Oocyte-specific acylglycerol kinase loss leads to infertility in mice via mitochondrial dysfunction

Abstract

The orderly architecture and energy supply of mitochondria, orchestrated by proteins, are essential for oocyte maturation. Acylglycerol kinase (AGK) is known to predominantly localize to the mitochondrial membrane and regulate multiple cellular processes related to Sengers syndrome and tumorigenesis. However, the functions of AGK in the ovaries remain largely unknown. Here, we generate mice with oocyte-specific disruption of Agk. Ablation of Agk in oocytes causes ovarian atrophy characterized by the arrest of early secondary follicles and a significantly reduced number of mature follicles, ultimately leading to female sterility. Furthermore, mitochondria in Agk-deficient oocytes exhibit significant structural and organizational defects. The developmental competence of Agk-deficient oocytes is compromised with decreased mitochondrial membrane potential, reduced mtDNA copy number, impaired ATP synthesis, and elevated reactive oxygen species. Mechanistically, single-cell RNA sequencing shows that genes associated with the mitochondrial oxidative phosphorylation (OXPHOS) pathway are among the most significantly downregulated in Agk-disrupted oocytes. Spatial metabolomics further identifies a substantial reduction in phosphate-associated metabolites linked to the OXPHOS pathway in Agk-deficient oocytes. These findings demonstrate that AGK plays an essential role in oocyte development and folliculogenesis by maintaining mitochondrial function. The present study identifies AGK as a potential diagnostic marker and therapeutic target for female infertility.

Oocyte-specific AGK deficiency causes multifaceted mitochondrial dysfunction and follicular development arrest, providing insights into the molecular mechanisms and a potential diagnostic and therapeutic target for female infertility.

Introduction

The quality of follicle development in the ovary is a primary determinant of successful pregnancy outcomes in both natural and assisted reproduction. Ovarian follicles progress through primordial, primary, secondary, and antral follicle stages¹. Secondary follicles are further subdivided into early and late stages based on the number of granulosa cell layers². The developmental continuum from primordial follicles to antral follicles undergoes substantial cellular and molecular reprogramming, including organelle restructuring, protein turnover, and modification¹. Critically, morphological and functional transformations of mitochondria are key determinants of oocyte developmental competence throughout folliculogenesis³. Emerging evidence has demonstrated that mitochondria play a crucial role in oocyte development by supplying ATP through the oxidative phosphorylation (OXPHOS) system^4,5.

Mitochondria possess their own unique DNA, known as mitochondrial DNA (mtDNA). The mtDNA is a double-stranded circular DNA with a length of 16.6 kb and contains 37 genes which encode 13 OXPHOS complex subunits, along with 22 transfer RNAs and 2 ribosomal RNAs required for intramitochondrial protein synthesis⁶. Oogenesis begins with approximately 200 copies of mtDNA in each primordial germ cell, which undergoes dynamic expansion to reach 50,000 to 550,000 copies in mature mouse and human oocytes, albeit with considerable variability across developmental stages⁷. Compared to somatic cells, oocytes contain a higher abundance of mitochondria, which are characteristically smaller and rounder⁸. These mitochondria transition from a circular shape in primary oocytes to a more elongated, oval form upon maturation³. Consequently, defects in mitochondrial structure, shape, number, mtDNA copy number, and energy synthesis in oocytes can lead to follicle atresia, oocyte maturation disturbance, and female infertility⁹. Indeed, mitochondrial dysfunction in oocytes, often arising from specific gene mutants, has been directly linked to such pathologies. For instance, oocyte-specific deletion of genes crucial for mitochondrial structural dynamics, such as mitofusin 1 (Mfn1) and dynamin-related protein 1 (Drp1, fission), disrupts mitochondrial architecture and leads to female sterility or subfertility, respectively^10,11. Similarly, mutations in genes essential for OXPHOS complex assembly (COX15) or the mitochondrial stress response (Clpp) result in severe oocyte developmental defects and infertility^12,13. Notably, human mutations in CLPP are also associated with premature ovarian insufficiency¹⁴. These examples underscore the critical dependence of female fertility on a diverse set of mitochondria-associated proteins.

Acylglycerol kinase (AGK) is an evolutionarily conserved enzyme, exhibiting 96% sequence homology between rodents and humans¹⁵. It is widely expressed in tissues including the heart, muscles, kidneys, and brain^15,16. AGK is also frequently overexpressed in various malignancies, such as uterine, cervical, esophageal, gastric, prostate, and breast cancer^16–18, implicating its important role in tumorigenesis. In somatic cells, AGK localizes to mitochondria and phosphorylates monoacylglycerol (MAG) and diacylglycerol (DAG) to generate phosphatidic acid (PA) and lysophosphatidic acid (LPA), respectively¹⁹. Furthermore, independent of its kinase activity, AGK has been shown to be essential for the assembly and function of the mitochondrial inner membrane TIM22 complex, a process critical for mitochondrial oxidative phosphorylation and cell proliferation^20,21. In addition, extramitochondrial AGK participates in glycolysis of CD8 T cells and platelet formation by interacting with PTEN or JAK2, respectively^15,22. The mechanism of mitochondrial or non-mitochondrial AGK action, as well as its enzymatic and non-enzymatic activities, is summarized in Supplementary Fig. 1. However, the expression profile and physiological functions of AGK in the reproductive system remain uncharacterized.

In this study, we generated an oocyte-specific Agk knockout (Agk^flox/flox; Dppa3-Cre) mouse model to elucidate the role of AGK in female fertility. Our results demonstrate that oocyte-specific ablation of Agk results in female infertility due to defective folliculogenesis characterized by arrest at the early secondary follicle stage and severe mitochondrial dysfunction. These findings identify AGK as a critical regulator of oocyte development and establish it as a potential diagnostic marker and therapeutic target for female infertility.

Results

Oocyte-specific disruption of Agk results in female infertility

To investigate the physiological role of AGK in the ovary, we first explored its spatial and temporal expression profile. AGK was widely expressed in female reproductive organs, including ovary, uterus, and oviduct (Fig. 1A). Within the ovary, the expression of Agk mRNA was the highest in the ovaries on the third day, followed by the seventh day, implicating a potential role in early ovarian development (Fig. 1B). Immunohistochemical analysis confirmed that AGK protein is expressed in both oocytes and surrounding granulosa cells (Fig. 1C). Based on its prominent expression and potential role in the ovary, we generated oocyte-specific Agk conditional knockout (Agk cKO) mice by crossing Agk^flox/flox mice with transgenic mice carrying developmental pluripotency associated 3 (Dppa3, also known as Stella or Pgc7) promoter-mediated Cre recombinase (Supplementary Fig. 2). This recombinase is specifically active at embryonic day 7 and continues to exert its active role during the primordial germ cell (PGC) stage of embryonic development²³. To validate the model, we confirmed the efficiency of Agk deletion. Quantitative PCR and immunohistochemistry revealed that Agk expression was successfully reduced at both the mRNA (Fig. 1D) and protein (Fig. 1E) levels specifically in the oocytes of Agk cKO mice, while remaining unaffected in granulosa cells (Fig. 1D).

Fig. 1. Expression profile of AGK and the reproductive phenotype of female Agk cKO mice.

A Western blot analysis of AGK protein expression across various organs (heart, liver, spleen, lung, kidney, brain, muscle, ovary, uterus, and oviduct) from 8-week-old wild-type mice. The bar graph below shows the quantitative analysis of three independent replicates (mean ± SD, and individual data points are shown). B RT-qPCR analysis of Agk mRNA expression in wild-type ovaries at postnatal developmental stages (3, 7, 14, 21, 35, and 56 days). n = 9. C Immunohistochemical staining for AGK protein in an 8-week-old wild-type mouse ovary. Oocytes and granulosa cells were marked with a red circle and a black arrow, respectively. Scale bar, 200 μm. D Relative Agk mRNA expression in oocytes (n = 16 for control group and n = 15 for Agk cKO group) and granulosa cells (n = 4 each group) from 8-week-old control and Agk cKO mice. E Immunohistochemical analysis of AGK expression in ovaries from 8-week-old control and Agk cKO mice, with quantification of AGK signal intensity in oocytes (n = 22 for the control group and n = 31 for Agk cKO group). Oocytes and granulosa cells were marked with a red circle and a black arrow, respectively. F Comparison of litter size between control and Agk cKO mice (n = 8 each group). G Representative ovarian morphology and ovary-to-body weight ratios from 8-week-old control and Agk cKO mice (n = 3 - 4 each group). H Representative images of vaginal smear cytology used for estrous cycle staging from control and Agk cKO mice. I, J Quantification of the duration and proportion of each estrous cycle stage in control and Agk cKO mice (n = 6 each group). K Serum concentrations of anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol (E2), and testosterone in control and Agk cKO mice (n = 7 - 8 each group). For all panels, data are presented as mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001.

Agk cKO female mice were viable, developed normally into adulthood, and were phenotypically indistinguishable from their control littermates. However, upon mating with wild-type males, Agk cKO females were found to be completely infertile (Fig. 1F). Anatomical examination revealed that the ovaries of Agk cKO mice were markedly smaller, with a significantly reduced ovary-to-body weight ratio compared to controls (Fig. 1G).

We next assessed the physiological consequences of oocyte Agk disruption. The estrus cycle of adult Agk cKO mice was irregular (Fig. 1H). Specifically, the proportion of diestrus was significantly increased at the expense of markedly reduced proestrus (Fig. 1I), and the total duration of metestrus and diestrus was prolonged (Fig. 1J). Consistent with ovarian dysfunction, serum hormone analysis showed significantly decreased levels of anti-Müllerian hormone (AMH), estradiol (E2), with no significant alterations in luteinizing hormone (LH), follicle-stimulating hormone (FSH), and testosterone levels observed in Agk cKO mice (Fig. 1K). Collectively, these results demonstrate that oocyte-specific Agk disruption impairs ovarian function, leading to hormonal imbalances, anovulation, and ultimately, female infertility.

Agk deficiency arrests folliculogenesis at the early secondary follicle stage

To determine the impact of oocyte-specific Agk deficiency on folliculogenesis, we performed quantitative histological analysis of the ovaries from control and Agk cKO mice at 2, 4, and 8 weeks of age. At 2 weeks, the ovaries of Agk cKO mice showed no discernible morphological differences compared to controls (Fig. 2A). However, a clear developmental defect emerged by 4 weeks, at which point Agk cKO ovaries contained a significantly higher number of early secondary follicles but fewer late secondary and antral follicles (Fig. 2B). This phenotype was exacerbated by 8 weeks, with a marked accumulation of primordial plus primary, and early secondary follicles, and a concomitant depletion of late secondary and antral follicles, as well as corpora lutea (Fig. 2C). Collectively, these results indicate that oocyte-specific Agk deficiency leads to progressive disruption of folliculogenesis, culminating in developmental arrest at early secondary follicle stage in adult mice.

Fig. 2. Defective folliculogenesis in Agk cKO mice.

Representative images of H&E-stained ovarian sections from 2-week-old (A), 4-week-old (B), and 8-week-old (C) control and Agk cKO mice, accompanied by quantification of follicles at differential stages. Follicle stages are defined as primordial (PrF), primary (PF), early secondary (ESF), late secondary (LSF), antral (AF), and corpus luteum (CL). Scale bar, 100 μm. Data are presented as mean ± SD (n = 4 per group). ns, not significant; *p < 0.05, **p < 0.01.

To functionally assess the developmental competence of Agk-deficient oocytes, we first conducted superovulation assays. Histological examination of ovaries 48 hours post-PMSG injection confirmed this lack of response; while control ovaries contained numerous large antral follicles, most follicles in Agk cKO ovaries had failed to progress beyond the early secondary stage (Fig. 3A). To further examine if this developmental failure was oocyte-intrinsic, we isolated germinal vesicle (GV) -stage oocytes and cultured them in vitro. In the control group, 84% of oocytes underwent germinal vesicle breakdown (GVBD). In the Agk cKO group, 55% of the oocytes underwent GVBD (Fig. 3B).

Fig. 3. Impaired ovarian response to gonadotropins and compromised oocyte meiotic competence in Agk cKO mice.

A H&E-stained ovarian section from 3-week-old control and Agk cKO mice following pregnant mare serum gonadotropin (PMSG) stimulation or vehicle control. Scale bar, 200 μm. B Quantification of the germinal vesicle breakdown (GVBD) rate in Agk cKO and control oocytes during in vitro maturation. Data are expressed as mean ± SD. n = 5, ***p < 0.001.

Despite this profound GVBD incompetence, immunofluorescence staining of the synaptonemal complex proteins SCP3 and SCP1 in ovaries from E18.5 to P5 revealed normal progression through the initial stages of prophase I (leptotene, zygotene, pachytene, diplotene) in Agk cKO oocytes (Supplementary Fig. 3).

Taken together, these results indicate that oocyte-intrinsic AGK function is essential for progression beyond the early secondary stage and for the meiotic resumption at puberty.

Agk disruption profoundly alters oocyte mitochondrial gene expression

To elucidate the molecular mechanisms by which AGK regulates oocyte and follicle development, we performed single-cell mRNA sequencing (Smart-seq3) on oocytes isolated from early growing follicles of 4-week-old Agk cKO and control mice. Principal component analysis (PCA) revealed that the transcriptomes of Agk cKO and control oocytes segregated into two distinct clusters, indicating a significant and consistent impact of Agk disruption on the oocyte gene expression profile (Fig. 4A).

Fig. 4. Transcriptomic alterations in 4-week-old Agk cKO oocytes.

A Principal component analysis (PCA) of transcriptomes in Agk cKO (n = 35) and control (n = 31) oocytes isolated from early secondary follicles of 4-week-old mice. B Volcano illustrating differentially expressed genes (DEGs) between Agk cKO and control oocytes from 4-week-old mice. C Heatmap of DEG expression profiles in Agk cKO and control oocytes from 4-week-old mice. D Protein-protein interaction (PPI) network analysis of proteins encoded by DEGs between Agk cKO and control oocytes from 4-week-old mice. Gene Ontology (GO) (E) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (F) pathway enrichment analysis of DEGs between Agk cKO and control oocytes from 4-week-old mice. G RT-qPCR validation of selected mitochondrial gene expression in Agk cKO and control oocytes (n = 6 each group). H The mRNA level of Tim22 from control (n = 31) and Agk cKO (n = 35) oocytes in Smart-seq3 sequencing data. Data were expressed as mean ± SD. ns, no significance, *p < 0.05, **p < 0.01.

We identified 136 differentially expressed genes (DEGs) between the two groups, with 56 genes upregulated and 80 downregulated in Agk cKO oocytes (Fig. 4B). Strikingly, a substantial proportion of the downregulated genes were mitochondrially encoded components of the electron transport chain, a pattern clearly visible in both volcano plot and heatmap analysis (Fig. 4B, C). Protein-protein interaction (PPI) network analysis of the DEGs further highlighted a highly interconnected cluster of these downregulated mitochondrial genes, including mt-Cytb, mt-Co1, mt-Co3, and mt-Nd1 to mt-Nd6 (Fig. 4D). Consistent with these observations, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses revealed a significant enrichment of downregulated genes in pathways related to electron transfer activity and oxidative phosphorylation (Fig. 4E, F).

This transcriptomic dysregulation was exacerbated at 8 weeks of age, with a greater number of DEGs identified (560 upregulated, 412 downregulated) and an even more pronounced enrichment of genes associated with mitochondrial oxidative phosphorylation (Supplementary Fig. 4). We validated the differential expression of key mitochondria-related transcripts by RT-qPCR, confirming the sequencing results (Fig. 4G). Notably, despite AGK’s established role as a component of the TIM22 complex in the inner membrane of mitochondria, we observed no significant changes in the transcript level of TIM22 in Agk-deficient oocytes (Fig. 4H). Collectively, these transcriptomic data suggest that AGK is essential for maintaining the expression of mitochondrially encoded OXPHOS subunits in oocytes, a function that appears to be distinct from the regulatory role of its known TIM22 complex partners.

Agk deficiency impairs mitochondrial architecture and function in oocytes

Building upon the transcriptomic data pointing to mitochondrial dysregulation, we next investigated the direct impact of Agk ablation on mitochondrial integrity within oocytes. Immunofluorescence staining confirmed that AGK co-localizes with the mitochondrial marker COXIV, affirming its presence in oocyte mitochondria (Fig. 5A). To examine the consequences of Agk loss, we employed transmission electron microscopy (TEM) to assess mitochondrial ultrastructure. In control oocytes, mitochondria exhibited a normal morphology, characterized by well-defined, organized cristae and even distribution throughout the cytoplasm. In stark contrast, mitochondria in Agk cKO oocytes were structurally aberrant, appearing swollen and aggregated, with a near-complete loss of internal cristae and the formation of vacuole-like structures (Fig. 5B). To determine if these severe structural defects corresponded with functional impairment, we assessed two key parameters of mitochondrial health. First, we measured the mitochondrial membrane potential (ΔΨm) using JC-1 staining and observed a significant reduction in Agk-deficient oocytes, indicative of compromised bioenergetic activity (Fig. 5C). Second, as mtDNA content is a critical determinant of oocyte quality, we quantified mtDNA copy number. Agk cKO oocytes contained significantly fewer mtDNA copies compared to control oocytes, indicating a failure in mitochondrial genome maintenance or replication (Fig. 5D). Taken together, these results demonstrate that AGK is indispensable for maintaining mitochondrial ultrastructure, bioenergetic capacity, and genomic content in oocytes, providing a direct mechanistic link between AGK deficiency and the observed arrest in folliculogenesis.

Fig. 5. Aberrant mitochondrial ultrastructure and function in Agk cKO oocytes.

A Co-immunofluorescence staining for AGK (green) and COX IV (red) in control and Agk cKO mouse ovaries. Nuclei were counterstained with Hoechst (blue). Scale bar, 50 μm. B Representative transmission electron micrographs of mitochondria in oocytes from 4-week-old control and Agk cKO mice. Scale bars, 1 μm (left panel) and 0.5 μm (other panels). C Assessment of mitochondrial membrane potential in control and Agk cKO oocytes using JC-1 staining. JC-1 aggregates (red fluorescence) indicate high potential, while monomers (green fluorescence) indicate low potential. Bar graphs show the ratio of red to green fluorescence intensity (n = 31 for control; n = 29 for Agk cKO). Scale bar, 50 μm. D Quantification of mitochondrial DNA (mtDNA) copy number per oocyte from control (n = 8) and Agk cKO (n = 7) mice. Data are expressed as mean ± SD. **p < 0.01, ***p < 0.001.

Agk deficiency disrupts mitochondrial energy metabolism and induces oxidative stress in oocytes

Given the observed mitochondrial structural defects, we next sought to determine their functional metabolic consequences in oocytes. A primary function of mitochondria is ATP synthesis; accordingly, we quantified intracellular ATP levels and found that ATP content was significantly diminished in Agk cKO oocytes compared to controls (Fig. 6A). To gain a broader understanding of the metabolic disruption, we performed spatial metabolomics on oocytes from early secondary follicles. Adjacent section HE staining was used to identify oocyte boundaries (red circles, Fig. 6B), enabling oocyte-specific metabolite quantification independent of surrounding somatic cells. PCA showed a clear separation between the Agk cKO and control groups, indicating a widespread metabolic reprogramming (Fig. 6C).

Fig. 6. Altered metabolic profile in Agk cKO oocytes.

A Quantification of ATP content per oocyte from 4-week-old control and Agk cKO mice (n = 18 each group). B Representative image showing the oocyte area within early secondary follicles demarcated for analysis based on adjacent H&E-stained sections. C Principal component analysis (PCA) of metabolites identified by spatial metabolomics in 4-week-old control and Agk cKO oocytes. D Volcano plot of differentially abundant metabolites in 4-week-old Agk cKO and control oocytes. Metabolites with a p-value < 0.05 and |log2 (Fold change)| > 1 are highlighted. E Representative mass spectrometry ion images of phosphate in ovarian sections from 4-week-old control and Agk cKO mice. The bar graph shows the relative abundance of phosphate in oocytes (control, n = 21; Agk cKO, n = 31). F Relative abundance of phosphatidic acid (PA) in control and Agk cKO ovarian sections. G Metabolic pathway enrichment analysis of differentially abundant metabolites in 4-week-old control and Agk cKO oocytes. H Heatmap of key significant differential metabolites. I Representative images of DCFH-DA fluorescence (green) and quantification of reactive oxygen species (ROS) level in 4-week-old control (n = 36) and Agk cKO (n = 33) oocytes. Scale bar, 20 μm. Data are expressed as mean ± SD. **p < 0.01, ***p < 0.001.

We identified numerous differentially abundant metabolites between the two conditions (Fig. 6D). Pathway analysis of these metabolites revealed significant alterations in core mitochondrial processes, including the glycerol phosphate shuttle, the electron transport chain (ETC), and the transfer of acetyl groups into mitochondria and glycolysis (Fig. 6G). In the metabolite intensity maps, red indicates higher relative abundance, blue indicates lower abundance, and intermediate colors represent mid-range levels. Consistent with ETC dysfunction, phosphate levels within this pathway were significantly decreased (Fig. 6E). Notably, pathways for thiamine (Vitamin B1) metabolism and cardiolipin biosynthesis were also significantly affected. These pathways are critical for generating essential coenzymes for key metabolic reactions and for synthesizing a key phospholipid required for the structural integrity of the inner mitochondrial membrane, respectively, reinforcing the link between AGK and mitochondrial homeostasis. As AGK is a lipid kinase that produces lysophosphatidic acid (LPA) and phosphatidic acid (PA), we specifically examined the levels of these metabolites. While LPA was undetectable in oocytes from either group, PA levels were significantly elevated in Agk-deficient oocytes (Fig. 6F). At the same time, phospholipid molecules such as PC and PE also showed significant increases (Fig. 6H and Supplementary Fig. 5). A common consequence of electron transport chain disruption is the overproduction of reactive oxygen species (ROS). We therefore measured intracellular ROS levels and found them to be significantly elevated in Agk cKO oocytes (Fig. 6I). Collectively, these findings demonstrate that AGK deficiency triggers a cascade of mitochondrial dysfunction, beginning with metabolic disruption and culminating in severe oxidative stress, thereby compromising the fundamental bioenergetic and redox state required for oocyte developmental competence.

Discussion

Acylglycerol kinase (AGK) is a mitochondrial lipid kinase that also functions as an integral subunit of the TIM22 protein import complex, highlighting its multifunctional nature. In humans, pathogenic mutations in AGK cause Sengers syndrome, a severe multisystem disorder characterized by congenital cataracts and mitochondrial myopathy, often leading to premature death²⁰. AGK has also been implicated in the progression of various cancers, including prostate tumors, nasopharyngeal cancer, ovarian cancer, breast cancer, hepatocellular carcinoma, and gliomas^{16,17,22,24–26}. In terms of the correlation between AGK and reproduction, clinical observations based on gene sequencing show that AGK mutations are related to premature ovarian failure^27,28. Here, our mouse model directly confirms this function and provide a mechanistic explanation for this clinical phenotype, demonstrating that AGK deficiency causes mitochondrial dysfunction, energy depletion, and meiotic arrest in oocytes, ultimately lead to premature ovarian failure and infertility. Our findings establish AGK as an essential regulator of folliculogenesis and oocyte competence, identifying it as a potential diagnostic target for certain forms of female infertility.

In this study, we utilized a Dppa3-Cre driver to excise Agk in primordial germ cells (PGCs) beginning at embryonic day 7²³. Interestingly, despite this early reduction, follicular development in Agk cKO mice appeared normal at 2 weeks of age. The developmental arrest phenotype only became evident at 4 weeks, manifesting as an accumulation of secondary follicles. This temporal disconnect suggests that AGK’s essential function is not required during the initial stages of primordial follicle formation but becomes critical during the postnatal wave of follicular growth. The accumulation of primordial and primary follicles in Agk cKO ovaries (Fig. 2C) can be explained by the developmental block at the early secondary follicle stage, which creates a ‘bottleneck’ in folliculogenesis. Under normal circumstances, primordial follicles are continuously recruited into the growing pool and progress through primary and secondary stages before either ovulating or undergoing atresia. In Agk cKO ovaries, follicles successfully initiate growth and transition from primordial to primary and early secondary stages; however, the developmental arrest at the early secondary stage prevents further progression. Consequently, the primordial/primary follicle pool is not being depleted at the normal rate, leading to an apparent accumulation, with no significant difference in the total number of follicles. In adulthood, the hormonal profile of Agk cKO mice, characterized by reduced AMH and estradiol (E2), is a classical endocrine signature of ovarian dysfunction^29,30. This pattern is consistent with findings in other models of premature ovarian insufficiency, such as the Fbxw7 cKO mouse³¹. The low E2 levels are a direct consequence of the arrest at the preantral stage, as antral follicles are the primary source of E2 production. The low AMH levels reflect the arrest before follicles reach the preantral/small antral stages, where granulosa cells normally secrete AMH. Additionally, oocyte mitochondrial dysfunction may further compromise follicle quality and AMH production. This endocrine profile mirrors human premature ovarian insufficiency (POI), suggesting that disruption of the AGK pathway could represent a novel etiological factor³². The developmental arrest underscores the indispensability of functional oocyte-granulosa cell bidirectional communication. Although granulosa cells retain AGK expression, they fail to proliferate. We propose that severe energy depletion and metabolic dysfunction in Agk cKO oocytes impair the synthesis and/or secretion of key paracrine factors (e.g., GDF9 and BMP15), which are essential for granulosa cell function and follicle progression beyond the primary stage³³. Collectively, these results position the oocyte as the central organizer of follicular development and demonstrate that mitochondrial dysfunction intrinsic to the oocyte can incapacitate the entire follicular unit. Therefore, the infertility observed in Agk cKO mice unequivocally originates from an oocyte-intrinsic defect caused by AGK deficiency.

Mechanistically, mitochondrial AGK has two well-described functions: its lipid kinase activity and its structural role within the TIM22 complex. As a kinase, AGK phosphorylates mono- and diacylglycerol to produce LPA and PA, respectively. PA is a key precursor for cardiolipin, a phospholipid essential for mitochondrial inner membrane structure and the stability of respiratory super complexes³⁴. LPA could have a pro-fission role or pro-fusion effect in membrane morphology when generated on the mitochondrial surface or on the inner surface of the mitochondrial membrane³⁵. Our spatial metabolomics analysis, however, presented a complex picture. While LPA was undetectable in oocytes from both genotypes, PA levels were significantly increased in Agk-deficient oocytes. This counterintuitive result suggests that the observed phenotype is not caused by a simple loss of PA production via AGK’s kinase activity; the elevated PA may instead reflect a compensatory response or a bottleneck in other lipid metabolic pathways. Moreover, AGK has previously been reported to have functions independent of esterase activity. Identified as a subunit of the human TIM22 complex, AGK also contributes to the stabilization of the complex and the regulation of mitochondrial carrier protein import and assembly²⁰. Here, we found that the expression of the TIM22 gene did not significantly change in Agk-knockout oocytes, suggesting that AGK may not function through the TIM22-mediated signaling pathway. In another experimental study on ovarian cancer cells, AGK was shown to not only sustain mitochondrial cristae morphogenesis but also enhance ROS production in a kinase-independent manner, while directly interacting with ribosomal protein L39²⁶. These results were consistent with our results. Collectively, our data suggest that a non-canonical, potentially structural or kinase-independent function of AGK may be dominant in regulating oocyte mitochondrial integrity. Agk^G126E/G126E knockin (KI) mice expressing the catalytically inactive mutant Agk have been generated³⁶. To evaluate whether the kinase activity of AGK plays a role in the female reproductive system, we have introduced this mouse for further research.

Oocyte maturation is an energy-intensive process that relies on a metabolic shift from glycolysis to mitochondrial oxidative phosphorylation (OXPHOS) to build the ATP reserves necessary for oocyte maturation and folliculogenesis^9,37,38. Our findings provide a clear molecular basis for the failure of this metabolic transition in Agk-deficient oocytes. The profound downregulation of multiple mitochondrially-encoded subunits of the electron transport chain (e.g., mt-Nd1-6, mt-Cytb, mt-Co1/3), corroborated by reduced phosphate levels in the metabolome, points to a severe impairment of OXPHOS (Fig. 7). This disruption of the core bioenergetic machinery leads to diminished ATP production, increased oxidative stress, and ultimately, the catastrophic failure of the oocyte to acquire developmental competence, leading to arrest at the secondary follicle stage.

Fig. 7. Dysregulation of the oxidative phosphorylation pathway in Agk cKO oocytes.

Pathway map of oxidative phosphorylation illustrating the downregulation of key oxidative phosphorylation components in Agk cKO oocyte. Significantly downregulated genes, including subunits of NADH dehydrogenase (Complex I), Complex III (mt-Cytb), and Complex IV, along with the downstream metabolite phosphate (Pi), are highlighted in red characters and arrows. Created in BioRender. Li, Y. (2026) https://BioRender.com/jgxuu8y.

A critical consequence of electron transport chain (ETC) dysfunction is the overproduction of reactive oxygen species (ROS), which are known to induce follicular arrest and impair fertilization when present in excess^10,39,40. Most ROS is produced by the mitochondrial electron transport chain (ETC) I and III complexes⁴¹. Some studies have displayed that downregulated mt-Cytb resulted in elevated ROS and decreased ATP production^41,42. Consistent with the severe disruption of ETC complexes I and III observed in our study, we detected significantly elevated ROS levels in Agk-deficient oocytes. This finding aligns with previous reports, such as the oocyte-specific deletion of Mfn1, where elevated ROS was directly linked to compromised developmental competence¹⁰.

The profound ATP deficit is another major consequence of OXPHOS failure, and is often associated with meiotic errors⁴³. Notably, despite the severe energy crisis in Agk cKO oocytes, we did not observe any overt defects in chromosomal dynamics during prophase I. This suggests that the developmental arrest is not caused by the activation of a specific meiotic checkpoint, but rather by a more fundamental failure in cellular viability and the inability to support the immense energetic demands of oocyte growth. Metabolomic profiling revealed elevated phospholipids, including several types of PA, PC, and PE. We propose that mitochondrial dysfunction perturbs phospholipid biosynthesis and trafficking, leading to the accumulation of these lipids or their misdistribution. Because PC and PE are major constituents of mitochondrial membranes, shifts in their levels are likely to alter bilayer fluidity, permeability, and stability, thereby compromising the normal operation of the electron transport chain⁴⁴. In parallel, increased ROS generated by impaired mitochondria further damages phospholipid molecules (e.g., via peroxidation), reinforcing a vicious cycle that exacerbates phospholipid metabolic dysregulation⁴⁵. This inferential model warrants follow-up experiments to distinguish primary mitochondrial defects from secondary phospholipid remodeling and to clarify the causal sequence. Therefore, the combination of mitochondrial structural collapse, genomic instability (mtDNA loss), bioenergetic failure (ATP depletion), and severe oxidative stress (ROS accumulation) collectively renders the oocyte incompetent for further development, whether AGK directly or indirectly regulates protein changes in mitochondrial complexes is worth further investigation.

The phenotype of follicular arrest driven by oocyte-intrinsic mitochondrial failure is a convergent endpoint for the disruption of several key regulatory genes. For instance, oocyte-specific deletion of Furin, Mfn1, or Clpp also leads to female infertility through mechanisms involving mitochondrial dysfunction^2,14,30. A key finding of our study is that the transcript levels of these known regulators were unaltered in Agk cKO oocytes. This indicates that AGK likely operates in a distinct or parallel pathway and is not an upstream transcriptional regulator of these other factors. This positions AGK as a novel, non-redundant component of the molecular machinery essential for maintaining mitochondrial homeostasis during folliculogenesis. The elucidation of its specific role provides a new dimension to our understanding of the complex genetic network that safeguards oocyte quality. Furthermore, we cannot rule out the possibility that the molecular changes that occur through nuclear transcription after the absence of AGK may eventually lead to the phenotypic manifestations of oocyte development arrest and mitochondrial dysfunction.

In conclusion, this study identifies AGK as an indispensable, oocyte-intrinsic factor for follicular development and female fertility in mice. We demonstrate that AGK deficiency triggers a catastrophic cascade of mitochondrial failures, encompassing structural disintegration, depletion of the mitochondrial genome, profound bioenergetic collapse, and severe oxidative stress. This multifaceted mitochondrial dysfunction culminates in the arrest of follicular development at the secondary stage, preventing the oocyte from acquiring developmental competence and ultimately causing infertility. Our work delineates a critical axis linking AGK to the maintenance of the mitochondrial transcriptome and bioenergetic fidelity in the female germline. These findings not only expand our fundamental understanding of ovarian physiology but also posit AGK as a candidate gene for investigation in clinical cases of unexplained female infertility potentially linked to oocyte-specific mitochondrial defects.

Methods

Ethical approval

We have complied with all relevant ethical regulations for animal use. All animal experiments were performed in strict accordance with the protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the International Peace Maternity and Child Health Hospital, School of Medicine, Shanghai Jiao Tong University (GKDW-A-2023-11).

Mice were housed under specific pathogen-free (SPF) conditions with a 12 h light/dark cycle and ad libitum access to food and water. Wild-type C57BL/6J mice and oocyte-specific Agk conditional knockout (Agk cKO) mice were used in this study. For fertility assessment, 8-week-old female mice were paired with 2- to 5-month-old males. For other experiments, female mice were sampled at different ages as indicated in the corresponding figure legends. All animals were maintained on a C57BL/6J genetic background. For tissue collection, mice were euthanized by cervical dislocation at the indicated ages. All efforts were made to minimize animal suffering.

Reagents and antibodies

M2 medium (Cat# M7167), mineral oil (Cat# M8410), and Hoechst 33342 (Cat# B2261) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Pregnant mare serum gonadotropin (PMSG, Cat# 11021) and human chorionic gonadotropin (hCG, Cat# 11032) were obtained from Ningbo Second Hormone Factory (Zhejiang, China). Rabbit polyclonal antibody of AGK (Cat#: ab137616), mouse monoclonal antibody of COX IV (Cat#: ab33985), mouse anti-SCP3 antibody (Cat#: ab97672), and rabbit anti-SCP1 antibody (Cat#: ab32363) were purchased from Abcam (Cambridge, MA, USA). Mouse GAPDH monoclonal antibody (Cat#: 60004) and HRP-conjugated secondary antibodies (Cat#: SA-00001 and SA-00002) were purchased from Proteintech (Wuhan, China). Donkey anti-Rabbit/mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ Plus 488/594 (Cat#: A32790, A32766, A21203, and A21207) were bought from ThermoFisher (Waltham, MA, USA).

Western blotting

Tissue lysates were prepared by homogenizing various mouse organs (liver, heart, spleen, kidney, lung, brain, muscle, ovary, uterus, and oviduct) in RIPA lysis buffer (Thermo Scientific, Cat# 89901) supplemented with a protease inhibitor cocktail (Roche, CO-RO). Lysates were incubated on ice for 30 minutes and subsequently clarified by centrifugation at 12,000 × g for 15 minutes at 4 °C. Protein concentration in the supernatant was determined using a BCA assay. Samples (20 μg) were denatured in 6x SDS loading buffer at 95 °C for 5 min. Proteins were separated by electrophoresis on 10% (w/v) sodium dodecyl sulfate-polyacrylamide gels and transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Cat# IPFL00010). Membranes were blocked in 5% non-fat milk in TBST for 1 h at room temperature and then incubated with primary antibodies (at a dilution of 1:1000 to 1:5000) overnight at 4 °C. Following washing, membranes were incubated with HRP-conjugated secondary antibodies (1:5000 dilution) for one hour at room temperature. Bands were visualized using an ECL Plus Western Blotting Detection System (GE Healthcare).

Immunofluorescence (IF) and immunohistochemistry (IHC)

Paraffin-embedded ovarian sections (5 μm) were heated at 60 °C for 1 hour, dewaxed in xylene, and rehydrated through a graded ethanol series. Antigen retrieval was performed by incubating slides in Tris-EDTA buffer (pH 9.0) in a microwave oven for 20 minutes. For IF, sections were permeabilized with 0.2% Triton X-100 for 15 min and blocked with 10% normal donkey serum (NDS) for 30 min. For IHC, endogenous peroxidase activity was quenched with 3% H₂O₂ for 10 min, followed by blocking with 10% NDS for 1 h. Sections were incubated with primary antibodies (IF, 1:100; IHC, 1:200) overnight at 4 °C. For IF, slides were washed and incubated with Alexa Fluor-conjugated secondary antibodies (1:1000) and Hoechst 33342 for 1 hour. For IHC, slides were incubated with a biotinylated secondary antibody and a streptavidin-enzyme conjugate, followed by development with a DAB substrate-chromogen mixture and counterstaining with hematoxylin. All slides were mounted with an anti-fade mounting medium and imaged using a Leica DMI8 confocal microscope. For quantification of IHC signal intensity, the immunohistochemical images were first imported into Image J and followed by setting the optical density calibration. Then, Regions of Interest (ROIs) are outlined, and the staining intensity and area within the ROIs are converted into Integral Optical Density (IOD) values. The IOD is then divided by the area of the ROI to obtain the average optical density per unit area. Finally, the average optical density values from the two groups are subjected to statistical analysis.

Establishment of Agk conditional knockout mouse model

A conditional allele for Agk (Agk^flox) was generated by flanking exon 3 with loxP sites via embryonic stem (ES) cell targeting in a C57BL/6J background. Chimeric mice were generated and bred with C57BL/6J mice to establish the Agk^flox/+ line, which was subsequently intercrossed to obtain homozygous Agk^flox/flox mice.

To generate oocyte-specific conditional knockout (Agk cKO) mouse, Agk ^flox/flox females were crossed with Dppa3-Cre male mice (C57BL/6Smoc-Dppa3^{em1 (IRES-Cre) Smoc}, Shanghai Model Organisms Center, NM-KI-00040), which express CRE recombinase in PGCs from E7.25. The resulting Agk ^flox/-; Dppa3-Cre offspring were then crossed with Agk^flox/flox mice to produce the experimental genotype, Agk^flox/flox; Dppa3-Cre (Agk cKO). Littermate Agk^flox/flox females lacking the Dppa3-Cre allele served as controls. Genotyping was performed by PCR using primers listed in Supplementary Table 1.

Fertility test

Eight-week-old Agk cKO and control females were continuously housed with wild-type C57BL/6 J males (2 females:1 male) for an 8-week period. Cages were monitored daily, and the number of litters and pups per female was recorded.

RNA extraction and quantitative RT-PCR (RT-qPCR)

Total RNA was extracted from whole ovaries using Super FastPure Cell RNA Isolation Kit (Vazyme, RC102-01). cDNA was synthesized using the HisyGo RT Red SuperMix for qPCR (+gDNA Wiper) (Vazyme, RT101-01). RT-qPCR was conducted using ChamQ Blue Universal SYBR qPCR Master Mix (Vazyme, Q312-02) on an ABI 7500 Real-Time PCR system. Gene expression was normalized to Gapdh or β-actin and calculated using the 2-ΔΔCT method. All experiments were performed in replicate. Primers are listed in Supplementary Table 2.

Hormone quantification by ELISA

Serum levels of follicle stimulating hormone (FSH), luteinizing hormone (LH), anti-Mullerian hormone (AMH), testosterone, and estradiol (E₂) were quantified using commercial ELISA kits (AMH: CSB-E13156m, Testosterone: CSB-E05101m, Cusabio; LH: E-EL-M3053, E2: E-OSEL-M0008, Elabscience; FSH: JLC2953, Gelatins) according to the manufacturer’s methods. In brief, the ELISA plate was prepared with blank wells, standard sample wells, and sample wells, and incubated for 90 minutes. Then, the wells were incubated with biotinylated antibody for 60 minutes at 37°C. After washing, the wells were incubated with HRP enzyme conjugate working solution for 30 minutes at 37 °C. Subsequently, the TMB substrate solution was added to wells for 15 minutes at 37 °C in the dark. The absorbance was analyzed with an ELISA reader and calculated for the concentration according to the standard curves.

Histology and follicle counting

Ovaries from 2-, 4-, and 8-week-old mice were fixed in 4% paraformaldehyde, paraffin-embedded, and serially sectioned at 5 µm thickness. Every fifth section was mounted and stained with hematoxylin and eosin (H&E). Follicles were classified as primordial/primary (Prf + PF), early secondary (<2 granulosa cell layers, ESF), late secondary (≥2 granulosa cell layers, LSF), antral (AF), or corpus luteum (CL). All follicles in the selected sections were counted to determine the distribution for each ovary.

Estrous cycle staging

Vaginal smears were collected daily at 9:00 AM for 16 consecutive days from 8-week-old female mice. Smears were stained with Giemsa, and the stage of the estrous cycle (proestrus, estrus, metestrus, diestrus) was determined by cytological analysis.

Superovulation analysis

3-week-old female mice were injected with 10 IU PMSG, followed by 10 IU hCG 48 h later. The cumulus-oocyte complexes were collected in M2 medium 16 h post-hCG. As no cumulus-oocyte complexes were obtained from Agk cKO mice, we collected the ovaries from 3-week-old female mice injected with PMSG for 48 h, to observe the follicle development potency by exogenous gonadotropin.

Oocyte collection and in vitro maturation

Germinal vesicle (GV)-stage oocytes were collected from the ovaries of 3-week-old females by puncturing antral follicles in M2 medium⁴⁶. Because Agk cKO ovaries yielded predominantly denuded oocytes due to follicular arrest at the early secondary stage with incomplete cumulus cell layers, we selected naked oocytes of similar diameter from both control and Agk cKO groups to ensure comparable experimental conditions. For in vitro maturation, GV oocytes were cultured in droplets of M2 medium under mineral oil at 37 °C in 5% CO₂ for 14–16 h. Maturation was assessed by scoring for germinal vesicle breakdown (GVBD).

Ovarian chromosome surface spreading for meiotic Prophase I analysis

Ovaries were collected from female mice (E18.5 - P5). Then transferred to a hypotonic extraction buffer (30 mM Tris-HCl, 50 mM sucrose, 17 mM trisodium citrate dihydrate, 5 mM EDTA, 0.5 mM DTT, pH 8.2) and incubated for 30 minutes. Chromosome spreading was performed on glass slides pre-coated with 1% paraformaldehyde (pH 9.2) containing 0.15% Triton X-100 and air-dry in a humid chamber overnight. Slides were blocked at 4 °C in blocking buffer (10% normal goat serum, 3% BSA, 0.05% Triton X-100 in PBS). Primary antibodies, anti-SCP3 to mark the lateral elements of the synaptonemal complex and anti-SCP1 to label the central element, were applied overnight at 4° (1:200, ab15090, ab97672, Abcam). Secondary antibodies stained for 1 hour at 37 °C. Chromatin was counterstained with Hoechst. Images were captured using a Leica DMI8 confocal microscope. Meiotic prophase I substages (Leptotene, Zygotene, Pachytene, and Diplotene) were determined based on the morphology of the synaptonemal complex as revealed by the SCP3 and SCP1 staining pattern.

Smart-seq3 RNA sequencing for oocytes

Single naked oocytes were isolated and processed for library construction using a modified Smart-seq3 protocol by Abiosciences (Beijing, China). Detailed protocols can be found at: Smart-seq3xpress.protocols.io: https://www.protocols.io/view/smart-seq3xpress-yxmvmk1yng3p/v3. Briefly, Oocytes were lysed in lysis buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, supplemented with protease inhibitor cocktail (Roche, 04693132001) and phosphatase inhibitor cocktail (Roche, 04906845001). cDNA was generated by reverse transcription and pre-amplified by PCR. Libraries were sequenced on an Illumina NovaSeq 6000 platform. Raw reads were processed using zUMIs (v2.8.2), aligned to the mouse genome (GRCm38) with STAR (v2.7.3), and quantified. Differential expression analysis was performed using the Seurat package in R. Genes with an absolute log2 fold change ≥ 1 and a p-value < 0.05 were considered differentially expressed. Gene Ontology (GO) and KEGG pathway enrichment analyses were conducted using Metascape⁴⁷. The sequencing data have been deposited in the Sequence Read Archive (SRA) under accession number PRJNA1223809.

Mitochondrial assays

Oocytes were incubated with JC-1 dye (Beyotime, C2003S) at a final concentration of 10 μg/ml for 20 min at 37 °C. The ratio of red (J-aggregates) to green (J-monomers) fluorescence was quantified using a confocal microscope and ImageJ software. Intracellular ROS levels were measured by incubating oocytes with 10 µM DCFH-DA (Beyotime, S0033S) for 20 min at 37 °C. Fluorescence intensity was quantified using a confocal microscope and ImageJ. Total DNA was extracted from single oocytes, and mtDNA content was quantified using a mouse mtDNA SYBR qPCR Kit (Tianjingsha, 14-141) against a standard curve. ATP levels in oocytes were measured using a luciferase-based enhanced ATP assay kit (MeilunBio, MA0440) on a luminometer, with quantification against a standard curve.

Transmission electron microscopy

Ovaries were fixed in 2.5% glutaraldehyde, post-fixed in 1% osmium tetroxide, dehydrated in ethanol, and embedded in resin. Ultrathin sections were stained with uranyl acetate and lead citrate and examined with a transmission electron microscope. Images were captured at 11,000× and 26,000× magnification.

Spatial metabolomics

The spatial metabolome was performed according to the steps provided by Applied Protein Technology Company (Shanghai, China). In brief, the post-collection ovarian tissue of 4-week-old mice was rapidly embedded in carboxymethylcellulose sodium salt (CMC) and sectioned into frozen slides with 10 μm thickness. Matrix-assisted laser desorption/ionization-mass spectrometry imaging (MALDI-MSI) was performed by TimsTOF fleX MADLI 2 (Bruker Daltonics, Bremen, Germany) to generate spatially resolved metabolic profiles with a focus on compounds under 1000 Da and lipids. The mass spectrometer was calibrated and set to scan a predefined mass range, capturing a broad spectrum of metabolites with a specified spatial resolution of 20 µm. Raw data underwent processing for spatial variation correction, baseline subtraction, and normalization using specialized software, with metabolite identifications confirmed against reference libraries or standards. For oocyte metabolic analysis, we circle each oocyte of the early secondary follicle and extract data from this region. Statistical analyses, including t-tests and ANOVA, were conducted to identify significant differences in metabolite levels. The web server for metabolomic data analysis, MetaboAnalyst (https://www.metaboanalyst.ca), was used to analyze metabolomics data. Spatial metabolomics data are available in the METASPACE under this link: https://metaspace2020.org/project/Agk-mouse.

Statistical analysis

Data are presented as mean ± SD from at least three independent experiments unless otherwise noted. Statistical significance between two groups was determined using a two-tailed Student’s t-test. Analyses were performed using GraphPad Prism 9. Significance was defined as p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Data were representative of at least three independent experiments unless otherwise specified. The sample size was calculated based on this website: https://hedwig.mgh.harvard.edu/sample_size/size.html.

Author contributions

T.L. and Y.L. Investigation, analysis, and writing; F.G., C.T., S.Z., and J.D. Methodology; Y.Z., W.L., N.W., and P.Z. Conceptualization, analysis, and writing. All authors made substantial contributions in critically revising the manuscript and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks Carla Tatone, who co-reviewed with Teresa Vergar, and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Rupinder Kaur and Kaliya Georgieva. A peer review file is available.

Data availability

All data generated or analyzed during this study are included in this published article and its Supplementary Information files. The uncropped and unprocessed western blot images are provided in the Supplementary Materials file (Supplementary Fig. 6). The numerical source data underlying all graphs and charts in the main figures are provided in the Supplementary Data file. Sequencing data from the oocyte-RNA-seq experiment were available in the Sequence Read Archive (SRA) under accession number PRJNA1223809. Spatial metabolomics data are available in the METASPACE database under the project name “Agk-mouse” (https://metaspace2020.org/project/Agk-mouse).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10073-7.