To investigate whether metformin impacts oocytes development from aged mice, 1065 GV (germinal vesicle) -stage oocytes from 9- to 10-month-old female mice were collected and cultured in M16 medium with 10, 20, or 50 μM metformin or control for 16 h (Figure 1A). The 10 μM metformin significantly increased the rate of polar body (PB1) extrusion compared to the control (10 μM metformin = 77.0 ± 4.28%, n = 280 vs. control = 67.9 ± 3.60%, n = 265; p < 0.05); note that the 10 µM concentration outperformed the other treatments (10 μM metformin = 77.0 ± 4.28%, n = 280 vs. 20 μM metformin = 74.60 ± 7.36%, n = 267 vs. 50 μM metformin = 74.65 ± 5.03%, n = 253) (Figure 1B). These results clearly suggest that treating oocytes from aged mice with metformin can significantly promote the IVM rate, and we used a 10 μM metformin concentration for all subsequent experiments, because it was the lower concentration that increased PB1 rate with respect to the control.

We next assessed the potential involvement of metformin in early embryonic development from aged mice by performing IVF with or without 10 μM metformin. We found no significant differences in the rate of fertilization between the two groups (metformin = 81.86 ± 12.36%, n = 295 vs. control = 90.79 ± 6.72%, n = 274; p > 0.05). However, the presence of metformin in the culture medium significantly increased by 9.4% (metformin = 74.02 ± 10.71%, n = 295 vs. control = 64.7 ± 5.98%, n = 274; p < 0.05) which developed into blastocysts (p < 0.05) (Figure 1E). These results suggest that metformin treatment of oocytes from aged mice can apparently improve both IVM rates and early embryonic development.

It has been widely reported that oocyte quality is determined by spindle morphology and chromosome alignment (Huang et al., 2015; Chen et al., 2016). Further, studies have reported that abnormal spindle morphology and aneuploidy in oocytes from aged mice can lead to decreased fertilization rates, increased risk of miscarriage, and birth defects in children (Martin-Montalvo et al., 2013; Barzilai et al., 2016; Cao et al., 2020). Our confocal microscopy analysis of MII-stage oocytes after IVM from aged mice revealed multiple spindle morphology and chromosome alignment abnormalities, including elongated spindles, an apparent lack of poles, and chromosome misalignment (Figure 2A). However, the metformin treatment group significantly reduced the proportion of abnormal chromosomes and spindles (metformin = 38.19 ± 5.24%, n = 71 vs. control = 64.60 ± 2.44%, n = 47; p < 0.01), as observing orderly aligned chromosomes on the equatorial plate and bipolar spindles (Figure 2C).

The CGs is regarded as an informative indicator of oocyte cytoplasmic maturation that can block polyspermy following fertilization (Walls and Hart, 2018). We assessed whether metformin affects the distribution of CGs of oocytes from aged mice using Lens culinaris agglutinin (LCA)-FITC staining and confocal microscopy. As the meiotic maturation is completed, normal distribution of CGs is distributed evenly in the oocyte subcortical region and except the CG-free domain near the chromosomes, whereas the abnormal distribution of CGs is discontinuous and weak signals in the oocyte cortex and without leaving CG-free domains. We found that the proportions of abnormal distribution of CGs were significantly reduced in the metformin treatment group, but more than 50% (metformin = 27.94 ± 4.01%, n = 60 vs. control = 51.98 ± 13.88%, n = 79; p < 0.05) lost this normal localization in the control group. (Figures 2B,D). Altogether, these results suggest that metformin may rescue the meiotic defects and recover the cytoplasmic maturation.

We next performed a single-cell transcriptome analysis of oocytes from aged mice, including both metformin-treated and control samples. Compared with controls, the metformin-treated oocytes had 295 up-regulated differentially expressed genes (DEGs) and 155 down-regulated genes (with the following cutoff criterion: adjusted p-values less than 0.05 (Figure 3A). Gene Ontology (GO) analysis of the top-30-ranking DEGs revealed enrichment for functional annotations including oxidative phosphorylation, oxygen binding, oxidoreductase activity, antioxidant activity, oxygen transport, cellular oxidant detoxification, and hydrogen peroxide metabolic process (Figure 3B). The expression trend data for 11 randomly selected genes in each group was verified using qPCR (Figure 3C). Numerous up-regulated genes have annotated functions related to anti-oxidative metabolism (Sod2), oocyte and embryonic development (Bmp15, Gdf9), autophagy (Tomm6), aging (Sirt3), and the mitochondrial function (Uqcrb, Mrps21). Some of the down-regulated genes had annotated functions concerning lipid binding (Golph3l) and metabolic processes (Acadsb) (Figure 3C). Our transcriptome analysis of MII-stage oocytes thus indicates that metformin treatment can mitigate oxidative stress in oocytes from aged mice.

ROS is one of the causes of age-related decline in fertility (Appasamy et al., 2008; Igarashi et al., 2015; Becatti et al., 2018), with specific reports of dysregulated mitochondrial redox balance, destabilization of mitochondrial DNA, as well as disrupted oocyte membrane function and deterioration (Droge, 2002; Tamura et al., 2020). We used carboxy-H2DCF diacetate to stain oocytes from aged mice with or without metformin treatment to evaluate whether metformin impacts ROS. Briefly, we found that the ROS level was significantly reduced in the metformin treatment oocytes compared to controls (Figures 4A,G; metformin = 31.78 ± 12.57, n = 98 vs. control = 56.22 ± 45.12, n = 92; p < 0.001).

Mitochondria are considered the major site of intracellular ROS production, as more than 90% of total cellular oxygen reduction involves electron carriers of the mitochondrial respiratory chain (Hamatani et al., 2004; Nohl et al., 2005). Given that previous studies of oocytes from aged mice have revealed mitochondrial dysfunctions such as defects in mitochondrial membrane potential (MMP), mitochondrial distribution (MD), and ATP levels (Zeng et al., 2018; Pasquariello et al., 2019; Cao et al., 2020), we focused our assessments of mitochondrial function on these aspects.

Pursing the idea that metformin may provide beneficial effect(s) against age-induced mitochondrial dysfunction of oocytes, we investigated the mitochondrial membrane potential (MMP) using a standard immunolabeling protocol with JC-1 wherein increasing membrane potential induces a green-to-red shift (Muhammad et al., 2020). This quantitative analysis confirmed that the red/green ratio was significantly increased in the metformin-treated oocytes compared to controls (Figures 4B,H; metformin = 1.09 ± 0.22, n = 75 vs. control = 1.0 ± 0.28, n = 68; p < 0.05). To further assess mitochondrial function, we used MitoTracker-Red staining to examine the mitochondrial distribution of oocytes from aged mice. Confocal microscopy showed that whereas around 60% (metformin = 41.15 ± 3.18%, n = 67 vs. control = 62.41 ± 8.72%, n = 70) of the oocytes from aged mice without metformin treatment had abnormally distributed mitochondria (i.e., asymmetric clustering), this proportion was significantly decreased by that given metformin treatment (Figures 4C,I). The distribution of mitochondria is known to dynamically impact ATP homeostasis, and mitochondrial disorders frequently lead to reduced ATP production (Wakim et al., 2017). Our analysis showed higher ATP levels in metformin-treated oocytes than in untreated controls (Figure 4J; metformin = 7.18 ± 4.21, n = 143 vs. control = 5.57 ± 2.36, n = 143; p < 0.001). Together, these results show that metformin treatment of oocytes from aged mice mitigates oxidative stress, specifically by reducing ROS levels and promoting mitochondrial function.

Given our observations of disrupted mitochondrial distribution, ATP production and MMP, we next evaluated whether metformin treatment influences mitochondrial ultrastructure in oocytes from aged mice. Our transmission electron microscopy (TEM) analysis showed that the metformin-treated oocytes had a significantly reduced extent of abnormal mitochondrial ultrastructure phenotypes (Figure 4K; metformin = 29.52 ± 6.33%, n = 14 vs. control = 36.92 ± 6.20%, n = 12; p < 0.01), which typically exhibit blurred cristae, narrowed inter-membrane spaces (Figures 4D, F1, F2). The normal mitochondrial ultrastructure had defined cristae, clearly visible intact inner and outer membranes, and well-defined intermembrane spaces (Figures 4E, F3, F4). Collectively, these results demonstrate that metformin can somehow improve age-related abnormal ultrastructure phenotypes of oocyte mitochondria.

Given the known impacts of oxidative stress in the induction of apoptosis (Chen et al., 2014; Dimozi et al., 2015), we next performed a TUNEL (Terminal deoxynucleotidyl transferase dUTP nick-end labeling) analysis to investigate potential impacts(s) of metformin on oocytes from aged mice. There were significantly fewer apoptotic oocytes in the metformin-treated group compared to the untreated controls (Figures 5A,D; metformin = 27.66 ± 2.34%, n = 54 vs. control = 42.48 ± 4.39%, n = 47; p < 0.01). Further, we found that the extent of caspase3 activation was significantly reduced in the metformin-treated group compared to the untreated controls (Figures 5B,E; metformin = 0.63 ± 0.30, n = 92 vs. control = 1.0 ± 0.32, n = 85; p < 0.05). Previous studies have shown that, during follicular development, oocytes block the induction of apoptosis by employing autophagy as a “survival mechanism” (Pepling, 2012). We conducted immunofluorescence staining of oocytes from aged mice against the autophagy marker protein LC3 (Microtubule-associated protein light chain 3) and found that metformin treatment significantly increased LC3 levels (Figures 5C,F; metformin = 1.14 ± 0.13, n = 71 vs. control = 1.0 ± 0.11, n = 52; p < 0.05). Thus, metformin treatment can both increase autophagy activity and decrease the extent of apoptosis in oocytes from aged mice.

SIRT3 (Sirtuin 3) is localized in the mitochondria and regulates the acetylation status of the most mitochondrial proteins (Sultana et al., 2016), often associated with aging (Ansari et al., 2017). Previous studies have shown that SIRT3 can deacetylate SOD2K68, consequently activating SOD2, thereby increasing its mitochondrial anti-oxidative activity (Someya et al., 2010; Liu et al., 2017; Cao et al., 2020), and acetylation levels of SOD2 are negatively associated with its enzymatic activity (Cao et al., 2020). However, whether metformin can mitigate mitochondrial ROS by SIRT3-mediated reduction of SOD2K68ac in IVM from oocytes from aged mice still needs to be investigated. We next assessed whether maternal age affects SIRT3 level in GV oocytes, immunoblotting of oocytes from old (42–45 weeks) and young mice (6–8 weeks) revealed that oocytes from aged mice had reduced SIRT3 expression compared with that in young mice (Figures 6A,E; young = 0.91 ± 0.08, n = 3 vs. aged = 0.62 ± 0.23, n = 3; p = 0.11). We next conducted SOD2K68ac staining using confocal microscopy in the metformin-treated group and the untreated controls from the MII-stage oocytes after IVM, and found that metformin-treated significantly reduced SOD2K68ac levels (Figures 6B,F; metformin = 13.35 ± 4.90, n = 66 vs. control = 15.11 ± 4.48, n = 67; p < 0.05). We further examined whether metformin in the IVM culture medium of oocytes from aged mice may affect the SOD2K68ac by regulating SIRT3, consequently reducing mitochondrial ROS in oocytes from aged mice. Fully grown GV oocytes from old mice were injected with exogenous Sirt3 siRNAs or PBS, as the negative control (Figure 6C).

Monitoring ROS levels in live oocytes using carboxy-H2DCF diacetate fluorescent dye with confocal microscopy showed that the ROS accumulation was significantly increased in the siSirt3 oocytes compared with control (PBS) oocytes. Moreover, treatment of siSirt3 oocytes with metformin led to a significant reduction in ROS as compared with the untreated siSirt3 oocytes (Figures 6D,G; siSirt3 = 2.04 ± 0.68, n = 11 vs. control = 1.0 ± 0.19, n = 12; p < 0.01; siSirt3 = 2.04 ± 0.68, n = 11 vs. siSirt3+metformin = 1.29 ± 0.29, n = 10; p < 0.01). Similarly, assessment of SOD2K68ac levels showed that siSirt3 oocytes had significantly increased SOD2K68ac levels compared to the control cells, and also showed that metformin treatment could partially rescue this increased SOD2K68ac phenotype (Figures 6D,H; siSirt3 = 2.60 ± 0.63, n = 22 vs. control = 0.97 ± 0.74, n = 20; p < 0.001; siSirt3 = 2.60 ± 0.63, n = 22 vs. siSirt3+metformin = 2.23 ± 0.81, n = 24; p < 0.05). Thus, metformin can mitigate mitochondrial ROS via SIRT3-induced reduction of SOD2K68ac in oocytes from aged mice.

Old ICR (Institute of Cancer Research) female mice from 42- to 45-week-old (when fertility declines rapidly) and young ICR female mice from 6- to 8-week-old were purchased from Beijing Vital River Experimental Animals Centre (Beijing, China). Mice were housed under a 12-h light: 12-h darkness cycle, in a controlled temperature and humidity animal facility, and were provided with water ad libitum and food.

To retrieve fully grown GV oocytes, mice were injected intraperitoneally with 10 IU Pregnant Mares Serum Gonadotropin (PMSG) (Ningbo Hormone Product Company, China). After 46–48 h, mice were ultimately sacrificed, cumulus-oocyte complexes were collected by manual rupturing of antral ovarian follicles, cumulus cells were removed by repeatedly pipetting. For IVM, denuded GV oocytes cultured were cultured in the small drops in M16 medium supplemented with or without different concentrations of metformin (10, 20, or 50 μM), subsequently covered with mineral oil were incubated under 6% CO₂, 5% O₂, and 90% N₂ at 37°C for 16 h to determine the PB1 extrusion rate.

For IVF, mice were intraperitoneally injected with 10 IU PMSG, followed 48 h later by 10 IU of human chorionic gonadotrophin (hCG) (Ningbo Hormone Product Company, China). After 16 h, mice were ultimately sacrificed, oviductal ampullae were broken to release cumulus-oocyte complexes. Meanwhile, the adult male mice were sacrificed, and sperm was obtained from cauda epididymis via orchidectomy. Subsequently cumulus-oocyte complexes fertilized with adult male sperm (the concentration of sperm was (1–2.5) × 10⁶/ml) in G-IVF medium (Vitrolife, Sweden) supplemented with or without 10 μM metformin. Zygotes covered with mineral oil were incubated under 6% CO₂, 5% O₂, and 90% N₂ at 37°C in G1 medium (Vitrolife, Sweden) to observe embryonic developmental potential with a stereomicroscope (Nikon SMZ1500).

For microinjection in knockdown experiments, 10 pL Sirt3-targeting siRNA (10 ng/μL) was injected into GV stage oocytes, and the same amount of RNase-free PBS was injected as a control. Oocytes were arrested at the GV stage in M16 medium containing 2.5 mM milrinone for 20 h to promote the synthesis of new protein. Then, oocytes were cultured in the M16 medium supplemented with (siRNA + metformin) or without (siRNA) 10 μM metformin for 16 h. RNA was obtained from RiboBio (Guangzhou, China) and the sequences used are listed in Supplementary Table S2.

For measurement of intracellular ROS levels, oocytes were incubated in M2 medium containing 10 mM carboxy-H2DCF diacetate (Beyotime, China) for 30 min at 37°C. After being washed 3 times in an M2 medium and subsequently mounted on glass slides, oocytes were imaged with a confocal microscope (Dragonfly, Andor Technology, United Kingdom). The fluorescence intensity for each oocyte was measured with ImageJ (National Institutes of Health, United States).

ATP measurement was performed using the luciferin–luciferase reaction (Bioluminescent Somatic Cell Assay Kit, Sigma, United States). Firstly, mixed for 5 s before detection, a linear regression of standard curve containing 11 ATP concentrations from 10 fmol to 10 pmol was used to determine oocyte ATP content. Then, ATP concentrations of oocytes were measured on an EnSpire Multimode Plate Reader (PerkinElmer, United States).

Immunofluorescence was performed as described previously (Cao et al., 2020). In brief, oocytes were fixed for 30 min with 4% paraformaldehyde, permeabilized for 20 min with 0.1% Triton X-100, and blocked with 1% BSA for 0.5 h. Samples were incubated with primary antibodies in 1% BSA for 1 h at room temperature. The primary antibodies were as follows: anti-active caspase-3 (Abcam, United States), anti-LC3 (Cell Signaling, United States), and anti-SIRT4 (Abcam, United States). After being washed three times for 5 min, samples were incubated with suitable secondary antibodies at room temperature for 1 h. To visualize the spindle, oocytes were probed with an anti-α-tubulin antibody (Sigma, United States). For the distribution of cortical granules staining, oocytes were probed with labeled lens culinaris agglutinin (Vectorlabs, United States). For the TUNEL staining, oocytes were probed with an in situ cell death kit (Roche, Swiss). To evaluate the mitochondrial membrane potential (MMP), oocytes were incubated in 2 μM JC-1 (Invitrogen, United States) for 30 min at 37°C. To detect the mitochondrial distribution, oocytes were incubated in 200 nM MitoTracker-Red (Invitrogen, United States) for 30 min at 37°C. To visualize chromosomes, oocytes were probed with DAPI (Solarbio, China) for 10 min. Oocytes were examined under a laser scanning confocal microscope (Dragonfly, Andor Technology, United Kingdom). Images were acquired by using the same confocal microscope settings within and between experiments. The mean fluorescence intensity of each oocyte was measured with ImageJ (National Institutes of Health, United States). The antibodies used in these experiments are shown in Supplementary Table S1.

For RNA sequencing, 15 MII stage oocytes after IVM from three mice were considered one group, and three replicates were assessed per group. The sample library was built with a smart-seq HT Kit (Takara, Japan) at Shanghai Sinomics Corporation and sequenced with an Illumina NovaSeq 6000 instrument (Illumina, United States). Raw data files are publicly available from the Gene Expression Omnibus (GEO) database under accession number GSE201098. Total RNA was extracted from samples using RNeasy Mini Kits (Qiagen, Germany) following the manufacturer’s protocol. Expression levels of mRNA were partially verified by qPCR experiments performed with a Light Cycler 480 (Roche, Swiss). The mRNA levels were normalized to endogenous GAPDH (Glyceraldehyde-3-phosphate dehydrogenase) mRNA levels using calculations performed with Microsoft Excel. Primer sequences are shown in Supplementary Table S2.

Oocytes from five mice treated with or without metformin were used for each sample. To analyze mitochondrial ultrastructure, oocytes were prepared for transmission electron microscopy (TEM). Morphometric analysis of mitochondrial ultrastructure was based on electron micrographs at 30,000-fold magnification.

For total protein extraction, a total of 100 oocytes were lysed in SDS buffer by boiling for 5 min. The sample was separated by 10% SDS-PAGE and transferred to a PVDF membrane (Bio-Rad), blocked with 5% skim milk diluted in Tries-buffered saline containing 0.05% Tween-20 (TBST) for 1 h at room temperature, and then incubated with primary antibody overnight at 4°C (anti-SIRT3 antibody, 1:500), incubation with HRP-conjugated secondary antibodies for 1 h at room temperature. Immunoreactive bands and molecular weight were detected using the Odyssey Infrared Imaging System (LI-COR Bioscience, United States). The antibodies used in these experiments are shown in Supplementary Table S1.

All experiments were replicated three or more times; data are presented as the mean ± SD unless otherwise indicated, and all % data were subjected to an arcsine-square-root transformation before statistical analysis. Differences between the two groups were analyzed for statistical significance using two-tailed unpaired Student’s t-tests. Comparisons between more than two groups were analyzed using a one-way ANOVA (Analysis of Variance) test implemented in GraphPad Prism 7 (GraphPad Software, San Diego, CA, United States). *p < 0.05; **p < 0.01, and ***p < 0.001.