Acta Neuropathologica

, Volume 123, Issue 2, pp 173–188

Parkin, PINK1 and mitochondrial integrity: emerging concepts of mitochondrial dysfunction in Parkinson’s disease

Authors

  • Anna Pilsl
    • Neurobiochemistry, Adolf Butenandt InstituteLudwig Maximilians University
    • Neurobiochemistry, Adolf Butenandt InstituteLudwig Maximilians University
    • German Center for Neurodegenerative Diseases (DZNE)
Review

DOI: 10.1007/s00401-011-0902-3

Cite this article as:
Pilsl, A. & Winklhofer, K.F. Acta Neuropathol (2012) 123: 173. doi:10.1007/s00401-011-0902-3

Abstract

Mitochondria are dynamic organelles which are essential for many cellular processes, such as ATP production by oxidative phosphorylation, lipid metabolism, assembly of iron sulfur clusters, regulation of calcium homeostasis, and cell death pathways. The dynamic changes in mitochondrial morphology, connectivity, and subcellular distribution are critically dependent on a highly regulated fusion and fission machinery. Mitochondrial function, dynamics, and quality control are vital for the maintenance of neuronal integrity. Indeed, there is mounting evidence that mitochondrial dysfunction plays a central role in several neurodegenerative diseases. In particular, the identification of genes linked to rare familial variants of Parkinson’s disease has fueled research on mitochondrial aspects of the disease etiopathogenesis. Studies on the function of parkin and PINK1, which are associated with autosomal recessive parkinsonism, provided compelling evidence that these proteins can functionally interact to maintain mitochondrial integrity and to promote clearance of damaged and dysfunctional mitochondria. In this review we will summarize current knowledge about the impact of parkin and PINK1 on mitochondria.

Keywords

MitochondriaMitophagyNeurodegenerationParkinParkinson’s diseasePINK1

Introduction

Parkinson’s disease (PD) is the most common movement disorder and the second most common neurodegenerative disease after Alzheimer’s disease. Advanced age is the most important risk factor for sporadic PD, and thus, in an aging population the incidence and prevalence of PD will significantly increase over the next decades. Pathologically, PD is characterized by a preferential loss of dopaminergic neurons in the substantia nigra pars compacta (SNc) which project to the striatum. Degeneration of the nigrostriatal pathway and depletion of striatal dopamine are responsible for the most prominent motor symptoms of PD patients, comprising bradykinesia, hypokinesia, rigidity, resting tremor, and postural instability. However, the neurodegenerative process is not limited to dopaminergic neurons; it also affects noradrenergic (locus coeruleus), serotonergic (dorsal raphe nucleus), and cholinergic (nucleus basalis of Meynert) systems, the cerebral cortex, brain stem, spinal cord, and the peripheral autonomic nervous system [9, 42, 87], explaining non-motor clinical manifestations, such as autonomic dysfunction, sleep abnormalities, depression, and cognitive impairment. Another pathological hallmark of PD is the presence of proteinaceous deposits in neuronal perikarya (Lewy bodies) and processes (Lewy neurites), which contain aggregated α-synuclein species [153]. Mounting evidence from different models suggests that not the final protein aggregates but rather oligomeric intermediates are responsible for neurotoxic processes (reviewed in [13, 183]). Consequently, the pathogenetic role and significance of Lewy bodies have remained elusive (reviewed in [12, 123]). Lewy bodies may confer a protective function by sequestering toxic misfolded synuclein species. On the other hand, they may act as a reservoir and source for toxic protein species, based on the fact that proteinaceous inclusions are dynamic [83, 96, 110, 173]. Notably, Lewy bodies are not found in all parkinsonian syndromes; they are absent in some familial cases and have not been reported in MPTP-induced parkinsonism in humans, emphasizing that PD is not a uniform disease entity [28, 89, 122].

The etiopathogenesis of sporadic PD, the most common form of parkinsonism, seems to be complex with variable contributions of genetic susceptibility and environmental factors (Fig. 1). A major breakthrough in PD research was the identification of monogenetic variants, which account for up to 10% of all PD cases. So far, six genes have conclusively been linked to PD (reviewed in [56, 105, 122, 149]). Autosomal-dominant parkinsonism is caused by mutations in the genes encoding α-synuclein or LRRK2 (leucine-rich repeat kinase 2, dardarin), whereas mutations in the genes encoding parkin, PINK1 (PTEN-induced kinase 1), DJ-1, or ATP13A2 lead to autosomal-recessive parkinsonism (Fig. 1). Recent insight into the function and dysfunction of PD-associated genes has reinforced the notion that mitochondrial dysfunction plays a crucial role in PD. First evidence for a possible link between mitochondria and PD was provided in the late 1970s/early 1980s when young drug addicts in the United States developed acute and irreversible parkinsonism after intravenously using a meperidine analog which was contaminated by MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) [28, 88]. MPTP crosses the blood–brain barrier, is oxidized to the toxic species MPP+ in glial cells, and can then be taken up by dopaminergic neurons via the dopamine transporter [68, 104]. In dopaminergic neurons MPP+ has been shown to inhibit complex I activity of the mitochondrial electron transport chain, resulting in a decrease in oxidative phosphorylation, and an increase in the generation of reactive oxygen and nitrogen species [17, 121, 133]. Prompted by insights into the molecular action of MPTP, complex I activities were analyzed in tissues from PD patients. Indeed, in post-mortem SNc samples from PD patients complex I activities were found to be reduced [141], whereas data from peripheral tissue, such as blood cells or skeletal muscle, are less consistent (reviewed in [140]).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-011-0902-3/MediaObjects/401_2011_902_Fig1_HTML.gif
Fig. 1

Etiology of Parkinson’s Disease (PD). Sporadic PD is a complex multifactorial disease with variable contributions of environmental factors, such as exposure to toxins and genetic susceptibility. Several polymorphisms have been identified in genome-wide association studies (GWAS) as genetic risk factors for PD, among them polymorphisms in the genes encoding α-synuclein, LRRK2, and tau [138, 151]. Familial forms account for up to 10% of PD cases. Autosomal-dominant parkinsonism is caused by mutations in the genes encoding α-synuclein or LRRK2, whereas mutations in the genes encoding parkin, PINK1, DJ-1, or ATP13A2 are associated with autosomal recessive parkinsonism

The interest in mitochondrial alterations linked to PD tremendously increased when it became evident that some PD-associated gene products have a direct or indirect impact on mitochondrial integrity (reviewed in [1, 7, 10, 11, 59, 103, 140, 170, 171, 181]). The most compelling link between PD genes and mitochondria has emerged from studies on PINK1 and parkin; we, therefore, concentrate on mitochondrial effects mediated by these proteins in the following sections.

PINK1: a mitochondrial kinase

The PINK1 gene encodes a 581 amino acid protein which is ubiquitously expressed. Similarly to parkin, the PINK1 protein is highly conserved between species. PINK1 contains an N-terminal mitochondrial targeting signal sequence, a transmembrane sequence, and a C-terminal serine-threonine kinase domain (Fig. 2). Mutations in the PINK1 gene are the second most common cause of autosomal recessive parkinsonism after mutations in the parkin gene [167]. About 30 pathogenic PINK1 mutations have been associated with the disease, for instance missense mutations, truncations, deletions, and large genomic rearrangements (reviewed in [122]). Pathogenic PINK1 mutations suggest a loss-of-function mechanism. Mutations cluster in the kinase domain and most mutations either compromise the kinase activity of PINK1 or interfere with protein stability (reviewed in [32, 73, 162]). Supporting a loss-of-function scenario, Drosophila PINK1 mutants display reduced life span, male sterility, and apoptotic flight muscle degeneration [24, 127, 187]. In contrast, deletion of PINK1 in mice does not result in an overt phenotype. Aged PINK1 knockout (KO) mice display only minor deficits, such as a decrease in total dopamine levels and an impaired synaptic plasticity in the striatum, but no nigrostriatal degeneration [49, 78, 79, 194]. However, mitochondrial respiration is impaired in the striatum of PINK1 knockout mice, and respiration deficits can be induced in the cortex by cellular stress [45]. In line with a bioenergetic deficit in the absence of PINK1, Morais et al. [114] reported a decrease in complex I activity in tissues from PINK1-deficient mice and flies. The mobilization of reserve pool synaptic vesicles at the neuromuscular junction of PINK1-deficient flies is impaired during rapid stimulation due to synaptic ATP depletion, indicating that synaptic activity cannot be maintained under increased energy demand in PINK1-deficient neurons.
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-011-0902-3/MediaObjects/401_2011_902_Fig2_HTML.gif
Fig. 2

Domain architecture of PINK1 and parkin. Protein domains are highlighted by colors; numbers below indicate amino acid residues

The subcellular localization of PINK1 is still extensively discussed. Although PINK1 contains a mitochondrial targeting signal and a putative transmembrane sequence [150], several groups have identified a proportion of the protein in the cytosol [4, 55, 97, 157, 176]. Zhou et al. [193] presented evidence that the PINK1 protein resides in the outer mitochondrial membrane with its kinase domain facing the cytoplasm. Interestingly, cytosolic PINK1 lacking the mitochondrial import sequence and the transmembrane domain was reported to protect against MPTP-induced toxicity in mice, suggesting that cytosolic PINK1 may have a function in neuronal survival [55]. The analysis of the subcellular localization of PINK1 has been hindered by the lack of an appropriate antibody detecting endogenous PINK1. Moreover, PINK1 is a low-abundance protein under basal conditions. Recent evidence indicated that PINK1 protein levels are kept low through instant shedding of the protein during or shortly after mitochondrial import, most likely via the proteolytic activity of PARL (presenilin-associated rhomboid-like protease) [31, 70, 109, 146, 179]. However, when the mitochondrial membrane potential is dissipated, PINK1 processing is inhibited and full-length PINK1 is stabilized at the outer mitochondrial membrane [4, 31, 70, 98, 109, 120]. Stabilization of PINK1 at the outer mitochondrial membrane is essential for parkin recruitment to dysfunctional mitochondria and subsequent mitophagy, which will be discussed in “Parkin and PINK1: a couple that promotes mitophagy”.

So far, three putative PINK1 substrates have been identified. Pridgeon et al. [132] observed that PINK1 interacts with the mitochondrial chaperone TRAP1 (TNF-receptor associated protein 1, also known as Hsp75). The authors of this study showed that PINK1 can phosphorylate TRAP1, which is important to mediate the protective activity of PINK1 against oxidative stress. Plun-Favreau and coworkers reported that PINK1 functionally interacts with the mitochondrial serine protease HtrA2/OMI (high-temperature regulation A2) [129]. Upon stimulation of the p38 stress kinase pathway, HtrA2/OMI is phosphorylated at a conserved serine residue in a PINK1-dependent manner. Investigation of this pathway in Drosophila suggested that HtrA2/OMI acts in a common pathway downstream of PINK1, but independently of parkin [156, 179, 192]. Recently, two studies reported that PINK1 can directly phosphorylate parkin [75, 145], providing evidence that both proteins act in a common pathway (see "Parkin and PINK1: common and separate pathways converging on mitochondria"). PINK1 has also been found in a multiprotein complex together with Miro and Milton [177]. The atypical GTPase Miro and the adaptor protein Milton are two proteins at the outer mitochondrial membrane that link kinesin heavy chains to mitochondria for anterograde axonal transport along microtubules. Thus, PINK1 may have a function in mitochondrial transport. Moreover, it is highly probable that the decrease in mitochondrial membrane potential in parkin- or PINK1-deficient cells (see “Mitochondrial dynamics: the fission and fusion confusion”) has an impact on the efficiency of mitochondrial transport [111].

An important physiological function of PINK1 is to increase the cellular stress resistance. Overexpression of PINK1 protects from cell death induced by various toxins, while PINK1 depletion increases the vulnerability to stress-induced cell death (reviewed in [32]). These findings suggest that PINK1 plays a role in maintaining cellular homeostasis under cellular stress. Several mechanisms have been proposed to explain the cytoprotective activity of PINK1, including effects on mitochondrial bioenergetics and calcium homeostasis. As mentioned above, a reduction in complex I enzymatic activity has been observed in Drosophila and mouse models lacking PINK1 [45, 114]. This defect in complex I activity could be responsible for decreased ATP levels, elevated reactive oxygen species (ROS), and compensatory up-regulation of antioxidative pathways in PINK1-deficient models [26, 53, 62, 185]. An imbalance in calcium homeostasis was observed in neurons derived from human embryonic stem cells when PINK1 is downregulated by RNA interference [43]. In this model, PINK1 deficiency resulted in mitochondrial calcium overload under cellular stress due to the dysfunction of the mitochondrial Na+/Ca2+ exchanger. In addition, there is substantial experimental evidence for a role of PINK1 in the regulation of mitochondrial dynamics and mitochondrial quality control, which will be addressed below.

Parkin: an E3 ubiquitin ligase

In 1998, mutations in the gene encoding parkin have been identified as a cause of autosomal recessive parkinsonism [77]. Since then, more than 100 pathogenic mutations have been reported resulting in a loss of parkin function (reviewed in [56, 57, 122]). Notably, inactivation of parkin may also play a role in sporadic PD. Severe oxidative, nitrosative, and dopamine stress can inactivate parkin by inducing its misfolding, and misfolded parkin species have indeed been identified in the brains of patients suffering from sporadic PD [23, 90, 91, 142, 174, 182, 184, 190].

Parkin, a 465 amino acid cytosolic protein, is characterized by a ubiquitin-like (UBL) domain at the N-terminus and a RBR (RING-between-RING) domain close to the C-terminus (Fig. 2). The RBR domain coordinates zinc ions and is composed of two RING fingers which flank an in-between RING (IBR) domain. An additional RING finger domain (RING0) has been identified between the UBL and RBR motifs, which also shows zinc-binding activity [64]. The domain architecture of parkin suggested that it acts as an E3 ubiquitin ligase, catalyzing the covalent attachment of ubiquitin to lysine residues within substrate proteins. Interestingly, parkin can mediate different modes of ubiquitination, ranging from monoubiquitination to polyubiquitination involving different lysine residues within the ubiquitin molecule (for example Lys48- and Lys63-linked ubiquitin chains). Due to the presence of 7 lysine residues within ubiquitin, polyubiquitin chains with different topologies and functions can be formed (reviewed in [3, 65, 82]). Ubiquitin linkage via Lys48 usually targets substrates for degradation by the proteasome, whereas other modes of ubiquitination serve a wide range of regulatory functions implicated in signal transduction, DNA repair, endocytosis, and autophagy. From intensive research over more than 10 years a long list of putative parkin substrates and various parkin activities have emerged (reviewed in [29, 113, 165, 180]). A common denominator of these studies is the remarkably wide and robust neuroprotective activity of parkin. The increased expression of parkin protects from cell death induced by mitochondrial toxins, excitotoxins, endoplasmic reticulum stress, and proteotoxic stress in various cellular and animal models [8, 27, 41, 60, 61, 66, 69, 100, 116, 128, 136, 154]. Several pathways have been implicated in the neuroprotective activity of parkin, such as the NF-κB pathway [60, 145], JNK signaling [15, 58], and PI3K signaling [40, 187]. Recently, the identification of PARIS (parkin-interacting substrate) as a new parkin substrate significantly increased our understanding of the neuroprotective activity of parkin [148]. Shin et al. could show that parkin regulates the levels of PARIS via the ubiquitin–proteasome system. PARIS represses the expression of PGC-1α (peroxisome proliferator-activated receptor gamma-coactivator 1-alpha) by binding to its promoter. PGC-1α acts as a transcriptional co-activator which regulates mitochondrial biogenesis [135, 139, 178]. Degradation of PARIS by parkin increases PGC-1α-dependent gene expression and mitochondrial biogenesis. Consequently, loss of parkin function leads to the accumulation of PARIS, suppressing mitochondrial biogenesis by PGC-1α.

In line with a prominent role of parkin in conferring neuroprotection, parkin gene expression is up-regulated in various stress paradigms [8, 60, 61, 66, 158]. We could recently demonstrate that ATF4, a transcription factor of the unfolded protein response, mediates the transcriptional up-regulation of parkin in response to mitochondrial, and endoplasmic reticulum stress, whereas c-Jun represses parkin expression [8].

Similarly to PINK1 KO mice, parkin-deficient mice do not develop nigrostriatal degeneration (reviewed in [30, 102, 161]. However, several reports provided evidence for mitochondrial alterations, at least for an increased vulnerability to mitochondrial damage in the absence of parkin. Expression of proteins involved in mitochondrial function (subunits of complexes I and IV) and in the oxidative stress response is decreased in the ventral midbrain of parkin KO mice [125]. Accordingly, the respiratory capacity of striatal mitochondria isolated from parkin-deficient mice is reduced [125]. Alterations in mitochondrial function reflected by a decrease in ATP production were also observed in skin fibroblasts derived from patients carrying pathogenic parkin mutations [54, 115].

Parkin and PINK1: common and separate pathways converging on mitochondria

Evidence for a functional or epistatic interaction of PINK1 and parkin was first provided by studies in Drosophila melanogaster. Flies deficient for parkin or PINK1 display similar phenotypes, including reduced life span, male sterility, and apoptotic flight muscle degeneration [24, 52, 127, 187]. Remarkably, overexpression of parkin in PINK1 mutant flies is able to restore the PINK1 mutant phenotype, while PINK1 cannot compensate for parkin loss of function, suggesting that PINK1 and parkin act in a common pathway with parkin acting downstream of PINK1. Consistent with this notion, parkin is able to rescue mitochondrial alterations induced by PINK1 deficiency in cultured human cells [39, 101]. In cellular models, acute silencing of parkin or PINK1 causes similar mitochondrial defects, including mitochondrial fragmentation with ultrastructural damage of cristae, a decrease in the mitochondrial membrane potential, and ATP production [26, 39, 101, 137].

Based on the observed epistatic interaction between parkin and PINK1, an important question is whether these proteins interact directly or indirectly. In principle, parkin and PINK1 could interact indirectly via parallel pathways or within a cascade of a linear pathway. As the kinase domain of PINK1 is exposed to the cytosol and parkin can be recruited to mitochondria in a PINK1-dependent manner, a direct interaction between parkin and PINK1 is also conceivable, at least under certain conditions. Support for such a direct interaction was provided by several studies. Upon overexpression of parkin and PINK1 in cultured cells, parkin can be co-purified with PINK1 and vice versa [113]. Kim et al. [75] reported that PINK1 recruits parkin to mitochondria by phosphorylating a threonine residue within the RING0 domain of parkin. Phosphorylation of parkin by PINK1 was also observed in another study, showing that the E3 ligase activity of parkin for catalyzing Lsy63-linked polyubiquitination is enhanced upon PINK1-induced phosphorylation of parkin [145]. Stabilization and upregulation of PINK1 by parkin was demonstrated in cultured cells upon the overexpression of both parkin and PINK1 [147, 166]. In an approach to analyze the interaction of parkin and PINK1 in vivo, endogenous parkin and PINK1 could be co-immunoprecipitated in lysates from the striatum and substantia nigra of rats [166] or from human brain [145].

Given that both parkin and PINK1 confer protection from stress-induced cell death, an interesting question in this context is whether the protective activities of parkin and PINK1 are dependent on each other. As outlined above, parkin can mediate protection via an increase of PGC-1α levels. Whether PINK1 is implicated in the neuroprotective pathway linking parkin, PARIS, and PGC-1α has not been reported so far. Notably, Berger et al. [6] demonstrated that parkin can prevent cytochrome c release induced by proapoptotic BH3 domains, while PINK1 cannot, indicating that parkin and PINK1 may also have different protective activites.

Parkin and PINK1: a couple that promotes mitophagy

Macroautophagy (in the following referred to as autophagy) is a major cellular degradation system (reviewed in [38, 189]). First, it executes housekeeping functions through bulk degradation of cellular components. Second, autophagy can be induced under certain stress conditions, for example upon starvation to recycle cellular material and to provide cells with metabolites that can be used for energy production. Autophagy involves the formation of a phagophore which elongates to engulf cargo in the autophagosome, a double-membrane vesicular structure. Upon fusion of the autophagosome with a lysosome the vesicular content is eventually degraded. In addition to the non-selective sequestration of cytoplasmic material, selective autophagic degradation of cellular structures, such as organelles and proteins aggregates, can occur. Selective autophagy requires specific labeling of the cargo, for example by ubiquitination, and subsequent binding of adaptor proteins to recruit the autophagic machinery to the cargo (reviewed in [76, 84]).

Increasing evidence points towards a dysfunction of autophagy in several neurodegenerative diseases (reviewed in [22, 44, 95, 112]). A compelling link between PD and selective autophagy of mitochondria, termed mitophagy, was recently discovered in Richard Youle’s laboratory. Narendra et al. [119] observed that parkin is targeted to dysfunctional mitochondria and induces their degradation by mitophagy. When CCCP (carbonyl cyanide 3-chlorophenylhydrazone), an uncoupling agent that dissipates the mitochondrial membrane potential, was added to HeLa cells, a robust recruitment of overexpressed parkin to uncoupled mitochondria was observed after 1 h of treatment. Strikingly, upon prolonged exposure to CCCP, mitochondria were cleared from parkin-expressing cells within 24 h as determined by immunostaining against several mitochondrial marker proteins. The degradation was specific for mitochondria, as peroxisomal abundance and distribution were unchanged [119]. In a next step, the role of PINK1 in parkin-mediated mitochondrial degradation was analyzed. Several groups found that expression of PINK1 is essential for the recruitment of parkin to dysfunctional mitochondria [47, 74, 107, 120, 172]. Downregulation of PINK1 by RNA interference resulted in decreased parkin recruitment to mitochondria following CCCP treatment [47]. Moreover, parkin targeting to mitochondria was completely abolished in fibroblasts from PINK1 KO mice and could be rescued by reintroduction of wild-type PINK1 [120]. Thus, in PINK1-deficient cells, parkin cannot induce mitophagy after CCCP treatment (Fig. 3). Several pathogenic PINK1 mutants as well as an artificial kinase-dead PINK1 mutant failed to guide parkin to dysfunctional mitochondria [48, 75, 107]. Overexpression of PINK1 per se or targeting of PINK1 to mitochondria via a heterologous mitochondrial membrane anchor derived from Tom20 or OPA3 was shown to be sufficient for the mitochondrial recruitment of parkin [120]. As outlined above, parkin may directly interact with PINK1 at the outer mitochondrial membrane, but it is also conceivable that the PINK1-dependent recruitment of parkin involves other proteins (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-011-0902-3/MediaObjects/401_2011_902_Fig3_HTML.gif
Fig. 3

Mitophagy is compromised in PINK1-deficient fibroblasts. Mouse embryonic fibroblasts (MEFs) derived from PINK1 knockout (KO) mice or wild-type mice were lentivirally transduced to express human wild-type parkin. Cells were left untreated or treated with CCCP (10 μM) to dissipate the mitochondrial membrane potential. After 24 h the cells were fixed and immunostained with antibodies against parkin (green) and Tom20 (red) to visualize mitochondria. Notably, in PINK1-deficient MEFs mitochondria are fragmented but not degraded upon CCCP treatment. In wild-type MEFs mitochondria are removed in parkin-expressing cells after CCCP treatment (cell surface is marked in white to highlight cells cleared from mitochondria)

https://static-content.springer.com/image/art%3A10.1007%2Fs00401-011-0902-3/MediaObjects/401_2011_902_Fig4_HTML.gif
Fig. 4

Sequential steps of parkin-induced mitophagy. On healthy mitochondria, steady-state levels of PINK1 are kept low due to rapid shedding, and parkin is mainly found in the cytosol. Upon dissipation of the mitochondrial membrane potential, for instance by treatment with the uncoupling agent CCCP, proteolytic processing of PINK1 is abolished, resulting in accumulation of full-length PINK1 at the outer mitochondrial membrane. PINK1 stabilization is essential to recruit parkin to uncoupled mitochondria, possibly via a direct or indirect interaction between PINK1 and parkin. Parkin-induced ubiquitination of outer mitochondrial membrane proteins allows remodeling of the mitochondrial surface, facilitating the recruitment of adaptor proteins that link damaged mitochondria to the autophagic machinery. Dysfunctional mitochondria are engulfed in autophagosomes and can be degraded after fusion with lysosomes

Notably, it was observed that full-length PINK1 accumulates after CCCP treatment, whereas under basal conditions PINK1 is cleaved, generating a shorter cytosolic fragment that is rapidly degraded by the proteasome [70]. The stabilization of PINK1 at mitochondria is thought to trigger parkin translocation to uncoupled mitochondria, and therefore, it has been of great interest to identify the protease(s) mediating PINK1 cleavage. Studies in Drosophila have shown a genetic link between PINK1, parkin, and Rhomboid-7, the fly homolog of the presenilin-associated rhomboid-like protease (PARL), suggesting that PARL mediates PINK1 cleavage [179]. Indeed, experimental evidence has been presented that PARL cleaves PINK1 in mammalian cells under basal conditions, but not upon loss of the mitochondrial membrane potential by CCCP treatment [31, 70, 109, 146]. Fibroblasts derived from PARL-deficient mice show an altered PINK1 cleavage pattern that can be reconstituted by reintroducing wild-type PARL, but not by a catalytically inactive PARL mutant [70, 146]. However, PARL seems not to be the only PINK1-cleaving enzyme, as processed PINK1 isoforms are present in fibroblasts from PARL KO mice [146]. PARL is a transmembrane protein at the inner mitochondrial membrane catalyzing intramembrane proteolysis, while PINK1 seems to be anchored to the outer mitochondrial membrane; thus, the exact molecular mechanism of PARL-induced PINK1 cleavage remains to be demonstrated.

Since the discovery of parkin-mediated mitophagy, extensive effort has been taken to dissect the molecular mechanisms of this pathway. Besides PINK1 and PARL, various other proteins have been shown to be involved in parkin-induced clearance of damaged mitochondria. One of them is p62 (also termed SQSTM1, sequestosome-1), a protein previously implicated in selective autophagy [36, 47, 93, 118, 124]. p62 is an autophagy adaptor protein that recognizes ubiquitinated substrates and interacts with LC3, a protein anchored at maturating autophagosomes. p62 therefore links ubiquitinated cargo to autophagosomes for selective degradation (reviewed in [85]). Geisler et al. [47] observed that mitophagy is compromised upon silencing of p62 in HeLa cells. Similar results were obtained in two other studies using transient p62 downregulation in cultured cells [36, 93]. However, Narendra et al. [118] reported that parkin-induced mitophagy is neither impaired in HeLa cells depleted of p62 nor in fibroblasts derived from p62 KO mice. The authors of this study concluded that p62 is necessary for mitochondrial clustering at the perinuclear region, but is dispensable for parkin-induced mitophagy, which is in line with data from Okatsu et al. [124]. Despite these controversies, a consistent finding of all studies is that p62 is recruited to uncoupled mitochondria in the presence of parkin.

Mitophagy requires cooperation of the lysosomal and proteasomal pathway

Mitophagy completion involves a series of sequential steps (Fig. 4). Upon dissipation of the mitochondrial membrane potential, parkin is recruited to uncoupled mitochondria in a PINK1-dependent manner. Damaged mitochondria are then engulfed by autophagosomes and degraded after fusion with lysosomes. Consequently, inhibition of lysosomal function using chemicals such as bafilomycin A1 (an inhibitor of lysosomal V-type ATPase that interferes with lysosomal acidification) or 3-methyladenine (a PI3 kinase inhibitor that prevents formation of autophagosomes) blocks degradation of mitochondria following CCCP treatment [119, 168]. Mitophagy is also compromised in fibroblasts from mice lacking essential proteins of the autophagic machinery, such as ATG5 or ATG7 [107, 119, 124]. In addition, several proteins implicated in the autophagosomal-lysosomal system were recently described to be involved in parkin-mediated mitophagy. The histone deacetylase HDAC6 binds to ubiquitinated protein aggregates and promotes autophagosome maturation [72, 92]. Lee et al. [93] reported that HDAC6 is enriched at uncoupled mitochondria in parkin-expressing cells and essential for parkin-mediated mitophagy, as parkin failed to remove CCCP-treated mitochondria in fibroblasts from HDAC6 KO mice. Ambra1 (activating molecule in Beclin1-regulated autophagy) was recently identified as a parkin-interacting protein by a tandem affinity purification approach [168]. Ambra1 is dispensable for the recruitment of parkin to mitochondria but is required for the autophagic clearance of mitochondria by activating class III PI3K, which is essential for phagophore formation.

There is substantial evidence indicating that mitophagy is critically dependent on the ubiquitin–proteasome system (reviewed in [99]). Proteasomal inhibitors, such as MG-132 and epoxomycin, were found to almost completely abolish mitophagy [16, 159, 191]. Furthermore, a strong ubiquitin signal has been found at mitochondria upon CCCP treatment in parkin-expressing cells but not in cells lacking parkin [107, 124]. These findings raised the question whether proteins located at the outer mitochondrial surface might be ubiquitinated by parkin. Indeed, Chan et al. [16] reported wide-spread ubiquitination and degradation of outer mitochondrial membrane proteins upon dissipation of the mitochondrial membrane potential in parkin-expressing cells. Notably, overexpression of the ubiquitin mutant Lys48Arg, which is incapable of forming Lys48-linked ubiquitin chains, interfered with parkin-dependent mitophagy. Moreover, parkin mutants defective in E3 ligase activity showed reduced polyubiquitination of mitochondrial proteins and were unable to degrade mitochondria [93, 107].

In addition to Lys48-linked ubiquitin chains, which serve as a tag for proteasomal degradation, non-degradative ubiquitin linkage via Lys63 or Lys27 was detected upon parkin-induced mitophagy [16, 47]. Another putative parkin substrate on mitochondria found by Geisler et al. [47] is the voltage-dependent anion channel VDAC1. This protein was found to be ubiquitinated by parkin upon CCCP treatment in HeLa cells. However, fibroblasts from VDAC1/3 double KO mice were not compromised in parkin-mediated mitochondrial clearance [118]. Thus, although VDAC1 seems to be a target for parkin-dependent ubiquitylation on mitochondria, it is obviously not an essential substrate for mitophagy induction.

In conclusion, several outer mitochondrial membrane proteins have been identified as targets for parkin-mediated ubiquitination, including proteins involved in the regulation of mitochondrial dynamics such as mitofusins (see “Mitochondrial dynamics: the fission and fusion confusion”). Both degradative and non-degradative ubiquitination occurs upon mitochondrial uncoupling, suggesting that remodeling of the outer mitochondrial membrane and degradation of outer membrane proteins primes mitochondria for their clearance via autophagy.

Does dysfunctional mitophagy contribute to PD pathogenesis?

Importantly, several pathogenic parkin mutations have been found to interfere with parkin function at distinct steps in the mitophagic process [16, 47, 48, 93, 107, 124]. Consistent with a critical role of PINK1 in parkin-induced mitophagy, pathogenic PINK1 mutants are also compromised in mitophagy execution [47, 48, 74, 107, 120]. In addition, translocation of overexpressed parkin to uncoupled mitochondria is abrogated in dopaminergic neurons differentiated from induced pluripotent stem cells derived from patients carrying pathogenic PINK1 mutations [144].

Most of the studies on parkin-induced mitophagy have been performed using established cell lines, such as tumor cells or mouse embryonic fibroblasts and treating cells with CCCP or valinomycin, which cause rapid uncoupling of all mitochondria. Therefore, it is crucial to prove the in vivo relevance of parkin-induced mitophagy by demonstrating that parkin- or PINK1-deficient neurons are impaired in the clearance of dysfunctional mitochondria. Berman and coworkers addressed this important question and provided insight into bioenergetic differences between tumor cells and neurons and how these differences affect parkin-induced mitophagy [169]. The authors of this study observed parkin recruitment to mitochondria following CCCP treatment in tumor cells (HeLa and SH-SY5Y cells), but not in primary rat cortical neurons or mixed striatal/midbrain neuronal cultures. Cultured tumor cells generate ATP preferentially by glycolysis, whereas neurons depend on ATP production through oxidative phosphorylation [83, 110, 173]. Strikingly, when Van Laar et al. [169] forced HeLa cells into dependence on mitochondrial respiration, parkin no longer translocated to mitochondria upon CCCP treatment. This finding may indicate that neurons, which rely on mitochondria for energy production, do not easily eliminate their mitochondria. In line with such a scenario, Larsson and co-workers recently reported on a mouse model to study mitochondrial effects of parkin in vivo. The authors of this study made use of mice with a dopaminergic neuron-specific loss of TFAM, the mitochondrial transcription factor A [37]. Loss of TFAM abolishes mitochondrial DNA (mtDNA) expression and results in mtDNA depletion causing severe respiratory chain deficiency. Dopaminergic neurons lacking TFAM display large, aggregated mitochondria; however, mitochondrial recruitment of parkin was not observed [155]. Surprisingly, neither the removal of dysfunctional mitochondria nor the neurodegeneration phenotype of TFAM-deficient mice was affected by the absence of parkin [155].

Mitochondrial dynamics: the fission and fusion confusion

The mitochondrial fusion and fission machinery

Mitochondria appear as bean-shaped solitary organelles in electron micrographs, but with the availability of life cell imaging it became obvious that mitochondria are highly dynamic organelles in terms of their shape, size, and subcellular distribution (reviewed in [28, 35, 89]). Mitochondria frequently fuse and divide and move along cytoskeletal tracks, thereby building up a dynamic mitochondrial network (see also accompanying reviews by Karboski/Neutzner and Licci/Oettinghaus). The dynamic features of mitochondria are regulated by the opposing events of mitochondrial fusion and fission. The components of the fusion and fission machinery are evolutionary conserved from yeast to mammals, and we are just beginning to understand the molecular function of these factors and their impact on many essential cellular processes, such as bioenergetics, quality control pathways, and the regulation of cell death and viability (reviewed in [5, 152]). Moreover, dysfunction of the mitochondrial fusion and fission machinery has been linked to neurodegenerative diseases, underscoring the important role of mitochondrial dynamics in maintaining a healthy mitochondrial population.

The core machinery of mitochondrial dynamics is made up of four GTPases, the activity of which can be modulated by a variety of regulatory and accessory components. As mitochondria are double-membrane-bound organelles, fusion and fission of mitochondria are mechanistically challenging. In mammals, outer mitochondrial membrane (OMM) fusion is mediated by mitofusins (MFN1 and MFN2), which are anchored to the OMM by two transmembrane domains with the main part of the protein facing the cytosol. For inner mitochondrial membrane (IMM) fusion, OPA1 (optic atrophy protein 1) is required. OPA1 is located in the intermembrane space between the OMM and IMM. OPA1 functions are regulated by alternative splicing and proteolysis, and a combination of long and short OPA1 isoforms has to be present for mitochondrial fusion activity (reviewed in [86, 186]). Mitochondrial fission is mediated by the dynamin-related cytosolic protein DRP1 (dynamin-related protein 1), which can be recruited to mitochondrial fission sites. In yeast, FIS1, a tail-anchored protein at the OMM has been shown to mediate recruitment of DRP1 to mitochondria by an indirect interaction [14]. Mammalian FIS1 is not required to recruit DRP1 to mitochondria and its role in mitochondrial fission is still being debated [94]. Assembly of DRP1 at mitochondria promotes formation of spirals that constrict and sever the membrane following GTP hydrolysis.

When mitochondrial dynamics are balanced, mammalian cells usually display a tubular network of mitochondria. An increase in mitochondrial fission results in small rod-like or spherical mitochondria, whereas a prevalence of mitochondrial fusion generates elongated, highly interconnected mitochondria. A shift in the balance between mitochondrial fusion and fission has consequences on mitochondrial bioenergetics, transport, and clearance; thus, mitochondrial dynamics need to be tightly regulated. Regulation involves posttranslational modifications of the core fission and fusion proteins, such as phosphorylation (DRP1), ubiquitination (DRP1, FIS1, MFN1/2), SUMOylation (DRP1) proteolytic processing (OPA1), or influence on the transcriptional level.

Mitochondrial dynamics and quality control

Mitochondrial fusion is important for mixing of mitochondrial contents and lipid membranes and for the maintenance of mtDNA (reviewed in [1921, 63]). Thus, fusion of dysfunctional with functional mitochondria can dilute defective mitochondria and promote functional complementation [117]. Indeed, mitochondrial hyperfusion induced by various stress stimuli is accompanied by an increase in mitochondrial ATP production and reflects an adaptive and protective stress response [163]. On the other hand, mitochondrial fission seems to be required for proper mitochondrial transport [67]. Moreover, the formation of smaller mitochondrial units by mitochondrial fission facilitates the removal of dysfunctional mitochondria with a lower mitochondrial membrane potential by mitophagy [164]. Vice versa, increased mitochondrial fusion, for example in response to nutrient deprivation, protects mitochondria from autophagic degradation [51, 134]. Hence, both mitochondrial fusion and fission contribute to mitochondrial quality control in a context-specific manner.

Mitochondrial dynamics and neurodegeneration

Neurons are particularly vulnerable to alterations in mitochondrial dynamics (reviewed in [18, 80, 81, 108, 143]). First, neurons are characterized by high rates of metabolic activity. Second, the dimensions and polarity of neurons require efficient transport of mitochondria to sites of high energy demand, such as dendrites and axons. Third, the postmitotic state of neurons necessitates highly efficient mitochondrial quality control systems (reviewed in [106, 160]). Thus, it is not surprising that mutations in genes of the mitochondrial fusion/fission machinery are associated with neurological diseases. Mutations in MFN2 are responsible for Charcot-Marie-Tooth type 2A, a peripheral neuropathy that affects both sensory and motor neurons [196]. Mutations in OPA1 are the most common cause of autosomal-dominant optic atrophy (ADOA), characterized by loss of retinal ganglion cells and atrophy of the optic nerve [2, 33]. More recently, PD genes have been linked to mitochondrial dynamics. Studies in Drosophila revealed that the phenotype of parkin- or PINK1-deficient flies can be rescued by increasing mitochondrial fission or decreasing fusion, leading to the conclusion that parkin and PINK1 promote mitochondrial fission [34, 126, 130, 188]. However, there is an obvious discrepancy to the effects of parkin and PINK1 on mitochondrial dynamics in mammalian models. In cultured cells and primary neurons the acute silencing of parkin or PINK1 induces a transient but robust increase in mitochondrial fragmentation accompanied by a decrease in the mitochondrial membrane potential and ATP production [25, 26, 39, 101, 137, 175, 177]. In contrast to the fly model, the mitochondrial phenotype in mammalian cells can be rescued by increasing mitochondrial fusion (by enhanced expression of MFN2 or OPA1) or decreasing fission (by enhanced expression of dominant negative DRP1) [25, 26, 101, 137]. Notably, mitochondrial fragmentation upon silencing of parkin or PINK1 does not occur in DRP1-deficient cells, indicating that the alterations in mitochondrial morphology and dynamics are associated with an increase in DRP1-induced mitochondrial fission [101, 137]. Moreover, both parkin and PINK1 can protect from mitochondrial fragmentation induced by DRP1 overexpression or cellular stress [8, 101, 137]. The mitochondrial phenotype observed after transiently reducing the expression of parkin or PINK1 can be compensated in stable parkin/PINK1-deficient cells, explaining why tissues from parkin/PINK1 KO mice do not display increased mitochondrial fragmentation under steady-state conditions. However, stable parkin/PINK1-deficient cells, including fibroblasts from patients carrying pathogenic mutations, seem to be more vulnerable to stress-induced mitochondrial fragmentation, indicating an imbalance in mitochondrial dynamics when compensating mechanisms are overloaded under cellular stress [54, 62, 115].

How can the discrepant observations in Drosophila and mammalian cells be explained? First, the time of phenotype analysis is crucial. Mitochondrial fragmentation upon downregulation of parkin or PINK1 is an early and transient phenomenon, which obviously has to be compensated in order to prevent irreversible cellular damage. Second, both compensating strategies and the regulation of mitochondrial dynamics may be different in mammalian cells compared with Drosophila cells (reviewed in [181]). In line with such a scenario, we indeed observed an increase in mitochondrial fragmentation in Drosophila S2 cells early upon silencing of parkin or PINK1, which was rapidly compensated by hyperfusion [101]. This observation may reflect an (over-) activation of mitochondrial fusion in fly cells, which does facilitate dilution of dysfunctional mitochondria, but does not favor elimination of damaged mitochondria by mitophagy. Thus, mitochondrial hyperfusion occurring in fly cells as a compensatory strategy may not be beneficial in the long run, particularly in tissues with high energy demands. By increasing mitochondrial fission smaller mitochondrial units are generated which can more easily be taken up by autophagosomes and removed via mitophagy, explaining why stimulating fission can rescue mitochondrial defects in parkin- or PINK1-deficient flies. This hypothesis is based on the assumption that basal mitophagy is still working in flies lacking parkin or PINK1, but the role of mitophagy in Drosophila models has not been addressed so far.

To add another layer of complexity, it has recently been reported that mammalian mitofusins (MFN1 and MFN2) and the Drosophila mitofusin ortholog MARF are degraded by the proteasome in a parkin-dependent manner in response to mitochondrial uncoupling [16, 46, 50, 131, 159, 195]. As a consequence of reduced mitofusin levels, parkin promotes mitophagy by inhibiting mitochondrial fusion (reviewed in [71]). However, parkin can induce mitophagy in cells derived from MFN1/2 double KO mice [159], indicating that ubiquitination and degradation of mitofusions are not required for mitophagy. As outlined above, inhibition of the proteasome suppresses parkin-induced mitophagy; thus, it is likely that ubiquitination and proteasomal degradation of other proteins at the OMM contribute to the induction of mitophagy. Of note, parkin-induced alterations of MFN levels is not a consistent finding; two studies using cultured human cells reported that parkin can decrease the levels of the pro-fission proteins DRP1 or FIS1, respectively [25, 175].

In conclusion, parkin may modulate mitochondrial dynamics in a context- and tissue-specific manner. It seems plausible that under moderate cellular stress parkin prevents mitochondrial fragmentation and favors mitochondrial fusion, allowing transfer of mitochondrial contents from healthy to damaged mitochondria. Under severe stress conditions when mitochondria are irreversibly damaged parkin may favor mitochondrial fission, which helps to sequester dysfunctional mitochondria and facilitates their elimination via mitophagy. Although direct effects of parkin on the mitochondrial fusion/fission machinery have been reported, it is also conceivable that indirect effects account for the influence on mitochondrial dynamics. The latter possibility also applies for PINK1, which could affect mitochondrial dynamics indirectly via modulating mitochondrial bioenergetics.

Perspective

The identification of genes associated with rare familial variants of PD launched a new era in PD research. Considerable progress has been made in recent years to shed light on the function and dysfunction of PD-associated genes. A consistent result of these efforts is that at least some of the PD genes interface with pathways regulating mitochondrial function, dynamics and quality control. This finding is particularly appealing in that it provides a rationale for a common etiopathogenic factor in sporadic and familial PD. A plethora of cell culture studies provided conclusive evidence for a role of parkin and PINK1 in maintaining mitochondrial integrity. Yet, a wide range of questions remains to be explored. The next challenging steps will be to unravel the mechanisms underlying the observed activities of parkin and PINK1 and most importantly, to evaluate their physiological and pathophysiological relevance in the organismal context. Finally, the ultimate aim would be to translate insights from basic research into novel therapeutic strategies targeting basic mitochondrial processes.

Acknowledgments

We thank Daniela Vogt Weisenhorn and Wolfgang Wurst for providing mouse embryonic fibroblasts from PINK1 KO mice. K.F.W. is supported by the Deutsche Forschungsgemeinschaft (SFB 596 “Molecular Mechanisms of Neurodegeneration”), the German Ministry for Education and Research (NGFN plus “Functional Genomics of Parkinson’s Disease”), the Helmholtz Alliance “Mental Health in an Ageing Society”, and the German Center for Neurodegenerative Diseases (DZNE).

Copyright information

© Springer-Verlag 2011