Nutrient deprivation induces mouse embryonic diapause mediated by Gator1 and Tsc2 Free

Adaptation to stress is an evolutionary strategy employed by organisms in response to harsh environments. In eukaryotes, organisms survive in complicated and dynamic environments, and face stresses such as starvation, heat, cold, chemical stimuli and hypoxia. Organisms can evolve different strategies to survive in these difficult environments. Diapause is a reversible process in which an organism enters a dormant state by suppressing energy-requiring activity such as cell proliferation and metabolism. This ‘wait-it-out’ strategy is essential for organisms to survive harsh stresses and revive when the environment is favorable for growth and development. Indeed, embryonic development is arrested at blastocyst stage during embryonic diapause, which is found in over 130 species of mammals (Ziegler, 1843; Renfree and Shaw, 2000).

Nutrient deficiency in the uterus during embryonic diapause in other mammals (such as rat and roe deer) has been found in recent reports (van der Weijden et al., 2021; Daniel and Krishnan, 1969; van der Weijden et al., 2019). Moreover, previous research has showed that mouse embryonic diapause can also be experimentally induced by ovariectomy (Yoshinaga and Adams, 1966; Weitlauf and Greenwald, 1968). Mouse blastocysts can be induced into a diapause-like state using mTOR inhibitor or Myc inhibitor in vitro, and after mTOR or Myc inhibition, blastocysts are characterized by hypo-transcription and hypometabolism (Hondo and Stewart, 2005; Boroviak et al., 2015; Hussein et al., 2020; Scognamiglio et al., 2016). Although many studies have focused on embryonic diapause (Stewart et al., 1992; Daikoku et al., 2011; Scognamiglio et al., 2016; Bulut-Karslioglu et al., 2016; Liu et al., 2020), the mechanism of embryonic diapause induced by the maternal environment is still unclear in mice.

Here, we observed that diapaused mouse blastocysts could be induced by pre-implantation maternal starvation in vivo and nutrient deprivation in vitro. We tested the hypothesis that mouse embryonic diapause is induced by decreased concentration of several nutrients in the uterine fluid (UF) of mice suffering from starvation or ovariectomy. Moreover, we uncovered that nutrient deprivation is sensed upstream of the mTORC1 pathway (Gator1 and Tsc2), which is essential for induction of mouse embryonic diapause.

Seasonal starvation is a severe stress widely experienced across animal species, therefore we first investigated whether starvation could trigger mouse embryonic diapause (Lindenfors et al., 2003). To better understand this condition, female mice were separated into individual cages at day 0.5 of pregnancy [embryonic day (E) 0.5], and food was removed (Starvation group; Sta) or left available (normal diet group; ND) in the cages at E2.0 (Fig. 1A). After 3.5 days of starvation (E5.5), we found that the implantation site was indeed absent in the uterus of Sta mice but retained in the uterus of ND mice, consistent with the report that implantation occurs at approximately the end of E4.5 in mice (Wang and Dey, 2006) (Fig. 1B). After flushing Sta mice uteri, ∼33 morphologically viable blastocysts with an obvious blastocoel were obtained (Fig. 1B). Because ND blastocysts at E5.5 were unavailable to harvest from the uterus, ND blastocysts harvested from ND mice at E4.5 were cultured in vitro to E5.5 as control. Immunofluorescence (IF) staining showed that the proportion of Cdx2⁺ and Oct4⁺ (Pou5f1), markers of trophectoderm (TE) and inner cell mass (ICM), respectively, were comparable between the ND and Sta groups (Fig. 1C). Thus, we indicated that pre-implantation maternal starvation compromised the success of mouse implantation.

To investigate whether Sta blastocysts resembled a state equivalent to diapause induction, we collected diapaused (diapaused group, DIA) blastocysts at E5.5 from normal mice which had undergone ovariectomy, as previously described (Liu et al., 2020). IF staining and morphological examination demonstrated that the DIA blastocysts had similar cell compositions and blastocyst area as Sta blastocysts harvested at E5.5 (Fig. 1C; Fig. S1A,B). DIA blastocysts reportedly exhibit significant reduction in cell proliferation (van der Weijden and Bulut-Karslioglu, 2021). We therefore examined cell proliferation in ND, DIA and Sta blastocysts at E5.5 by staining with the thymidine analogue 5-ethynyl-2′-deoxyuridine (EdU), a marker of replication in the S phase. The results revealed that EdU⁺ cells mainly clustered at ICM and polar TE in ND blastocysts, which is consistent with the role of polar TE in TE proliferation (van der Weijden and Bulut-Karslioglu, 2021). Notably, the number of EdU⁺ cells was significantly lower in the Sta and DIA blastocysts than in ND blastocysts at E5.5 (Fig. 1D). Therefore, these data indicate that blastocysts obtained by maternal starvation suppress cell proliferation.

It was recently shown that INK128 (INK), a mTOR inhibitor, can induce diapause-like or paused state in mouse blastocysts or embryonic stem cells (Bulut-Karslioglu et al., 2016). mTOR comprises mTOR complex 1 and 2 (mTORC1 and mTORC2; Fig. S1C). In order to determine which mTOR complex was essential for diapause induction, we knocked out Raptor (R-edited; major functional components of mTORC1) or Rictor (Ri-edited; major functional components of mTORC2) using CRISPR/Cas9-mediated zygote injection (Table S1; supplementary Materials and Methods, Mouse zygote injection and embryo transplantation). Embryo genotyping analysis illustrated that all the sgRNA-injected embryos (n=16) had introduced mutations (indel) at the target site (Fig. S1D-F; supplementary Materials and Methods, Mouse embryo genotyping analysis). We found that all wild-type (WT) and Ri-edited blastocysts cultured in KSOM medium collapsed before E9.5, while Ri-edited and R+Ri-edited blastocysts remained morphologically viable until E22.5 (Fig. S1G). EdU staining showed that R-edited blastocysts at E6.5 exhibited low cell proliferation, whereas it was comparable between the Ri-edited and WT groups (Fig. S1H). Therefore, mTORC1 was essential for diapause induction.

To examine whether mTORC1 contributed to maternal starvation-induced blastocyst diapause in mice, we conducted immunostaining for downstream markers of mTORC1 activity, including phosphorylated RPS6 (p-RPS6) and 4EBP1 (p-4EBP1) (González et al., 2020; Wu and Storey, 2021) in Sta and DIA blastocysts at E5.5 (Fig. S1I). The p-RPS6 and p-4EBP1 signal in Sta and DIA blastocysts was almost undetectable compared with that of ND blastocysts, as previously described in diapaused blastocysts (Bulut-Karslioglu et al., 2016; Liu et al., 2020) (Fig. 1E,F). It indicates that mTORC1 activity was suppressed in mouse diapaused blastocysts induced by maternal starvation. These results suggested that Sta blastocysts at E5.5 were in the process of diapause induction.

Previous studies have indicated that implantation and embryogenesis occur after transferring diapaused blastocysts into receptor mice uteri (Liu et al., 2020; supplementary Materials and Methods, Mouse zygote injection and embryo transplantation). Given that a long time of starvation was lethal for mice, Sta blastocysts after E5.5 were unavailable. Sta blastocysts at E5.5 were transferred into pseudo-pregnant mice, and DIA blastocysts at E5.5 and ND blastocysts at E5.5 as control were additionally transferred, to examine whether the diapaused state of these blastocysts was reversible. We found that the birth rate of Sta and DIA blastocysts at E5.5 was 51.33±4.67% (N=3) and 45±2.89% (N=3), respectively, which was higher than ND control at E5.5 (N=4, 28.75±4.27%; Fig. 1G,H; Table S2). To examine whether maternal starvation impacted the health of fetuses, we measured the weight of fetuses and placentas, crown-rump length and placenta diameter. We found that these parameters were overall comparable among the ND, Sta and DIA groups (Fig. S1J,K). These collective results suggest that pre-implantation maternal starvation induces mouse embryonic diapause.

Deprivation of glucose and fetal bovine serum can induce a reversible dormant state in cultured mouse embryonic stem cells via mTORC1 repression (van der Weijden and Bulut-Karslioglu, 2021). To uncover the factors which contribute to the suppression of mTORC1 activity caused by maternal starvation, we examined whether mouse blastocysts cultured in vitro under nutrient deprivation could be triggered to enter a diapaused state. KSOM culture media was used, consisting of protein components [bovine serum albumin (BSA), L-glutamine (Gln) and amino acid (AA)], carbohydrate components [lactate (Lac), pyruvate (Pyr) and glucose (Glc)] and other substances that maintained osmotic pressure. Mouse zygotes were cultured in complete KSOM medium until reaching the blastocyst stage at E4.5 and transferred to other medium, and blastocysts left in KSOM medium (KSOM group) served as a control (Fig. 2A). Removal of the protein or carbohydrate components from KSOM culture medium (K-P and K-C groups, respectively) resulted in significantly extended blastocyst survival time in vitro, with a maximum survival time of E13.5 and E17.5, respectively, while blastocysts cultured in complete KSOM collapsed at E8.5, consistent with reports that blastocysts cultured in KSOM in vitro collapsed at E7.5-E8.5 (Bulut-Karslioglu et al., 2016) (Fig. S2A). Additionally, removal of both protein and carbohydrate components (K-PC group) led to an extended blastocyst survival time in vitro (Fig. 2B). Blastocysts cultured in KSOM first expanded at E5.5 then gradually collapsed at E7.5 (Fig. S2B). However, blastocyst area during blastocyst K-PC culture from E5.5 to E11.5 was almost consistent (Fig. S2C). IF staining illustrated that the proportion of Cdx2⁺ and Oct4⁺ was comparable between K-PC and DIA blastocysts at E8.5 (Fig. 2C; Fig. S2D). It indicates that nutrient deprivation can delay collapse for over a week (Movie 1).

To examine whether nutrient deprivation induces diapaused blastocysts, we examined cell proliferation by EdU staining. Cell count indicated that the number of EdU⁺ cells was significantly lower in K-P, K-C and K-PC blastocysts compared with blastocysts cultured in KSOM after 48 h of culturing (E6.5) (Fig. 2D; Fig. S2E). Subsequent IF detection of p-RPS6 and p-4EBP1 to determine whether mTORC1 activity contributed to these effects revealed a pronounced reduction in both marker signals in K-P and K-C blastocysts compared with KSOM controls at E5.5. Additionally, K-PC blastocysts also had a reduction in p-RPS6 and p-4EBP1 levels, comparable (but not greater than) with K-P and K-C groups at E5.5 (Fig. 2E; Fig. S2F,G). These results indicate that nutrient deprivation is capable of inducing a diapaused state in cultured mouse blastocysts.

Previous reports have shown that diapaused blastocysts suppress cell apoptosis (Bulut-Karslioglu et al., 2016; Scognamiglio et al., 2016). To quantify apoptotic levels, blastocysts were immunostained with cleaved caspase-3 (Asp175) antibody (CC3), a standard marker of apoptosis. IF staining showed a gradual increase in CC3⁺ puncta from E4.5 to E7.5 in ND blastocysts, suggesting that elevated apoptosis may contribute to the observed collapse of ND blastocysts at E7.5 (Fig. S2H). Detection of embryonic apoptosis at E7.5 illustrated a significant decrease in CC3⁺ signal in blastocysts of both DIA and K-PC groups, compared with that of KSOM control (Fig. S2I). It suggests that nutrient deprivation inhibits cell apoptosis of blastocysts.

Based on the above observations, that diapaused blastocysts could be induced by nutrient deprivation in vitro, we next sought to determine which components were essential for this process. To define the minimal conditions required for the K-P induced diapaused state, we first selectively supplemented K-P medium with each protein component individually (Fig. S3A). Based on our results above, we defined nutrients capable of activating mTORC1 as essential nutrients for diapause induction. Therefore, we examined p-RPS6 levels in these blastocysts and performed survival experiments to prove it. The p-RPS6 signal was comparable with that of K-P blastocysts following the addition of BSA (K-P+BSA) or Gln (K-P+Gln), whereas supplementation with all 20 amino acids (K-P+AA) restored p-RPS6 to levels comparable with that of the KSOM group in cultured blastocysts (Fig. S3B). To examine more closely which of the 20 amino acids deprivation were involved in mTORC1 inhibition and the induction of diapause, we supplemented K-P medium with either essential (K-P+EAAs) or non-essential (K-P+NEAAs) amino acids for mice. The results indicated that K-P+EAAs but not K-P+NEAAs, led to restored p-RPS6 levels comparable with that of KSOM group in cultured blastocysts (Fig. S3B). Additionally, all blastocysts cultured in KSOM, K-P+AA and K-P+EAA medium collapsed at E9.5, whereas 70%-80% of K-P blastocysts remained morphologically viable (Fig. S3C). These results suggest that deprivation of EAAs is essential for mouse diapause induction in vitro. We then supplemented K-P medium with each of the EAAs individually and performed IF staining. This demonstrated that supplementation with Arg (K-P+Arg), Leu (K-P+Leu), Ile (K-P+Ile) or Lys (K-P+Lys), but not other EAAs, could increase p-RPS6 signal compared with that of K-P group in cultured blastocysts (Fig. 2F,G; Fig. S3D-F). Consistent with the results of IF detection, the addition of Arg, Leu, Ile or Lys in K-P significantly attenuated the extended survival time of K-P blastocysts, with collapse occurring before E10.5 (Fig. 2H). These cumulative data suggest that deprivation for Arg, Lys, Ile and Leu (hereafter referred to ALIL) is essential for mouse diapause induction in vitro.

Similarly, supplementing K-C medium with each carbohydrate component individually was performed to identify which of these was necessary for diapause induction. IF staining demonstrated that p-RPS6 levels could be rescued by addition of Glc (K-C+Glc) or Lac (K-C+Lac), but not pyruvate (K-C+Pyr; Fig. 2FfG; Fig. S3G). Moreover, all blastocysts cultured in K-C+Glc and K-C+Lac collapsed at E12.5, while 80% of those in K-C were morphologically viable (Fig. 2H). These collective data indicates that deprivation of Glc and Lac (hereafter referred to GL) is essential for mouse diapause induction in vitro.

We next examined whether blastocysts cultured in KOSM medium with removal of both ALIL and GL (K-ALILGL) would result in similar effects to observations in K-PC blastocysts. We found that K-ALILGL blastocysts remained morphologically viable for several days and the blastocyst area and cell composition were comparable with those of the K-PC group at E8.5 (Fig. S3H-J). Moreover, blastocyst survival time was significantly extended in K-ALILGL blastocysts (Fig. S3K). Cell proliferation, mTORC1 activity and apoptosis were significantly decreased in K-ALILGL blastocysts compared with KSOM controls, which is similar to the effects of the K-PC group (Fig. S3L-O). Thus, these results indicated that deprivation of Arg, Leu, Ile, Lys, Glc and Lac in culture medium was the overall minimal condition required to induce blastocysts into diapaused state in vitro.

To examine whether these diapaused blastocysts induced by nutrient deprivation were reversible, K-PC and K-ALILGL blastocysts at E8.5 were transferred into pseudo-pregnant mice, as were DIA blastocysts at E8.5 as control. Because KSOM blastocysts were collapsed and lacked ICM at E8.5, they were not transferred to receptors (Fig. S4A). We found that the birth rates of K-PC (N=3; 38.75±4.02%) and K-ALILGL (N=3; 36±6.00%) blastocysts at E8.5 were comparable with that of DIA controls at E8.5 (N=3; 45±2.89%) and the KSOM group at E4.5 (N=5, 52±3.74%). Moreover, the birth rate of the KSOM group at E5.5 (N=6, 30.82±6.42%) was lower than in the E4.5 group, but it was comparable with that of the diapaused blastocyst at E8.5 (Fig. 2I; Table S2). Although K-PC blastocysts could maintain a high birth rate, the reversibility of the diapaused state was compromised after a long culture time (Fig. S4B). Notably, blastocysts cultured in KSOM with INK (INK group) at E8.5 were transferred to receptors and the birth rate was 11.11±0.32%, which was significantly lower than that of the K-PC group (Fig. S4C). All fetuses in DIA, K-PC and K-ALILGL groups were born healthy (Fig. S4D-F). These collective results suggest that nutrient deprivation induces diapaused blastocysts of mice in vitro.

It is well established that a maternal diet low in protein during the preimplantation stage in mice causes a reduction in the concentration of amino acids in UF (Eckert et al., 2012). Based on the above observations and previous reports that mouse UF is composed of free amino acids, lactate and pyruvate (Harris et al., 2005), we speculated that ovariectomy and maternal starvation led to a decreased concentration of protein and carbohydrate components in UF, which thus induced embryonic diapause. Because of the small amount of UF in pregnant mice (Hoversland and Weitlauf, 1981), we ligated the mice uterus at E0.5 and collected UF from ND, DIA and Sta mice uteri at E4.5. Amino acids (Arg, Leu, Ile, Lys, Glu, Asp, Gln and Asn) were measured by high-performance liquid chromatography (HPLC) and carbohydrate components (Glc and Lac) were measured by colorimetry (Fig. 3A). Colorimetry showed that the concentration of Glc and Lac decreased significantly in the UF of DIA and Sta compared with that of the ND control (Fig. 3B). Moreover, HPLC showed that the concentration of essential amino acids (Arg, Leu, Ile and Lys) was significantly lower in the DIA and Sta groups compared with that of ND control, while the concentration of non-essential amino acid (Glu, Asp, Gln, and Asn) in these treatments were comparable (Fig. 3C; Fig. S5A,B). These results suggested that ovariectomy and maternal starvation led to a reduced concentration of Arg, Leu, Ile, Lys, Glc and Lac in the uterus of mice.

Diapaused blastocysts can attach to the uterine endometrium and start the implantation after it has been activated by estradiol (E2) in the uterus (Paria et al., 1998). However, incubating diapaused blastocysts with E2 in vitro did not result in blastocyst activation, suggesting that E2 does not act directly on blastocysts in the uterus. Therefore, we next sought to examine whether E2 activated diapaused blastocysts by increasing the concentrations of Arg, Leu, Ile, Lys, Glc and Lac in the uterus. To test this proposition, DIA mice were injected for 3 days with progesterone (DIA+P4) or E2 (DIA+E2) and UF was collected as described above (Fig. 3D). Colorimetry results showed that the concentration of Glc and Lac increased significantly in DIA+E2 group, and HPLC results showed that concentration of Arg, Leu, Ile, and Lys, but not Glu, Asp, Gln, and Asn increased in DIA+E2 group, compared with that of DIA+P4 control (Fig. 3E,F; Fig. S5C,D). In order to explore whether elevation of ALILGL concentration in the uterus was capable of activating diapaused blastocysts in DIA and Sta mice, we directly injected ALILGL into the uterus of DIA and Sta mice (DIA+ALILGL and Sta+ALILGL group respectively), and collected blastocysts from uteri at 24 h after injection. Morphological examination showed that DIA+ALILGL and Sta+ALILGL blastocysts were significantly bigger in blastocoel size compared with the DIA and Sta groups (Fig. S5E). IF detection showed that DIA+ALILGL and Sta+ALILGL blastocysts were able to recover EdU⁺ cell count and p-RPS6 levels to be comparable with those of the E2-activated DIA group (DIA-act) (Fig. 3G-J). Furthermore, in light of previous studies that ornithine decarboxylase (Odc1) was crucial factor for blastocyst activation (Liu et al., 2020; Fenelon and Murphy, 2017), IF staining of Odc1 showed that DIA and Sta blastocysts were activated by ALILGL injection (Fig. S5F,G). Therefore, these results suggested that E2 activated diapaused blastocysts by increasing the concentrations of Arg, Leu, Ile, Lys, Glc and Lac in the uteri.

As the UF mixture accumulated from E0.5 to E4.5 in mice uteri, quantification by HPLC and colorimetry could not reflect the bona fide concentration of six components in UF during embryonic diapause. Therefore, we examined the minimum concentration of ALILGL required to induce embryonic diapause in cultured blastocysts by gradient dilution (from 1- to 100-fold dilution) and found that a 50-fold decrease in concentration of ALILGL in KSOM medium (K+0.02) led to a significant decrease in cell proliferation and mTORC1 activity comparable with that of K-ALILGL in cultured blastocysts (Fig. S6A,B). This indicated that mouse blastocysts cultured in medium reducing ALILGL concentration could mimic ovariectomy-induced diapaused blastocysts. Taken together, these results suggested that the concentration of amino acids and carbohydrates decreases in UF during mouse embryonic diapause

To thoroughly characterize the process by which mice blastocysts entered into the diapaused state induced by ovariectomy, maternal starvation and nutrient deprivation, we executed RNA-sequencing (RNA-seq) analyses for ND blastocysts at E4.5 and E5.5 (hereafter named as E4.5 and E5.5 control), and DIA, Sta, K-PC and K+0.02 blastocysts at E5.5. Principal component analysis (PCA), Pearson correlation and hierarchical clustering indicated that Sta, K-PC and K+0.02 blastocysts exhibited transcriptional similarity to DIA compared with E4.5 and E5.5 control (Fig. 4A-C). These results suggest that DIA, Sta, K-PC and K+0.02 blastocysts were in the process of diapause induction.

Consistent with previous studies on diapaused blastocysts, gene expression associated with translation (e.g. Eif2d and Eif3a), DNA replication (e.g. Ssrp1), ribosome biogenesis (e.g. Rpl3 and Rpl5), mRNA processing (e.g. Tardbp and Sf3a1), cell cycle (e.g. Ccne1, Cdca5, Cdc6, Mcm3 and Mcm5) and metabolic process were all reduced significantly in DIA, Sta, K-PC and K+0.02 blastocysts compared with E4.5 and E5.5 control. Notably, gene expression associated with gluconeogenesis (e.g. Fbp1 and Fbp2) was increased in E4.5, whereas that of glycolytic process (e.g. Hk2, Pkm and Pfkl) was increased in E5.5, which might be caused by the requirement of glycolysis in peri-implantation (Malkowska et al., 2022) (Fig. 4D-F; Fig. S7A-E; Tables S3-S8). In light of our finding that deficiency of Arg, Leu, Ile, Lys, Glc and Lac are essential for inducing the diapaused state, we found that gene expression associated with response to glucose and amino acid stimulus, and ALIL metabolism were strikingly lower in DIA, Sta, K-PC and K+0.02 blastocysts compared with E4.5 and E5.5 control (Fig. S7E-J; Tables S9-S13). Notably, expression of Sesn2 which has been shown to sense Leu and upregulate mTORC1 activity (Wolfson et al., 2016), and Odc1 decreased markedly in diapaused blastocysts. These cumulative data indicate that DIA, Sta, K-PC and K+0.02 blastocysts are characterized by reduction of cell proliferation and metabolism at the transcriptomic level.

Differentially expressed genes (DEGs) analysis showed that the number of DEGs between K+0.02 and DIA (K+0.02 versus DIA) was lower than that of K-PC versus DIA. Moreover, there are 77.62% (652 of 840) of DEGs in DIA versus E5.5 overlapped with that of K+0.02 versus E5.5, whereas only 39.64% (333 of 840) overlapped with that of K-PC versus E5.5 (Fig. S7K-L). Therefore, based on the observation that K+0.02 blastocysts exhibited more transcriptional similarity with the DIA group than K-PC, we chose K+0.02 for subsequent analysis. DEGs were analyzed in DIA, Sta and K+0.02 blastocysts and were compared with the E5.5 control individually. For convenience, upregulated and downregulated DEGs in DIA, Sta and K+0.02 versus E5.5 were named DIA-up, DIA-down, Sta-up, Sta-down, K+0.02-up and K+0.02-down, respectively. Notably, 70.67% (306 of 433) of the DIA-up and 46.84% (111 of 237) of the Sta-up were overlapped with K+0.02-up, and 85.01% (346 of 407) of the DIA-down and 58.51% (141 of 241) of the Sta-down were overlapped with K+0.02-down. Moreover, we found 55 overlapping genes of upregulation and 100 overlapped genes of downregulation among DIA versus E5.5, Sta versus E5.5 and K+0.02 versus E5.5 (Fig. 4G). Gene ontology (GO) analysis demonstrated that overlapping genes of upregulation were enriched in ‘macrophage activation involved in immune response’ and ‘endocytosis’, which was the cellular response for nutrient deprivation. The overlapping genes of downregulation were enriched in ‘sterol, steroid and cholesterol biosynthetic process’, ‘steroid and cholesterol metabolic process’, ‘G1/S transition of mitotic cell cycle’ and ‘protein phosphorylation’, consistent with our above observation (Fig. 4H). Taken together, these results indicate that DIA, Sta and K+0.02 blastocysts have similar features in the suppression of cell cycle and metabolism.

To better understand the mechanism underlying nutrient deprivation-induced mTORC1 suppression, we examined the sensor of amino acid deprivation and glucose deprivation in blastocysts. Gator1 is comprised of three subunits (Depdc5, Nprl2 and Nprl3), and is capable of sensing amino acid deprivation and preventing mTORC1 from phosphorylating by Rheb in mouse embryonic fibroblasts (Bar-Peled et al., 2013) (Fig. S8A). Therefore, we knocked out Nprl2, Nprl3 or Depdc5 using CRISPR/Cas9 individually to examine whether Gator1 regulated the K-P induced diapaused state (Table S1). Embryo genotyping analysis illustrated that all the sgRNA-injected embryos had introduced mutations at the target site (Fig. S8B,C). Moreover, Nprl2 protein significantly decreased in Nprl2-edited blastocysts compared with WT control (Fig. S8D). Knocking out Nprl2, Nplr3 and Depdc5 did not impair blastocyst development (Fig. S8E). Notably, we found that knocking out Nprl2, Nplr3 and Depdc5 attenuated the extended survival time of K-P blastocysts, with collapse occurring before E8.5, comparable with that of KSOM control (Fig. S8F-H). IF staining of EdU and p-RPS6 showed that K-P-dependent suppression of cell proliferation and mTORC1 activity was compromised in Nprl2-edited, Nprl3-edited and Depdc5-edited blastocysts, but mTORC1 activity of Nprl2-edited blastocysts was still inhibited in the K-C group (Fig. 5A,B; Fig. S8I,J). These results indicated that amino acid deprivation-dependent Gator1 activation was essential for diapause induction.

Glucose deprivation causes Tsc2 to prevent Rheb from phosphorylating mTORC1 (Inoki et al., 2003a,b) (Fig. S8A). Therefore, we knocked out Tsc2 to examine whether Gator1 regulated the K-C induced diapaused state (Table S1). Knockout efficiency was determined by embryo genotyping analysis and protein identification (Fig. S9A-D). Tsc2-edited embryos could development into blastocysts normally (Fig. S9E). Notably, knocking out Tsc2 attenuated the extended survival time of K-C blastocysts, with collapse occurring before E8.5, comparable with that of KSOM control (Fig. S9F). IF staining of EdU and p-RPS6 showed that K-C-dependent suppression of cell proliferation and mTORC1 activity was compromised in Tsc2-edited blastocysts, but mTORC1 activity was still inhibited in the K-P group (Fig. 5C,D; Fig. S9G). These data indicated that amino-acid-deprivation-dependent Gator1 activation and carbohydrate-deprivation-dependent Tsc2 activation are specific mechanisms, consistent with previous studies (Bar-Peled et al., 2013; Demetriades et al., 2014). To eliminate any concerns regarding off-target impacts of the CRISPR/Cas9 system, we chose different sgRNAs from the same exon of Nprl2 and Tsc2 (sgNprl2-2 and sgTsc2-2; Fig. S10A). Similarly, blastocyst rate, indel rate, blastocyst survival time and mTORC1 activity were examined (Fig. S10A-D). These data indicate that amino acid starvation-dependent Gator1 activation or carbohydrate starvation-dependent Tsc2 activation are required to induce diapaused blastocysts.

To examine whether knocking out Nprl2 or Tsc2 individually could compromise K-PC-induced diapaused state, we cultured Nprl2-edited or Tsc2-edited blastocysts in K-PC medium. We found that these blastocysts still exhibited mTORC1 suppression, suggesting that the K-PC-induced diapaused state might be regulated by a combination of Nprl2 and Tsc2 (Fig. S10E). Notably, knocking out both Nprl2 and Tsc2 (N2T2-edited) attenuated the extended survival time in K-PC blastocysts, with collapse occurring before E8.5, comparable with the KSOM control (Fig. 5E). Moreover, N2T2-edited blastocysts recovered levels of cell proliferation, apoptosis and mTORC1 activity in K-PC and DIA group comparable with those of WT control (Fig. 5F,G; Fig. S10F). To examine whether mTORC1 hyperactivation led to recovery of the levels of cell proliferation and p-RPS6 signal in N2T2-edited K-PC blastocysts, we cultured N2T2-edited blastocysts in K-PC supplemented with INK (K-PC+INK). We found that cell proliferation and mTORC1 activity exhibited pronounced reduction and extended blastocyst survival time upon mTORC1 inhibition (Fig. S10G-I). This indicated that knocking out Nprl2 and Tsc2 hyperactivated mTORC1 leading to compromised induction of the diapaused state. These collective data indicate that both amino acid deprivation-dependent Gator1 activation and glucose deprivation-dependent Tsc2 activation are required to induce nutrient-deprivation-dependent embryonic diapause.

To examine whether knockout of both Nprl2 and Tsc2 could compromise the formation of ovariectomy-induced embryonic diapause, we transferred N2T2-edited blastocysts into the uteri of ovariectomized mice. Although collection efficiency in the DIA group (number of diapaused blastocysts obtained by flushing uterus/number of blastocysts transferred into uterus×100%) was unchanged in N2T2-edited blastocysts, over 50% were collapsed blastocysts at E6.5, whereas all DIA blastocysts were morphologically viable in the WT control (Fig. 5H; Fig. S10J; Table S14). This suggests that Gator1 and Tsc2 activation are also required to induce ovariectomy-dependent embryonic diapause in mice. Taken together, these data indicate that amino acid deprivation-dependent Gator1 activation and glucose deprivation-dependent Tsc2 activation are required to induce embryonic diapause.

As embryonic diapause is regarded as a reproductive strategy against harsh environment, it is important to decipher the effect of environmental stresses on the blastocyst. In our study, we found that low concentration of ALILGL in UF of mice induced embryonic diapause. This conclusion was further proven by the following results. Concentration of these six components increased in the UF of ovariectomized mice treated with estradiol (E2). Blastocysts cultured in a low concentration of ALILGL in culture medium (K+0.02) could mimic maternal-starvation-induced (Sta) and ovariectomy-induced (DIA) diapaused blastocysts in birth rate and transcriptome analysis. Notably, we uncovered that Gator1 and Tsc2 were upstream of mTORC1 during induction of embryonic diapause. They sensed amino acid starvation and carbohydrate starvation, leading to mTORC1 inhibition. These results suggested that nutrient deprivation induces embryonic diapause mediated by Gator1 and Tsc2.

In morphology, Sta blastocysts are different from DIA blastocysts. This might be caused by the different methods used to obtain diapaused blastocysts between maternal starvation and ovariectomy. Maternal starvation starts at E2.0 and lasts until E5.5. As for ovariectomy, E4.5 blastocysts are transferred into ovariectomized receptors and we obtained DIA blastocysts at E5.5. Therefore, based on our results, Sta blastocysts develop in nutrient deprivation during blastocyst formation, whereas DIA blastocysts develop in nutrient deficiency after blastocyst mutation, which might account for the different blastocyst morphology.

Previous reports indicate that mTOR inhibition induces mouse embryonic diapause in vitro (Bulut-Karslioglu et al., 2016). In our report, we revealed that nutrient deficiency in uteri activated upstream of mTORC1, leading to mTORC1 inhibition. This finding enhances our understanding of the crucial role of mTOR in mouse diapause induction. However, the downstream of mTORC1 inhibition by nutrient deficiency need to be further investigated.

In our report, we revealed that nutrient deficiency is essential for mouse diapause induction. In fact, recent studies have indicated the relationship between nutrient composition in the uterus and embryonic diapause in European roe deer. By examining the proteome of UF and transcriptome of blastocysts of roe deer, researchers have found a gradual decrease in protein abundance during embryonic diapause, followed by an increase during embryonic activation (van der Weijden et al., 2019, 2021). These findings provide a comprehensive overview of the relationship between UF proteins and embryonic diapause. Therefore, nutrient deficiency might be the main factor of diapause induction in other mammals.

Although we indicated that nutrient deficiency is essential for induction of mouse embryonic diapause, whether E2 injection activates diapaused blastocysts mediated by nutrient restoration in the mouse uterus is still unclear. In our report, we found that the concentration of ALILGL increases in mouse UF after E2 injection. This is consistent with previous studies that have shown that E2 promotes the transportation of amino acids and Glc in uterine luminal cells (Walters et al., 1981) and injection of E2 increases the concentration of amino acids in the uterine lumen (Roskoski and Steiner, 1967). Moreover, we found that injecting ALILGL directly into the uterus of DIA and Sta mice increases expression of Odc1 in diapaused blastocysts, consistent with activating blastocysts after E2 injection. Therefore, whether nutrient restoration can rescue mouse embryonic diapause needs to be further investigated.

Reversibility is an essential feature of the diapaused blastocysts. Although previous studies have indicated that Let-7a (Mirlet7a-1) overexpression could also induce mouse embryonic diapause by mTOR inhibition, whether ovariectomy or harsh environment impacted expression of Let-7a in blastocysts remains unclear. Few offspring could develop from blastocysts after overexpression of Let-7a for 3 (25.00%) or 4 days (14.50%) (Liu et al., 2020), and birth rate was significantly lower than that of DIA blastocysts at E8.5 (45±2.89%) in our report. It revealed that Let-7a overexpression was not the main factor for inducing mouse embryonic diapause. Here, we demonstrated that birth rate was comparable among K-PC (38.75±4.02%), K-ALILGL (36±6.00%) and DIA blastocysts at E8.5. Therefore, nutrient deficiency is a main factor for inducing embryonic diapause. Moreover, few offspring could develop from blastocysts treated with mTOR inhibitor (10%) for 5 days (Bulut-Karslioglu et al., 2016), which suggests that nutrient deficiency might impact other signal pathways outside of mTORC1.

Frozen-thawed embryo transfer, a process of assisted reproductive technology, is widely applied in synchronization between embryonic development and the window of implantation. However, there is still no consensus on when cryopreserved embryos can be thawed and transferred, because of their transience and limitation of the methods to detect the window of implantation (Lessey, 2011). Moreover, many clinical studies have shown that frozen embryo transfer increases rates of hypertensive disorders in pregnancy and risk of placenta accreta, low birthweight, preterm birth and macrosomia (Kalinderis et al., 2021; Lee et al., 2022). In our report, we propose a short-term, easy and non-supplemented method to preserve mouse blastocysts with low cell proliferation and metabolism. Therefore, it is possible to develop new methods based on human diapause to preserve human blastocysts before transferring.

Cancer cells enhance autophagy against nutrient deprivation-induced apoptosis and reduce their cell cycle and metabolism similar to blastocysts cultured in nutrient deprivation conditions (Zhang et al., 2013; Dalby et al., 2010). Recently, it has been found that dormant cancer cells which persist after chemotherapy resemble the state of diapaused blastocysts (Dhimolea et al., 2021; Rehman et al., 2021). Therefore, understanding diapause may benefit treatment of cancer by reducing cancer relapse.

Consequently, we uncovered the role of nutrient deficiency in mice diapause induction. Our study will help in the investigation of the molecular mechanisms of mouse embryonic diapause.

B6D2F1 (C57BL/6 X DBA2J) mice (7-8 weeks old) were used for zygote collection. ICR females were used for recipients. ICR mice (10-14 weeks old) were used for uterine embryo transfer. The use and care of animals and all animal procedures complied with the ethical guidelines of the Biomedical Research Ethics Committee, the Institute of Biochemistry and Cell Biology and the Center for Excellence in Brain Science and Intelligence Technology, Shanghai Institutes for Biological Science, Chinese Academy of Sciences.

At E5.5, mice were anesthetized by intraperitoneal injection of chloral hydrate, and then intravenously injected with Chicago Sky Blue (Sigma-Aldrich, C8679) to identify the implantation site. The number of implantation sites was recorded. If no implantation sites were observed, the uterus of the mouse was flushed to detect the number of blastocysts.

Blastocysts were collected at indicated time-points after detection of the copulatory plug by flushing the uteri of pregnant females using M2 medium (Merck, MR-051).

Superovulation was induced in female B6D2F1 (C57BL/6 X DBA2J) mice (7-8 weeks old) by injecting 5 IU of pregnant mare serum gonadotropin (PMSG, Merck, S20130055-75IU) followed by 5 IU of human chorionic gonadotropin (hCG, Sigma-Aldrich, CG10-10VL). Females were mated to B6D2F1 males 48 h later, and fertilized embryos were collected from oviducts 20 h post-hCG injection. Zygotes cultured in drops of KSOM medium (homemade) were covered by a layer of mineral oil (Sigma-Aldrich, M8410) in a humidified incubator under 5% CO₂ at 37°C. The formula of homemade complete nutrient KSOM, K-PC, K-ALILGL and K+0.02 medium is shown in the supplementary Materials and Methods (Mouse embryo culturing). Subsequent embryo culture was performed in 5% O₂, 5% CO₂ at 37°C in KSOM medium and 200 nM INK128 (Medchem Express, HY-13328) after optimization of concentrations.

B6D2F1 females (8 weeks) were mated naturally overnight with B6D2F1 males and plug-positive females were housed individually the following morning (E0.5). On the second evening (E2.0), females were assigned randomly to either the ND or Sta group until the uterus was flushed at E4.5 or E5.5. In the Sta group, foods were removed from the cage.

For DIA blastocyst induction, B6D2F1 blastocysts at E4.5 were transferred into the uteri of ovariectomized pseudo-pregnant ICR mice at E2.5 morning (08:00-09:00) and mice received daily 2 mg P4 (Sigma-Aldrich, P0130-25G) injections (subcutaneously). DIA blastocysts at E5.5-E8.5 meant that blastocysts were left in receptor uteri for 1 day (E5.5) to 4 days (E8.5). Moreover, N2T2-edited blastocysts were used in experiments. For the UF collection experiment, B6D2F1 females had an ovariectomy at E3.5. For diapaused blastocyst reactivation, ovariectomized mice were injected with both E2 (25 ng per mouse, Sigma-Aldrich, E8875-250MG) and P4 (2 mg per mouse) subcutaneously. In some experiments, diapaused blastocysts were activated by uterine injection of 3 µl mixture of Arg (526.67 mg/l, Targetmol, 11009), Leu (218.33 mg/l, Targetmol, 61819), Ile (218.75 mg/l, Targetmol, T0063), Lys (302.08 mg/l, Sigma-Aldrich, L5501), Glc (13.33 mM, Sigma-Aldrich, G7528) and Lac (666.67 mM, Sigma-Aldrich, L4263) 2 days after P4 injection in the DIA group and after 3 days of starvation in the Sta group. Activated blastocysts were collected on the second day (24 h) after injection by flushing the uterus. For induction of diapaused blastocysts in vitro, B6D2F1 blastocysts at E4.5 were transferred into K-PC or K+0.02 medium at 37°C under 5% CO₂ in air.

For immunofluorescence (IF) staining, diapaused blastocysts were fixed in 4% paraformaldehyde for 20 min, washed with PBS and permeabilized with 0.2% Triton X-100 in PBS for 20 min. After blocking in 1% BSA in PBS at room temperature for 2 h, the blastocysts were incubated overnight at 4°C with the following primary antibodies in blocking solution: Oct4 (Santa Cruz Biotechnology, sc-5279, 1:200), Cdx2 (Abcam, ab76541, 1:200), phospho-4EBP1 (Thr37/46, Cell Signaling Technology, 2885T, 1:200), phospho-RPS6 (Ser240/244, Abcam, ab215214, 1:200), cleaved caspase-3 (Asp175, Cell Signaling Technology, 9664T, 1:200) and Odc1 (Santa Cruz Biotechnology, sc-5279, 1:200). Embryos were washed in PBS-Tween20, 0.2% BSA, incubated with fluorescence-conjugated secondary antibodies (Jackson ImmunoResearch, 715-545-150 and 711-165-152, 1:500) for 2 h at room temperature, and mounted using fluodish mounting medium with DAPI (Beyotime, C1005, 1:1000). Imaging was performed using an Olympus FV10i confocal microscope with z-stacking at 3 μm intervals.

For cell count, we used CELLPOSE (Stringer et al., 2021) (cp model) to segment the cell boundary in confocal images and ImageJ (3D suite) to count the cell number.

For cell proliferation analysis, blastocysts were incubated for 2 h with EdU which was diluted to 10 μM in KSOM. All samples were processed according to the manufacturer's instructions (Click-iT EdU Alexa Fluor 488 and 555 Imaging Kit, Thermo Fisher Scientific, C10425 and C10338). EdU incorporation was detected by Click-iT chemistry with anazide-modified Alexa Fluor 488 and 555 from the kit.

After mating to B6D2F1 males, each uterine horn of B6D2F1 females at E0.5 was ligated at the posterior end, close to the cervix. Mice of the ND group were fed a normal diet. For the DIA group, pregnant mice were ovariectomized at E2.5 morning (08:00-09:00) and mice received daily 2 mg P4 injections (subcutaneously). For Sta group, pregnant mice were subject to maternal starvation at E2.0.

The mice were sacrificed at E4.5 and UF was collected in a 1.5 ml centrifuge tube using a 1 ml syringe. Samples were centrifuged at 10,000 g for 10 min at 4°C and supernatant was transferred to a fresh tube. Detail for the process of deproteinization of UF samples is shown in supplementary Materials and Methods (Uterine fluid collection in mice).

For glucose analysis, glucose was quantified in deproteinized UF samples of mice on ND, ovariectomy and Sta treatments using a fluorometric glucose assay kit (ab65333, Abcam). Each sample was assayed in duplicate. Details for the process of glucose and lactate measurement is shown in supplementary Materials and Methods (Glucose and lactate analysis).

Amino acids, including essential and non-essential, were determined in UF samples of mice on ND, ovariectomy and Sta treatments using fluorometric HPLC methods involving precolumn derivatization with o-phthaldialdehyde. Details for the process of amino acid measurement is shown in supplementary Materials and Methods (Amino acid analysis).

E4.5 blastocysts were harvested from mice at E4.5 with normal diet by flushing mouse uterus. Sta blastocysts were harvested from mice at E5.5 with maternal starvation. DIA blastocysts were harvested from mice at E5.5 which were ovariectomized at E3.5 and maintained by P4 injection. K-PC blastocysts were cultured in K-PC medium until E5.5. K+0.02 blastocysts were cultured in K+0.02 medium until E5.5.

The cDNA was sequenced with the Illumina NovaseqTM 6000 sequence platform. Details for the process of RNA-seq, differentially expressed genes (DEG) analysis, Pearson correlation analysis, principal component analysis (PCA), hierarchical clustering and GO enrichment analysis are shown in supplementary Materials and Methods [RNA-seq; Differentially expressed genes (DEGs) analysis; Pearson correlation analysis of replicas; Principal component analysis; Hierarchical clustering; GO Enrichment Analysis].

To measure the mean fluorescence intensity (MFI), we used the measurement tool in ImageJ to calculate MFI of the entire embryo size (circle in brightfield images). We used the same set of parameters on an Fv10i microscope in all the images for the same experiment.

R and Prism were used to perform statistical analysis between groups. For RNA-seq analysis, genes with a false discovery rate (FDR) below 0.05 and absolute fold change≥2 were considered differentially expressed genes. P≤0.05 was considered significant: *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001. P>0.05 was considered not significant (ns). All the statistical details of the experiments and the figures they correspond to can be found in the figure legends. Data are represented as mean+s.d. and unpaired two-tailed Student's t-test was used unless specified otherwise.

We thank all the staff at the Non-human Primate Facility of the CAS Center for Excellence in Brain Science and Intelligence Technology for their assistance with the experiments.

RNA-seq data generated during this study have been deposited in GEO under accession number GSE220804. All other relevant data can be found within the article and its supplementary information.

This article has an associated ‘The people behind the papers’ interview with some of the authors.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202091.reviewer-comments.pdf