Mechanisms of mitochondrial dysfunction in premature ovarian insufficiency

Abstract

Premature ovarian insufficiency (POI) is a clinical syndrome characterized by the decline of ovarian function in women under the age of 40, and its core pathological feature is the irreversible depletion of the ovarian reserve. Mitochondria serve as the central hub for energy metabolism and signal integration in ovarian cells, and their dysfunction drives POI progression through multiple cell death pathways and signaling cascades. This review systematically examines the multidimensional mechanisms by which mitochondrial damage leads to follicle depletion and proposes, for the first time, an integrated regulatory model termed “mitochondrial damage – follicle fate decision.” The review further identifies contradictory evidence, model limitations, and gaps in clinical translation within current research. Regarding therapeutic strategies, we evaluate the current status, safety risks, and ethical barriers of mitochondrial nutrients, stem cell therapy, gene editing, and mitochondrial replacement therapy. Finally, we propose a future research framework centered on mitochondrial dynamics and quality control, emphasizing the need for multi-omics integration and personalized interventions.

Introduction

Premature ovarian insufficiency (POI) is a clinical syndrome characterized by the decline of ovarian function in women under the age of 40. Its primary manifestations include menstrual disorders such as oligomenorrhea or amenorrhea, elevated gonadotropin levels, and reduced estradiol levels [1]. Epidemiological data indicate a rising prevalence of this condition over the past two decades, with a current global prevalence of approximately 3.5% among women. In terms of etiological composition, iatrogenic factors account for about 11.2% of cases, while autoimmune factors contribute to approximately 10.5% [2]. POI has emerged as a significant clinical issue affecting female fertility [3]. Accumulating evidence suggests that the pathogenesis of POI is frequently accompanied by excessive apoptosis of ovarian granulosa cells (GCs) and oocytes, with mitochondrial dysfunction identified as a critical trigger of this apoptotic process [4]. Therefore, investigating the underlying mitochondrial mechanisms holds substantial significance for the prevention and treatment of POI.

Mitochondria generate ATP via oxidative phosphorylation, supplying the energy required for most cellular processes [5]. As highly conserved semi-autonomous organelles, mitochondria also participate in maintaining intracellular calcium homeostasis [6], regulating cell differentiation, and playing pivotal roles in key biological processes such as programmed cell death [7, 8].

Within ovarian physiology, mitochondria perform multiple essential functions. They not only provide the necessary energy substrate for oocyte maturation, fertilization, and early embryonic development, but also regulate the metabolic activity of granulosa cells and support steroid hormone biosynthesis, thereby directly influencing follicular growth and overall ovarian function [9]. Healthy mitochondria tightly regulate the balance between energy production and ROS generation [10]., which is crucial for ensuring oocyte quality and preserving ovarian reserve.

Conversely, mitochondrial dysfunction initiates a cascade of pathological responses: insufficient ATP production leads to disruptions in cellular energy metabolism, while excessive accumulation of ROS induces oxidative stress damage, subsequently aberrantly activating apoptotic signaling pathways. These alterations lead to premature follicular depletion, reduced oocyte quality, and accelerated ovarian decline. Together, they represent key mechanisms driving reproductive endocrine disorders such as POI [11]. Fig. 1 summarizes the four primary functions of mitochondria in the ovary.

Fig. 1.

Mitochondrial functions in the ovary. The diagram illustrates four functional roles of mitochondria in the ovary: (1) energy supply; (2) calcium signaling dynamics; (3) steroid hormone synthesis; (4) support of the oocyte

Methods

This study systematically searched the literature available up to 2026 in databases including PubMed and Web of Science, covering recent advances in genetics, biochemistry, and clinical research. The inclusion criteria comprised in vitro and in vivo experimental studies, clinical trials, and review articles. The main search keywords included: “premature ovarian insufficiency”, “mitochondrial dysfunction”, “ovarian reserve”, “granulosa cell death”, “follicle activation”, “apoptosis”, “ferroptosis”, “PANoptosis”, along with their corresponding English terms. This study followed the standard procedures for systematic reviews, performed a graded assessment of the strength of evidence for the included literature, and provided a critical integration and analysis of the relevant mechanisms and therapeutic strategies in the review.

Roles of mitochondria in ovarian physiology

Functions of mitochondria in oocytes

Energy supply

Within the cytoplasm of oocytes, mitochondria exhibit a distinct morphology. They predominantly appear as spherical structures with a diameter of ≤ 1 μm, containing only a few short cristae that enclose an electron-dense matrix [12]. Despite this seemingly simplified architecture, these mitochondria remain capable of performing oxidative phosphorylation and serve as the primary source of ATP production in oocytes and during early embryonic development [13].

From a metabolic perspective, oocyte growth and maturation primarily rely on the utilization of pyruvate via oxidative phosphorylation. During the resumption of meiosis, metaphase I oocytes demonstrate significantly higher pyruvate consumption compared to immature prophase I oocytes and fully matured metaphase II oocytes [14]. This is further supported by studies using oocyte-specific knockout of Pdha1, a gene encoding a catalytic subunit of the pyruvate dehydrogenase complex [15], which confirms the essential role of this metabolic pathway in successful oocyte maturation. It is noteworthy that key enzymes involved in glycolysis are highly expressed in the surrounding cumulus cells but are present at very low levels within the oocyte itself [16].

High-quality oocyte development—including processes such as meiotic spindle assembly, accurate chromosome segregation, protein translation, intracellular signaling, and calcium homeostasis maintenance—is highly dependent on a consistent and sufficient supply of high-energy substrates, primarily ATP [9]. Recent studies have revealed that total ATP content significantly declines with oocyte aging, indicating a progressive decrease in overall metabolic efficiency [17]. In murine models, aged oocytes exhibit impaired dynamic regulation of ATP during fertilization, which directly compromises their developmental potential [18]. Collectively, this evidence underscores that normal mitochondrial function and efficient energy provision are critical for sustaining oocyte viability and ensuring female fertility.

Mitochondrial regulation of calcium signaling dynamics

Mitochondria are fundamental in modulating intracellular calcium (Ca²⁺) homeostasis, primarily through their capacity for regulated Ca²⁺ uptake. Functioning as dynamic Ca²⁺ buffers, mitochondria attenuate cytosolic Ca²⁺ fluctuations and shape the spatiotemporal profile of Ca²⁺ signals [19]. Within the mitochondrial matrix, Ca²⁺ acts as a key activator of oxidative phosphorylation by simultaneously upregulating tricarboxylic acid cycle dehydrogenase activity, enhancing electron transport chain efficiency, and directly stimulating F₀/F₁ ATP synthase to promote ATP synthesis [20].

During mammalian fertilization, sperm entry triggers periodic cytosolic Ca²⁺ oscillations that are essential for oocyte meiotic resumption and embryonic activation. These oscillations are transmitted to mitochondria, where they stimulate metabolic activity. Conversely, disruption of mitochondrial oxidative phosphorylation reduces ATP availability, impairing Ca²⁺ reuptake mechanisms and leading to aberrant endoplasmic reticulum Ca²⁺ release, thereby destabilizing cellular Ca²⁺ homeostasis. Thus, sustained mitochondrial ATP production is critical for maintaining low resting Ca²⁺ levels in oocytes and for supporting prolonged Ca²⁺ oscillations post-fertilization [21]. In metaphase II-arrested oocytes, mitochondrial oxidative phosphorylation constitutes the principal ATP source, underscoring its indispensability for maintaining developmental competence prior to fertilization.

Mitochondrial function in granulosa cells

Steroid hormone synthesis

Mitochondria serve as a central hub for the efficient synthesis of estradiol [22]. Acting as the core organelle for steroidogenesis, mitochondria not only supply ATP via oxidative phosphorylation for the biosynthetic reactions, but also generate the essential reducing equivalents (NADPH) required for cytoplasmic hydroxylation reactions through the electron transport chain and associated metabolic pathways, such as the malate-aspartate shuttle [23]. The rate-limiting step of steroid synthesis – the conversion of cholesterol to pregnenolone by P450scc enzyme (CYP11A1) on the mitochondrial inner membrane – occurs directly within this compartment [24]. This critical step is tightly regulated by the Steroidogenic Acute Regulatory protein(StAR) [25]. The formation of functional microdomains via mitochondria-associated endoplasmic reticulum membranes (MAMs) facilitates the efficient transport and metabolic coordination of substrates like cholesterol [26]. Thus, by integrating multiple functions including “energy supply, generation of reducing power, housing the rate-limiting reaction, and providing a membrane platform for enzyme complexes,” mitochondria constitute an irreplaceable control center for the estrogen synthesis pathway in granulosa cells. The functional integrity of mitochondria directly dictates the output of steroid hormones and, consequently, the normal development of follicles.

Support of the oocyte

During mammalian follicular development, cumulus granulosa cells and the oocyte form a tightly coupled metabolic syncytium via gap junctions (primarily composed of Connexin 37 and Connexin 43 proteins) traversing the zona pellucida [27]. Within this structure, granulosa cells efficiently convert glucose to pyruvate and lactate through highly active glycolysis (the “Warburg effect”) [28]. Concurrently, their mitochondria generate reduced glutathione (GSH) via the glutamate-cystine antiporter and glutathione synthesis enzymes (e.g., glutamate-cysteine ligase and glutathione synthetase) [29]. These key metabolites (energy substrates and antioxidants) are transferred directly into the oocyte cytoplasm through the gap junction channels [16]. For the oocyte, which has a relatively limited intrinsic capacity for aerobic oxidation, these imported energy substrates serve as the primary source for mitochondrial oxidative phosphorylation and ATP synthesis [30]. Recent research further reveals the exquisite precision of this metabolic coupling. For instance, oocyte-secreted paracrine factors (such as GDF9 and BMP15) provide feedback to modulate glycolytic gene expression in granulosa cells [31, 32]. Moreover, the key mitochondrial regulatory protein SIRT3 in granulosa cells maintains glycolytic flux and pyruvate output by deacetylating and activating the AMPK-PFK2 signaling axis [33].

Mechanisms of mitochondrial damage-induced ovarian reserve depletion

Mitochondrial damage contributes to ovarian reserve depletion through multiple mechanisms, including the induction of granulosa cell death, aberrant activation of primordial follicles, disruption of redox homeostasis, and impairment of mitochondrial quality control [34]. This review proposes a conceptual framework termed the “mitochondrial damage-follicle fate decision” model. The core hypothesis is that the severity of mitochondrial damage determines the specific fate of follicles through three key parameters: the intensity of reactive oxygen species (ROS) burst, the degree of mitochondrial membrane potential collapse, and the compensatory efficiency of mitophagy.

When mitochondrial membrane potential declines and the electron transport chain becomes dysfunctional, excessive ROS are released into the cytoplasm. These ROS subsequently attack mitochondrial membrane lipids, proteins, and mitochondrial DNA, leading to further collapse of membrane potential and impairment of ATP synthesis. In parallel, ROS activate apoptotic and other cell death signaling pathways [34]. These interconnected events establish a vicious cycle that exacerbates mitochondrial injury and accelerates ovarian reserve depletion.Table 1 lists the studies on the impact of mitochondrial dysfunction on ovarian reserve.

Table 1.

Research on the impact of mitochondrial dysfunction on ovarian reserve

Drugs/Targets	Pathway of Action	References	Model	Outcomes
Resveratrol, Ex-527	SIRT1/P53/BAX	[35]	3-NPA-treated GCs, POI mouse model	Resveratrol suppressed apoptatic markers (P53, BAX, CASPASE3), restored BCL2 expression, mitigated ROS-induced POI
Intravenous administration of mitochondria	Apoptosis	[36]	Cyclophosphamide-induced POI mice	Intravenous mitochondria administration improves ovarian function, increases mitochondrial activity, reduces apoptosis rate
MSC-Mito + PQQ	SIRT1/ATM/p53	[37]	Cyclophosphamide + Busulfan-induced POI mice	Combined therapy improves ovarian function, increases mitochondrial biogenesis, inhibits DNA damage-mediated apoptosis
LARS2	LARS2/E2F1/MFN2	[38]	POI patient GCs	LARS2 downregulation inhibits GC proliferation, promotes apoptosis, induces mitochondrial dysfunction
BMSCs + Moxibustion	PI3K/Akt	[39]	Cyclophosphamide-induced POI rats	Combined therapy reduces ROS levels, upregulates mitochondrial membrane potential, inhibits excessive mitophagy
SIRT3	SIRT3/CPT2	[33]	Goat granulosa cells	Enhanced mitochondrial function, promoted GC proliferation, inhibited apoptosis
Mfn2	Mfn2/Bax	[40]	Cisplatin-induced POI mice	Mfn2 low expression linked to mitochondrial damage and ovarian tissue apoptosis
Esculentoside A	PPARγ /BCL-2	[41]	Cyclophosphamide-induced POI mice	Restored ovarian structure, suppressed GC apoptosis, promoted folliculogenesis
TMAO	TMAO/NF-κB	[42]	POI mice	TMAO triggered GC apoptosis via mitochondrial pathway, causing ovarian insufficiency
FOXJ2/	FOXJ2/MCU	[43]	Amh-cre; Foxj2tg/tg mice	Reduced fertility, hormonal abnormalities, follicular atresia, POI phenotype
MRPL50	MRPL50 deficiency	[44]	Patient cells/Drosophila	Autosomal recessive syndromic POI with hearing loss and renal/cardiac dysfunction
hUMSCs	hUMSCs/GSK3β	[45]	Cyclophosphamide-induced POI mice	hUMSCs restored ovarian function by regulating mitochondrial dynamics
Hypoxic MSC-Exos	SIRT3/PGC-1α	[46]	POI rats	Hypoxic preconditioning enhanced therapeutic effect via mitochondrial improvement

The mitochondrial apoptosis pathway

The mitochondrial apoptosis pathway is among the most extensively studied mechanisms, with substantial evidence supporting its role in POI pathogenesis.The imbalance between BCL-2 and BAX, leading to cytochrome c release and caspase cascade activation, has been repeatedly validated in various POI animal models. Under pathological conditions, the BCL-2/BAX ratio decreases due to reduced anti-apoptotic proteins (BCL-2, BCL-XL) and increased pro-apoptotic proteins (BAX, BIM) [47]. This allows BAX to translocate to the mitochondrial outer membrane and oligomerize into pores, causing membrane potential collapse and cytochrome c release. Cytochrome c then forms the apoptosome with Apaf-1, activating caspase-9 and downstream caspase-3, which ultimately leads to DNA fragmentation, nuclear breakdown, and apoptotic cell death [48].Wenqi Chen and colleagues further demonstrated that downregulation of Mfn2 reduces BCL-2 expression and increases BAX levels, thereby promoting cytochrome c release and caspase cascade activation, suggesting that Mfn2 may be a potential therapeutic target for POI [40]. However, whether Mfn2 exerts consistent effects across different etiologies of POI, such as chemotherapy-induced, autoimmune, and idiopathic forms, and what upstream signals regulate Mfn2 expression remain unclear. Clinical sample analyses have shown a trend of decreased BCL-2 expression and increased BAX expression in ovarian tissues from POI patients, although some studies have reported no significant change in BCL-2 levels [49]. This discrepancy suggests possible compensatory regulation by other BCL-2 family members, such as MCL-1 and BID. These contradictory findings indicate that current mechanistic models have not fully incorporated the fine-tuning of BCL-2 family functions by post-translational modifications, and also suggest that the molecular characteristics of the apoptotic pathway may be heterogeneous across different etiologies of POI.

Mitochondrial calcium overload

The calcium overload axis promotes the opening of the mitochondrial permeability transition pore through the FOXJ2-MCU(mitochondrial calcium uniporter)-Ca²⁺mt pathway, and functional impairment of mitochondria-associated endoplasmic reticulum membranes can further amplify calcium signaling disturbances [43]. The transcription factor FOXJ2 is significantly upregulated in POI patients. FOXJ2 directly binds to the promoter region of the mitochondrial calcium uniporter, enhancing its transcriptional activity and leading to marked upregulation of MCU protein expression. This promotes excessive calcium influx into the mitochondrial matrix, disrupts intramitochondrial calcium homeostasis, and induces severe calcium overload. Mitochondrial calcium overload subsequently triggers aberrant opening of the mitochondrial permeability transition pore, dissipates mitochondrial membrane potential, and activates the mitochondrial apoptosis pathway. MCU knockdown or administration of the MCU inhibitor Ru360 effectively reverses this process [43]. Mitochondria-associated endoplasmic reticulum membranes are specialized contact sites between the endoplasmic reticulum and mitochondria that facilitate precise calcium transfer. Under physiological conditions, calcium released through inositol 1,4,5-trisphosphate receptors subsequently enters the mitochondrial matrix via voltage-dependent anion channels. Functional impairment of these contact sites may result in either excessive or insufficient calcium release, thereby disrupting mitochondrial membrane potential and triggering cell death pathways [50, 51]. These findings provide a theoretical basis for MCU-targeted therapeutic strategies.

However, it should be noted that although the FOXJ2-MCU pathway has been validated in specific POI subtypes, such as FOXJ2 overexpression models, its expression variation frequency and clinical relevance in human POI patients lack validation from large-scale cohort studies. The in vivo application of the MCU inhibitor Ru360 faces challenges related to specificity and toxicity, and clinical translation remains distant.

PANoptosis

PANoptosis is a novel form of programmed cell death that integrates features of apoptosis, pyroptosis, and necroptosis, but its level of evidence in POI is still in the early stages. Mitochondrial dysfunction leads to the escape of mitochondrial DNA into the cytoplasm, where newly synthesized mitochondrial DNA serves as a key trigger for NLRP3 inflammasome activation [52]. Once assembled, the NLRP3 inflammasome recruits and activates caspase-1, promoting the release of pro-inflammatory cytokines such as IL-1β and IL-18, which subsequently induce pyroptosis. In POI models, inflammasome activation resulting from mitochondrial damage not only triggers pyroptosis but also engages apoptotic pathways through caspase-8, ultimately manifesting as a mixed PANoptosis phenotype that drives severe granulosa cell depletion [53].

Recent studies have identified potential therapeutic interventions targeting these pathways. Proanthocyanidins have been shown to inhibit cellular senescence through modulation of the Sirt1-p53-p21 signaling pathway, indirectly reducing inflammasome activation and delaying POI progression [54].

However, this mechanism has currently been supported only by a few in vitro experiments, and its actual contribution in the in vivo physiological context requires further validation. It remains a relatively speculative emerging mechanism. Although mitochondrial DNA leakage and NLRP3 inflammasome activation have been confirmed in various diseases, the correlation between circulating or local ovarian mitochondrial DNA levels and disease severity in POI has only been reported in small-sample studies, and potential confounding effects of other damage-associated molecular patterns have not been excluded [53]. Direct evidence for PANoptosis in POI mainly relies on transcriptomic analysis and limited histological observations from single studies, and functional validation through gene knockout or specific inhibitors is still lacking.

Ferroptosis

Research on the ferroptosis pathway also has evident limitations. Ferroptosis is a form of programmed cell death associated with iron-dependent accumulation of lipid peroxidation, and its morphological features include reduction or disappearance of mitochondrial cristae, rupture of the mitochondrial outer membrane, and mitochondrial membrane condensation [55]. Mitochondria serve as central hubs for intracellular iron metabolism and reactive oxygen species production. Damaged mitochondria exhibit abnormal accumulation of free iron, while disruption of the electron transport chain leads to excessive reactive oxygen species production. Excess reactive oxygen species promote iron participation in the Fenton reaction, generating hydroxyl radicals that initiate lipid peroxidation of cellular membranes [56, 57]. In POI models, glutathione levels are significantly reduced and GPX4 activity is decreased, impairing the cellular capacity to eliminate lipid peroxides [58]. hnRNPA2B1 binds to and stabilizes SLC7A11 mRNA. SLC7A11 is a core component of the glutathione antioxidant system and functions to inhibit the accumulation of lipid peroxides, thereby protecting granulosa cells from ferroptosis [59]. However, existing data are mainly derived from in vitro granulosa cell experiments, and large-scale cohort studies confirming the association between key markers such as GPX4 and SLC7A11 and the clinical phenotype of POI in human samples are lacking.

Current evidence for ferroptosis in POI comes primarily from cisplatin- or D-galactose-induced animal models, and its presence and contribution in natural aging or autoimmune POI remain unclear. Although decreased GPX4 activity and glutathione depletion have been reported, the interactions between these changes and other pathways such as apoptosis and autophagy, including whether cross-activation or compensation occurs, have not been systematically evaluated. Furthermore, the finding that hnRNPA2B1 inhibits ferroptosis by regulating SLC7A11 currently exists only in a single cell-based study and lacks validation through in vivo loss-of-function experiments. Fig. 2 summarizes the pathways described above.

Fig. 2.

Cell death pathways triggered by mitochondrial damage. Abbreviations: BCL-2,B-cell Lymphoma 2; BAX, BCL-2-Associated X Protein; FOXJ2,Forkhead Box J2; MCU, Mitochondrial calcium uniporter; MPTP, Mitochondrial permeability transition pore; mtDNA, Mitochondrial DNA; NLRP3,NOD-like Receptor Family Pyrin Domain Containing 3; IL-1β,Interleukin-1 beta; IL-18,Interleukin-18; Fe²⁺,Iron ion; ROS, Reactive oxygen species; GPX4,Glutathione Peroxidase 4

Accelerated primordial follicle activation

Accelerated primordial follicle activation is another key mechanism by which mitochondrial damage leads to depletion of the ovarian reserve. When mitochondrial membrane potential declines or the electron transport chain is impaired, excessive reactive oxygen species are released into the cytoplasm. These reactive oxygen species directly oxidize the active site of PTEN, inducing conformational changes that inactivate the phosphatase [60]. PTEN inactivation results in elevated levels of PI(3,4,5)P₃ at the plasma membrane, which recruits and activates downstream AKT kinase. AKT subsequently phosphorylates and inhibits TSC2, thereby relieving its suppression of Rheb GTPase and activating mTORC1. As a central regulator of cellular growth and metabolism, mTORC1 promotes protein synthesis and lipid biosynthesis while suppressing autophagy. The activation of mTORC1 enhances anabolic metabolism in both oocytes and granulosa cells, driving the transition of primordial follicles into primary follicles. Under normal conditions, the FOXO3A transcription factor resides in the nucleus to maintain follicle quiescence. Upon activation of the PI3K/AKT signaling pathway, AKT phosphorylates FOXO3A, leading to its nuclear export and subsequent inactivation, thereby relieving the suppressive effect on follicle activation [61, 62]. Accumulation of mitochondrial reactive oxygen species also induces cytoskeletal remodeling, particularly F-actin polymerization, which is thought to inhibit the activity of the Hippo pathway core kinases LATS1/2 [63]. Inactivation of LATS1/2 prevents phosphorylation of YAP/TAZ, allowing their nuclear translocation. Once in the nucleus, YAP binds to TEAD family transcription factors and upregulates the transcription of growth-related genes, particularly CTGF and cyclin D, which collectively promote granulosa cell proliferation and follicle growth [64]. In addition, anti-Müllerian hormone is expressed in growing follicles and specifically localized to granulosa cells, where it serves as an inhibitor of primordial follicle activation. Mitochondrial damage impairs granulosa cell function, leading to reduced anti-Müllerian hormone secretion. Decreased anti-Müllerian hormone levels relieve the inhibition of primordial follicle activation, resulting in excessive follicular recruitment [65]. However, activated follicles fail to sustain development under conditions of insufficient metabolic support, leading to large-scale apoptosis rather than maturation into competent follicles. Given that the primordial follicle pool is finite, approximately six to seven million at birth, premature activation leads to a sharp decline in the number of remaining follicles, ultimately manifesting as premature depletion of the ovarian reserve [62]. Fig. 3 summarizes the pathways described above.

Fig. 3.

Signaling pathways regulating follicle activation. Abbreviations: PI3K, Phosphoinositide 3-Kinase; AKT, Protein Kinase B(PKB); PTEN, Phosphatase and Tensin Homolog; mTOR, Mammalian Target of Rapamycin; FOXO3A, Forkhead Box O3; YAP, Yes-Associated Protein; TEAD, TEA Domain Transcription Factor; CTGF, Connective Tissue Growth Factor; AMH, Anti-Müllerian hormone; GCs, Granulosa cells

Most of the evidence for this pathway comes from conditional gene knockout mouse models, but the “global activation” seen in these models differs from the “activation followed by death” observed in POI. In POI, mitochondrial damage simultaneously provides both an activation signal and an energy crisis: activated follicles cannot complete development due to insufficient ATP and eventually undergo apoptosis. This means that any intervention aimed at promoting follicle activation, such as PTEN inhibitors, may be counterproductive in POI unless mitochondrial function is enhanced at the same time.

Mitochondrial damage as the central hub determining follicle fate

Based on the above findings, this review proposes an integrated model in which mitochondrial dysfunction serves as the initial insult and, through three interconnected axes involving reactive oxygen species, calcium homeostasis imbalance, and mitophagy inhibition, cross-regulates multiple cell death pathways and follicle fate decisions. Reactive oxygen species are both a direct product of mitochondrial damage and an upstream trigger that activates apoptosis, ferroptosis, and inflammatory signals, forming a self-amplifying loop in which reactive oxygen species lead to mitochondrial membrane potential collapse and subsequent apoptosis initiation. The calcium overload axis promotes the opening of the mitochondrial permeability transition pore through the FOXJ2-MCU-Ca²⁺mt pathway, and functional impairment of mitochondria-associated endoplasmic reticulum membranes can further amplify calcium signaling disturbances, creating positive feedback with the apoptotic pathway [43]. Inhibition of mitophagy leads to the accumulation of damaged mitochondria, which exacerbates reactive oxygen species production and the buildup of cytotoxic metabolites. Mitophagy selectively eliminates damaged mitochondria, thereby maintaining metabolic homeostasis and reducing oxidative stress [66]. The PINK1/Parkin pathway is the primary mechanism governing this quality control system. PINK1 is rapidly degraded in healthy mitochondria but accumulates on the outer membrane of damaged mitochondria, where it recruits Parkin, an E3 ubiquitin ligase. Activated Parkin catalyzes polyubiquitination of mitochondrial outer membrane proteins, facilitating recognition by autophagy receptors and subsequent encapsulation of damaged mitochondria within autophagosomes [67]. mTOR serves as a core kinase that integrates nutrient sensing and autophagic suppression. In POI, particularly in chemotherapy-induced models, the PI3K/AKT/mTOR axis is frequently activated, which inhibits normal autophagic flux and leads to the accumulation of damaged mitochondria [68]. Activated mTOR further suppresses mitophagy by inhibiting the expression of autophagy-related genes, resulting in intracellular accumulation of damaged mitochondria and triggering cellular senescence signals. Cells subsequently enter a senescence-associated phenotype characterized by cell cycle arrest and secretion of senescence-associated secretory phenotype factors, accelerating ovarian tissue aging [69]. This mechanism is best supported by evidence in chemotherapy-induced POI, but the regulatory network in idiopathic POI remains to be elucidated. This integrated model emphasizes that the various pathways do not exist in isolation but rather form a complex regulatory network through cross-talk nodes such as reactive oxygen species, calcium signaling, and the PI3K/AKT/mTOR axis. Therefore, targeting a single pathway may be insufficient to achieve satisfactory therapeutic effects, and future treatment strategies should focus on coordinated modulation of multiple pathways. Fig. 4 provides a cohesive overview of the interconnected mechanisms.

Fig. 4.

Mitochondrial damage affects the ovarian reserve. Abbreviations: ROS: Reactive oxygen species; PTEN: Phosphatase and Tensin Homolog; mTORC1:Mammalian Target of Rapamycin Complex 1; LATS1/2:Large Tumor Suppressor Kinase 1/2; YAP: Yes-Associated Protein; TEAD: TEA Domain Transcription Factor; CTGF: Connective Tissue Growth Factor; PINK1:PTEN-induced putative kinase 1; AMH: Anti-Müllerian hormone; BAX: BCL-2-Associated X Protein; BCL-2:B-cell Lymphoma 2; MCU: Mitochondrial calcium uniporter; FOXJ2:Forkhead Box J2; MPTP: Mitochondrial permeability transition pore; NLRP3:NOD-like Receptor Family Pyrin Domain Containing 3

Mitochondria-targeted therapeutic strategies and future directions

Mitochondrial nutrients and antioxidants

Preclinical and clinical studies have investigated mitochondrial nutrients and antioxidants as interventions for POI. CoQ10 deficiency has been suggested as a potential risk factor for POI, as the serum CoQ10 to total cholesterol ratio is significantly lower in POI patients than in healthy controls. Pretreatment with CoQ10 has been shown to improve ovarian response and embryo quality in patients with diminished ovarian reserve, possibly by restoring mitochondrial energy metabolism and reducing oxidative stress [70]. Alpha-lipoic acid and N-acetylcysteine have also demonstrated protective effects in POI models by modulating signaling pathways and replenishing antioxidant defenses [71, 72]. However, these studies have limitations, including small sample sizes, lack of placebo controls, low oral bioavailability, and non-standardized dosing. The serum CoQ10 to total cholesterol ratio as a POI risk marker has not been validated in large cohorts. Therefore, these nutrients should only be considered as adjunctive options.

Lifestyle and metabolic interventions

Caloric restriction and exercise are potential adjunctive approaches for addressing mitochondrial dysfunction. Caloric restriction upregulates PGC-1α, increasing mitochondrial biogenesis and reducing reactive oxygen species production. Aerobic exercise stimulates AMPK and PGC-1α, potentially enhancing mitochondrial function [73, 74]. However, no randomized controlled trials have validated the effects of these interventions on ovarian reserve in POI patients. Moreover, excessive caloric restriction may worsen osteoporosis, a concern given the low estrogen levels in POI patients [75]. Thus, only general health advice can be given, and these interventions should not be recommended as specific treatments for POI.

Gene editing

Gene editing using CRISPR/Cas9 could theoretically correct nuclear gene mutations (such as LARS2, CLPP, and MRPL50) that cause mitochondrial dysfunction in POI [38, 44, 76, 77]. However, this approach faces major safety and ethical barriers for a non-life-threatening condition. Off-target effects may cause unintended cuts in tumor suppressor genes, posing cancer risk. Mosaicism from embryonic editing introduces genetic uncertainty. Delivery systems such as adeno-associated virus vectors have limited cargo capacity and insertional mutagenesis risk, while non-viral vectors have low transfection efficiency in ovarian cells. Additionally, Cas9 may mistakenly cut mitochondrial DNA due to sequence homology. Therefore, applying gene editing to POI treatment remains premature [78, 79].

Mitochondrial replacement therapy

Mitochondrial replacement therapy aims to replace defective maternal mitochondria with healthy donor mitochondria at the egg or embryonic stage [80]. Since 2019, the European Society of Human Reproduction and Embryology has explicitly opposed its clinical use [81]. Key concerns include mitochondrial heteroplasmy, where residual mutant DNA may selectively amplify and cause late-onset disease years after birth. A large randomized trial showed that autologous mitochondrial transfer did not improve embryo euploidy rates and was terminated early [82]. Therefore, this therapy is highly experimental and should not be used for POI.

Stem cell therapy

Mesenchymal stem cell therapy has shown reduced follicular atresia in animal models, potentially through paracrine effects or mitochondrial transfer. However, major concerns remain [45, 46, 83]. Tumorigenicity is the most serious risk, as long-term studies have shown potential for malignant transformation [84, 85]. Immune rejection from allogeneic sources has not been fully resolved. The mechanism of action remains unclear—whether benefits come from mitochondrial donation, paracrine factors, or inflammation suppression—leading to poor treatment standardization. Most studies have short follow-up periods insufficient to detect long-term tumors. In the absence of large, randomized, double-blind, controlled trials with long-term follow-up, stem cell therapy for POI should be viewed with caution.

Clinical translation: current status and future directions

Current clinical biomarkers and their limitations

At present, no mitochondria-specific functional marker is available for clinical use in POI. Serum anti-Müllerian hormone remains the most valuable clinical indicator [86], but it reflects the number of growing follicles rather than mitochondrial health. Circulating mitochondrial DNA copy number shows bidirectional changes in POI patients, with elevated levels indicating damage and reduced levels suggesting mitochondrial depletion, and no standardized threshold has been established. Mitochondrial stress-related factors such as GDF15 and FGF21 are elevated in POI, but their correlations with traditional markers such as FSH and antral follicle count are weak. Future prospective cohort studies should combine measurements of the oxidized to reduced glutathione ratio in follicular fluid, mitochondrial membrane potential in granulosa cells, and serum CoQ10 levels to establish a multi-parameter mitochondrial function score. Such a score could help predict POI progression and guide individualized treatment.

The translational gap from animal models to humans

The majority of current mechanistic studies on POI are based on acute chemically induced models, such as those using VCD, cyclophosphamide, or D-galactose, or on ovarian hyperstimulation models in young mice. These models differ significantly from human POI, particularly the idiopathic or autoimmune forms. The human follicular pool, approximately one million at birth, is much smaller than that of mice, which is about five thousand, leading to different sensitivities to injury. Baseline mitochondrial dynamics parameters, such as the Drp1 to Mfn2 ratio, also differ between mice and humans. Furthermore, human POI progresses slowly over months to years, whereas acute models complete their course within days. Therefore, positive results from animal models cannot be directly extrapolated to humans. Future research should prioritize the use of human ovarian cortex xenograft models or patient-derived organoids to bridge this translational gap.

Future research priorities

Multicenter prospective cohorts should be established to systematically assess mitochondrial function parameters and biomarkers in POI patients across different etiologies, including chemotherapy-induced, autoimmune, and idiopathic forms. Human ovarian tissue culture combined with lentivirus-mediated gene intervention should be used to directly validate the relative contributions of various pathways in the human ovary, rather than continuing to accumulate low-evidence animal data. Ovarian-targeted delivery systems for mitochondrial antioxidants, such as CoQ10 encapsulated in nanoparticles, should be developed to improve bioavailability and tissue specificity.

Conclusion

This review has systematically and critically examined the mechanisms by which mitochondrial damage leads to depletion of the ovarian reserve in POI. We have proposed for the first time an integrated network model termed “mitochondrial damage – follicle fate decision.” Based on evidence strength, we have classified apoptosis and excessive follicle activation as confirmed pathways, ferroptosis and PANoptosis as emerging pathways, and MAMs-related calcium signaling disruption as a hypothetical mechanism. We have clearly identified existing contradictions in the literature, such as non-consistent changes in the BCL-2 to BAX ratio, evidence gaps including the lack of human data and the uncertain authenticity of PANoptosis in POI, and therapeutic misconceptions, such as the inappropriateness of mitochondrial replacement therapy for POI and unresolved risks associated with stem cell treatment. From a clinical translation perspective, current evidence relies excessively on acute mouse models and lacks prospective cohort studies in POI patients as well as standardized mitochondrial functional biomarkers. Future efforts should prioritize human tissue validation, organoid models, and rigorously designed clinical trials to move mitochondrial-targeted interventions from mechanistically interesting to clinically effective.

Abbreviations

POI

Premature ovarian insufficiency

GCs

Granulosa cells

ATP

Adenosine triphosphate

ROS

Reactive oxygen species

NADPH

Nicotinamide adenine dinucleotide phosphate hydrogen

CYP11A1

Cytochrome P450 family 11 subfamily A member 1

StAR

Steroidogenic acute regulatory protein

MAMs

Mitochondria-associated endoplasmic reticulum membranes

GSH

Glutathione

GCL

Glutamate-cysteine ligase

GS

Glutathione synthetase

GDF9

Growth differentiation factor 9

BMP15

Bone morphogenetic protein 15

SIRT3

Sirtuin 3

AMPK

AMP-activated protein kinase

PFK2

6-Phosphofructo-2-kinase

BCL-2

B-cell lymphoma 2

BCL-XL

B-cell lymphoma-extra large

BAX

BCL-2-associated X protein

BIM

BCL-2 interacting mediator of cell death

ΔΨm

Mitochondrial membrane potential

Apaf-1

Apoptotic protease activating factor-1

DNA

Deoxyribonucleic acid

Mfn2

Mitofusin 2

AQP

Aquaporin

DBT

Danggui Buxue Tang

TMAO

Trimethylamine N-oxide

PTEN

Phosphatase and tensin homolog

AKT

Protein kinase B (PKB)

MCU

Mitochondrial calcium uniporter

FOXJ2

Forkhead box J2

MPTP

Mitochondrial permeability transition pore

ER

Endoplasmic reticulum

IP3Rs

Inositol 1,4,5-trisphosphate receptors

ER

Endoplasmic reticulum

IP3Rs

Inositol 1,4,5-trisphosphate receptors

VDACs

Voltage-dependent anion channels

mtDNA

Mitochondrial DNA

NLRP3

NOD-like receptor family pyrin domain containing 3

IL-1β

Interleukin-1 beta

IL-18

Interleukin-18

Sirt1

Sirtuin 1

p53

Tumor protein p53

p21

Cyclin-dependent kinase inhibitor 1A

Fe²⁺/Fe³⁺

Iron ion

GPX4

Glutathione peroxidase 4

hnRNPA2B1

Heterogeneous nuclear ribonucleoprotein A2/B1

SLC7A11

Solute carrier family 7 member 11

PI3K

Phosphoinositide 3-kinase

mTOR

Mammalian target of rapamycin

TSC2

Tuberous sclerosis complex 2

mTORC

Mammalian target of rapamycin complex 1

FOXO3A

Forkhead box O3

LATS1/2

Large tumor suppressor kinase 1/2

YAP

Yes-associated protein

TAZ

Transcriptional coactivator with PDZ-binding motif

TEAD

TEA domain transcription factor

CTGF

Connective tissue growth factor

AMH

Anti-Müllerian hormone

SOD

Superoxide dismutase

MAPK

Mitogen-activated protein kinase

HO-1

Heme oxygenase-1

Keap1

Kelch-like ECH-associated protein 1

IKK

IκB kinase

TNF-α

Tumor necrosis factor-α

IL-6

Interleukin-6

MT1

Melatonin receptor 1A

MT2

Melatonin receptor 1B

PGC-1α

Peroxisome proliferator-activated receptor gamma coactivator 1-alpha

CoQ10

Coenzyme Q10

PINK1

PTEN-induced putative kinase 1

p62

Sequestosome-1

OPTN

Optineurin

ATGs

Autophagy-related genes

TRIM28

Tripartite motif containing 28

LC3B-II

Microtubule-associated protein 1 light chain 3 beta

ALA

Alpha-lipoic acid

cAMP

Cyclic adenosine monophosphate

NAC

N-acetylcysteine

CR

Caloric restriction

CLPP

Caseinolytic mitochondrial matrix peptidase proteolytic subunit

CLPB

Caseinolytic mitochondrial matrix peptide chaperone B

tRNA

Transfer ribonucleic acid

LARS2

Mitochondrial leucyl-tRNA synthetase

MRPL50

Mitochondrial ribosomal protein L50

CRISPR

Clustered regularly interspaced short palindromic repeats

Cas9

CRISPR-associated protein 9

hUMSCs

Human umbilical cord mesenchymal stem cells

GSK3β

Glycogen synthase kinase-3 beta

Drp1

Dynamin-related protein 1

MSCs

Mesenchymal stem cells

Authors’ contributions

Author contributions All authors contributed to the study conception and design. Chenyu Jia conceptualized the idea, edited the entire draft, and prepared the manuscript. Investigations were performed by Huihui Li, Ruotong Ju, Puhua Zhang. Project administrations were performed by Li Peng, Tingting Xue, Xinyu Zhu, Shu Wang. Validations were performed by Jiali Luo, Ruixiang Zhu. Financial support is performed by Xuan Jing (Corresponding Author), Xiangrong Cui (Corresponding Author). All authors read and approved the final manuscript.

Funding

This research was funded by The Natural Science Foundation of Shanxi Province, China (No. 202403021221275 to X.J.); Scientific Research Project of Shanxi Provincial Health Commission (No. 2025QM021 to XR.C); The Major Special Research Project on High-Quality Development of Health Undertakings in Shanxi Province (No. SXWSJKZD25011 to XR.C); Shanxi Scholarship Council of China (No. 2025-225 to X.J.); Special Research Project for High-quality Development of Health Care in Shanxi Province (No. 14001); Sanjin Talents Program for Young Top-notch Talents in Healthcare (No. SJYC2025500); The Natural Science Foundation of Shanxi Province (No. 202403021221265 to CY.H); Science and Technology Innovation Project of Colleges and Universities, Department of Education of Shanxi Province (No. 2023L097 to to CY.H); Research Project of Shanxi Provincial Health Commission (No. 2021114 to to CY.H).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.