Bioenergetic Defects and Oxidative Stress
The vulnerability of neurons to mitochondrial dysfunction can be pinned down to their high metabolic activity and energetic requirement, especially to maintain ionic gradients across membranes and for neurotransmission . The electron transport chain (ETC) of mitochondria is composed of four multiprotein complexes, named complexes I–IV, which couple the transport of electrons to the creation of a proton gradient across the inner mitochondrial membrane. Dissipation of the gradient by movement of protons through complex V drives the synthesis of adenosine triphosphate (ATP). Inefficiency in ETC function is recognized as one of the major cellular generators of reactive oxygen species (ROS) .
The most commonly observed bioenergetic deficit in PD is an impaired activity of mitochondrial complex I (NADH:ubiquinone oxidoreductase), the main site of entry of electrons into the respiratory chain. Complex I deficiency is linked to increased ROS accumulation, depletion of ATP, and dopaminergic neuron death . Evidence obtained from postmortem samples revealed a 30–40% reduced activity of complex I in the SN of patients with PD , as well as decreased activity and impaired assembly of complex I in the frontal cortex . Outside the brain, inhibition of complex I has been observed in platelets and skeletal muscles of PD patients, although there is no consensus on such data [8, 9].
To specifically address the question of whether complex I deficiency is etiological in PD, transgenic mice were recently generated that lack expression of a subunit of complex I (NADH:ubiquinone oxidoreductase Fe–S protein, Ndufs4) in midbrain dopaminergic neurons. Surprisingly, no significant neuronal loss was observed, and animals had no overt symptoms of parkinsonism . However, the Ndufs4-knockout neurons were more vulnerable to complex I inhibition, and dopamine release was reduced from striatal axon terminals, implying that disruption of striatal dopamine homeostasis may be directly correlated with mitochondrial dysfunction .
Toxin-based Models of Mitochondrial Dysfunction in PD
The involvement of mitochondrial complex I, and of ETC dysfunction in general, in the pathogenesis of PD is strongly suggested by the fact that toxin-induced disruption of the ETC triggers degeneration of dopaminergic neurons and causes a PD-like phenotype in flies, rodents, and humans. Such toxins typically inhibit activity of the ETC while increasing ROS and mitochondrial permeability transition (mPT), thus impairing ATP synthesis and inducing a bioenergetic crisis.
Evidence implicating mitochondrial dysfunction in the pathogenesis of PD historically emerged with the discovery that humans accidentally exposed to the recreational drug 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) developed an acute form of parkinsonism. It was subsequently found that administration of MPTP to mice, rats, and primates also reproduced the major clinical and neuropathological hallmarks of PD [11-13]. MPTP rapidly crosses the blood–brain barrier into the brain, where it is metabolized into its toxic metabolite 1-methyl-4-phenylpyridinium ion (MPP+). MPP+ is subsequently taken up by dopamine transporters into SN neurons and actively concentrated in mitochondria ; mice lacking the dopamine transporter are resistant to MPTP toxicity . At the mitochondrial level, MPP+ blocks the transfer of electrons through complex I and inhibits the Krebs cycle enzyme α-ketoglutarate dehydrogenase, thereby eliciting a decrease in mitochondrial ATP production . A more crucial effect might be the increased free radical generation which leads to significant oxidative stress and activation of pro-apoptotic pathways. The neurotoxin MPTP has been used extensively for reproducing the motor symptoms of PD in animal models and for unraveling the mechanisms and consequences of nigrostriatal neuron loss in vivo .
A complex I inhibitor closely related in chemical structure to MPP+ is the ammonium herbicide, paraquat. Paraquat (1,1′-dimethyl-4,4′-bipyridinium) is an even more powerful inducer of ROS generation than MPTP, although the binding affinity to complex I is lower . In humans, exposure to paraquat has been associated with a higher incidence of PD , while administration to rodents reproduces a selective degeneration of dopaminergic neurons, increased oxidative stress, and aggregation of the synaptic protein α-synuclein . Of note, induction of ROS and consequent lipid peroxidation by paraquat in mouse striatum involves nitric oxide and the increased expression of nitric oxide synthase .
The pesticide rotenone, another well-known inhibitor of mitochondrial complex I, impairs ATP synthesis and stimulates increased ROS formation by mitochondria . Epidemiological studies have reported a significantly increased risk of developing PD for humans exposed to rotenone . Indeed, rotenone rat models of PD accurately reproduce the cardinal features of PD, including selective dopaminergic neuron degeneration, loss of the nigrostriatal pathway [24, 25] and motor symptoms such as ataxia, bradykinesia, and trembling in the forelimbs [26, 27]. These rats also display Lewy body-like cytoplasmic inclusions . Collectively, these features have made the rotenone rat model one of the most widely utilized to study neuroprotective modalities in PD. More recently, other mitochondrial ETC inhibitors have been identified that induce parkinsonism in humans or animal models. These include trichloroethylene, tetrahydro-beta-carbolines, pyridaben, tebufenpyrad, fenazaquin, and fenpyroximate. They all result in a similar loss of nigrostriatal dopaminergic neurons and a PD-like motor phenotype [13, 29-31].
Genetic-based Models of Mitochondrial Dysfunction in PD
There is a striking convergence of the biochemical abnormalities from several inherited genetic defects with those found in sporadic cases. This denotes that there may be common pathways to dopaminergic cell loss and dysfunction, and thus common target areas for pharmacological treatment which may slow disease progression.
Alpha-synuclein (α-synuclein) is a 140-amino acid protein found enriched in presynaptic terminals and associated with synaptic vesicle membranes . α-Synuclein is encoded by the SCNA gene, and missense mutations in SCNA were the first autosomal-dominant PD-associated mutations to be identified . The three most common mutations segregating with familial PD are Ala53Thr, Ala30Pro, and Glu46Lys [33, 34]. Patients harboring mutations in SCNA cannot be clinically distinguished from sporadic PD, although onset is typically earlier and the disease more aggressive. Neuropathological findings are also similar, with nigrostriatal degeneration and the defining deposition of amorphous Lewy bodies and Lewy neurites in the brain. Genetic studies prompted the subsequent finding of fibrillar α-synuclein as the main structural component of Lewy inclusions . The α-synuclein protein displays an increased propensity for aggregation due to the presence of a central hydrophobic domain . Aggregation propensity is augmented by increased protein concentration; thus, a higher dosage of wild-type SCNA, as in gene duplication or triplication, is associated with a more aggressive form of familial PD [37, 38]. All this brought the α-synuclein protein to the fore in the investigation of PD pathogenesis.
Accumulated evidence from both in vitro and in vivo studies postulates a major pathogenic role for α-synuclein in mitochondrial dysfunction, thereby providing a link between protein aggregation, mitochondrial damage, and neurodegeneration. Transgenic mice expressing human A53T α-synuclein developed mitochondrial DNA (mtDNA) damage and degeneration , with mitochondrial degeneration preceding the onset of motor symptoms . A53T was found to directly localize to mitochondrial membranes and selectively inhibit mitochondrial complex I, thus suggesting a common mechanism with ETC inhibitors . In primary cortical neurons that overexpress mutant A53T α-synuclein, there is massive mitochondrial destruction and loss that is associated with a bioenergetic deficit and neuronal degeneration . Similarly, α-synuclein-overexpressing transgenic mice and neuronal cells exhibit mitochondrial dysfunction, increased mitochondrial DNA damage, and impaired ETC activity [43, 44]. Interestingly, α-synuclein-knockout (SNCA−/−) mice also showed reductions in ETC activity, although not in complex I. These alterations in mitochondrial function were associated with abnormalities in mitochondrial membrane properties, in particular a reduction in content of the phospholipid cardiolipin . Furthermore, SNCA−/− mice have been shown to be less sensitive to administration of mitochondrial toxins, such as MPTP, 3-nitropropionic acid, and malonate . As mitochondrial toxins also promote α-synuclein aggregation, these findings suggest a functional link between mitochondrial dysfunction and accumulation of the misfolded α-synuclein protein . Interestingly, also in the eukaryotic yeast Saccharomyces cerevisiae, there was a requirement for functional mitochondria in mediating α-synuclein-induced cell death .
A plausible mechanism for explaining mitochondrial dysfunction is the direct interaction of α-synuclein with mitochondrial components, especially mitochondrial membranes. Various in vitro investigations have correlated localization of α-synuclein aggregates to mitochondria with increased mitochondrial membrane damage and dysfunction [49-51]. Intriguingly, a cryptic mitochondrial-targeting signal has been identified in the N-terminal amino acid sequence of α-synuclein, effectively targeting the protein to the inner mitochondrial membrane, whereupon it inhibits complex I and increases ROS formation [49, 52]. Another mechanism contributing to mitochondrial degeneration might involve the fragmentation of mitochondria upon direct association of α-synuclein to mitochondrial membranes [53, 54].
Point mutations in the leucine-rich repeat kinase 2 (LRRK2) gene have been found in around 7% of familial PD cases with late-onset autosomal-dominant parkinsonism; mutations have also been reported in a few sporadic cases [55, 56]. Patients with mutations in LRRK2 have a clinical phenotype that closely resembles the sporadic form, and neuropathological findings typically include intracellular Lewy bodies and nigrostriatal neuronal loss . LRRK codes for a large protein of 2527 amino acids, with all of the identified mutations occurring in its multiple functional domains . Most pathogenic mutations, including the most frequent Gly2019Ser point mutation, increase the kinase activity of LRRK2 and mediate neuronal toxicity, although its substrates have yet to be determined . The main site of action of LRRK is again the mitochondrion. Transient expression of mutant LRRK2 in neuronal cells activates mitochondrial-mediated apoptotic mechanisms of cell death . Caenorhabditis elegans nematodes expressing mutant LRRK2 showed selective vulnerability of dopaminergic neurons to the mitochondrial inhibitor rotenone . Although predominantly cytosolic, LRRK2 can associate directly with the outer mitochondrial membrane, as demonstrated using synaptosomal cytosolic fractions from mammalian brain . LRRK2 may also induce mitochondrial dysfunction by disturbing mitochondrial dynamics, particularly as it controls the expression of the mitochondrial fission factor dynamin-like protein 1 (DLP1) .
Genetic mutations of the Parkin gene contribute to around 50% of familial PD cases having an autosomal-recessive mode of inheritance and disease onset before the age of 45 years . Clinical manifestations closely resemble those of idiopathic PD, accompanied by loss of nigral neurons in the brain, although Lewy body deposition is typically absent . Numerous in vitro and in vivo studies suggest that parkin function is intimately related to mitochondrial protection, especially from oxidative damage. Thus, parkin inhibited swelling of mitochondria and associated cytochrome c release in ceramide-treated cells and in cell-free assays [66, 67]. Localization of parkin at the outer mitochondrial membrane has been observed in parkin-overexpressing cells and adult mouse brain [66, 68]. Although the precise mechanism is not yet established, the ability of parkin to protect mitochondria from insult by chemical toxins has been demonstrated in a wide range of cellular and animal models [69, 70]. Transgenic mice knocked out for parkin manifest impaired oxidative phosphorylation in striatal mitochondria, which are swollen and have fragmented cristae . Drosophila models representing loss of parkin function have a shortened life span while showing degeneration of dopaminergic neurons and severe mitochondrial pathology [72, 73]. Even mitochondria isolated from nonnervous tissues of parkin-mutant patients, such as leukocytes and fibroblasts, have various morphological abnormalities and functional deficits [74, 75].
Parkin may additionally have an inherent role in regulating mitochondrial dynamics. Under normal physiological conditions, parkin is recruited selectively to impaired mitochondria, promotes mitophagy, and mediates mitochondrial elimination by catalyzing ubiquitination of targeted mitochondria . Disease-associated parkin mutations, on the other hand, induce defective recognition and ubiquitination of dysfunctional mitochondria .
Another strong piece of evidence for the etiological role of mitochondria in PD comes from the discovery of families with early-onset parkinsonism caused by mutations in PINK1 . PINK1 is a putative mitochondrial serine/threonine kinase that plays an important neuroprotective role. In cell culture studies, wild-type, but not mutant, PINK1 protects against apoptotic cell death induced by MPTP, oxidative stress, and activation of the mPT pore [79-81]. PINK1 import into mitochondria occurs via an N-terminal mitochondrial-targeting sequence. Most investigations have shown that PINK1 attaches to the inner mitochondrial membrane, while others propose that PINK1 localizes to the intermembrane space [82, 83]. Conversely, PINK1 suppression or knockdown in human dopaminergic neurons increases the cellular susceptibility to apoptosis via the mitochondrial pathway, with increased oxidative stress markers, defective mitochondrial respiration, and abnormal mitochondrial morphology [84-87]. SN dopaminergic neurons prepared from PINK1 null mice exhibited fragmented mitochondria, significant ROS generation, and a depolarized mitochondrial membrane potential . Intriguingly, a functional link between parkin and PINK1 has also been found. First, the pathological phenotype of parkin-mutant flies closely resembles those of PINK1 mutant flies . Second, and more significantly, the effects of loss of PINK1 function in flies, including ATP depletion, shortened life span, and degeneration of dopaminergic neurons, could be rescued by means of transgenic expression of parkin [89, 90].
Mutations in the DJ-1 gene have been linked with rare cases of early-onset autosomal-recessive PD and are responsible for 1–2% of early-onset forms of the disease . The overarching functional role of DJ-1 appears to be as an antioxidative stress sensor, particularly against oxidative stress in association with mitochondrial dysfunction . Accordingly, downregulation of DJ-1 in neuronal cell models causes increased vulnerability to cell death by oxidative stress and mitochondrial toxins . DJ-1 loss-of-function cellular models have a defect in the assembly of complex I and hence in mitochondrial respiration and morphology [93, 94]. Likewise, DJ-1 knockout mice suffered increased dopaminergic neuronal degeneration upon challenge by MPTP and paraquat [95, 96], while DJ-1-deficient Drosophila flies were extremely sensitive to paraquat and rotenone . Under basal conditions, DJ-1 is mostly a cytosolic protein, with a limited endogenous pool in the mitochondrial matrix and intermembrane space . Oxidant exposure, including exposure to the complex I inhibitor rotenone, triggers relocalization of DJ-1 to mitochondria, correlating with neuroprotection [99, 100]. Overexpression of DJ-1 resulted in an increased capacity to withstand these insults and reduced intracellular ROS . It has been further demonstrated that DJ-1 function can be linked to other proteins implicated in PD pathogenesis, namely α-synuclein, PINK1, and parkin. In cellular models, wild-type DJ-1 inhibited the aggregation of wild-type α-synuclein, in part by demonstrating chaperoning activity [102, 103]. DJ-1 also works in parallel to the PINK1/parkin pathway to maintain mitochondrial function and regulate mitophagy .
Genetic Defects in Mitochondrial DNA
Various somatic mtDNA deletions or rearrangements have been reported in patients with PD [105, 106]. Cytoplasmic hybrid (cybrid) models, in which platelet mtDNA from PD patients is expressed in neural cell lines, have provided particularly convincing evidence . For example, defects in complex I activity could be transferred from patients with sporadic PD to mitochondrial-deficient cybrid cell lines, which exhibited a decrease in mitochondrial membrane potential, impaired respiratory capacity, and abnormal Ca2+ sequestration by mitochondria [108, 109]. Significantly, PD cybrid cells were rescued by delivery of normal mtDNA to the defective mitochondria . Conditional knockout mice have been generated with reduced or defective mtDNA expression. These mice develop a progressive motor phenotype associated with loss of midbrain dopamine neurons [111, 112].