Mitochondria in neurodegeneration

Abstract

The brain is one of the most energetically demanding tissues in the human body, and mitochondrial pathology is strongly implicated in chronic neurodegenerative diseases. In contrast to acute brain injuries in which bioenergetics and cell death play dominant roles, studies modeling familial neurodegeneration implicate a more complex and nuanced relationship involving the entire mitochondrial life cycle. Recent literature on mitochondrial mechanisms in Parkinson’s disease, Alzheimer’s disease, frontotemporal dementia, Huntington’s disease, and amyotrophic lateral sclerosis is reviewed with an emphasis on mitochondrial quality control, transport and synaptodendritic calcium homeostasis. Potential neuroprotective interventions include targeting the mitochondrial kinase PTEN-induced kinase 1 (PINK1), which plays a role in regulating not only multiple facets of mitochondrial biology, but also neuronal morphogenesis and dendritic arborization.

Historical Considerations

Given that neurons derive most of their ATP from oxidative phosphorylation, it is not surprising that perturbations in mitochondrial content, quality and function are implicated in diseases of the nervous system. Certainly bioenergetic dysfunction plays a key role in ischemic brain injury [1], and the nervous system is a consistent target of primary mitochondrial diseases [2]. Yet whether the mitochondrial alterations observed in chronic age-related neurodegenerative diseases represent cause or consequence of disease processes remained less clear until a set of key studies in the late 20^th and early 21^st centuries.

In 1979, a meperidine analog was found to cause acute onset of an illness resembling Parkinson’s disease (PD) [3], but its significance remained obscure until a 1983 report that contamination of designer opioids with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) caused a cluster of cases [4]. In 1985, the key discovery that the active MPTP metabolite 1-methyl-4-phenylpyridinium (MPP+) inhibits NADH dehydrogenase (complex I) of the mitochondrial electron transport chain (ETC) [5] ushered in a new age in neurodegenerative disease research. Other important advances included identification of a systemic complex I deficiency in sporadic Parkinson’s disease [6], followed by reports of ETC deficiencies in brain tissues of Alzheimer’s disease (AD) (complex I and complex IV) and Huntington’s disease (HD) (complexes II and III) [7]. The linkage of PD to environmental exposure to pesticides such as rotenone, another complex I inhibitor, generated additional rodent models showing selective neurodegeneration in response to systemic exposure [8]. Unfortunately, clinical trials involving antioxidants and other metabolic factors identified using these models have thus far yielded disappointing results. This implicates a more nuanced relationship for the role of mitochondria in neurodegenerative disease pathogenesis.

Mitochondria are integrally involved in many important processes other than oxidative phosphorylation, including calcium homeostasis, lipid biosynthesis and cell death. The seminal discovery in 2004 that mutations in PTEN-induced kinase 1 (PINK1) represent the second most frequent cause of familial autosomal recessive PD [9, 10] stimulated a new wave of interest in the role of altered mitochondrial dynamics in neurodegeneration [11-13]. Indeed, the rapidly expanding availability of genetic models for PD, AD, HD, frontotemporal lobar dementia (FTD) and amyotrophic lateral sclerosis (ALS) ushers in a broader view examining the entire lifecycle of the mitochondrion, from its biogenesis, fission-fusion, and transport to and from synaptic regions to quality control mechanisms that mediate degradation of its components or organelle-level turnover by selective mitochondrial autophagy (mitophagy). In addition, advances in basic mitochondrial biology coupled with the recognition that mitochondria are regulated differently in neurons compared to proliferative and less polarized cell types, continue to support intense interest in the role of mitochondria in neurodegeneration and neuroprotection.

In this review, I will discuss recent advances in mitochondrial quality control and calcium homeostasis with an emphasis on the synaptodendritic degeneration that is frequently observed in the earlier stages of many neurodegenerative diseases. I will touch upon unique features of synaptic mitochondrial regulation, ending with a brief discussion of therapeutic implications.

Mitochondrial Quality Control in Neurodegeneration

In addition to ineffectual ATP generation, damaged mitochondria produce higher levels of reactive oxygen species and can release pro-apoptotic mediators. Repair or removal of damaged mitochondria plays a key role in maintaining cellular health. Macroautophagy represents one of the major mechanisms for sequestering damaged mitochondria for delivery to lysosomes for degradation. The discovery that PINK1 and parkin cooperate to label depolarized mitochondria for autophagy [14, 15] triggered an intense wave of interest in the potential role of mitophagy in neurodegenerative diseases. These seminal studies further amplified interest in the therapeutic upregulation of autophagy, a cellular degradation system that is capable of clearing insoluble protein aggregates as well as damaged organelles [16].

Mitophagy

While the process of PINK1/parkin-dependent mitophagy been reviewed in depth elsewhere [17, 18], this pathway of ubiquitin-mediated mitophagy is triggered when mitochondria are no longer able to effectively import, process and release cleaved PINK1 back to the cytosol. Under these conditions, full-length PINK1 accumulates at the mitochondrial surface and phosphorylates ubiquitin and parkin, resulting in parkin-mediated ubiquitination of mitochondrial surface proteins (Fig.1a-c). Adapter proteins that bind to both ubiquitin and the autophagy protein microtubule associated protein-2 light chain 3 (LC3) recruit ubiquitinated mitochondria to nascent autophagosomes, which mature into acidified autolysosomes for degradation of the contents [16].

Figure 1. Roles of PINK1 in the mitochondrial life cycle of neurons.

PINK1 functions as a sensor for both mitochondrial function (right) and dysfunction (left). (a) Severe mitochondrial depolarization and/or protein import deficits trigger accumulation of full-length PINK1 at the mitochondrial surface where TOM complexes scaffold PINK1 dimerization allowing trans-autophosphorylation (not shown). This activates the ubiquitin kinase activity of PINK1, resulting in activation of parkin via phosphorylation of its ubiquitin-like domain and binding of phosphorylated ubiquitin. Polyubiquitination of mitochondrial outer membrane proteins accompanied by release of mitochondrial-ER contact sites via VCP- UFD1-NPL4 activity (not shown) allows recruitment of damaged mitochondria into nascent autophagosomes through interaction with bifunctional receptors that bind both ubiquitin and LC3. (b) Sequestration and degradation of damaged mitochondria by autophagy prevents cytosolic release of reactive oxygen species and pro-death proteins, and limits extracellular release of mitochondrial damage-associated molecular patterns that trigger inflammation. (c) Protection against cell death comes at a cost to neurons, however, with retraction of synapto-dendritic arbors. (d) In functioning mitochondria, internalized and presumably cleaved PINK1 functions to regulate respiratory complexes, cristae morphology, mitochondrial chaperones and mitochondrial calcium efflux. At the mitochondrial surface, PINK1 and parkin facilitate mitochondrial repair/biogenesis through effects on local translation and import. (e, f) Cleaved PINK1 is also released from mitochondria into the cytosol, where it can be itself ubiquitylated for degradation, or function as a neuronal pro-growth signal through several mechanisms. (e) In the direct pathway, cleaved PINK1 interacts with PKA and VCP-p47 to promote p47 phosphorylation and dendritic arborization. (f) PINK1 also regulates mitochondrial dynamics and mitochondrial transport, sending signals to the nucleus to promote mitochondrial biogenesis by cooperating with parkin to degrade PARIS, a transcriptional repressor of PGC-1α. The ability to replace damaged peri-synaptic mitochondria is crucial for maintaining neuronal structure and function. (g) Small molecule strategies to inhibit PINK1 degradation and/or promote its activity have the potential to protect neurons through any of these mitoprotective pathways.

The stabilization of PINK1 at the surface of the damaged mitochondrion initiates depolarization-induced mitophagy. Yet how is PINK1 activated at the mitochondrial surface? Prior structural studies revealed that phosphorylation of insect PINK1 at a residue homologous to human PINK1 S228 (S205 in Tc-PINK1; S202 in PhPINK1) initiates a conformation change involving kinking of the αC helix and transition of a structurally disordered Insert 3 to form a binding pocket for ubiquitin [19]. While phosphorylation at this site is required for ubiquitin kinase activity, it is not required for PINK1 to phosphorylate other substrates [20]. A recent set of structural studies employing different insect forms of PINK1 illustrate how dimerization results in trans-autophosphorylation of this serine residue [21, 22]. Using crystallization and docking studies of Tribolium castaneum PINK1 (Tc-PINK1), one group showed that part of Insert 2 is necessary for dimerization, and that a motif formed by approximation of helices in the N- and C-terminal domains of PINK1 binds to the translocase of the outer membrane (TOM) complex to position the PINK1 dimer for activation [22]. Crystallography and cryo-EM of PINK1 from Pediculus humanus corporis further suggest that oxidation of C169 decreases PINK1 activity towards ubiquitin, raising the possibility that human PINK1 may exhibit redox-sensitive functions [21].

Interestingly, TOM is not the only large protein complex that binds PINK1. An unbiased immunoprecipitation-mass spectrometry study in HEK293 cells confirmed interaction of PINK1 with CDC37 and TOM complexes, additionally revealing valosin-containing protein (VCP) as one of the top ranked interactors with endogenous protein interactions confirmed in human frontal cortex [23]. VCP is a homo-hexamer that forms a barrel-like structure with 12 ATPase domains [24]. During CCCP/FCCP-induced mitochondrial depolarization, some studies have suggested that VCP translocates to the mitochondria to participate in PINK1-mediated mitophagy in HeLa cells and MEFs [25]. Mechanisms by which VCP may facilitate mitophagy include degradation of Marf in Drosophila [26], removal of MFN2 in HeLa cells [27], and the release of ER-mitochondria contact sites in U2OS cells and iPSC-derived dopamine neurons [28]. Interestingly, cleaved cytosolic PINK1 shows an even greater avidity for VCP, acting to promote growth and branching of dendrites in primary cortical neurons through phosphorylation of the VCP cofactor NSFL1C/p47 [23].

In addition to PD, which can be caused by recessive mutations of either PINK1 or parkin, mitophagy disruption has been reported in primary neurons from the mutant hAPP transgenic mouse model of AD [29] and more recently in immortalized striatal neurons from a knockin mouse model of HD [30]. A closer examination of gene mutations that cause diseases in the ALS-FTD spectrum reveals several proteins that function in mitophagy, such optineurin, sequestosome 1/p62 and TANK binding kinase 1 [31].

While these findings in multiple disease states suggest that impaired mitochondrial quality control could represent a common mechanism in neurodegeneration, it is important to keep in mind that each of these proteins also has functions other than mitophagy [32]. For example, PINK1 and parkin also cooperate to regulate mitochondrial biogenesis by degrading a repressor of proliferator-activated receptor gamma coactivator-1-alpha (PGC-1α) in SH-SY5Y cells and DA neurons of conditional Pink1-RNAi transgenic mice [33]. PINK1 and parkin can regulate localized translation at the mitochondrial surface in HEK293 cells, with altered nuclear encoded respiratory chain mRNA expression in Drosophila and dopamine neurons induced from fibroblasts [34]. PINK1 and parkin may also regulate mitochondrial protein import in HEK293 cells and primary patient-derived fibroblasts [35]. Cytosolic PINK1 has been implicated in regulating pro-survival signaling pathways such as BDNF during development and Akt in SH-SY5Y cells [36, 37]. Moreover, cytosolic PINK1 shows enhanced interaction with VCP, acting to promote differentiation and dendritic arborization of primary cortical neurons [23] (Fig.1d-f).

At this time, it is unclear which of the many functions of PINK1 or Parkin, in relation to mitochondrial biology or in relation to other signaling pathways, contribute to disease pathogenesis when lost. Thus, it is important to study the normal biology of these proteins throughout brain development in addition to effects of disease-causative mutations and aging.

Moreover, there are other pathways by which damaged mitochondria can be recognized for mitophagy, and this functional redundancy emphasizes the importance of mitochondrial quality control. In vivo mouse and Drosophila knockout studies reveal that basal mitophagy proceeds normally in the absence of PINK1 and Parkin [38, 39], although they do play a role in mitophagy induced in Drosophila larvae by hypoxia or high doses of rotenone [40] and in ionophore-induced mitophagy in dopamine neurons differentiated from engineered PINK1 knockout iPSCs [41]. In contrast, engineered respiratory chain deficiency in the MitoPark mouse leads to clearance of defective mitochondria through a parkin-independent mechanism [42]. Pathways of PINK1- and parkin-independent mitophagy have been recently reviewed [43].

Upregulation of alternative mitophagy pathways could explain the increases in mitophagy observed in PINK1-deficient systems [11]. In primary neurons treated with sublethal doses of parkinsonian toxins 6-hydroxydopamine or rotenone, mitophagy is mediated by phospholipid scramblase-3 catalyzed externalization of cardiolipin, an inner mitochondrial membrane phospholipid, to the surface of the mitochondrion [44]. Subsequently, the intermembrane space protein NM23-H4 was also implicated along with OPA1 in cardiolipin-mediated mitophagy in MLE-12, HeLa and SH-SY5Y cells [45]. The upregulation of mitophagy receptors such as Nix has been described as a compensatory mechanism in fibroblasts biopsied from a homozygous parkin mutation carrier that has remained disease-free [46]. Damaged mitochondria may also be removed from neurons linked to AD or ALS through mitophagy-independent mechanisms involving late endosomes [47], and axonal mitochondria may be transferred to astrocytes for clearance at the optic nerve head [48]. It is possible that declining efficiencies of these alternative or compensatory pathways (or lysosomal function in general during aging) may account for the delayed onset of neurodegeneration, which typically occurs decades after birth even in families with early-onset mutations.

Intra-mitochondrial proteostasis

While autophagy represents a major mechanism of quality control, excessive autophagy contributes to muscle atrophy and to the retraction of neuronal processes important for maintaining synaptic contacts [49-51]. Neurons require sufficient mitochondria to support axonal and dendritic extension. Selective mitophagic clearance has been shown to precede dendritic retraction in response to mutations in LRRK2 [52]. RNAi to reduce expression of autophagy proteins can prevent both mitochondrial loss and simplification of the dendritic arbor [52]. Inhibiting mitochondrial fission also acts to curb excessive mitophagy in LRRK2 G2019S iPSC-derived dopaminergic neurons [53]. It is perhaps not surprising the neurons may show a higher threshold for mitophagy [54], as mitochondrial clearance that is not balanced by biogenesis is detrimental to neurons [55].

With lesser degrees of mitochondrial damage, other quality control pathways may be engaged resulting in more focused degradation. For example, PINK1 and parkin participate in the formation of mitochondrial derived vesicles containing only outer membrane components, which are delivered into the endolysosomal system [56]. Moreover, in Drosophila, PINK1 and Parkin are implicated in regulating respiratory chain protein turnover through a mechanism independent of the autophagy protein Atg7 [57]. Interestingly, respiratory chain proteins represent some of the most long-lived proteins in the brain [58, 59], suggesting that whole-scale mitochondrial turnover by mitophagy is either rare or that some mechanism exists for sparing respiratory supercomplexes.

Interestingly, PINK1 may regulate the activity of the intermembrane space serine protease HtrA2/Omi [60]. HtrA2 is phosphorylated in a PINK1-dependent manner upon activation of the p38 pathway. HtrA2 is not required for PINK1-parkin dependent mitophagy but does seem to play a parkin-independent role in maintaining mitochondrial homeostasis downstream of PINK1 [61]. Recently it has been proposed that HtrA2 mediates compensatory upregulation of UCP2-SIRT3-PGC1 signaling following brain ischemia-reperfusion injury [62]. Phosphorylation of TRAP1, also known as mitochondrial HSP75, which has been localized in different studies to either the intermembrane space or matrix fractions [63, 64] is also regulated by PINK1 [63]. In Drosophila and PINK1-silenced SH-SY5Y cells, overexpression of TRAP1 rescues mitochondrial fragmentation, but has no effect in rescuing loss of Parkin function [65]. PINK1 also regulates phosphorylation of complex I subunits that face the intermembrane space [66]. A super-resolution microscopy study shows that the kinase domain of PINK1 colocalizes with complex I inside energized mitochondria [67]. While these findings suggest that PINK1 could function as a kinase within energized mitochondria (Fig. 1d), it is also possible that it regulates mitochondrial proteins from the cytosol prior to mitochondrial import.

The AAA+ proteases LONP1, CLPXP and AFG2L2 are believed to mediate proteostasis in the mitochondrial matrix. Deletion of LONP1 results in accumulation of protein aggregates in the mitochondrial matrix, accompanied by accumulation of PINK1 and activation of the integrated stress response [68]. The integrated stress response acts to halt protein synthesis to allow for cellular recovery. CLPXP plays an important role in the mitochondrial unfolded protein response (mtUPR), which involves retrograde signaling to the nucleus via the transcription factors CHOP and ATF4 to induce expression of mitochondrial chaperones and proteases [69].

TAR-DNA binding protein 43 (TDP-43) is a protein that aggregates in ALS and a subset of FTDs. TDP-43 aggregation dysregulates mitochondrial membrane potential and ROS to activate the mtUPR [70]. While this study suggested that LONP1 plays an important role in limiting the buildup of TDP-43 at the mitochondria, other studies suggest that TDP-43 may impact mitochondrial homeostasis from the cytosol by intercepting nuclear encoded mitochondrial proteins and specific miRNAs [71].

Synaptic Mitochondria and the Maintenance of Neuronal Architecture

The function of neurons in communication, often across vast cellular distances, is inextricably tied to their morphology. Neuronal soma support an unusually vast network of axonal and dendritic branches that support numerous synaptic contacts. Differences in dendritic morphology regulate in vivo firing patterns of individual neurons [72], which contribute not only to behavior, but also organismal stress responses and susceptibility to neurodegeneration [73, 74]. Structural remodeling of synaptic contacts in turn contribute to the ability to learn from experiences.

Mitochondria are likely to play important roles both in development and in maintenance of these elaborate synapto-dendritic and axonal arbors. PINK1-deficiency impairs neurogenesis in the zebrafish dopaminergic system [75], the mouse hippocampal formation [76] and in human iPSC-derived organoids [77]. In contrast, elevating the levels of PINK1 by inhibiting degradation of endogenous PINK1 confers protection against MPP+-induced dendritic degeneration [78].

Mitochondrial Dynamics

Mitochondrial dynamics involving fission-fusion and transport play key roles throughout the lifecycle of the neuron. Most of the earliest reports of PINK1 and Parkin function describe effects on mitochondrial dynamics, although PINK1 deficiency causes mitochondrial fragmentation in otherwise untreated mammalian cells [11], but promotes fusion in Drosophila [79] and depolarized mitochondria due to impaired mitofusin degradation [27]. LRRK2 also regulates mitochondrial dynamics through interaction with the fission protein Drp1 [80].

Drp1-mediated mitochondrial fission is important not only for mitophagy and cell death, but also during mitosis and to permit transport of mitochondria out to peri-synaptic regions (Reviewed in [81]). Fission-fusion cycles serve to sequester damaged mitochondrial segments, or allow repair through mixing of mtDNA and other mitochondrial constituents [82]. An interesting recent study using Cos-7 cells and cardiomyocytes suggest that fission associated with mitochondrial biogenesis occurs in the midzone in concert with the adaptor MFF, whereas peripheral fission mediated by FIS1 leads to mitophagy and is preceded by lysosomal contact [83]. This is consistent with earlier observations that Drp1-dependent fission triggered by PINK1-deficiency requires components of the autophagy machinery [11]. It would be interesting to determine whether fission events that are required for effective mitochondrial transport and dendritogenesis [84] also involves distinct mediators in addition to Drp1.

Microtubule-dependent transport and synaptic mitochondria

As shown in a landmark 2004 study, the delivery and proper positioning of mitochondria within dendrites plays a key role in dendritic extension and branching, as well as in regulating the stability versus plasticity of individual synaptic contacts [85]. Indeed, peri-synaptic mitochondria represent the primary energy source that powers stimulus-induced protein translation [86]. It is thus not surprising that synaptic mitochondrial dysfunction has been linked to aging-related memory loss at ages where non-synaptic mitochondria still show normal function [87].

Mutations in the microtubule associated protein Tau, which cause several forms of dementia, offered the first clues that altered transport of synaptic vesicles or mitochondria may contribute to neurodegeneration. Recently, it has been shown that Tau mediates stress-induced losses in synaptic mitochondrial localization, resulting in dendritic atrophy and memory impairments [88]. PINK1 regulates mitochondrial motility in axons [89] and into dendrites [90, 91], and VCP mediates mitochondrial axonal transport [92]. PINK1 and VCP also cooperate to enhance dendritic arborization through effects on NSFL1C/p47 phosphorylation [23]. Interestingly, a recent study suggests that huntingtin normally functions to facilitate localized glycolysis for bioenergetic support of fast axonal transport [93]. In contrast mutant huntingtin fragments act to impair mitochondrial transport along microtubules [94].

Given the vast cellular distance between the soma and synaptic terminals, which can be several feet in primary motor and sensory neurons, local translation is necessary for mitochondrial protein biogenesis, as assessed using synaptosomes isolated from mouse hippocampi and adjacent cortices [95]. The transport of mRNAs along axons and dendrites is critically important to this process. The observation that several ALS-FTD linked mutations impact RNA handling, while others are involved in regulating cellular transport stimulated renewed interest in microtubule-based transport in neurodegeneration. Indeed, TDP-43 mutations impair the microtubule transport of mRNA granules in Drosophila larval neurons, mouse cortical neurons and ALS patient iPSC-derived motor neurons [96].

Proteomic studies indicate that synaptic mitochondria are distinct from non-synaptic mitochondria in mouse brains. Synaptic mitochondria show an enrichment in oxidative stress-related proteins such as Mn-SOD and tricarboxylic acid cycle proteins [97]. This may reflect observations that dendritic regions are more susceptible to mitochondrial damage, resulting in enhanced bioenergetic and redox perturbations compared to the soma [98, 99]. In particular, perisynaptic mitochondria are exposed to greater degrees of calcium flux due to synaptic activity. There may also be intrinsic differences that render synaptic mitochondria more sensitive to calcium overload than nonsynaptic mitochondria [100].

Mitochondrial calcium dysregulation and synaptic neurodegeneration.

Mitochondria play a key role in buffering calcium entering neuronal subcompartments during synaptic activity. The mitochondrial calcium uniporter (MCU) complex is capable of mediating high-capacity uptake of calcium into the mitochondrial matrix during synaptic spiking [101], and a brain specific MICU3 isoform confers a higher affinity to calcium in neurons [102]. Following this high-capacity uptake, mitochondria release calcium through activity of the mitochondrial sodium calcium exchanger (NCLX) [103]. Whereas mitochondrial calcium uptake plays a physiological role in upregulating ATP generation in response to synaptic activity [104], a recent set of studies implicate unbalanced mitochondrial calcium uptake and release in mediating synaptic degeneration in models of familial PD and AD.

In early studies of PINK1 loss of function, SH-SY5Y cells were more susceptible to mitochondrial calcium overload due to deficient calcium extrusion [105]. It was subsequently discovered that NCLX is activated by PKA, and introduction of a constitutively active NCLX-S258D phosphomimic rescues PINK1 knockout neurons from dopamine-induced cell death [106]. PINK1 was subsequently shown to phosphorylate PKA in its activation loop [23] as well as phosphorylating LETM1 at T192 [107], another mitochondrial calcium transporter with mixed activity [103, 107].

Defective mitochondrial calcium efflux is also observed in AD models. In the 3xTg-AD mouse model, defective mitochondrial calcium efflux precedes the development of amyloid and tau pathology, and memory loss is modulated by increasing or decreasing NCLX expression [108]. Furthermore, defective mitochondrial calcium efflux is also observed in iPSC-derived neurons with the 10 + 16 MAPT mutation that is linked to FTD [109]. These changes confer increased susceptibility to mitochondrial depolarization in response to calcium stimuli.

Primary neuron studies in mutant LRRK2 models of PD provided a mechanistic linkage between synaptic hyperactivity, mitochondrial calcium overload and synaptic neurodegeneration [110]. Using genetically encoded calcium sensors, the commonly studied LRRK2 mutations were found to not only elicit increased activity-dependent calcium influx through glutamate receptors and L-type calcium channels in primary cortical neurons [52, 111], but also to upregulate MCU expression in SH-SY5Y cells, mutant LRRK2 patient fibroblasts and human mid-frontal cortex, resulting in increased mitochondrial calcium flux in primary cortical neurons and patient fibroblasts [110]. This glutamate-triggered mitochondrial calcium overload triggers selective dendritic mitophagy in primary cortical neurons, which precedes dendritic retraction [52, 110].

Treatment with MCU inhibitors, MCU RNAi or constitutively active NCLX-S258D prevented both dendritic mitophagy and dendritic atrophy in cortical neurons and neurite shortening in SH-SY5Y cells [110]. MCU inhibitors are also neuroprotective in pink1−/− zebrafish [112], Abeta treated microglia [113], and a transgenic mouse model of cerebral beta-amyloidosis [114]. Recently it was shown that adenoviral driven MCU overexpression is sufficient to cause cortical neuron degeneration in the absence of synaptic hyperactivity [115].

Translational considerations

All facets of the mitochondrial life cycle are affected in chronic neurodegenerative diseases. These processes include mitochondrial biogenesis, fission-fusion, and transport to distal perisynaptic regions. Regulation of respiratory function, lipid biosynthesis, calcium homeostasis and mitochondrial stress signaling to the nucleus are also impacted, as well as local repair and proteolytic turnover of damaged segments by mitophagy.

With all these potential mechanisms by which mitochondrial dysfunction could contribute to neurodegeneration, it is still unclear which of the function(s) impacted by a given gene mutation plays a causal role in neurodegeneration. It suffices to state that the most heavily studied function and/or model system may not necessarily yield effective therapeutic strategies. While causation has traditionally been studied using reductionist approaches to isolate significant effects of a single perturbation, it is possible that disease is triggered following a network of small changes over many years, each of which may be insufficient by itself to cause significant synaptic loss or neuron cell death. For these reasons, studies aimed to understand the entire range of biological processes regulated by a given gene product are critically important. In addition, the relationship of mechanisms triggered by gene mutations to the multifactorial processes that contribute to sporadic neurodegeneration may require approaches utilizing the tools of systems biology.

Irrespective of which mitochondrial perturbations represent cause, consequence or amplifying factors of the neurodegenerative process in different diseases, the kinase PINK1 has been implicated in regulating a wide swath of mitochondrially-related processes (Fig.1). Increasing PINK1 levels is generally neuroprotective (and cytoprotective in a host of other tissues as well). Diminished expression of wild type PINK1 has also been observed in sporadic AD brain tissues [116] For these reasons, PINK represents an interesting therapeutic target even if there is still debate over the relative contributions of specific PINK1-regulated pathways in a given disease context.

Small molecule strategies targeting PINK1

One of the earliest activators of PINK1 function is based on kinetin riboside triphosphate (KTP) [117]. Using a form of KTP with a gamma thiophosphate (KTPγS), it was shown that both wild type PINK1 and its G309D mutant was able to more effectively utilize KTP compared to ATP. The enhanced activity was observed using either PINK1 autophosphorylation or phosphorylation of a TRAP1 peptide, and treatment of cells with the precursor kinetin elicited phosphorylation of Bcl-xL with appropriate modulation of other downstream effects of PINK1 signaling. However, kinetin treatment had no effects in preventing α-synuclein toxicity in vivo [118]. Subsequent efforts have been directed at chemical modifications to improve precursor cellular penetration and/or bioconversion to KTP [119].

Using the emerging structural information discussed above [21, 22], it may be feasible to identify or design small molecules that would promote PINK1 dimerization or stabilize conformations that promote autophosphorylation. Engineered molecular scaffolds have the potential to enhance PINK1 activation independently of mitochondrial depolarization. Although PINK1 autophosphorylation at S228 is important for creating a substrate-binding pocket for ubiquitin, phosphorylation at this site is not required for other substrates [20]. As the growing body of literature on cytosolic functions of either full-length or cleaved PINK1 emerges, strategies to enhance important functions that are independent of mitophagy will likely prove valuable.

Given that PINK1 is rapidly degraded within cells, strategies to prolong its half-life within cells would serve to upregulate PINK1 in its endogenous setting. PINK1 may be constitutively degraded through the action of ubiquitin ligases, although the role of a proposed N-end rule process is controversial [120]. It is also important to note that different pathways may be involved in degrading full-length versus cleaved PINK1 under basal versus stressed states.

Recently, FBXO7, another protein implicated in PD pathogenesis [121], was found to regulate the half-life of both full length PINK1 and a cleaved fragment. In vitro assays confirmed that FBXO7 increases PINK1 polyubiquitylation and identified several lysine residues important in regulating its stability in MLE12 cells [78]. A small molecule BC1464 reduced PINK1 interaction with FBXO7, stabilized PINK1 expression in cells and enhanced the detection of phospho-ubiquitin and phospho-PKA in SH-SY5Y cells. BC1464 not only reduced cell death induced by MPP+ in multiple models, but also showed a striking protection of the dendritic arbor in toxin-treated primary cortical neurons [78].

Although beyond the scope of the current review, PINK1 has been implicated in suppressing neuroinflammation in primary glial cultures and knockout mice [122, 123], although an earlier study demonstrated potentiation of IL-1β inflammatory signaling in HEK293 cells [124]. When administered i.p., BC1464 reduced inflammation in vivo in an experimental pneumonia model [78]. Interestingly, PINK1 knockout mice spontaneously develop lung fibrosis, possibly due to mtDNA-mediated inflammation [125].

PINK1 expression is stabilized in cells by interaction with the CDC37/HSP90 chaperone system [126] and by interaction with BAG2 and BAG5 [127, 128]. In contrast, it undergoes further acceleration of degradation in neuronally-differentiated SH-SY5Y cells chronically treated with low dose complex I inhibitors due to upregulation of BAG6 [129]. Its ability to form higher molecular weight complexes with other proteins such as VCP [23], which cooperates with proteins enriched in the nervous system to regulate spinogenesis [130], may also modulate its half-life. Future studies are needed to determine whether enhanced expression and/or activity of PINK1 binding partners act to stabilize PINK1 expression and promote its pathway-specific signaling.

Targeting other mitochondrial pathways

Another possible strategy to enhance ubiquitin-mediated mitophagy, the pathway regulated by PINK1 and parkin, involves modulation of deubiquitinases. In RPE1 and U2OS cells, USP30 has been shown to oppose PINK1-mediated mitophagy by removing ubiquitin from the mitochondrial surface [131]. These authors propose that this serves to prevent tonic activation of mitophagy. A small molecule targeting USP30 acts to enhance cardiac mitophagy in vivo [132], although it is unclear why mitophagy was not induced in other tissues. More research is needed, however, to determine if the presence of USP30 plays a protective role by setting a higher threshold for activation of mitophagy in perisynaptic structures that would poorly tolerate mitochondrial loss.

Based on the availability of small molecules that regulate MCU activity, as well as known protective effects of calcium uptake inhibitors such as NMDA receptor antagonists, normalizing peri-synaptic mitochondrial calcium fluxes may serve to protect against early stages of excitatory dendritic mitochondrial toxicity. However, care must be taken in this regard given the physiological roles of mitochondrial calcium uptake in regulating ATP generation. For example, in the PS2 model of AD, there is decreased, rather than increased mitochondrial calcium uptake, impacting ATP production [133]. Moreover, reduced mitochondrial calcium buffering may act to increase susceptibility to classic cytosolic mechanisms of excitotoxicity.

Strategies to enhance NCLX or LETM1 activity may prove more promising, as enhanced NCLX expression is protective in primary cortical neurons and the 3xTg model of AD [108, 110], even when the primary perturbation involves upregulation of MCU [110]. In this regard, promoting PKA-dependent phosphorylation and activation of NCLX is another potential strategy [106]. As PINK1 promotes PKA activation [23], strategies to facilitate their interaction represent another potential avenue to promote synaptic mitochondrial function through both mitochondrial and non-mitochondrial mechanisms.

Conclusions

The discovery of PINK1 mutations that cause early onset familial neurodegeneration provided the first genetic evidence supporting a causal role of mitochondrial dysregulation in neurodegeneration. While bioenergetics and cell death regulation are two prominent mitochondrial functions that have been heavily studied, a growing number of neurodegenerative disease models for PD, AD, ALS/FTD and HD implicate multiple additional facets of mitochondrial dysregulation in neurodegeneration.

One major theme involves homeostatic processes that regulate various mitochondrial quality control pathways. These range from autophagic clearance of entire mitochondrial segments to mechanisms that mediate more localized turnover of damaged components, and mitochondrial stress signaling that modulates transcription of nuclear encoded chaperones and antioxidants. Another emerging theme reflects the unique functional demands placed on synaptic mitochondria, and the role of mitochondrial calcium dyshomeostasis in synaptodendritic degeneration.

Given established roles of PINK1 in regulating mitochondrial quality control, mitochondrial biogenesis and transport, mitochondrial calcium handling and synaptodendritic arborization, strategies to upregulate PINK1 function has the potential to mediate neuroprotection through multiple mitochondrial pathways.

Abbreviations:

AD

Alzheimer’s disease

ALS

amyotrophic lateral sclerosis

ETC

electron transport chain

FTD

frontotemporal lobar dementia

HD

Huntington’s disease

iPSC

induced pluripotent stem cell

LC3

microtubule associated protein 2 light chain 3

LRRK2

leucine rich repeat kinase 2

MCU

mitochondrial calcium uniporter

MPP+

1-methyl-4-phenylpyridinium

MPTP

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine

NCLX

mitochondrial sodium calcium exchanger

mtUPR

mitochondrial unfolded protein response

PD

Parkinson’s disease

PINK1

PTEN-induced kinase 1

PKA

protein kinase A

TDP-43

TAR-DNA binding protein 43

TOM

translocase of the outer membrane

VCP

valosin containing protein