Electron Transport Disturbances and Neurodegeneration: From Albert Szent-Györgyi's Concept (Szeged) till Novel Approaches to Boost Mitochondrial Bioenergetics

Among genes associated with an autosomal recessive inheritance of familial PD, parkin, a ubiquitin E3 ligase, is responsible for the polyubiquitin tagging of toxic protein aggregates for proteasomal degradation [80]. In addition, parkin appears to be involved in antioxidant functions through regulating SOD activity and glutathione levels [81] and may play important roles in mitochondrial transcription via its association with mitochondrial transcription factor A (Tfam) [82]. Accordingly, parkin-deficient mice display decreased expression of complex I and IV subunits accompanied by a diminished antioxidant capacity and enhanced oxidative damage [83], whereas parkin-deficient flies develop abnormal mitochondria and exhibit an increased vulnerability to paraquat [84]. Among in vitro conditions, overexpression of wild-type parkin reduced, whereas that of the mutant allele aggravated cell death induced by different oxidative stimuli including H2O2 and MPP+ intoxication, which aggravation was accompanied by increased levels of oxidative stress markers and a decreased amount of glutathione [85]. In line with these, transfection of cell lines with mutant parkin leads to an increased presence of markers of oxidative and nitrative injuries [86]. Notably, however, the potential of wild-type parkin overexpression to protect against oxidative insults in vitro has been challenged [87], and another study found no increase in vulnerability to different regimens of MPTP exposure in parkin-deficient mice [88].

Parkin appears to share common pathway with phosphatase and tensin homologue- (PTEN-) induced putative kinase 1 (PINK1), another protein associated with autosomal recessive PD, with PINK1 acting upstream of parkin [89]. Indeed, both proteins are involved in the regulation mitochondrial dynamics [28, 29, 90, 91], and phenotypes associated with PINK1-deficiency have been repeatedly reported to be rescued by parkin [89, 92–94]. In experimental models, PINK1-deficiency has been associated with impaired mitochondrial respiration (most consistently complex I deficiency) [95, 96], decreased energy production [92, 93, 97, 98], elevated ROS production [98, 99], impaired mitochondrial calcium handling [98–100], mitochondrial morphological alterations [28, 29, 89–91, 93, 98], and an increased susceptibility to mPTP [99–101]. Furthermore, PINK1-deficiency exacerbates neurodegeneration evoked by MPP+ in vitro and MPTP in vivo [102]. In addition to parkin, tumor necrosis factor (TNF) receptor-associated protein 1 (TRAP1), a mitochondrial molecular chaperone also known as heat shock protein 75 (Hsp75), has been postulated to be another possible downstream target of PINK1, through which PINK1 activity can prevent the release of cytochrome c and a subsequent apoptosis [103]. The functional association between TRAP1 and PINK1 has gained further support by more recent studies [104, 105], consistently suggesting that TRAP1 acts downstream of PINK1 and in parallel with parkin when mediating amelioration in mitochondrial dysfunction.

Mutations of DJ-1, an oxidative stress sensor capable of modulating glutathione metabolism and mitochondrial transcription under mitochondrial stress [106], leads to autosomal recessive familial PD. The protein is suggested to function in parallel with PINK1/parkin pathway in maintaining mitochondrial function among oxidative conditions [107]. At the experimental level, DJ-1-deficiency has been associated with increased ROS production [108–110], impaired mitochondrial respiration [109] or energy production [110], mitochondrial morphological abnormalities [109], increased opening of the mPTP [110], as well as an increased sensitivity to oxidative stressors including MPTP [111], paraquat [112], and H2O2 [112].

Apart from genes identified in monogenic familial PD, a number of genes have been associated with the development of sporadic PD as modifying or susceptibility factors, including genes involved in mitochondrial functions such as mtDNA polymerase gamma 1 (POLG1) [113, 114] and complex I subunit ND5 [115]. In addition to these, an increasing body of evidence suggests that PGC-1α may add important contributions to the pathogenesis of PD. Indeed, a comprehensive genome-wide meta-analysis found a set of 425 PGC-1α-responsive nuclear-encoded mitochondrial genes underexpressed in sporadic PD, representing pinpoint defects in glucose metabolism and mitochondrial ETC [116]. Furthermore, associations of single nucleotide polymorphisms (SNPs) of PGC-1α have been reported with the risk of PD, the age of onset and the longevity [117]. These appear to be in correspondence with decreased expression of PGC-1α and its target gene NRF-1 in the SN and striatum of PD patients as well as in the midbrain of conditional parkin knockout mice [118]. In line with decreased ATP production and impairments in mitochondrial OXPHOS [119] and antioxidant responses [6], PGC-1α-deficient mice display enhanced susceptibility to MPTP toxicity [6]. Corresponding to observations that mitochondrial dysfunction can promote the aggregation of α-synuclein [120], reduced expression of PGC-1α in vitro lead to enhanced α-synuclein oligomerization [121]. This effect was, however, not confirmed in PGC-1α-deficient mice, suggesting a more complex scenario for mitochondrial dysfunction-induced α-synuclein aggregation in vivo [122]. Supporting a potential therapeutic relevance in PD, overexpression of PGC-1α demonstrated neuroprotection against α-synuclein- and rotenone-induced toxicity in vitro [116] and in a parkin interacting substrate (PARIS) overexpression model of PD in vivo [118].

In line with these, transgenic overexpression and resveratrol-induced activation of PGC-1α (via deacetylation by Sirt-1) both rendered neuroprotection against MPTP toxicity in mice [123]. Similarly, the administration of pioglitazone, an agonist of PPARγ that enhances the activity and expression of PGC-1α [124], was also protective in MPTP studies [125, 126]. These altogether suggest that a deficient expression and/or function of PGC-1α and its target genes may play important roles in the development of sporadic PD, which may be of therapeutic relevance in the future. Notably, however, contrasting results have also been published reporting that adenoviral overexpression of PGC-1α aggravated MPTP-mediated damage in mice [127] and was ineffective against mutant α-synuclein-mediated toxicity in rats [128] and that a sustained overexpression of PGC-1α to high levels per se evoked the degeneration of nigral neurons in rats [128]. These findings necessitate further investigations and draw the attention to the possibility that a sustained overactivation of mitochondrial biogenesis aiming to restore mitochondrial functions may also have deleterious consequences on the long run. This issue needs to be clarified in the future.

The search for effective neuroprotective compounds capable of modifying the disease course in PD is still extensive; as molecules targeting mitochondrial dysfunction in PD though provided promising results in experimental models of PD [129–131], they were ineffective in clinical trials [132]. In line with the data on a decreased function of PGC-1α in PD and on the therapeutic potential of its activation, there is a hope that transcriptional activation of mitochondrial biogenesis and antioxidant responses via PGC-1α activation may hold therapeutic value. A phase II safety and futility clinical trial with the PPARγ agonist pioglitazone on patients with early PD has recently been completed, and the results are about to be published in the near future (NCT01280123).

Huntington's disease (HD) is a monogenic, progressive neurodegenerative disease of autosomal dominant inheritance. The genetic alteration is the expansion of CAG trinucleotide repeat sequence on the interesting transcript 15 (IT15) gene on chromosome 4 encoding huntingtin, with increasing number of repeat associating with earlier onset and more rapid progression [133, 134]. The disease onset is usually between 40 and 50 years of age, presenting with behavioral alterations and hyperkinesia in the early stages, subsequently associating with pyramidal symptoms, dystonia, dementia, and psychosis. The pathognomonic alteration is the preferential loss of the striatal γ-aminobutyric acid (GABA)-ergic medium-sized spiny projection neurons (MSNs) and the presence of intracytoplasmic and intranuclear protein inclusions of mutant huntingtin widely distributed in neuronal as well as extraneuronal tissues. The characteristic decreased activity of respiratory complex II, especially in the striatum, has early linked HD to mitochondrial dysfunction [135] (Figure 5). Since then, deficiency in complex II is the most consistent and predominant alteration reported in HD; however, the involvement of other respiratory complexes has also been suggested [136, 137]. The concept of mitochondrial dysfunction mediating the pathological process induced by mutant huntingtin is consistent with an increased presence of oxidative stress [137–141], which is well-reflected by the increased amount of mtDNA mutations observed in HD patients [142].

Further alterations supporting the primary role of mitochondrial dysfunction in mediating the effects of mutant huntingtin include a transcriptional [143] and/or functional [144] repression of PGC-1α, disturbances in mitochondrial trafficking [145], a gradually decreasing mitochondrial number [146], and an impairment of mitochondrial calcium handling [147] with an enhanced sensitivity to calcium-induced opening of mPTP and cytochrome c-mediated cell death [148, 149].

In line with the predominant biochemical alteration, irreversible inhibition of complex II by 3-nitropropionic acid effectively recapitulates most of the clinical and histopathological characteristics of HD, including the preferential neuronal loss of GABAergic MSNs within the striatum [150, 151]. Similar alterations can be evoked via reversible complex II blockade by malonate [152]. The relevance of complex II dysfunction in HD is highlighted by the fact that mutant huntingtin leads to decreased expression (at the protein level) of the 30 kDa iron-sulfur (Ip) subunit and the 70 kDa FAD (Fp) subunit of complex II in the striatum of HD patients [153] and in vitro lentiviral models [153, 154]. Rather similar pattern of alterations was found to underlie complex II deficiency in transgenic HD mice and in a lentiviral model of HD in rats [155]. Complex II dysfunction appears to be causative in HD as its overexpression demonstrated marked restorative effect in these models [153–155]. Alterations in complex II assembly were accompanied by reduced mitochondrial biogenesis in transgenic animals, prompting the authors to suggest the possible contribution of PGC-1α repression [155]. Indeed the expression of PGC-1α has been found to be downregulated in the striatum of HD patients [143, 144, 146, 156], in transgenic HD animals [24, 143, 144, 157–159], and in in vitro HD models [143, 156, 159, 160]. Correspondingly, the decreased expression of several PGC-1α target genes has been identified in the striatum of HD patients [144, 146] and transgenic HD mice [144, 157, 158].

A possible mechanism through which mutant huntingtin can lead to the downregulation of PGC-1α can be secondary to its effect to enhance the expression [161] and activity of NR2B subunit-containing NMDARs [162], features characteristic of transgenic HD mice [163], which in turn results in a decreased striatal CREB signaling [163], and a subsequent downregulation of PGC-1α [164]. In addition, a decreased expression of transducer of regulated CREB-binding protein 1 (TORC1), an activator of CREB-mediated PGC-1α expression, has been found in post mortem HD striatum, in transgenic HD mice, and in an in vitro HD model, which may contribute to the downregulation of PGC-1α expression in HD [160], whereas others suggest that PGC-1α repression may be secondary to the downregulation of PPARγ in HD [159]. Supplementing the alterations in PGC-1α expression in HD, Johri et al. reported decreased protein level of the functionally active N-truncated splice variant of PGC-1α (NT-PGC-1α) in the striatum of low-grade HD patients and young asymptomatic transgenic N171-82Q and R6/2 HD mice, whereas its level was found to be elevated in high-grade HD patients and in older symptomatic transgenic HD mice [156]. This pattern has been recently supported by our study on transgenic N171-82Q HD mice at the mRNA level, revealing a significantly upregulated expression of NT-PGC-1α in both the striatum and overlying cortex of older symptomatic HD mice compared to wild-type and young HD counterparts [24]. These changes were accompanied by a decreased expression of full-length PGC-1α in the striatum and cortex of young transgenic mice, corresponding to prior observations.

A main novelty of this study included a previously unidentified consistent elevation of both the full-length and the N-truncated isoforms of PGC-1α in the cerebellum of transgenic HD mice, which may underlie the relative resistance of cerebellar neurons to degeneration in HD [24]. This study further provided evidence for a consistent striatal upregulation of both the full-length and the N-truncated isoforms of PGC-1α following acute but not chronic injury due to 3-nitropropionic acid intoxication in mice [24]. This possibly compensatory elevation corresponds to prior findings of PGC-1α upregulation in response to ROS challenge in vitro and highlights the role of PGC-1α in concerting antioxidant responses [6, 23]. Supporting a potential therapeutic relevance, a line of evidence suggests that mechanisms associated with the upregulation of PGC-1α can exert neuroprotection in experimental models of HD. Indeed, the administration of PPARγ agonist thiazolidinediones (such as rosiglitazone and pioglitazone) was proven to be protective in transgenic HD mice [158, 159, 165], in an intrastriatal quinolinic acid-induced rat toxin model of HD [166], in 3-nitropropropionic acid-induced murine toxin model of HD [167], and in in vitro HD models [159, 165, 168]. Similarly, the pan-PPAR agonist bezafibrate exerted protection in transgenic HD mice [169]. Furthermore, TORC1 activation displayed protective and restorative effects on viability and mitochondrial functions in a striatal HD cell line model exposed to 3-nitropropionic acid [160]. The protective effect of resveratrol, a polyphenol with potent Sirt-1/PGC-1α-activating properties, has also been demonstrated in transgenic murine and nematode models [170], in a 3-nitropropion acid-induced murine model [171], and in an in vitro model of HD [170].

Notably, however, the potency of resveratrol to significantly elevate the expression of PGC-1α and its target genes within the striatum has recently been questioned [172], which necessitates further investigations. Considering that early attempts with mitochondria-targeted molecules being neuroprotective in experimental HD models [131, 173, 174] provided little success at the clinical level [132], the transcriptional activation of mitochondrial respiration and biogenesis may hold novel therapeutic potential as a more system-based approach. A phase III clinical trial with resveratrol is just about to recruit its participants (NCT02336633).

The corresponding set of evidence implicating a potential pathogenetic role of PGC-1α repression in mitochondrial dysfunction in HD as well as early observations of striatal alterations in PGC-1α-deficient mice [175, 176] suggested that such animals may serve as experimental models for HD; however, findings of a recent detailed neuropathological evaluation of mice lacking the expression of full-length PGC-1α has indicated that it might not indeed be the case and turned our attention to another group of diseases where systemic mitochondrial dysfunction is pathognomonic [122].

Mitochondrial diseases are a group of multisystemic disorders where the characteristic pathologies affecting organs with high energy demand (i.e., brain, liver, heart, skeletal muscle, and kidney) are due to mitochondrial dysfunction as a consequence of a genetic alteration either in the mtDNA or in the nDNA. The deleterious loss of functions may affect several components of proper mitochondrial functioning, including genes encoding respiratory complex subunits, proteins responsible for mtDNA transcription/translation, mitochondrial tRNAs and rRNAs, and nuclear-encoded ancillary proteins of mitochondrial function [177]. The diseases are distributed to characteristic syndromes based on the clinical manifestation and the observed neuropathological alterations, including Kearns-Sayre syndrome, Leigh syndrome, mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS), myoclonic epilepsy with ragged-red fibres (MERRF), neuropathy, ataxia, retinitis pigmentosa (NARP), and mitochondrial neurogastrointestinal encephalopathy (MNGIE) [178–181]. In these diseases, impaired ATP production with various defects in respiratory complexes and excess ROS production in the affected tissues has widely been documented and has excessively been reviewed [182]. Though in somewhat different patterns, mitochondrial encephalopathies are collectively characterized by various degrees of tissue vacuolation in the white and gray matter of the CNS, accompanied by region-selective reactive astrocytosis with or without neurodegeneration.

A number of genetically modified murine strains have been developed to model diseases with mitochondrial defects, however, with variable outcomes [122]. On the one hand, many of the genetic modifications lead to embryonic or early postnatal mortality due to multisystemic insufficiency (e.g., complete knockouts of CREB [183], Tfam [184], NRF-1 [185], NRF-2 [186], ERRγ [187], POLG1 [188], synthesis of cytochrome c oxidase 2 (SCO2) [189], and optic atrophy 1 (OPA1) [190]). On the other hand, a remarkable proportion of viable models, surprisingly, does not have any pathological changes in the CNS (e.g., complete knockouts of adenine nucleotide translocator 1 (ANT1) [191], PPARγ [192], ERRα [193], and SURF1 [194]; ΔmtDNA Mito-Mice [195]; and Twinkle mutant “Deletor” mice [196]). Genetic models exhibiting a neuropathology closely reminiscent of human mitochondrial leukoencephalopathies include mice deficient in Mn-SOD [197], in thymidine phosphorylase and uridine phosphorylase (TP/UP) [198], and in NADH dehydrogenase [ubiquinone] iron-sulfur protein 4 (NDUFS4) [199]. In addition to these, our recent neuropathological analysis on mice deficient in the expression of full-length PGC-1α revealed widespread spongy vacuolation predominating in the white matter of the striatum, thalamus, cerebellum, and the brainstem, accompanied by moderate to severe reactive astrogliosis in the pontomedullary brainstem and the cerebellar nuclei, corresponding to a pattern of alterations characteristic of the spongiform leukoencephalopathy seen in Kearns-Sayre syndrome [122]. This is especially interesting in light of the facts that experimental animals used for modeling cardiomyopathy in Kearns-Sayre syndrome are the tissue-specific knockouts of Tfam [200, 201], a gene under the regulation PGC-1α, and the expression of which is severely downregulated in PGC-1α-deficient mice [119, 175].

Notably, no indirect or direct signs indicative striatal neuronal degeneration and/or loss were observed in our study [122], which corresponds to the independent observations of Lucas et al. [202], both publications drawing the conclusion that PGC-1α-deficiency per se is not sufficient to evoke HD-like pathology, contrasting to what had previously been suggested.

Considering the spectrum of roles of PGC-1α in regulating and promoting mitochondrial functions and the fact that a number of genes involved in disease-causing mutations and/or that involved in modeling mitochondrial disease have direct or indirect interactions with PGC-1α (e.g., ANT-1, POLG1, Tfam, NRF-1, NRF-2, PPARs, ERRs, Mn-SOD, and CREB), the rationale for PGC-1α induction to provide symptomatic benefit in these currently intractable groups of diseases can be accepted [203]. Indeed, transgenic overexpression or bezafibrate-induced expression of PGC-1α delayed the onset of symptoms in a cytochrome c oxidase-deficient murine model of mitochondrial myopathy [204]. Similarly, transgenic overexpression of PGC-1α ameliorated the phenotype and increased the activity of mitochondrial respiratory complexes in POLG1 mutant “Mutator” mice [205]. Furthermore, adenoviral overexpression of PGC-1α partially restored respiratory deficits in fibroblasts obtained from patients with mitochondrial disease of various origin (though to different efficacy) and in MELAS cybrids [206]. These altogether suggest a potential therapeutic relevance of boosting mitochondrial biogenesis via PGC-1α-mediated approaches in diseases with genetic mitochondrial disorder.

Since the revelation of the essential role electrons, originating from reducing equivalents that arise from cytosolic and/or mitochondrial metabolic processes, in cellular bioenergetics via their flow through metal-containing electron transport complexes in the mitochondrial inner membrane, an armada of evidence has accumulated linking the impaired function of this system to degenerative diseases of the CNS. While initial attempts to compensate for such alterations showed as much promise at the experimental level as deep disappointment they caused at the clinical level, novel strategies with more system-based approaches aiming to render protection via improving mitochondrial bioenergetics at a transcriptional level may open up new therapeutic perspectives and boost pharmacological research. With more and more evidence linking PGC-1α and its target genes to the pathogenesis of neurodegenerative diseases including PD, HD, and mitochondrial disorders, pharmacological manipulations to restore and/or activate PGC-1α may provide valuable tools in the therapy of these currently intractable diseases.