AOC1 regulates labor initiation through spermidine-induced autophagy of placental trophoblast cells via EIF5A hypusination Authors: Huaiyan Chen, Peihua Long, Zhe Wang, Ruoheng Du, Chagui Zheng, Zhuo Li, Yihuan Xu, Qiaozhen Peng, Xuesong Sui, Yanyu Sui, Xiang Jiang, Qin Li, Lu Gao Published: 2026 ABSTRACT Parturition depends on precise communication between the mother and fetus. While fetal lung signals are known to help initiate labor, the role of the placenta has remained unclear. Here we show that in steroid receptor coactivator (Src)−1 and −2 double-knockout mice, reduced placental amine oxidase, copper-containing 1 (Aoc1) leads to increased spermidine levels. In trophoblast cells, spermidine induces autophagy via hypusination of eukaryotic translation initiation factor 5 A (EIF5A), reducing estrogen and prostaglandin production. Estrogen reciprocally increases Aoc1 expression via estrogen receptor-α (ERα) in concert with SRC-1/2, forming a feedback loop maintaining placental autophagy homeostasis. AOC1 levels are elevated in preterm labor placentas from both mice and humans. Placenta-specific Aoc1 knockout dramatically delays labor by increasing trophoblast autophagy. Importantly, spermidine supplementation rescues inflammation-induced preterm labor in mice. Our findings reveal that placental AOC1–spermidine–EIF5A–autophagy axis is essential for parturition timing and offer a potential therapeutic strategy for preterm birth. Introduction Preterm birth (delivery occurring before 37 weeks of gestation) is the most common pathological complication of pregnancy and the leading cause of perinatal mortality and morbidity1. Globally, preterm birth accounts for 5–15% of all live births, representing 33% of neonatal deaths and 17.4% of deaths among children under 5 years old2,3. Preterm infants who survive face significantly greater risks of infectious diseases, hemorrhagic diseases, neurodevelopmental disorders, respiratory and gastrointestinal complications, and even adult-onset diseases than do term infants4. Despite awareness of these consequences, the incidence of preterm birth has not declined substantially over recent decades5, primarily because of an incomplete understanding of the mechanisms underlying human labor initiation6. Pregnancy involves continuous bidirectional crosstalk between the mother and fetus. Maternal physiological adaptations occur to support fetal growth and development7–9. Concurrently, the maturing fetus synthesizes and secretes factors that both regulate maternal adaptations10,11 and signal the initiation of parturition12. Studies indicate that fetal maturation—particularly activation of the hypothalamic–pituitary–adrenal (HPA) axis— increases fetal adrenal glucocorticoid production. This increase promotes fetal lung development, and crucially, the signals generated by the fetal lung can trigger labor13–15, highlighting the pivotal role of fetal-originating factors in parturition initiation. As a unique fetal organ during pregnancy, the placenta serves not only as the primary site for maternal–fetal nutrient/gas exchange but also as a critical endocrine organ that secretes hormones and factors that regulate maternal adaptation during pregnancy and even contributes to the initiation of labor. Studies have shown that corticotrophin-releasing hormone (CRH), a neuroendocrine factor specifically secreted by the human placenta, can increase the intrauterine estrogen/progesterone ratio, thus promoting uterine contraction and facilitating the onset of labor16. However, studies have also reported normal parturition timing in Crh knockout mice17, indicating that additional placental factors are involved in labor initiation. The placenta possesses robust metabolic activity and plays a supportive role in fetal growth18. It performs the dual functions of delivering nutrients to the fetus and transporting metabolic waste to the maternal circulation, thereby ensuring the proper execution of fetal developmental programming. Disruption of these regulatory programs can lead to maternal–fetal complications19. Emerging evidence suggests that the insufficient capacity of the placenta to maintain bioenergetic and metabolic homeostasis during pregnancy may contribute to spontaneous preterm birth. Consistently, significant metabolic alterations have been identified in the placentas of spontaneous preterm births20. However, how precisely these metabolic changes influence the timing of parturition remains unclear. Our previous study demonstrated that wild-type mice carrying fetuses doubly deficient in steroid receptor coactivators (Src)−1 and Src-2 (Src-1/−2d/d) manifested a delay in parturition of 1–2 days, caused by a deficiency in surfactant protein A (SP-A) and platelet-activating factor (PAF) production by the fetal lungs13. To investigate whether placental dysfunction contributes to this phenotype and identify placenta-derived parturition signals, the current study investigated alterations in gene expression in the placentas of Src-1/−2 double-knockout (dKO) fetuses. In this work, we identify a significant downregulation of amine oxidase, copper-containing 1 (Aoc1), which leads to the accumulation of its substrate, spermidine, in the placentas of Src-1/−2 dKO fetuses. This increase in spermidine induces autophagy in placental trophoblast cells specifically via the hypusination of eukaryotic translation initiation factor 5 A (EIF5A), which in turn suppresses the production of the key contraction-promoting hormones, estrogen and prostaglandin. We further demonstrate that estrogen, through estrogen receptor-α (ERα) and in concert with SRC-1/−2, forms a positive feedback loop by transcriptionally upregulating Aoc1, thereby maintaining placental homeostasis. The physiological relevance of this pathway is supported by the elevated AOC1 expression in placentas from both human and mouse preterm labor cases. Conversely, placenta-specific Aoc1 knockdown or knockout in vivo causes a significant delay in labor, confirming its causal role. Finally, we reveal that spermidine supplementation can rescue lipopolysaccharide (LPS)-induced preterm labor in mice, highlighting the therapeutic potential of targeting this pathway. Our findings establish the placental AOC1-spermidine-EIF5A hypusination-autophagy axis as a critical determinant of parturition initiation. Results Aoc1 is a core differentially expressed gene in Src-1/−2 dKO placentas We performed RNA-sequencing analysis on placentas from wild-type (WT), Src-1 single knockout (Src-1 KO), Src-2 single knockout (Src-2 KO), and Src-1/Src-2 double knockout (Src-1/−2 dKO) fetuses at 18.5 days post-coitum (dpc). The genotypes of all fetuses were confirmed by PCR analysis of tail DNA, and placental tissues from confirmed WT, Src-1 KO, Src-2 KO, and Src-1/−2 dKO fetuses were collected for subsequent qPCR and western blotting analyses (Supplementary Fig. 1a–d). Significantly differentially expressed genes (DEGs) were identified in the Src-1 KO versus WT, Src-2 KO versus WT, Src-1/−2 dKO versus Src-1 KO, Src-1/−2 dKO versus Src-2 KO, and Src-1/−2 dKO versus WT comparisons (Fig. 1a, b). The top 50 genes with the greatest fold changes (padj <0.05) between the Src-1/-2 dKO and WT placentas are listed in the heatmap (Fig. 1c). We performed gene ontology (GO) analysis on all the DEGs between Src-1/−2 dKO and WT mice, and the top ten molecular function pathways that were significantly altered are listed in Fig. 1d. Among these pathways, half were closely related to amine oxidase activity. The DEGs included in these pathways were diamine oxidase (Aoc1, also known as Dao or Abp1), diamine oxidase like protein 1 (Doxl1), diamine oxidase like protein 2 (Doxl2), and Gm19172, all of which fell within the top 50 highly variable genes (Fig. 1c). Gm19172 is a pseudogene that is minimally expressed in the placenta (Supplementary Table 1). The Doxl1 and Doxl2 genes were lost during the process of evolution and are present in the mouse genome but not in the human genome21. Only Aoc1 is conservatively expressed in rats, mice, and humans21. Since delayed labor occurred only when the fetuses were deficient in both Src-1 and Src−2, we focused on the DEGs in the placentas of Src-1/−2 dKO mice compared with those in WT mice, especially the 25 DEGs common to Src-1/−2 dKO compared with those in WT, Src-1/-2 dKO compared with those in Src-1 KO, and Src-1/−2 dKO compared with those in Src-2 KO, indicating that these DEGs are caused by a synergistic effect of Src-1 and Src−2 double deficiency in addition to Src-1 or Src−2 single knockout (Fig. 1e). Interestingly, we identified Aoc1 as the sole overlapping gene between the GO‑enriched amine metabolism–related genes (Fig. 1d) and the 25 common DEGs (Fig. 1e), suggesting that the Aoc1-encoded protein diamine oxidase (AOC1) of the placenta may play an important role in controlling parturition timing.Fig. 1RNA expression profile and GO analysis of the placentas from WT, Src-1 KO, Src−2 KO and Src-1/-2 dKO fetuses at 18.5 dpc.a Histogram showing differentially expressed gene counts between the comparison groups based on RNA-seq. The screening criteria included a |log2FC | ≥ 1 and a q (adjusted P) value < 0.05. b Venn diagram showing the number of DEGs between each group. c Heatmaps of the top 50 genes whose expression differed between Src-1/−2 dKO and WT placentas. d Horizontal bar chart demonstrating the top ten GO terms related to molecular functions in which the DEGs between the Src-1/−2 dKO and WT mice were enriched. The numbers of DEGs involved in each molecular function are listed alongside the corresponding bar. e Heatmap of the 25 common DEGs across three comparisons: Src-1/−2 dKO vs WT, vs Src-1 KO, and vs Src−2 KO. dpc: days post-coitum. WT: wild type, Src-1 KO: Src-1−/−/−2+/+, Src-2 KO: Src-1+/+/−2−/−, Src-1/−2 dKO: Src-1−/−/−2−/−. Source data are provided as a Source Data file. RNA expression profile and GO analysis of the placentas from WT, Src-1 KO, Src−2 KO and Src-1/-2 dKO fetuses at 18.5 dpc. a Histogram showing differentially expressed gene counts between the comparison groups based on RNA-seq. The screening criteria included a |log2FC | ≥ 1 and a q (adjusted P) value < 0.05. b Venn diagram showing the number of DEGs between each group. c Heatmaps of the top 50 genes whose expression differed between Src-1/−2 dKO and WT placentas. d Horizontal bar chart demonstrating the top ten GO terms related to molecular functions in which the DEGs between the Src-1/−2 dKO and WT mice were enriched. The numbers of DEGs involved in each molecular function are listed alongside the corresponding bar. e Heatmap of the 25 common DEGs across three comparisons: Src-1/−2 dKO vs WT, vs Src-1 KO, and vs Src−2 KO. dpc: days post-coitum. WT: wild type, Src-1 KO: Src-1−/−/−2+/+, Src-2 KO: Src-1+/+/−2−/−, Src-1/−2 dKO: Src-1−/−/−2−/−. Source data are provided as a Source Data file. The expression of polyamine-metabolizing enzymes is significantly decreased in the placentas of Src-1/−2 dKO fetal mice Next, we validated the RNA-seq findings by quantifying the mRNA expression of Doxl1, Doxl2 and Aoc1 via RT–qPCR and AOC1 protein expression via western blotting in the placentas of WT and Src-1/-2 dKO fetal mice. Consistent with the RNA-seq data, the Doxl1, Doxl2, and Aoc1 mRNA levels were significantly lower in the Src-1/−2 dKO placentas than in the WT control placentas (Fig. 2a–c). Similarly, AOC1 protein expression decreased significantly in the Src-1/−2 dKO placentas (Fig. 2d). Immunohistochemical (IHC) analysis further confirmed the decrease in AOC1 protein expression at 18.5 dpc across placental compartments—including the labyrinth, junctional zone, and decidua—in Src-1/−2 dKO fetuses compared with WT fetuses (Fig. 2e–g).Fig. 2Validation of DEG mRNA and protein expression in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc.a–c mRNA levels of Doxl1 (a), Doxl2 (b) and Aoc1 (c) in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. n = 8 biological replicates of the WT (a–c), n = 8 biological replicates of the dKO (a), and n = 7 biological replicates of the dKO (b,c). Two-sided nonparametric Mann‒Whitney tests were used to analyze the data. d Representative immunoblots of AOC1 protein expression in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. Densitometric quantitation of immunoblot scans (relative to ACTB as a loading control) is also shown as a bar graph; n = 6 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. e–g Immunohistochemical staining of AOC1 in the labyrinth (e), junctional zone (f), and decidua (g) from the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. IgG was used instead of the primary antibody as a negative control (NC). The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. Scale bars: 50 μm in (e–g). dpc days post-coitum. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Validation of DEG mRNA and protein expression in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. a–c mRNA levels of Doxl1 (a), Doxl2 (b) and Aoc1 (c) in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. n = 8 biological replicates of the WT (a–c), n = 8 biological replicates of the dKO (a), and n = 7 biological replicates of the dKO (b,c). Two-sided nonparametric Mann‒Whitney tests were used to analyze the data. d Representative immunoblots of AOC1 protein expression in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. Densitometric quantitation of immunoblot scans (relative to ACTB as a loading control) is also shown as a bar graph; n = 6 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. e–g Immunohistochemical staining of AOC1 in the labyrinth (e), junctional zone (f), and decidua (g) from the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. IgG was used instead of the primary antibody as a negative control (NC). The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. Scale bars: 50 μm in (e–g). dpc days post-coitum. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Dynamic changes in AOC1 expression in the placentas of WT mice during late gestation AOC1 protein levels were significantly greater in the placentas of WT fetuses at labor (approximately 19.5 dpc) than at 15.5 dpc (Fig. 3a). From 16.5 dpc until 18.5 dpc, the protein expression of AOC1 increased significantly with the progression of pregnancy and remained high until parturition (Fig. 3a). However, these alterations were absent in the Src-1/−2 dKO mice (Supplementary Fig. 2). Additionally, IHC staining revealed temporal increases in AOC1 abundance in WT mouse placentas from 15.5 dpc to labor, as observed in the labyrinth, junctional zone and decidua (Fig. 3b–d).Fig. 3Dynamic changes in AOC1 expression in the placentas of WT mice during late gestation.a Representative immunoblots of gestational changes in AOC1 protein expression in the placentas of WT fetuses at 15.5 dpc. Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 4 biological replicates for each gestational age. Two-sided unpaired t tests or nonparametric Mann‒Whitney tests, as appropriate, were used to analyze the data. b–d Immunohistochemical staining of AOC1+ in the labyrinth (b), junctional zone (c), and decidua (d) of placentas from WT mice at 15.5 dpc, 17.5 dpc, and labor. IgG was used instead of the primary antibody in the NC group. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. Scale bars: 50 μm in (b–d). dpc: days post-coitum. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Dynamic changes in AOC1 expression in the placentas of WT mice during late gestation. a Representative immunoblots of gestational changes in AOC1 protein expression in the placentas of WT fetuses at 15.5 dpc. Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 4 biological replicates for each gestational age. Two-sided unpaired t tests or nonparametric Mann‒Whitney tests, as appropriate, were used to analyze the data. b–d Immunohistochemical staining of AOC1+ in the labyrinth (b), junctional zone (c), and decidua (d) of placentas from WT mice at 15.5 dpc, 17.5 dpc, and labor. IgG was used instead of the primary antibody in the NC group. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. Scale bars: 50 μm in (b–d). dpc: days post-coitum. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Changes in the levels of polyamine metabolites in the placentas of Src-1/-2 dKO mice Polyamines, a class of small aliphatic amines widely distributed in prokaryotic and eukaryotic tissues and cells, are primarily represented by spermine, spermidine and putrescine. The main cellular source of polyamines is ornithine. Ornithine is converted to putrescine by ornithine decarboxylase (ODC), the rate-limiting enzyme of polyamine synthesis. Putrescine is subsequently converted to spermidine under the action of spermidine synthase (SRM), and spermidine is then converted to spermine by spermine synthase (SMS)22. Polyamines are primarily degraded by amine oxidase. Specifically, putrescine and spermidine can be metabolized into aldehydes and other products by the diamine oxidase AOC123. As previously mentioned, AOC1 expression increased in the placenta during late gestation under physiological conditions but decreased in the placentas of Src-1/-2 dKO fetuses. We therefore investigated whether the expression of enzymes involved in polyamine synthesis was altered in Src-1/−2 dKO placentas. We found that both the mRNA and protein levels of Odc were upregulated in the Src-1/-2 dKO placentas (Fig. 4a, b). In contrast, Srm expression was unchanged (Supplementary Fig. 1e), and Sms abundance was undetectable. These findings suggest that polyamine anabolism is significantly upregulated in Src-1/−2 dKO placentas.Fig. 4Changes in the levels of polyamine metabolites in the placentas of Src-1/−2 dKO mice.a mRNA expression of ornithine decarboxylase (Odc) in the placentas of the Src-1/−2 dKO mice at 18.5 dpc. n = 8 biological replicates in each group. Two-sided nonparametric Mann‒Whitney tests were used to analyze the data. The data are shown as the mean ± s.e.m. b Representative immunoblots of ornithine decarboxylase (ODC) protein levels in the placentas of Src-1/−2 dKO mice at 18.5 dpc. Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 3 biological replicates in each group. One-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. c-h Histograms showing the contents of spermidine (c), putrescine (d), spermine (e), citrulline (f), ornithine (g) and arginine (h) in the placentas of WT mice and Src-1/−2 dKO mice at 18.5 dpc. n = 5 biological replicates for each group. Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. Data are shown as box-and-whisker with median (middle line), 25–75th percentiles (box), and min–max values (whiskers). ns: not significant. i Immunohistochemical staining for spermidine (SPD) in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. IgG was used instead of the primary antibody as a negative control (NC). Scale bars: 20 μm. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Changes in the levels of polyamine metabolites in the placentas of Src-1/−2 dKO mice. a mRNA expression of ornithine decarboxylase (Odc) in the placentas of the Src-1/−2 dKO mice at 18.5 dpc. n = 8 biological replicates in each group. Two-sided nonparametric Mann‒Whitney tests were used to analyze the data. The data are shown as the mean ± s.e.m. b Representative immunoblots of ornithine decarboxylase (ODC) protein levels in the placentas of Src-1/−2 dKO mice at 18.5 dpc. Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 3 biological replicates in each group. One-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. c-h Histograms showing the contents of spermidine (c), putrescine (d), spermine (e), citrulline (f), ornithine (g) and arginine (h) in the placentas of WT mice and Src-1/−2 dKO mice at 18.5 dpc. n = 5 biological replicates for each group. Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. Data are shown as box-and-whisker with median (middle line), 25–75th percentiles (box), and min–max values (whiskers). ns: not significant. i Immunohistochemical staining for spermidine (SPD) in the placentas of WT and Src-1/−2 dKO mice at 18.5 dpc. IgG was used instead of the primary antibody as a negative control (NC). Scale bars: 20 μm. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. To compare the levels of polyamine metabolites in the placentas of the Src-1/−2 dKO and WT mice, ultrahigh-performance liquid chromatography–mass spectrometry (UPLC–MS) was used. The spermidine concentration was significantly greater in the Src-1/−2 dKO placenta than in the WT control placenta (Fig. 4c), whereas the concentrations of putrescine and spermine were not significantly different between the Src-1/−2 dKO and WT placentas (Fig. 4d, e). In addition to serving as the substrate for polyamine synthesis, ornithine also participates in the urea cycle24,25. Given the close relationship between polyamine synthesis and the urea cycle, we next measured the concentrations of amino acids in the urea cycle. Citrulline and ornithine levels were significantly lower in the Src-1/-2 dKO placentas than in the WT placentas (Fig. 4f, g). Conversely, the content of arginine, an intermediate metabolite in the urea cycle, did not significantly differ between the Src-1/−2 dKO placentas and the WT control placentas (Fig. 4h). These results indicate that the activity of the urea cycle is relatively suppressed, whereas the activity of the polyamine synthesis pathway is substantially activated in the Src-1/−2 dKO placentas (Supplementary Fig. 3). Furthermore, immunohistochemical (IHC) analysis revealed that the expression of spermidine (SPD) in the placenta, particularly within the cytoplasm of placental trophoblasts at 18.5 dpc, was dramatically greater in Src-1/−2 dKO fetuses than in WT fetuses (Fig. 4i), which is consistent with the UPLC–MS results. Therefore, we focused on the effects of AOC1 and spermidine on placental functions and investigated their contributions to the initiation of labor. While spermidine has been reported to be closely associated with cell senescence, SA-β-gal staining revealed no evidence of delayed placental aging or senescence in the Src-1/−2 dKO mice compared with the WT controls (Supplementary Fig. 4). Spermidine enhances placental trophoblast autophagy via hypusination-mediated EIF5A activation Different concentrations of spermidine did not affect β-gal staining in human primary trophoblast cells (Supplementary Fig. 5), further excluding its involvement in regulating placental cellular senescence. To investigate the effect of spermidine on placental autophagy, human primary trophoblast cells were treated with either DFMO (to inhibit the production of endogenous spermidine) or exogenous spermidine. Exogenous spermidine significantly increased, while DFMO significantly decreased, the expression of LC3B-II (Fig. 5a, b). Spermidine induces cellular autophagy through diverse mechanisms, such as modulating the acetyltransferase activity of E1A-associated protein p300 (EP300)25,26 and upregulating the expression of various autophagy-related transcripts, such as ATG7, ATG11 and ATG15, through the inhibition of histone acetyltransferases (HAT1)27. However, the mRNA levels of EP300, HAT1, and ATG7 were not affected by the different concentrations of spermidine in human placental trophoblasts (there are no homologous genes of ATG11 and ATG15 in humans) (Supplementary Fig. 6).Fig. 5Spermidine increases autophagy in placental trophoblast cells through hypusination-mediated EIF5A activation followed by reduced production of estrogen and prostaglandin.a Representative immunoblots of LC3B, EIF5A, TFEB and EIF5A hypusination in human primary trophoblast cells treated with DFMO, exogenous spermidine, the specific EIF5A hypusination inhibitor GC-7, EIF5A siRNA or an autophagy inhibitor b–e Densitometric quantitation of the LC3B-II (b), EIF5A hypusination (c), EIF5A (d) and TFEB (e) immunoblots relative to tubulin as a loading control. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. f Representative images of human primary trophoblast cells infected with Ad-mCherry-GFP-LC3B adenovirus after drug treatment. Under nonautophagic conditions, the mCherry-GFP-LC3B fusion protein exhibited diffuse yellow fluorescence (resulting from the combined signals of mCherry and GFP) in the cytoplasm when observed by fluorescence microscopy. During autophagy, mCherry-GFP-LC3B aggregates on autophagosome membranes, appearing as yellow punctate structures (LC3B dots or punctae). Following the fusion of autophagosomes with lysosomes, GFP fluorescence is quenched, causing these punctae to appear red. Thus, the increase in both red dots and yellow dots indicates the activation of autophagy. The boxes outline the positive signals. Scale bars: 20 μm. g–j Histograms showing the levels of IL-1β (g), IL-18 (h), E2 (i) and PGF2α (j) in the supernatant of human primary trophoblast cells treated with DFMO, exogenous spermidine, the specific EIF5A hypusination inhibitor GC-7 or EIF5A siRNA or the autophagy inhibitor 3-MA determined by ELISA. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Spermidine increases autophagy in placental trophoblast cells through hypusination-mediated EIF5A activation followed by reduced production of estrogen and prostaglandin. a Representative immunoblots of LC3B, EIF5A, TFEB and EIF5A hypusination in human primary trophoblast cells treated with DFMO, exogenous spermidine, the specific EIF5A hypusination inhibitor GC-7, EIF5A siRNA or an autophagy inhibitor b–e Densitometric quantitation of the LC3B-II (b), EIF5A hypusination (c), EIF5A (d) and TFEB (e) immunoblots relative to tubulin as a loading control. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. f Representative images of human primary trophoblast cells infected with Ad-mCherry-GFP-LC3B adenovirus after drug treatment. Under nonautophagic conditions, the mCherry-GFP-LC3B fusion protein exhibited diffuse yellow fluorescence (resulting from the combined signals of mCherry and GFP) in the cytoplasm when observed by fluorescence microscopy. During autophagy, mCherry-GFP-LC3B aggregates on autophagosome membranes, appearing as yellow punctate structures (LC3B dots or punctae). Following the fusion of autophagosomes with lysosomes, GFP fluorescence is quenched, causing these punctae to appear red. Thus, the increase in both red dots and yellow dots indicates the activation of autophagy. The boxes outline the positive signals. Scale bars: 20 μm. g–j Histograms showing the levels of IL-1β (g), IL-18 (h), E2 (i) and PGF2α (j) in the supernatant of human primary trophoblast cells treated with DFMO, exogenous spermidine, the specific EIF5A hypusination inhibitor GC-7 or EIF5A siRNA or the autophagy inhibitor 3-MA determined by ELISA. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Notably, spermidine serves as a precursor for the synthesis of the unconventional amino acid hypusine, which has only one known target of posttranslational modification, eukaryotic translation initiation factor 5 A (EIF5A)28. The hypusination level of EIF5A was significantly increased by spermidine treatment and decreased by DFMO (Fig. 5a, c), whereas total EIF5A protein expression was not affected by either exogenous or endogenous spermidine (Fig. 5a, d). Crucially, inhibiting EIF5A hypusination using the specific inhibitor GC-7 and reducing EIF5A expression via RNA interference both significantly reversed spermidine-induced LC3B-II expression (Fig. 5a, b). These results indicate that spermidine accumulation enhances autophagy in placental trophoblast cells primarily through EIF5A hypusination. Furthermore, exogenous spermidine significantly increased the expression of the core autophagy-related protein TFEB in trophoblast cells, which was reversed by either inhibiting EIF5A hypusination or reducing EIF5A expression (Fig. 5e). We further infected human primary trophoblasts with Ad-mCherry-GFP-LC3B adenovirus to monitor autophagic flux in the presence of DFMO or spermidine alone or in combination with GC-7, si-EIF5A, and the autophagy inhibitor 3-MA. DFMO treatment decreased autophagic flux, as indicated by reduced LC3B expression. In contrast, the number of LC3B punctae (yellow dots merged by mCherry and GFP) increased significantly after treatment with spermidine, and this effect was reversed by treatment with GC-7, si-EIF5A or 3-MA, as indicated by diffuse yellow cytoplasmic fluorescence (Fig. 5f). Similar effects were observed in the JEG-3 trophoblast cell line (Supplementary Fig. 7a–f). Previous studies have reported that EIF5A assists in the translation of the autophagy-related protein ATG3 in a hypusination-dependent manner29,30. However, we did not observe an increase in ATG3 protein expression in placental trophoblast cells upon spermidine treatment (Supplementary Fig. 8). To confirm the specific roles of spermidine and AOC1 in regulating autophagy in trophoblast cells, we infected human primary trophoblast cells with AOC1 shRNA lentiviral particles, lentiviral activation particles or control lentiviral particles, and the results revealed that knockdown of AOC1 promoted endogenous spermidine accumulation, accompanied by increased autophagic flux, while overexpression of AOC1 increased endogenous spermidine degradation, accompanied by suppressed autophagic flux (Supplementary Fig. 9), which is consistent with the trends observed in our previous pharmacological experiments. Exogenous spermidine reduces estrogen and prostaglandin production by enhancing trophoblast autophagy Previous studies have indicated that reduced autophagy leads to the accumulation of damaged mitochondria, generating excessive amounts of reactive oxygen species (ROS). This elevation in ROS can amplify inflammatory responses and activate the NLRP3 inflammasome, thereby promoting the formation of inflammatory corpuscles. These structures activate caspase-1 and ultimately trigger the production and release of the proinflammatory cytokines IL-1β and IL-18, which are involved in labor induction31. However, in our experiments, treatment with exogenous spermidine alone or in combination with the EIF5A hypusination inhibitor GC-7 or the autophagy inhibitor 3-MA did not alter the production of IL-1β or IL-18 in the supernatants of primary human placental trophoblast cells (Fig. 5g, h). In contrast, we observed that exogenous spermidine significantly reduced the production of both estrogen and prostaglandins, which could be reversed by GC-7, EIF5A RNA interference or the autophagy inhibitor 3-MA (Fig. 5i, j). These findings collectively suggest that spermidine-induced autophagy may inhibit labor initiation by suppressing the release of key hormones (estrogen and prostaglandins) from placental trophoblasts, which in turn can prime uterine smooth muscle contraction. Similar effects were observed in the JEG-3 trophoblast cell line (Supplementary Fig. 7g–j). E2/ERα, in concert with SRC-1 and SRC-2, regulates the transcription of Aoc1 and controls local spermidine homeostasis in the placenta SRCs enhance nuclear receptor-mediated gene transcription by interacting with nuclear receptor–steroid hormone complexes, although they do not directly bind gene promoters. Given our earlier findings in Src-knockout mice, steroids likely play a key role in regulating polyamine metabolism. As a major endocrine organ, the placenta secretes increasing levels of steroid hormones, including estrogen (E2) and progesterone (P4), throughout gestation. We therefore added E2 to the culture medium of the JEG-3 trophoblast cell line and found that E2 treatment significantly increased AOC1 protein expression, and this effect was reversed by the estrogen receptor inhibitor ICI182,780 (Fig. 6a). Treatment with progesterone did not result in significant changes in AOC1 protein expression in either the human trophoblast cell line JEG-3 or primary human trophoblast cells (Supplementary Fig. 10). To investigate the transcriptional regulation of Aoc1 by estrogen, we used the LASAGNA-Search 2.0 tool with TRANSFAC to identify five putative estrogen-responsive elements (EREs) within the proximal promoter region (-2000 to +200 bp relative to the transcription start site) of the murine Aoc1 gene (Fig. 6b). ChIP‒qPCR using placental tissue from WT and Src-1/-2 dKO fetal mice at 18.5 dpc revealed significant enrichment of endogenous ERα, SRC-1 and SRC−2 at these EREs on the Aoc1 promoter in WT placentas. This enrichment was substantially diminished in the Src-1/-2 dKO mice (Fig. 6c–e).Fig. 6E2 and ERα increase Aoc1 promoter transactivation in an SRC-1/SRC−2-dependent manner.a Effects of estrogen on AOC1 protein expression in JEG-3 cell line. Western blotting was used to detect the expression of AOC1 after treatment with estrogen (10-7 M) and its receptor blocker ICI182,780 (10-6 M). Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 3 biological replicates in ctr, n = 4 biological replicates in estrogen treatment with or without ICI182,780. Two-sided unpaired t tests were used to analyze the data. The data are presented as the mean ± s.e.m. b Consensus motifs containing putative ERE–binding sites in the mouse Aoc1 promoter between -2,000 and +200 bp relative to transcription start site. Statistical significance of the predictions is determined by LASAGNA-Search 2.0. The reported P values are provided by the tool (exact P value available on the LASAGNA-Search output page). c–e Chromatin immunoprecipitation (ChIP) was performed with WT or Src-1/−2 dKO placental tissues using antibodies against ERα (c), SRC-1 (d), or SRC−2 (e). Enrichment of these transcription factors at ERE-binding sites within the Aoc1 promoter was determined by quantitative PCR. n = 3 biological replicates for each group of (c); n = 5 biological replicates for each group of panel (d) and (e). Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. The data are presented as the mean ± s.e.m. f AOC1: Luciferase reporter constructs containing putative EREs and their corresponding mutant sequence of the human AOC1 gene. g Effects of mutation of EREs (MUT) and knockdown (KD) of SRC-1, SRC−2, or SRC-1 and SRC−2 on estrogen-induced AOC1 promoter activity. JEG-3 cells cotransfected with these constructs, ERα and Renilla luciferase were treated with estrogen (10-7 M) for 24 h and luciferase assays were performed. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. E2 and ERα increase Aoc1 promoter transactivation in an SRC-1/SRC−2-dependent manner. a Effects of estrogen on AOC1 protein expression in JEG-3 cell line. Western blotting was used to detect the expression of AOC1 after treatment with estrogen (10-7 M) and its receptor blocker ICI182,780 (10-6 M). Densitometric quantitation of immunoblots (relative to ACTB as a loading control) is also shown as a bar graph. n = 3 biological replicates in ctr, n = 4 biological replicates in estrogen treatment with or without ICI182,780. Two-sided unpaired t tests were used to analyze the data. The data are presented as the mean ± s.e.m. b Consensus motifs containing putative ERE–binding sites in the mouse Aoc1 promoter between -2,000 and +200 bp relative to transcription start site. Statistical significance of the predictions is determined by LASAGNA-Search 2.0. The reported P values are provided by the tool (exact P value available on the LASAGNA-Search output page). c–e Chromatin immunoprecipitation (ChIP) was performed with WT or Src-1/−2 dKO placental tissues using antibodies against ERα (c), SRC-1 (d), or SRC−2 (e). Enrichment of these transcription factors at ERE-binding sites within the Aoc1 promoter was determined by quantitative PCR. n = 3 biological replicates for each group of (c); n = 5 biological replicates for each group of panel (d) and (e). Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. The data are presented as the mean ± s.e.m. f AOC1: Luciferase reporter constructs containing putative EREs and their corresponding mutant sequence of the human AOC1 gene. g Effects of mutation of EREs (MUT) and knockdown (KD) of SRC-1, SRC−2, or SRC-1 and SRC−2 on estrogen-induced AOC1 promoter activity. JEG-3 cells cotransfected with these constructs, ERα and Renilla luciferase were treated with estrogen (10-7 M) for 24 h and luciferase assays were performed. n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data, and nonparametric tests were used if the variance homogeneity was not met. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Similarly, we identified four putative EREs in the proximal promoter of the human AOC1 gene (Fig. 6f). To functionally define the estrogen-responsive region(s), the AOC1 promoter containing putative EREs (WT) or their corresponding mutated (MUT) sequences was subcloned and inserted into the pGL4.12 luciferase reporter vector. These constructs, along with an ERα expression plasmid and a Renilla luciferase control, were transfected into JEG-3 cells. Compared with treatment with the empty vector (MOCK) or the mutated promoter construct, treatment with E2 (10⁻⁷ M) for 24 h significantly increased the luciferase activity driven by the WT promoter containing all four EREs (Fig. 6g). The critical role of SRC-1 and SRC−2 in mediating the effects of estrogen was demonstrated by siRNA knockdown. Depletion of endogenous SRC-1, SRC−2, or both effectively blocked the E2-induced activation of the AOC1 WT promoter (Fig. 6g). Collectively, these results suggest a reciprocal feedback loop between estrogen signaling and AOC1/spermidine levels in the placenta during pregnancy that is crucial for maintaining placental autophagy homeostasis. Disruption of this loop may impair trophoblast autophagy and hormone secretion, potentially contributing to aberrant labor timing (either preterm or delayed delivery). AOC1 expression is significantly elevated in placentas from both mouse and human preterm labor cases To further investigate the correlation between placental AOC1 expression and preterm labor, we assessed AOC1 levels in placentas from both RU486-induced and LPS-induced preterm labor mouse models and clinical preterm labor samples. Western blot analysis revealed significantly higher AOC1 expression in the placentas of both RU486-induced and LPS-induced preterm labor (PTL) mice than in those of the corresponding time-matched preterm-not-in-labor (PNIL) controls (Fig. 7a, Supplementary Fig. 11). These findings align with the observed decrease in AOC1 expression in the placentas of delayed labor mice (Fig. 2d).Fig. 7Expression of AOC1 and spermidine in mouse and human preterm placentas.a Protein expression of AOC1 in the placentas of preterm labor (PTL) mice induced by RU486. Western blotting was used to detect the protein level of AOC1. Quantitative protein analysis revealed that placental AOC1 expression was significantly greater in the RU486 group than in the time-matched preterm-not-in-labor (PNIL) group (15.5 dpc). n = 4 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. dpc: days post-coitum. b Immunohistochemical staining of AOC1 in human PNIL and TNIL placentas. Compared with that in PNIL (36+2 weeks), the expression of AOC1 in TNIL (38+6 weeks) was increased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. PNIL: preterm not in labor; TNIL: term not in labor. c Immunohistochemical staining of AOC1 in human PNIL and PTL placenta. Compared with that in the PNIL group (36+2 weeks), the expression of AOC1 in the PTL group (36 weeks) increased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. d Immunohistochemical staining of spermidine in human PNIL (36+2 weeks) and PTL (36 weeks) placentas. Compared with that in PNIL, the expression of spermidine in PTL decreased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. PNIL preterm not in labor, PTL, preterm in labor, SPD spermidine. IgG was used instead of the primary antibody as a negative control (NC). Scale bars: 50 μm in (b–d). The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Expression of AOC1 and spermidine in mouse and human preterm placentas. a Protein expression of AOC1 in the placentas of preterm labor (PTL) mice induced by RU486. Western blotting was used to detect the protein level of AOC1. Quantitative protein analysis revealed that placental AOC1 expression was significantly greater in the RU486 group than in the time-matched preterm-not-in-labor (PNIL) group (15.5 dpc). n = 4 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. dpc: days post-coitum. b Immunohistochemical staining of AOC1 in human PNIL and TNIL placentas. Compared with that in PNIL (36+2 weeks), the expression of AOC1 in TNIL (38+6 weeks) was increased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. PNIL: preterm not in labor; TNIL: term not in labor. c Immunohistochemical staining of AOC1 in human PNIL and PTL placenta. Compared with that in the PNIL group (36+2 weeks), the expression of AOC1 in the PTL group (36 weeks) increased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. d Immunohistochemical staining of spermidine in human PNIL (36+2 weeks) and PTL (36 weeks) placentas. Compared with that in PNIL, the expression of spermidine in PTL decreased. The percentage of positive area was quantified using ImageJ (IHC Profiler). n = 3 biological replicates in each group. Two-sided unpaired t tests were used to analyze the data. PNIL preterm not in labor, PTL, preterm in labor, SPD spermidine. IgG was used instead of the primary antibody as a negative control (NC). Scale bars: 50 μm in (b–d). The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Furthermore, we assessed AOC1 expression in human placentas collected from preterm labor (PTL, gestational age: 34–36 weeks), preterm-not-in-labor (PNIL, gestational age: 34–36 weeks), and term-not-in-labor (TNIL, gestational age: 38 weeks) patients (Supplementary Table 2). AOC1 expression was higher in TNIL placentas than in PNIL placentas (Fig. 7b), reinforcing the dynamic increase in placental AOC1 expression throughout gestation. Notably, AOC1 expression was significantly greater in PTL placentas than in PNIL placentas (Fig. 7c). Given the comparable gestational ages of these PNIL and PTL groups, elevated AOC1 expression is likely associated with the pathogenesis of preterm labor rather than gestational age alone. In support of these findings, IHC analysis revealed a significant decrease in spermidine levels in PTL placentas compared with those in PNIL controls (Fig. 7d). Collectively, these results demonstrate that the dysregulated expression of AOC1 in the placenta is strongly associated with preterm labor occurrence, a correlation that is conserved across both mice and humans. Specific knockdown or knockout of Aoc1 in the placenta causes delayed labor To further elucidate the causal link between AOC1 and labor initiation in vivo, we performed intraplacental injection of an adeno-associated virus (AAV) carrying either an Aoc1-targeting shRNA sequence (Prl3b1-AAV2/9-shAoc1) or a control shRNA sequence (Prl3b1-AAV2/9) at 13.0 dpc (Fig. 8a). AOC1 expression was significantly knocked down specifically in the placentas (Fig. 8b, c), whereas its expression in fetal lungs, fetal livers and myometrium was unaffected by infection with Prl3b1-AAV2/9-shAoc1 (Supplementary Fig. 12). Notably, compared with those receiving the control virus, the gestational period was substantially prolonged, and labor initiation was delayed in pregnant mice injected with Prl3b1-AAV2/9-shAoc1 (20.26 ± 0.22 dpc versus 19.51 ± 0.10 dpc) (Fig. 8d). Furthermore, IHC analysis revealed dramatic increases in spermidine expression in both the cytoplasm and nucleus of placental trophoblasts in mice injected with Prl3b1-AAV2/9-shAoc1 compared with those in control mice (Fig. 8e), confirming efficient Aoc1 knockdown in the placenta.Fig. 8Specific Aoc1 knockdown or knockout in the placenta causes delayed labor.a Diagram of DNA modification in adeno-associated virus and its injection in mice. Mouse Prl3b1-promoter sequence was added to the backbone of AAV (pAAV2/9-U6-GFP), which allows tissue-specific interference of Aoc1 in placenta. b Representative immunoblots of AOC1 expression in the placentas from the Prl3b1-AAV2/9-shAoc1 group and the Prl3b1-AAV2/9 group. n = 3. c Paraffin sections show widespread and intense expression of GFP and effective knockdown of AOC1 in the placentas of mice injected with GFP-tagged, Prl3b1-AAV2/9-shAoc1 or Prl3b1-AAV2/9 viruses. d The average gestation length of mice injected intraplacentally with Prl3b1-AAV2/9-shAoc1 or Prl3b1-AAV2/9 viruses. n = 5. e Immunohistochemical staining of spermidine in AAV-injected placental tissue. n = 3. SPD: spermidine. f Hybridization scheme of Aoc1 trophoblast-specific knockout mediated by Cyp19-cre. g The average gestation length in days of female Aoc1f/f mice mated with male Aoc1f/f mice with or without Cyp19-cre. n = 8. h Representative immunoblots of AOC1 expression in the placentas of Aoc1f/f (control) and Cyp19-cre+Aoc1f/f (cKO) mice. n = 4 in control, n = 6 in cKO. i Immunohistochemical staining of spermidine in placental tissue from control and cKO mice. n = 3. SPD: spermidine. j Immunoblots of LC3B, EIF5A, TFEB and EIF5A hypusination levels in the placentas of control and cKO mice. k Densitometric quantitation of LC3B-II, TFEB, EIF5A hypusination and EIF5A immunoblot results relative to loading control, ACTB. n = 3. ns: not significant. l Immunofluorescence of AOC1 and LC3B in control and cKO placentas. Scale bars: 50 μm (c), 20 μm (e,i), 500 μm and 20 μm (l). Densitometric quantitation of immunoblots (normalized to ACTB or TUBULIN) is also shown as a bar graph in (b) and (h). IgG was used instead of the primary antibody as a negative control and the percentage of positive area was quantified using ImageJ in (e) and (i). Two-sided unpaired t tests were used to analyze the data, and the data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Figure 8a was created in BioRender. Liu, Y. (2026) https://BioRender.com/dmkbkgb. Figure 8f was created in BioRender. Liu, Y. (2026) https://BioRender.com/l3tmhq1. Specific Aoc1 knockdown or knockout in the placenta causes delayed labor. a Diagram of DNA modification in adeno-associated virus and its injection in mice. Mouse Prl3b1-promoter sequence was added to the backbone of AAV (pAAV2/9-U6-GFP), which allows tissue-specific interference of Aoc1 in placenta. b Representative immunoblots of AOC1 expression in the placentas from the Prl3b1-AAV2/9-shAoc1 group and the Prl3b1-AAV2/9 group. n = 3. c Paraffin sections show widespread and intense expression of GFP and effective knockdown of AOC1 in the placentas of mice injected with GFP-tagged, Prl3b1-AAV2/9-shAoc1 or Prl3b1-AAV2/9 viruses. d The average gestation length of mice injected intraplacentally with Prl3b1-AAV2/9-shAoc1 or Prl3b1-AAV2/9 viruses. n = 5. e Immunohistochemical staining of spermidine in AAV-injected placental tissue. n = 3. SPD: spermidine. f Hybridization scheme of Aoc1 trophoblast-specific knockout mediated by Cyp19-cre. g The average gestation length in days of female Aoc1f/f mice mated with male Aoc1f/f mice with or without Cyp19-cre. n = 8. h Representative immunoblots of AOC1 expression in the placentas of Aoc1f/f (control) and Cyp19-cre+Aoc1f/f (cKO) mice. n = 4 in control, n = 6 in cKO. i Immunohistochemical staining of spermidine in placental tissue from control and cKO mice. n = 3. SPD: spermidine. j Immunoblots of LC3B, EIF5A, TFEB and EIF5A hypusination levels in the placentas of control and cKO mice. k Densitometric quantitation of LC3B-II, TFEB, EIF5A hypusination and EIF5A immunoblot results relative to loading control, ACTB. n = 3. ns: not significant. l Immunofluorescence of AOC1 and LC3B in control and cKO placentas. Scale bars: 50 μm (c), 20 μm (e,i), 500 μm and 20 μm (l). Densitometric quantitation of immunoblots (normalized to ACTB or TUBULIN) is also shown as a bar graph in (b) and (h). IgG was used instead of the primary antibody as a negative control and the percentage of positive area was quantified using ImageJ in (e) and (i). Two-sided unpaired t tests were used to analyze the data, and the data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Figure 8a was created in BioRender. Liu, Y. (2026) https://BioRender.com/dmkbkgb. Figure 8f was created in BioRender. Liu, Y. (2026) https://BioRender.com/l3tmhq1. Next, we generated trophoblast-specific Aoc1 knockout mice by breeding Cyp19-cre mice with Aoc1flf mice (Fig. 8f). Female Aoc1flf mice mated with Cyp19-cre+ Aoc1 flf males exhibited significantly delayed labor compared with those mated with Aoc1 flf males (22.45 ± 1.17 dpc versus 20.04 ± 0.93 dpc, Fig. 8g). AOC1 expression was significantly lower in the placentas of Cyp19-cre+; Aoc1 flf (cKO) mice than in those of Aoc1 flf (control) mice (Fig. 8h), whereas spermidine levels were significantly greater in the placental trophoblasts of cKO mice (Fig. 8i). Interestingly, autophagy levels were elevated in cKO placentas, which was accompanied by increases in TFEB expression and hypusination levels of EIF5A (Fig. 8j-l), which aligned with the effects of spermidine on placental trophoblast cells. Supplementation with spermidine counteracts LPS-induced preterm labor in mice Given the significant increase in spermidine in the placentas of both Aoc1-knockdown and Aoc1-knockout mice, we further explored whether supplementation with spermidine could counteract the preterm labor phenotype in a mouse model. Pregnant C57BL/6 J mice received daily intraperitoneal injections of spermidine (2 mg/kg) or PBS (control) at 8:30 AM from 0.5 dpc to 16.5 dpc. At 17.0 dpc, the mice from both groups were subcutaneously injected with LPS (1 μg/mouse) to induce preterm labor (Fig. 9a). Strikingly, while 80% of the control mice (8/10) delivered prematurely at 17.5 dpc, spermidine-treated mice exhibited a substantially prolonged gestational age (18.75 ± 1.36 dpc vs. 17.6 ± 0.21 dpc in controls), with 50% delivering at ≥ 19.5 dpc (Fig. 9b). These results suggest that spermidine protects against LPS-induced preterm labor in vivo.Fig. 9The addition of exogenous spermidine counteracts LPS-induced preterm labor in mice.a Schematic diagram of drug administration to pregnant C57BL/6 J mice. b Percentage of remaining pregnant C57BL/6 J mice injected with or without spermidine after LPS administration. n = 10 pregnant mice per group. The log-rank (Mantel‒Cox) test was used to analyze the data. *P < 0.05, compared with the control group. c Representative immunoblots of the contraction-associated proteins CX43, OXTR, PTGFR, and COX2 in human uterine smooth muscle cells after treatment with spermidine. d Densitometric quantitation of the CX43, OXTR, PTGFR, and COX2 immunoblots relative to TUBULIN as a loading control. Quantitative protein analysis revealed that exogenous spermidine significantly decreased the expression of the contraction-associated proteins CX43, PTGFR, and COX2. n = 6 biological replicates for CX43, n = 4 biological replicates for OXTR, PTGFR and COX2. Two-sided paired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Figure 9a was created in BioRender. Liu, Y. (2026) https://BioRender.com/cdnpv8a. The addition of exogenous spermidine counteracts LPS-induced preterm labor in mice. a Schematic diagram of drug administration to pregnant C57BL/6 J mice. b Percentage of remaining pregnant C57BL/6 J mice injected with or without spermidine after LPS administration. n = 10 pregnant mice per group. The log-rank (Mantel‒Cox) test was used to analyze the data. *P < 0.05, compared with the control group. c Representative immunoblots of the contraction-associated proteins CX43, OXTR, PTGFR, and COX2 in human uterine smooth muscle cells after treatment with spermidine. d Densitometric quantitation of the CX43, OXTR, PTGFR, and COX2 immunoblots relative to TUBULIN as a loading control. Quantitative protein analysis revealed that exogenous spermidine significantly decreased the expression of the contraction-associated proteins CX43, PTGFR, and COX2. n = 6 biological replicates for CX43, n = 4 biological replicates for OXTR, PTGFR and COX2. Two-sided paired t tests were used to analyze the data. The data are shown as the mean ± s.e.m. Source data are provided as a Source Data file. Figure 9a was created in BioRender. Liu, Y. (2026) https://BioRender.com/cdnpv8a. To investigate whether spermidine directly affects uterine contraction, we treated human uterine smooth muscle cells (hTERT-USMCs) with spermidine. Treatment with spermidine significantly decreased the expression of contraction-associated proteins (CAPs), including connexin-43 (CX43), cyclooxygenase 2 (COX2), and prostaglandin F2α receptor (PTGFR). Although not statistically significant, the protein expression of the oxytocin receptor (OXTR) also showed a decreasing trend (Fig. 9c, d). These results suggest that spermidine not only indirectly modulates labor initiation by inducing autophagy and subsequently reducing the placental release of uterine smooth muscle contraction-promoting hormones (e.g., E2 and PGF2α) but also directly inhibits parturition by attenuating uterine smooth muscle cell contraction. Discussion Steroid receptor coregulators—a superfamily of transcription modulators comprising coactivators and corepressors—play critical roles in regulating steroid hormone-dependent transcription across tissues. While they lack direct DNA-binding capacity, these coregulators influence diverse physiological processes and disease states32–34, including female reproductive disorders35. Although delayed parturition in Src-1/Src−2-deficient mice has been previously attributed to fetal defects (e.g., impaired L-arginine metabolism and surfactant production in fetal lungs)13,14, the contribution of placental factors remains unexplored. Here, integrated RNA sequencing and metabolomic analysis revealed significant downregulation of the diamine oxidase gene Aoc1 and concomitant spermidine accumulation in the placentas of Src-1/−2 dKO mice. AOC1 deficiency has been implicated in high-risk pregnancies36. During pregnancy, maternal plasma AOC1 levels significantly increase, and the AOC1 concentration in maternal plasma during delivery is much greater than that in fetal plasma. However, during preterm delivery, the ratio of AOC1 content in maternal plasma to that in fetal plasma significantly decreases37, suggesting that the increase in AOC1 expression during preterm labor is largely from the fetuses other rather than from the mothers. These findings align with our observation of gestational upregulation in wild-type placentas, which is further increased in PTL placentas from both mice and humans and significantly decreased in the placentas from Src-1/-2 dKO fetuses. Critically, placenta-specific Aoc1 knockdown (via Prl3b1-AAV2/9-shAoc1) and knockout (via Cyp19-Cre; Aoc1f/f) both delayed labor, establishing a causal role for placental AOC1 in parturition initiation. Spermidine is a polyamine that has a wide range of biological functions in the body and plays important roles in the establishment of pregnancy and embryo/fetal development38–41. Inhibiting polyamine synthesis with DFMO can lead to abnormal attachment of the placental chorioallantoic membrane, a lack of a trophoblastic layer, and defects in placental labyrinthine development42, suggesting that polyamines are essential for placental development. During late pregnancy, the upregulated biosynthetic activity of polyamines in the placenta is critical for the high demand for polyamines in growing fetuses43. Our discovery of disrupted homeostasis between the urea cycle and polyamine metabolism in the placentas of Src-1/−2 dKO fetuses—characterized by a relative attenuation of the urea cycle alongside an increase in polyamine synthesis—provides corroborating evidence for placental polyamine metabolism dysregulation from a complementary perspective. Among the polyamines, putrescine has been shown to increase mTOR signaling and protein synthesis in trophoblasts44. However, the upregulation of Odc, together with unchanged Srm mRNA levels and a concomitant increase in spermidine—but not putrescine—in our system may be due to the restricted availability of its direct substrate ornithine and/or the feedback regulation of increased spermidine to rapidly induce the expression of ODC antizyme (OAZ), which binds to ODC and leads to a decrease in ODC enzyme activity45. This potential feedback loop may limit the excessive accumulation of putrescine, representing a finely tuned dynamic balance among polyamines. However, the specific mechanism underlying this contradictory phenomenon warrants future investigation. Spermidine markedly extends the lifespan of yeast, flies, worms, mice, and humans, which is mediated by the induction of autophagy27,46. However, the conserved role of spermidine in autophagy induction in placental contexts and its involvement in labor initiation remain unexplored. In the current study, we demonstrate that spermidine, either endogenous or exogenous, induces placental trophoblast autophagy. Autophagy is a major mechanism of cellular adaptation to stress and is the most important pathway for the turnover of cytoplasmic structures, including whole organelles, thus facilitating the rejuvenation of important cell structures47–49. Notably, we did not observe delayed aging or senescence in the placentas of Src-1/−2 dKO mice or in trophoblast cells treated with spermidine, suggesting that spermidine-induced autophagy may have biological functions other than affecting trophoblast senescence. We found that spermidine-triggered autophagy reduced placental estrogen and prostaglandin production, revealing a mechanism for modulating labor-triggering hormones. Mechanistically, spermidine-induced autophagy in trophoblasts depends specifically on the hypusination of eukaryotic translation initiation factor 5 A (EIF5A), a unique modification derived from spermidine in nature28. The hypusination of EIF5A is generated from the butylamine group on spermidine and the lysine residue conserved in the nitrogen terminal domain of EIF5A50,51. EIF5A is widely present in eukaryotes and archaea and plays an important role in connecting the translation initiation and extension stages52. Numerous experiments have shown that EIF5A is associated with cell proliferation and apoptosis. Inhibiting EIF5A in the early-gestational trophoblast cell line HTR8 can reduce cell proliferation, migration, and invasion ability while increasing apoptosis53,54. While spermidine can inhibit EP300 to promote autophagy, the transcript levels of canonical autophagy genes (e.g., ATG7) remained unchanged in our models. Instead, hypusinated EIF5A upregulated TFEB, which is consistent with reports linking EIF5A hypusination to TFEB-mediated autophagy in B cells55. This pathway diverges from the role of EIF5A in ATG3 translation29,30, warranting further investigation into the autophagy targets of spermidine. Various studies, mostly in vitro, have revealed that spermidine strongly relaxes spontaneous contraction and agonist-induced contraction of vascular smooth muscle56, airway smooth muscle57,58, gastric smooth muscle59,60, and uterine smooth muscle61. These findings are validated by our in vivo experiments in which the intraperitoneal injection of spermidine significantly counteracted the preterm labor induced by LPS in a mouse model through its relaxant effects on uterine smooth muscle mediated by the inhibition of contraction-associated proteins, such as CX43, COX-2 and PTGFR. Interestingly, our results indicate that spermidine can also affect the contraction of uterine smooth muscle by inducing autophagy in the placenta and subsequently decreasing the release of estrogen and prostaglandins, which are important contraction-primed hormones. Taken together, the results of our study suggest that spermidine has synergistic relaxing effects on the contraction of uterine smooth muscle through the dual mechanisms of indirect placental signaling and direct myometrial action, indicating that spermidine is a potential tocolytic both in vitro and in vivo. Previous studies have shown that steroid hormones can regulate the expression of the uterine ornithine decarboxylase ODC and adenosylmethionine dehydrogenase AMD1 (which catalyzes methionine to become an amino donor for polyamine synthesis)62,63, suggesting that polyamine metabolism is regulated by steroid hormones. In mouse decidua, estrogen has been observed to stimulate Aoc1 transcription mediated by the transcription factor CCAAT/enhancer-binding protein-beta (C/EBPβ)64. Similarly, in the present study, we found that estrogen treatment increased Aoc1 transcription in placental trophoblasts via ERα and required SRC-1/SRC−2 recruitment to Aoc1 promoter EREs. Previous work from our group has shown that pregnant Src-1/−2-deficient females or WT females bred to Src-1/-2-deficient males failed to exhibit a decrease in serum progesterone levels at term because of delayed luteolysis, which partially contributed to the delayed labor of these females13. Furthermore, progesterone treatment did not result in significant changes in AOC1 protein expression in either the human trophoblast cell line JEG-3 or primary trophoblast cells, and spermidine did not result in significant changes in progesterone production in either JEG-3 or primary trophoblast cells (Supplementary Fig. 13), supporting our findings that estrogen/ERα and Aoc1 form a specific feedback loop in placental trophoblasts. Although the level of circulating progesterone remains high, it contributes to delayed labor primarily through systemic mechanisms rather than via local regulation of Aoc1 in the placenta. Interestingly, our recent studies revealed that the expression of C/EBPβ is reduced in the lungs of Src-1/-2 dKO fetuses15 and that glucocorticoid (GC)-induced C/EBPβ expression is essential for the transcriptional activation of Arg1 in an SRC-1/SRC−2-dependent manner14. Whether the expression of C/EBPβ is also reduced in the placentas of Src-1/-2 dKO mice and contributes to Aoc1 dysregulation merits further investigation. In the current study, we found that the expression of a key enzyme involved in polyamine metabolism, AOC1, was developmentally upregulated in the placenta during late gestation, limiting spermidine accumulation. In Src-1/−2 dKO mice, AOC1 deficiency increased the level of spermidine, which had dual anticontractile effects. Spermidine induced autophagy in placental trophoblast cells through the hypusination modification of EIF5A and led to decreases in estrogen and prostaglandin levels. On the other hand, spermidine directly suppressed the expression of CAPs in uterine smooth muscle. Overall, the synergistic effects of spermidine inhibited the contraction of the myometrium, demonstrating its potential for the prevention of preterm labor (Fig. 10).Fig. 10Schematic diagram of the regulation of Aoc1 and its role in placental endocrine activity and parturition.E2 and ERα increase the transcription of Aoc1 in an SRC-1/SRC−2-dependent manner by binding to ERE-binding site elements in the Aoc1 promoter. Deficiency of Aoc1 in the placenta causes the accumulation of the Aoc1 substrate spermidine and induces autophagy in trophoblast cells, leading to decreases in estrogen and prostaglandin levels. In an endocrine or paracrine manner, spermidine decreases the expression of CAPs in uterine smooth muscle cells. Overall, the synergistic effects of spermidine inhibit the contraction of the myometrium, leading to delayed labor in mice. Figure 10 was created in BioRender. Liu, Y. (2026) https://BioRender.com/5dxu27v. Schematic diagram of the regulation of Aoc1 and its role in placental endocrine activity and parturition. E2 and ERα increase the transcription of Aoc1 in an SRC-1/SRC−2-dependent manner by binding to ERE-binding site elements in the Aoc1 promoter. Deficiency of Aoc1 in the placenta causes the accumulation of the Aoc1 substrate spermidine and induces autophagy in trophoblast cells, leading to decreases in estrogen and prostaglandin levels. In an endocrine or paracrine manner, spermidine decreases the expression of CAPs in uterine smooth muscle cells. Overall, the synergistic effects of spermidine inhibit the contraction of the myometrium, leading to delayed labor in mice. Figure 10 was created in BioRender. Liu, Y. (2026) https://BioRender.com/5dxu27v. Methods Mice and drug treatment All animal procedures were approved by the Animal Care and Use Committee of Naval Medical University, Shanghai (ethics approval number: 20210308117). All animal care and treatments were carried out in compliance with Naval Medical University Animal Care. The mice were housed under a 12-h-dark/12-h-light cycle, a temperature of 23 °C–25 °C, and a relative humidity of 50–60%, were pathogen free and allowed free access to standard pellet chow. All the mice used in this study were on a C57BL/6 J background, and all the genetically modified mice were constructed and maintained on this background. Since this study focused on the timing of pregnancy and delivery, female animals were primarily used, although male animals were also used for mating. However, the sex of the delivered pups was not recorded, and no sex-specific analyses were conducted. The mice were used at 8–12 weeks of age for mating. The specific protocols for each animal experiment were as follows: Steroid receptor coactivator-1/−2 (Src-1/−2)-deficient mice Src-1-/- mice and Src-2+/- mice were obtained from our in-house stock. Src-1+/-/Src−2+/- mice were generated by crossing Src-1-/- mice with Src−2+/- mice. Src-1+/-/Src−2+/- males and females were caged at 6:00 pm at a ratio of 1:2 and separated at 8:00 am the following day; gestational age was considered 0.5 dpc when a vaginal plug was observed. Placenta (each placenta was divided into two parts, half of which were used to extract total RNA, and the other half were used to extract protein) and the corresponding amniotic fluid and uterus were collected at 18.5 dpc from timed-pregnant mice that were euthanized by cervical dislocation. Some of the placentas were left intact, fixed in 4% paraformaldehyde for 24 h and embedded in paraffin after dehydration for subsequent immunohistochemistry analysis. The remaining tissues were frozen immediately in liquid nitrogen and then stored at -80 °C until analysis. Tails of fetuses from Src-deficient females were stored at -20 °C for subsequent DNA isolation and genotyping. For genotyping, DNA was extracted from the tails of the mice using a Mouse Tail Kit (Foregene, Chengdu, China) and amplified by PCR. The sequences of primers used for genotyping are listed in Supplementary Table 3. PCR was performed using 2× Taq DNA polymerase (Vazyme Biotech, P112-AA) under the following conditions: initial denaturation at 95 °C for 5 min; 35 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s; and a final extension at 72 °C for 5 min. PCR products were size fractionated on 2% agarose gels and visualized with ethidium bromide. The expected band sizes were 300 bp for the wild-type Src-1 allele, 680 bp for the knockout Src-1 allele, 250 bp for the wild-type Src−2 allele and 500 bp for the knockout Src-2 allele. Samples from the Src-1+/+/Src-2+/+ (WT), Src-1−/−/Src-2+/+ (Src-1 KO), Src-1+/+/Src−2−/− (Src-2 KO) and Src-1−/−/Src−2−/− (Src-1/−2 dKO) genotypes were selected for subsequent experimental studies. Sample IDs and genotype information for RNA sequencing are detailed in Supplementary Table 4. Timed-pregnant wild-type mice C57BL/6 J wild-type (WT) mice were purchased from Shanghai Bikai Corporation (Shanghai, China). WT females were mated with WT males, and the gestational age was recorded as described above. Placentas were harvested from timed-pregnant mice at approximately 10:00 am at 15.5 dpc, 16.5 dpc, 17.5 dpc, 18.5 dpc, and labor (the birth of the first pup). Sample processing and preservation were performed as described above. Construction of the RU486-induced premature mouse model RU486 (approximately 3 mg/kg body weight) was subcutaneously injected into WT mice at 15.0 dpc, and the control group was injected with the same volume of PBS. Delivery was monitored for the two groups. Placentas were collected during labor (birth of first pup) from RU486-injected mice and control mice, frozen in liquid nitrogen, and stored at −80 °C for subsequent experiments. Generation of conditional knockout (cKO) mice Aoc1-floxed mice were obtained from GemPharmatech Company (Nanjing, Jiangsu). Aoc1-floxed mice (Aoc1f/f) carrying loxP sites flanking exons 1 − 4 of the Aoc1 gene were crossed with cre recombinase-expressing mice under the control of a trophoblast−specific promoter, Cyp1965,66. Cyp19-cre mice were donated by Dr. Haibin Wang from Xiamen University. To generate trophoblast-specific knockout mice, female Aoc1f/f mice were crossed with male Cyp19-Cre transgenic mice. Male offspring with the genotype Cyp19-Cre+, Aoc1f/+ were then bred with Aoc1f/f female mice to obtain cKO (Cyp19-Cre+, Aoc1f/f) and control (Aoc1f/f) littermates. Genotyping was performed by PCR of tail DNA using primers for loxP (F: 5’-GTCACACATGGGCCATAATCATGG-3’; R: 5’-CCTCTTACCCTACAAACTCCTTGG-3’) and Cyp19-cre (F: 5’-CCACGACCAAGTGACAGCAA-3’; R: 5’-TGACCAGAGTCATCCTTAGCG-3’). Aoc1-Floxed alleles (449 bp) and Cyp19-Cre (approximately 500 bp) were detected on 2% agarose gels and visualized with ethidium bromide. Spermidine supplementation in vivo Starting at 0.5 dpc, timed-pregnant WT mice were injected intraperitoneally (i.p.) with spermidine (Merck Millipore, Darmstadt, Germany) at a dosage of 2 mg/kg or vehicle (25 μl in PBS) at 8:30 am once a day until 16.5 dpc. Pregnant mice from both groups were injected with LPS (1 µg/mouse) subcutaneously at 19:00 on 16.5 dpc (17.0 dpc). The time of labor was defined as the delivery of the first pup. Amniotic fluid, placenta and uterus were collected during labor (the birth of the first pup) from timed-pregnant mice that were euthanized by cervical dislocation. The samples were rinsed in ice-cold 1× PBS, flash frozen in liquid nitrogen and stored at −80 °C until analysis. Collection of clinical samples Human placental tissue samples were collected from pregnant women who underwent delivery at Shanghai First Maternity and Infant Hospital and Shanghai Changhai Hospital (affiliated with Naval Medical University). Written informed consent was obtained from all participants, and the study was approved by the Ethics Committees of Medicine of Naval Medical University (ethics approval number: 20210308117) and the Ethics Committee of Shanghai First Maternity and Infant Hospital (ethical approval licenses no. KS21166 and KS22305). The information of the enrolled patients is listed in Supplementary Table 2. Clinical placentas were divided into three groups according to pregnancy outcomes: preterm not in labor (PNIL), preterm labor (PTL), term not in labor (TNIL) and term in labor (TIL). Pregnant women with infection, preeclampsia, gestational diabetes and other pregnancy-related diseases were excluded. The preterm delivery group was not given drug intervention before sampling. At the same time, the clinical data of the patients were recorded. All participants were female; sex was determined on the basis of biological sex recorded in hospital medical records. The ages of the participants ranged from 30 to 34 years. No participant compensation was provided. The number of participants per experiment is indicated in the corresponding figure legend. Given that this study focused on pregnancy and placental biology, only female participants were included; therefore, sex-disaggregated analysis was not applicable. This study did not include gender as a social construct variable, as only biological sex was relevant to placental tissue sampling. After delivery of the placenta, a portion of the placental villous tissue was excised using sterile instruments, and the remaining tissue samples were completely destroyed after sampling and were not preserved. The obtained placental samples were rinsed with precooled normal saline at least 3 times. The red trophoblast in the middle of the placenta was cut into small pieces of corn grain size with scissors. Term placental tissues were used for primary cell isolation, and the remaining tissues or tissues from the other groups were fixed in 4% paraformaldehyde for 24 h and embedded in paraffin after dehydration for subsequent immunohistochemistry analysis. RNA sequencing Total RNA was extracted from the placentas of WT, Src-1-/- (Src-1-KO), Src−2-/- (Src-2-KO), and Src-1-/-/Src-2-/- (dKO) fetuses (four samples in each group) at 18.5 dpc. After RNA purity and integrity were assessed, 2 μg of total RNA per sample was extracted and used for cDNA library construction. Transcriptome sequencing was performed on an Illumina NovaSeq Control Software (v1.7). After low-quality sequences and contaminants were removed from the 150-bp paired-end (PE) raw reads, the high-quality sequences (clean reads) were subsequently analyzed. In particular, HISAT2 (v2.2.1) was used for RNA-seq alignment, featureCounts (v2.0.1) were used for gene quantification, and differential gene expression was identified between sample groups by applying the R (v4.2.1) package DESeq2 (v1.34.0). A gene was considered a DEG when it met the following criteria: a twofold cutoff change (log2-fold change ≥ 1) in transcript levels and a q (adjusted P) value < 0.05. GO analysis and KEGG datasets were used to calculate the significance of the enrichment of DEGs in each pathway. Reverse transcription quantitative PCR Total RNA was extracted from the placentas using TRIzol. Total RNA was quantified by a microplate reader (Biotek, USA). Afterward, 0.6 μg of total RNA was reverse transcribed using a PrimeScript RT reagent kit with gDNA Eraser (Takara, Japan). Gene expression levels were quantified by a CFX Connect Real-Time System (Serial No. BR003353 Bio-Rad, USA) using Thunderbird qPCR Mix (Toyobo, Taiwan, China). Amplification of the housekeeping gene β-actin was performed for sample loading correction and normalization. The cycling conditions were 95 °C for 1 min, followed by 49 cycles of 95 °C for 15 s and 60 °C for 1 min. The specificity of the primers was verified by examining the melting curve. Distilled water was used in place of cDNA as a negative control for all the reactions. Each sample was amplified in triplicate, and the mean Cq value was calculated. The abundance of mRNA was quantified using the 2-ΔΔCt method. The primer sequences are listed in Supplementary Table 5. All PCR primers were synthesized by Sangon Biotech (Shanghai, China). Western blotting Approximately 50 mg of tissue or 1×106 trophoblast cells were homogenized in cold RIPA lysis buffer (Solarbio, Beijing, China) supplemented with protease inhibitor (Abcam, UK) and PMSF (1 mM) to extract total protein. The tissue homogenates were sonicated in an ice bath and boiled at 99 °C for 10 min. The protein samples (30μg) were separated by 10% or 12.5% SDS‒PAGE and subsequently transferred to PVDF membranes (Merck Millipore, Germany). After blocking in 5% nonfat milk for 2 h, the membranes were incubated under constant shaking overnight at 4 °C with specific primary antibodies, including AOC1 (Abcam, ab231558, 1:200), ornithine decarboxylase (ODC, Santa Cruz Biotechnology, sc-390366, 1:1,000), LC3B (Abcam, ab192890, 1:2,000), EIF5A (Santa Cruz Biotechnology, sc-390202, 1:1,000), hypusine (AtaGenix, YP587013, 1:1,000), TFEB (Beyotime, AF8130, 1:2000), COX2 (Abcam, ab15191, 1:1000), PTGFR (Thermo Scientific, PA5-110237, 1:1,000), CX43 (Abcam, ab11370, 1:2,000), OXTR (Abcam, ab181077, 1:2000), TUBULIN (Abcam, ab7291, 1:1000) and β-ACTIN (Abcam, ab8227, 1:2000). Immunoreactive bands were visualized using Peroxidase-AffiniPure goat anti-mouse IgG (115-035-020; 1:5000; Jackson, USA) or Peroxidase-AffiniPure goat anti-rabbit IgG (111-035-045; 1:5000; Jackson) and ECL solution (Thermo Scientific), and the bands were subsequently analyzed using a Tanon chemiluminescent imaging system (version 5200 multi). The immunoblots were quantified using ImageJ software (v1.4.3.67). Relative protein expression levels were normalized to the corresponding β-actin or tubulin expression level as a loading control. Immunohistochemistry Placenta tissues were fixed in 4% paraformaldehyde for 24 h and embedded in paraffin after dehydration. Sections (5 μm) were deparaffinized, rehydrated, and boiled for 15 min in 10 mM citrate buffer (pH 6). Nonspecific background staining was blocked by incubation with 3% hydrogen peroxide (Sigma, USA) at room temperature for 25 min in the dark. The sections were blocked in 3% bovine serum albumin (BSA) for 30 min and then incubated at 4 °C overnight with primary antibodies against AOC1 (Abcam, ab231558, 1:50), spermidine (Abcam, ab7318, 1:50), or IgG (negative control). The following day, the sections were incubated with a horseradish peroxidase-conjugated secondary antibody (GK500705; GeneTech, Shanghai, China) at room temperature for 40 min. The bound antibodies were detected by adding one drop of DAB chromogenic agent to 1 ml of DAB buffer (GeneTech). Nuclei were counterstained using a hematoxylin counterstain reagent (Servicebio, Wuhan, China). The sections were then dehydrated and analyzed using a Nikon Digital Sight microscope (Model DS-Fi2). UPLC‒MS/MS analysis of polyamine metabolites Sample information and preparation For the quantification of polyamines, a total of 10 mouse placental tissue samples were analyzed (5 Src-1/−2 dKO placentas and 5 wild-type controls). For amino acid quantification, a total of 10 mouse placental tissue samples were analyzed (5 Src-1/−2 dKO placentas and 5 wild-type controls). Each group consisted of 5 independent biological replicates (i.e., 5 individual mouse placentas). No technical replicates were included in the metabolomics analysis; each biological sample was extracted and analyzed once. All tissue samples were immediately snap-frozen in liquid nitrogen after removal from the body and stored at −80 °C until analysis. Placental tissue samples (approximately 10 mg each) were precisely weighed and placed into 2 mL EP tubes. A steel ball and 100 μL of 80% methanol were added, followed by homogenization at 60 Hz for 3 min. The homogenates were subsequently centrifuged at 15,000 ×g for 20 min at 4 °C. The supernatants were subsequently collected and diluted with 80% methanol. For the polyamine analysis, the samples were diluted 10-fold. For amino acid analysis, the samples were diluted 10-fold and 100-fold and then mixed 1:1 with an internal standard solution (200 ng/mL). All the samples were vortexed and transferred into vials for subsequent analysis. UPLC‒MS/MS Analysis Polyamines: Chromatographic separation was performed on an Agilent 1290 Infinity UHPLC system (Agilent Technologies, USA) equipped with an XBridge BEH C18 column (2.5 μm, 2.1 × 100 mm). The mobile phase consisted of 10 mM ammonium formate–0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The gradient elution program was as follows: 0–1 min, 5% B; 1–1.5 min, 95% B; and 1.5–3 min, 95% B, with a post‑time of 3 min. The flow rate was 0.4 mL/min, the injection volume was 1 μL, and the column temperature was 25 °C. Mass spectrometric detection was performed on an Agilent 6470 triple quadrupole MS/MS system (Agilent Technologies, USA) in positive ion mode. The optimized parameters were as follows: a gas temperature of 350 °C, a gas flow of 11 L/min, a nebulizer pressure of 40 psi, a sheath gas heater of 350 °C, a sheath gas flow of 11 L/min, and a capillary voltage of 4000 V. The multiple reaction monitoring (MRM) transitions were m/z 203.1 → 129.1 (fragment 95 V, CE 15 V) for spermine, m/z 146.2 → 129.1 (fragment 80 V, CE 8 V) for spermidine, and m/z 89.0 → 30.2 (fragment 80 V, CE 25 V) for putrescine. Amino acids: Chromatographic separation was performed on a SH-MADZU LC-20A system (Shimadzu Corporation, Japan) equipped with an XBridge BEH Amide column (2.5 μm, 2.1 × 100 mm). The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B), with isocratic elution of 30% A and 70% B for 5 min. The flow rate was 0.4 mL/min, the injection volume was 1 μL, and the column temperature was 25 °C. Mass spectrometric detection was performed on a SHIMADZU LCMS 8040 triple quadrupole MS/MS system (Shimadzu Corporation, Japan) in positive ion mode. The instrument parameters were as follows: nebulizing gas flow rate of 3 L/min, DL temperature of 250 °C, heat block temperature of 400 °C, and drying gas flow rate of 15 L/min. MRM transitions were m/z 176.1 → 159.1 (CE -10 V) for citrulline, m/z 175.1 → 116.1 (CE -14 V) for arginine, m/z 133.1 → 116.1 (CE -15 V) for ornithine, and m/z 181.1 → 74.1 (CE −21 V) for the internal standard. Data preprocessing Standard curves were constructed for quantification. For polyamines, mixed standard stock solutions (100 μg/mL) were serially diluted with 80% methanol to obtain working solutions of various concentrations. For amino acids, the serially diluted standards were mixed 1:1 with internal standard solution (200 ng/mL). All standard working solutions and sample solutions were analyzed under the above chromatographic and mass spectrometric conditions. Each sample solution was injected in duplicate. Peak determination and peak area integration were performed with MassHunter Workstation software (Agilent, Version B.08.00) for polyamines and LabSolutions LCMS (Shimadzu, Version 6.50) for amino acids. Standard curves were plotted with the standard concentration (X) versus the peak area (Y) for polyamines or versus the peak area ratio of the analyte to the internal standard (Y) for amino acids. Sample concentrations were calculated on the basis of the corresponding standard curves and dilution factors. Quantitative detection of amino analytes by UPLC‒MS/MS was performed using an optimized method previously reported67. Cell culture and treatment Human primary trophoblast cells were isolated as previously described68. Specifically, term placental villous tissue was minced and digested with 0.25% trypsin plus 0.02% DNase I at 37 °C for 30 min, and this process was repeated three times. The dissociated cells were filtered, subjected to density gradient centrifugation, and plated into 6-well or 24-well culture plates in phenol red-free culture medium supplemented with 10% FBS and 1% penicillin–streptomycin under standard tissue culture conditions. After 48–72 h, adherent trophoblast cells were used for subsequent experiments. Purity was verified by CK7 immunostaining. Then, 100 μM spermidine (Absin, abs42027652), 1 mM difluoromethylornithine (DFMO, Enzo Life Sciences, ALX-270-283-M050) or spd combined with 10 μM GC7 (Merck Millipore, 259545) or si-EIF5A (200 nM) was added to the trophoblast cells for 24–48 h or combined with 5 mM 3-MA (Sigma, 189490) for 2 h. Cells were collected from each treatment with at least three biological replicates for western blotting. For immunofluorescence, cells were infected with adenovirus expressing the mCherry-GFP-LC3B fusion protein (Beyotime, C3011) 24 h before drug treatment to detect autophagic flow. For estrogen treatment, human trophoblast JEG-3 cells (ATCC, HTB-36™, female origin) in 6-well culture plates were incubated with 10-7 M 17β-estradiol (Sigma, E2758) alone or in combination with 10-6 M ERα inhibitor ICI182,780 (Sigma, V900926) for 48 h, after which the cells were collected from each treatment (with the same volume of solvent as a control), and at least three biological replicates were performed for western blotting. HTERT-HSMC cells (a human smooth muscle cell line generated from primary culture of human myometrial cells (ATCC, PCS-460-011™) were infected with expression vectors containing human telomerase reverse transcriptase (hTERT) according to a previous publication69, which was kindly provided by Professor Carole Mendelson’s laboratory at the University of Texas, USA. The cells were plated into 6-well culture plates in culture medium supplemented with 10% FBS and 1% penicillin–streptomycin under standard tissue culture conditions for 24 h. Then, the cells were incubated with spermidine (Absin, abs42027652) at a final concentration of 10-6 M for 48 h. The cells were collected from each treatment, and at least three biological replicates were performed for western blotting. Prior to the experimental treatments, the cells described above were precultured overnight in phenol red-free, charcoal-stripped serum medium to deplete any residual hormone effects. ELISA The supernatant of the cell sample was centrifuged at 2–8 °C and 900 ×g for 15 minutes, and it was immediately used for the experiment or stored at −20 °C or −80 °C after separation. The assays were conducted according to the instructions of the kit. The basic steps were as follows: A total of 100 μL of the sample to be tested was added to the coated reaction well. The plate was sealed with plate sealer and incubated at 37 °C for 1–2 h. The buffer was carefully removed and the wells were washed 3–5 times. Next, 100 μL of diluted biotinylated antibody working solution was added to each well. The plate was sealed with a sealing membrane and incubated at 37 °C for 1 h. After the wells were washed, 100 μL of the diluted enzyme conjugate working solution was added to each well. The plate sealer was used to seal the plate, which was subsequently incubated at 37 °C in the dark for 30 min. The chromogenic substrate and 100 μL of TMB substrate solution were added to each well, and the samples were allowed to react in the dark at 37 °C for 10–30 min until an obvious color gradient was detected in the wells containing the serially diluted standard. To each reaction well, 100 μL of 2 M sulfuric acid was added, and the color changed from blue to yellow. Within 10 min, the O.D. of each well was measured at 450 nm on the enzyme labeling instrument after the blank control well was zeroed. The standard curve was constructed according to the concentration and O.D. of the standard, and then the sample concentration was calculated according to the standard curve equation. RNA interference To knockdown EIF5A mRNA, sequence-specific small interfering RNA (siRNA) targeting human EIF5A mRNA was used53. The sequences (si-EIF5A, sense: 5’-GGCCAUGUAAGAUCGUCGAdTdT-3’; antisense: 5’-UCGACGAUCUUACAUGGGCCdTdT-3’; si-NC, sense: 5’-UUCUCCGAACGUGUCACGUTT-3’; antisense: 5’-ACGUGACACGUUCGGAGAATT-3’) were synthesized by GenePharma (Shanghai, China). The cultured trophoblast cells were transfected with siRNA-EIF5A or siRNA-NC using LipofectamineTM 3000 (Invitrogen) for 6 h, followed by an 18 h incubation with DMEM alone. Cells were harvested and stored at -80 °C until analysis. Generation of tissue-specific shAoc1 adeno-associated virus (AAV) and placental injection The backbone of adeno-associated virus (pAAV2/9-U6-GFP) was modified by the addition of a Prl3b1-promoter sequence, which allows tissue-specific delivery into the placenta70. A modified AAV vector carrying either a control short hairpin RNA (shRNA) sequence (TTCTCCGAACGTGTCACGT, Prl3b1-AAV2/9) or an Aoc1-targeting shRNA sequence (GCCACTTTGACTCAAACTTTA, Prl3b1-AAV2/9-shAoc1) was designed and generated by Genechem (Shanghai, China). Timed-pregnant C57BL/6 J WT mice were anesthetized with isoflurane via a precision vaporizer at 13.0 dpc. The fur covering the abdominal region was shaved, and an incision was made to expose the uterine horns. Each placenta in both uterine horns was injected with 5 μl of Prl3b1-AAV2/9-shAoc1 (5×1012 VG/ml) or Prl3b1-AAV2/9 (5×1012 VG/ml) particles using microsyringes (Gaoge Tech Company, Shanghai); all the placentas from the same female were injected with the same particles. Specifically, the maternal side of each placenta near the uterine myometrium was targeted. During injection, visible blood vessels were carefully avoided, and a shallow needle angle was maintained to prevent contact with the fetus, ensuring accurate viral delivery into the placental tissue. After the operation, the uterine horns were returned to the abdominal cavity, and the incision was sutured. The mice were kept on a warming pad and returned to their cages after recovery from surgery. The mice were allowed to deliver for analysis of the parturition timing or were sacrificed at labor to harvest the amniotic fluid, placenta, fetal lung, fetal liver, and myometrium. Some of the tissues were fixed in 4% paraformaldehyde for 24 h and embedded in paraffin after dehydration for subsequent immunofluorescence analysis. To precisely determine the time to labor, beginning at 18.5 dpc, the mouse cages were monitored every half hour from 7:00 am to 12:00 pm and every hour from 12:00 pm to 6:00 am, and the delivery of the first pup was recorded as the time to labor. Immunofluorescence Placenta sections were prepared as described above. After deparaffinization and rehydration, antigen retrieval was performed by boiling in 10 mM citrate buffer (pH 6). The slides were blocked in BSA for 30 min and stained with AOC1 antibody (sc-515908; 1:100; Santa Cruz) at 4 °C overnight. The next day, Cy3 AffiniPure goat anti-mouse IgG (GB21301; 1:300; Servicebio) was added to the slides, which were subsequently incubated at room temperature for 40 min. The nuclear DNA was labeled with DAPI, and the slides were mounted with Fluoromount. The stained slides were analyzed using a Nikon fluorescence microscope (Eclipse Ti). ChIP–qPCR Placenta tissues isolated from WT and Src-1/-2 dKO fetuses at 18.5 dpc were finely minced and immediately incubated in 1% formaldehyde for 15 min at room temperature. The pelleted tissues were washed, resuspended in SDS lysis buffer supplemented with a protease inhibitor cocktail, and then sonicated. Sheared chromatin was diluted in ChIP dilution buffer with Dynabeads™ Protein G (Millipore, CS200638) and incubated with anti–ERα (Abcam, ab32063, 4 μg/25 μg chromatin), anti-SRC-1 (Bethyl Laboratories, A300-344A, 1:165), anti-SRC−2 (Bethyl Laboratories, A300-346A, 1:165) or IgG (Abcam, ab171870, 1:1,000) antibodies at 4 °C overnight. The beads were subsequently washed following the manufacturer’s instructions (Millipore, 17-20000). Crosslinking was reversed by incubation with ChIP Elution Buffer/Prok at 62 °C for 2 h with shaking. DNA was then purified using spin columns following the manufacturer’s instructions (Millipore, 17-20000). Quantitative PCR was performed using primer pairs (listed in Supplementary Table 6) to amplify putative binding sites of ERα in the promoter of Aoc1. The primers were designed carefully to ensure that each primer pair selected only one specific binding site and to avoid nonspecific amplification of the adjacent binding site. The fold binding was calculated as the expression of the ERE binding site from the chromatin DNA immunoprecipitated with a specific antibody versus the expression of the ERE binding site from the chromatin DNA immunoprecipitated with IgG. Luciferase assay Human AOC1 promoters containing putative EREs (WT) or corresponding mutant (MUT) sequences were synthesized by Sangon Biotech Company (Shanghai) and subcloned and inserted into the pGL4.12 vector (Promega) to generate luciferase reporter plasmids. Both the constructs and the mutations were confirmed by DNA sequencing by Sangon Biotech Company (Shanghai). SRC-1 and SRC-2 transcripts in JEG-3 cells were knocked down by transfection with preannealed Silencer Select siRNA duplexes (s16461 for SRC-1 and/or AM16708 for SRC-2; Thermo Fisher) using Lipofectamine RNAiMAX transfection reagent (Invitrogen). Silencer negative control #1 siRNA (AM4611; Thermo Fisher) was transfected as a negative control. Twenty-four hours after transfection, WT, SRC-1, SRC-2, or SRC-1/−2 double-knockdown JEG-3 cells were transfected with AOC1 WT or MUT-luciferase reporter constructs and ERα and Renilla luciferase plasmids using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturer’s protocol. The ERα and Renilla luciferase plasmids were obtained from our laboratory’s plasmid repository and have been described previously13. All plasmids and newly generated reagents are available upon reasonable request from the corresponding author. Estrogen (10–7 M) was added to the medium for 24 h, after which whole-cell lysates were prepared using Passive Lysis Buffer (Promega). Luciferase assays were performed using the Dual Luciferase Reporter Assay System (Promega) according to the manufacturer’s protocol in a GloMax plate-reading luminometer (Promega). Statistics and reproducibility All the results are presented as the mean ± SEM. All the statistical analyses were performed using GraphPad Prism software (version 9.5.1). A normality test was performed on all the data, and one-way ANOVA followed by Tukey’s multiple-comparison test or an unpaired t test (or paired t test according to the experimental conditions) was used if the data followed a normal distribution; if the data did not follow a normal distribution, nonparametric methods (the Kruskal–Wallis test or Mann–Whitney test) were used to analyze the data. In all of the tests, P < 0.05 was considered to indicate statistical significance. The sample sizes (n) and statistical test details are provided in the respective figure legends. For all experiments, particularly representative images (e.g., micrographs in Figs. 5f, 8c, l) were repeated independently at least 3 times with similar results. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Source Data