A tale of two pathways: regulation of proteostasis by UPR^mt and MDPs

Abstract

Mitochondrial fitness is critical to organismal health and its impairment is associated with aging and age-related diseases. As such, numerous quality control mechanisms exist to preserve mitochondrial stability, including the unfolded protein response of the mitochondria (UPR^mt). The UPR^mt is a conserved mechanism that drives the transcriptional activation of mitochondrial chaperones, proteases, autophagy (mitophagy), and metabolism to promote restoration of mitochondrial function under stress conditions. UPR^mt has direct ramifications in aging, and its activation is often ascribed to improve health whereas its dysfunction tends to correlate with disease. This review pairs a description of the most recent findings within the field of UPR^mt with a more poorly understood field: mitochondria-derived peptides (MDPs). Similar to UPR^mt, MDPs are microproteins derived from the mitochondria that can impact organismal health and longevity. We then highlight a tantalizing interconnection between UPR^mt and MDPs wherein both mechanisms may be efficiently coordinated to maintain organismal health.

Introduction

Mitochondrial fitness deteriorates over time. The gradual loss of mitochondrial function results in deleterious consequences in organismal health and contributes to aging and age-related pathologies including neurodegenerative diseases such as Alzheimer’s (AD) and Parkinson’s (PD). Considering its significant involvement in multiple cellular processes such as energy production, apoptotic regulation, calcium signaling, and amino acid and lipid metabolism, it is not surprising that quality control mechanisms exist to protect the mitochondria under detrimental conditions, such as the unfolded protein response (UPR^mt). The UPR^mt transcriptionally activates genes required for mitigating damage to the organelle under conditions of stress. While the UPR^mt was originally identified as a critical mechanism to preserve mitochondrial health, recent evidence has uncovered its importance in other cellular processes, including the heat shock response [1], the endoplasmic reticulum [2], and the actin cytoskeleton [3]•. These studies argue that mitochondrial fitness and function can directly impact general cellular homeostasis, rather than solely as a byproduct of loss of mitochondrial function. Despite extensive studies in cellular and organismal models such as Drosophila melanogaster and Caenorhabditis elegans, unique mechanisms and molecular machineries dedicated to UPR^mt are yet to be fully elucidated in mammalian cells, which poses more questions for future unraveling. Here, we briefly review the most recent literature on how the UPR^mt, potentially in coordination with specialized microproteins derived from mitochondria, can directly impact protein homeostasis and organellar dysfunction, as well as how the activation of UPR^mt plays a role in the amelioration or progression of age-related diseases.

Diamonds in the rough: mitochondria derived peptides in age-related disease

Over 18,000 proteins have been validated from an estimated 20,000–25,000 genes within the human genome [4]. However, bioinformatic analysis of all potential open reading frames (ORF) within the genome identified millions of theoretical proteins – and some even embedded within annotated genes. Further characterizations of these ORFs identified millions of short open reading frames (sORF), defined primarily as those with peptide sequences shorter than 100 amino acids [5]. While thousands of microprotein products from these sORFs have been predicted from ribosome profiling experiments, proteomic detection has been lacking likely due to their small, low-abundant, and primarily hydrophobic features [6]. Human mtDNA also contain hundreds of sORFs, which have similarly been elusive to those encoded in the nuclear genome. The putative proteins encoded by these sORFs are only beginning to be understood as protein detection methods improve. These microproteins called mitochondrial derived peptides (MDPs) have numerous functions and are mechanistically implicated in many diseases [7]. Currently, there are eight major MDPs that have been functionally characterized, all of which show promising therapeutic potential in proteostasis-related diseases. Here, we briefly describe these best-characterized MDPs and cover their therapeutic potential focusing primarily on neuronal health and function (Fig. 1).

Fig. 1. Mitochondria-derived peptides impact organismal health and aging.

Humanin was the first functionalized mitochondria derived peptide (MDP) in the early 2000s, and since then, multiple other MDPs, including MOTS-c and SHLP1–6, were functionalized in various age-related diseases. Specifically, all MDPs have promise as a novel therapeutic agent: humanin encoded in the 16S rRNA has direct implications in AD and other neurocognitive diseases. MOTS-c, a putative exercise mimetic encoded in the 12S rRNA can improve metabolism and prevent pathology associated with metabolic diseases like obesity and diabetes. Finally, SHLPs encoded in the 16s rRNA can improve mitochondrial function and stress resilience to promote beta cell health.

Humanin (HN) was the first identified mitochondrial derived peptide through screening of cDNA libraries from AD brains in 2001 [8] and found by two other labs shortly after [9,10]. Due to its discovery in AD patient brains, HN has been studied as a potential therapeutic agent for neurodegenerative diseases. What makes HN unique is that unlike many other drugs that have potential negative side effects, HN has the potential to be one of the first drugs that can target AD with minimal, if not beneficial, side effects [18]. Perhaps this is the most unique aspect of humanin as a promising agent. Indeed, HN levels were found to be significantly lower in patients with AD [11] and specific genetic variants of humanin were found to be of higher risk for AD [12]. AD is the most common form of dementia and is attributed to increases in amyloid beta (Aβ) peptide production derived from the amyloid precursor protein (APP) on neuronal membranes. These amyloid peptides aggregate forming toxic oligomers and fibrils which form plaques and can contribute to neuronal death [13]. HN has been proposed to modulate amyloid aggregation by binding to Aβ peptides at aa17–28 [14,15]. Acetylcholinesterase (AChE), implicated in Aβ aggregation binds the similar regions of the Aβ peptide as HN suggesting that HN may block AChE mediated aggregation through competitive inhibition [14,16]. Neuronal cells (SH-SY5Y) treated with soluble Aβ42 were less viable, a phenotype that was restored in the presence of HN [17]. Additionally, 18-month-old mice treated with a HN analog had improved cognition and reduced circulating inflammatory cytokines. HN also has a role in inhibiting apoptosis through blocking BAX activation in neuronal cells which in part explain cognitive benefits with age [10]. In addition to its ability to mitigate neurodegeneration, HN has been implicated in longevity. Specifically, overexpression of HN was sufficient to increase lifespan in C. elegans [18], and while lifespan extension was not seen in middle-aged mice injected with HN, healthspan as measured by memory and metabolic parameters was significantly improved [9,14]. Finally, human offspring of centenarians display higher levels of circulating humanin, suggesting the impact of HN on longevity may be conserved [11].

Since HN’s discovery, 7 additional MDPs have been identified: mitochondrial ORF of the 12S rRNA type-c (MOTS-c; [19]) and six small humanin-like peptide (SHLP) family members [20]. These MDPs have also been proposed as protective against AD neuropathology, though the links are less established than those with humanin. Instead, a majority of MOTS-c and SHLP studies have been focused on metabolic dysfunction, including type II diabetes mellitus (T2DM). Specifically, MOTS-c has been implicated in regulating insulin sensitivity where its increase results in improved glucose clearance in vivo [19]. 32-month-old mice had decreased MOTS-c levels in skeletal muscle and serum accompanied with reduced glucose clearance which were attenuated with MOTS-c treatment. Humans with T2DM also have decreased levels of serum HN and MOTS-c, further implicating MDPs involvement in metabolic homeostasis [21]. SHLPs also modulate metabolism where SHLP2 and SHLP3 increased oxygen consumption rate providing further evidence of MDPs in enhancing metabolism [20]. T2DM is associated with Islet amyloid polypeptide (iAPP) aggregation, promoting amyloid deposits and contributing to β cell death. SHLP2 and HN analog humaninS14G, prevent aggregation by binding to iAPP [22]. Finally, similar to humanin, MOTS-c is implicated in longevity whereby it is described as an exercise mimetic and its supplementation can enhance physical performance in young, middle-age, and old mice and ultimately result in a trend towards significance for lifespan extension [23]•.

A clear direction for future MDP research is to maximize methods and techniques for the identification, detection, and functionalization of additional MDPs. For example, the ability to sensitively detect age associated MDPs may reveal novel biomarkers for aging and age-related diseases. A big question is whether MDP levels decline during the aging process and where mitochondrial dysfunction with aging is a cause of this – for example, as mtDNA mutations accumulate with age, it is possible that coding sequences for MDPs are compromised, resulting in lower production of functional MDPs. In addition, identification of novel MDPs may reveal promising therapeutic interventions. Indeed, a functional analog of MOTS-c is currently under clinical trials for nonalcoholic steatohepatitis and obesity. Finally, while Cobb et al eloquently demonstrated that MDPs almost definitively originate solely from mitochondria [20], a link to the nucleus still possible. Indeed, there are MDP-like sequences encoded in nuclear DNA that can potentially have similar function to their cognate MDPs [24], suggesting that perhaps MDP and MDP-like nuclear encoded proteins may serve either overlapping, synergistic, antagonistic, or redundant functions and that mito-nuclear balance of even microproteins are an essential regulation in protein biology.

Entanglement of mitochondrial unfolded protein response (UPR^mt)

One major challenge in the field of MDP biology is to ascribe a mechanistic model whereby MDPs can drive organismal health. One potential model is that MDPs can function characteristically as a transcriptional regulator to drive expression of genes responsible for stimulating mitochondrial function. For example, the MDP MOTS-c has been associated with the activation of the unfolded protein response of the mitochondria (UPR^mt), an evolutionary conserved mechanism that regulates mitochondrial homeostasis under deleterious conditions. Specifically, UPR^mt is a coordinated process between the nucleus and the mitochondrial network that can trigger the assembly of chaperones and proteases, to ensure accurate protein quality control [25–27]. The upregulation of UPR^mt-related genes reestablishes a stable mitochondrial environment and mechanistically combat unfavorable conditions such as mtDNA mutations, mitochondrial protein imbalance, altered membrane permeability, damaged mitochondrial respiratory chain, and protein aggregate accumulation [25–27]. Chaperones localized within the matrix, such as mtHsp70 (from the Hsp70 family) [28] and Hsp60/Hsp10 chaperonin [29], are responsible for protein import, folding, and maturation while proteases roaming the matrix and those integrated in the inner membrane known as “mitoproteases”, degrade misfolded or misassembled proteins within the mitochondria [30]. Mitoproteases comprise membrane-integrated AAA+ metalloproteases including the intramembrane space-oriented iAAA and matrix-oriented mAAA, as well as serine metalloproteases such as LON and ClpXP in the mitochondrial matrix [31,32]. Although there are extensive data on genes and proteins involved in the activation of UPR^mt, specific mechanistic triggers and pathways involved is yet to be resolved.

Cumulative evidence demonstrates a link between UPR^mt-related genes and aging. For instance, LON protease activity is known to decline with age, a process that contributes to the accumulation of misfolded proteins in the mitochondria over time and can affect stress responses induced by UPR^mt [32,33]. Similar to LON, the presequence of the peptidase responsible for degrading amyloid-beta peptides in the mitochondria, known as PreP protein, also declines with age and is known to aggravate Alzheimer’s disease (AD) [34,35]. The imports and exports of nuclear-encoded precursor proteins can also be initiated by UPR^mt-related genes which when impaired, can have direct ramifications to aging and age-related diseases. Protein aggregates accumulates in the mitochondria when presequence processing is dysfunctional and precursor proteins fail to reach maturation [32,34]. Impairments in the presequence processing can occur due to detrimental point mutations in the main mitochondrial processing protease (MPP) and results to severe neurodegeneration and childhood deaths in human patients [36]. To immediately combat mitochondrial dysfunction, a nuclear HMG-box transcription factor, ROX1, was found to translocate to the mitochondria in yeast cells and initiate an early UPR^mt by engaging to mtDNA in a TFAM-like manner [14]. In turn, early UPR^mt induction via ROX1 translocation results to enhanced protein refolding and degradation [14]. To what extent can the upregulation of UPR^mt-related genes facilitate mitochondrial stress responses and adapt to detrimental and diseased states are still being explored.

Although the UPR^mt was first observed in mammalian cells during mtDNA depletion [37,38], many of its regulatory pathways were discovered in model organisms such as Drosophila melanogaster, with most of the studies done in Caenorhabditis elegans [39–42]. Due to its ease in manipulation and transparency, C. elegans has been an ideal model in investigating key proteins associated with UPR^mt using fluorescent reporters. It is using C. elegans models that UPR^mt activation by CLPP-1 cleavage was found to result in a peptide efflux via HAF-1 transporter [43]. In turn, this peptide efflux was suggested to negatively localize ATFS-1 to the mitochondria and promote further LONP-1 degradation [39]. Subsequently, the accumulation of ATFS-1 in C. elegans, together with cofactors UBL-5 and DVE-1, stimulates the expression of known UPR^mt target genes including hsp-60 and hsp-6 chaperones –known homologs of mammalian Hsp60/Hsp70 [27,39]. It has also been recently highlighted that an ATFS-1-dependent activation of UPR^mt increases mitochondrial import [44]••. This was quite paradoxical to the requirement of decreased mitochondrial import driving nuclear accumulation of ATFS-1 for UPR^mt activation. Intriguingly, it seems that the increased mitochondrial import was context dependent (i.e. preferentially drives import of mitochondrial chaperones to restore mitochondrial function) and was actually required for the beneficial effects of UPR^mt activation on organismal health [44].

Mitochondrial import driving organismal health is not unique to C. elegans: as mitochondrial function declines with age, mitochondrial permeability transition pores (mPTP) increase in response to high ROS concentration or excessive cytosolic Ca²⁺ and possess detrimental effects on mitochondrial import [45]. The composition of mPTP proteins is still currently debatable, however higher frequency of mPTP formation is observed in the early onset of age-related diseases such as Alzheimer’s and Parkinson’s, and late stages of cardiovascular disease and stroke [46–48]. Further evidence has suggested that F-ATP synthase (complex V) is the most probable mPTP candidate and may be involved in mPTP-related diseases [49,50]. Importantly, loss of F-ATP synthase subunit OSCP/atp-3 during development leads to longer UPR^mt and increases longevity. On the contrary, loss of OSCP/atp-3 subunit during adulthood leads to formation of mPTP and activates a reversible ATFS-1-dependent UPR^mt [46]••.

In contrast to the pathways described in C. elegans, most aspects of the mammalian UPR^mt are yet to be fully understood and whether a homologous response to mitochondrial stress exists has been challenged [51]. In mammalian cells, mitochondrial proteotoxic stress was reported to induce transcription of C/EBP Homologous Protein (CHOP) transcription factor which, in turn, initiates the upregulation of UPR^mt-related genes. In this process, CHOP binds to the UPR^mt gene promoter via two conserved sequences known as the mitochondrial UPR elements 1 and 2 (MURE1 and MURE2) [52,53]. However, whether CHOP initiates UPR^mt initiation remains in-question since the transcription factors allowing MURE1 and MURE2 binding are still unclear [37,54]. Recent evidence indicates that CHOP may be a fine-tuning regulator of mitochondrial integrated stress response (ISR), together with C/EBPβ, and ATF4, instead of its proposed function as a transcriptional activator of UPR^mt [37]. Although it was first characterized in yeast, the ISR was found to be highly conserved in mammals [55]. This is in direct contrast to the “canonical” ATFS-1-driven UPR^mt described in C. elegans, although some suggested that ATF5, mammalian homolog to ATFS-1, mediates UPR^mt in mammalian cells [56]. However, upon proteosome inhibition, ATF5 can still be detected in mammalian cells which poses more questions on the role of ATF5 in UPR^mt. Despite these contradicting theories, several mitokines have been proposed to induce UPR^mt in mammalian cells including fibroblast growth factor 21 (FGF21) which also regulates a portion of ISR independent of ATF4 in cardiomyocytes and other tissues, and growth differentiation factor 15 (GDF15) in skeletal muscles and mice models [57,58].

In addition to molecular triggers, UPR^mt can also be activated in a cell-non-autonomous manner (Fig. 2), which dictates a global cellular response against mitochondrial dysfunction across multiple tissues [59–61]. Deleterious intracellular changes caused by an accumulation of disease-related protein aggregates in a specific tissue can influence mitochondrial function, UPR^mt, and proteostasis in distal organs. For instance, accumulation of protein aggregates in neurons trigger the expression of Wnt ligand/EGL-20 in C. elegans, which promotes a serotonin-dependent UPR^mt activation from the nervous system to peripheral tissues leading to enhanced proteostasis [59,62]. In addition, activation of UPR^mt specifically in neurons both in the presence [60] and absence [63] of stress can result in distal activation of UPR^mt in nonneuronal cells and improved organismal health and longevity. These data suggest that increased UPR^mt activity in C. elegans can actually result in improved longevity. However, another study that surveyed multiple methods of UPR^mt induction and found no direct correlation of UPR^mt induction with lifespan. In fact, they found that certain conditions that significantly induce UPR^mt can actually negatively influence lifespan [64], showing that the association between UPR^mt and longevity is complex and likely context-dependent (for more detailed review of UPR^mt on longevity, please refer to [65]).

Fig. 2. Non-autonomous communication of UPR^mt between tissues.

MOTS-c, an MDP found in neurons, and FGF21, a mitokine found in skeletal muscles, are proposed to activate UPR^mt and modulate key processes in distal tissues such as thermogenesis in adipose tissues, as well as glucose tolerance and insulin sensitivity in pancreas that results in body weight reduction. An overall systemic response triggered by UPR^mt activation has also been shown in other models such as C. elegans. Neurodegenerative-related protein aggregation in the nervous system of C. elegans upregulates UPR^mt, resulting in the systemic regulation of proteostasis in peripheral tissues. An increasing accumulation of protein aggregates in the C. elegans germline upon stress also reduces mitochondrial content and induce the UPR^mt in distal somatic cells (e.g. the nervous system, muscles, and intestine) through Wnt signaling, leading to the accumulation of protein aggregates in somatic tissues.

Interestingly, despite the overall improvement in organismal proteostasis initiated by non-autonomous signals in organisms with protein aggregation specifically in neurons, accumulation of protein aggregates in distal tissues can have detrimental effects on organismal proteostasis [66]•. In C. elegans germline, protein aggregates have been shown to reduce mitochondrial content in distal somatic cells via Wnt signaling. In addition, accumulated protein aggregates in the germline activates somatic UPR^mt and was associated with mitochondrial fragmentation, inducing a further accumulation of disease-related protein aggregates in the nervous system, muscles, and intestine. These findings highlight tissue-specific variations in the induction of the UPR^mt and the effects of cell non-autonomous signaling in the overall organismal response to combat extensive protein aggregation. Distinct organisms may also exhibit different regulatory mechanisms of the UPR^mt. This begs the question of how extensive and to what degree can the activation of UPR^mt affect the whole organismal system.

While the UPR^mt field and its impact on aging and age-related diseases is a far more developed field than MDP biology, there are still significant questions that remain unanswered. First, the major controversy that exists between directly linking UPR^mt and longevity must be addressed. For example, while various methods of inducing mitochondrial dysfunction results in mitohormesis that drives lifespan extension [42,60,67], inhibition of complex II genes and many other methods fail to extend lifespan despite robust activation of UPR^mt [64]. One plausible explanation is that the extent of mitochondrial stress/dysfunction is variable, and there is a fine line between sufficient stress to activate a beneficial UPR^mt versus inducing irreparable damage. Regardless, further work is necessary to truly understand the effects of UPR^mt on longevity. This is especially true in mammalian systems where UPR^mt is far more complex with the major involvement of other stress response regulators like the heat-shock transcription factor HSF1 and the oxidative stress regulator NRF2 [3], as well as components of the ISR [51].

Concluding remarks

There exist many signaling molecules and tissue-to-tissue communication involved in UPR^mt and some of these genes and regulators are shared across different species. However, a general consensus of how UPR^mt is triggered as well as key molecular pathways unique to UPR^mt in mammalian cells, are yet to be established. This raise questions as to whether UPR^mt is governed by a single cascade of reactions or an amalgamation of various stress responses, or whether it is only present in certain organisms. Further investigation, with a combination of experimental models and high-throughput analyses, could offer insights on the exclusive features and conserved characteristics of UPR^mt and verify its existence and importance to mammalian systems.

However, accumulating evidence suggests an undeniable role for the UPR^MT, potentially in coordination with MDPs, in maintaining proteostasis and metabolic function during aging. An obvious next question is understanding how UPR^MT activation can impact MDP generation and vice versa? Indeed, a recent study showed in a mitohormesis model that neurons experiencing mitochondrial stress display an increase in expression of MOTS-c [68]••. Specifically, mitoribosomal stress is applied to POMC neurons, which resulted in non-autonomous activation of UPR^mt and increased thermogenesis in adipose tissue, which improved metabolism and increased resistance to obesity in mice. In this paradigm, it was clear that induction of mitochondrial stress in POMC neurons drove induction of MOTS-c, thus suggesting a tantalizing model whereby activation of UPR^mt drives production of MDPs to improve cellular health. Understanding mechanistically how mitochondrial stress drives MDP synthesis and how MDPs function downstream of UPR^mt activation to promote organismal health are exciting avenues of future research. For example, can MDPs function as an additional retrograde signal (along with canonical UPR^MT transcription factors) to promote activation of UPR^mt and other genes required for increased proteostasis and general stress resilience? Indeed, MOTS-c interacts with numerous stress-responsive transcription factors including NRF2 and ATF1/7 [69]. Finally, in a mitohormesis model, could synthesis and secretion of MDPs serve as a potential mitokine signal? Can MDPs be secreted from neurons and trafficked to other cell types to promote nonautonomous UPR^MT? MDPs are found in circulating serum [7,70], and thus it is possible it can serve as an endocrine signal.

In thinking of future clinical applications in aging and age-related diseases, it is clear that robust activation of the UPR^mt may not be a viable option. While mitohormesis models show that UPR^mt activation downstream of mitochondrial dysfunction extends lifespan [42,67,71], mitochondrial dysfunction is directly causative of multiple chronic diseases in humans [72], which makes intentional dysregulation of mitochondria a highly unattractive therapeutic. Moreover, UPR^mt activation has been correlated with multiple disease states, including cancer [73], which suggests that even if UPR^mt activation can be accomplished in the absence of stress, several unwanted side effects may blunt the potentially positive effects of such a therapy. Thus, it is imperative to dissect the downstream effects of UPR^mt activation to better curate a method that can promote only its beneficial effects, while avoiding negative consequences of chronic UPR^mt activity. An alternative therapy could be the usage of MDPs. Both humanin and MOTS-c have primarily been associated with improved health and mitigation of certain disease, including metabolic syndrome [21] and AD [14,15], and do not suffer the caveats of the more global and pleiotropic effects of UPR^mt activation.

Highlights.

• Mitochondria-derived peptides (MDPs) impact proteostasis and neuronal health

• Identification of novel MDPs can reveal new therapies for age-related diseases

• Mitochondrial permeability transition pore promotes aging

• Neuronal UPR^mt promotes whole organism health through non-autonomous signaling

• Germline mitochondrial proteostasis prevents protein aggregation in distal tissues