials are degraded (610). Autophagy is important for various physiological processes such as starvation adaptation and intracellular quality control.

To investigate whether autophagy occurs in fertilized oocytes, we used autophagy-indicator mice, in which green fluorescent protein (GFP)–fused LC3, a mammalian Atg8 homolog present on autophagosomes, is systemically expressed (11, 12). We collected oocytes and embryos from superovulated GFP-LC3 female mice after mating with wild-type males (13). Although metaphase II (MII) oocytes showed almost no GFP-LC3 dots, a number of dots that represented autophagosomes appeared in fertilized embryos at the one- to four-cell stage (Fig. 1, A and B, and fig. S1A). Electron microscopy of two-cell embryos confirmed the presence of autophagic vacuoles (Fig. 1C and fig. S2). Induction of autophagy was also confirmed by LC3 conversion (11), which was increased in two-cell embryos relative to MII oocytes (Fig. 1D). Additionally, phosphorylation of S6 kinase was reduced after fertilization, which suggested that mTOR, a negative regulator of autophagy, was inactivated (fig. S3). Thus, formation of autophagosomes is up-regulated immediately after fertilization.

Fig. 1.

Induction of autophagy after fertilization. (A) Autophagosome formation in preimplantation embryos. MII oocytes and embryos were obtained from superovulated GFP-LC3 transgenic females. Times (in hours) after human chorionic gonadotropin (hCG) treatment are indicated. Scale bar, 20 μm. (B) Quantification of GFP-LC3 dots in oocytes and embryos. Numbers of GFP-LC3 dots per embryo image are shown. Each value represents the mean ± SE of at least 10 oocytes or embryos from the indicated stages. (C) Conventional and immuno–electron microscopy analysis of embryos at the two-cell stage. An autophagosome and autolysosome are indicated by an arrow and closed arrowhead, respectively. Localization of endogenous LC3 is indicated by open arrowheads. Black squares indicate the enlarged areas shown in insets. Scale bar, 500 nm. (D) Conversion of LC3-I (cytosolic) to LC3-II (autophagosome-bound) in fertilized embryos. Whole lysates from 300 MII oocytes and two-cell embryos were loaded and analyzed by immunoblotting. Asterisk indicates an LC3-I degradation product. (E) Autophagy induction in embryos 6 hours after IVF or parthenogenetically activated with SrCl2. Scale bar, 20 μm.

Early embryos contained a large number of lysosomes stained with LysoTracker (Molecular Probes) (fig. S1B) (14). Some of the GFP-LC3 dots colocalized with LysoTracker, suggesting that autophagosomes had indeed fused with lysosomes. Total expression levels of GFP-LC3 rapidly decreased after the four-cell stage (fig. S1C). Because LC3 in the inner autophagosomal membrane is degraded upon fusion with lysosomes (fig. S1D) (15, 16), the rapid disappearance of GFP-LC3 protein suggests that autophagic flux is up-regulated in fertilized embryos.

We next determined whether autophagy is induced simply as a result of starvation after ovulation. MII oocytes were collected from GFP-LC3 mice and inseminated in vitro with wild-type sperm. The number of GFP-LC3 dots remained very small until 3 hours, but numerous dots appeared by 4 hours after in vitro fertilization (IVF) and increased in number thereafter (Fig. 1E and movie S1). Autophagy was not induced in unfertilized oocytes cultured under the same conditions. We also observed that unfertilized oocytes in the oviduct showed only very small numbers of GFP-LC3 dots, which suggests that the occurrence of autophagy depends on fertilization rather than ovulation (fig. S1F). After induction, autophagy was transiently suppressed between the late one-cell stage and the middle of the two-cell stage, and was then reactivated (fig. S1E). Transient suppression of autophagy has also been observed during mitosis in various mammalian cell lines (17).

We then sought to determine whether fertilization itself or subsequent events are sufficient to induce autophagy. Oocytes isolated from GFP-LC3 mice were parthenogenetically activated with strontium and cultured in vitro. In these oocytes, GFP-LC3 dot generation was observed with a time course similar to that observed for in vitro fertilized embryos (Fig. 1E). Because strontium induces repetitive intracellular calcium release in a manner similar to that seen after normal fertilization (18), calcium oscillation may be an inducer of autophagy.

Atg5 is an essential factor for autophagosome formation (19). Atg5–/– mice appear grossly normal at birth but die within 1 day of delivery, which implies that autophagy may not be important during embryogenesis (20). However, these Atg5–/– mice were generated by mating Atg5+/– mice, so maternally inherited Atg5 protein in the cytoplasm of Atg5-null oocytes might have rescued the autophagy-deficient phenotype during early embryogenesis. Indeed, we observed normal levels of autophagy in Atg5–/– two-cell embryos derived from Atg5+/– mice (fig. S4). This is consistent with our previous observation that only very small amounts of Atg5 are required for autophagy (21). To analyze the functional relevance of autophagy during early embryogenesis, we generated oocyte-specific Atg5-deficient mice, in which both the Atg5 gene and Atg5 protein could be deleted in oocytes. Mice bearing an Atg5flox allele (22) were crossed to a transgenic line expressing Cre recombinase under the control of the Zp3 promoter, which is active in growing oocytes (Zp3-Cre mice) (23).

The ovaries of Atg5flox/–;Zp3-Cre mice contained all developmental stages of oocyte and were morphologically indistinguishable from ovaries observed in Atg5flox/+;Zp3-Cre mice (fig. S5A). Normal numbers of oocytes were recovered from superovulated Atg5flox/–;Zp3-Cre females (fig. S5B) and could be fertilized normally (fig. S5C). Atg5 mRNA expression was completely suppressed in MII oocytes collected from Atg5flox/–;Zp3-Cre females (Fig. 2A). Thus, autophagy is not essential for oogenesis or fertilization. We next determined whether autophagy was induced in these oocytes after fertilization. In two-cell embryos collected from Atg5flox/–;Zp3-Cre;GFP-LC3 females that had been crossed with wild-type males, GFP-LC3 dots were not observed; in contrast, GFP-LC3 dots were extensively generated in embryos from Atg5flox/+;Zp3-Cre (Fig. 2B) and wild-type females (Fig. 1A). In the autophagy-deficient two-cell embryos, GFP-LC3 accumulated in the nucleus; similar events are often observed in other cell types under autophagy-inactive conditions, although the physiological importance is unknown (24). Thus, oocytes from Atg5flox/–;Zp3-Cre females are completely defective in autophagy, and this cannot be rescued at the two-cell stage by wild-type sperm.

Fig. 2.

Post-fertilization autophagy is essential for development. (A) Generation of oocyte-specific Atg5-deficient mice, as shown by reverse transcription polymerase chain reaction analysis of Atg5 and actin mRNA levels in MII oocytes derived from wild-type, Atg5flox/+;Zp3-Cre, and Atg5flox/–;Zp3-Cre female mice. (B) Absence of GFP-LC3 dot formation in Atg5-deficient embryos. Embryos from Atg5flox/+;Zp3-Cre;GFP-LC3 and Atg5flox/–;Zp3-Cre;GFP-LC3 females crossed with wild-type males were isolated at 48 hours after hCG injection. Scale bar, 20 μm. (C) Genotypic distribution of offspring obtained from Atg5flox/–;Zp3-Cre females crossed with Atg5+/– males. (D) Average litter size of indicated mating pairs. Numbers above each bar are numbers of pregnancies. Each value represents the mean ± SD.

We then examined in vivo development of the oocyte-specific Atg5-deficient mice. After mating of Atg5flox/–;Zp3-Cre females and Atg5+/– males, we obtained only Atg5Δ/+ and Atg5–/+ (Δ indicates the floxed allele successfully rearranged by the Cre recombinase) pups; we did not obtain either Atg5Δ/– or Atg5–/– (Fig. 2C and fig. S5D) pups. Thus, fertilization of Atg5-null oocytes with Atg5-null sperm resulted in embryonic lethality.

In contrast, Atg5-null oocytes were able to produce pups when fertilized with wild-type sperm, which suggests that the deficiency in autophagy can be rescued by zygote-derived Atg5. We thus determined the stage at which rescue started. In embryos derived from Atg5flox/–;Zp3-Cre;GFP-LC3 females mated with wild-type males, zygotic Atg5 mRNA was detected as early as the two-cell stage (fig. S6A). However, induction of autophagy was not observed immediately; large numbers of GFP-LC3 dots appeared only after the eight-cell stage (fig. S6B). At the eight-cell stage, the nuclear accumulation of GFP-LC3 also disappeared. The timing of rescue seemed late, given the rapid onset of post-fertilization autophagy in wild-type embryos (Fig. 1A). Consistently, we noticed that the litter sizes of Atg5flox/–;Zp3-Cre females mated with Atg5+/+ males (Fig. 2D, yellow) were smaller than those of Atg5flox/+;Zp3-Cre or wild-type females crossed with Atg5+/– males (Fig. 2D, blue and green). Thus, restoration of autophagy at the eight-cell stage is insufficient to completely rescue embryonic viability. However, we cannot exclude the possibility that, despite normal ovulation and fertilization rates, autophagy-deficient oocytes are slightly compromised, resulting in low birth rate. In either case, the clear dependence of survival on sperm genotype (Fig. 2C) suggests a critical role of autophagy after, rather than before, fertilization.

We next determined the stage at which development was perturbed in Atg5-deficient embryos by both in vivo and in vitro experiments. Atg5flox/–;Zp3-Cre females were mated with Atg5+/– males, and embryos were collected and examined at various stages. Development seemed to proceed normally until the four-cell stage; however, when embryos were collected at embryonic day 3.5 (E3.5), fewer than half of the embryos reached the blastocyst stage, while most of the control embryos became blastocysts (Fig. 3A). Even after culturing these E3.5 embryos for an additional 24 hours in vitro, the abnormal embryos did not develop and remained at the four- or eight-cell stage (fig. S7).

Fig. 3.

Autophagy deficiency causes embryonic lethality at the four- to eight-cell stage. (A) Defective in vivo development of autophagy-deficient embryos. E3.5 embryos from Atg5+/– or Atg5flox/–;Zp3-Cre females mated with Atg5+/– males are shown. Arrowheads indicate normally developing blastocysts; arrows indicate developmentally retarded embryos. Scale bar, 100 μm. (B) Defective in vitro development of autophagy-deficient embryos. Embryos were collected from Atg5flox/+;Zp3-Cre and Atg5flox/–;Zp3-Cre females mated with Atg5+/– or wild-type males at the two-cell stage (E1.5) and subsequently cultured in vitro for 2 days until the blastocyst stage (E3.5). Representative bright-field photographs of embryos at each stage are shown. Scale bar, 100 μm. (C) Distributions of embryonic developmental stage at each time point. Data represent means ± SE of three to five different experiments.

For the in vitro experiments, E1.5 embryos were collected and cultured for 2 days. Again, embryos obtained from Atg5flox/–;Zp3-Cre females crossed with Atg5+/– males developed to the four- to eight-cell stage almost normally (Fig. 3B; Fig. 3C, red). However, after 48 hours in culture, only 17% of embryos reached the blastocyst stage (all of which were Atg5Δ/+ or Atg5+/–, fig. S8), and 30% of embryos (some of which appeared fragmented) remained at the four- to eight-cell stage (Fig. 3, B and C, E3.5). Reduced numbers of blastocysts (53%) were detected at 48 hours in matings of Atg5flox/–;Zp3-Cre females and wild-type males (Fig. 3C, yellow), consistent with the results above (Fig. 2D). Taken together, these data suggest that autophagy deficiency causes a developmental block at the four- to eight-cell stage.

One possible mechanism underlying the developmental defects of autophagy-deficient embryos might be impaired protein recycling. Although there was no apparent difference in protein synthesis rates up to the two-cell stage, [35S]methionine incorporation into proteins in autophagy-defective embryos was reduced to about 70% of that of wild-type embryos at the four- and eight-cell stages (Fig. 4). Suppression of protein synthesis to similar levels with cycloheximide also caused a developmental block, confirming the importance of new protein synthesis during this period (fig. S9). The impaired protein synthesis in autophagy-deficient embryos is likely due to amino acid insufficiency; however, we cannot exclude the possibility that the protein synthesis defects are a secondary effect of a role of autophagy in the removal of obsolete maternal factors, or for energy production within embryos.

Fig. 4.

Reduced protein synthesis in autophagy-defective embryos. Oocytes and embryos were radiolabeled with [35S]methionine for 2 hours. Total lysates were resolved by SDS–polyacrylamide gel electrophoresis and subjected to autoradiography. 35S incorporation was measured and normalized against signals from the wild type. Numbers of oocytes or embryos used are shown above each lane. Data represent means ± SE of three independent experiments. WT, embryos derived from wild-type females crossed with wild-type males; f/-, embryos derived from Atg5flox/–;Zp3-Cre females crossed with Atg5+/– males.

Our results show that autophagy is up-regulated shortly after fertilization and is essential for preimplantation development. These results are not inconsistent with the finding that conventional Atg5–/– mice survive preimplantation development (20). Rather, these two different mouse models clearly demonstrate the specific importance of autophagy during very early post-fertilization development, in which maternally inherited Atg5 protein remains in the cytoplasm. Autophagy may be dispensable for later development. We do not know whether post-fertilization autophagy is specifically important for mammals. In contrast to birds, fish, and amphibians, mammalian preimplantation development progresses very slowly without extracellular nutrient stores; thus, autophagy may be a unique strategy to support mammalian development.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Movie S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article