Dysregulated Gln-Glu-α-ketoglutarate axis impairs maternal decidualization and increases the risk of recurrent spontaneous miscarriage

Summary

Recurrent spontaneous miscarriage (RSM) affects 1%–2% of fertile women worldwide and poses a risk of future pregnancy complications. Increasing evidence has indicated that defective endometrial stromal decidualization is a potential cause of RSM. Here, we perform liquid chromatography with mass spectrometry (LC-MS)-based metabolite profiling in human endometrial stromal cells (ESCs) and differentiated ESCs (DESCs) and find that accumulated α-ketoglutarate (αKG) derived from activated glutaminolysis contributes to maternal decidualization. Contrarily, ESCs obtained from patients with RSM show glutaminolysis blockade and aberrant decidualization. We further find that enhanced Gln-Glu-αKG flux decreases histone methylation and supports ATP production during decidualization. In vivo, feeding mice a Glu-free diet leads to a reduction of αKG, impaired decidualization, and an increase of fetal loss rate. Isotopic tracing approaches demonstrate Gln-dependent oxidative metabolism as a prevalent direction during decidualization. Our results demonstrate an essential prerequisite of Gln-Glu-αKG flux to regulate maternal decidualization, suggesting αKG supplementation as a putative strategy to rectify deficient decidualization in patients with RSM.

Graphical abstract

Highlights

• •Gln metabolism is activated during the decidualization
• •Patients with RSM show Gln metabolism blockade and aberrant decidualization
• •αKG supplementation rescues the impaired decidualization in patients with RSM

Tang et al. report that accumulated αKG derived from activated glutaminolysis contributes to maternal decidualization. Glutaminolysis blockade or reduction of αKG impairs decidualization and increases the risk for recurrent spontaneous miscarriage (RSM). Supplementation of αKG rescues the impaired decidualization in patients with RSM.

Introduction

Decidualization is a complex biological process in which extensive morphological, functional, and metabolic changes take place in endometrial stromal cells (ESCs) to promote the development of an implanting embryo.^1,2 Deficiencies in decidualization are associated with female reproductive system diseases, such as recurrent spontaneous miscarriage (RSM), recurrent implantation failure, and preeclampsia.^3,4 As a highly heterogeneous and distressing disorder, RSM is defined as two or more losses during the first trimester. It affects 1%–2% of fertile women worldwide.⁵ Fetal chromosomal abnormalities make up 50%–60% of RSM, and other factors, such as anatomical uterine abnormalities, endometrial infections, endocrine abnormalities, inherited thrombophilias, and antiphospholipid syndrome account for a small percentage.⁶ However, a large proportion of RSM cases remain unexplained.⁷ Furthermore, the frequency of euploid loss increases with each additional miscarriage, thus decreasing the likelihood of a successful pregnancy.⁸ It has been suggested that uterine factors, especially uterine decidualization, might drive intrinsically higher-order miscarriages.

Decidualization is a multistep process during which ESCs convert from fibroblast-like cells into large polygonal cells (usually named decidual stromal cells [DSCs]) that are rich in cytoplasmic glycogen and lipid droplets.^9,10 ESC polyploidy is a unique phenomenon that occurs during decidual cell differentiation in rodents.¹¹ DSCs secrete prolactin (PRL) and insulin-like growth factor binding protein-1 (IGFBP-1), which are key regulators of decidualization and have been widely used as markers of decidualization in human.¹² Decidualization requires prodigious anabolic and energetic requirements and utilizes distinct metabolic pathways for cell proliferation and differentiation.¹³ However, the feature and effects of metabolic dynamics that support these complex processes are not well characterized, and whether blocked metabolic dynamics can lead to impaired decidualization and increase the likelihood of another spontaneous abortion in patients with RSM needs to be understood further.

Here, we report that rewiring of Gln metabolism is required during the decidualization. An increase of αKG from Gln metabolism flux decreased histone methylation and supported ATP production. αKG supplementation rescued decidualization of ESCs from patients with RSM through αKG-mediated H3K27me3 demethylation in the promoters of decidual marker genes. These results highlight the crucial role of glutaminolysis in supporting decidualization and suggest that modulation of Gln metabolism would be a promising strategy for RSM therapy.

Results

The activated Gln metabolism is critical for decidualization

We collected 14 healthy endometrium samples and isolated high purity of ESCs (Figure S1A). Then, ESCs were decidualized into differentiated ESCs (DESCs) successfully in vitro over 9 days, characterized by the significant upregulation of decidual markers PRL and IGFBP1 (Figure S1B). Paired ESCs and DESCs were collected and subjected to liquid chromatography with mass spectrometry (LC-MS)-based untargeted metabolite profiling. A total of 108 differential metabolites were detected between ESCs and DESCs (Figure 1A). KEGG and GO pathway analyses revealed that Gln metabolism was significantly altered during decidualization (Figures 1B and 1C). Notably, a marked reduction in Glu was observed (ESCs versus DESCs, p = 0.037) (Table S1). Next, using nuclear magnetic resonance (NMR), we confirmed that Glu levels were notably decreased after in vitro decidualization, while Gln levels were similar (Figure 1D). Moreover, enzymes involved in Gln catabolism such as glutaminase 1 (GLS1), Glu dehydrogenase 1 (GLUD1), glutamic-oxaloacetic transaminase 2 (GOT2), glutamic-pyruvic transaminase 2 (GPT2), and succinate dehydrogenase (SDHB) were upregulated in DESCs compared with ESCs. In contrast, isocitrate dehydrogenase 2 (IDH2), involved in the conversion of αKG to citrate, did not exhibit significant changes during decidualization (Figures 1E and 1F). These results indicated that glutaminolysis was activated in the process of decidualization. The expression levels of subtypes of glutaminolysis-related enzymes are listed in Table S5. Very low expression levels were observed in several subtypes such as GLS2, GPT1, GLUD2, LDHB, and SDHA. Although both GOT1 and GOT2 were rich in ESCs and DESCs, only GOT2, the inner-membrane mitochondrial form, was further investigated since glutaminolysis occurs in the mitochondria. In addition, the expression levels of GLS1, GPT2, GOT2, and GLUD1 were examined in endometrial biopsies across the natural human menstrual cycle. Decidualization process occurred in early-secretory phases, while middle-secretory phases were the window of implantation (Figure S2A). The results showed that the expression of GLS1, GPT2, GOT2, and GLUD1 was obviously increased in endometrial biopsies from secretory phases compared with that in proliferative phases (Figures S2B and S2C). Overall, this revealed that there might be an activated glutaminolysis flux in both in vitro decidualization and in vivo decidualization in a natural human menstrual cycle (Figure 1F).

Figure 1.

Gln metabolism is activated to support maternal decidualization

(A–C) Metabolites in ESCs and decidualized ESCs (DESCs) from healthy control endometrium (n = 14) were analyzed via untargeted LC-MS metabolomics. Pathway analysis was performed and is shown as a bubble plot in the bottom panel.

(D) Relative quantification of Glu and Gln levels in ESCs and paired DESCs was performed by NMR (n = 6).

(E) Western blot analysis of Gln-metabolism-related enzymes in ESCs and paired DESCs from healthy patients (n = 9). Blot quantification and statistics are shown in the right panel.

(F) A Gln pathway map during decidualization is shown.

(G) qPCR analysis of the relative mRNA expression of PRL and IGFBP1 in DESCs with or without Gln-metabolism-related enzyme inhibitors (BPTES or AOAA) (n = 3).

(H) LC-MS analysis of absolute quantification of αKG levels in DESCs with or without inhibitors BPTES and AOAA (n = 3).

(I) Effects of knockdown for Gln metabolism enzymes GLS1, GPT2, and GOT2 on the relative mRNA expression of PRL and IGFBP1 in DESCs were analyzed by qPCR (n = 3).

(J) LC-MS analysis of absolute quantification of αKG levels in DESCs with or without the knockout of siGLS1, siGOT2, and siGPT2 (n = 3).

Data are presented as the mean ± SEM and were analyzed by two-sided unpaired Student’s t tests or two-sided paired Student’s t tests. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, not significant.

Then, to investigate whether Gln metabolism plays a vital role during decidualization, the GLS1 inhibitor bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl) ethyl sulfide (BPTES), transaminase (GOT and GPT) inhibitor aminooxyacetic acid (AOAA), and GLUD inhibitor (−)-epigallocatechin gallate (EGCG) were added to the in vitro decidualization system, respectively. The expression of PRL and IGFBP1 genes, which are markers of successful decidualization, was significantly decreased after treatment with inhibitors targeting GLS1 or transaminases (Figure 1G), while cell viability was not significantly affected (Figures S2D and S2E), indicating that decidualization was impaired by inhibition of GLS1 or transaminases. Moreover, a marked decrease of αKG was observed in DESCs after treatment with BPTES or AOAA (Figure 1H). In sharp contrast, EGCG treatment blocked neither the decidualization process nor Gln-Glu-αKG flux in DESCs (Figures S2H and S2I). In addition, small interfering RNA (siRNA)-mediated knockdown of GLS1 or transaminases resulted in impaired decidualization (Figure 1I), and a significant decrease of αKG resulted from blocked Gln-Glu-αKG flux (Figures 1J and S2K). These findings revealed that blocking Gln metabolism in mitochondria critically disrupted the decidualization process.

Blocked Gln-Glu-αKG flux is associated with impaired decidualization in patients with RSM

We next investigated the link between Gln metabolism and female reproductive system diseases caused by deficient decidualization. Although ESCs from healthy subjects and patients with RSM had similar basal expression levels of PRL and IGFBP1 (Figure S2L), ESCs from patients with RSM had deficient decidualization undergoing treatment with steroid hormones with much lower fold changes in the expression of PRL and IGFBP1 (Figure 2A). Based on the fold changes in PRL expression in DESCs versus ESCs, samples were distributed into high and low groups with the cutoff of 100 in both patients with RSM and controls. The chi-squared test showed a significant difference among patients and controls (p < 0.0001; Table S2). It was interesting that Glu levels were increased in DESCs compared with ESCs from the low groups during in vitro decidualization (Figure 2B). However, Glu levels were decreased in DESCs compared with ESCs from healthy control subjects in the high group during in vitro decidualization (Figures 1D and 2B). In addition, in order to figure out whether there were potential differences in Glu levels among ESCs from different groups, Glu levels in ESCs from healthy subjects (high group), healthy subjects (low group), and patients with RSM (low group) were evaluated. A notable decrease in Glu level was present in ESCs from the low groups versus that from the high group (Figure 2B). Furthermore, the basal protein expression levels of GLS1, GLUD1, GOT2, and GPT2 had no significant changes in ESCs from normal subjects and patients with RSM (Figure S2M). It is worth noting that GLS1, GLUD1, GOT2, and GPT2 were significantly induced in DESCs versus ESCs from control subjects (high group) but were not upregulated in DESCs versus ESCs from patients with RSM (low group) (Figure 1E and 2C). Importantly, the protein levels of GLS1, GLUD1, GOT2, and GPT2 were significantly decreased in first trimester decidual cells obtained from RSM pregnancies versus voluntary termination of normal pregnancies (Figure S2N). Both the in situ data and in vitro results indicated that dysregulated Gln metabolism happened frequently in patients with RSM with poor decidualization. To further show the causal link between activated Gln metabolism and successful decidualization, Glu and its downstream metabolite αKG were administered to ESCs from patients with RSM (low group) in the in vitro decidualization culture system. Remarkably, impaired decidualization could be rescued by supplementation with dimethyl-αKG (DM-αKG), a membrane-permeant αKG analog that could be hydrolyzed to a native αKG by intracellular esterase, but not with Glu (Figures 2D and S2O). After treatment with D-2-hydroxyglutarate (D-2-HG), rather than L-2-HG, a competitive inhibitor of αKG-dependent dioxygenases, the in vitro decidualization of ESCs from control subjects (high group) was suppressed (Figure 2E). Then, it was confirmed that a notable accumulation of αKG was detected in DESCs compared with paired ESCs in both healthy subjects (high group) and patients with RSM (high group) (Figure 2F). Contrarily, the αKG level was not changed in the in vitro decidualization of ESCs from healthy subjects (low group) and patients with RSM (low group) (Figure 2F). These findings suggested that under oxidative conditions, the accumulation of αKG produced from Glu could be a rate-limiting step for restoring successful decidualization rather than Glu, per se.

Figure 2.

Blocked Gln-Glu-αKG flux is associated with abnormal epigenetic programming, impaired decidualization, and the onset of RSM

(A) Transcription levels of the decidual markers PRL and IGFBP1 in ESCs and paired DESCs from the healthy (n = 43) and RSM (n = 29) groups were analyzed by qPCR.

(B) The Glu levels in ESCs and paired DESCs in the high group from healthy controls (n = 5) and low groups from healthy controls (n = 6) and patients with RSM (n = 6) were detected by a Glu assay kit.

(C) Western blot analysis of Gln metabolism-related enzymes in DESCs and paired ESCs from patients with RSM (n = 4). Blot quantification and statistics are shown in the right panel.

(D) Levels of PRL and IGFBP1 in DESCs from patients with RSM supplemented with DM-αKG (n = 3).

(E) Effects of (D/L)-2-HG on the relative mRNA expression of the decidual markers PRL and IGFBP1 in DESCs from healthy subjects were detected by qPCR (n = 4).

(F) Relative αKG levels were detected in DESCs versus ESCs in both of the low and high groups from healthy controls (n = 8) and patients with RSM (n = 8).

(G) Histone methylation levels in DESCs and paired ESCs from healthy controls (n = 4).

(H) Histone methylation levels in DESCs and paired ESCs from patients with RSM (n = 6). Blot quantification and statistics are shown in the right panel.

(I) Effects of the demethylase inhibitors GSKJ1 on the relative mRNA expression of the decidual markers PRL and IGFBP1 in DESCs from healthy subjects were detected by qPCR (n = 3).

(J) H3K27me3 levels in the PRL and IGFBP1 promoter regions were analyzed by ChIP-qPCR assay. Normal rabbit immunoglobulin G (IgG) was used as a negative control. Data were normalized to input and are expressed as mean fold enrichment over input relative to an IgG control (n = 3).

(K) The oxygen consumption rate (OCR) and spare respiratory capacity (SRC) of ESCs and DESCs were measured under the treatment with oligomycin (oligo), FCCP, or rotenone plus antimycin A (Rot/AA) (n = 3).

Data are presented as the mean ± SEM and were analyzed by chi-squared test, two-sided unpaired Student’s t tests, or two-sided paired Student’s t tests. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ns, not significant.

αKG-mediated histone modification reprogramming regulates decidualization

Emerging studies have reported that endometrial decidualization is accompanied by genome-wide histone modification reprogramming.¹⁴ Intracellular αKG has been implicated in the regulation of histone demethylases.^15,16,17 Thus, we speculate that αKG could have important implications for the regulation of histone demethylation in decidualization. To test this, we first examined whether the accumulation of αKG during decidualization could affect the methylation levels of histone Lys. After in vitro decidualization, DESCs exhibited significant decreases in H3K4me3, H3K9me3, and H3K27me3 and a notable increase in αKG concentration compared with paired ESCs from normal controls (high groups) (Figures 2F and 2G). Conversely, no significant changes in H3K4me3, H3K9me3, H3K27me3, or αKG concentration were found in the in vitro decidualization of ESCs from patients with RSM (low groups) (Figures 2F and 2H). Since there were no specific inhibitors for H3K4 and H3K9 demethylation, GSKJ1, which is a specific potent inhibitor for H3K27 demethylation, was used to confirm the effects of histone methylation on the in vitro decidualization. The results showed downregulated expression of decidual marker genes PRL and IGFBP1 in DESCs (Figure 2I). Besides restoring the expression of decidual markers, supplementation of DM-αKG also restored the downregulation of H3K27me3 in DESCs from patients with RSM (low groups) (Figure S2P). Then, to confirm dynamics of H3K27me3 levels in the promoters of PRL and IGFBP1 upon decidualization, chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) was performed. The results showed that H3K27me3 was notably less enriched in the PRL and IGFBP1 promoters in DESCs than in paired ESCs. However, enrichment of H3K27me3 in the PRL and IGFBP1 promoters was kept in the high level during the in vitro decidualization of ESCs from patients with RSM (Figure 2J). These results indicated that Gln metabolism was rewired to maintain αKG pools, favoring active demethylation of H3K27 in the promotor regions of PRL and IGFBP1 during decidualization.

αKG produced by glutaminolysis supports cell bioenergetics during decidualization

Decidualization is a multistep process accompanied by a significant production of extracellular matrix, which is conceivably an energy-expensive process.^18,19 Hence, to investigate whether the accumulation of αKG via glutaminolysis during decidualization also promoted energy production, we determined the oxygen consumption rates (OCRs) of ESCs and paired DESCs using a Seahorse extracellular flux analyzer. As expected, a higher spare respiratory capacity (SRC), which is the extra capacity available in cells to produce energy in response to increased stress or work, was obtained in DESCs than that in paired ESCs. The increase in SRC was suppressed when Gln was deprived during in vitro decidualization but was rescued after the addition of DM-αKG (Figure 2K). These results indicated that αKG produced by glutaminolysis supported cell bioenergetics during decidualization.

Decreased Glu intake contributes to pregnancy loss via impairing decidualization in mice

We next determined whether the link between Gln metabolism and decidualization found in humans also exists in mice. First, we confirmed that the Glu was markedly decreased with the gestational time from nonpregnancy (NPG) to gestational day (GD) 8.5 and remained largely stable from GD8.5 to GD10.5. Then, there was a notable drop on GD12.5 (Figures 3A and S3A). Subsequently, we found a slight accumulation in downstream metabolite αKG from NPG to GD8.5 (without statistical significance). The significant changes were observed on GD12.5 (Figures 3B and S3A). Then, to explore the possible connection between decreased Glu and abnormal pregnancy, mice were distributed into a standard diet (StD) group and a Glu-free diet (GfD) group. The GfD led to decreased Glu and αKG levels in the mouse uterus (Figure S3B). Mice in the GfD group suffered increased embryo loss at GD12.5 compared with the StD group (Figures 3C, 3D, and S3C). Additionally, lower embryo weight (Figure 3E) and impaired decidualization marked by a decrease in DtPrP expression were determined in the GfD group of mice on GD12.5 (Figure 3F). However, the embryo weight was not altered between two groups of mice on GD8.5 or GD10.5 (Figure S3D). Moreover, implantation sites were similar in the two groups of mice, indicating that ovulation and fertilization were not impaired by Glu-free feeding (Figure S3E). These results suggest that Glu deficiency in the mouse uterus leads to impaired decidualization and embryo development. Furthermore, to exclude the direct effects of Glu on embryo development and underline the importance of Glu metabolism on decidualization, we assessed the ability of the endometrium to mount a pseudo-decidualization reaction using an artificial decidualization under diet experiments (Figure 3G). A marked decrease of Glu level was seen and a notable accumulation of αKG was generated in the oil-treated horn (Figures 3H and 3I). Artificial decidualization was not totally precluded but severely compromised in mice from the GfD group. Compared with the mice in the StD group, mice in the GfD group had a remarkably diminished response to the decidual stimuli, characterized by smaller and lighter decidoma (Figures 3G and 3J), and poor stromal differentiation, characterized by lower expression of decidual marker DtPrP (Figure 3K). These abnormalities could be rescued in the mice fed a GfD supplied with 3% αKG (GfD-αKG), suggesting that impaired decidualization is caused by the decreased flux through the Glu-αKG axis (Figures 3G, 3J, and 3K). The expression levels of Gln-metabolism-related enzymes were further estimated in oil-treated and control horns by immunofluorescence staining. The results showed that GLS1, GPT2, GOT2, and GLUD1 were more upregulated in the artificial decidualized horns than those in the control horns (Figure 3L). Overall, these findings revealed that activated Gln catabolism played a crucial role in supporting decidualization in mice.

Figure 3.

Decreased Glu intake contributes to pregnancy loss and impaired decidualization in mice

(A) Targeted metabolites in uterine horns from NPG uterus (n = 6) and pregnant uterus without embryo or placenta (GD12.5, n = 6) were analyzed by LC-MS.

(B) LC-MS analysis of the absolute quantification of αKG in uterine horns and uterine decidua from NPG (n = 5) and GD12.5 (n = 5), respectively.

(C) Representative photograph (black arrow: absorption point) of uterine horns from the mice fed with standard diet (StD) and Glu-free diet (GfD) at GD12.5.

(D) Embryo absorption rates were analyzed as the ratio of the number of resorption fetus/(viable fetus + resorption fetus) in the two groups of mice (n = 9).

(E) Embryo weights were analyzed in the two groups of mice (n = 9).

(F) The mRNA levels of the decidualization-related gene DtPrP were analyzed by qPCR in the uterus of pregnant mice from the StD (n = 9), GfD without absorption rates (n = 9), and GfD with absorption sites (n = 11).

(G) Representative pictures showing the gross morphology of oil-treated uterine horns and untreated horns were collected 5 days after oil injection from the mice fed with different diet.

(H and I) LC-MS analysis of the absolute quantification of Gln, Glu, and αKG (n = 7).

(J) The ratio of the wet weight in the oil-treated horn to the wet weight in the untreated horn in different groups (n = 7).

(K) Relative expression of DtPrP in different groups (n = 7).

(L) Representative immunofluorescent staining of Gln metabolism-related enzymes (red) and nuclear marker DAPI (blue) from the oil-treated horns and paired untreated horns of StD mice. L, luminal epithelium; G, glandular epithelium; S, stromal cells. Scale bars: 250 μm and 50 μm.

Data are presented as the mean ± SEM and were analyzed by two-sided unpaired Student’s t tests. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.00.

Enhanced Gln-Glu-αKG flux during decidualization is verified using stable isotope tracing

We next tracked U-¹³C-¹⁵N-labeled Gln to measure the labeling of intermediates in an LC-MS-based stable isotope tracing study (Figure 4A). The consumption of media ¹³C₅-¹⁵N₂-Gln concentration was increased during the in vitro decidualization (Figure 4B). Compared with ESCs, m+5 Gln in DESCs was reduced by 17.8% (Figure 4C), while downstream metabolites derived from m+5 Gln, namely, m+5 Glu and m+5 αKG, were significantly decreased by 8% and 9%, respectively, during the in vitro decidualization (Figures 4D and 4E). Other downstream metabolites such as m+4 succinate, m+4 fumarate, m+4 malate, and m+4 citrate were decreased by 5%, 6%, 6%, and 4%, respectively, with no significance (Figures 4F–4I). Oxidative metabolism of [U-13C] Gln was predicted to produce m+4 succinate, m+4 fumarate, m+4 malate, m+4 Asp, and m+4 citrate by directly converting Glu to αKG to succinate. In contrast, reductive carboxylation was predicted to generate m+5 citrate and m+3 oxaloacetate (OAA), m+3 Asp, m+3 malate, and m+3 fumarate through back flux via IDH (Figure 4A). Therefore, by measuring the m+4:m+5 citrate ratio, the relative contribution of oxidative metabolism versus reductive carboxylation arising from Gln-Glu-αKG flux could be confidently quantified. We found that the m+4:m+5 ratio of citrate was increased, while m+5 citrate only accounted for 2%, much less flux than m+4 citrate (18.7%) (Figure 4F). Moreover, the labeled TCA cycle intermediates yielded from reductive carboxylation revealed much less flux (e.g., m+5 citrate, m+3 malate, m+3 fumarate) or were even not detected (e.g., m+3 Asp) (Figures 4F and 4H–4J). These findings verified that oxidative metabolism was the main pathway of Gln-derived Asp production (Figure 4J). Overall, these findings implied that αKG oxidative flux prevailed over reductive carboxylation and became the prevalent direction during decidualization. These findings were in agreement with our observation that IDH2 was not upregulated, while the oxidative decarboxylation-related enzyme SDHB was markedly increased (Figure 1E). However, in the analysis of the labeled metabolic intermediates in gluconeogenesis, there was little metabolic flux from glutamine to pyruvate or acetyl-coenzyme A (CoA) (Figures 4K and 4L). Taken together, these data uncovered the main metabolic flux of Gln metabolism in which Gln was converted to Glu and αKG, followed by oxidation of αKG in the TCA cycle during decidualization.

Figure 4.

Carbon atom and nitrogen atom transition map for [U-¹³C₅-¹⁵N₂] Gln

(A) Schematics of representative isotopologs produced from [U-¹³C₅-¹⁵N₂]-Gln labeling through Gln-dependent decidualization. M + n: a metabolite with n carbon atoms labeled with ¹³C or ¹⁵N.

(B) Consumption rate in culture of ¹³C⁵-¹⁵N² Gln for 36 h via LC-MS analysis.

(C–L) ESCs and paired DESCs (n = 5) were cultured for 36 h in the presence of tracer [U-¹³C] Gln, followed by MS of the indicated metabolites. The proportion of ¹³C in each metabolite was calculated from the mass isotopomer distribution determined by ion chromatography (IC)-MS or LC-MS.

(M) Relative ammonia excretion from ESCs and DESCs was measured by ammonia assay kits (n = 9).

(N–P) The ¹⁵N-labeled fraction of amino acids in ESCs and DESCs cultured with medium containing 2.5 mM [U-¹⁵N]-Gln for 36 h.

Data are presented as the mean ± SEM and were analyzed by two-sided paired Student’s t tests. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

In addition to serving as a carbon source, Gln metabolism also provides nitrogen for nucleotide biosynthesis.^20,21 Thus, we measured the level of ammonia in the culture medium. Intriguingly, as shown in Figure 4M, the concentration of ammonia was significantly decreased in DESCs compared with ESCs. This meant that ammonia was safely removed by some other metabolic pathways and that Gln-nitrogen should be enriched in some other intracellular metabolites under decidualization. Then, we traced the nitrogen in Gln in ESCs and DESCs using [¹⁵N₂] Gln. The utilization of m+2 Gln was increased in DESCs compared with ESCs (Figure 4N). In contrast, there was an increase of m+1 (amide-15N) Gln, which might be partly produced at the expense of Glu since Gln synthetase (GS) was also markedly increased during decidualization (Figures 4N, 1E, and 1F). This might explain that the notable decrease of ammonia concentration during decidualization could result from the activation of GS. Additionally, m+1 Glu and m+1 Asp showed significantly increased incorporation during the decidualization (Figures 4O and 4P). Increased incorporation of m+1 Asp might derive from Gln catabolism, as Asp-producing transaminase enzyme GOT2 was activated during the in vitro decidualization of ESCs from normal subjects (high group) (Figure 1E). Nevertheless, m+1 Ala and m+1 Ser showed notably decreased incorporation during decidualization. Other nonessential amino acids, such as Asn, Leu, and Ile, showed no significant difference in the labeling rates between ESCs and DESCs (Figure S4).

Discussion

Recently, the significance of reprogrammed metabolism to support ESCs to continue proliferation and differentiation as well as supply energy during decidualization has been emphasized.²² Although increased glucose uptake, Warburg-like glycolysis, and lactate shuttle have been reported to play critical roles during decidualization in driving bioenergetic needs,^13,22,23 it is clear that highly proliferative and differential cells need additional supplies of biosynthetic precursors not met by glucose metabolism.²⁴ An increased proportion of the biosynthetic requirements under aerobic conditions are reported to be met by Gln metabolism.^25,26 In this regard, we reported that an increased proportion of the biosynthetic requirements under aerobic conditions were met by Gln metabolism. Our study revealed the metabolic role and fate of Gln beyond anaplerotic influx into the TCA cycle to supply carbon to the cycle so that intermediate pools are maintained as well as energy is yielded. In particular, we uncovered the fundamental support of αKG produced from Gln metabolism for successful decidualization. The present results emphasized the crucial roles of αKG in establishing an epigenetic landscape and supplying energy during decidualization. αKG supports decidualization via JMJD3-dependent H3K27 demethylation in promotor regions of the decidual markers PRL and IGFBP1. The ten-eleven translocation family of DNA hydroxylases are also αKG-dependent dioxygenases,²⁷ but DNA methylation status is stable during decidualization.^28,29 It remains unknown what directs this specificity, which may be due to the differential sensitivities of αKG-dependent dioxygenases to cofactors.

Intriguingly, it was D-2HG, rather than L-2HG, functioning as a competitive inhibitor of αKG-dependent dioxygenases to block differentiation of ESCs during the decidualization. Elevated D-2HG resulted in enhanced repressive histone methylation, thus impairing the expression of the canonical decidual marker genes PRL and IGFBP1. The stereoisomers of 2HG, D-2HG and L-2HG, both function as potent inhibitors of αKG-dependent enzymes.³⁰ D-2HG is believed to be an oncometabolite produced either via somatic mutations in IDH1/2 or reverse flux through IDH1/2.³¹ The data presented here also confirmed that Gln-derived αKG oxidative flux prevailed over reductive carboxylation during decidualization. This may reduce the production of D-2HG to support decidualization in the physiological state. Although the molecular details remain unclear, this may result from the different affinity of 2-HG isoforms for substrate binding. Given that D-2HG could affect decidualization, it will be worthwhile to further investigate whether somatic mutations in IDH1/2 can be found to have impaired decidualization in some patients with RSM.

In summary, we uncovered the fundamental support of Gln-Glu-αKG metabolism flux for successful decidualization through Gln-Glu-αKG-dependent H3K27 demethylation in promotor regions of the decidual markers PRL and IGFBP1, while patients with RSM exhibited impaired decidualization owing to deficient Gln metabolism. In addition, Glu levels might become a biosensor for successful decidualization; however, this application warrants confirmation in larger cohorts. Overall, the present study provides a new finding that reprogramming Gln metabolism is an essential prerequisite for rewiring epigenetic modification to promote endometrial decidualization. It also sheds creative light on the possible strategy of αKG for RSM therapy.

Limitations of the study

The current study has some limitations. A GfD was used to verify the roles of Gln metabolism on the decidualization in vivo. Utilizing the conditional knockout approach would provide further insight into the effectiveness of the key genes involved in Gln metabolism on the decidualization. Despite the remarkable effectiveness of αKG to rescue impaired decidualization in patients with RSM, this clinical effectiveness warrants confirmation in larger cohorts. Furthermore, it will be worthwhile to further investigate the molecular details in regulating differential sensitivities of αKG-dependent dioxygenases to cofactors and different affinitities of 2-HG isoforms for substrate binding.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-GLS1	Abcam	Cat#ab156876
Rabbit polyclonal anti-GLS1	Proteintech	Cat#12855-1-AP
Rabbit monoclonal anti-GS	Abcam	Cat#ab176562
Rabbit polyclonal anti-GLUD1	Abcam	Cat#ab168352
Rabbit polyclonal anti-GPT2	Affinity	Cat#DF9183
Rabbit polyclonal anti-GPT2	Proteintech	Cat#16757-1-AP
Rabbit monoclonal anti-GOT2	Abcam	Cat#ab171739
Rabbit monoclonal anti-IDH2	Cell Signaling Technology	Cat#56439S
Rabbit monoclonal anti-SDHB	Abcam	Cat# ab175225
Rabbit polyclonal anti-LDHA	ABclonal	Cat# A1146
Rabbit monoclonal anti-H3	Cell Signaling Technology	Cat#4499T
Rabbit monoclonal anti-H3K4me3	Cell Signaling Technology	Cat#9751T
Rabbit monoclonal anti-H3K9me3	Cell Signaling Technology	Cat#13969T
Rabbit monoclonal anti-H3K27me3	Cell Signaling Technology	Cat#9733T
Rabbit polyclonal anti-DIO2	Proteintech	Cat#26513-1-AP
Rabbit polyclonal anti-SCARA5	Novus	Cat# NBP1-72121
Rabbit monoclonal anti-β-actin	Cell Signaling Technology	Cat#4970S
HRP-conjugated goat anti-rabbit IgG	Cell Signaling Technology	Cat#14708S
Alexa Fluor® 488 Mouse anti-human Vimentin	Biolegend	Cat#562338
Alexa Fluor® 647 Mouse anti-human Cytokeratin 7	Biolegend	Cat#563614
Alexa Fluor® 488-conjugated anti-human SCARA5	Invitrogen	Cat#673527
Biological samples
Human endometrial tissues from RSM patients	Shanghai First Maternity and Infant Hospital	See Table S3 for details
Human endometrial tissues from normal subjects	Shanghai First Maternity and Infant Hospital	See Table S3 for details
Human decidua tissues from RSM patients	Shanghai First Maternity and Infant Hospital	See Table S3 for details
Human decidua tissues from normal subjects	Shanghai First Maternity and Infant Hospital	See Table S3 for details
Chemicals, peptides, and recombinant proteins
DMEM /F12, no phenol red	Gibco	Cat#21041025
DMEM (1X), no Glutamine, no phenol red	Gibco	LOT#2234876
XF DMEM Base Medium	Agilent	Cat#103575-100
Fetal bovine serum (FBS)	Bioind	Cat#04-001-1A
Penicillin/streptomycin/ Amphotericin (PSA, 100X)	Thermo	Cat#15240112
Phosphate-buffered saline (PBS, 1X)	Corning	LOT# 21020007
Trypsin-EDTA (0.25%), phenol red	Thermo	Cat#25200072
Collagenase IV	Sigma-Aldrich	Cat# C5138
DNase type I	Sigma-Aldrich	Cat#DN25
β-Estradiol (E2)	Sigma-Aldrich	Cat#50-28-2
N6,2′-O-Dibutyryladenosine 3′,5′-cyclic monophosphate sodium salt (cAMP)	Sigma-Aldrich	Cat# D0627
Medroxyprogesterone 17-acetate (MPA)	Sigma-Aldrich	Cat# M1629
BPTES	Selleck	Cat# S7753
AOAA	Selleck	Cat# S4989
EGCG	Selleck	Cat# S2250
D-2HG	Toronto Research Chemicals	Cat#1391068-16-8
L-2HG	Selleck	Cat# S1325
GSKJ1	Selleck	Cat# S7581
L-Glutamic acid	Sigma-Aldrich	Cat# G1251
DM-αKG	Sigma-Aldrich	Cat#349631
D-(+)-Glucose	Sigma-Aldrich	Cat#G7021
L-Glutamine (13C5, 99%;15N2, 99%)	Cambridge Isotope Laboratories	Cat# CNLM-1275-H-PK
DMSO	Sigma-Aldrich	Cat#D2650
RIPA lysis buffer	Beyotime Biotechnology	Cat#P0013B
protease inhibitor cocktail	Roche	Cat#4906845001
TRIzol	Thermo	Cat#15596018
Methanol (HPLC, >=99.9%)	Sigma-Aldrich	Cat#34860
DAPI	Thermo	Cat#62248
Critical commercial assays
Lipofectamine 3000 Transfection Kit	Invitrogen	LOT#2135042
high-capacity cDNA reverse transcription kit	Takara	Cat# RR036A
SYBR Green dye	Takara	Cat# RR820
Glutamate assay kit	Sigma-Aldrich	Cat# MAK004
Ammonia assay kit	Sigma-Aldrich	Cat# AA0100
ChIP kit	Cell Signaling Technology	Cat#9005
CCK-8 kit	Dojindo	LOT#KH741
XF Cell Mito Stress Test Kit	Agilent	Cat#103015-100
Micro BCA Protein Assay Kit	Thermo	LOT#UH289148
Deposited data
LC-MS raw data files	www.ebi.ac.uk/metabolights/	MTBLS7561
Experimental models: Organisms/strains
C57BL/6 mice	Gem Pharmatech	C57BL/6
Oligonucleotides
The List of Oligonucleotides Aequences for qPCR and siRNAs	This Paper	Table S4
Software and algorithms
GraphPad Prism version 8	Graphpad Software Inc.	https://www.graphpad.com/scientific-software/prism/
ImageJ	ImageJ	https://imagej.nih.gov/ij/
FlowJo (v10)	BD Biosciences	https://www.flowjo.com/
Seahorse wave software	Seahorse bioscience	https://www.agilent.com.cn/

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Liping Jin (jinlp01@163.com).

Materials availability

This study did not generate new unique reagents.

Experimental model and subject details

Human sample collection

Endometrial tissues were obtained from normal women and RSM patients with a normal menstrual cycle who were undergoing hysteroscopic surgery for benign reasons unrelated to endometrial dysfunction, such as mild intrauterine adhesion, endometrial polyps, and hysteromyoma at Shanghai First Maternity and Infant Hospital, Tongji University School of Medicine. Subjects did not receive any hormonal treatment for at least 3 months prior to the procedure. Surplus tissues from endometrial biopsies obtained for diagnostic purposes were used for this study. The endometrium samples for ESC isolation were collected from proliferative phase. The endometrium samples for immunohistochemistry were collected from proliferative phase, early-secretory phase, mid-secretory phase and late-secretory phase. Demographics of subjects for in vitro decidualization were summarized in Table S3.

Decidua tissues were obtained from voluntary termination of normal pregnancies (age ranged from 24 to 34, 30 ± 1.5 years; gestational age at sampling, 53 ± 3.3 days, mean ± SEM and RSM pregnancies (age ranged from 27 to 35, 31 ± 1.2 years; gestational age at sampling, 52 ± 2.2 days, mean ± SEM). Patients who had experienced chromosomal abnormality, infection, anatomic deformation, endocrine abnormalities as well as immune diseases were excluded.

All tissues were collected under sterile conditions and transported to the laboratory on ice in Dulbecco's modified Eagle's medium (DMEM) /F12 (21041025; Gibco) supplemented with 1% penicillin/streptomycin/ Amphotericin (PSA, 15240112; Thermo). This study was approved by the Scientific and Ethical Committee of the Shanghai First Maternity and Infant Hospital affiliated with Tongji University. All study participants provided written informed consent.

Animal experiments

C57BL/6 mice were purchased from the Gem Pharmatech (Nanjing, China). The specific pathogen-free grade mice were housed in the animal facility of Tongji University, Shanghai, China. All animal experiments were performed in accordance with National Institutes of Health published Guide for the Care and Use of Laboratory Animals and were approved by the Biological Research Ethics Committee of Tongji University. To verify the roles of glutamine metabolism on the decidualization in vivo, a glutamate-free diet was applied for the natural pregnancy mouse model and artificial induced decidualization mouse model. For the natural pregnancy mouse model, eight-weeks old female C57BL/6 mice (weight: 20–23 g; n = 24) were divided into two equal groups: standard diet (StD) group and glutamate-free (GfD) diet group (standard diet without glutamate). They were usually fed for 2 weeks in the animal facility before coitus. Then, adult C57/BL6 female mice were mated with C57/BL6 male mice (8–10 weeks old) to establish pregnancy. The day of copulatory plug appearance was arbitrarily designated as day 0.5 of gestation. Mice were sacrificed on gestational day (GD) 8.5, 10.5 and 12.5. The mouse uterus without embryo or placenta were collected for metabolite content and gene expression analysis. The transcription levels of DtPrP in mouse uterus were analyzed by real-time qPCR. The primer sequences were described in Table S4. In addition, the weight of fetus, litter size and absorption rate were recorded. Embryo absorption rate was calculated as follows: percentage fetal loss = R / (R + V) × 100, where R represents the number of hemorrhagic implantations (sites of fetal loss) and V represents the number of viable surviving fetuses.

For the artificially induced decidualization mouse model, eight-weeks old female C57BL/6 mice (weight: 20–23 g; n = 21) were divided into three equal groups: standard diet (StD) group, glutamate-free (GfD) diet group (standard diet without glutamate) and GfD supplied with 3% αKG (GfD-αKG) group. They were usually fed for 2 weeks in the animal facility before coitus. Then, adult C57/BL6 female mice were mated with C57/BL6 vasectomized male mice (8–10 weeks old) to induce pseudopregnancy. The day of copulatory plug appearance was arbitrarily designated as GD 0.5 of gestation. 10ul of sesame oil was injected into one uterine horn on GD 3.5 of pseudopregnancy while non-injected contralateral horn served as a control. Mice were sacrificed on GD 8.5 of pseudopregnancy. The mouse deciduoma were collected for metabolomics, gene expression and immunofluorescence analysis.

Method details

Isolation of ESC and DSC

Endometrial tissues were washed with phosphate-buffered saline (PBS) and finely minced and enzymatically digested with collagenase (5 mg /ml; C5138; Sigma-Aldrich) and deoxyribonuclease (DNase) type I (100 μg / μl; DN25; Sigma-Aldrich) with constant agitation for 40 min at 37°C. The resulting dispersion was filtered through 40 μm nylon strainers (BD Falcon, USA). The filtrate was then centrifuged at 1200 rpm for 5 min to further remove the leukocytes and erythrocytes. The ESCs were resuspended in DMEM/F-12 containing 10% fetal bovine serum (FBS, 04-001-1A; BI) and 1% PSA, plated on culture flasks, and incubated at 37 °C in 5% CO2.

DSC were isolated from decidua tissues which were obtained from normal pregnancies and RSM pregnancies according to previously described methods.³²

Flow cytometry (FCM) analysis

To identify and evaluate the purity of ESCs in human endometrium, the cells were stained with Alexa Fluor® 488 Mouse anti-human Vimentin antibody (562338, Biolegend) and Alexa Fluor® 647 Mouse anti-human Cytokeratin 7 antibody (563614; Biolegend). To identify and evaluate the expression of SCARA5 in ESCs and DESCs, the cells were stained with Alexa Fluor® 488-conjugated anti-human SCARA5 (673527; Invitrogen).

Cell culture and treatment

ESCs were cultured in DMEM/F12 media supplemented with 10% FBS and 1% PSA at 37 °C under 5% CO2 humidified air. An improved stimulation model was used to induced ESCs into DESCs in vitro decidualization with E2 (1×10^-8 M; 50-28-2; Sigma-Aldrich), MPA (1×10^-6 M; M1629; Sigma-Aldrich) and cAMP (5×10^-5 M; D0627; Sigma-Aldrich) for nine days. As compared to the tranditional stimulation model (1×10^-6 M MPA and 5 × 10⁻⁴ M cAMP for eight days),¹⁰ our improved stimulation model seemed to suppress the arising of senescent decidual cells in the in vitro decidualization and was more effective in detecting the deficiency of in vitro decidualization of ESCs from RSM patients (Figures S1B–S1E). GLS1 inhibitor BPTES (S7753; Selleck Chemicals), transaminase inhibitor AOAA (S4989; Selleck Chemicals), GLUD inhibitor EGCG (S2250; Selleck Chemicals), histone demethylases inhibitor D-2HG (1391068-16-8; Toronto Research Chemicals), L-2HG (S1325; Selleck Chemicals) and H3K27me3 inhibitor GSKJ1 (S7581; Selleck Chemicals) were added into the culture medium in vitro decidualization with no effects on cell viability (Figures S2D–S2G). siRNAs targeted GLS1, GPT2, GOT2 genes were purchased from GenePharma (Shanghai, China) and transfected into cells using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. The sequences of siRNA targets used in this study were listed in Table S2. In addition, cells were supplied with metabolites such as glutamate (g1251; Sigma-Aldrich), DM-αKG (349631; Sigma-Aldrich) in the in vitro decidualization culture system to investigate the causal link between activated glutamine metabolism and successful decidualization.

RNA extraction and quantitative reverse transcription PCR (qPCR)

Total RNA was isolated using TRIzol reagent (15596018; Thermo). A total of 500 ng RNA was used for first-strand cDNA synthesis with a high-capacity cDNA reverse transcription kit (RR036A; Takara). qPCR were performed using a StepOne analyzer (Applied Biosystems, Carlsbad, Carlsbad, CA, USA) with SYBR Green dye (RR820; Takara) according to the manufacturer’s protocol. β-actin was used as a reference gene. Relative quantification was performed by means of the $2^{\mathsf{-}\mathsf{\Delta }\mathsf{\Delta }\mathsf{C}\mathsf{t}}$ method. The primers were listed in Table S2.

UHPLC-HRMS/MS UHPLC-MS/MS GC/MS

Cell samples (about 1×10⁷ cells per sample) were processed by 5 cycles of 1 min ultra-sonication and 1 min interval in ice-water bath and stood for 30 min at −20°C. After centrifugation at 15000 rcf for 15 min at 4°C, 1 mL supernatant was evaporated to dryness. The residues were reconstituted in 50 μL of 50% aqueous acetonitrile (1:1, v/v) containing 50 mM ammonium acetate prior to UHPLC-MS/MS analysis and UHPLC-HRMS analysis. After UHPLC-MS/MS and UHPLC-HRMS analysis, 10 μL of 50% aqueous acetonitrile was evaporated to dryness under nitrogen stream. The dry residues were reconstituted in 20 μL of 30 mg/mL methoxyamine hydrochloride in pyridine, and the resulting mixture was incubated at 37°C for 60 min. A 30 μL of MTBSTFA (with 1% TBDMCS) was added into the mixture and derivatized at 70°C for 60 min prior to GC-MS metabolomics analysis.

Chromatographic separation was performed on a ThermoFisher Ultimate 3000 UHPLC system with a Waters BEH Amide column (2.1 mm × 100 mm, 1.7 μm). The injection volume was 2 μL and the flow rate was 0.35 mL/min. The column temperature was 30°C. The mobile phases consisted of water (phase A) and acetonitrile/water (90:10, v/v) (phase B), both with 10 mM ammonium formate (pH = 9). A binary gradient elution system of mobile phase A (water) and mobile phase B (ACN: water = 95:5, v/v) was used. Both A and B contained 2 mM ammonium formate and 0.3% formic acid. The flow rate was 0.4 mL/min and column temperature was 40°C. The gradient elution was performed as follows: 100% B maintained for 1 min, then changed to 50% B from 1 to 12 min, and hold for 1 min. The eluents were analyzed on a ThermoFisher Q Exactive™ Hybrid Quadrupole-Orbitrap™ Mass Spectrometry (QE) in Heated Electrospray Ionization Negative (HESI-) mode. Spray voltage was set to 3500 V. Capillary and Probe Heater Temperature were 250°C and 300°C, respectively. Sheath gas flow rate was 35 arbitrary unit (Arb), and Aux gas flow rate was 10 Arb. S-Lens RF Level was 50 Arb. The full scan was operated at a high-resolution of 70000 FWHM (m/z = 200) at a range of 70-400 m/z with AGC Target setting at 3×10⁶.

Details regarding metabolomics analysis, such as the quality control, data processing, determination of metabolites, and selection of differential metabolites were followed by the guideline.³³ All the MS spectra were processed using the software package qualitative analysis and profinder (Agilent Technologies, Inc., USA). The absolute concentration of glutamate was calculated with the standard curve of glutamate. The concentration of all metabolites was normalized to 1×10⁷ cells for cell samples.

Nuclear magnetic resonance (NMR)

ESCs and DESCs were subjected to NMR analysis and normalized to 1×10⁷ cells. Briefly, the dried extracts were dissolved in 570 μL of phosphate buffer (0.15 M, K₂HPO₄-NaH₂PO₄, pH 7.43) containing 80% D₂O (v/v) and tri-methylsilyl propionate (TSP, 0.2915 mM). Then the mixture was centrifuged at 16000 g for 10 min at 4°C. Then 530 μL each supernatant was transferred into a standard 5 mm NMR tube for further analysis. All the one-dimensional ¹H NMR spectra were acquired at 298 K on a Bruker Advance III 600 MHz NMR spectrometer (600.13 MHz for proton frequency) equipped with a quaternary cryogenic inverse probe (Bruker Biospin, Germany) using the first increment of the gradient selected NOESY pulse sequence (NOESYGPPR1DQ). Sixty-four transients were collected into 32 k data points with a spectral width of 20 ppm for each sample. All the NMR spectra were processed using the software package TOPSPIN (V3.6.0, Bruker Biospin, Germany). For ¹H NMR spectra, an exponential window function was employed with a line broadening factor of 1 Hz and zero-filled to 128 k prior to Fourier transformation. Each spectrum was then phase and baseline-corrected manually with the chemical shift referenced to TSP (δ 0.00). The spectral regions were then integrated into bins with the width of 0.002 ppm (1.2 Hz) using AMIX software package (V3.8.3, Bruker Biospin). The absolute concentration of metabolites was calculated with the known concentration of TSP.

Western blot analysis

Lysates were prepared using RIPA lysis buffer (Beyotime Biotechnology, China) supplemented with a protease inhibitor cocktail (4906845001; Roche). Proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to a PVDF membrane. The membrane was blocked with 5% non-fat milk for 2 h at room temperature and incubated with antibodies against GLS1 (1:1000; ab156876; Abcam), GS (1:1000; ab176562; Abcam), GLUD1 (1:1000; ab168352; Abcam), GPT2 (1:1000; DF9183; Affinity), GOT2 (1:1000; ab171739; Abcam), IDH2 (1:1000; 56439S; CST), SDHB (1:4000; ab175225; Abcam), LDHA (1:2000; A1146; ABclonal), H3 (1:1000; 4499T; CST), H3K4me3 (1:1000; 9751T; CST), H3K9me3 (1:1000; 13969T; CST), H3K27me3 (1:1000; 9733T; CST), DIO2 (1:1000; 26513-1-AP; Proteintech), SCARA5 (1:1000; NBP1-72121; Novus) and β-actin (1:1000; 4970S; CST) at 4°C overnight. Then, the blots were incubated for 1 h at room temperature with an HRP-conjugated goat anti-rabbit IgG (1:3000; 14708S; CST). The immunoreactive bands were detected with an enhanced chemiluminescence solution (Millipore) and then visualized using a FluorChem E imaging instrument (Protein Simple, San Jose, CA, USA). Then, imageJ was applied to quantify the relative protein levels. First, we determined the background-subtracted densities of targeted proteins and ACTB proteins. Then, we divided the grayscale values of targeted proteins by the grayscale values of ACTB in their respective lanes. The resulting normalized values were used to determine the p-values, fold changes and/or graphs.

Analysis of glutamate concentration

The levels of glutamate from ESCs and DESCs (normalized to 1×10⁶ cells) were detected with the Glutamate assay kit according to the protocol (MAK004; Sigma-Aldrich).

Immunostaining

Paraffin sections were technically supported by Wuhan Servicebio Technology Co., Ltd. (China). Human endometrium was labeled with anti-human GLS1 (1:100; ab156876; Abcam), GPT2 (1:100; DF9183; Affinity), GOT2 (1:100; ab171739; Abcam) and GLUD1 (1:100; ab168352; Abcam) antibodies overnight at 4°C in a humid chamber. After washing three times with PBS, the sections were incubated with dunkey anti-rabbit IgG H&L preadsorbed antibody for 50 min and the signal was visualized with GTvision TMIII Immunohistochemical Detection Kit (Servicebio; G1211).

Uterus tissues of oil-treated and untreated mice were labeled with rabbit anti-mouse GLS1 (1:100; 12855-1-AP; Proteintech), GPT2 (1:100; 16757-1-AP; Proteintech), GOT2 (1:100; ab171739; Abcam) and GLUD1 (1:100; ab168352; Abcam) antibodies overnight at 4°C placed in a wet box. After washing three times with PBS, the sections were incubated with Cy5 conjugated Goat Anti-rabbit IgG (H+L) (Alexa Fluor® 648; GB27303). And the nuclei were stained with 4ʹ,6-diamidino-2-phenylindole (DAPI; Servicebio; G1012) for 10 min and kept in dark place. The sections were observed and took pictures under a fluorescence microscope (NIKON ECLIPSE C1).

Measurements of oxygen consumption

Cells (40,000 cells/well) were seeded in a 24-well XFe Cell Culture Microplate and incubated overnight (n = 3). The mitochondrial respiratory function of cells was examined using a Seahorse XFe 24 Extracellular Flux Analyzer (Agilent, CA, USA). A sensor cartridge needs to be hydrated overnight. Reagents were loaded into hydrated sensor cartridge ports as follows: Port A: 56 μL Oligomycin (ATP synthase inhibitor) at 1.0 μM, Port B: 62 μL carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP) (uncoupling agent) at 2.0 μM, Port C: 69 μL Rotenone/antimycin A (rotenone, mitochondrial complex-I inhibitor; antimycin A, mitochondrial complex-III inhibitor) at 0.5 μM. Cell culture media were replaced by FBS-free medium consisting of XF Base Medium supplemented with 10 mM glucose, 1 mM pyruvate, and 2 mM glutamine (Agilent). When the calibration of the sensor cartridge passed, the XF Cell-Culture Microplate was put into the instrument for three measurement cycles. All data were normalized by protein content and analyzed using Wave software.

Measurement of ammonium in medium

Ammonium ions react with α-ketoglutarate and NADPH in the presence of L-glutamate dehydrogenase to form L-glutamate and oxidized NADP+. The decrease in absorbance at 340 nm, due to the oxidation of NADPH, is proportional to the ammonia concentration. The amount of ammonium ions in the cell culture medium was determined using the Ammonia Assay Kit (AA0100; Sigma-Aldrich) according to the manufacturer’s instructions.

Stable isotope tracing experiment of [U-13C-15N] glutamine

We applied a stable isotope tracing (SIT) strategy to obtain all possible fates of [U-13C-15N] glutamine (CNLM-1275-H-PK; Cambridge Isotope Laboratories), and quantitative information on the relative incorporation of glutamine–derived metabolites based on the stable isotope labeling pattern. SIT used the Agilent untargeted metabolite profiling software [MassHunter Qualitative Analysis 6.0, MassProfinder 8.0 and MassProfiler Professional (MPP 13.0)] for an initial untargeted identification of differentially-expressed metabolites in cells and media. Differential ion features then serve as targets for generating curated isotopologues that are expected to contain incorporated stable isotope elements. 2 mM of [U-¹³C-¹⁵N] glutamine was added to the DMEM/F12 medium without glutamine in the last 36 h during the nine days in vitro decidualization. The stable isotope distribution of individual metabolites was measured by LC-MS as described above.

Chromatin immunoprecipitation quantitative PCR (ChIP-qPCR)

ChIP assays were conducted using a CST ChIP kit (9005; CST). Firstly, cells were crosslinked with 1 % formaldehyde for 10 min and DNA was sonicated into fragments ranging in size from 250 bp to750 bp. Sheared chromatin was immunoprecipitated with antibodies against H3K27me3 and non-specific rabbit IgG overnight at 4°C and the precipitated DNA fragments were quantified by real time qPCR using the primers listed in Table S2.

CCK-8 assay

The viabilities of cells treated with inhibitors associated with glutamine metabolism were assessed using a CCK-8 kit (Dojindo, Shanghai, China) according to the manufacturer’s instructions.

The next generation RNA sequencing

Total RNA of ESCs and paired DESCs were isolated using the Trizol Reagent. Then, the concentration, quality and integrity were determined using a NanoDrop spectrophotometer. Three micrograms of RNA were used as input material for the RNA sample preparations. Then sequencing libraries were generated and was then sequenced on NovaSeq 6000 platform (Illumina) by Shanghai Personal Biotechnology Cp.Ltd.

Quantification and statistical analysis

Data are presented as the mean ± SEM. Statistical differences were analyzed using GraphPad Prism version 8 (GraphPad software, San Diego, CA, USA). Chi-square tests were performed for proportion, while two-sided Student’s t-tests were performed for continuous data. If unpaired data did not have similar variances, the t-test with Welch’s correction was performed. Differences were considered statistically significant if the P values were less than 0.05. Significance is indicated as follows: ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

Published: May 2, 2023

Data and code availability

• •The LC-MS data were deposited in the MetaboLights repository (www.ebi.ac.uk/metabolights/) under the accession number MTBLS7561. Data generated and/or analyzed in this study, excluding identifying personal information, are available from the lead author Liping Jin (jinlp01@163.com) with reasonable request to protect research participant privacy.
• •This paper does not report the original code.
• •Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.