All of the experiments were performed using female mice with C57BL/6J strain, which were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). Mice were fed in the cages (32 cm × 16 cm × 16 cm) under controlled environmental conditions. The experimental protocols were approved by the Animal Use and Care Committee of Qingdao Agricultural University.

A group of 35-day-old female mice were selected for the experiments. First, the mice were adapted for 1 week, and during this period, self-control data including Y-maze tests (YMTs), forced swimming tests (FSTs), and sucrose preference tests (SPTs) were determined. Then, mice were divided into the control and CUMS-treated groups. Following that, the mice in the CUMS-treated group were fed for 6 weeks under the CUMS environment, according to the previous description (Xu et al., 2016; Si et al., 2018; Sun et al., 2018). After the consequent CUMS treatment, the depression-like behaviors of mice were accounted for by measuring the indexes of YMTs, FSTs, and SPTs. Mice with significant decreases in the values of SPT and the YMT above 20% of their self-control values, as well as a significant increase in values of FST above 15% of their self-control values, were defined as CUMS-induced depression-like mice. Only the depression-like mice have been used in the following research studies.

After CUMS treatment, the mice were anesthetized, and blood samples were collected in heparinized centrifuge tubes. The obtained serum samples were prepared and stored at −80°C until analysis. Serum levels of mouse E₂ and AMH were specifically measured with the enzyme-linked immunosorbent assay (ELISA) kits (Lanpai Biotechnology Co., Ltd., Shanghai, China).

Following euthanasia, the mouse ovaries were removed and dissected immediately. Then, after fixing in 4% paraformaldehyde (PFA) at 4°C overnight, the fixed ovaries were cut into sections (5 μm thickness) and mounted on glass slides, then dehydrated, and embedded in paraffin. Following deparaffinization, the slides were stained with hematoxylin and eosin (HE) for histological analysis of follicles.

In order to count the number of follicles in an ovary, the paraffin-embedded ovaries were cut serially into sections of 5 μm thickness, and then, every fifth section was mounted on slides, according to the previous description (Tilly, 2003). Based on the well-accepted standards set by Pedersen and Peters, ovarian follicles at various developmental stages were counted in collected sections of one ovary (Pedersen and Peters, 1968). In order to avoid the repetitive counting of the same follicles, only follicles containing oocytes with a visible nucleolus were counted.

First, the mice were injected with PMSG (7 IU), and then, euthanasia was carried out after 48 h. After dissecting the ovary, the large antral follicles were punctured in an M2 medium, and GCs were extruded. The GCs were harvested and washed with PBS, then mixed with TRIzol, and stored at −80°C.

Total RNA was extracted from the ovary or GCs using TRIzol, according to the manufacturer’s instructions. cDNA was synthesized using the PrimeScript RT Reagent Kit (RR037A; TaKaRa Bio Inc., Shiga, Japan). Real-time PCR was performed using a Bio-Rad CFX96 Touch system. Amplification of mRNA was performed in a 20 μl reaction with 1 μl of sample cDNA, 0.5 μl of each primer (10 nmol·l⁻¹) (Supplementary Table S1), 10 μl of 2×qPCR Mastermix (Green), and 8 μl of ddH₂O. PCR cycles were run at 94°C for 2 min, followed by 45 cycles of denaturation for 15 s at 94°C, annealing and elongation for 30 s at 60°C, and melt curve 65–95°C increase of 0.5°C for 5 s. All values were normalized to GAPDH to balance potential irregularities in the RNA concentration. All experiment runs were repeated in three replications. The 2^−ΔΔCt method was used to calculate the fold changes in gene expression.

TUNEL assays were performed with the In Situ Cell Death Detection Kit (Roche Diagnostics, Basel, Switzerland, 11684795910), according to the manufacturer’s instructions. As described by Song et al. (2015), after heating at 60°C for 2 h, the slides of ovary sections were first washed in xylene, then rehydrated through a graded series of ethanol and double distilled water, followed by treatment with proteinase K for 15 min at room temperature, finally incubated with the TUNEL reaction mixture in a humidified atmosphere for 60 min at 37°C in the dark, and counterstained with Hoechst for 5 min. The sections were observed for fluorescein labeling under a fluorescence microscope (Nikon, Tokyo, Japan).

The frozen ovaries were thawed on ice, then grinded, and resuspended using 120 µl of precooled 50% methanol buffer. The metabolites were extracted from 20 µl of each sample. After vertexing for 1 min and incubating for 10 min at room temperature, the sample was stored at −20°C overnight. After centrifugation at 4,000 g for 20 min, the supernatant was transferred to 96-well plates. Additionally, a pooled quality control (QC) sample was generated by mixing 10 µl of each extraction combination together. A TripleTOF 5600 Plus high-resolution tandem mass spectrometer (SCIEX, Warrington, United Kingdom) was used to evaluate all of the samples, which could be analyzed in both the positive and negative ion modes. In this study, an ultra-performance liquid chromatography (UPLC) system was used to perform the chromatographic separation (SCIEX, United Kingdom). The acquired LC-MS data pretreatment was performed and analyzed by LC-Bio Technology Co., Ltd. (LC-Bio, Hangzhou, China).

Sequencing library preparation and RNA-seq were conducted at LC-Bio Technology Co., Ltd. (LC-Bio, Hangzhou, China). RNA samples with high purity (OD 260/280 ≥ 2.0) and high integrity (RIN >7) were used for cDNA library construction. The cDNA library construction, mapping reads, and FPKM calculations were conducted according to a previous study (Shen et al., 2019). Differentially expressed genes defined by p < 0.05 and an absolute fold change >1.5 were used for analysis.

For upregulated and downregulated gene list, pathway and process enrichment analysis has been separately carried out using the “Metascape” Gene Annotation and Analysis Resource (https://metascape.org/gp/index.html#/main/step1). In order to collect and group terms based on their membership similarities, terms with a p-value less than 0.01, a minimum count of 3, and an enrichment factor greater than 1.5 (the enrichment factor is the ratio between the observed counts and the counts expected by chance) are collected and grouped into clusters.

All the experiments were repeated at least three times, and all the measured data were presented as the mean ± SD. Student’s t-test was adopted to analyze the comparison between control and depression-like mice with SPSS statistical software (IBM Co., NY). The differences between groups were considered statistically significant if p < 0.05.

To produce mice with depression-like behaviors, female mice aged 35 days post-partum (dpp) were treated by CUMS or without stress for 6 weeks. After CUMS treatment, depression-like behaviors were assessed by behavioral tests (SPTs, YMTs, and FSTs). Since the three behavioral parameters of the treated mice varied so much, it was observed that only the mice that changed significantly in all three tests were classified as depression-like mice (Figure 1A). As shown in Figure 1B, after 6 weeks of treatment, the SPT values of the CUMS treatment mice were significantly lower than those mentioned earlier (56.88 ± 8.54% versus 78.79 ± 5.63%, p < 0.001, and n = 14). Also, the YMT values in the CUMS treatment group were also significantly lower than those mentioned earlier (49.91 ± 9.20% versus 78.43 ± 6.13%, p < 0.001, and n = 14) (Figure 1C). Significantly decreased FSTs values were also found, and the immobile time was 245.1 ± 24.86 s after CUMS treatment, while it was 151.6 ± 45.23 s before the CUMS treatments (p < 0.001, n = 14) (Figure 1D). A significant difference was also observed in such behavioral test values between the control and CUMS mice after 6 weeks (Figures 1B–D). According to the aforementioned behavioral tests, 14 mice out of a total of 70 met the criteria for depression. We considered these 14 mice as depression-like mice (20.00%, Table 1).

A growing body of research demonstrates that psychological stress in women is linked to ovarian developmental anomalies, including follicle and oocyte malformations (Zhai et al., 2020). But the mechanisms remain unclear. In order to know the effect of CUMS treatment on ovarian developments in female mice, serum hormone levels were measured in two groups of mice. The levels of E₂ (p < 0.05, Figure 2A) and AMH (p < 0.01, Figure 2B) were significantly decreased in depression-like mice. Additionally, we found degraded oocytes were significantly piled up in depression-like mouse ovaries, suggesting more prematurely active follicles underwent atresia (Figures 2C,D, arrows). The typical morphologies of different-stage follicles including the primordial follicle, primary follicle, secondary follicle, and antral follicle were shown by H&E staining (Figure 2E). Then, total ovarian sections were conducted to compare the number of ovarian follicles, a remarkable decrease in total follicles was found (p < 0.05), and the number of second follicles in depression-like mouse ovaries was significantly lower (p < 0.01) (Figure 2F). These results suggest that mouse ovarian follicle development might be interrupted by CUMS treatment, and clinical characteristics of ovarian insufficiency in human patients were observed in depression-like mice.

To test whether the reduction in serum E₂ and ovary total follicles (especially the second follicles) in depression-like mice is caused by the apoptosis in granulosa cells, we first examined the ovary section by H&E staining. As shown in Figure 2C, degraded oocytes were significantly piled up in treated ovaries, suggesting more prematurely active follicles underwent atresia. TUNEL assays suggested that granulosa cells in the depression-like mouse ovaries did undergo more apoptosis (Figures 3A,B). For further confirmation, the expression of Bcl2 and Bax in granulosa cells, which are markers of apoptosis, was performed by real-time PCR. As shown in Figures 3C–E, in comparison with the control mice, there was a significant difference in the expression of Bcl2 and Bcl2/Bax in depression-like mice’s granulosa cells. These results indicate that the reduction in serum E₂ and the defective follicle development in depression-like mice are associated with the apoptosis in granulosa cells.

To determine the changes in the overall lipid composition and distribution in ovaries in response to CUMS treatment, a non-targeted lipidomic strategy by liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis was performed to reveal differential lipids between depression-like mice (n = 6) and controls (n = 6). We detected over 1,001 different lipid species in ovaries, consisting of 340 triacylglycerols (TGs), 129 phosphatidylcholines (PCs), 87 monogalactosyldiacylglycerols (MGDGs), 66 lysophophatidylcholines (LPCs), 63 sphingomyelins (SMs), 60 phosphatidylethanolamines (PEs), 33 diglycerides (DGs), and 223 other lipid classes (Figure 4A). The abbreviations of these quantified lipid classes are shown in Supplementary Table S2. The orthogonal projections to latent structure discriminant analysis (OPLS-DA) plot shows a clear separation of two classes (depression-like vs. control, Figure 4B), in details R2 = 0.97 and Q2 = −0.51 (Supplementary Figure S1). From the OPLS-DA model, 14 features were selected that differed between the two groups with variables of importance for the projection of VIP (variable importance in projection) > 1.0, p < 0.05, and FC (fold change) ≥ 1.5 (or FC ≤ 0.67) (Figure 4C; Supplementary Table S3). We found that CUMS treatment significantly decreased the contents of the following lipid metabolisms in the ovary, including LPC (36:3-SN1 and 34:0-SN2), TG (O-10:0_20:0_20:0 and O-20:1_16:0_18:0), campesterol ester (SE 28:1/22:4 and SE 28:1/24:4), cholesteryl ester (CE 24:4), N-acyl ornithine (NAOrn), coenzyme Q 10 (CoQ10), PC (22:6_22:6 and O-18:1_20:4), sterol sulfate (ST 27:1; O; S), and phosphatidylserine (PS 20:2_20:2) (Figure 4C). However, the increased level of TAG (8:0_8:0_10:0) in the ovaries of depression-like mice was related to fat digestion and absorption, vitamin digestion and absorption, and insulin resistance pathways (Supplementary Figure S2–4). Additionally, CE is supported by the ovarian steroidogenesis pathway (Supplementary Figure S5), and the decreased level of CE might be associated with the decreased level of E₂ in the serum (Figure 4D). These results suggest that the overall composition of lipid classes was altered in the depression-like mouse ovaries.

To further understand how CUMS treatment affects follicle development, mouse ovaries from the control and depression-like groups were collected for RNA-seq. The total number of reads generated from each sample varied from 40218254 to 52603858, which were in a unique manner, with mapping rates between 97.81 and 97.89 (Supplementary Table S4). In total, at least 24,838 genes were expressed in the control and CUMS-treated samples. Among them, 423 significantly differentially expressed genes (DEGs) and 212 upregulated and 211 downregulated DEGs (fold change, FC ≥ 1.5 or FC ≤ 0.67, p < 0.05, depression-like/control) were identified (Figure 5A; Supplementary Table S1). Pathway and process enrichment analysis was performed with these DEGs, and the results showed that the downregulated genes were enriched in gene ontology annotations of the mitotic cell cycle, regulation of cell cycle processes, and pathway annotations of the cell cycle (Figure 5B). However, the upregulated genes were annotated as being involved in striated muscle contraction, skeletal myofibril assembly, circadian rhythm, and regulation of the extrinsic apoptotic signaling pathway (Figure 5C). Notably, our results showed that 60 genes associated with the GO:0000278 mitotic cell cycle were found to be significantly decreased in the CUMS-treated ovaries (Figure 5D; Supplementary Table S5). Five genes associated with the GO:1902041 regulation of the extrinsic apoptotic signaling pathway via death domain receptors and 15 genes associated with GO:0008285 negative regulation of cell population proliferation were found to be significantly increased in the depression-like mouse ovaries (Figure 5D; Supplementary Table S6). Then, real-time PCR was employed to confirm the validity of RNA-seq analysis. Two downregulated cell cycle-associated genes (Il1b and Tgfa) and three upregulated cell population proliferation-associated genes (Fgf2, Cdkn1a, and Serpine 1) were selected to amplify. These results indicated a good agreement between RNA-seq data and real-time PCR (Supplementary Figure S6). Taken together, cell death and growth-related pathways (including the mitotic cell cycle, regulation of the extrinsic apoptotic signaling pathway via death domain receptors, and negative regulation of cell population proliferation) were likely implicated in the defective follicle development of depression-like mice.