Dimethyl sulfoxide (DMSO), 6-aminonicotinamide (6-AN), and 17β-estradiol were purchased from Sigma-Aldrich (St Louis, MO, USA). High-glucose Dulbecco’s Modified Eagle Medium (high-glucose DMEM), penicillin–streptomycin, trypsin-EDTA (0.25%), and fetal bovine serum were purchased from Thermo Fisher Scientific (Gibco, USA). Cell Counting Kit-8 (CCK-8) was purchased from Yeasen (China). Water (HPLC Grade), acetonitrile (HPLC Grade, 99.95%), 2-propanol (HPLC Grade, 99.9%), methanol (HPLC Grade, 99.9%), ammonium hydroxide (LC–MS grade), and acetic acid (HPLC Grade, 99.7%) were ordered from Fisher Chemical. MTBE (HPLC Grade, 99.9%) was ordered from Sigma-Aldrich. Formic acid (LC-MS Grade) was ordered from Fisher Scientific.

Twelve infertility patients were recruited before undergoing hysteroscopy combined with laparoscopy. Six patients were treated with estrogen (Estrogen Group) and the remainder were not (Control Group) before they all underwent hysteroscopy combined with laparoscopy. Patients in the Estrogen Group were given estradiol valerate (Progynova) at an oral dose of 2 mg daily for 7 days before hysteroscopy and this was continued for 21 days. Endometrial tissue of both estrogen-treated and control groups were collected during surgery and frozen at −80°C until metabolite extraction. The use of endometrial tissue was approved by the Obstetrics and Gynecology Hospital of Fudan University’s ethics board (the hospital’s ethics board (NO.2021-132), Shanghai, China and consent was obtained from each patient.

The Ishikawa cells maintained in our laboratory are immortal endometrial tumor cells that can perfectly mimic the in vitro behavior of endometrial epithelium cells (24). Ishikawa cells are derived from a well-differentiated adenocarcinoma of human endometrial epithelium that expresses functional steroid receptors for estradiol (E2) and progesterone (P4) (25). The cell line represents an ideal model of normal endometrial epithelium cells owing to its phenotypic similarity and response to steroids, similar to physiological conditions (26). The Ishikawa cells were seeded at a density of 2,000 cells/well for 24 h until they were attached to the 96-well plate for cell proliferation measurement. For the metabolomics analysis, Ishikawa cells were seeded at a density of 2 × 10⁵ cells/well for 24 h until they were attached to the 24-well plate. After serum fasting for 24 h, the cells were treated with 100 nM estrogen or DMSO (vehicle control) for 48 h until their metabolites were extracted.

For the Ishikawa cells cultured in the 96-well plate, cell culture media were replaced by serum free DMEM and “fasting” for 24 h before drug treatments. Two cell proliferation assays were processed: (1) the fasted cells were treated with 5 µM of the PPP metabolism inhibitor 6-aminonicotinamide (6-AN); or 100 nM estrogen; or 100 nM estrogen along with 5 µM 6-AN; or DMSO (vehicle) for 0 h, 24 h, 48 h, 72 h, or 96 h, and subsequent cell proliferation was measured using a CCK8 kit (Yeasen); (2) the fasted cells were treated with 10 µM 6-AN; or 100 nM estrogen; or 100 nM estrogen along with 10 µM 6-AN; or DMSO (vehicle) for 0 h, 24 h, 48 h, and 72 h; cell proliferation was also measured using a CCK8 kit (Yeasen).

Metabolites were extracted from endometrial tissue samples and Ishikawa cell samples following a published protocol (27). Briefly: tissue samples (2 mg) were placed in microcentrifuge tubes and smashed with stainless steel beads at 4°C for 30 min. The homogenous samples were then transferred to 15-ml centrifuge tubes and 4.5 ml ice-cold 80% methanol was added. The Ishikawa cell samples were placed in 15-ml centrifuge tubes with 4.5 ml of ice-cold 80% methanol added. The mixture (either homogenous tissue or cells) was incubated at −80°C for 2 h and centrifuged 14,000 × g for 10 min at 4°C. The supernatant was collected into three 1.5-ml microcentrifuge tubes while the pellets were discarded. The supernatant was dried in a SpeedVac at room temperature, and the metabolites were stored at −80°C until analyzed by UHPLC-MS.

Metabolites extracted from endometrial tissue samples and Ishikawa cells were reconstituted with 100 µl of acetonitrile:water (v:v 50:50), and 5-µl sample solutions were injected into the UHPLC-MS. The metabolites were acquired using a targeted metabolomics method that was modified from a published protocol (28). The UHPLC was equipped with an HILIC column (XBridge Amide 3.5 µm, 4.6 × 100 mm) and samples were eluted with a gradient. Two buffers, buffer A (95% water and 5% acetonitrile with 20 mM ammonium hydroxide and 20 mM ammonium acetate, pH 9.0) and buffer B (acetonitrile), were used in the gradient. The total flow of the gradient was 0.25 ml/min, which started from 0 to 0.1 min, 85% B; 3.5 min, 32% B; 12 min, 2% B; 16.5 min, 2% B; and 16–17 min, 85% B. Metabolites in the samples were acquired by a QTRAP 5500+ (AB Sciex) mass spectrometer using targeted MRM methods containing 297 transitions. All transition peaks were integrated on a MultiQuant (AB Sciex) to obtain a metabolomics peak list.

The metabolomics data were analyzed using MetaboAnalyst 5.0 (29) and GraphPad (Prism 8). Heatmap, PCA score plot, and VIP score were analyzed using MetaboAnalyst. Student’s t-test was performed to evaluate significant metabolite differences between E2 and the control group (*p < 0.05, **p < 0.01, n.s., not significant) using GraphPad (Prism 8). The metabolic pathway analysis referred to the KEGG (https://www.kegg.jp/) metabolic pathway map (30).

Metabolomics based on cutting-edge UHPLC-MS techniques have been widely used in the discovery of multiple clinical mechanisms and biomarkers in recent years. We used these advanced tools to analyze metabolome alterations in the endometrial tissue of infertility patients who were treated with estrogen ( Figure 1A ). We collected endometrial tissue from 12 infertility patients; half of them were treated with E2 and the remainder were not. All the tissues were processed using the same metabolite extraction procedure and injected into the UHPLC-MS. The metabolomic results indicated that estrogen can indeed affect metabolic changes in patient’s endometrial tissues ( Figure 1B ), and most of them are associated with PPP metabolism. We then checked the metabolic intermediates correlated with the PPP and glycolysis ( Figure 1C ). Both the glycolysis metabolites (pyruvate, fructose-6-phosphate, fructose-1,6-phosphate, 1,3-diphosphateglycerate, phosphoenolpyruvate, and glyceraldehyde-3-phosphate) and PPP intermediates (sedoheptulose-1,7-phosphate, erythrose-4-phosphate, and ribose-phosphate) were upregulated after estrogen treatment ( Figure 1C ). This implies that the oral intake of estrogen can affect glucose usage in infertility patients and that glucose catabolism is promoted in general. However, we did not observe an increase in glucose aerobic oxidation ( Supplemental Figure 1 ). Levels of NAD+ and NADH fell in the estrogen-treated tissues; this indicated that glucose may flux to glycolysis and the PPP instead of aerobic oxidation catabolism.

Although we observed that glucose fluxes to glycolysis and the PPP in the endometrial tissues of infertility patients following estrogen treatment, it is still unclear if the endometrial epithelium can be directly affected by estrogen. To evaluate the effects of estrogen on endometrial epithelium cells, we treated the Ishikawa cell line with 100 nM estrogen and profiled the metabolic alterations. In vitro cell metabolomics results indicated that estrogen can extensively alter the metabolites of endometrial epithelium cells ( Figure 2A ). Most of the metabolites were elevated by 2 mg/day estrogen treatment ( Figure 2B ). This implies that estrogen could promote the metabolism of Ishikawa cells to generate more substrates for cell proliferation. To take a close look at the altered metabolites, we preformed metabolite set enrichment analysis (MSEA) using MetaboAnalyst 5.0 (29). The metabolite ontology results indicated that 12 metabolic pathways were significantly changed, and the top one was the PPP ( Figure 2C ), which was consistent with the findings in estrogen-treated endometrial tissue from infertility patients ( Figure 1C ). We then profiled the metabolic intermediates correlated to the PPP, which were all elevated with estrogen treatment as we expected ( Figure 2D ). It is interesting that 5-phosphoribosyl-1-pyrophosphate (PRPP) was also upregulated by estrogen. PRPP is one of the key metabolites that link glucose metabolism to nucleotide synthesis ( Figure 2E ); therefore, increased PRPP could imply that de novo nucleotide biosynthesis is stimulated by estrogen in endometrial epithelium cells. This result is consistent with Oliver’s discovery in 1972, who observed that the rate of purine de novo synthesis in an immature rat uterus was doubled at 6 h after 17β-estradiol administration (33).

Metabolically, the de novo biosynthesis of nucleotides is normally linked to promote cell proliferation, especially in cancerous cells (34, 35). As we found that estrogen may affect endometrial epithelium cell glucose metabolism via the PPP, we set out to determine whether estrogen affects cell proliferation through such glucose catabolism. First, Ishikawa cells were treated with two different doses of 6-aminonicotinamide (5 µM and 10 µM 6-AN) and no significant cell proliferation changes were observed ( Figures 3A, B , orange line vs. black line). 6-AN is a 6-phosphogluconate dehydrogenase inhibitor that turns glucose-6-phosphate into 6-phosphogluconate, and it inhibits glucose fluxing to PPP while it does not perturb glycolysis and aerobic respiration. Such results indicate that merely blocking the flux of glucose to the PPP is unlikely to affect endometrial epithelium cell proliferation. We then treated endometrial epithelium cells with both estrogen and 6-AN, and surprisingly noted a decelerated growth of endometrial epithelium cells to be even slower than the vehicle-treated cells, regardless of the volume of 6-AN that was added to the cell culture ( Figures 3A, B , blue line vs. black line). The cells treated with only estrogen achieved an accelerated growth as we expected ( Figures 3A, B red line vs. black line). Such results imply that the estrogen-promoted endometrial epithelium proliferation might be a pentose phosphate metabolism-dependent mechanism; meanwhile, turning off PPP metabolism, the endometrial epithelium cell growth would be slowed by estrogen.

It is known that estrogen can induce cell proliferation and stimulate cell growth (36–40), but there is still little evidence indicating that estrogen can alter such biological processes through metabolic pathways (41). Sun et al. showed that E2 can promote breast cancer and lymphangioleiomyomatosis tumor addiction to the PPP with upregulated G6PD enzyme activity (42). Salama et al. proposed that an E2-induced hESC proliferation would be closely linked to fluxes of glucose metabolism that upregulate aerobic glycolysis leading to such proliferation associated with PKM2 (43). Imbert-Fernandez et al. showed that E2 might also upregulate multiple glycolytic enzyme expressions and activities such as PFKFB3, resulting in increased glucose uptake and fructose 2,6-bisphosphate (F2,6BP) concentration (44). All of this evidence indicates that E2 can manipulate cellular metabolism to alter cell proliferation and other biological processes. Profiling the metabolome of estrogen-treated endometrial tissues, we believe that such hormones can alter glucose-associated metabolism in infertility patients. Glucose catabolism is one of the essential biochemical processes occurring in cells as glycolysis and the tricarboxylic acid cycle (TCA) provide energy for cell survival ( Supplemental Figure 1A ), while the PPP offers the substrates (such as nucleotides) for cell proliferation (45). As we observed that several PPP intermediates were upregulated in endometrial tissues after estrogen treatment, it can be hypothesized that estrogen stimulates the activation of pentose phosphate metabolism to promote endometrial cell proliferation. Using an endometrial epithelium immortal cell line (Ishikawa cells), we validated the promotion of cell growth function of estrogen as Ishikawa cells grew significantly faster than the vehicle-treated cells. Meanwhile, after blocking metabolism of the PPP with a chemical inhibitor (6-AN), the growth of Ishikawa cells was not affected. Surprisingly, when the endometrial epithelium cells were treated with both estrogen and 6-AN at the same time, the growth rate was slower than the vehicle controls. Such unexpected results imply that the cell growth promotion function of estrogen not only is highly dependent on activation of the PPP, but also can be regulated by the inhibition of pentose phosphate metabolism. Sun et al. proposed that the survival of estradiol-treated lymphangioleiomyomatosis xenograft mice was attenuated by the depletion of G6PD (42); Forbes et al. also observed the increased metabolic flux of the PPP in MCF-7 cells undergoing treatment with estradiol (46). Once the PPP is inhibited and glucose cannot flux to ribose and nucleotide synthesis, the estrogen treatment on endometrial epithelium diverts to preserving energy and nutrients rather than undergoing cell proliferation. Such high-dosage-induced E2-associated cell proliferation inhibition was also observed decades ago by Lewis-Wambi and Jordan in breast cancer. They found that high-dose estrogen-induced tumor regression is linked to extrinsic (Fas/FasL) and intrinsic (mitochondria) pathways (47). Srivastava et al. found that a high dosage of E2 can inhibit the differentiation of murine bone marrow monocytes and RAW 264.7 cells into mature osteoclasts (48). Researchers also found an inducible apoptosis effection of long-term high-dose E2 in cancer cells (49, 50). All this evidence implies a detrimental side to high-dosage E2 treatment.

Normally, estrogen can enter cells and bind to estrogen receptors (ESR) so that it can be delivered to the nucleus as a complex to stimulate a cascade of biochemical reactions ( Figure 4 ). It is widely known that ESR complexes can promote the expression of G6pd and the translation of glucose 6 phosphate dehydrogenase (G6PD) (51–53). G6PD is a critical enzyme in the conversion of glucose 6-phosphate to yield 6-phosphogluconate, and this biochemical reaction is also associated with one molecule of NADP⁺ being transformed to NADPH. Once the PPP is activated, de novo biosynthesis of nucleotides can largely supply the materials (deoxyribonucleic acid for DNA and ribonucleic acid for RNA) for cell proliferation ( Figure 4A ). However, when G6PD activity is inhibited with 6-AN, E2-ESR can promote the expression of G6PD without stimulating de novo biosynthesis of nucleotides. Cells might sense this signal of proliferation at an early time point, but without sufficient DNA and RNA, the cell cannot proliferate, or may even slow down the growth rate to preserve energy for cell survival ( Figure 4B ). Thus, we observed a decline in cell proliferation when 6-AN was added to the E2-treated cells ( Figure 3 ).

In our research, the Ishikawa cells we used are cancer cells; this was because it is difficult to obtain normal cells for experiments. Although they can perfectly mimic the in vitro behavior of endometrial epithelium cells (24), the experimental data obtained may have certain limitations and therefore need to be further verified by systems such as organoids in future work.