Targeting Mitophagy in Alzheimer’s Disease

INTRODUCTION

Mitochondria are respiratory organelles in cells that play the essential role of producing adenosine triphosphate (ATP) through oxidative phosphorylation (OXPHOS). Additionally, they perform a variety of other cellular functions such as maintaining calcium homeostasis [1–3], signal transduction [4–6], and mediation of cell survival and death [7–9]. During OXPHOS, mitochondrial enzymes inevitably produce reactive oxygen species (ROS) in the form of superoxide (O₂^•–) and hydrogen peroxide (H₂O₂) [10–12]. Although the electron transport chain is the major site of mitochondrial ROS production, there are a number of other ROS sources in mitochondria, such as some of the Krebs cycle enzymes (aconitase, α-ketoglutarate dehydrogenase complex, pyruvate dehydrogenase complex) and monoamine oxidase in outer mitochondrial membrane (OMM) [13, 14]. Mitochondria also have antioxidant defense systems to neutralize ROS. Superoxide dismutase isozymes (SOD), MnSOD (matrix-bound), and Cu/ZnSOD (intermembrane space) catalyze the dismutation of two superoxide molecules to H₂O₂ [11]. The main H₂O₂ degrading pathways in mitochondria are glutathione and thioredoxin systems. These require the reductive power of NADPH to eliminate ROS [15].

The physiological outcome of ROS is dependent on its abundance and the overall cell redox state. ROS generally build up when the antioxidant systems are overwhelmed. Low levels of ROS function as signaling molecules in some processes, while high quantities readily react with biomolecules, which is potentially harmful for all cell processes [16]. Mitochondria that are no longer able to sustain the energy requirements of their host cells can contribute further to ROS production, resulting an increasing cycle of ROS generation [17]. Therefore, it is important that a healthy population of mitochondria is maintained through mitophagy, the efficient removal of dysfunctional mitochondria as well as biogenesis of nascent mitochondria. Cellular bioenergetic homeostasis is dependent on these mitochondrial quality control mechanisms [18, 19], ensuring the effective management of mitochondrial turnover. The human brain is a major consumer of oxygen, requiring up to 20% of the total used by the body [20], making mitochondrial quality control vital to this organ. Deficits in mitophagy are seen in neurodegenerative diseases such as Parkinson’s disease [21, 22] and Alzheimer’s disease (AD) [23–25]. The aim of this review is to focus the impairment of mitophagy in AD and highlight the potential benefits of inducing mitophagy as a therapeutic target in AD.

HOMEOSTASIS OF MITOCHONDRIA

Maintaining mitochondrial homeostasis in terms of mitochondrial dynamics, biosynthesis, and clearance is critical for neurons. Mitochondria continuously move along cytoskeletal microtubule tracks between soma and synapses [26, 27]. While in transit, mitochondria exchange materials through two main processes. Fusion is a process where organelles join with each other to share resources [28]. Fission is the process when a single mitochondrion splits into daughter organelles [29–31]. This is a dynamic process to maintain homeostasis and mitochondrial quality.

A shift in the balance of mitochondrial dynamics may alter the rates of fusion and fission. The fusion process fortifies functional mitochondria for preservation by allowing dilution of defects within the mitochondrial pool. The fission process allows segregation of material which can then be isolated into separate daughter organelles. This uneven division of material between the two daughter organelles subsequently allows the selective elimination of defective portions [32, 33].

The mitochondrial fusion process has two sequential steps involving the fusion of the OMM, followed by the same of the inner mitochondrial membrane (IMM). In mammals, the fusion process is regulated by the Dynamin-related GTPases, Mitofusins (MFN1 and MFN2), and Optic atrophy 1 (OPA1). Within the mitochondria, OMM fusion is mediated by MFN1 and MFN2 while the IMM fusion is mediated by OPA1 [34, 35]. The fission process is driven by FIS1 and Dynamin-related protein 1 (DRP1) [31] which are recruited to mitochondria from the cytosol.

A study by Lutz et al. (2009) in SH-SY5Y cells showed an increase in DRP1, representing an overall increase in mitochondrial fragmentation, is induced by loss of the mitochondrial proteins PTEN-induced putative kinase protein 1 (PINK1: Park6) and Parkin (Park2), a mitochondrial E3 Ubiquitinin (Ub) ligase [36]. PINK1 and Parkin are considered key players in mitophagy and are further described in this review. With the presence of Parkin, DRP1 is ubiquitylated for disposal via the Ub-proteasome system (UPS), which degrades most short-lived intracellular proteins [37, 38]. It is evident that mitochondrial dynamics, especially the fission process may have links to the process of mitophagy [36].

MITOPHAGY: SELECTIVE AUTOPHAGY OF DAMAGED MITOCHONDRIA

Autophagy is the major self-degradation process in cells that is triggered by nutrient deprivation or cellular stress. This process recycles superfluous cells, cell organelles, and cytosolic proteins through lysosome-mediated degradation. This is a rapid process mediated by several specific autophagy-related proteins [39–41]. Autophagy can be categorized into three major types based on the specific components being degraded.

First, micro-autophagy degrades cytosolic components via direct engulfment by lysosomes. In the second, macro-autophagy, portions of cytoplasm and cellular organelles are sequestered by phagophores, forming a vesicle termed the autophagosome. The double membrane of autophagosome fuses with that of lysosomes forming an autophagolysosome, where the contents are degraded by lysosomal proteases. This is also known as non-selective autophagy. Lastly, selective autophagy targets specific tissues, cells, pathogens, protein aggregates, and damaged organelles to form the autophagosome. The ATG8 family of proteins on the phagophore act as cargo receptors to distinguish poly-Ub chains on the target cargo. Chaperone-mediated autophagy is a subtype of selective autophagy in which heat shock proteins, such as Hsp70, transport target proteins to lysosomes. Damaged mitochondria produce increased ROS levels resulting in mitochondrial permeability transition pore (MPTP) opening and Cytochrome c (CYT-C) release, potentially leading to apoptotic cell death.

At this point, mitochondrial autophagy known as mitophagy is initiated as a measure for cell survival [42]. Mitophagy ensures mitochondrial integrity and efficient function by preventing the accumulation of dysfunctional mitochondria, hence, it is recognized as an essential element of neuroprotection [43, 44]. While originally described by Lemasters in 2005 [45], there are now several types of mitophagy currently recognized. NIP3-like protein X (NIX) mediates programmed-mitophagy at specific developmental stages in certain cell types. For instance, during retinal ganglion cell differentiation [46] and maturation of cardiomyocytes [47] and erythroid cells [48], mitophagy occurs as an essential process of development. During programmed-mitophagy, cells switch to an alternative ATP source such as glycolysis. In mammalian cells, there are three major classes of mitophagy: type 1 is basal mitophagy, type 2 is PINK1/Parkin-mediated, and type 3 is mitochondrial vesicle-based mitophagy.

MOLECULAR SWITCHES OF PINK1/PARKIN-MEDIATED MITOPHAGY

The sophisticated and interwoven mechanisms of mitophagy are tightly regulated at different controlling points. In healthy mitochondria, PINK1 is targeted to mitochondria due to its N-terminal mitochondrial targeting sequence. Its importation into the mitochondrial matrix occurs via the TOMM and translocase of the inner mitochondrial membrane (TIMM) complexes. This is closely coupled with its cleavage by the mitochondrial processing peptidases (MPPs) [53]. This is a sensitive system, where reduced MPP activity can cause sufficient PINK1 accumulation for Parkin recruitment and subsequent mitophagy.

The remaining transmembrane fragment of PINK1 (a 52 kDa fragment) is subsequently cleaved by the PARL protease and released into the cytoplasm. These PINK1 fragments act as mitophagy inhibitors in the cytosol, which bind directly to Parkin and block its translocation to the OMM [72]. One structural control over Parkin is its occurrence in cytoplasm in an autoinhibitory form. This autoinhibition is modulated by a repressor element in its structural sites, preventing the binding of Parkin to the RING1 domain. Binding of Parkin to the RING1 domain is essential as the RING1 domain subsequently binds to the E2-binding site, which needs to be bound to E2-Ub conjugating enzymes during the Ub transfers. Additionally, a structural interaction between RING0 and RING2 provides a hindrance for the Parkin catalytic site of Cys⁴³¹ [73]. Moreover, the expression of Parkin is regulated by a transcriptional repressor, c-Jun [74].

Post-translational modifications, such as phosphorylation and ubiquitylation regulate a feed forward amplification loop in activating mitophagy through Parkin. When mitochondrial depolarization occurs, importation of full length PINK1 into the mitochondrial matrix via the TOMM and TIMM complexes is prevented, leading to its accumulation on the OMM. Moreover, triggered by an unknown mechanism, degradation of cleaved-PINK1 fragments in the cytosol occurs by the UPS, which releases the Parkin to be recruited to OMM. Following PINK1-mediated phosphorylation of Parkin on the damaged mitochondria, Parkin ubiquitylates itself and many other mitochondrial proteins [75]. These Parkin-ubiquitylated Ub conjugates are in turn phosphorylated by PINK1, forming further cycles of Parkin activation, leading to increased mitophagy [76]. Moreover, ubiquitylation by Parkin involves conjugation of the C-terminal glycine residue of Ub on to the ɛ-amino group of lysine (K) residues in the substrates.

Structurally, Ub has seven K residues. Therefore, based on which linkage is used to conjugate, there are multiple distinct poly-Ub chain patterns for substrates which are represented by different functional modes. For instance, it has been shown that substrates ubiquitylation with K-48 and K-63 linked chains, are directed via UPS and autophagy adapters, respectively. However, there is debate over which linkages are important for PINK1/Parkin-mediated mitophagy. It has been later reported that non-canonical linkages, K6 and K11, are important for mitophagy as the overexpression of K6R and K11R mutant Ub showed deficits in mitophagy [77]. The ubiquitylating activity of Parkin is regulated by deubiquitylation.

Deubiquitylating enzymes (DUBs) hydrolyze the peptide bonds that link Ub chains. Ub-specific proteases (USPs) are the largest class of DUBs. USP30 is one such DUB, which is integral to the OMM, that has recently been shown to regulate mitophagy upstream of PINK1. It does this through modulation of PINK1 substrate availability, which determines the potential for mitophagy initiation [78]. USP14 promotes mitophagy in the absence of the well-characterized mediators of mitophagy, PINK1 and Parkin. Critical to USP14-induced mitophagy is the exposure of the LC3 receptor PHB2 by mitochondrial fragmentation and mitochondrial membrane rupture [79]. Ma et al. (2016) have reported that ROS induce Peroxiredoxin 6 (PRDX6), a protein in the family of ubiquitously expressed antioxidant enzymes PRDX, to be recruited to damaged mitochondria and initiate PINK1/Parkin-mediated mitophagy [80].

Mitochondrial uncoupling protein 2 (UCP2) is an IMM protein involved in the control of uncoupling of proton flux within the mitochondria. The protons generated during OXPHOS, re-enter the matrix via UCPs [81]. It has been shown in retinal ganglion cells and retinal tissues that deletion of the Ucp2 gene increases the levels of mitophagy [82]. Furthermore, it has been reported that mouse lung endothelial cells lacking UCP2 exhibit increased mitophagy through the PINK1/Parkin pathway [83]. However, the exact mechanisms of UCP2 induction of mitophagy still remain to be identified.

Interestingly, there is evidence that PINK1 and Parkin modulate the expression of tumor protein p53 (TP53), a tumor suppressor where its mutant forms are partly responsible for tumor initiation and for early stages of tumor development [84]. There is feedback regulation of TP53 that affects the expression of PINK1 and Parkin proteins. As recently uncovered by Checler et al. (2018), TP53 acts as an anti-mitophagic effector by reducing the protein levels and mRNA levels of PINK1 by repressing the activity of PINK1 promoter [85]. Also, this promoter has been found to encompass a TP53-responsive element that may reduce the TP53-dependent PINK1 promoter activity. Moreover, it has been shown using both in vitro and in vivo mouse models, that a TP53 gene deficiency increases expression of the autophagy receptors OPTN, NDP52, and LC3-II and reduces TIMM23, TOMM20, and HSP60 [85, 86]. The expression of autophagy receptors is also induced by PINK1 itself, by activation of TP53 through phosphorylation. In this feedback loop, PINK1-phosphorylated TP53 translocates into the nucleus, leading to PINK1 transcriptional reduction. This reduces TP53 phosphorylation and subsequent translocation to the nucleus. Similarly, Parkin-mediated transcriptional reduction of TP53 can also be considered a cellular regulator that prevents the effects of TP53-dependent PINK1 downregulation [85].

The mitochondrial 18 kDa translocator protein (TSPO), a component of MPTP, is another molecule that regulates the mitophagy process. Four to six molecules of TSPO, together with one molecule of Voltage-dependent anion channel 1 (VDAC1), Bcl-2, hexokinase, Cyclophylin D (CYP-D), and adenine nucleotide translocator (ANT) form the MPTP [87]. It is suggested that mitochondrial production of ROS is mediated via TSPO ligands by activating VDAC1 to induce activation of the mitochondrial apoptosis pathway [88]. It has been shown that TSPO-induced ROS production prevents Parkin-mediated ubiquitylation of mitochondria [89]. On the other hand, it has been shown that Parkin activity is retarded by S-nitrosylation events [90]. Moreover, it seems that mitophagy and autophagy are regulated by ROS in a concentration-dependent manner, where moderate levels of ROS activate mitophagy while higher levels of ROS activate autophagy in general [91].

Despite all these regulatory measures, diminished removal of damaged mitochondria as a result of reduced function of these mitophagy pathways is increasingly being recognized as a key feature of the aging process [92]. It is clear that dysfunction of these mitophagy quality control pathways leads to either deficient or excessive mitophagy. Both of these are common factors in the pathogenesis of many diseases.

EVALUATING MITOPHAGIC FLUX USING IN VITRO AND IN VIVO DISEASE MODELS

As modulating mitophagy is an emerging therapeutic target in neurogenerative diseases, including AD, it is necessary that reliable techniques are developed to detect mitophagy to enable better understanding of its molecular mechanisms. It is challenging to distinguish mitophagy from autophagy as all cargoes eventually undergo degradation. Moreover, to study mitophagy in cell lines, transfection of Parkin is essential unless the cell lines endogenously express Parkin [93].

Analyzing mitochondrial mass is the simplest method of monitoring mitophagy. Flow cytometry or immunocytochemistry can be used in combination with mitotracker probes or mitochondrial marker protein antibodies for this purpose. Mitotracker probes are negatively charged and once they enter mitochondria, they react with the reduced thiols present in the matrix. However, in live cell imaging, care should be taken in selecting suitable MMP-insensitive mitotrackers, such as MitoTracker Green [94] to achieve dye retention in mitochondria. Alternatively, the mitochondrial to nuclear DNA ratio can be used to evaluate mitophagy. Transmission electron microscopy is also a powerful method to visualize autophagosomes and autophagolysosomes to detect the disappearance of mitochondria. However, disadvantages are the need for proper expertise and quality of sample preparations [95].

Use of mitophagy-modulating chemical compounds is pivotal in studying this process in vitro. Being a dynamic multi-step process, the amount of mitophagy assessed at a given time point (snapshot measurements) is not an appropriate quantification. Instead, autophagic or mitophagic flux is a better measure. Autophagic or mitophagic flux is the rate of autophagic or mitophagic degradation activity, which is often quantified by the difference of a degradable autophagosomal marker, such as p62 and LC3-I/II, in the presence or absence of lysosomal inhibitors [96]. Bafilomycin A and chloroquine are the most widely used of these inhibitors for lysosomal blockade. Among other mitophagy modulators used as controls in in vitro studies, Carbonyl cyanide m-chlorophenylhydrazone (CCCP) is a potent mitochondrial uncoupler that induces mitophagy by increasing the permeability of protons across mitochondrial membranes and depolarizing mitochondria. It can also induce mitophagy by generating ROS [57]. Many studies have used CCCP to induce mitophagy [97–99]; however, it is reported that CCCP interferes with lysosomal function and autophagosomal degradation in both yeast and mammalian cells. It also inhibits starvation-induced mitophagy in mammalian cells and blocks the induction of mitophagy in yeast cells [100]. Therefore, inducers of mitophagy that affect MMP should be carefully controlled in a time-dependent manner when used together with organelle-specific stains and trackers. Antimycin A and oligomycin mixture (AO) has also been used in many studies to induce mitophagy [98, 102]. Antimycin A is a complex III inhibitor [103, 104] and oligomycin inhibits ATPase synthase in the OXPHOS pathway [105, 106]. Cyclosporin A inhibits mitophagy by blocking the MPTPs [98]. However, it should be remembered that these mitophagy modulators may cause side effects by triggering other cellular pathways [107].

In using western blotting, one common approach to evaluate autophagy or mitophagy is detecting the ratio of LC3-II/LC3-I, but this may appear unreliable as it reflects the induction of autophagic sequestration and is not a direct measure of autophagic flux per se [108, 109]. Therefore, mitophagy should be assessed by simultaneous analysis of several autophagosomal markers such as LC3I/II and p62 [110], along with various mitochondrial inner/outer/matrix proteins as an estimate for mitophagy [111]. It should be noted that OMM proteins such as TOMM20 are destined to be degraded by the UPS, independent to mitophagy [112]. Indeed, TOMM20 does not appear to be a consistently reliable marker [60, 113]. Similarly, VDAC1 is also not a good mitochondrial marker to assess mitophagy as it is involved in many biochemical pathways. Overall, mitochondrial markers such as complex Va (CVa), complex III Core 1 (Core 1), TIMM23, CYP-D, and HSP60 are suggested to be more suitable markers to monitor mitophagy [60, 113].

Fluorescence-based co-localization of autophagosomal markers with mitochondrial markers is another strategy to study mitophagy, both in cells and in vivo models [114]. In this approach, the mitophagy index is the ratio of the mitochondrial probe co-localized with an autophagy marker, expressed as a percentage of the total mitochondrial probe [115]. Lysosomal markers, such as lysosomal activation membrane protein 2 (LAMP2), can be combined with mitochondrial trackers or immunostained with TOMM20, TIMM23, VDAC1, CYT-C, and Cytochrome c Oxidase subunit II (COXII) antibodies [111, 115]. LC3 is also commonly analyzed through green fluorescent protein (GFP) tagging. Similarly, the GFP-LC3 plasmid can be co-transfected with mitochondrially-targeted florescent probes such as mitochondrial-localized red fluorescent protein (Mito-DsRed). Mito Timer is a mutant form of DsRed which allows irreversible fluorescence shifts from green to red when the protein is oxidized. This mitochondrial-matrix targeted probe is versatile in studying mitophagy [116]. There are dual-fluorescent reporters that combine the detection of mitochondria in the acidic environment within the lysosomes. Mitochondria-targeted Rosella (mtRosella) is one such molecular biosensor that combines DsRed fused to a pH sensitive GFP variant. Mitochondria-targeted Keima (mt-Keima) is a ratiometric pH-sensitive fluorescent protein targeted to the mitochondrial matrix. This enables differential imaging of mitochondria in cytoplasm and lysosomes. Furthermore, the mito-QC assay is based on the expression of a functionally inert, tandem mCherry-GFP tag fused to the mitochondrial targeting sequence of the OMM protein, FIS1 (comprising of amino acids 101–152), in which the signals can be used to monitor lysosomal delivery. Inside the lysosomes, mCherry fluorescence remains stable, but GFP fluorescence becomes quenched by the acidic microenvironment resulting in appearance of punctate mCherry-only foci that can be easily quantified as an index of cellular mitophagy [102].

Studying mitophagy using in vivo models is facilitated by the same principles discussed previously. There are transgenic Caenorhabditis elegans expressing mtRosella in body-wall muscle cells that facilitate the evaluation of mitophagy, using the GFP/DsRed ratio of mtRosella fluorescence [115]. Transgenic mouse models expressing the mt-Keima fluorescent protein can be used to assess mitophagy in a wide range of physiological conditions. However, using the mt-Keima in tissues has several limitations, such as spectral overlap and incompatibility with tissue fixation procedures. Transgenic mito-QC is a novel reporter mouse model that overcomes the limitations of mt-Keima and permits the high-resolution study of mammalian mitophagy. Mito-QC mice are produced by constitutive knock-in of mCherry GFP-FIS1^101–152 (mito-QC) in the Rosa26 locus of C57BL/6 to express mito-QC in all tissues at different levels, while permitting visualization of the entire mitochondrial network [107]. All these methods provide valuable tools to better understand the mitophagy process despite their own specific limitations. For instance, it has recently been revealed using mito-QC mice, that basal mammalian mitophagy occurs independent of PINK1 [58]. However, it is generally advisable to use a combination of techniques such as fluorescence imaging and western blotting to assess changes in mitophagy. Overall, it is reasonable to suggest that more robust and reliable methods of mitophagy evaluation would broaden the current understanding of molecular mitophagy, especially in the context of pathologies such as AD.

ALZHEIMER’S DISEASE

AD is a devastating neurodegenerative disorder first described by Dr. Alois Alzheimer in 1907 [117, 118]. There are two types of AD arbitrarily based on the age at onset. Early-onset AD (EOAD) or familial AD (FAD) [119] exhibits autosomal dominant inheritance and is linked to mutations in the genes encoding the presenilins (PS-1 and PS-2) and amyloid precursor protein (APP) [120]. EOAD makes up less than 5% of all AD cases while late-onset AD (LOAD) or sporadic AD [121] is the most common form of the disease that constitutes approximately 50-70% of clinically diagnosed dementia cases, worldwide. The progressive nature of the disease is characterized by cognitive deficits exacerbated over time [122], leading to death within a decade or more from their first manifestation. The key risk factors for AD are age, Apolipoprotein E allele status (APOE ɛ4), sex, head injury, and lifestyle [123]. The AD brain is classically characterized by extracellular deposition of amyloid-β (Aβ) protein aggregates, as senile plaques, and intracellular neurofibrillary tangles (NFTs), composed of hyperphosphorylated forms of the microtubule (MT)-associated protein tau [124].

According to the amyloid cascade hypothesis, accumulation of Aβ aggregates is the key initiator of the cascade of events that lead to neurodegeneration in AD, which includes tau hyperphosphorylation, subsequent NFT formation and oxidative and inflammatory processes [124]. Although there have been reiterations and recent challenges to the validity of the hypothesis (see recent reviews, [125–128]), genetic, biochemical, and more recent evidence from neuroimaging studies [129–132] posit Aβ accumulation as an early, key feature of the disease progression. The Aβ peptides are produced from a two-step cleavage process of the membrane spanning amyloid-β protein precursor (AβPP), initially by β-site APP cleaving enzyme 1 (BACE1, formerly β-secretase) and γ-secretase enzymes, respectively. This process generates multiple Aβ peptides that vary in amino acid length, where the 40 or 42 amino acid peptides (Aβ_1–40 and Aβ_1–42, respectively) are predominately produced. The longer Aβ₄₂ peptides are thought to be more neurotoxic than shorter forms, due to increased propensity to aggregate and generate Aβ oligomers. The accumulation of these aggregates is thought to occur as a result of dysregulated enzymatic processing of AβPP or impaired Aβ clearance from the brain [133]. Many factors individually or cumulatively contribute to a gradual accumulation of Aβ plaques in the brain over at least two decades, until a critical point is reached with manifestation of clinical symptoms [129]. It is reported that pre-fibrillar Aβ initiates tau hyperphosphorylation resulting in disassembly of MTs and the subsequent formation of NFTs in the AD brain [134]. Hyperphosphorylated tau is correlated to cognitive impairment in AD [135, 136], which is driven through synaptic dysfunction and neuronal loss [137, 138]. Those who remain cognitively unaffected but have abundant Aβ plaques and NFTs are termed as Non-Demented with Alzheimer’s Neuropathology (NDAN) [139]. However, the severity of the cognitive impairment in most AD cases is reported to correlate well with the burden of neocortical NFT [140].

Tau proteins function primarily to stabilize the microtubules of axons and are directly involved in their assembly [141–143] through interactions with tubulin. Befitting their function, they are abundant in the axons of neurons [144]. Tau is a phospho-protein [145], where its dephosphorylated state is more efficient at promoting microtubule assembly [146]. There was only modest interest in tau, until the finding that it formed the paired helical filaments, seen as NFTs in the AD brain, where it was found to be abnormally phosphorylated [147, 148].

There are six isoforms of tau produced in the brain, all from a single gene produced alternative mRNA splicing [149]. The six isoforms may be grouped by the number of microtubule binding repeats. Three of the isoforms have three microtubule binding repeats (3RTau), while the others have four binding repeats (4RTau). In the AD affected brain, all six isoforms are present, but there is evidence for a shift toward the expression of more 4RTau, as determined by cDNA analysis of human AD brains [150]. However, a more recent animal study suggests that both types of tau may have their own modes of pathogenesis in the brain [151].

Hyperphosphorylation of tau is common to many neurodegenerative diseases that present with the accumulation of its filaments, often referred to as “tauopathies” [152], suggesting that this abnormal phosphorylation is required for cell toxicity. Some of the genetic forms of these show pronounced tau pathology, but often with no amyloid pathology [153–156]. It is well known that the phosphorylation state of tau determines its ability to bind to the microtubule [157, 158]. However, the mechanism of phospho-tau accumulation is not so well understood. There is evidence that certain tau species in the AD brain localize in the soma of neurons [159], suggesting specific roles for phospho-tau in neuronal death. In fact, it has been reported that tau toxicity is enhanced by the presence of Aβ suggesting that synergistic pathological mechanisms are at play [160].

MITOCHONDRIAL DYSFUNCTION IN AD

It is becoming increasingly recognized that mitochondrial dysfunction [161, 162] and defects in the selective elimination of impaired mitochondria [163] play an integral role in AD pathogenesis [24, 165]. Glucose hypometabolism in affected brain areas indicates the impairment of mitochondrial function and turnover that may contribute to neurodegeneration in AD [166, 167]. Although the mechanisms are not yet completely understood, growing evidence from cellular and animal AD models suggests that Aβ and tau proteins interact with mitochondrial proteins to trigger mitochondrial dysfunction through several pathways, such as impairment of OXPHOS, elevation of ROS production and alteration of mitochondrial dynamics. This suggests that an effective treatment for AD may need to include targets that address mitochondrial function and turnover [168, 169].

According to the mitochondrial cascade hypothesis proposed by Swerdlow, Burns, and Khan [23, 171], the bioenergetic deficit resulting from mitochondrial dysfunction is a primary event in the neurodegenerative process of AD. There are other hypotheses which address the origin of AD-related mitochondrial dysfunction. The Alu hypothesis, proposed recently by Larsen et al. (2017), posits that the mitochondrial dysfunction is due to deleterious activity of primate-specific retrotransposons, the Alu elements (short interspersed elements, also known as SINEs) in neurons [172, 173]. For instance, located on chromosome 19, at a closely adjacent locus to APOE, the TOMM40 homolog that encodes a β-barrel protein critical for mitochondrial preprotein-transport, is vulnerable to structural variations by Alu element insertions. Repeated insertion of Alu elements into TOMM40 introns 6 and 9 contributes to transcriptional noise, through enhanced nonsense-mediated decay and/or the production of alternative TOMM40 isoforms. This results in aberrations in mitochondrial protein import leading to mitochondrial dysfunction [172].

Accumulation of AβPP, Aβ, and hyperphosphorylated tau in mitochondrial membranes and compartments is another factor that contributes to mitochondrial dysfunction in AD. Notably, in human cortical neuronal cells, it has been reported that AβPP, with its chimeric N-terminal signal, is targeted to mitochondria in addition to plasma membranes [174]. In this regard, mitochondria-associated endoplasmic reticulum (ER) membranes (MAM) have gained attention as a potential key role player in the pathogenesis of AD. MAM are enriched with lipid rafts which can carry AβPP, PS-1, PS-2, Aβ, and γ-secretase [175], making MAM the possible sites of Aβ production [176]. The MAM hypothesis of AD suggests that MAM functions are increased significantly in AD [177]. Thus, AβPP directed to neuronal mitochondria accumulates on mitochondrial protein import channels and in mitochondrial compartments in a membrane-bound form [174, 178].

Soluble oligomers of Aβ also interact with many mitochondrial membrane proteins, including ANT, components of TOMM, TIMM, CYP-D, and UCPs [179]. Many studies have indicated that membrane-spanning channels could be created by Aβ interaction with lipid membranes (Aβ channels) in mitochondria [180, 181]. This uncontrolled passage of different ions, including Ca²⁺, disrupts mitochondrial homeostasis leading to neuronal death [161]. Furthermore, it is documented that Aβ itself induces activation of glutamate N-methyl-D-aspartate receptors (NMDARs) [182] and/or downregulation of Ca²⁺ binding proteins, Ca²⁺ dependent kinases and Ca²⁺ transporters, causing excessive release of calcium from the ER [183, 184]. Mitochondrial targeting presequences are cleaved off by MPPs and degraded by mitochondrial presequence peptidases (PreP) which also degrade Aβ [185]. Both Aβ and AβPP within mitochondria may lead to inhibition of PreP, resulting in aberrant mitochondrial protein profiles and accumulation of Aβ [186].

Another enzyme localized in mitochondria that can facilitate Aβ-induced stress is 17β-Hydroxysteroid dehydrogenase type 10 (Alternate name: Aβ-binding alcohol dehydrogenase (ABAD) is a misnomer predicated on the mistaken belief that this enzyme is an alcohol dehydrogenase) [187]. It provides a direct link between Aβ and cytotoxicity via mitochondrial oxidative stress. Neurons cultured from a transgenic mouse model overexpressing a mutant form of AβPP and ABAD (mAβPP/ABAD) have shown reduced COX activity and hence, aberrations in mitochondrial metabolism [188]. This could be due to obstructed entry of nuclear-encoded COX subunits IV and Vb into mitochondria, leading to diminished capacity to produce ATP. Moreover, the effects of COX subunit III messenger RNA (mRNA), a marker of mitochondrial energy metabolism caused by neuronal accumulation of NFTs in the AD brain, have been investigated [189]. In situ hybridization techniques have revealed that the levels of COX subunit III mRNA are decreased in AD-affected mid temporal cortex compared to NFT-free neurons of AD brains [189]. As the metabolic needs of neurons are mainly determined by synaptic transmission, it has been suggested that damaged mitochondria have a role in impaired synaptic function and decline of neuronal activity during the early stages of AD pathogenesis [189].

Aberrations in mitochondrial biogenesis and dynamics, such as transport, fission, and fusion, are also features of mitochondrial dysfunction seen in AD [190]. Recent studies in AD mouse models, have shown reduced levels of peroxisome proliferator-activated receptor γ coactivator 1 (PGC-1α), nuclear respiratory factors 1 and 2 (NRF1, NRF2), and mitochondrial transcription factor A (TFAM) suggesting that mitochondrial biogenesis is impaired in AD [191, 192]. Rui et al. (2006) showed that exposing cultured hippocampal neurons to Aβ severely impaired mitochondrial transport without inducing significant cell death or morphological changes [193]. Deficits in mitochondrial motility may arise due to disturbed MT-dependent transport to the soma. Studies have also revealed that presynaptic dysfunction is due to lack of mitochondria in synapses, a result of their impaired bidirectional transport [194, 195]. Another study showed that exposure of Aβ-derived diffusible ligands to primary rat hippocampal neurons results in depletion of OPA1, MFN1, and MFN2 proteins in cells, suggesting inhibition of the mitochondrial fusion process [196]. Furthermore, the cells also had increased levels of FIS1, the mitochondrial fission protein. Consistent with this, Aβ has been found to-induce production of NO leading to nitrosylation at Cys⁶⁴⁴ on the fission-inducing protein, DRP1 (forming SNO-DRP1). This resulted in a marked increase in DRP1 activation and the subsequent mitochondrial fission, synaptic loss, and neuronal damage [197].

It has been reported that phosphorylated tau interacts with VDAC1, blocking MMTPs and leading to mitochondrial dysfunction [198]. As shown in the P301L Tau mouse model, phospho-tau reduces mitochondrial complex V levels [199]. Moreover, evidence exists showing that hyperphosphorylated tau leads to abnormal mitochondrial transport [200]. It is known that tau, being a microtubule-associated protein, influences the transport of molecules through interactions with dynein and kinesin motor proteins [201]. It is thought to be part of a system that regulates the balance of microtubule traffic in the cell [201]. This microtubule network is also thought to be responsible for the movement of mitochondria [202–206]. Movement and positioning of mitochondria is thought to be tightly controlled, since these organelles have other important roles beyond their main function of energy production. This includes cell proliferation [207], stress response [208], apoptosis [209], and Ca²⁺ homeostasis [210]. There is evidence in AD mouse models that the transport of mitochondria is affected, where there is increased retrograde transport [2] and decreased anterograde transport [201]. This aberrant transport was associated with dysfunctional mitochondrial fusion and fission. In fact, it has been reported that interactions of hyperphosphorylated tau and DRP1 caused excessive mitochondrial fission in the brains of APP, APP/PS1, and 3xTg-AD mice [211]. Moreover, it has been reported in HEK293 cells and primary hippocampal neurons of rats that human wild-type full length tau increased mitochondrial fusion, as shown by the increased fusion proteins OPA-1, MFN1, and MFN2 [212]. These changes in mitochondrial dynamics lead to the mitophagy deficits and subsequent mitochondrial dysfunction seen in AD.

MITOPHAGY IMPAIRMENT IN AD

It is increasingly being recognized that mitophagy dysfunction can be considered a new hallmark of sporadic AD [24, 213]. However, as in the case of Aβ, whether mitophagy impairment is a cause or an outcome is still unknown. Diminished mitophagy has been shown in AD patient brains, accompanied with a depletion of cytosolic Parkin as the disease progresses [2]. There is evidence that increased PINK1 expression attenuates Aβ accumulation and Aβ-induced mitochondrial and synaptic dysfunction. Furthermore, PINK1 knock out (PINK1^–/–) mutant APP mice have impaired mitochondrial and synaptic function, coupled with raised Aβ pathology and oxidative stress, compared to non-transgenic mice [214].

Hyperphosphorylated tau can also affect this system, as shown by in vitro studies. Here, it directly localizes and accumulates in mitochondria, leading to an increase of MMP and impaired PINK1/Parkin activation resulting in subsequent deficits of mitophagy [215]. It is also recently reported in neuroblastoma cells that tau can attenuate mitophagy by reducing the translocation of Parkin to mitochondria [216]. However, mitophagy can also attenuate AD-related tau hyperphosphorylation in human neuronal cells and reverse memory impairment as shown in transgenic tau nematodes and mice [217].

While being considered as part of the UPS, Parkin has shown protection against intracellular lentivirus-generated Aβ toxicity in a cell model and Aβ-induced neuronal degeneration in rat brain [218]. Moreover, Parkin overexpression has been demonstrated to reverse the mitophagy failure found in human fibroblasts from sporadic AD patients [213]. Also, tau overexpression is necessary to generate a mitophagy failure in AβPP overexpressing cells [219]. It has been reported that N-terminal truncated tau induces aberrant Parkin recruitment, thus leading to excessive mitophagy and contributing to synaptic failure [220].

Thus, accumulation of dysfunctional mitochondria in AD-affected neurons is likely attributable to inadequate mitophagy capacity in the AD brain. Mitophagy is a specific form of autophagy in which the final steps of lysosome-based cargo degradation share the same machinery. As evidenced by excessively large autophagic vacuoles (AVs) of Aβ peptides in the neurons in AD [221], it is suggestive that neuronal mitophagy capacity is used to clear the toxic Aβ peptides resulting in overloaded and functionally saturated lysosomes and associated AVs. Studies reporting prominent autophagic accumulations of mitochondria in AD affected brains [222–225] suggest that the autophagic machinery is still competent in AD affected neurons. However, the flux is impaired in the final stages of the mitophagy process, during the fusion of mitophagosomes with lysosomes [226].

OPEN IN VIEWER

Fig.1

PINK1/Parkin-mediated mitophagy. Dissipated MMP in damaged mitochondria result in OMM stabilization and subsequent activation of PINK1 on the OMM. Targets of phosphorylated PINK1 sites (Ub and MFN2) recruit Parkin to the OMM. Parkin ubiquitylates numerous OMM proteins that attract numerous downstream adaptors and receptors. A) Specific autophagy adaptor-mediated mitophagy. Autophagy adaptors such as p62/SQSTM1 and NBR1 mediate mitochondrial engulfment by binding to LC3-II molecules on the phagophores. B) OMM protein receptor-mediated mitophagy. The OMM-localized proteins (such as BNIP3, NIX, and FUNDC1) can directly bind to LC3-II molecules on the phagophores, through their phospho-regulated LIR domains. C) OMM lipid receptor-mediated mitophagy. OMM-localized lipids can also function as mitophagy receptors. In damaged mitochondria, the OMM-resident Cardiolipins translocated through specific transporters, directly bind to LC3-II molecules on the phagophores.

It has been demonstrated that neurons subjected to oxidative damage show a significant increase in mtDNA and COX. These are localized in the neuronal cytoplasm and associated with lipofuscin, which may have arisen due to low mitophagy activity [227]. These accumulations of damaged mitochondria in the cytoplasm can act as pattern recognition molecules or damage-associated molecular patterns which activate the Nod Like Receptor family pyrin domain containing 3 inflammasome [228]. This may account for the neuroinflammation in AD, due to reduced mitophagy and define the mitophagy process as a mechanism of innate immunity [229].

In the mitophagy process, degradation of defective cargo is carried out by the lysosomal network in cells. Defects in lysosomal acidification often contribute to failure of proteolysis. It is reported that AD-linked genetic mutations disrupt lysosomal acidification and proteolysis, resulting in reduced autophagy and mitophagy. Lysosomal maturation exhibits a PS-1 dependency [230], where PS-1 has been reported to modulate lysosomal Ca²⁺ homeostasis, affecting the vacuolar ATPase-mediated lysosomal acidification [231]. This is supported with evidence of increased lysosomal pH in PS-1-deficient blastocytes, neurons, and fibroblasts from AD patients carrying PS-1 mutations and in neurons from a PS-1/APP AD mouse model [232]. Furthermore, PS-1-mediated disruption of the autophagy process is shown by Martin-Maestro et al. (2017), who have demonstrated that there is deregulation of autophagy proteins in AD patient-derived fibroblasts and neurons from induced pluripotent stem cells harboring the PS-1 mutation A246E [233].

It is also shown that the APOE ɛ4 allele is responsible for lysosomal leakage of Cathepsin D [234]. Moreover, intraneuronal Aβ has been shown to accumulate in lysosomes giving rise to ROS generation, resulting in lysosomal membrane permeabilization and subsequent cell death [235]. Considering the susceptibility of lysosomal damage due to Aβ, processing of damaged lysosomes for efficient clearance by lysophagy, the autophagy of dysfunctional lysosomes [236], may be a prerequisite for efficient mitophagy to occur. As indicated in Table 1, AD-based evidence suggests on dysregulation of mitochondrial dynamics and mitophagy-related proteins, and defective lysosomal clearance are key phenomena in AD brains. Overall, it is obvious that defective mitophagy is at least in part, related to autophagic-lysosomal dysfunction in AD [237–239]. Therefore, strategies to improve lysosomal function are a vital component in inducing mitophagy as a novel therapeutic target for AD.

INTERVENTIONS FOR IMPROVING MITOPHAGY

The presence of mitophagy deficits in AD pathology suggests that inducing mitophagy is a key therapeutic target for AD. Transcription factor EB (TFEB) and FOXO are main transcriptional regulators of autophagy. In addition to their role in general autophagy, they have also been shown as key players in mitophagy [247].

TFEB activation occurs in response to low nutrient or energy status to upregulate expression of autophagic and lysosomal genes, through global transcriptional activation of PGC-1α [248–250]. In addition to its role in general autophagy and mitophagy, the TFEB also regulates lysosomal biogenesis [251]. The levels of TFEB in the nucleus reportedly correspond to varying cellular demands for autophagosome-lysosome function [252]. In the de-activated state, TFEB resides in the cytoplasm, associated with the mammalian target of Rapamycin (mTOR) complex, which is localized in lysosomes at the cytoplasmic side [253]. Alternatively, mTOR1 inhibition promotes the translocation of TFEB to the nucleus and increases the downstream effect of mitophagy, through transcriptional activity of PINK1 and Parkin [254, 255].

Silent information regulator of transcription 1 protein (SIRT3) deacetylates and activates FOXO3 to promote its nuclear translocation for transcriptional activation of PINK [256, 257]. Further, the activity of FOXO3 during the process of mitophagy has been evaluated in SH-SY5Y cells [258]. In this study, the levels of the mitophagy marker proteins Beclin 1, PINK, and Parkin were shown to be increased, following induction of mitophagy by MnCl₂ and this was accompanied by FOXO3 nuclear retention. A more recent study carried out in SH-SY5Y and induced pluripotent stem cell (iPSC)-derived neurons also has shown that Akt signaling increases mitophagy via regulating PINK 1 levels [259]. Additionally, FOXO3 is activated by the Phosphoinositide-3-kinase/Protein kinase B (PI3K/Akt) pathway depending on mTORC2 activation [260].

It has been shown that mitophagy appear to improve in lifestyle interventions such as exercise and low-caloric diet. In fact, an increasing number of studies has indicated the importance of exercise in longevity and healthy ageing [261–263]. One of the ways it is thought to provide benefits is through protection against agents that can cause cell death. It has been reported that exercise induces SIRT1 expression in cortical neurons, which deacetylates SOD₂ and CYP-D in neuronal mitochondria to provide protection against excitotoxicity [264]. Furthermore, during exercise, mitochondrial biogenesis is induced by increased PGC-1α levels [265, 266]. However, a study by Philp et al. (2011) has indicated that upregulated PGC-1α levels are not due to the deacetylase activity of SIRT1, but are due to changes in the general control of amino-acid synthesis 5 (GCN5) acetyltransferase activity after exercise [267]. Alternately, the high energy consumption during exercise disrupts the balance between AMP and ATP, leading to AMPK activation followed by induction of PGC-1α [268].

During exercise, PGC-1α promotes ROS generation via increased mitochondrial metabolism, leading to an elevated oxygen consumption in muscle fibers. This ROS generation is dependent on other exercise-induced changes in muscles including increased CO₂ production, raised temperature and decrease in cellular pH [269]. However, PGC-1α has also been reported to induce ROS-detoxifying enzymes by Nrf2 activation [270]. In addition to the ROS-mediated adaptive responses, known as hormesis, there is emerging evidence that exercise induces autophagy and mitophagy in skeletal muscle [271–273]. Recently, it has been demonstrated in a mouse model that exercise promoted the nuclear translocation of TFEB in the brain cortex, resulting in upregulated transcription of genes associated with autophagy and lysosomes, through the AMPK-SIRT1-TFEB pathway [274]. Furthermore, it has been reported that acute exercise induces mitophagy through the AMPK-dependent activation of ULK1 in the skeletal muscle [275]. A human study indicates that mitophagy occurs only at a later stage and not during, or soon after high intensity endurance exercise [276]. This notion is consistent with the work by Ogborn et al. (2015), who showed that the mitophagy proteins, PINK1 and Parkin, do not change in response to exercise, particularly in the muscle of aged subjects immediately after a single bout of resistance training [277].

Similarly, a low-caloric diet is increasingly recognized as the most effective and promising intervention for slowing or preventing the pathogenesis and progression of AD [278]. A reason for this is suggested to be the induction of autophagy [278]. Caloric restriction (CR), defined as reducing ad libitum calorie intake by 10–50% without leading to malnutrition, has many beneficial effects for healthy aging and extending lifespan [279, 280]. TFEB is induced by CR through an autoregulatory feedback loop and exerts a global transcriptional control on lipid catabolism via PGC-1α [281]. Moreover, as evidenced in Sacchoromyces cerevisiae and C. elegans studies, CR increases cellular OXPHOS levels, giving rise to elevated amounts of ROS. This is thought to improve longevity by activation of the hormetic ROS defense pathways [282–284]. Caloric restriction has multiple effects, including activation of SIRT1, suppression of mTOR [285], expression of SIRT3, AMPK, and PGC-1α in skeletal muscle [286]. However, it has been shown that CR only preserves mitochondrial respiratory function, rather than increasing mitochondrial abundance [287]. In addition, a study conducted in rat cortical neurons showed that CR induces autophagy, stimulated by upregulated levels of neuropeptide Y (NPY) and ghrelin [288], a regulatory peripheral gut hormone that increases appetite. Additionally, Cui et al. (2013) have confirmed that CR markedly increases the expression of autophagosome markers in the kidneys of Fischer 344 rats [289]. CR drives SIRT1/3 and AMPK activation and mTOR inhibition, which may lead to mitophagy.

While exercise and CR clearly have potential benefits, they may not always be a practical option. There are substances being investigated which mimic the effects of CR, known as caloric restriction mimetics [290]. Among these compounds, are polyphenols, which have been identified to influence mitochondrial turnover [291].

NUTRACEUTICAL-BASED STIMULATION OF MITOPHAGY AS A THERAPEUTIC TARGET IN AD

As mitophagy deficits hold a pivotal role in the overall mitochondrial dysfunction seen in AD, it is tempting to consider modulating mitophagy as a therapeutic target for AD [292]. With failures of clinical trials for various synthetic compounds, attention has been turning towards interventions derived from natural sources, such as plant extracts and their active components. Research is increasingly becoming focused on the bioactive components of food, the nutraceuticals, which have additional benefits besides their basic nutritional values [293]. There are a number of mitophagy-inducing compounds reported to date [294], such as urolithin A, spermidine and tomatidine.

Urolithin A is a blood-brain-barrier (BBB)-permeable compound derived from gut microbial activity on ellagitannins [295], a class of polyphenols. It was first described as a mitophagy-inducing and anti-aging compound in C. elegans. It has also been shown to exert anti-AD properties via Aβ fibrillation effects in vitro [295]. Urolithin A also reduced Aβ pathology in an APP/PS-1 mouse model and inhibited several common tau-phosphorylation sites in the C. elegans tau model (BR5270) through the PINK1/Parkin/NIX pathways [217]. Mitophagy induced by urolithin A was also showed to be protective against neuroinflammation [217]. However, there is a contradictory report showing that urolithin A does not induce mitophagy but instead, reinforces ischemia-induced autophagy [297]. Therefore, more studies on the mechanisms regarding the mitophagy-inducing capacity of this compound are needed, even though clinical trials are already being conducted for urolithin A [298].

Spermidine is a polyamine naturally occurring in semen, which mediates multiple processes of the cell cycle and apoptosis [299]. It is also found in a variety of foods including vegetables and dairy products [300]. While physiological levels of spermidine decline with age, it has been found that spermidine extends longevity in certain model organisms [301, 302]. A study conducted in cardiomyocytes revealed that spermidine delays the development of hypertensive heart disease by reducing arterial blood pressure by improving global arginine bioavailability and protects from hypertension-associated renal damage [303]. Spermidine is also documented to be neuroprotective, conferring protection from age-induced memory impairment due to its ability to induce autophagy [304–306]. Furthermore, it has been studied for promoting cognitive and brain health in older individuals with subjective cognitive decline [307, 308]. In fact, spermidine has been shown to promote longevity as a result of its ability to induce autophagy. [309]. Spermidine has also been reported to induce PINK1/Parkin-mediated mitophagy [310]. In a study by Qi et al. (2016), spermidine acted as a trigger for mitophagy by inducing mitochondrial depolarization and by activating a protein named Ataxia telangiectasia Mutated (ATM) in vitro [310]. ATM has been previously reported to regulate mitochondrial function, including mitophagy [311, 312]. It has been speculated that ATM is activated by spermidine-generated ROS, resulting in accumulation of PINK1 and translocation of Parkin to damaged mitochondria to trigger mitophagy [310].

Tomatoes contain many bioactive compounds, including lycopene, α-tomatine, tomatidine, and P3 factor, all of which are reported to have potential health-promoting effects in humans [313]. Despite their activities, including antibiotic, anti-inflammatory, antioxidative, cardiovascular, and immuno-stimulating effects in different experimental models, the current understanding of their roles is still incomplete. A recent study in C. elegans has revealed that tomatidine induces mitophagy through mitochondrial hormesis, by mildly inducing ROS production. This activates the SKN-1/Nrf2 pathway and other cellular antioxidant response pathways, followed by increased mitophagy [314].

However, these mitophagy-inducing compounds may not exert their full potency when consumed as food products. Reasons are that these active compounds usually occur in minute quantities and may result in interactions with other nutrients. More research should be focused on this aspect in the development of nutraceuticals as therapeutic or preventative agents against AD.

CONCLUSIONS AND FUTURE DIRECTIONS

Mitochondrial quality control is maintained through a balance of mitochondrial dynamics, biogenesis, and mitophagy. Mitophagy is a subtle and complex mechanism through which damaged mitochondria are disposed inside cells. Although much progress has been made recently in understanding the molecular mechanisms associated with mitophagy, much remains to be answered. Hence, more robust, physiological models and methods of assessing mitophagy are vital. Moreover, it is important to understand the role of mitophagy in complex pathologies such as AD, especially in a context where mitophagy stimulation confers neuroprotection.

The neurodegenerative processes that lead to AD remain to be fully understood. While the accumulation of Aβ and hyperphosphorylated tau protein are considered to have key roles, investigating the relevant cellular pathways that are impaired in the early stages of the disease process is paramount. Mitochondrial dysfunction is an early feature of the disease, but it is yet to be established as a cause or consequence of accumulating pathology. However, it is increasingly being recognized as a key contributor to the disease process. Restoration of mitochondrial function could offer potential therapeutic targets; however, an alternative is to promote removal of damaged and dysfunctional organelles. In this regard, enhancing the process of mitophagy has shown encouraging results in AD research. Plant-derived compounds, such as urolithin A, spermidine, and tomatidine are gaining much attention in this regard. Further research into these nutraceutical compounds to restore mitophagy and mitochondrial capacity in AD is essential to determine whether they may provide effective therapeutic strategies to combat this global disease.