The Broad Impact of TOM40 on Neurodegenerative Diseases in Aging

Abstract

Mitochondrial dysfunction is an important factor in the pathogenesis of age-related diseases, including neurodegenerative diseases like Alzheimer’s and Parkinson’s spectrum disorders. A polymorphism in Translocase of the Outer Mitochondrial Membrane – 40 kD (TOMM40) is associated with risk and age-of onset of late-onset AD, and is the only nuclear- encoded gene identified in genetic studies to date that presumably contributes to LOAD-related mitochondria dysfunction. In this review, we describe the TOM40-mediated mitochondrial protein import mechanism, and discuss the evidence linking TOM40 with Alzheimer’s (AD) and Parkinson’s (PD) diseases. All but 36 of the >~1,500 mitochondrial proteins are encoded by the nucleus and are synthesized on cytoplasmic ribosomes, and most of these are imported into mitochondria through the TOM complex, of which TOM40 is the central pore, mediating communication between the cytoplasm and the mitochondrial interior. APP enters and obstructs the TOM40 pore, inhibiting import of OXPHOS-related proteins and disrupting the mitochondrial redox balance. Other pathogenic proteins, such as Aβ and alpha-synuclein, readily pass through the pore and cause toxic effects by directly inhibiting mitochondrial enzymes. Healthy mitochondria normally import and degrade the PD-related protein Pink1, but Pink1 exits mitochondria if the membrane potential collapses and initiates Parkin-mediated mitophagy. Under normal circumstances, this process helps clear dysfunctional mitochondria and contributes to cellular health, but PINK1 mutations associated with PD exit mitochondria with intact membrane potentials, disrupting mitochondrial dynamics, leading to pathology. Thus, TOM40 plays a central role in the mitochondrial dysfunction that underlies age-related neurodegenerative diseases. Learning about the factors that control TOM40 levels and activity, and how TOM40, specifically, and the TOM complex, generally, interacts with potentially pathogenic proteins, will provide deeper insights to AD and PD pathogenesis, and possibly new targets for preventative and/or therapeutic treatments.

Introduction

The Significance of the Mitochondria in Health and Disease

Mitochondria are involved in almost every aspect of cellular metabolism and are, arguably, the most complex organelles in eukaryotic cells, genetically, functionally and structurally. It is no surprise that mitochondrial dysfunction and disturbances in mitochondrial dynamics contribute to the etiologies of multiple age-related diseases [1,2] including neurodegenerative diseases such as Parkinson’s disease (PD), Amyotrophic lateral sclerosis (ALS), Huntington’s disease (HD) and late-onset Alzheimer’s disease (LOAD) [3-6]. The iconic mitochondrial two membrane arrangement encloses two aqueous spaces, the matrix, enclosed by the inner membrane, and the intermembrane space [7]. Each of these compositionally unique domains hosts one or more cellular process, that include metabolism and ATP synthesis [8], ion homeostasis [8], small-molecule biosynthesis [9,10], cellular redox status maintenance [11], apoptosis [12], innate immunity [13-16], and viral [17] and bacterial [18] infections. Distinctively, of all cellular organelles, mitochondria are the only ones that use protein products encoded by two genomes. The mitochondrial genome, (mt) DNA, localized within the matrix, encodes 37 genes, including 13 essential subunits of the oxidative phosphorylation system plus the rRNA and tRNA molecules needed for their expression [19,20]. The remaining 99% of the >1,500 mitochondrial proteins [21-23] are synthesized as preproteins on cytoplasmic ribosomes from mRNAs encoded by nuclear genes, and are assembled post-translationally after passing through TOM40, the central pore of the mitochondrial protein import apparatus. This review focuses on the role of mitochondrial biogenesis generally, and specifically on the role of TOM40 in late onset neurodegenerative diseases, including dementias of the AD type and Lewy Body (LB) disorders, primarily PD.

The Central Players in the Mitochondria Function: TOM40 and TOM Complex

Mitochondrial biogenesis occurs by the coordinated actions of a series of complex cellular machines (reviewed in [24-26]). The TOM (Translocase of the Outer Membrane) complex is the main entry portal for most cytoplasmically synthesized mitochondrial proteins and TOM40 is the key subunit of the complex, the pore through which the vast majority of imported proteins must pass. The complete TOMcomplex contains seven subunits: TOM40, TOM22, two proprotein receptors, TOM20, and TOM70, and three smaller proteins, TOM5, TOM6 and TOM7 (Figure 1A). In general, imported proteins bind to one of the receptors and then, with the assistance of TOM22 and TOM5, are passed into the TOM40 channel. TOM40 possesses chaperone-like properties [27], which helps expedite movement of the preproteins through the channel, and it facilitates interactions of proteins transiting through the outer membrane with the appropriate complexes that assemble them into the mitochondrion [28]. TOM6 is involved in assembly and maintenance of the TOM complex, and TOM7 is involved in its disassembly [24,25]. In addition to its role in protein translocation, TOM22 is also key to organizing the TOM complex and together with TOM7 controls the unit’s dynamics [29-31]. TOM connects with distinct complexes to sort incoming proteins to their appropriate destinations. The SAM complex inserts proteins into the outer membrane. The MIA/Erv1 disulfide relay imports cysteine-containing proteins into the intermembrane space. TOM cooperates with the TIM (Translocase of the Inner Membrane) 23 complexes to import matrix-targeted proteins, and TOM, the TIM23 complex and other small TIM proteins that reside in the intermembrane space direct the translocation of oxidative phosphorylation (OXPHOS) and metabolite transporter proteins to the inner membrane. Protein sorting is accomplished by recognition of targeting signals within the sequences of the imported proteins by the appropriate complexes.

Figure 1.

a: Schematic representation of normal mitochondrial protein uptake. Proteins with cleavable terminal targeting signals (positively charged, red) are targeted to the TIM23 complex for release into the inner membrane or into the matrix. Some matrix proteins travel from TIM23 to the Mia40/Erv1 complex for redox-mediated folding. Proteins with multiple membrane-spanning domains have internal targeting sequences (uncharged, red); these are targeted to the TIM22 complex, for lateral release into the inner membrane. Outer membrane β-barrel proteins, such as TOM40 itself, contain non-cleavable targeting domains and are transferred from TOM to intermembrane space chaperones via SAM (not shown) to TOB, required for insertion into the outer membrane. Outer membrane proteins with a-helical segments use multiple pathways for insertion into the outer membrane, some involving MIM1, but none involving TOM.

b: Model of the possible hijacking of mitochondrial protein uptake by pathogenic proteins. Pink1, α-synuclein and APP possess alternative mitochondrial targeting signals. Pink1, is targeted to the TIM23 complex where it is subject to proteolytic degradation by the Mitochondrial Proteolytic Peptidase and possibly other proteases. If the Δψ collapses, wild-type Pink1, escapes and elicits Park2 binding to mitochondria, leading eventually to autophagy of the damaged mitochondria. PD pathogenic mutant proteins escape proteolysis and exit mitochondria, even in the presence of an intact membrane potential, leading to loss of healthy mitochondria. α-synuclein targets inner membrane proteins, such as cytochrome c oxidase. Aβ peptides enter mitochondria through the TOM40 pore and target matrix and inner membrane proteins. APP enters and blocks the TOM40 channel and interferes with mitochondrial oxidative processes.

In addition to its well-known role as a mitochondrial protein import translocase, ectopic TOM40 was identified as a protein termed p38.5 (Haymaker) on the cell surface of NK(natural killer) cell – susceptible cell lines. It was not detected on NK-resistant cells [32]. The purified protein bound to NK cells but not to non-NK-lineage cells, and both the purified protein and antibodies raised against it blocked NK activity [32]. This research group subsequently showed the p38.5 protein was associated with intracellular membrane fractions, in addition to the plasma membrane, and found its sequence was identical to the human TOMM40 sequence [33]. Although we will not discuss ectopic TOM40 further in this review, we do note other “prototypical” mitochondrial proteins also occur ectopically, notably subunits of the mitochondrial ATP synthase [34-36], and it is possible such translocations represent a form of cellular stress response.

The Essential Role of TOM40 in the Organism Level

Considering the critical role TOM40 plays in mitochondrial biosynthesis, it is not surprising that TOM40 is essential for life in most eukaryotic organisms, as demonstrated by knock-out genetic studies using S. cerevisiae [37], N. crassa [38], and mice [39] and using shRNA in human cell culture [40]. Moreover, reducing TOM40 levels in C. elegans retarded growth [41], as did introducing assembly-defective TOM40 mutations in N. crassa [38].

The C. elegans RNAi screen showed that reduced TOMM40 levels arrested growth between the 1st and 3rd larval stages; TOMM40 knock-out animals exhibited a phenotype similar to the knock-downs [41]. In neither case were there detectable developmental defects or structural or functional defects in the worms’ feeding apparatus [41], indicating the animals ability to feed was intact and TOMM40 per se was required for continual growth. TOMM40 knock-down collapsed the mitochondrial membrane potential, blocked the uptake of mitochondrially targeted proteins and elicited the mitochondrial stress response, but markers of cytoplasmic and endoplasmic reticulum stress were not affected. Consistent with these data, reducing TOM40 levels in HEK cells using “moderate” amounts of RNAi also dissipated the mitochondrial membrane potential [42]. Unexpectedly, in C. elegans the TOMM40 knock down also suppressed DAF-28/insulin secretion, which represents the major metabolic insulin in C. elegans. The secretion of two related neuropeptides, ANF and INS-22 was unaffected, and therefore this was not a generalized defect but was specific for insulin. These results show that reducing TOM40 levels causes mitochondrial dysfunction, disrupting processes that critically depend on healthy mitochondria.

Knock-down mice have provided important additional insights to the role TOM40 plays in health and disease. TOMM40^-/- homozygous knockdowns died in embryonic stage E1. By contrast, TOMM40^+/- mice exhibited normal embryonic development and growth, but after two years of age they had 30% higher mortality rates than homozygous littermates [39]. Early in life, the hemizygotes developed heart conduction impairments and peripheral neuropathy. In addition, there were reduced numbers of dopaminergic neurons in aged (two year) hemizygotes compared with their age-matched normal littermates [39]. Histologically, mutant mitochondria in both young and old hemizygous mice appeared darker, with less space between cristae, than normal mitochondria from comparable aged animals. Although histologically the cristae appeared normal otherwise, oxygen consumption in both heart and brain mitochondria was severely affected in old but not young hemizygotes vs. their respective littermates. The amounts of OXPHOS complexes III and IV were also reduced in the two year-old hemizygous mice vs. their normal littermates [39].

Involvement of TOMM40 in Late Onset Neurodegenerative Diseases – Genetic Evidence

APOE ε4 is the strongest and most replicated genetic risk factor for LOAD. APOE is located on chromosome 19 (19q13.32) in the tight gene cluster TOMM40-APOE-APOC1-APOC4-APOC2 [43] that is in strong linkage disequilibrium (LD) block. Genome wide association studies (GWAs) reported that the strongest association signal (by a wide margin) was at the APOE LD region [44,45]. This top association signal was attributed to the APOE ε4 haplotype, however, it has been acknowledged that other genetic factor/s within this LD block may also explain the strong genome-wide significant signal and contribute, in part, to disease risk attributed to this genomic region. A number of groups discovered that TOMM40 SNPs are associated with LOAD or related endophenotypes [46-50,51-54]. Additional studies reported TOMM40 association with LOAD risk [55], cognitive performance [56] and hippocampal atrophy [57] independently of APOE. Valant found that fourteen SNPs within the TOMM40 locus were associated with LOAD risk, and it is noteworthy they also found SNPs within the TOMM20 and TOMM7 genes that were independently associated with AD. The results of association tests of SNPs in TOMM22 and TOMM70 with LOAD did not reach statistical significance (P = 0.082 and 0.11 for TOMM22 and TOMM70, respectively) but were suggestive [58]. These results further implicate the TOM complex in the etiology of LOAD.

Recently, a large GWA study implicated the APOE genomic region in the etiology of dementia with lewy body (DLB) [59], and it is not surprising that several highly significant associated SNPs were mapped within the TOMM40 gene. We replicated the association of TOMM40 SNPs with DLB pathology in a small size study group (Ncase/control= 47/132, p=0.01, unpublished data). Furthermore, we tested the associations of TOMM40 locus with LB pathology using postmortem-confirmed neuropathology of Lewy body variant of AD (LBvAD) cases (N=102) compared to AD control (N=384), and demonstrated a marginal association of TOMM40-SNP rs11668327 with increased risk for LB pathology in LOAD (p=0.04, unpublished data).

In order to pinpoint the biologically relevant gene and causal genetic variant/s that underlie the robust genetic associations of neurodegenerative diseases with this high LD region there is a need for advanced genomic strategies and follow-up functional and biological investigations using cell and molecular biology approaches. Herein we will outline fine mapping analysis that supports the role of TOMM40 in the etiology of LOAD.

Deep sequencing of this region followed by phylogenetic analysis identified a variable length deoxythymidine homopolymer, rs10524‘523’ (‘523’) located in intron 6 of the TOMM40 gene that is associated with risk for LOAD in Caucasian subjects. Three major allele groups were defined based on the distribution of ‘T’-lengths, S, L, VL, and specific allele lengths of this variable poly-T tract were linked with specific APOE alleles [60-62]. Our team originally reported that the ‘523’ VL allele of TOMM40 is associated with earlier age of LOAD onset relative to the ‘523’ S allele in APOE ε3/4 subjects, i.e. in the context of haplotypes where the ‘523’ L allele is linked to an APOE ε4 allele and the ‘523’ VL or S allele is linked to an APOE ε3 allele [63-65].

Since the original discovery of the association between ‘523’ and age of LOAD onset [66,67] a growing number of studies have reinforced our original observations. Johnson et al. discovered memory recall is impaired in APOE ε3/VL carriers dose-dependently on the VL allele, and also discovered a VL dose-dependent decrease in gray matter volume in the posterior cingulate and the medial ventral precuneus, two regions affected early in the course of AD [68]. The posterior cingulate is hypometabolic in mild cognitive impairment [69] and mild AD [70] and in presymptomatic APOE ε4 carriers [71,72], and early mitochondrial dysfunction has been identified in this region and has been hypothesized to contribute to the hypometabolism [73]. Caselli et al. confirmed the APOE ε3/3//VL/VL genotype was associated with decreased early memory performance. Before the age of 60, APOE ε3/3//VL/VL subjects failed to demonstrate a normal test/re-test result on the auditory verbal learning test, a finding that is also seen in early- LOAD. By contrast, the APOE ε3/3//S/S and the APOE ε4/4//L/L subjects gave normal test/re-test outcomes at those ages [74]. Beyond about 60 years of age, both the VL/VL and S/S groups gave similar results [74]. Hayden et al. [56] and Greenbaum et al. [75] used a similar strategy to study the effects of the S and VL alleles on age-related defects on cognitive performance in cognitively normal elderly populations. Hayden et al. found among APOE ε3/3 homozygotes, the S/S homozygotes performed better than S/VL heterozygotes on tests of verbal memory recall and executive function [56]. Greenbaum et al. found the S/S homozygotes performed significantly better than VL/VL homozygotes on tests of executive function and episodic memory. When Greenbaum et al. included S/VL heterozygotes in their analysis; they found the adjusted mean for executive function of the S/VL subpopulation was intermediate between the S/S and VL/VL groups. Together, these data suggest the ‘523’ genotypes affect cognitive performance in the elderly; in particular cognitive domains that are preferentially affected in early stage AD. Furthermore the genetic influence is mediated by the effects of the extreme alleles, and that the effects of the S allele are protective vs. those of the VL allele for many of the subphenotypes that contribute to changes in cognitive performance during presymptomatic LOAD. Bruno et al. also found the S allele was protective in studies that used levels of cortisol [76] or neurofilament light chain [77] as surrogate markers of damage associated with LOAD. Bekris et al. also demonstrated that the TOMM40 ‘523’ is associated with neuropathology of AD; in particular, the L allele dose-dependently was associated with neuritic tangles and a higher frequency of pathologically defined AD [78].

However other studies reported conflicting data. Jun et al. failed to replicate the association between TOMM40 and risk for LOAD [79]. Cruchaga et al. and Maruszak et al. also did not detect an APOE-independent effect of ‘523’ [80,81] on LOAD risk, however, both groups found an association between ‘523’ and LOAD risk in APOE ε3/3 homozygotes, and that the S allele was over-represented in this population while the VL allele was under-represented [80,81]. Several technical and methodological differences between their approaches and ours might explain their failure to identify APOE-independent effects of TOMM40 [82]. Our group has performed additional analysis on a cohort of Caucasian subjects with carefully ascertained ages of onset of cognitive impairment or probable AD [83]. We replicated the findings that the VL allele is associated with earlier ages of LOAD onset in APOE ε3/4//VL/L individuals compared with the ages of onset in APOE ε3/4//S/L individuals. We also found the proportion of the population with dementia who are S/S or VL/VL homozygotes changes rapidly [83]. Among younger members (65-74 years of age), the proportions of S/S and VL/VL homozygotes are roughly equal, but in older ages (75-84 years of age) the proportion of the total population who are S/S homozygotes is enriched by approximately 8% relative to the proportion who are VL/VL. As the results of Johnson et al., Hayden et al. and Casselli et al. suggest, the effects of the VL/VL genotype may be associated with presymptomatic events that are masked by later pathology. Collectively, these studies provided genetic evidences for the contribution of TOMM40 locus to LOAD and related endophenotypes. In the next sections we will describe follow-up functional studies using cell and molecular biology approaches that lend support to the significant role of TOM40 in LOAD and a broader spectrum of neurodegenerative disorders including PD.

Involvement of TOM40 and TOM in Alzheimer’s and Parkinson’s diseases – Biochemical and Cell Biological Evidence

Mitochondrial dysfunction is associated with both inheritable and sporadic forms of AD [3,73,84-87], due, in part, to internalized Aβ peptide and perhaps APP, mediated by the TOM40 pore. Mitochondria internalize APP and amyloid-beta peptide (Aβ) in vitro [88,89], in culture [88-90], and in brain samples from human AD patients [91-93] and from APP-over expressing mice [88,90,92]. APP and Aβ peptides enter mitochondria through TOM40 [88,89]. Intact APP molecules are not released into the mitochondrial interiors, but are arrested at the TOM complex in the TOM40 pore, in a complex with TOM23, and penetrate far enough into the intermembrane space to interact with the inner membrane protein TIM44 [88]. Possibly, γ-secretase complexes in the mitochondria release pathogenic Aβ peptides from the trapped APP [94,95]. Internalized Aβ is associated with the inner membrane fraction and/or the matrix [89,90]. Two of the main targets mediating Aβ effects on mitochondrial function are Aβ-related alcohol dehydrogenase (ABAD) [96], and the mitochondrial peptide processing enzyme PreP [97,98]. ABAD catalyzes the oxidation of short chain fatty acids, and helps protect against metabolic damage caused by conditions such as stroke [96], but Aβ-mediated inhibition of ABAD causes oxidative stress, leading to mitochondrial and cellular damage. PreP cleaves presequences from mitochondrial precursor proteins, and inhibition prevents maturation of matrix- and some inner membrane-directed proteins, and causes accumulation of mitochondrial preproteins and processing intermediates [98,99]. The Aβ-mediated inhibition of both enzymes, ABAD and PreP, can account for mitochondrial dysfunction at multiple levels, including impaired respiration [98,100,101], oxidative phosphorylation [88,90,91,101], and ATP production [88,102], defects in maintaining the inner membrane potential [88,98,102] and ion homeostasis [100], dysregulated ROS production [91,98,102], increased oxidative damage [90], abnormal dynamics [103] and increased apoptosis [102]. Mitochondrial Aβ levels were associated with mitochondrial dysfunction in a regionally specific manner in Aβ overexpressing mice [100], i.e. hippocampal and cortical mitochondria showed the highest levels of mitochondrial dysfunction, while striatal mitochondria were moderately affected, and amygdalar mitochondria were minimally affected. Moreover, these differences in mitochondrial Aβ were correlated with impaired cognitive function [100,104].

ApoE, and especially ApoE ε4 (1-272), is another possible cause of mitochondrial dysfunction associated with LOAD. ApoE fragments are found in detergent soluble and insoluble human brain fractions and are more abundant in fractions from LOAD cases than from age-matched normal controls [105]. In transgenic mice, they cause neurodegeneration and behavioral changes reminiscent of human LOAD [106]. On transfection into mouse neuroblastoma N2A cells, the APOE ε4 (1-272) fragment reduced cell survival and a portion of the expressed protein co-localized with mitochondria, and reduced the mitochondrial membrane potential [107]. Fractionating mouse brain extracts on affinity columns carrying either full length ApoE or ApoE (1 – 272) fragment, revealed several mitochondrial proteins specifically bound to ApoE (1-272), including ubiquinol cytochrome c reductase core protein 2, cytochrome c, and subunit 4 of cytochrome c oxidase [108]. ApoE and ApoE (1-272) proteins were recovered in the mitochondrial fraction of transfected N2A cells. The fragment modestly inhibited complex III and complex IV activities vs. the full-length protein, but cell survival rates, ATP production and membrane potential did not differ between the cell lines [108]. Although effects of the ApoE (1-272) fragment are modest in tissue culture, they may be stronger in other cell types resident in brain, such as microglial cells, or interactions between microglial cells and neurons in the intact brain may make them more potent. It is tempting to speculate mitochondrial localization of ApoE and ApoE (1-272) is due to interactions between the ApoE backbone and the TOM complex receptor subunits, since mitochondrial localization was dependent on three positive charges in the receptor-binding domain (aa 1-170) [107], and possibly this region constitutes a cryptic mitochondrial localization signal.

Parkinson’s disease (PD) also exhibits sporadic and familial forms and mitochondrial dysfunction was implicated in the etiology of both forms of PD. To date, at least seven genes were identified to cause the inherited rare form of PD. Products of three PD-genes, SNCA, PARK2 and PINK1 demonstrated interactions with the TOM complex and TOM40 and will be discussed herein.

Mutations in or multicopy variations of the SNCA gene, encoding the alpha-synuclein protein, cause autosomal dominant PD. Genome-wide association (GWAs) and candidate gene association analyses also implicated SNCA as a highly significant genetic risk factor for the common sporadic form of PD [109-121,122]. Bender et al. investigated alpha-synuclein induced mitochondrial dysfunction and reported that TOM40 levels vary reciprocally with alpha-synuclein levels in brain tissue from human subjects with PD as well as brain tissue from transgenic alpha-synuclein over expressing mice [123]. A portion of cellular alpha-synuclein localizes to mitochondria [124] and is imported via the TOM complex [125,126], where it inhibits complex I [125] and possibly other members of the oxidative phosphorylation chain [127], and elicits autophagy [128]. Transfection of rat B103 neuroblastoma cells with either the wild-type or the pathogenic A53T mutant alpha-synucleins caused loss of TOM40. Over expression with A30P alpha-synuclein, another pathogenic form did not [123], and the authors suggested the A30P mutation may alter the affinity of the mutant alpha-synuclein for the mitochondrial membrane, making it a poor substrate for interaction with the TOM complex; this bears further investigation. In brain samples collected from alpha-synuclein over-expressing mice and in the transfected neuroblastoma cells, the effect of alpha-synuclein on TOM40 levels was specific, in that TOM20 levels did not change in the cells, nor did the levels of oxidative phosphorylation complex III, II, or IV in the transgenic animals. These results suggest neither over-expression of the WT alpha-synuclein or expression of the pathogenic A53T mutant form altered overall abundance of mitochondria, but measurements using MitoTracker indicate treatment did deplete mitochondria. Further work is needed to resolve this paradox. Furthermore, over-expression of the wild-type and of the pathogenic A53T alpha-synuclein caused oxidative damage in neuroblastoma cells, and in the alpha-synuclein transgenic mice there was extensive mtDNA damage in neurons that expressed the transgene. In the neuroblastoma cells, co-expression of TOM40 and alpha-synuclein or the A53T mutant prevented the damage caused by expression of the alpha-synuclein by itself. Moreover, injection of lentivirus expressing TOM40 into the hippocampal and cortical regions of the transgenic mice reduced signs of cellular oxidative damage, appeared to improve mitochondrial function, as measured by tissue ATP levels, and diminished signs of inflammation and neurodegeneration [123]. Recalling the chaperon-like ability of TOM40 to bind disordered proteins [27], it is possible that before newly transcribed TOM40 proteins reached mitochondria they interacted with disordered α-synuclein in the cytosol, and these extra-mitochondrial complexes were recognized by cellular proteasomes for degradation. This would allow healthy mitochondria to replace alpha-synuclein-targeted mitochondria after clearance of the latter by autophagy.

Mutations in PARK2 and in PINK1 encoding the parkin RBR E3 ubiquitin-protein ligase and the PTEN-induced putative kinase 1, respectively, cause the autosomal recessive form of PD. These gene products act on the same pathway [129-131], and the disease-causing mutations lead to similar pathologies. PARK2/PINK1-mediated PD pathology also involves interactions with the TOM complex. Pink1, is recruited into mitochondria, enters via the TOM40 channel, and degraded in a membrane potential-dependent manner [132,133]. If the inner membrane potential collapses resulting in membrane depolarization, Pink1, associates with TOM via TOM22 in a high molecular weight complex134. In this form it recruits and activates Park2 [132,135], leading to destruction of outer membrane proteins and eventual mitophagy [135]. Initially Park2 is recruited to TOM70 (or TOM70A) or a near-by site [42]; at later times it is associated with TOM22 and subsequently with other outer membrane proteins [136]. These results suggest Park2 eliminates members of the TOM complex initially, triggering the loss of other outer membrane proteins and leading to mitophagy [42]. Pathogenic mutations in either protein disrupt this pathway [42,135], resulting in the accumulation of dysfunctional mitochondria.

In order to better understand the interactions of TOM40 and proteins involved in PD and AD, it is of interest to outline several findings from the research of Gaucher’s disease (GD). GD is caused by defects in sphingomyelin metabolism, and shares with AD, PD and related Lewy Body disease [137] the presence of aggregated, insoluble proteins. Large APP clusters characterize a mouse model of type III neuronopathic GD [138] that was generated by crossing mice that over-express sphingomyelin activator protein with mice bearing point mutations in acid β-glucosidase [139,140]. In many cases the APP clusters overlap with ones containing α-synuclein [138]. Anti-APP antibodies co-localized with anti-TOM40 antibodies, and anti-α-synuclein antibodies localized with anti-TOM20 antibodies [138]. In isolated neurons from these mice, APP co-localized with both TOM40 and COX IV to about the same extents [138]. Possibly, at least a portion of the APP that was bound in the TOM40 channel extended into the mitochondria and interacted with components of the inner membrane. α-synuclein co-localized with TOM20 and, to a much lesser extent, with TOM40 [138]. These abnormal localizations interfered with normal mitochondrial function since mitochondria isolated from the neuronal cells were bioenergetically compromised compared with those isolated from neurons from control mice [138]. These data are consistent with the results for APP discussed previously, in that it can clog the TOM40 channel and interfere with a host of down-stream mitochondrial functions. The fact that α-synuclein does not directly target TOM40 as extensively as we might expect, suggests the aggregated forms that develop in these models became trapped on the preprotein receptor.

The cell biological and biochemical evidence summarized above point unambiguously to the roles TOM40 and the TOM complex play in mitochondrial dysfunction that underlies PD and AD. Inhibition of TOM40 and blockage of mitochondrial protein import also underlie the etiologies of Huntington’s disease [141] and amyotrophic lateral sclerosis (ALS) [142]. Figure 1 summarizes the cellular interactions of TOM40 and TOM complex with the proteins involved in AD and PD pathogenesis. The normal pathway of mitochondrial protein import in healthy cells is described (Figure 1A) and a predicted model for the pathogenic mechanism is proposed (Figure 1B).

The Regulation of TOMM40 Gene Expression in Health and Disease

It has been suggested that changes in expression levels of normal proteins in the brain may be an important mechanism in the etiology of neurodegenerative diseases including AD and PD [143]. The full understanding of how TOMM40 contributes to LOAD-related phenotypes has involved functional analysis.

We showed recently that TOMM40 mRNA levels are significantly increased in LOAD versus control brain [144]. However, there are inconsistent reports in the literature regarding the relationship between LOAD status and TOMM40 expression. Lee et al. found that TOMM40 mRNA was reduced in peripheral blood samples from LOAD subjects compared to matched controls [145]. In subsequent studies they showed that TOMM40 expression remained suppressed vs. normal controls at one- and two-year follow-up points [146,147]. Tissue-specific regulation of expression of the TOMM40 gene may explain, at least in part, the contradicting results. Another study detected a correlation between TOMM40 mRNA levels and LOAD progression, but was inconclusive regarding the direction of the change, i.e. TOMM40 mRNA levels were higher in the frontal cortex of the majority of LOAD patients, but in the rest of the LOAD samples TOMM40 mRNA was down-regulated [148]. We would like to comment here that gene expression comparisons between LOAD versus control using whole brain tissues might have a limitation due to differences in the cell type composition resulting from neuronal cell loss in LOAD brains. Thus, experiments to measure gene expression in homogenous cell populations (i.e. neurons and glia) are warranted, and these experiments will also advance the understanding of the cellular mechanism in the basis of the disease. It is interesting to note that TOMM40-mRNA levels in LBvAD were further increased compared to LOAD only brains (our unpublished data), indicating that a comparison analysis of TOMM40-mRNA level across a broad range of AD and LB pathologies is of great interest.

To explore the functional consequences of the LOAD associated TOMM40-‘523’ variant we compared the effects of the VL- and the S-TOMM40 alleles on TOMM40 and APOE mRNAs expression in post mortem brain regions that are vulnerable to AD pathology, obtained from LOAD and cognitively normal aged individuals who were homozygous for the APOE ε3 allele. The mRNA expression levelsof both TOMM40 and APOE were greater in the occipital and temporal lobes from VL homozygotes compared with S homozygotes [144]. Moreover, a luciferase reporter system, with the intact native genomic context flanking the ‘523’ preserved, mimics the in vivo effect of ‘523’ on gene expression, that is to say the VL poly-T led to greater expression than the S poly-T [144]. We showed for the first time the functional significance of the ‘523’ locus in vivo in human tissues. These finding were in agreement with Bekris et al. that revealed a complex transcriptional regulatory region for TOMM40 and APOE expression that extends throughout both genes and is influenced by multiple polymorphisms including the ‘523’ locus [149]. In that study, Bekris et al. fused a promoter DNA fragment with a relative short putative enhancer sequence that contained the ‘523’ locus and demonstrated that ‘523’ length influenced TOMM40 promoter activity. In contrast, two other studies have investigated the association between ‘523’ and TOMM40 mRNA expression in human and reported negative results. The first study used human fibroblast cell lines derived from cognitively-healthy, APOE ε3/4 donors and found no significant differences in TOMM40 mRNA expression [150], and furthermore, this study found no significant effect of the ‘523’ on levels of TOM40 protein or on mitochondrial function and morphology. A second group analyzed the expression of TOMM40 and APOE in parietal cortex from subjects chosen without regard to APOE genotype and also did not detect an association between ‘523’ and TOMM40 or APOE mRNA levels [80]. That these experiments did not detect an association may be explained by the different tissues types assayed, very small sample size (especially when the analysis was repeated using a specific APOE genotype), and differences in the RNA analysis methodologies and study designs.

The Biological Consequences of Increased TOM40 Levels

We demonstrated increased TOMM40-mRNA expression in LOAD brains compared to healthy control, specifically in regions vulnerable to AD-related pathology [144]. Therefore we are investigating the biological effects of enhanced TOM40 expression in HeLa cell cultures that were stably transfected with a vector expressing full-length TOM40 protein or with a vector expressing the selection vector alone (controls). The TOM40 subcellular distribution is the same in the control and the over-expressing cells (data not shown). The chaperonin GRP75/Mortalin participates in mitochondrial protein import [151,152], and protects against oxidative stress [153,154] and its expression is down-regulated in PD [155,156] and AD [157]. In this cell-line model it resides, at least in part, in the mitochondrial matrix. Figure 2 shows over-expression of the TOM40 import channel did not affect levels of either GRP75 or of the oxidative phosphorylation complex IV, which resides on the mitochondrial inner membrane. Figure 3 shows, by contrast, over- expression of TOM40 did increase the abundance of complex I, another inner membrane protein, and of another matrix protein α-ketoglutarate dehydrogenase (OGDH). These results suggest the effect of over-expressing TOM40 is independent of subsequent importation pathway steps. Both complex I [158,159] and OGDH [160,161] are mitochondrial sources of reactive oxygen species (ROS). For this reason, we also investigated whether TOM40 over-expression is associated with altered ROS production, and observed that decreasing glucose in the growing media (from 25 mM to 5 mM) caused a mild increase in ROS in TOM40 over-expressing cells but not in the control HeLa C cells. We are currently replicating these experiments using neuronal cell-lines and performing further investigations to explore the implications of these results.

Figure 2.

Levels of GRP75 and TOM40 (A) and Complex IV activity (B) in control (transfected with an empty vector) and TOM40 over-expressing cells. Cultures at 75 –85% confluence were harvested in RIPA buffer supplemented with protease inhibitors and processed for determination of GRP75 and TOM40 by Western blotting (A) and Complex IV amount/activity using an ELISA/activity kit from MitoSciences according to the manufacturer’s instructions. Both HeLaC3 and TOM Mix cultures were processed simultaneously.

Figure 3.

Complex I (A) and ODGH (B) levels in control (transfected with an empty vector) and TOM40 over-expressing cell lines. Cultures were harvested at 75 – 85% confluence in RIPA buffer without protease inhibitors (A) or fortified with protease inhibitors (B). After one freeze-thaw cycle the respective mitochondrial enzyme complexes were measured using activity/ELISA methodology with kits from MitoSciences according to the manufacturer’s instructions. C: ROS levels were determined in control and TOM40 over-expressing cells after a mild hypoglycemic shock. Growth medium was removed from cultures once they had attained 75 – 85% confluence and was replaced with either fresh medium containing 25 mM glucose (standard) or 5 mM glucose (low glucose). After 24 hours the medium was replaced with Earl’s buffered salts solution containing 5 mM malate, 5 mM pyruvate and 10μM 5’,6’-chloromethoxy-2,7-dichlorodihydro fluorescein diacetate to quantify ROS. For all three experiments, all cell lines were processed simultaneously.

Concluding Remarks

Mitochondria underlie many cellular processes and it is not surprising functional and structural mitochondrial defects contribute to the pathogenesis of age-related diseases, including neurodegenerative diseases [162,163]. Here, we summarized the genetic and cell biological evidence linking TOMM40 and its cognate protein TOM40 with many of these disease-related changes in mitochondrial function. Human studies, in particular genetic association and phylogenetic analysis suggested TOMM40 as a genetic factor for LOAD risk, age of onset and other LOAD-endophenotypes, and provided evidence linking a gene encoded by the nuclear genome with mitochondrial dysfunction associated with LOAD. Animal models and cell-culture experiments showed that altered mitochondrial protein import or alterations in the amounts or distributions of proteins associated with mitochondrial play prominent roles in many neurodegenerative diseases, including LOAD and PD. Potentially, changes in the activity or abundance of any of the other members of the TOM complex also could contribute to disease etiology, and we anticipate future work will unravel these intriguing connections. We are developing new cell and animal models that will allow us to learn how TOMM40 regulation of gene and protein expression and variations in TOM40 abundance, contribute to the pathogenesis of LOAD and presumably other neurodegenerative conditions in aging. The availability of these models will enable us to isolate effects of TOMM40 from effects of all other genes. In addition to providing basic knowledge about how altered TOMM40 expression affects fertility, survival to adult-hood and life-span, they will also show how alterations in the expression of one subunit of the TOM complex influences expression of other members of the complex, the activity of the TOM complex, and mitochondrial protein composition and function, and whether there is tissue-selectivity to these processes. By creating hybrids with animal models of other neurodegenerative diseases, we may be able to obtain a more detailed understanding of how TOMM40 contributes to the pathogenesis of diseases such as PD and LOAD, HD and ALS. Knowledge gained may lead to identification of tractable therapeutic targets and development of new drugs for the delay or treatment of LOAD, PD and other neurodegenerative diseases, and will support mechanism-of-action studies for both new and re- purposed drugs, which, in turn, will complement the knowledge we gain from human genetics and cell biological studies.