AMPK/mTOR downregulated autophagy enhances aberrant endometrial decidualization in folate-deficient pregnant mice

2.1 Animals and ethical approval

The Animal Facility of Chongqing Medical University provided the Kunming mice. All animal experiments were approved by the Laboratory Animal Centre of Chongqing Medical University, China (Certificate No.: SYXK (YU) 20120001) and the Ethics Committee of Chongqing Medical University. Mice were housed in a pathogen-free animal room under a controlled environment (12 h light/12 h dark).

The folate-deficient mouse model was established as previously reported (Gao et al., 2012; Liao et al., 2015). Mature female mice (25 ± 2.3 g) were randomly divided into two groups: the folate deficiency group (FD) fed a diet with no folate (Research Diets) and the normal diet group (control) fed normal fodder. After 5 weeks, mice were mated with mature males. The day a vaginal plug was found was considered the first day of pregnancy (D1). In mice, decidualization occurs with blastocyst attachment to the endometrium on Day 5 (D5) and is vigorous on Day 6–Day 8 of pregnancy (D6–8). Uterine tissues were collected from D6 to D8 during the critical period of mice endometrial decidualization and stored at −80°C for subsequent experiments. Serum folate levels were measured by electro-chemiluminescence (Zhao et al., 2006).

Uterine decidualization autophagy was blocked by the specific autophagy inhibitor 3-methyladenine (3-MA, Sigma) and intensified by the specific autophagy inducer rapamycin (Rapa, Sigma) (L. Wu, Feng, et al., 2013). Female pregnant mice were given a daily intraperitoneal injection from D1 to D6/7/8 with 2 mg/kg body-weight/day (mg/kg-bw/d) rapamycin or 15 mg/kg-bw/d 3-MA (Carmignac et al., 2011; X. Wu et al., 2014). The control group was given an equal volume of PBS using the same method.

2.2 Mouse endometrial stromal cell (mESCs) isolation and culture

The mESCs were isolated as previous study (Kover et al., 1995; Tan et al., 2004). The mice uterus tissues were minced and digested in enzyme I solution containing 6 mg/ml dispase II (Roche, MHG, GER) and 1% typsin (Sigma) at 4°C for 2 h, followed by 37°C in a 5% CO₂ incubator for 30 min. Subsequently, a 70 μm cell strainer was utilized to filter the supernatant after further digestion using 0.05% collagenase, and the stromal cells were collected by centrifugation. The stromal cells were resuspended with complete medium and plated on culture plates in the 5% CO₂ incubator at 37°C. After incubating about 1 h, the culture medium was changed to eliminate the nonadherent cells. Based on mice diet with or without folate, the stromal cells were cultured in Dulbecco's modified Eagle medium (DMEM)/F12 medium (Sigma) or folate-free roswell park memorial institute (RPMI) 1640 medium (Gibco), both containing 20% charcoal-stripped fetal bovine serum (Gibco) and 0.1% penicillin-streptomycin (Beyotime Biotechnology). Meanwhile, the mESCs were induced to decidualization in vitro by 10 nM β-estradiol (E2, Sigma) and 1 mM P4 (Sigma) for 72 h. The mESCs autophagy was blocked by 5 mM 3-MA and intensified by 100 nM rapamycin in a culture medium. The mESCs were isolated on more than three pregnant mice from each group.

2.3 Plasmid transfection

Microtubule-associated protein 1 light chain 3 (LC3) plays an important role in autophagy. To clearly realize LC3 expression level in the endometrium, plasmids containing monmer Cherry-green fluorescent protein-microtubule-associated protein 1 light chain 3 (mCherry-GFP-LC3, Gene Pharma) were transfected into the mESCs. According to the manufacturer's protocol, the mESCs were cultured in 24-well plates containing DMEM (Sigma) with 10% fetal serum (Gibco) (not containing antibiotics) for 6 h, then the prepared plasmids were added to the medium along with the Lipofectamine® 3000 Transfection kit (Invitrogen) and cultured for an additional 6 h. After successful transfection, the medium was changed to a complete medium containing 20% charcoal-stripped fetal bovine serum (Gibco) and 0.1% penicillin-streptomycin (Beyotime) for subsequent experiments. Cell images were taken by fluorescence microscopy (BX43, Olympus). A minimum of 10 microscopic views of cells were detected in each group, and each sample was assayed three times.

2.4 Transmission electron microscopy (TEM)

To explore whether autophagy occurs in endometrial cells, we collected uterine tissue and primary cells to observe autophagosomes under TEM (H7500, Hitachi). Samples were fixed in 2.5% glutaraldehyde in 0.1 M PBS buffer (pH 7.2) and 1% osmium tetroxide (OsO4) at 4°C, then postfixed in 1% OsO4 and dehydrated in an ascending series of alcohols. Subsequently, the samples were infiltrated in Araldite-Epon. Autophagosomes were photographed with a transmission electron microscope. TEM was performed on three pregnant mice in vivo and mESCs in vitro from each group, and each experiment was analyzed at least three times.

2.5 Immunofluorescence

Collected tissues were fixed in 4% paraformaldehyde (PFA) in PBS, transferred to 70% ethyl alcohol, and embedded in paraffin. After antigen retrieval, slides (5 µm) were incubated with the primary anti-LC3 antibody (1:200) overnight at 4°C. In addition, primary mESCs were plated at about 50%–70% confluence on slides (10 mm x 10 mm) and treated with hormone or no hormone. The slides were fixed by methanol and blocked by 5% bovine serum albumin (BSA), then we incubated the primary anti-LC3 antibody (1:200) overnight at 4°C. The slides were then incubated with secondary antibodies. Cell nuclei were stained using the 4, 6-diamidino-2-phenylindole stain (DAPI, Beyotime). Green fluorescence was defined as LC3 positive. All slides were viewed directly using a fluorescence microscope at a wavelength of 488 nm. Immunofluorescence was performed on four to five pregnant mice in vivo and mESCs in vitro from each group, and each experiment was analyzed at least three times.

2.6 Western blot analysis

According to previous reports (Li et al., 2015; Liao et al., 2015), protein samples were extracted from decidual tissues collected at various time points (D6, D7, D8) using cell lysis buffer (Beyotime). Then, samples were boiled in 5× sodium dodecylsulfate (SDS) sample loading buffer for 15 min. Cellular protein samples were collected using 1× SDS sample loading buffer, then boiled for 15 min. Subsequently, protein samples were analyzed using a 10% SDS-polyacrylamide gel (Beyotime). After electrophoresis, proteins were transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad). The membranes were blocked for 75 min at 37°C in 10% nonfat milk or 10% BSA and then incubated with the appropriate primary antibodies (Table S1) overnight at 4°C. After three washes with PBST (15 min each), the membranes were incubated with specific secondary antibodies for 1 h at 37°C. The bands were washed with PBST and visualized by chemiluminescence reagents (WBKLS0500, Millipore) and ChemiDocTM XRS + (Bio-Rad). Finally, target band densities were quantified with Quantity One version 4.4.0 software and normalized for loading using β-actin. Western blot was performed on more than three pregnant mice in vivo and mESCs in vitro from each group, and each experiment was analyzed at least three times.

2.7 Quantitative real-time PCR (qRT-PCR)

Total cellular RNA was isolated from the different groups using the RNAiso Plus Reagent (Takara). Reverse transcription was measured with a Prime ScriptTM RT reagent Kit (Takara) according to the manufacturer's protocols. The qRT-PCR was performed using a Bio-Rad CFX Manager 3.1 Detection System (Bio-Rad) with gene-specific primers (Table S2) and an SYBR Green PCR Kit (Takara). Experiments were conducted in triplicate for each sample, and the 2^-ΔΔCt method was used to calculate the relative gene expression levels in different samples, with β-actin as the internal control. The specific primers for qRT-PCR were designed and synthesized by Sangon Biotech Co. Ltd. qRT-PCR was performed on more than three pregnant mice in vivo and mESCs in vitro from each group, and each experiment was analyzed at least three times.

2.8 Co-immunoprecipitation (Co-IP)

The tissue lysates were extracted from endometrial decidual tissues were collected at various time points (D6 and D7) and were precleared with Protein A/G-magnetic beads (Bimake Biotechnology) for 20 min. The solution was collected after magnetic separation. The Cathepsin L primary antibody was added to the solution and incubated overnight at 4°C. The tissue lysates further incubated for 1 h with Protein A/G-magnetic beads. After magnetic separation, the immune precipitates were collected and washed with cold lysis buffer three times. Finally, the mixture of precipitates and loading buffer was boiled and then used to detect Cathepsin L and Hoxa10 protein levels by Western blot. Co-IP was performed on six pregnant mice from each group, and each experiment was analyzed at least three times.

2.9 Statistical analysis

All experiments were replicated at least three times. Statistical analysis was performed using GraphPad Prism 6.0 (GraphPad Software). Quantitative data were displayed as the mean ± SEM for each experiment. Statistical significance of the differences was determined using Student's t tests and one-way analysis of variance tests. Values of p < 0.05 were considered statistically significant.

3.1 Endometrial decidualization was impaired by folate deficiency in early pregnant mice and in primary mESCs

Folate is crucial for endometrial decidualization (Li et al., 2015; Liao et al., 2015). Serum folate concentration of folate-deficient pregnant mice was significantly lower than that in the control group (Figure 1a, p < 0.001), which was consistent with our previous study (Q. Chen et al., 2018). It indicated the successful establishment of a folate-deficient mouse model.

To validate whether folate deficiency could impair endometrial decidualization, the key proteins expressions of decidualization were detected, such as homeobox (Hoxa10), bone morphogenetic protein 2 (Bmp2), matrix metalloproteinase 2 (Mmp2), matrix metalloproteinase 9 (Mmp9), and progesterone receptor (PR) (Das et al., 2009; Sroga et al., 2012). The proteins expressions of Hoxa10, Bmp2, Mmp2, and Mmp9 were significantly decreased in the folate-deficient group compared with the normal diet group on D6, D7, and D8 which is the decidualization period in mice in vivo (Figure 1b,c, p < 0.05). It suggested that endometrial decidualization was impaired by folate deficiency, which was corroborated by our previous report (Liao et al., 2015).

The uterus is a complex organ comprising multiple types of cells, such as natural killer cells, endometrial epithelial cells, and stromal cells. During decidualization, stromal cells started transforming into decidual cells (C. Zhang et al., 2019). To further validate the effects of folate deficiency on the differentiation of endometrial stromal cells into decidual cells, the mESCs were isolated from folate deficient and normal mice, respectively (Figure S1), and were decidualized artificially by hormone in vitro. After being decidualized, the qRT-PCR results showed that the mRNA expression levels of Hoxa10 and Bmp2 were significantly decreased in the folate-deficient group when compared with that in the normal group (Figure 1d, p < 0.05). Western blot results showed that the protein expressions of Hoxa10, Bmp2, Mmp2, and PR were also significantly decreased in the folate-deficient group compared with the control group after mESCs were decidualized (Figure 1e,f, p < 0.05).

3.2 Autophagy was inhibited by folate deficiency during endometrial decidualization in early pregnant mice

There is some increasing evidence that connects autophagy with the normal development and health maintenance of the endometrium (Avagliano et al., 2015; J. Choi et al., 2012; Nakashima et al., 2017). Therefore, the role of autophagy in folate deficiency-impaired endometrial decidualization is described here. TEM was initially applied to observe the numbers of single- or double-membrane autophagosomes in the endometrium of pregnant mice on D6, D7, and D8. Compared with the normal diet group, the number of endometrial autophagosomes was drastically decreased in the folate-deficient group (Figure 2a). Immunofluorescence was performed to analyze the distribution of the autophagic marker LC3 protein in decidual cells on D8 during early pregnancy. The LC3 signaling accumulated in the normal diet group, while only weak signals were observed in the folate deficiency group (Figure 2b). The mRNA expression levels of LC3 were also significantly decreased in the folate-deficient group on D6, D7, and D8 by qRT-PCR (Figure 2c, p < 0.05). In addition, as a marker of autophagosomal membranes, LC3 including LC-I and LC3-II. After autophagosome formation, LC3-I can be transformed into LC3-II by conjugation with phosphatidylethanolamine located in the autophagosome membrane, where it is fused and degraded. Thus, LC3-II accumulation could reflect the number of autophagosomes within a cell (S. Choi, Shin, et al., 2014; Ko et al., 2017). Western blot results showed that folate deficiency dramatically reduced the conversion of LC3-I to LC3-II on D6, D7, and D8 (Figure 2d,e, p < 0.01). These data indicated that autophagy was significantly downregulated in folate deficiency mice during early pregnancy.

Additionally, we examined protein expressions of autophagy at different stages using Western blot. Compared with the normal diet group, protein expression levels of Atg12, Atg3, Atg16, and Atg7 involved in autophagosome formation were obviously decreased in the folate-deficient group (Figure 2d,e, p < 0.05). This result suggested that autophagosome formation was suppressed by folate deficiency. Furthermore, lysosomal-associated membrane protein 2 (Lamp2), Cathepsin B, and Cathepsin L were reported to play an essential role in lysosome-autophagosome fusion and the degradation process (Klionsky et al., 2012). The decreased protein expression levels of Lamp2, Cathepsin B, and Cathepsin L in the folate-deficient group (Figure 2f,g, p < 0.05) indicate that folate deficiency not only impaired autophagosome-lysosome fusion but also disturbed lysosomal degradation. These results also provided the evidence that autophagy was inhibited by folate deficiency in early pregnancy mice during endometrial decidualization.

3.3 Autophagy was inhibited by folate deficiency during decidualization in primary mESCs

We further wondered whether autophagy was reduced by folate deficiency in the primary mESCs model in vitro. A notably decreased number of autophagosomes was observed by TEM in the decidualized mESCs isolated from folate-deficient mice (Figure 3a). By immunostaining, numerous puncta of LC3 were detected in the control group, but only weak expression was exhibited in the decidualized mESCs isolated from folate-deficient mice (Figure 3b). After mESCs were decidualized by hormone, the mRNA expression levels of LC3 were also significantly decreased in the folate-deficient group when compared with that in the control group (Figure 3c, p < 0.05). Western blot analysis also showed that, after mESCs decidualized by hormone, the protein level of LC3-II/I was significantly decreased in folate-deficient group (Figure 3d,e, p < 0.05).

Meanwhile, obviously downregulated expression levels of autophagosome formation-related markers of Atg3, Atg5, Atg16, and Atg7 were found in the decidualized mESCs isolated from folate-deficient mice (Figure 3d,e, p < 0.05). Additionally, after mESCs were decidualized by hormone, the protein levels of Cathepsin L and Cathepsin B, which are the lysosomal degradation-related markers, were also significantly decreased in the folate-deficient group (Figure 3f,g, p < 0.05). Furthermore, autophagic flux was examined using a mCherry-GFP-LC3 transfection assay for assessing the autophagosomal-lysosomal fusion step (Kimura et al., 2007). By this approach, autophagosomes could exhibit green (GFP) and red (mCherry) fluorescence, resulting in yellow signals in merged images. After fusion with lysosomes, acidic autolysosomes quench GFP fluorescence and preferentially exhibit red fluorescence. Our results showed that the fusion step of autophagosomes with lysosomes was notably inhibited by folate deficiency in the decidualized mESCs (Figure 3h).

These results showed that autophagy could be inhibited by folate deficiency in the decidualized mESCs in vitro.

3.4 Aberrant decidualization induced by folate deficiency was reversed by activating autophagy in early pregnant mice and in primary mESCs

To further address whether folate deficiency-impaired endometrial decidualization was mediated by inhibited autophagy, autophagy was activated by the specific autophagy inducer rapamycin, which inhibits the mammalian target of rapamycin (mTOR) pathway by binding mTORC1 (Y. J. Kim et al., 2013). Besides this, autophagy could be blocked by the specific inhibitor 3-MA (Y. T. Wu et al., 2010). According to the results shown in Figure 4a,b, the number of embryo implantation sites on D6 was significantly reduced under the role of folate deficiency (control: 14.00 ± 0.84, FD: 7.60 ± 0.51, p < 0.001). After the mice received combined treatment with a folate-deficient diet and rapamycin, rapamycin could significantly reverse the inhibitory effect of folate deficiency on the number of embryo implantation sites (FD: 7.60 ± 0.51, FD + Rapa: 12.00 ± 0.84, p < 0.01). After combined treatment with a folate-deficient diet and 3-MA, the specific inhibitor of autophagy, 3-MA could enhance the inhibitory effect of folate deficiency on the number of embryo implantation sites (FD: 7.60 ± 0.51, FD + 3-MA: 5.80 ± 0.58, p < 0.05). Western blot results subsequently showed that the protein expression levels of Atg5 and Beclin1 were induced by rapamycin and reduced by 3-MA, indicating that autophagy was successfully activated by rapamycin and inhibited by 3-MA (Figure 4c). Further Western blot results showed that the decreased expression level of the endometrial decidualization related protein Bmp2 in the condition of folate deficiency could also be reversed by autophagy inducer rapamycin on D6, D7, and D8 during early pregnancy. However, it was not reversed by autophagy inhibitor 3-MA (Figure 4d). These results indicated that rapamycin may be able to reverse abnormal endometrial decidualization induced by folate deficiency while 3-MA could not. It suggested that autophagy played a critical role in the impaired endometrial decidualization induced by folate deficiency.

In addition, autophagy also was successfully induced by rapamycin and blocked by 3-MA in the decidualized mESCs (Figure 4e). Western blot analysis showed that the folate deficient-decreased protein expression levels of Hoxa10, Bmp2, and Mmp2 were significantly increased by autophagy inducer rapamycin and were further decreased by autophagy inhibitor 3-MA (Figure 4f).

Treatment with autophagy inducer rapamycin reversed the adverse impact of folate deficiency on endometrial decidualization both in vivo and in vitro, suggesting that a significant role of autophagy in folate deficiency-impaired endometrial decidualization.

3.5 Endometrial AMPK/mTOR signaling was disrupted in the impaired decidualization by folate deficiency in early pregnant mice and in primary mESCs

It has been reported that inhibition of adenosine 5‘-monophosphate (AMP)-activated protein kinase (AMPK) can activate the mTOR signaling pathway (Liu et al., 2017; Y. Zhou et al., 2017), which acted as one of the AMPK downstream molecules accompanied by its phosphorylated form (p-mTOR). To determine whether the AMPK/mTOR signaling pathway is involved in the folate deficiency-mediated regulation of autophagy, the AMPK/mTOR-related protein expression levels were evaluated by Western blot in vitro/vivo study. The results showed that folate deficiency significantly decreased AMPK phosphorylation and increased mTOR phosphorylation without inducing the expression of total AMPK or mTOR in the pregnant mice on D6, D7, and D8 (Figure 5a). Additionally, the protein expression level of mTOR downstream molecules p70 S6 kinase 1 (S6K1) phosphorylation was notably increased while eIF4E binding protein 1 (4E-BP1) phosphorylation was also obviously decreased in the folate-deficient group of pregnant mice (Figure 5a). The same results were obtained in the decidualized mESCs in vitro (Figure 5c). When the mTOR pathway was inhibited by rapamycin, the decreased protein expression of LC3-II/I, Atg5, and Beclin1 due to folate deficiency were significantly increased both in vitro/vivo studies (Figure 4c,e), suggesting that AMPK/mTOR was disrupted by folate deficiency to down-regulate autophagy during endometrial decidualization. Furthermore, rapamycin significantly reversed the inhibitory effect of folate deficiency on the number of embryo implantation sites (Figure 4a,b) and the endometrial decidualization-related protein expression levels of Hoxa10, Bmp2, and Mmp2 (Figure 4d,f). All data, both in vivo and in vitro studies, suggested that folate deficiency could reduce autophagy by disrupting AMPK/mTOR signaling, resulting in aberrant endometrial decidualization and adverse pregnancy outcomes.

In addition, recent research has demonstrated a new axis of AMPK-SKP2-CARM1 signaling to initiate autophagy (Shin et al., 2016), and our research team reported that AMPK-SKP2-CARM1 signaling activated autophagy to disrupt primordial folliculogenesis in mice ovary exposure of Di (2-ethylhexyl) phthalate (Y. Zhang et al., 2018). We further explored whether the AMPK-SKP2-CARM1 signaling pathway was involved in folate deficiency-impaired aberrant endometrial decidualization in vitro and in vivo. The results revealed that the protein expression levels of p-AMPK and its downstream effector the transcription factor EB (TFEB) were markedly decreased, and the protein expression levels of SKP2 p95 were obviously increased by folate deficiency in vitro and in vivo (Figures 5b and d). These data suggested that the AMPK-SKP2-CARM1 signaling pathway could participate in the process of aberrant endometrial decidualization in folate-deficient pregnant mice.

Lastly, we further explored how the impairment of folate deficiency-induced decidualization was mediated by inhibited autophagy using Co-IP. The lysates were pulled down using the Cathepsin L antibody, and the precipitates were used to detect the endometrial decidualization-related protein levels by Western blot. The result on immunoprecipitation showed that Hoxa10 interacts with Cathepsin L. Furthermore, the interaction between Cathepsin L and Hoxa10 was notably reduced in the folate deficiency group of pregnant mice on D6 and D7 (Figure 5e,f).