Mitochondrial dysfunction in Alzheimer's disease: Guiding the path to targeted therapies

Abstract

Alzheimer's disease (AD) is characterized by progressive neurodegeneration, marked by the accumulation of amyloid-β (Aβ) plaques and tau tangles. Emerging evidence suggests that mitochondrial dysfunction plays a pivotal role in AD pathogenesis, driven by impairments in mitochondrial quality control (MQC) mechanisms. MQC is crucial for maintaining mitochondrial integrity through processes such as proteostasis, mitochondrial dynamics, mitophagy, and precise communication with other subcellular organelles. In AD, disruptions in these processes lead to bioenergetic failure, gene dysregulation, the accumulation of damaged mitochondria, neuroinflammation, and lipid homeostasis impairment, further exacerbating neurodegeneration. This review elucidates the molecular pathways involved in MQC and their pathological relevance in AD, highlighting recent discoveries related to mitochondrial mechanisms underlying neurodegeneration. Furthermore, we explore potential therapeutic strategies targeting mitochondrial dysfunction, including gene therapy and pharmacological interventions, offering new avenues for slowing AD progression. The complex interplay between mitochondrial health and neurodegeneration underscores the need for innovative approaches to restore mitochondrial function and mitigate the onset and progression of AD.

Introduction

Alzheimer's disease (AD) is the most common form of neurodegeneration, affecting an estimated 6.9 million Americans aged 65 and older, with a projected increase to 13.8 million by 2060 [1]. The initial clinical symptoms of AD vary among individuals but often include a decline in non-memory cognitive functions such as word-finding, visuospatial issues, and impaired reasoning or judgment [2]. As the disease progresses to its later stages, symptoms include significant cognitive decline, difficulty communicating, and the need for full-time assistance with daily personal care, which places a heavy burden on caregivers and family members [2]. In 2024, the projected cost of AD and other neurodegenerative diseases in the United States is estimated to be $360 billion, potentially rising to nearly $1 trillion by 2050 [1]. The substantial human and economic toll underscores the urgent need for advancements in the understanding, prevention, and treatment of AD.

In recent decades, extensive efforts have been made to elucidate the mechanisms underlying AD pathogenesis. Beyond the well-known Amyloid and Tau hypotheses [3,4], increasing attention is directed toward the mitochondrial hypothesis. Mitochondrial integrity and function are compromised early in the disease, both in the context of amyloidogenesis and tauopathy, highlighting a potentially convergent pathway in AD development [5]. This early mitochondrial dysfunction is now recognized as a critical factor in the progression of AD, contributing to synaptic failure, neuroinflammation, lipid disturbance, oxidative stress, and neuronal loss [[5], [6], [7], [8]]. Therefore, understanding mitochondrial abnormalities in AD could provide crucial insights into the disease's pathogenesis, offering new therapeutic targets for the development of effective treatments.

Mitochondria are dynamic organelles essential for cellular energy production via oxidative phosphorylation (OXPHOS) and ATP generation, particularly crucial for the high energy demands of neuronal cells [9,10]. Normal mitochondrial function encompasses energy production, calcium homeostasis, metabolism, and redox regulation, all vital for cellular growth, proliferation, and survival [11,12]. During OXPHOS, mitochondria also produce and eliminate reactive oxygen species (ROS). Excessive ROS can impair protein function and trigger inflammatory responses, leading to neuronal death [13]. Therefore, under stress or disease conditions, mitochondrial quality control (MQC) mechanisms are necessary to maintain mitochondrial function. MQC includes the mitochondrial unfolded protein response (UPR^mt), which rescues protein homeostasis [14]; mitochondrial dynamics and transportation, which maintain mitochondrial morphology [15,16]; and mitophagy pathways, which eliminate damaged mitochondria [17]. Collectively, these mechanisms ensure the preservation of mitochondrial integrity and functionality, thereby supporting overall cellular health and resilience in the face of physiological and pathological challenges.

In AD, these MQC mechanisms become compromised, leading to mitochondrial dysfunction. Impairments such as defective mitochondrial import [18], excessive activation of UPR^mt [19,20], overproduction of ROS [21], disruptions in mitochondrial dynamics like fragmentation [22], and impaired mitophagy result in the accumulation of damaged mitochondria [23]. These abnormalities destabilize the mitochondrial genome, causing bioenergetic failure and the release of mitochondrial DNA (mtDNA), which further exacerbates AD progression [24,25]. Given the key role of mitochondria in neuronal health, these MQC impairments highlight the importance of targeting mitochondrial function in AD. Recent therapeutic strategies have focused on restoring MQC, showing promise in mitigating mitochondrial dysfunction and improving overall cellular health. This review summarizes the molecular basis of MQC, encompassing its mechanistic pathways, physiological functions, and pathological relevance in AD, and discuss potential MQC-targeted therapeutic strategies for AD.

Mitochondrial proteostasis and AD

The mitochondrial proteostasis system refers to the maintenance of protein homeostasis within the mitochondria. Mitochondria are unique in their separation of proteome and genome from the rest of the cell by inner and outer membranes. They contain 1100–1300 proteins, but only ∼37 of these proteins are encoded by the mtDNA [26,27]. This indicates that the majority of mitochondrial proteins are translated in the cytoplasm and imported into the mitochondria. Both mitochondrial and nuclear-encoded proteins work together to sustain mitochondrial functions, including energy homeostasis and metabolism [28,29]. The import of mitochondrial precursor proteins, surveillance, and regulation of mitochondrial proteins are crucial for maintaining mitochondrial protein homeostasis. Nuclear-encoded mitochondrial precursor proteins are synthesized by cytosolic ribosomes and have mitochondrial targeting sequences, with 70 ​% residing within the N-terminus [[30], [31], [32]]. The mitochondrial protein import system includes several complexes which include TOMM, MIM, SAM, MIA, TIMM23, TIMM22, and PAM [33]. TOMM is the most critical as it first contacts the majority of nuclear-encoded mitochondrial proteins, facilitating their entry into the intermembrane space [33]. The precursor proteins are recognized by TOMM20, transferred to TOMM22, and then translocated through the TOMM40 channel into the mitochondrial matrix (Fig. 1a). Once inside, the targeting sequence is proteolytically removed by mitochondrial processing protease (MPP) [33]. Chaperones such as mitochondrial heat shock protein 70 (mtHsp70) and mtHsp60 (chaperonin) then fold the proteins [34]. The mature proteins are recognized by specific machinery like the TIMM or MIA complexes, which escort them to their destinations. This intricate import system is essential for maintaining mitochondrial function and cellular energy homeostasis, underscoring the complexity and precision of mitochondrial protein import.

Fig. 1.

Mitochondrial protein import and the mitochondrial unfolded protein response. (a) Nuclear-encoded mitochondrial precursor proteins are imported into the mitochondria via the TOM complex. Precursor proteins with matrix-targeting presequences are further translocated by the TIM23 complex, where mitochondrial processing peptidase (MPP) removes the presequences, and chaperones like mtHsp70 fold the proteins into their mature forms (b) Mitochondrial proteostasis involves chaperones (e.g., mtHsp). and proteases (e.g., Lon peptidase 1, LONP1) to manage misfolded and damaged proteins. When overwhelmed, the mitochondrial unfolded protein response (UPR^mt) is activated, triggering stress signals. (c) Kinases JNK and PKR activate transcription factor c-Jun, which binds to AP-1 elements and induces CHOP and C/EBPβ transcription. The CHOP–C/EBPβ complex binds to CHOP elements flanked by MURE1 and MURE2, promoting UPR^mt gene expression. (d) The integrated stress response is activated through UPR^mt. Kinases GCN2 and PERK phosphorylate eIF2α, leading to the nuclear translocation of eIF2α, which upregulates the translation of CHOP, ATF4, and ATF5, thereby promoting stress response genes. (e) In AD, amyloid-β (Aβ) peptides inhibit the import of nuclear-encoded mitochondrial precursor proteins via the TOM complex. f. Aβ is imported into the mitochondria via the TOM complex. Aβ accumulates in the OMM or be imported into the IMM via the TIM complex. Although Aβ can be degraded by mitochondrial presequence peptidase (PreP), it may inhibit PreP function. The accumulation of Aβ peptides aggregates triggers UPR^mt activation.

The mitochondrial proteostasis system includes chaperones and proteases to manage misfolded and damaged proteins. Chaperones such as mtHsp10 [35], mtHsp40 (DNAJ) [36], mtHsp60 [37], mtHsp70 [38,39], mtHsp90 [40], mtHsp100 [41], and small heat shock proteins (sHsp) [42] recognize and refold misfolded proteins. Proteases such as Lon peptidase 1 (LONP1) [43], ClpP [44], OMA1 [45], and HtrA serine peptidase 2 (HTRA2) [46] located in the inner mitochondrial space (IMS), along with m-AAA [47], and i-AAA [48] in the IMM, recognize and degrade misfolded and damaged proteins. This refolding and degrading mechanism is essential for a healthy proteome. However, when overwhelmed, chaperones and proteases activate the UPR^mt to alleviate proteotoxicity and re-establish proteostasis [49]. The UPR^mt is a retrograde signal between the mitochondria and nucleus that induces the production of mitochondrial chaperones, proteases, and antioxidant enzymes [50]. It is triggered by accumulation of unfolded proteins, excessive ROS, depletion of mtDNA, and or disruptions in OXPHOS complexes [51] (Fig. 1b). In mammalian cells, UPR^mt activation involves multiple pathways (Fig. 1c). Under mitochondrial stress, the kinases JNK and PKR phosphorylate and activate the transcription factor c-Jun, which binds to AP-1 elements, activating CHOP and C/EBP-β transcription [52]. The CHOP–C/EBP-β complex binds to CHOP elements flanked by mitochondrial unfolded protein response elements MURE1 and MURE2, promoting UPR^mt response genes such as HSP60, ClpP, and mitochondrial import proteins. Additionally, CHOP, ATF4, and ATF5 activate UPR^mt [52]. The integrated stress response (ISR) is activated, leading to eIF2α phosphorylation by GCN2 and PERK kinases, triggered by amino acid depletion, excessive ROS, and ER stress [52] (Fig. 1d). Phosphorylated eIF2α reduces protein synthesis and preferentially translates mRNAs encoding CHOP, ATF4, and ATF5, which induce mitochondrial proteases and chaperones, including HSP60, mtHSP70, and LONP1 [52]. These mitochondrial proteases and chaperones aid in alleviating accumulated misfolded and damaged proteins, helping restore balance to the mitochondrial proteostasis system. Dysfunction in IMS proteostasis can stimulate a protective response mediated by SIRT3 [53,54]. SIRT3 deacetylates and activates FOXO3a, which translocates to the nucleus, acting as a transcription factor to induce antioxidant enzymes such as SOD2, PGC-1α, and CAT [55,56]. In the mitochondria, SIRT3-activated SOD2 converts O₂˙ into H₂O₂, which CAT breaks down into water and oxygen O₂ [55]. This protective mechanism is crucial for mitigating oxidative damage and maintaining cellular redox balance, thereby ensuring mitochondrial and overall cellular health.

The mitochondrial proteome is altered in AD, affecting mitochondrial protein import, mitochondrial proteases, and the UPR^mt. Aβ peptides from the cytosol coaggregate with mitochondrial precursor proteins [57]. These Aβ-coaggregates have their mitochondrial targeting sequence exposed, which is recognized by the TOM complex [58]. These Aβ-coaggregates can either clog or inhibit the TOM complex, leading to dysfunction in mitochondrial protein import, or enter the mitochondria [[18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58]] (Fig. 1e). Once inside, the Aβ-coaggregates are translocated to the IMM by the TIM23 complex [[18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59]].

In addition to Aβ-coaggregates clogging or inhibiting import proteins, Aβ-coaggregates can prevent essential mitochondrial proteins from reaching their destination resulting in mitochondrial dysfunction [18]. Additionally, Aβ peptides with positively charged amino acid residues at the N-terminal cation-binding domain can be recognized and imported into the mitochondria by the TOM complex, leading to the accumulation of Aβ within the IMS (Fig. 1f) [61]. Aβ is also recognized and imported into the IMM by the TIM23 complex, causing further accumulation within both the IMS and IMM [18]. Within the IMM, mitochondrial presequence peptidase (PreP) degrades Aβ peptides [62]. However, Aβ can inhibit PreP's function in degrading presequences and processing protein intermediates after translocation into the IMM [63]. This inhibition results in the accumulation of non-folded mitochondrial preproteins and processing intermediates, decreasing mitochondrial function [63]. Dysfunctional preprotein maturation leads to rapid protein degradation and an imbalanced organellar proteome. Consequently, these disruptions contribute significantly to the pathophysiology of AD.

The relationship between the UPR^mt and AD has only recently been explored. The UPR^mt is activated in both 3- and 9-month-old APP/PS1 mice brains and within SH-SY5Y cells exposed to Aβ treatment [19]. UPR^mt-related genes are upregulated in AD post-mortem prefrontal cortex samples [64]. Additionally, a deficiency in the mitochondrial peptidase PITRM1 induces UPR^mt and increases Aβ accumulation within cerebral organoids generated from human induced pluripotent stem cells (iPSCs) [65]. Pharmacological inhibition of UPR^mt results in a drastic accumulation of Aβ [65]. Therefore, future studies should aim to clarify whether both inhibition and persistent activation of UPR^mt contribute to AD.

Mitochondrial dynamics & transport and AD

Mitochondria are highly dynamic organelles involved in ATP production, cellular metabolism, stress responses, and the maintenance of homeostasis. Mitochondria continuously undergo fission and fusion processes. Mitochondrial dynamics regulate the length, size, and connectivity of mitochondria, contributing to maintain the mitochondrial morphology [66]. Maintain the regular morphology of cristae and mitochondria is crucial to support the cell daily activity. Mitochondrial contact site and cristae organizing system (MICOS) in the IMM is involves in the formation and maintenance of cristae structure [67]. MICOS complex contains Mic60 (Mitofilin), Mic10 (MINOS1), Mic19 (CHCHD3), Mic25 (CHCHD6), Mic23, Mic27, Mic13, and Mic14 (CHCHD10) which form MIB complex with SAM and DnaJC11 across the IMM and OMM [68]. During mitosis, mitochondrial fission generates two daughter mitochondria that inherit the same mtDNA and proteins, ensuring proper function after cell division. Selective removal of damaged or dysfunctional mitochondria during mitosis also supports normal cellular function [69]. During mouse embryogenesis and cell differentiation, mitochondrial morphology and localization change. For example, in neuronal development, mitochondria transition from a fragmented to an elongated morphology, indicating that fusion occurs to shift energy production from glycolysis to OXPHOS. As synapses form, mitochondria translocate to areas requiring more energy, such as presynaptic terminals [70,71]. Under cellular stress, such as bacterial or viral infection, upregulation of mitochondrial fission promotes apoptosis to eliminate damaged cells [72,73]. Mitochondrial dynamics also regulate metabolism under different nutrient conditions: starvation induces mitochondrial fusion to enhance ATP production via OXPHOS, while in nutrient-rich environments, cells undergo mitochondrial fission to reduce ATP production. These processes are tightly regulated by specific proteins to support cellular activity [74].

Mitochondrial fission involves the division of a mitochondrion into two, separating both the outer mitochondrial membrane (OMM) and inner mitochondrial membrane (IMM) without losing soluble proteins from the matrix. The GTPase dynamin-related protein 1 (DRP1) translocates to the OMM, where it interacts with other proteins, such as mitochondrial fission protein 1 (FIS1), mitochondrial dynamics proteins 49 and 51 ​kDa (MiD49 and MiD51), and mitochondrial fission factor (MFF), to mediate OMM separation [[75], [76], [77], [78]]. Phosphorylation of MFF triggers DRP1 recruitment and activates mitochondrial fission. DRP1's function is regulated by various post-translational modifications, including phosphorylation, which controls its GTPase activity. Several proteins regulate mitochondrial fission through the phosphorylation and dephosphorylation of DRP1 at specific sites, such as Ser-616 and Ser-637 [79]. For instance, mitochondrial serine/threonine-protein phosphatase PGAM5 dephosphorylates DRP1 at Ser-637 to induce GTPase activity and promote fission, while cyclin-dependent kinase (CDK) 1/Cyclin B or CDK5 phosphorylates DRP1 at Ser-616 during mitosis to activate fission [80,81]. PTEN-induced kinase 1 (PINK1) and mitogen-activated protein kinase (MAPK1) also phosphorylate DRP1 at Ser-616, promoting mitochondrial fission [82,83].

Mitochondrial fusion involves the merging of the OMM and IMM. The dynamin GTPases mitofusin 1 (MFN1) and mitofusin 2 (MFN2) form heterodimers or homodimers to mediate OMM fusion [71]. Following OMM fusion, IMM and matrix fusion occurs, a critical step during mitochondrial fusion since the matrix contains essential functional components such as mtDNA, metabolites, and proteins. Optic atrophy 1 (OPA1), a dynamin-related GTPase, interacts with cardiolipin to maintain cristae structure. OPA1 is regulated through post-transcriptional modifications and alternative splicing, producing long and short forms of the protein [[74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84]]. The long form promotes mitochondrial fusion and supports cellular homeostasis, while the short form leads to loss of fusion, bioenergetics, and cristae disruption. The balance between mitochondrial fusion and fission is crucial for maintaining cellular homeostasis and activity. An imbalance can cause mitochondrial dysfunction, leading to cell cycle arrest and apoptosis. Mitochondrial dynamics, including fission, fusion, and transport, are involved in processes like mitophagy and distribution, ensuring normal mitochondrial function in healthy cells (Fig. 2a). Accumulation of damaged mitochondria contributes to cell loss and senescence, playing a role in the development of neurodegenerative diseases.

Fig. 2.

Mitochondria dynamic and transport. (a) Mitochondrial Dynamics in Healthy Cells: Mitochondrial dynamics in healthy cells is characterized by a balance between fission and fusion, essential for maintaining cellular homeostasis. During mitochondrial fusion, MFN1 and MFN2, located on opposing OMMs, form homo- or hetero-oligomeric complexes, facilitating the tethering of two mitochondria. Fusion of the inner mitochondrial membrane is mediated by OPA1, resulting in the formation of an enlarged mitochondrion. Conversely, during mitochondrial fission, DRP1 is recruited to the OMM by fission receptors such as FIS1. DRP1 oligomerizes, wrapping around and constricting the mitochondrion, ultimately leading to its division into two daughter mitochondria. (b) In neurons, mitochondria are transported along microtubules from the soma to the synapse through anterograde transport. The outer mitochondrial membrane protein Miro-1 forms a complex with MTX1, MTX2, and KLC-1, facilitating kinesin-1-mediated anterograde transport. Retrograde transport, moving mitochondria back toward the soma, is mediated by a complex of Miro-1, MTX2, and TRAK-1, which interacts with dynein to enable this process. (c) In AD, mitochondrial dynamics are disrupted. The downregulation of MFN1, MFN2, and OPA1 reduces mitochondrial fusion, while the upregulation of DRP1 leads to excessive mitochondrial fission. (d) Mitochondrial transport dysfunction in AD. The hyperphosphorylation of tau leads to tau's disassociation from the microtubule leading to microtubule instability. Microtubule instability leads to damage and disintegration of axonal microtubules impair mitochondrial transport. Accumulation of mitochondria in the soma leads to perinuclear clustering, increasing ROS production. Reduced mitochondrial transport to the synapse contributes to synaptic dysfunction.

Neuronal cells are polarized with long axons and dendrites, making proper mitochondrial distribution essential. Mitochondrial transport along microtubules ensures redistribution to provide sufficient ATP for neurotransmission at synaptic terminals and along axons [85]. The complex responsible for mitochondrial transport includes motor and adaptor proteins (Fig. 2b). Kinesin-related protein (KIF5), dynein, and myosin are motor proteins that facilitate mitochondrial movement along microtubules [85]. The heavy chain of KIF5 has a head domain with ATPase activity that binds to microtubules and a tail domain that binds to cargo. KIF5 transports mitochondria to distal axons and synaptic terminals, while dynein moves them toward the soma [86]. Dynein consists of catalytic heavy chains, intermediate chains, light intermediate chains, and light chains. Like KIF5, dynein's heavy chains serve as motor domains that bind to microtubules [87]. Myosin, in contrast, is responsible for short-distance mitochondrial transport [88]. Dynein directly binds to both mitochondria and microtubules, whereas KIF5 requires adaptor proteins to attach to mitochondria [86]. Several mitochondrial outer membrane (MOM) proteins serve as adaptors. Mitochondrial Rho (Miro), a GTPase, binds to Milton, also known as trafficking of kinesin-binding protein (TRAK), forming a complex that attaches to KIF5's tail domain [89,90]. Miro contains calcium-binding sites regulated by calcium signaling and synaptic activity. Upregulation of Miro enhances complex formation. When mitochondria reach energy-deficient regions, calcium binds to Miro, causing KIF5 to dissociate from microtubules, and mitochondria stabilize in axons via anchoring protein syntaphilin [[89], [90], [91]]. Metaxin 2 (MTX2) bind to the Miro1 and function together with Metaxin 1 (MTX1) to contribute kinesin-mediated mitochondrial movement with KLC1. MTX2/Miro complex also bind to TRAK1 and support dynein-mediated mitochondrial trafficking [92]. Mitochondrial dynamics-related proteins have been shown to interact with Miro proteins, supporting mitochondrial transport. Mitochondrial dynamics are tightly regulated by proteins to maintain regular cellular activity under varying conditions. Thus, an imbalance in mitochondrial dynamics during disease development can lead to abnormal cellular functions.

In the progression of AD, the accumulation of Aβ and hyperphosphorylated Tau alters mitochondrial mass by influencing mitochondrial dynamics, transport and morphology. Imbalances in mitochondrial fission and fusion have been observed in various brain regions associated with AD [[21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93]]. In both AD mouse models and human patients, reduced levels of MFN1 and MFN2 and overexpression of DRP1 have been detected in the hippocampus and cerebral cortex. In the APPSWE/PS1 transgenic mice model, mitochondrial fission is upregulated while mitochondrial fusion is downregulated [93]. The 5xFAD mice model, which overexpresses APP695 with the Swedish (K670 ​N, M671L), Florida (I716V), and London (V717I) familial AD mutations, along with human PS with two familial AD mutations, also shows disturbed mitochondrial dynamics in an age- and gender-dependent manner. For instance, the mitochondrial fusion protein MFN2 decreases in 2- and 8-month-old female mice, while the mitochondrial fission protein MFF is phosphorylated in 6- and 8-month-old 5xFAD mice. FIS1 levels decrease in both male and female mice at 4 months old. DRP1, another key fission protein, decreases in female mice at 2 months old and in males at 4 months old [94]. Tau, a microtubule-associated protein, is a hallmark of neurological disorders known as tauopathies. During aging, elevated Tau levels in mitochondria downregulate mitochondrial fusion proteins like OPA1 [95]. In the Tau P301S transgenic mice model, which is commonly used to study tauopathies, mitochondria become fragmented, and mitochondrial length decreases in neuronal cells. Overexpression of MFN2 specifically in neuronal cells of P301S mice has been shown to attenuate mitochondrial fragmentation, neurodegeneration, synaptic loss, cognitive deficits, and neuronal inflammation [96].

In AD patients and mouse models, mitochondrial fission is significantly influenced, with DRP1 being the most studied protein. DRP1 translocates from the cytosol to the mitochondria, binds to FIS1, and forms a complex that splits the mitochondria [76]. Phosphorylation at Ser-616 and dephosphorylation at Ser-637 activate DRP1, leading to increased mitochondrial fission [79]. Overexpression of DRP1 has been observed in the brain tissue of AD patients [97]. In neuronal cell cultures, phosphorylated Tau and toxic Aβ phosphorylate DRP1 and facilitate its translocation to the mitochondria, disrupting mitochondrial dynamics [98]. Excessive mitochondrial fission in neuronal cells can lead to mitochondrial fragmentation, ultimately inducing cell death and neurodegeneration (Fig. 2c). Impairment of morphology can not only be induced by imbalance of mitochondrial fission and fusion, MICOS impairment also place a role in it. In APP knock in mice, deficiency of CHCHD6 caused MICOS loss, promote APP processing, mitochondrial damage, neuronal cholesterol accumulation, and cognitive deficits [99]. In addition to the imbalance between mitochondrial fission and fusion, dysfunction in mitochondrial transport also affects cellular activity. During AD development, the aggregation of toxic Aβ and hyperphosphorylated Tau promotes disease-like cytoskeletal abnormalities, disrupting the axonal transport of organelles, including mitochondria (Fig. 2d) [85]. Peri-nuclear mitochondrial clustering increased ROS level in the soma which damage the neuronal cell [100]. Also, lack of mitochondria at synaptic terminals and axons leads to synaptic degeneration, contributing to neuronal disorders [101]. Although the exact mechanisms behind mitochondrial transport disorders remain unknown, evidence from both in vitro and in vivo models suggests that stabilizing axonal transport can ameliorate cognitive deficits. In APP/PS1 double transgenic mice, stabilizing axonal transport with drugs can mitigate cognitive deficits [102,103]. In the 5xFAD mice model, decreased tubulin acetylation destabilizes the cytoskeleton, making it more susceptible to damage [104]. Tau is a protein with six isoforms and several phosphorylated species that normally associate with microtubules in mature neurons to maintain cytoskeletal stability. Abnormal Tau leads to the formation of neurofibrillary tangles, disrupting cytoskeletal stability. In various cell and animal models, reduced mitochondrial motility accompanies the aggregation of neurofibrillary tangles [[105], [106], [107], [108], [109], [110]]. Abnormal mitochondrial dynamics can lead not only to cellular loss but also to cellular senescence. In mouse models, downregulation of PGAM5 increases DRP1 phosphorylation at Ser-637, inducing mitochondrial fusion and cellular senescence. Overexpression of mitochondrial fusion prevents the removal of damaged mitochondria within the cell, and the accumulation of damaged mitochondria contributes to cellular senescence in the context of AD [111].

Mitophagy and AD

Mitophagy is the selective degradation of dysfunctional mitochondria via lysosomes, crucial for maintaining mitochondrial health, regulating numbers, and preventing ROS and cellular stress [112,113]. It balances functional mitochondria with autophagic degradation, ensuring homeostasis. During this process, damaged mitochondria are enclosed by a phagophore, forming a mitophagosome, which fuses with a lysosome for enzymatic breakdown into recyclable components [112]. Neurons rely heavily on mitochondria, making mitophagy vital for cellular health. As lysosomes are mainly in the soma, damaged mitochondria must be transported back for degradation via specialized mechanisms [114].

Fig. 3a shows both the PINK1/Parkin dependent and independent pathways. The PINK1/Parkin pathway is a prominent mechanism in mitophagy regulation. PINK1 and Parkin function as the first steps of a signaling pathway that activates mitochondrial quality control pathways in response to mitochondrial damage [115,116]. Under basal conditions, PINK1 is recruited and imported into mitochondria through the TOM-TIM complex, then cleaved by IMM-bound proteases such as matrix processing peptidase (MPP) and prescenilins-associated rhomboid-like protein (PARL) and subsequently degraded by the proteasome, leading to undetectable basal levels of PINK1 [[117], [118], [119], [120], [121]]. However, mitochondrial stressors such as membrane depolarization, mitochondrial complex dysfunction, and proteotoxicity, leading to PINK1 accumulation on the OMM [[112], [113], [114], [115], [116], [117], [118], [119], [120], [121], [122]]. The accumulation of OMM PINK1 leads to subsequent homodimerization and then autophosphorylation thus promoting kinase activation and facilitates the recruitment and binding of Parkin [[17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109], [110], [111], [112], [113], [114], [115], [116], [117], [118]]. Parkin is an E3 ubiquitin ligase that has minimal basal activity under normal conditions due to intramolecular interactions that blocks the active site and compete with E2 ligase binding [123]. Once mitochondrial damage occurs and PINK1 accumulation on the OMM occurs, PINK1 activates Parkin through two mechanisms. The first mechanism occurs through PINK1 phosphorylation of ubiquitin at S65, which competes with an autoinhibitory domain within Parkin and stabilizes it in an active conformation [115]. The second mechanism occurs by PINK1 direct phosphorylation of Parkin on S65 in Parkin's ubiquitin-like domain, which induces conformational changes allowing for the binding of the charged E2 ligase [115]. These two mechanisms increase Parkin's E3 ubiquitin ligase activity. Within the first 2 ​hours of Parkin recruitment and activation, Parkin ubiquitinates many OMM and mitochondrial matrix proteins then targets IMM proteins [115,[124], [125], [126]]. These include cytosolic proteins that can act as mitophagy receptors including p62, OPTEN, NDP52, NBR1, and TAX1BP1 that will recognize and bind to p-S65-Ub proteins [[127], [128], [129]]. The formation of the ubiquitin chains on the mitochondria then leads to the recruitment and binding of autophagy receptors. In PINK1/Parkin-independent mitophagy, phagosomes are recruited directly around mitochondria either via OMM receptors containing LIR motifs (such as NIX/BNIP3L, BNIP3, FUNDC1, BCL2L13, FKBP8, DISC1, AMBRA1, or MCL1) or through recognition of exposed cardiolipin (CL) on the OMM [[130], [131], [132], [133], [134], [135], [136], [137]]. These autophagy proteins lead to the assembly of microtubule-associated protein 1A/1Blight chain 3 (LC3)-positive phagophores. Phagophores sequester damaged mitochondria and deliver them to the lysosomes for degradation via acidic hydrolases [138,139]. This orchestrated process ensures mitochondrial quality control, vital for cellular function and neuronal health.

Fig. 3.

PINK1/Parkin dependent and independent mitophagy (a) Mitochondrial stress, such as mitochondrial membrane depolarization, triggers mitophagy, a degradation process mediated by either PINK1/Parkin-dependent or PINK1/Parkin-independent pathways. In the PINK1/Parkin-dependent pathway, mitochondrial stress leads to the accumulation and stabilization of PTEN-induced putative kinase 1 (PINK1) on the OMM. Stabilized PINK1 recruits the E3 ubiquitin ligase Parkin, which is phosphorylated by PINK1 at the ubiquitin-like domain (S65). This phosphorylation enhances Parkin's E3 ligase activity, leading to the ubiquitination of OMM proteins. Cytosolic mitophagy receptors, such as p62, recognize these polyubiquitinated proteins via their LC3-interacting region (LIR) motifs, facilitating the recruitment of autophagosomes through interaction with LC3-II on the autophagosome membrane. In the PINK1/Parkin-independent pathway, autophagosomes are recruited either through the exposure of cardiolipin (CL) on the OMM or via LIR motif-containing OMM receptors, such as FUNDC1, NIX/BNIP3L, and BNIP3. Following autophagosome engulfment of the damaged mitochondria, mitophagosomes fuse with lysosomes to form mitolysosomes, where damaged mitochondria are degraded by lysosomal acidic hydrolases. (b) Mitophagy impairments in AD. Aβ peptides inhibit mitophagy, characterized by upregulation of PINK1, Parkin, and LC3-II, along with the accumulation of damaged mitochondria. Human phosphorylated tau binds to the OMM, preventing PINK1 stabilization and subsequent recruitment of Parkin, thereby inhibiting mitophagy and leading to the buildup of damaged mitochondria. Additionally, expression of the tau P301L mutation leads to its interaction with Parkin, blocking its translocation to mitochondria. This impaired mitophagy results in the accumulation of damaged mitochondria, increased ROS production, and neuronal death.

Impaired mitophagy in AD has been linked to increased production of ROS, accumulation of mtDNA, proteins, and autophagic vacuoles with oxidative damage, and the presence of damaged mitochondria [23]. Fig. 3b shows both the PINK1/Parkin dependent and independent pathways in AD. In the hippocampus of early-stage AD patients, elevated PINK1 levels have been observed, while late-stage AD is marked by increased Parkin levels, suggesting a reduction in mitophagy flux due to a defect in the initiation of the PINK1/Parkin-dependent pathway [[23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109], [110], [111], [112], [113], [114], [115], [116], [117], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127], [128], [129], [130], [131], [132], [133], [134], [135], [136], [137], [138], [139], [140]]. Additionally, PINK1 mRNA levels are reduced in the hippocampus of late-stage AD patients [141]. Basal mitophagy levels in the hippocampus of postmortem AD patients are reduced by 30–50 ​%, with these patients displaying an accumulation of damaged mitochondria characterized by reduced size, disorganized cristae, and diminished ATP production [142]. There are also defects in the initiation of mitophagy, such as reduced recruitment of activated LC3 to mitochondria, dysfunction in the AMPK signaling cascade, and impaired fusion of mitophagosomes with lysosomes [142]. Several mitophagy-related proteins, including AMBRA1, ATG5, ATG12, BCL-1, BNIP3, BNIP3L, FUNDC1, OPTEN, PI3K class III, ULK1, and VCP/P97, are downregulated in the brains of AD patients [143]. Mitochondrial fractions from late-stage AD brain samples reveal increased p62 levels, an elevated LC3II/I ratio, and decreased PINK1 and Parkin levels, indicating a failure in mitophagy recruitment [144]. In AD patient iPSC-derived neuronal cells, reduced expression levels of LC3 and TFEB lead to a decrease in mitophagy-related proteins [145].

The effects of Aβ peptide on mitophagy depend on the duration of exposure. In rat pheochromocytoma (PC12) cells, a 12 hour treatment with Aβ peptide impairs mitophagy, as indicated by reductions in PINK1, Parkin, BCL-1, and LC3II/I ratios, along with the accumulation of p62 [146]. Conversely, a 24 hour exposure to Aβ peptide in PC12 ​cells results in increased levels of Parkin, BCL-1, and the LC3II/I ratio [147]. Similarly, a 72 hour treatment with Aβ peptide in murine embryonic hippocampal neurons upregulates the expression of PINK1, Parkin, LC3II, and BCL-1, alongside a decrease in mitochondrial membrane potential, reduced ATP production, and increased mitochondrial ROS production [148]. Additionally, treatment of HEK293T cells with oligomeric Aβ initially induces mitophagy, but this is followed by a blockade of the process, characterized by increased levels of Parkin, an elevated LC3II/I ratio, and the accumulation of p62 within the mitochondrial fraction, along with the buildup of autophagosomes and mitochondria [149]. Rats subjected to intracerebroventricular injections of Aβ peptide exhibit reduced levels of PINK1, Parkin, and BCL-1, along with the accumulation of p62 in the hippocampus [150].

Tau has also been shown to interfere with mitophagy. Overexpression of human tau (hTau) induces mitophagy deficits in the hippocampus of hTau transgenic mice, primary hippocampal neurons, and HEK293T cells [151,152]. These deficits are characterized by the accumulation of total and phosphorylated tau on the OMM and an increase in mitochondrial membrane potential, which hinders PINK1 stabilization on the OMM, thereby preventing the recruitment of Parkin. However, overexpression of Parkin can rescue the mitophagy deficits induced by hTau [151]. The tau P301L mutant also inhibits mitophagy in C. Elegans and neuroblastoma cell models by interacting with Parkin's projection domain, preventing its translocation to mitochondria [153]. Contrarily, NH2-Tau fragments in mature hippocampal primary neurons trigger alterations in mitochondrial structure and neuronal synapses while enhancing mitophagy flux by recruiting Parkin to the mitochondria [154]. In both in vitro (overexpression of 2N4R, 1N4R, and 2N3R tau in SH-SY5Y cells) and in vivo (transgenic nematodes expressing human tau, 3xTg-AD mice) models, stimulation of mitophagy reduces tau hyperphosphorylation [[142], [143], [144], [145], [146], [147], [148], [149], [150], [151], [152]]. Collectively, these findings suggest that impaired mitophagy contributes to tau pathogenesis and underscore the distinct pathological features between different models of tau phosphorylation.

Genome instability and AD

Mitochondrial genome stability is crucial for maintaining regular mitochondrial function in healthy cells. Replication of mtDNA is required to maintain stable copy numbers supporting cellular energy demands while transcription of mtDNA into mitochondrial RNA (mtRNA), is required to translate OXPHOS proteins essential for electron transport chain (ETC) and overall mitochondrial bioenergetics (Fig. 4a) [155]. When this stability is compromised, mitochondrial dysfunction can occur, particularly in the context of AD (Fig. 4b). Unlike nuclear DNA, mtDNA lacks extensive non-coding regions and histones, making it more susceptible to mutations [156]. Additionally, mtDNA is located near sites of ROS production, increasing the likelihood of oxidative damage and mutations. Elevated oxidative stress and Aβ accumulation in AD exacerbate mtDNA mutations and damage, impairing the BER mechanism and other repair pathways. This impairment leads to the accumulation of mtDNA mutations, disrupting OXPHOS and resulting in energy failure [157]. Mutation in the mtDNA transcription and translation related protein has also been detected in AD. TFAM is the key protein regulate mtDNA transcription and translation. TFAM will transcript from nucleus, translate in the cytosol, and translocate to the mitochondria. Mutation in TFAM could cause decrease in mtDNA copy number [158]. Mitochondrial transcription factor 4 (MTERF4) is another protein has been altered in APP/PS1 mice [159]. In AD patients, increased oxidative mtDNA mutations and decreased mtDNA copy numbers are observed, contributing to metabolic comorbidities.

Fig. 4.

Mitochondrial genome stability. (a) Mitochondrial DNA (mtDNA) replicates to maintain stable copy numbers, supporting cellular energy demands. mtDNA is transcribed into mitochondrial RNA (mtRNA), which is translated into OXPHOS proteins essential for the electron transport chain (ETC) and overall mitochondrial bioenergetics. (b) In AD, mtDNA copy numbers decrease, and mtDNA becomes more susceptible to oxidative damage. This leads to impaired OXPHOS protein production, resulting in mitochondrial bioenergetic deficits. (c) Increased release of mtDNA through VDAC1 and mPTP into the cytoplasm activates the cGAS-STING pathway. cGAS detects mtDNA, catalyzing the formation of cGAMP, which activates STING. This triggers TANK-binding kinase 1 (TBK1) to phosphorylate interferon regulatory factor 3 (IRF3), promoting the transcription of NF-κB, proinflammatory cytokines, and interferon response proteins.

The release of mtDNA serves as a messenger that facilitates communication between neurons and innate immune cells, such as microglia, during AD progression [160]. In the APP/PS1 mouse model, mtDNA copy numbers decrease, while mtDNA methylation increases [161,162]. The oxidative stress in APP/PS1 mice leads to mtDNA mutations, disrupting transcription and replication [157]. Unlike nuclear DNA, mtDNA lacks modifications, leading cells to recognize it as a foreign molecular, similar to viral or bacterial DNA, which activates the cGAS-STING pathway (Fig. 4c) [163,164]. In AD patients and various mouse models, the accumulation of Aβ plaques and hyperphosphorylated Tau triggers mitochondrial dysfunction, resulting in the release of significant amounts of mtDNA [161,162]. This release activates the cGAS-STING pathway, causing neuronal inflammation primarily within microglia, leading to microgliosis and neuronal loss. Overexpression of the cGAS-STING pathway promotes cell death and further DNA release into the extracellular space. These foreign DNA fragments induce cellular stress in healthy cells, perpetuating the activation of the cGAS-STING pathway [168,169]. Beyond inducing cell death, mtDNA release also contributes to cellular senescence. Senescent cells release a senescence-associated secretory phenotype (SASP), which can convert healthy cells into senescent cells. The accumulation of senescent cells in the brain exacerbates neurodegeneration in AD [170]. Additionally, mtDNA damage directly leads to mitochondrial bioenergetic deficiency, with glucose hypometabolism consistently recognized as an early indicator of disease onset. Since bioenergetic failure in AD has been extensively reviewed elsewhere [[171], [172], [173], [174]], it will not be the focus of this review.

These studies demonstrate that MQC is disrupted in AD in both in vitro and in vivo models. Key processes, including mitochondrial protein import and the UPR^mt are impaired. Aβ peptides disrupt precursor protein import, accumulate in mitochondrial membranes, and overactivate UPR^mt. Altered mitochondrial dynamics, characterized by decreased fusion proteins (OPA1, MFN1, MFN2) and increased fission protein (DRP1), lead to mitochondrial fragmentation. Axonal damage disrupts mitochondrial transport, causing perinuclear clustering and increased ROS in the soma. Reduced mitophagy flux hampers the removal of damaged mitochondria, while mitochondrial genome instability, marked by reduced mtDNA copy numbers, diminishes ETC proteins and activates the pro-inflammatory cGAS-STING pathway. These findings highlight the interconnected nature of mitochondrial dysfunction and its critical importance of therapeutic potential in AD.

Therapeutics Development Enhancing Mitochondrial Function

Gene therapy targeting mitochondrial pathways

Adeno-associated virus (AAV)-based therapies have revolutionized the field of gene therapy, offering a versatile platform for the targeted delivery of therapeutic genes to treat a wide range of genetic and acquired diseases. The ability of AAV vectors to achieve long-term expression with minimal immunogenicity has made them a cornerstone for addressing disorders. Furthermore, their tissue-specific tropism and capacity for modification enhance their therapeutic precision. Despite challenges such as scaling up production, immune responses, and limited payload capacity, ongoing advancements in AAV vector engineering continue to expand their clinical utility, highlighting their transformative impact on modern medicine [183]. In the context of AD, this technique can potentially address mitochondrial dysfunction by enhancing mtDNA stability, bioenergetics, dynamics, and mitophagy. By stabilizing mitochondrial function, reducing oxidative stress, and improving neuronal health, AAV gene therapy holds potential for mitigating neurodegeneration in AD.

Table 1 displays multiple mitochondrial targeted AAV gene therapies within AD models. NDI1, an NADH dehydrogenase located in yeast mitochondria, is integral to the electron transport chain and apoptosis, functioning analogously to mammalian complex I. The introduction of NDH1 can effectively compensate for deficiencies in mammalian mitochondrial complex I [184]. Stereotaxic injection of AAV9-NDI1 into the hippocampus of an AD mouse model (created by Aβ_1-42 injection) improved mitochondrial function, reduced histopathological abnormalities, and alleviated neurological deficits without causing adverse effects in healthy mice [175]. DJ-1 is a multifunctional protein involved in numerous cellular processes, including the regulation of apoptosis and pro-survival signaling, PINK1/parkin-mediated mitophagy, inflammatory responses, protection against oxidative stress through NRF2 activation, mitochondrial dynamics, mitochondrial-ER communication, and chaperone activity [185]. Delivery of AAV9-DJ1 to the hippocampus of 8-month-old APP/PS1 mice improved cognitive function and increased levels of NRF2, LC3, phosphorylated AMPK, and BCL-1 [176]. DJ-1 also enhanced antioxidant defenses, including total superoxide dismutase activity, total antioxidant capacity, and glutathione peroxidase activity [176]. Thioredoxin-1 (Trx-1) is a critical regulator of cellular functions, including redox homeostasis, cell proliferation, DNA synthesis, transcription factor modulation, and apoptosis control [186]. Acting as an antioxidant, Trx-1 directly scavenges ROS. Overexpression of Trx-1 via AAV-Trx-1 delivery to the hippocampus of APP/PS1 mice enhanced learning and memory, reduced hippocampal Aβ deposition, and restored mitochondrial biogenesis by upregulating key proteins such as AMPK, Sirt1, and PGC-1α [177]. PGC-1α is a transcriptional coactivator that plays a pivotal role in cellular energy metabolism, regulating metabolic gene expression, oxidative phosphorylation, and mitochondrial biogenesis [187]. In APP23 transgenic mice, overexpression of PGC-1α via a lentiviral vector delivered to the hippocampus and cortex during the preclinical stage of AD [178]. Four months post-injection treated mice exhibited improved spatial and recognition memory, reduced Aβ deposition, and decreased BACE1 expression. PGC-1α overexpression also attenuated neuroinflammation, lowered proinflammatory cytokine levels, and suppressed microglial activation, preserving pyramidal neurons in the CA3 region and enhancing neurotrophic factor expression [178]. In APP/PS1 mice, AAV-mediated PGC-1α overexpression alleviated deficits in spatial memory, working memory, and sensorimotor gating. Additionally, it restored mitochondrial homeostasis by balancing fission and fusion dynamics, increasing mitochondrial fusion proteins (OPA1, MFN2) while reducing fission proteins (DRP1, FIS1) [179].

Table 1.

Mitochondria AAV based therapeutics in Alzheimer's disease.

AAV-Protein	Protein function	Model	Effects	Reference
AAV9-NDI1	Yeast mitochondrial NADH dehydrogenase; electron transport chain; apoptosis	SH-SY5Y Aβ1-42 ​cell model, Aβ1-42 hippocampal injected mice.	Restored oxidative phosphorylation; reduces ROS production	Li et al., 2024
AAV9-DJ-1	Regulation of apoptosis; PINK1/parkin-mediated mitophagy; Antioxidant response through NRF2 activation	APP/PS1	Enhanced cognitive function; increased levels of antioxidant genes (NRF2, SOD, GPx)	Peng et al., 2023
AAV-Trx-1	Maintains redox homeostasis; Antioxidant ROS scavenger	APP/PS1	Improved learning and memory; Reduced hippocampal Aβ deposition; increased expression of APRK, Sirt1, PGC-1α	Jia et al., 2023
AAV-PGC-1α	Regulator of oxidative phosphorylation; Regulator of mitochondrial biogenesis	APP/PS1	Improved learning and memory; Increase mitochondrial fusion protein (OPA1, MFN2); reduced mitochondrial fission protein (DRP1, FIS1)	Wang et al., 2022
AAV2-CEND1	Neuronal lineage-specific modulator; Regulates mitochondrial dynamics via DRP1	5xFAD	Improved learning and memory; Increase mitochondrial fusion protein (OPA1, MFN2); reduced mitochondrial fission protein (DRP1)	Xie et al., 2022
AAV5-CHCHD6	MICO complex component; Key component for mitochondrial crista formation	APPNL−F-G KI AD	Reduced APP accumulation in the MAM; lowered C99 fragment; improved spatial working memory; restored synaptic function	Shang et al., 2022
AAV9-MICU3	Regulates mitochondrial calcium uptake	Tg-SwDI AD	Improved behavioral performance; enhance cerebral blood flow; reduce Aβ deposition	Zhou et al., 2023
AAV2-PINK1	Mitochondrial protein kinase; Mitophagy	APPSwInd (J20) AD	Alleviated mitochondrial dysfunction and oxidative stress; increased ATP; Increased mitophagy; reduced Aβ accumulation	Du et al., 2017
AAV-MCL1	OMM LIR motif receptor; Mitophagy; Regulates apoptosis	APP/PS1	Improved learning and memory; reduced hippocampal Aβ deposition;	Cen et al., 2020
DISC1	OMM LIR motif receptor; Mitophagy; Mitochondrial dynamics;	APP/PS1	Reduced expression of BACE1; reduced hippocampal Aβ deposition; improved learning and memory	Deng et al., 2016

CEND1 is a neuronal lineage-specific modulator that synchronizes cell cycle exit and differentiation of neuronal progenitors during nervous system development [188]. It is also localized in presynaptic mitochondria, where it regulates mitochondrial dynamics via DRP1 modulation. CEND1 depletion increases mitochondrial fission through DRP1 upregulation [188]. In 5xFAD mice, AAV-mediated overexpression of CEND1 in the hippocampus restored mitochondrial function, alleviated synaptic deficits, and improved learning and memory [180]. This overexpression promoted mitochondrial fusion by upregulating MFN2 and OPA1 levels while downregulating DRP1 expression [180]. Similarly, CHCHD6, a core component of the MICO complex, is critical for mitochondrial crista formation. AAV5-eGFP-CHCHD6 delivery to the hippocampus of APPNL-F-G knock-in Alzheimer's disease mouse models reduced APP accumulation in the MAM and lowered C99 fragment levels, suggesting reduced APP processing [99]. CHCHD6 overexpression also improved spatial working memory and restored hippocampal levels of synaptophysin and PSD95, proteins essential for synaptic function [99].

Mitochondrial calcium uptake 3 (MICU3) regulates mitochondrial calcium uptake by forming heterodimers with MICU1, enhancing MCU-dependent mitochondrial Ca²⁺ uptake [189]. The MICU1-MICU3 heterodimer promotes synaptic flexibility in neuronal cells by increasing mitochondrial calcium uptake at presynapses [189]. In Tg-SwDI transgenic AD mouse models, AAV9-mediated overexpression of MICU3 improved behavioral performance, enhanced cerebral blood flow, and significantly reduced Aβ deposition by modulating Aβ metabolism [181]. MICU3 overexpression also mitigated oxidative stress, mitochondrial dysfunction, ATP depletion, and mtDNA reduction in these models [181]. Targeting mitophagy-related proteins, such as PINK1, MCL1, and DISC1, has demonstrated improvements in cognitive and pathological features of AD. Overexpression of the mitochondrial protein kinase PINK1 via AAV2 delivery to the hippocampus of APPSwInd (J20) mice reduced cerebral and mitochondrial Aβ accumulation [141]. PINK1 overexpression alleviated mitochondrial dysfunction and oxidative stress, evidenced by increased ATP levels and decreased mitochondrial ROS. It also activated autophagy signaling by enhancing the recruitment of autophagy receptors NDP52 and OPTN to damaged mitochondria and increasing levels of the autophagy marker LC3-II [141]. PINK1-mediated autophagy reduced Aβ accumulation, improved synaptic function, and enhanced learning and memory in APPSwInd mice [141]. Similarly, AAV-mediated overexpression of OMM LIR motif receptors MCL1 or DISC1 in APP/PS1 mice produced comparable effects, including improved cognitive function and reduced extracellular Aβ plaque burden [[136], [137], [138], [139], [140], [141], [142], [143], [144], [145], [146], [147], [148], [149], [150], [151], [152], [153], [154], [155], [156], [157], [158], [159], [160], [161], [162], [163], [164], [165], [166], [167], [168], [169], [170], [171], [172], [173], [174], [175], [176], [177], [178], [179], [180], [181], [182]]. Collectively, these findings underscore the potential of gene therapy targeting mitochondrial pathways in AD treatment.

Mitochondria-targeted pharmacological strategy

Pharmacological strategies targeting mitochondrial function offer promising therapeutic avenues for AD. Various reagents have been reported to influence mitochondrial activities, particularly oxidative stress and energy metabolism, as summarized in Refs. [[190], [191], [192]]. However, many of these reagents may only indirectly affect mitochondrial function. This review focuses on pharmacological agents that selectively target mitochondrial proteins to develop AD therapies shown in Table 2.

Table 2.

Mitochondrial focused therapeutic agents in Alzheimer's disease.

Agent	Molecular action	Model	Effects	Reference
MitoQ	Antioxidant	3xTg AD	Reduces ROS levels; reduces Aβ levels; Improves synaptic function	McManus et al., 2011; Young et al., 2019
Szeto-Schiller tetrapeptides 31 (SS-31)	Binds to cardiolipin	APP/PS1	Rescues mitochondrial morphology; reduces ROS levels; Reduces Aβ levels	Mitchell et al., 2020; Jia et al., 2023
Nicotinamide riboside (NR)	NAD+ precursor	APP/PS1	Suppresses cGAS-STING pathway; reduces Aβ levels; Enhances mitophagy	Hou et al., 2021
DA1	Binds to ATAD3A and suppresses its oligomerization	5xFAD	Improves mitochondrial dynamics; reduces Aβ levels	Zhoa et al., 2022
Mdivi-1	Inhibits DRP1	APP/PS1	Improves mitochondrial dynamics; reduces Aβ levels; reduces ROS levels	Baek et al., 2017
P110	Blocks the interaction between DRP1 and FIS1	5xFAD	Improves mitochondrial dynamics; reduces Aβ levels; reduces ROS levels	Joshi et al., 2018
UMI-77	Mitophagy activator; BH3-mimetic for MCL-1	APP/PS1	Enhances mitophagy; reduces Aβ levels; reduces neuroinflammation	Chen et al., 2020
Loganin	Promotes OPTN-mediated mitophagy	3xTg AD	Enhances mitophagy; Improves synaptic plasticity; improves mitochondrial dynamics	Zhou et al., 2023

Mitochondria-targeted antioxidant therapies represent a promising approach to restoring mitochondrial function in AD. Compounds such as MitoQ (ubiquinone) [193], MitoVitE (mitochondrially-targeted vitamin E) [194,195], and SS-31 (a tetrapeptide) [196,197] selectively accumulate within mitochondria, providing greater protection against oxidative damage compared to non-targeted antioxidants. MitoQ, for example, is a modified ubiquinone conjugated to triphenylphosphonium (TPP), allowing it to easily penetrate mitochondrial membranes and accumulate within the IMM. Once localized, MitoQ protects neurons from oxidative damage by converting H₂O₂ into H₂O and O₂, reducing ROS levels, and mimicking the activity of endogenous CoQ [198]. In AD models, MitoQ treatment has been shown to improve memory retention, reduce Aβ accumulation, decrease Tau hyperphosphorylation, and enhance neurite growth [199]. Long-term treatment with the antioxidant peptide SS-31 in APP/PS1 mice reduced mitochondrial ROS, improved synaptic integrity, decreased expression of mitochondrial fission proteins DRP1 and FIS1, and lowered Aβ levels [200]. These findings underscore the potential of mitochondria-targeted antioxidants in mitigating AD pathology and improving cognitive function. Nicotinamide adenine dinucleotide (NAD+), a vital metabolite in human cells, plays a key role in processes such as DNA repair and mitophagy. Hou et al. [201] demonstrated that treatment of APP/PS1 mutant mice with the NAD ​+ ​precursor nicotinamide riboside (NR) for 5 months increased brain NAD ​+ ​levels, reduced proinflammatory cytokine expression, and decreased microglia and astrocyte activation. Elevated cGAS-STING signaling observed in AD mice was normalized by NR treatment, which also induced mitophagy and improved cognitive and synaptic functions in the APP/PS1 mutant mice [201].

ATAD3A is a nuclear-encoded mitochondrial membrane protein that plays a critical role in maintaining MAMs, regulating mitochondrial dynamics, and ensuring mitochondrial genome stability. In various neurodegenerative diseases, including AD, oligomerization and accumulation of ATAD3A have been observed in hippocampal neurons in both patients and mouse models [202,203]. In 5xFAD mouse models and cell cultures, ATAD3A oligomers are enriched in the MAM fraction, leading to impaired MAM integrity and resulting in synaptic loss [203]. Genetic downregulation of ATAD3A in AD mouse models has been shown to restore cognitive function, reduce neuronal inflammation, and attenuate AD pathology [203]. Furthermore, inhibition of ATAD3A oligomerization using the inhibitor DA1 was found to suppress neuronal inflammation, reduce AD pathology, and protect against neurodegeneration. In these models, ATAD3A oligomerization also inhibits CYP46A1-mediated cholesterol metabolism, which promotes the processing of APP, contributing to AD pathology [203]. Thus, targeting ATAD3A represents a potential therapeutic approach to mitigate neurodegeneration and AD progression.

In AD, abnormal mitochondrial dynamics are characterized by increased mitochondrial fission and decreased fusion. Targeting mitochondrial fission proteins, such as DRP1, has shown potential in restoring balance to these dynamics. Mitochondrial division inhibitor-1 (Mdivi-1), for example, inhibits DRP1 to improve mitochondrial function [204,205]. In APP/PS1 mutant mice, treatment with Mdivi-1 restores mitochondrial length, decreased ROS production, and reduced hippocampal and cortex Aβ accumulation [206]. P110 is a seven-amino acid peptide that specifically blocks the interaction between DRP1 and FIS1, inhibits DRP1 enzymatic activity, and prevents DRP1 translocation to the mitochondria [207,208]. In 5xFAD mutant mice, P110 treatment reduces excessive mitochondrial fission, Aβ accumulation, energetic failure, and oxidative stress by inhibiting DRP1 activity [209].

Enhancing mitophagy can alleviate neuronal stress by clearing damaged mitochondria, a crucial factor in neurodegenerative diseases like Alzheimer's. In APP/PS1 mice, treatment with the BH3-mimetic UMI-77, which targets MCL1, has been shown to promote mitophagy [136]. This increase in mitophagy leads to a reduction in damaged mitochondria, decreased proinflammatory cytokines, and lower Aβ accumulation. Loganin, an iridoid glycoside derived from Corni Fructus, promotes mitophagy and improves learning and memory deficits, reduces Aβ accumulation, and restores mitochondrial dynamics in 3xTg AD mouse models [210]. Notably, loganin decreased the levels of LC3-II, p62, PINK1, and Parkin proteins, yet enhanced the localization of the mitophagy receptor OPTN to mitochondria, promoting mitophagy through a PINK1-Parkin independent pathway. Molecular docking analysis revealed that loganin has a strong binding affinity with OPTN [210]. These findings suggest that targeting mitophagy pathways may be a promising therapeutic strategy for alleviating mitochondrial dysfunction and neurodegeneration in AD.

Perspectives

The complex and multifactorial nature of AD presents significant challenges for both researchers and clinicians. The emerging mitochondrial hypothesis offers a promising new perspective on understanding AD beyond traditional frameworks. Mitochondria, far from being mere energy generators, are dynamic organelles crucial for maintaining cellular homeostasis, especially in neurons with high energy demands. Disruptions in MQC processes highlight a convergent pathway where mitochondrial dysfunction drives AD progression, positioning mitochondria at the heart of disease pathology. Given their critical role in AD, targeting mitochondrial dysfunction represents a novel and potentially transformative therapeutic strategy. Recent advances in gene therapy and pharmacological interventions aimed at restoring mitochondrial function have shown considerable promise in preclinical models. Therapies that enhance mitochondrial biogenesis, stabilize mitochondrial dynamics, or promote mitophagy could potentially mitigate the cascade of neurodegenerative events triggered by mitochondrial impairment.

Looking forward, several critical areas of research should be prioritized to advance the field. A key challenge is to elucidate the precise mechanisms by which mitochondrial dysfunction interacts with other pathological features of AD, such as Aβ deposition, tau hyperphosphorylation, and chronic neuroinflammation. Understanding these interactions will clarify whether mitochondrial dysfunction acts as a primary driver or a secondary amplifier of AD pathology, guiding the development of more targeted therapies. Additionally, while preclinical studies have demonstrated the feasibility of mitochondrial-targeted approaches, their translation to clinical settings requires careful evaluation of long-term safety and efficacy. Future studies should address the potential risks of mitochondrial interventions, such as unintended effects on mitophagy or mitochondrial dynamics, and optimize therapeutic windows and dosing regimens.

Another important direction is the development of biomarkers to monitor mitochondrial dysfunction in vivo. Such biomarkers would enable patient stratification, allowing for personalized treatment approaches based on specific mitochondrial impairments. Advances in biomarker development could also support the real-time assessment of treatment efficacy, facilitating the rapid translation of preclinical findings to clinical trials. Furthermore, the role of mitochondrial dysfunction in non-neuronal cells, particularly microglia and astrocytes, warrants greater attention. These glial cells contribute significantly to the inflammatory and metabolic landscape of the brain in AD, and their mitochondrial health may directly influence disease progression. Investigating mitochondrial dysfunction across cell types could open new therapeutic avenues targeting the brain's immune-metabolic axis. Finally, integrating mitochondrial-targeted therapies with established approaches that address Aβ and tau pathology may provide a more comprehensive strategy for managing AD. Combination therapies have the potential to address the multifactorial nature of the disease more effectively, reducing the burden of neurodegeneration and improving clinical outcomes.

In summary, understanding mitochondrial dysfunction in AD not only deepens our knowledge of disease mechanisms but also expands the scope of therapeutic possibilities. By addressing the outlined research priorities, we can advance innovative, targeted treatments that have the potential to slow or halt the progression of this devastating disease. The integration of mitochondrial-focused strategies with existing therapeutic approaches holds promise for developing more effective and personalized treatments, offering hope to millions of patients and their families.

Author Contributions

K.M.P., and Z.L. wrote the manuscript and X.Q. revised the manuscript. All authors reviewed and approved the manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.