Mechanisms Linking Mitochondria Dysfunction and Proteostasis Failure

Abstract

Maintaining cellular protein homeostasis (proteostasis) is an essential task of all eukaryotes. Proteostasis failure worsens with aging and is considered a cause of and a therapeutic target for age-related diseases including neurodegenerative disorders. The cellular networks regulating proteostasis and the pathogenic events driving proteostasis failure in disease remain poorly understood. Model organism studies in yeast and Drosophila reveal an intriguing link between mitochondrial function and proteostasis. Here we review recent findings of mitochondrial outer membrane-associated mRNA translation, how this process is sensitive to mitochondrial dysfunction and constantly surveyed by ribosome-associated quality control, and how defects in this process generate aberrant proteins with unusual C-terminal extensions that promote aggregation and drive proteostasis failure. We also discuss the implications for human diseases.

An Intimate Connection between Mitochondrial Function and Cytosolic Proteostasis

Proteostasis refers to a cellular state in which protein synthesis, folding, and degradation are maintained at a homeostatic state such that an intact yet dynamic proteome is preserved to meet the changing demands of cellular functions under different conditions [1]. It is generally believed that cellular mechanisms preserving proteostasis decline with age, contributing to the pathogenesis of age-related diseases [2]. Previous studies have emphasized the importance of quality control of fully synthesized mature proteins in proteostasis maintenance. However, the life cycle of a protein begins when it is synthesized on ribosomes, and a ribosome-associated quality control (RQC) mechanism is closely surveying the quality of nascent peptides [3, 4]. Defects in RQC can lead to perturbed proteostasis and is associated with neurodegeneration in mammals [5, 6]. Recent studies in yeast and metazoans highlight the intimate connection between RQC and mitochondrial health [7, 8]. Mitochondria are archetypal organelles in biology. In addition to producing ATP, these morphologically dynamic organelles possess highly varied proteome landscapes, are home to a vast array of metabolic pathways, and control essential cellular signaling processes from Ca²⁺ homeostasis to apoptosis [9]. We are still learning new roles of mitochondria as important signaling organelles [10, 11]. For example, in yeast mitochondria serve as a site where heat shock-induced misfolded proteins are imported and degraded [12], and ROS released by defective mitochondria has been shown to influence cytosolic translation machinery, presumably through redox regulation of ribosomal proteins [13]. These studies strongly support that there exist an intimate link between mitochondrial function and cytosolic protein translation and proteostatsis (Figure 1).

Figure 1. Possible Mechanisms of Mitochondrial Involvement in Cellular Proteostasis.

Protein synthesis, chaperone-assisted protein folding, and degradation of aberrant proteins represent major events in the proteostasis network. These events are carried out by mega-machineries and consume a large amount of ATP, which is provided primarily by mitochondria under normal conditions (I). Under extreme conditions, such as when yeast cells are under heat shock, many cellular proteins are denatured and misfolded and form aggregates. Persistent stress will overwhelm the cytosolic protein QC systems. While a majority of misfolded proteins are targeted for degradation by the cytosolic proteasome, some of the protein aggregates are imported into mitochondria through the TOM/TIM complex and targeted for degradation by mitochondrial proteases. If the burden of protein aggregates exceeds the capacity of mitochondrial proteases, mitophagy may be triggered to target the whole organelle containing aberrant proteins for degradation (II). Cells also synthesize certain nuclear-encoded mitochondrial proteins on mitochondrial surfaces using the cytosolic ribosomes. This process ensures that components of the multi-protein complexes such as the respiratory chain complexes, which are encoded by both the nuclear and mitochondrial genomes and are translated by both cytosolic and mitochondrial ribosomes, are made at the right stoichiometry. Dysregulation in this process can perturb mitochondrial proteostasis and trigger mitochondrial unfolded protein response (III).

Proteostasis failure manifested as formation of aberrant protein aggregates is a defining feature of age-related neurodegenerative diseases [14–16]. These hallmark aggregates include those composed of the amyloid beta peptide in Alzheimer’s disease (AD) (the amyloid plaques), α-Synuclein in Parkinson’s disease (PD) and other synucleinopathies (the Lewy bodies), tau in AD and related tauopathies (the neurofibrillary tangles), SOD1, TDP-43, FUS, and dipeptide repeats (DPRs) in amyotrophic laterosclerosis (ALS), and polyglutamine tracks in trinucleotide repeat disorders such as Huntington’s disease (HD). While protein aggregation may serve some normal function in biology and be protective at times [17], its persistence in cells can impair essential cellular proteins or structures and be detrimental. Another common pathological feature of neurodegenerative diseases is mitochondrial dysfunction [16, 18, 19]. Mitochondrial dysfunction has been profoundly implicated in other major human diseases, including cancer, diabetes, and also the aging process [20]. Gene products disrupted in major neurodegenerative disease such as AD and PD associate with or enter into mitochondria and regulate essential mitochondrial functions [19, 21–24]. Moreover, at least in the case of PD, certain environmental toxins [25] and genetic lesions [26] cause disease by specifically targeting mitochondria, supporting a primary role of mitochondrial dysfunction in disease pathogenesis. How mitochondrial dysfunction arises and moreover contributes to the often tissue- and cell type-specific neurodegenerative disease pathologies remains enigmatic. While mitochondria metabolism are important to neuronal function, including synaptic transmission [27], and bioenergetics failure may partially explain the susceptibility of energy-demanding neuromuscular tissues, energy supplementation has been largely ineffective in treating diseases associated with mitochondrial dysfunction, necessitating therapeutic strategies targeting other pathogenic events originating from mitochondrial impairment [20].

The co-existence of proteostasis failure and mitochondrial dysfunction as hallmarks of neurodegenerative diseases raises the tantalizing possibility that these two pathologies are mechanistically linked. Consistent with this notion, in animal PD model featuring prominent mitochondrial dysfunction, genetic manipulation of protein synthesis has proven to be beneficial [7, 28–31]. Similarly, in a mouse model of Leigh syndrome caused by genetic disruption of mitochondrial complex-I subunit NDUFS1, treatment with rapamycin, a specific inhibitor of the mechanistic target of rapamycin (mTOR) pathway that stimulates cytosolic mRNA translation and other anabolic processes, enhances survival and attenuates disease progression [32]. In AD, conserved features of mitochondrial stress response have been observed [33]. In worm and mouse models of AD, genetic or pharmacological manipulation of mitochondrial translation or mitophagy (see Glossary) reduces amyloid aggregation, improves cognitive capacity, and increases animal fitness [33, 34]. However, molecular mechanisms linking mitochondria to proteostasis regulation that are relevant to disease remain to be delineated [35]. Recent studies start to provide some clues.

Cytosolic Ribosomes are Recruited to Mitochondrial Outer Membrane

Mitochondria are complex organelles consisting of the outer membrane (MOM), inner membrane (IMM), intermembrane space (IMS), and matrix. Of the more than 1000 protein components of the human mitochondria [36], only 13 are encoded by the mitochondrial genome. Therefore, most mitochondrial proteins are encoded by the nuclear genome, synthesized in the cytosol, and delivered to mitochondria, primarily through the TOM/TIM complex (see Glossary). The conventional view is that these proteins are fully synthesized in the cytoplasm as pre-proteins and then imported into mitochondria with the help of mitochondrial targeting sequences [37], which are subsequently cleaved from the mature protein after import. Studies in yeast and metazoans however have challenged this view by demonstrating that many mitochondrial proteins are synthesized on MOM and engage in co-translational import (see Glossary) [29, 38, 39]. These new findings corroborate with a previous observation made by Ronald Butow and colleagues in the 1970s. By electron microscopy, they showed that cytosolic ribosomes were present on MOM and their abundance responded to cellular metabolic status [40]. Recent studies in yeast using electron cryo-tomography offered structural insights on the cytosolic ribosomes associated with MOM [41]. The logic of deploying a co-translational import mechanism at the mitochondria is not immediately clear. It is possible that co-translational import of these often hydrophobic, membrane-embedded mitochondrial proteins serves to avoid their misfolding, aggregation, or mistargeting in the cytoplasm. It is also possible that by making the multi-subunit complexes such as complex-I on demand, through coordinating the synthesis of nuclear- and mitochondrial DNA-encoded subunits, cells may prevent the accumulation of unassembled subunits in the cytosol or inside mitochondria, which could perturb proteostasis at both locations [35]. Other styles of protein import into mitochondria may also exist (e.g., an ER surface retrieval pathway is shown to safeguard mitochondrial membrane protein import in yeast [42]).

Co-translational import of nuclear-encoded mitochondrial proteins is regulated by the PINK1-Parkin pathway (see Glossary) [29], mutations in which cause familial Parkinson’s disease. The PINK1-Parkin pathway was initially shown in Drosophila to critically maintain mitochondrial function [43–45]. Previous studies have emphasized its role in mitophagy, a cellular process that targets dysfunctional or damaged mitochondria for autophagic clearance [46, 47]. Recent studies reveal that PINK1 and Parkin also regulate the recruitment of nuclear-encoded respiratory chain component (nRCC) mRNAs including complex-I 30 kD (C-I30) mRNA to MOM and direct their local translation in both Drosophila neuromuscular tissues and mammalian induced dopaminergic neurons (iDN) (Figure 2). Defects in this process affects mitochondrial structure and function and contributes to Parkinson’s disease pathogenesis in Drosophila and human iDN models [29]. It appears that co-translational import of nRCC mRNA into the mitochondria represents an earlier event in the continuum of PINK1/Parkin-directed mitochondrial quality control (MQC) program (Box 1). nRCC mRNAs have also been shown to be localized to axonal mitochondria [48], and the position of axonal mitochondria plays an important role in coordinating intra-axonal protein synthesis and marking axonal branching sites [49]. It will be interesting to test if the PINK1/Parkin pathway is involved.

Figure 2. MOM-Associated mRNA Translation and Its Regulation by PINK1/Parkin.

Select nRCC mRNAs such as C-I30 mRNA are recruited to mitochondrial surfaces, and the newly synthesized proteins are co-translationally imported into mitochondria through the TOM/TIM complex in Drosophila neuromuscular tissues and mammalian cells. This process is regulated by the PINK1/Parkin pathway in metazoans. PINK1 plays a role first in recruiting the mRNA/ribosome/mRNP complex from the cytosol to MOM (Step I). During this process, the mRNA/ribosome/mRNP complex is translationally repressed by RNA-binding proteins such as Pumillio and Gloround/hnRNP-F bound to 3′-UTRs of the mRNA. Upon arrival of the mRNP complex at MOM, PINK1 releases translational repression by promoting the degradation of the repressors by Parkin and further recruiting translational activators such as PABP and eIF4G (Step II). Since C-I30 and PINK1/Parkin are not present in yeast, this mechanism is specific to metazoans. It remains to be seen whether similar mechanisms, but different players are involved in regulating MOM-associated mRNA translation in yeast.

Box 1: Mitochondrial Quality Control Mechanism and Function.

Mitochondria are dynamic and complex organelles with essential roles in many aspects of biology, from energy production to intermediary metabolism to apoptosis. While a great deal has been learned about the physiological and pathophysiological roles of mitochondria, how eukaryotes monitor mitochondrial health and maintain organelle function, however, are relatively underexplored. In its simplest form, mitochondrial quality control (MQC) can be envisioned as a cellular response mechanism to deal with dysfunctional mitochondria. This can be achieved by refolding or removal of misfolded/damaged proteins by chaperones/proteases from the inside and outside of the mitochondria, fusion with healthy mitochondria to restore organelle function via complementation and exchange of metabolites, or autophagic removal of badly damaged mitochondria that are beyond repair [46, 47, 84]. A mechanistic understanding and holistic view of this critical cellular process currently is still lacking. Drosophila has served as a model system to uncover basic mechanisms underlying the regulation and function of MQC, thanks to the powerful tools available in this model organism and the generally high-level conservation of the mechanisms and function of mitochondrial regulation. Genetic studies in Drosophila have delineated a conserved signaling pathway governing MQC. This pathway was initially defined by Pten-induced kinase 1 (PINK1) and the E3 ubiquitin ligase Parkin, with PINK1 acting upstream of Parkin [43–45]. The physiological importance of MQC to postmitotic neurons is demonstrated by the association of mutations in PINK1 and Parkin with familial Parkinson’s disease (PD) featuring degeneration of dopaminergic (DA) neurons, which has also been observed in Drosophila PINK1 and Parkin mutants. Although the involvement of the PINK1-Parkin pathway in mitophagy regulation is broad and has been extended to mammalian systems, it is now becoming clear that there exists PINK1/Parkin-independent mitophagy pathways [85–87], and that before its triggering of mitophagy, the PINK1-Parkin pathway performs other functions to help preserve MOM-associated RQC, mitochondrial proteostasis, and even cytosolic proteostasis, suggesting that PINK1-Parkin signaling is an integral part of the cellular proteostasis network [7, 29, 31, 88].

Mitochondrial Outer Membrane-Associated Translation is Stalled by Mitochondrial Stress

The strategic positioning of select nRCC mRNAs on MOM for co-translational import also makes their translation sensitive to the functional status of mitochondria. For example, a drop in mitochondrial membrane potential, the driving force of protein import through the TIM complex [50], would cause a slowdown of the import of MOM-associated nascent peptides. This would lead to a collision of translating ribosomes on MOM, a triggering event of ribosome-associated quality control (RQC) [51, 52]. RQC is a recently described cellular quality control mechanism that handles stalled translation (Box 2). During nascent peptide chain (NPC) synthesis, the translating ribosomes could be stalled for a number of reasons, including mRNA defects, strong mRNA secondary structures, insufficient supply of aminoacyl-tRNAs, and electrostatic interaction between NPCs and the ribosome exit tunnel [3, 4]. Consequently, the co-translational QC complexes are recruited to stalled ribosomes to target the aberrant translation products and the template mRNAs for degradation. Factors such as ZNF598 and Rack1 sense colliding ribosomes, while Dom34/Pelo, HBS1, and ABCE1 separate 40S and 60S ribosome subunits and recycle them. Key RQC factors involved in protein quality control include Listerin (an E3 ubiquitin ligase), RQC1, RQC2 (Tae2/Clbn/NEMF), and Cdc48/p97/VCP. RNA quality control involves the 5′ to 3′ exonuclease Xrn1 and the exosome complex that degrades mRNA in the 3′ to 5′ direction. Although most of these factors are conserved in metazoans, their functions in RQC are largely limited to studies in yeast using artificial mRNA substrates [4] until recently.

Box 2: Co-translational Quality Control Mechanisms and Function.

Proteome integrity is essential for cellular health. Quality control of cellular proteins starts when proteins are being synthesized on ribosomes. Ubiquitination and degradation of nascent peptide chains (NPCs) still associated with ribosomes is a widespread phenomenon [89]. It is estimated that more than 30% of newly synthesized peptides are ubiquitinated in mammalian cells [90]. An inter-connected network of co-translational QC pathways, including the nonsense mediated decay (NMD), no-go decay (NGD), non-stop decay (NSD), and RQC pathways are deployed to survey the quality of translation. These distinct pathways target aberrant mRNAs containing premature termination codons (PTCs), no stop codons, or undergoing translational stalling due to mRNA damage, secondary structure, or aminoacyl tRNA shortage [3]. A common theme of these various types of co-translational QC is the splitting of stalled ribosomes before the clearance of stalled NPCs. A fundamental question is whether the protein C-terminal extension (CTE) by a CAT-tailing like process happens to all aberrant NPCs associated with the split 60S ribosome. The eRFs have been implicated in the resolution of more than one type of stalled ribosomes. In the case of PTC-containing mRNAs targeted by the NMD pathway, the NMD factors SMG1 and UPF1 form a complex with eRF1/eRF3 (the SURF complex), which interacts with the exon junction complex (EJC) containing UPF2, UPF3, eIF4AIII, Y14, and Magoh to trigger degradation of PTC-containing mRNAs [91]. The faulty NPCs are also degraded during NMD [92], but the mechanism remains poorly understood. Two components of the EJC, UPF3 and Y14, are found to act in the MISTERMINATE of C-I30 [7]. Inhibition of UPF3 was sufficient to induce C-I30-u formation and phenocopy the PINK1 mutant, which was rescued by eRF1 overexpression. The mechanism of action of UPF3 in NMD is not well understood. It acts as a nucleocytoplasmic shuttling factors, interacts with Y14 and NMD factors UPF1/UPF2, and mutations in UPF3B cause mental retardation [93]. UPF3, but not UPF2, is shown to be required for the E3 ligase activity of UPF1. Moreover, UPF3B participates in translational termination by interacting with eRF factors [94]. It would be interesting to test whether the termination function of UPF3B is important for its role in C-I30-u regulation, and whether it is related to the pathogenesis of mental retardation.

Using both cultured mammalian cells and in vivo Drosophila models, it was found that when mitochondria suffer damage, there is an early attempt to repair the damage, via PINK1-dependent recruitment of C-I30 mRNA and its associated ribosome/RNP complex to MOM in mammalian cells [31]. However, persistent mitochondrial damage results in stalled translation of C-I30 mRNA on MOM, recruitment of RQC factors and ribosome separation/recycling factors to C-I30 mRNP, and remodeling of the C-I30 mRNP complex. One of the key events occurring in C-I30 mRNP is the ubiquitination of ABCE1 by the ribosome-associated NOT4 E3 ligase, resulting in ABCE1 polyubiquitination [31]. Ubiquitinated ABCE1 then recruits autophagy receptors to C-I30 mRNP, triggering the autophagy of damaged mitochondria. This chain of events is critical for mitochondrial regulation and maintenance as genetic manipulation of key factors including ABCE1, effectively rescues the PINK1 mutant phenotypes. Moreover, downregulation of ABCE1 and HBS1 is found in transcriptome analyses of PD patient brain samples [31], supporting the importance of these factors and RQC in PD pathogenesis.

Inefficient Resolution of Stalled Translation Generates Aberrant Proteins with Unusual C-terminal Extension by CAT-tailing in Yeast and MISTERMINATE in Metazoans

Consistent with stalled translation of C-I30 mRNA upon mitochondrial stress, there is reduced C-I30 protein abundance in PINK1 or Parkin mutant flies or when wild-type animals are exposed to mitochondrial toxins [7]. Detailed analysis of C-I30 protein revealed the generation of an upshifted C-I30 band (C-I30-u) under mitochondrial stress conditions. After excluding posttranslational modifications (ubiquitination, phosphorylation, MTS-cleavage) and alternative splicing as possible sources of C-I30-u generation, and after demonstration of C-terminal extension (CTE) as possible mechanism of C-I30-u formation, a picture of C-I30 modification by a C-terminal Ala and Thr addition (CAT-tailing) (see Glossary) - like phenomenon started to emerge [7].

Studies in yeast have shown that NPCs on stalled ribosomes can be modified while still attached to 60S subunits, in a 40S- and mRNA template-independent CAT-tailing process [53]. So far CAT-tailing has been studied mainly using artificial substrates and under conditions preventing the degradation stalled NPCs [4]. The physiological role of CAT-tailing in the context of intact RQC is unknown. It has been proposed that CAT-tailing may induce the heat shock response [54], expose lysine residues in stalled NPCs hidden in the exit tunnel by pushing them out of the exit tunnel for ubiquitination by Listerin [55], or drive degradation of stalled NPCs on and off the ribosomes [56]. However, whether CAT-tailing-like CTE of stalled NPCs happens in metazoans, the compositions of the CTEs, and the pathological consequences of such CTEs in disease settings was not known due to the lack of identified endogenous CAT-tailed substrates. Studies in yeast have shown that the cytosolic Vms1 protein protects cells of mitochondrial toxicity from ribosome-stalled proteins, and that Vms1 or its mammalian counterpart cleaves peptidyl-tRNA bond and antagonizes CAT-tailing activity [8, 57–60]. But the CAT-tailing studies were done using reporter constructs. So far, no endogenous yeast protein is shown to be CAT-tailed, and C-I30 is not conserved in yeast.

Various evidences support Drosophila and mammalian C-I30-u as the first endogenous CAT-tailed substrate: i) Its reduction by knocking down Clbn/NEMF or AARS and TARS, factors needed for the CAT-tailing reaction; ii) Its reduction by overexpression of the Vms1/ANKZF1 family of mitochondrial stress-responsive factors that antagonize CAT-tailing activity; iii) Its reduction by overexpression of various RQC factors including ABCE1, VCP, Pelo; and iv) Mass Spec identification of CTEs containing A and T [7]. Distinct from CAT-tails in yeast that consist exclusively of A and T, CTEs of C-I30-u purified from PINK1 mutant flies also contain other amino acids (e.g. S, Y, C, and E/P). Moreover, in contrast to the CAT-tails in yeast, which is added to the C-terminal poly-K of GFP reporter without a stop codon [53], CTE in C-I30-u is added to C-I30 protein by ribosomes stalled at the canonical stop codon. This is caused by mitochondrial stress induced impairment of translation termination factor eRF1 and recycling factor ABCE1 [7]. Thus, although the CTE in C-I30-u in metazoans is generated by a mechanism thematically like CAT-tailing in yeast, there are fundamental differences. This process is termed MISTERMINATE (see Glossary) (Figure 3, Key Figure).

Figure 3, Key Figure. Mechanism of MISTERMINATE and Its Role in Disease.

In response to mitochondrial stress, the levels of translation termination factor eRF1 and the recycling factor ABCE1 are reduced. While the exact mechanism remains to be elucidated, ABCE1, an iron-sulfur cluster containing protein that is redox-sensitive and forms a complex with eRF1, might be the primary target. Loss of ABCE1/eRF1 creates ribosome-stalling signals that activate the RQC machinery. In the process of RQC, C-terminal extensions (CTEs) containing A, T, and select other amino acids are added to C-I30 protein, generating C-I30-CTE in a process termed MISTERMINATE. C-I30-CTE can be imported into mitochondria where it is assembled into C-I and may also form protein aggregates, resulting in mitochondrial toxicity. C-I30-CTE can also be released into the cytosol, where it forms protein aggregates in a CTE-dependent manner and such aggregates also recruit cellular meta-stable proteins, altering cytosolic proteostasis. Persistent activation of the RQC process on MOM may also titrate RQC factors away from their normal substrates, creating a deficit of overall cellular RQC activity. All of these events may contribute to disease pathogenesis.

Elucidation of the MISTERMINATE process in C-I30-u formation highlights the importance of translational termination and ribosome recycling factors in RQC. The availability of functional eRF1/eRF3 can strongly influence the amount of ubiquitinated NPCs associated with 60S in yeast [61]. Studies in yeast also showed that the ribosome recycling factor Dom34, the yeast homolog of Pelo, functions in rescuing ribosomes wondering past canonical stop codons. Such events occur more often under stress conditions, and Dom34 and ABCE1 functionally interact to limit translation re-initiation by terminating ribosomes in the 3′-UTR [62]. The relevance of this process to human brain aging and age-related brain diseases is highlighted by recent findings of ribosome-associated 3′-UTRs that lack 5′-UTR and open reading frame [63, 64]. These isolated 3′-UTRs are proposed to be the mRNA remnants of RNA quality control of stalled translation. They accumulate in an age-specific and brain-region specific manner and encode short peptides [64]. Moreover, their accumulation correlates with mitochondrial dysfunction and oxidative stress, and can be induced by the depletion of ABCE1 [64]. Thus, as in MISTERMINATE, mitochondrial stress and impairment of the translational termination/recycling factor ABCE1 results in altered RNA and protein quality control in the human brain. The pathogenic role of these aberrant mRNA and protein products remains to be tested.

ABCE1 is a Fe-S cluster-containing protein, and the Fe-S domain of ABCE1 interacts with eRF1 [65]. Fe-S association with ABCE1 is sensitive to cellular redox state and essential for ABCE1 function in translation termination and ribosome recycling [66]. ABCE1 may serve as a key component of the translation machinery that senses mitochondrial health. In addition, at least in yeast cells, the physiological level of ABCE1 is below the threshold sufficient for complete ribosome recycling [62], and ABCE1 behaves as an intrinsically meta-stable protein that can be sequestered by protein aggregates formed by other proteins [67]. Future studies will test whether ABCE1 represents a weak link between mitochondrial function and cellular proteostasis.

Mitochondrial Stress-induced C-I30 Protein with C-terminal Extension Can Form Aggregates and Perturb Mitochondrial and Cytosolic Proteostasis

In animal model studies the presence of C-I30-u is shown to correlate strictly with toxicity in neuromuscular tissues [7]. In both fly models and mammalian cells, C-I30-u is imported into mitochondria, assembled into C-I and super-complexes of the respiratory chain, and imposes mitochondrial toxicity [7]. Moreover, in both systems, C-I30-u is also released into the cytosol, presumably by the action of Vms1/ANKZF1, where it forms protein aggregates. Such aggregates recruit cellular meta-stable proteins and perturb the proteostasis network, as indicated by the recruitment of molecular chaperones and activation of signaling pathways responsive to proteostasis stress [7]. These findings are consistent with the propensity of A- and T-containing peptide polymers to form β-sheet and amyloid structures [68], and recent finding demonstrating failure in the timely removal of CAT-tailed proteins can cause proteotoxicity in yeast [69–71].

Coordination between Mitochondrial function and Proteostasis in Health and Diseases

It is unlikely that the formation of CAT-tails in yeast and MISTERMINATE-generated CTEs in metazoans is intended initially to cause proteotoxicity. Protein aggregate formation is a general phenomenon in biology, with possible functions ranging from consolidating aberrant proteins for quality control by the UPS or autophagy systems, or simply preventing their interactions with other cellular proteins, to serving normal roles in development, translational regulation, innate immune response, stress granule formation, or learning and memory [17, 72–76]. Under normal conditions, protein aggregates formed in response to stress are resolved after stress stimuli are removed. However, under aging or pathological conditions, such aggregates persist, resulting in sequestration of key cellular proteins like ABCE1, causing loss of essential cellular function and cell death (Box 3). These aggregates may also recruit disease-associated metastable proteins such as amyloid-β, tau, α-Synuclein, TDP-43, FUS, SOD1, or DPRs, driving the formation of disease-defining protein aggregates (Figure 4).

Box 3: Pathogenic Mechanisms of Protein Conformational Diseases.

Protein conformational diseases are characterized by disease-associated proteins acquiring aberrant conformations that can self-propagate to spread the disease. This category of diseases includes the infectious prion diseases and the seemingly non-infectious, age-related neurodegenerative diseases. The former is composed of the Creutefeldt-Jakob disease (CJD) in humans, scrapie in sheep, chronic waste disease in deer and elk, and mad cow disease [79]. The latter include AD, PD, tauopathy, ALS, and multiple system atrophy [79–81]. A key tenet of conformational disease pathogenesis is that an aberrantly folded prion conformer would have the ability to force another prion protein in its native conformation to adopt the aberrant conformation. In the case of human CJD, 10–20% of the patients carry familial mutations in the PRNP gene that encodes PrP^C. The genetic mutations may directly induce the formation of PrP^Sc, the infectious and disease-causing conformer, or facilitate the adoption of PrP^Sc conformation by mutant PrP^C. However, in sporadic CJD patients, which accounts for the majority of CJD cases, no mutations in PRNP genes are found. How PrP^Sc arises in this situation is not understood [79]. The same holds true for the candidate prion-like conformers that cause other neurodegenerative diseases, such as Aβ, tau, α-Synuclein, and SOD1, where familial mutations in protein sequences account for only a small portion of the disease cases. Even in the familial cases, the diseases tend to be late-onset, despite the presence of the mutated proteins after birth. This may suggest that the process of prion formation is a sporadic process, probably aided by other cellular cofactors or post-translational modifications of the disease proteins. It is also possible that the prion conformers are constantly turned over by cellular quality control mechanisms, and that only with age when such quality control activities decline [2], the toxic conformers are able to accumulate and reach a threshold level that leads to disease.

Figure 4. The Normal and Pathological Effect of MISTERMINATE-Generated CTEs.

Based on the propensity of A/T rich sequences to form beta-sheet structures, and the heterogeneity of the sequence and composition of A/T rich CTEs, it is proposed here that MISTERMINATE-generated C-I30-CTEs, or other RQC substrates containing CTEs (Protein X-CTEs), may adopt different shaped beta-sheet structures, which will lead to the formation of CTE-dependent aggregates. The normal function of CTE-mediated aggregation may be to concentrate these stalled ribosome-generated peptides, making it more efficient for cellular quality control machinery to refold them or target them for degradation. These aggregates are expected to be resolved after the stress signal is removed under normal conditions. However, under disease conditions when the stress signals persist or when RQC is impaired, CTE-containing proteins will accumulate and start to interact with other meta-stable proteins. These interactions will drive conformational changes of the meta-stable proteins to different conformers based on the CTEs involved, analogous to the prion-like conversions. This may represent one of the mechanisms by which normal cellular proteins acquire prion-like conformations to form diverse prion “strains” in sporadic forms of neurodegenerative diseases.

Recent studies in yeast and metazoans emphasize the particular importance of RQC in maintaining mitochondrial function by protecting the integrity of nuclear-encoded mitochondrial proteins that are co-translationally imported [7, 8]. A corollary of the finding of MISTERMINATE in linking mitochondrial stress to proteostasis failure is that mitochondrial dysfunction may contribute significantly to the pathological consequences of defective co-translational QC. Indeed, mutations in key RQC factors Listerin [5] and HBS1 [6] are linked to neurodegenration in mouse models. Mutations in VCP are involved in the Inclusion Bodies Myopathy with Paget’s disease of bone and Frontotemporal Dementia (IBMPFD), familial ALS, and Charcot-Marie-Tooth Disease Type 2Y [77], and ANKZF1 defects are linked to infantile-onset inflammatory bowel disease [78]. Future studies will test whether defective RQC of MOM-associated translation and mitochondrial function is causally involved in these disease conditions, especially the neurodegeneration aspect.

While mechanisms like MISTERMINATE may potentially explain the intimate relationship between mitochondrial dysfunction and proteostasis failure in disease, a fundamental question is whether the accumulation of CTE forms of many cellular proteins are involved, or selective few (e.g., C-I30-u) is the initial trigger for neurodegeneration. Available data suggest that accumulation of C-I30-u alone is sufficient to cause mitochondrial- and cyto-toxicity in mammalian cells [7]. Given the heterogeneity of MISTERMINATE-generated CTEs, it would be interesting to test whether CTE variants have the capacity to “seed” metastable proteins, resulting in the generation of diverse “strains” of prion (see Glossary) - like protein aggregates found in various neurodegenerative diseases [79–81] (Figure 4).

Therapeutic Implications

Although the biochemical mechanisms of CAT-tailing in yeast and MISTERMINATE in metazoans remain to be elucidated, available data suggest that these non-templated peptide chain extension processes are distinct from canonical peptide chain elongation, offering an opportunity to inhibit this process without adversely inhibiting normal translation. For example, anisomycin, an antibiotic produced by Streptomyces griseolus, which inhibits eukaryotic protein synthesis by targeting the peptidyl-transfer center of 60S, potently inhibits CAT-tailing in vitro in yeast lysates [82] and MISTERMINATE in vivo in fly models and mammalian cells [7] in a manner distinct from other translational inhibitors. More work is needed to determine the exact mechanism of anisomycin action in these processes, as anisomycin can act independent of canonical translation [83]. Future preclinical studies in rodent models and patient-derived induced neuronal models are needed before anisomycin or its derivatives can be tested as candidate therapeutics.

Genetic studies in fly models and human cells also indicate that activation of VCP or Vms1/ANKZF1 and inhibition of Clbn/NEMF are effective approaches in inhibiting MISTERMINATE and preventing mitochondrial stress-induced neuromuscular degeneration [7]. Small molecule compounds or gene therapy-based biologicals targeting these molecules may be pursued as novel therapies for disease conditions with prominent mitochondrial dysfunction. Finally, a potential application of MISTERMINATE-generated CTE is that it could be used as a biomarker for detecting mitochondrial stress in various age-related diseases. A specific antibody against a major subtype of MISTERMINATE-generated CTE populations could help achieve this goal. C-I30-u may be just the first example of CTE-containing proteins that accumulate on MOM under disease conditions. It is thus imperative to further characterize the MISTERMINATE-generated CTEs under different disease conditions to discover novel sequences that could serve as biomarkers.

Concluding Remarks

An intimate relationship between mitochondrial function and proteostasis exists across eukaryotes. Quality control of co-translationally imported mitochondrial proteins on MOM-associated ribosomes is critical for not only mitochondrial proteostasis but also global proteostasis in the cytosol. This provides a molecular framework for future studies to elucidate the mechanistic link between mitochondrial function and cytosolic proteostasis in health and disease. A limitation of current research is the paucity of knowledge of MISTERMINATE substrates, the compositions of the CTEs in these substrates, and the repertoire of disease-associated meta-stable proteins that are recruited by the CTEs. Many outstanding questions remain to be explored (see Outstanding Questions). For example, what is the nature of the mitochondrial stress signals that trigger MISTERMINATE? Do aging-related signaling pathways impinge on MISTERMINATE? What is the significance of the heterogeneity of the CTE generated by MISTERMINATE? Do different diseases produce different CTEs? Is the accumulation of MISTERMINATE substrates a contributing factor in Listerin mutation-induced neurodegeneration in mouse? Could mitochondrial dysfunction and MISTERMINATE explain the regional and cell type-specificity of neurodegenerative diseases? Finally, is MISTERMINATE a viable target for therapeutic development? These questions will be avidly explored in the coming years and exciting new insights are expected to emerge.

Outstanding Questions.

What is the nature of mitochondrial stress signals that impinge on MOM-associated RQC and MISTERMINATE? Is mitochondrial membrane potential, Ca²⁺, ROS, or certain metabolites involved?

What cellular signaling pathways act upstream to regulate MOM-associated RQC and MISTERMINATE? Are aging-related pathways involved?

What is the repertoire of MISTERMINATE-generated CTEs in different diseases? And what are the normal functions of the CAT-tails and CTEs induced by MISTERMINATE?

Can Listerin-like E3 ligase reach and modify the stalled NPCs on MOM? Are MISTERMINATE-induced CTEs involved in the neurodegeneration observed in the Listerin mutant mouse?

Could mitochondrial dysfunction and MISTERMINATE explain the regional- and cell type-specificity of neurodegenerative diseases?

Could MISTERMINATE-induced CTEs induce the formation of “prion”-like seeds of meta-stable proteins that form aggregates in the absence of disease-causing mutations in sporadic forms of conformational diseases?

Could RQC failure result in the accumulation of defective ribosome-associated products (DRiPs) (see Glossary) that can induce immune response?

Can new therapeutic approaches be devised by targeting mitochondrial protein import, mitochondrial proteostasis, MISTERMINATE/CAT-tailing, or RQC?

Highlights.

• Co-translational import of nuclear-encoded mitochondrial proteins is a common feature in eukaryotes

• Defects in synthesis of nuclear-encoded mitochondrial proteins on mitochondrial surface are found in Parkinson’s disease

• Mitochondrial stress can cause inefficient translation termination and/or recycling of mitochondrial outer membrane-associated ribosomes, triggering translation stalling and ribosome-associated quality control

• Translation stalling-induced protein C-terminal extension can cause protein aggregation, proteostasis failure, and contribute to disease

• Mitochondrial stress-induced aberrant protein C-terminal extension mechanistically links mitochondrial dysfunction with cytosolic proteostasis failure

• Mitochondrial dysfunction and proteostasis failure, along with oxidative stress and inflammation, are hallmarks of aging and age-related diseases

Glossary

CAT-tailing

Unusual protein C-terminal extension by random Ala and Thr addition, which occurs to stalled nascent peptide chains still attached on the 60S subunit of stalled ribosomes after the 40S and the template mRNA have been removed. Its physiological role remains unclear

Co-translational import

A cell biological process whereby nuclear-encoded mitochondrial protein mRNAs are translated by cytoplasmic ribosomes on mitochondrial surface and the nascent peptides are simultaneously imported into mitochondria through the import complex

DRiP

Abbreviation for defective ribosome-associated peptides, referring to peptides derived from newly synthesized proteins that are somehow defective in sequence or folding and degraded. Such peptides are thought to be the major source of MHC class I-presented peptides to T-cells

MISTERMINATE

Stands for mitochondrial stress-induced translational termination impairment and protein carboxyl terminal extension. Since such C-terminally extended proteins are aggregation prone, MISTERMINATE is considered a process that link mitochondrial stress to cytosolic proteostasis failure seen in neurodegenerative diseases

Mitophagy

Cell biological process by which dysfunctional or defective mitochondria are targeted for degradation by the autophagy machinery. Mitophagy is considered a cytoprotective mechanism. However, under certain conditions, excessive mitophagy can be detrimental to cells

PINK1-Parkin pathway

A genetic pathway in which PINK1 and Parkin, two genes mutated in familial Parkinson’s disease, work together to protect mitochondrial health, with PINK1 working upstream of Parkin. This pathway was initially defined through genetic studies in Drosophila and later shown to be conserved in mammals. PINK1 (Pten-induced kinase 1) encodes a Ser/Thr kinase targeted to mitochondria, and Parkin encodes an E3-ubiquitin ligase

Prion

A term coined by Stanley Prusiner for infectious protein assemblies that can convert normal proteins to pathogenic conformations and thus transmit disease between individuals. Although prion was initially discovered in Scrapie, an infectious degenerative disease affecting sheep and goats, recent studies suggest that the seemingly non-infectious major neurodegenerative diseases including Alzheimer’s disease and Parkinson’s disease may be caused by agents with prion-like properties, such as amyloid beta, tau and α-Synuclein

TOM/TIM complex

stands for translocase of outer membrane (TOM)/translocase of inner membrane (TIM) complex, a protein complex located on mitochondrial outer and inner membranes that conduct protein import from the cytosol into mitochondrial compartments. It consists of proteins that form import channels and peripheral proteins that assist in protein docking, translocating, and folding