If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Laboratory of Functional Neurogenetics, Department of Neurodegeneration, Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen, GermanyInterfaculty Institute of Biochemistry, University of Tübingen, Tübingen, Germany
Laboratory of Functional Neurogenetics, Department of Neurodegeneration, Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen, GermanyInterfaculty Institute of Biochemistry, University of Tübingen, Tübingen, GermanyGerman Center for Neurodegenerative Diseases (DZNE), Tübingen, Germany
Mitochondria are important organelles in eukaryotes. Turnover and quality control of mitochondria are regulated at the transcriptional and posttranslational level by several cellular mechanisms. Removal of defective mitochondrial proteins is mediated by mitochondria resident proteases or by proteasomal degradation of individual proteins. Clearance of bulk mitochondria occurs via a selective form of autophagy termed mitophagy. In yeast and some developing metazoan cells (e.g., oocytes and reticulocytes), mitochondria are largely removed by ubiquitin-independent mechanisms. In such cases, the regulation of mitophagy is mediated via phosphorylation of mitochondria-anchored autophagy receptors. On the other hand, ubiquitin-dependent recruitment of cytosolic autophagy receptors occurs in situations of cellular stress or disease, where dysfunctional mitochondria would cause oxidative damage. In mammalian cells, a well-studied ubiquitin-dependent mitophagy pathway induced by mitochondrial depolarization is regulated by the mitochondrial protein kinase PINK1, which upon activation recruits the ubiquitin ligase parkin. Here, we review mechanisms of mitophagy with an emphasis on posttranslational modifications that regulate various mitophagy pathways. We describe the autophagy components involved with particular emphasis on posttranslational modifications. We detail the phosphorylations mediated by PINK1 and parkin-mediated ubiquitylations of mitochondrial proteins that can be modulated by deubiquitylating enzymes. We also discuss the role of accessory factors regulating mitochondrial fission/fusion and the interplay with pro- and antiapoptotic Bcl-2 family members. Comprehensive knowledge of the processes of mitophagy is essential for the understanding of vital mitochondrial turnover in health and disease.
Mitochondria, development, aging, and Parkinson's disease (PD)
Mitochondria are specialized organelles present in most eukaryotic cells. In addition to energy production, mitochondria play important roles in nutrient and lipid metabolism as well as in apoptosis. Two specialized membranes compartmentalize mitochondria. This feature, together with remnants of an own genome (mitochondrial DNA; mtDNA), is likely due to their endosymbiont heritage. Although the vast majority of 1136 proteins in human mitochondria (www.broadinstitute.org/files/shared/metabolism/mitocarta/human.mitocarta3.0) are encoded in the nucleus, the mitochondrial genome codes for 13 proteins, all of which are components of the respiratory chain (
). During oxidative phosphorylation, electrons flow through the respiratory chain in the mitochondrial inner membrane (MIM). This process builds up the membrane potential (proton gradient) that drives ATP synthase. As a by-product, reactive oxygen species are produced (
). As their energy production is obligatorily dependent on oxidative phosphorylation, neurons are particularly vulnerable to mitochondrial dysfunctions. Oxidative damage is aggravated in age-dependent neurodegenerative diseases, such as PD (
). Indeed, the two most common recessive PD gene products, phosphatase and tensin homolog (PTEN)-induced kinase 1 (PINK1) and the E3 ubiquitin ligase parkin, are enzymes that mediate the autophagic removal of mitochondria (mitophagy). This mitophagy pathway might offer therapeutic targets for the treatment of PD (
). Moreover, mitochondrial turnover is important for cellular homeostasis, and in a few cell types (reticulocytes, germ cells) mitochondria are eliminated during normal development.
Given the tremendous importance of mitophagy in development, aging, and neurodegeneration, it is imperative to understand the cellular mechanisms regulating cargo-selective autophagy of mitochondria. Enormous progress in understanding this process in diverse cells types has been made in the past two decades, unraveling key players including the aforementioned PINK1, parkin, and other mitochondrial outer membrane (MOM) ubiquitin ligases as well as ubiquitin-binding autophagy adaptors. Clearly, mitophagy is embedded in an intricate signaling network integrating mitochondrial elimination signals, complex posttranslational modifications (PTMs) of MOM proteins, and the coupling to autophagy propagation toward autophagolysosomal degradation. These complex signaling events throughout the entire process of mitophagy are beginning to be understood, also thanks to recent proteomic studies. This review describes how the basic elements of the autophagy machinery and their coupling to cargo-selective mitochondrial removal are connected via PTMs mediated by mitophagy-regulating enzymes.
Brief introduction to autophagy
Macroautophagy (hereafter autophagy) is a major intracellular, tightly regulated catabolic process in eukaryotes that delivers cytosolic components including protein aggregates, damaged organelles, and invasive pathogens for lysosomal degradation (
). This process is activated in response to various stressful conditions such as nutrient starvation, growth factors deprivation, hypoxia and infections, and is essential for maintenance of cellular homeostasis (
). The core mechanism comprises the sequestration of “to-be-degraded” cargo into the cup-shaped isolation membrane termed phagophore, which expanses and seals into a double-membraned sphere-termed the autophagosome- which engulfs autophagic cargo. Upon maturation, the outer membrane of autophagosome fuses with the vacuole (in yeast and plants) or endosomes and lysosome (in metazoans), leading to the degradation of the autophagic body together with its cargo by lysosomal catabolic enzymes (
Autophagy is activated by cellular triggers such as amino acid deprivation, which inhibits the master cell growth regulator serine/threonine kinase “mammalian target of rapamycin” (mTOR), and reduced energy levels activating AMP-activated protein kinase (AMPK) (
). In high nutrient conditions, the mTOR complex 1 (mTORC1) binds and phosphorylates the uncoordinated protein 51-like kinase 1 (ULK1) at residue S757, which disrupts the interaction with inactive AMPK (
). Upon starvation, AMPK is activated, which blocks mTORC1 activity and initiates autophagy at the transcriptional level through dephosphorylation of the master transcription factor EB (TFEB) that is negatively regulated by mTORC1 (Fig. 1). Dephosphorylated TFEB is released from mTORC1 and translocates to the nucleus where it stimulates coordinated lysosomal expression and regulation (CLEAR) by binding to the CLEAR-box sequence (5′-GTCACGTGAC-3′) present in the regulatory region of many lysosomal and autophagy-associated genes (
). Moreover, inactive mTORC1 dissociates from the ULK1 complex, which reduces ULK1 S757 phosphorylation. Instead, starvation-activated AMPK binds to ULK1 and phosphorylates it on multiple distinct sites (
Activated ULK1-bearing numerous PTMs forms an oligomeric megacomplex together with its cofactors, autophagy-related protein ATG13, focal adhesion kinase-interacting protein of 200 kDa (FIP200) and ATG101, which is translocated to ER-associated isolation membrane assembly sites (
). ATG14L together with beclin-1 associates with the lipid kinase vacuolar protein sorting-associated protein VPS34 and the pseudokinase VPS15 to form a PI3K-III (Fig. 1). The ULK1 complex component ATG13 targets ULK1 to its substrate ATG14L for S29 phosphorylation (
). Such regulation of PI3K promotes the local production of phosphatidylinositol 3-phosphate (PI3P) at an ER structure called “omegasome” (Fig. 1), recruiting the PI3P-binding double FYVE domain-containing protein DFCP1 (
Phagophore elongation requires two conjugation systems, involving the E1 ubiquitin ligase ATG7 and the E2 ubiquitin-conjugating enzyme ATG10, which links the ubiquitin-like protein ATG12 to ATG5 (Fig. 1). ATG16L1 is recruited to the phagophore via binding to WIPI2 and conjugated to ATG5-ATG12, forming an ATG5-ATG12-ATG16L1 complex (
). The other conjugation system targets the ATG8 proteins, which can be subgrouped into LC3 (microtubule-associated protein light chain 3) and GABARAP (γ-amino-butyric acid receptor-associated protein) family members. Nascent pro-ATG8 precursors are processed by ATG4 cysteine proteases. The proteolytically processed ATG8 proteins expose a C-terminal glycine, allowing conjugation to phosphatidylethanolamine by the ATG3 E2 conjugating enzyme after activation by ATG7 (Fig. 1). Such ATG8 lipidations promote phagophore expansion (
). Lipidated ATG8 proteins play an important role in the recruitment of cargo destined to lysosomal degradation into autophagosome, a process mediated by autophagy receptors, which brings part of the machinery to the autophagosomal membrane. The ULK1 complex regulates the membrane supply for sealing and involves the ATG9 trafficking system. Additionally, ULK1 may also phosphorylate various autophagy receptors, thus playing a role in selective autophagy including mitophagy (
Following phagophore sealing, the autophagosome undergoes maturation that results in clearance of ATGs from the autophagosomal outer membrane and recruitment of two machineries: one responsible for lysosomal delivery, comprised of kinesin motor; and the other responsible for autophagosome fusion with the lysosome, comprised of a soluble N-ethylmaleimide-sensitive factor-attachment protein receptor (SNARE), including syntaxin17 and synaptosomal-associated protein 29 on the autophagosome and vesicle-associated membrane protein 8 on the lysosome (Fig. 1). The lysosomes are recruited and tethered to autophagosomes via the homotypic fusion and protein-sorting (HOPS) complex, which is composed of VPS11, VPS16, VPS18, VPS33A, VPS39, and VPS41 (
Whereas bulk autophagy mediates a nonspecific engulfment and degradation of cytoplasmic material, selective autophagy is mediated by specific autophagy receptors. The main role of selective autophagy is to maintain the intracellular homeostasis by selectively degrading distinct cellular components such as damaged mitochondria (mitophagy), endoplasmic reticulum (ERphagy), lysosomes (lysophagy) invading pathogens (xenophagy), aggregated proteins (aggrephagy), and others (
). The molecular machinery of selective autophagy must efficiently identify the cargo and sequester it within autophagosomes. Autophagic receptors/adaptors that contain ubiquitin-interacting motifs (e.g., p62 sequestosome-1, optineurin, nuclear dot protein of 52 kDa (NDP52), next to breast cancer type 1 susceptibility gene 1 protein (NBR1) and autophagy-linked FYVE protein) connect their ubiquitylated cargo with the autophagosome via LC3-interacting region (LIR) motifs (
). On the other hand, autophagy receptors such as the B cell lymphoma-2 (Bcl-2) 19 kDa protein-interacting proteins BNIP3 and BNIP3L/NIX or the FUN14 domain-containing protein 1 (FUNDC1) act independently of ubiquitin to deliver cargo to phagophores via their LIR motifs directly (
General principles of cargo-selective mitochondrial autophagy
Mitophagy is the cellular catabolic process responsible for maintenance of mitochondrial homeostasis and quality control. Various developmental, environmental, physiological, and pathological stimuli activate mitophagosome formation resulting in mitochondrial clearance (
). While the Atg11-Atg32 interaction is crucial for mitochondria recognition as a cargo and recruits the targeted mitochondria to the phagophore assembly site, the Atg8-Atg32 interaction through Atg32 LIR motif mediates phagophore expansion. The transcriptional levels of Atg32 may be upregulated due to oxidative stress or starvation conditions via suppression of TOR and release of transcriptional suppression of Atg32 by the Ume6-Sin3-Rpd3 complex (
). Such phosphorylation facilitates Atg32 interaction with Atg11. Especially, phosphorylation at S114 is essential for successful mitophagy, since substitution of this serine residue to alanine prevents Atg11-Atg32 interaction and abolishes mitophagy (
Regulatory mechanisms of mitophagy in mammalian cells
Two pathways are responsible for mitophagy mediation in mammalian cells, the ubiquitin-mediated pathway and receptor-mediated pathway. The ubiquitin-mediated pathway is dependent on autophagy adaptors/receptors such as p62, optineurin, transient axonal glycoprotein 1 binding protein 1, NDP52, and NBR1 that contain ubiquitin-binding domains in addition to their LIR motif (
). These adaptors/receptors sequester the ubiquitylated mitochondrial proteins through their ubiquitin-binding domain and recruit the mitochondria into autophagosomes by interacting with lipidated ATG8s through their LIR motifs (
), which ubiquitylates MOM proteins and induces mitophagy. On the other hand, parkin ubiquitylation-independent mitophagy pathways are mediated by autophagy receptors such as NIX, BNIP3, FUNDC1, Bcl-2-like protein 13, and FK506-binding protein 8 (FKBP8) (
). All of these receptors undergo MOM incorporation and recruit lipidated ATG8s via their LIR motif. Such recruitments label damaged mitochondria as a cargo and promote their engulfment into the growing phagophore.
In addition to protein receptors that mediate mitophagy, several lipids are also implicated in this process. Cardiolipin was reported to act as mitophagy receptor in neuronal cells. Upon mitochondrial stress, cardiolipin translocates from the MIM to the MOM and there, directly interacts with the N-terminus of LC3, mediating clearance of damaged mitochondria (
): (i) quality control removing dysfunctional mitochondria to maintain cellular homeostasis and viability (see below) and (ii) during organismal development, for example, when red blood cells lose all their organelles in the terminal step of erythropoiesis. Upregulation of the autophagy receptor NIX during terminal erythroid differentiation is the key trigger for cargo-selective autophagic removal of mitochondria from red blood cells (
) offered tremendous insights into the cell biology of mitochondrial quality control. Under normal conditions, PINK1 is effectively transported into mitochondria, where it is immediately degraded. As the mitochondrial protein import machinery depends on ΔΨm, damaged mitochondria with collapsed ΔΨm can no longer import PINK1. Instead, PINK1 massively accumulates on the MOM (
Under healthy conditions, PINK1 is imported into mitochondria via an N-terminal mitochondrial targeting sequence (MTS). Once imported into the mitochondrial matrix, the MTS of pre-PINK1 is cleaved off by the mitochondrial processing protease (
When the kinase PINK1 cannot be imported for mitochondrial processing and clearance, it accumulates on depolarized mitochondria and thus acts as a molecular switch to trigger mitophagy. Following such accumulation at the MOM, PINK1 becomes autophosphorylated at S228 and S402, which is important for PINK1 catalytic activity and the recruitment of parkin (
). The individual contributions of phosphorylations at S228 and S402 for PINK1 functions are not entirely clear. Functional PINK1 autophosphorylation of S228 in mammals is conserved in insects, where the homologous residue S346 in Drosophila melanogaster is needed for PINK1 activity and parkin translocation to mitochondria (
). Charging the really interesting new gene (RING)-type E3 enzyme parkin with ubiquitin requires the sequential action of the E1 ubiquitin-activating enzyme and a cognate E2 conjugating enzyme. In the resting state, parkin is folded in an auto-inhibited manner (
). When such structural reorganization of the Ubl domain and E2 binding site of parkin has taken place, the E3 ligase parkin promotes the attachment of ubiquitin to specific MOM substrates during mitophagy (Fig. 2B). These events are antagonized by PTEN-L, the first discovered ubiquitin phosphatase (
At the onset of mitophagy, parkin is very efficiently translocated from the cytosol to the MOM of depolarized mitochondria, such that in model cells lines practically the entire immunofluorescence signal is found associated with mitochondria within an hour. Active PINK1 on the MOM is necessary and sufficient for this step (
). More specifically, targeting parkin catalytic activity by site-directed mutagenesis (C431S, C431F, and G430D) or rendering the PINK1 site unphosphorylatable (S65A) delay or compromise parkin translocation to the MOM after mitochondria depolarization (
A straightforward model would posit that PINK1 accumulated on depolarized mitochondria would phosphorylate ubiquitin and parkin at the respective S65 residues, thereby activating the parkin ubiquitin ligase holoenzyme and enriching it on mitochondria by engagement with MOM substrates. Indeed, phosphorylation of parkin at S65 in the Ubl occurs without parkin reaching the MOM in cells expressing catalytically inactive parkin (C431S) (
). Activation of parkin promotes its translocation from the cytosol to depolarized mitochondria. Accumulating on depolarized mitochondria, phospho-ubiquitin stimulated parkin strongly promotes the ubiquitylation of MOM proteins. Such ubiquitin molecules become substrates for further PINK1 phosphorylation, generating a feed-forward amplification loop at the MOM.
A specific subset of E2 ubiquitin-conjugating enzymes was identified that regulated parkin-dependent mitophagy (
). Curiously, even the combined knockdown of these parkin coenzymes UBE2N, UBE2L3, and UBE2D2/3 only delays parkin recruitment and subsequent mitochondrial ubiquitylations and p62 localization to depolarized mitochondria, indicating additional regulatory steps for parkin recruitment and the process of mitophagy.
It seems hard to envision that cells rely for the intricate control of mitophagy on random vicinity of cytosolic ubiquitin and parkin with PINK1 accumulated on depolarized mitochondria. New insights into ubiquitin chain structures during mitophagy recently reveal similar relative amounts of mono-ubiquitylated and single-branched ubiquitin species in HeLa cells expressing catalytically active or inactive mutant (C431S) parkin, suggesting the presence of a mono-ubiquitin and single-branched ubiquitin coat in depolarized mitochondria (
). Likewise, recent global ubiquitylome analyses show similar amounts of mono-ubiquitylated and single-branched ubiquitylation species in induced neurons expressing both wild-type or nonphosphorylable parkin (S65 A) after 6 h of mitochondria depolarization (
) that would become directly phosphorylated by PINK1 on the MOM of depolarized mitochondria, generating a phosphorylated MOM-ubiquitin coat available for parkin to ubiquitylate (Fig. 2A). In addition, the mitochondrial ubiquitin ligase MUL1 acts upstream of the PINK1/parkin pathway, restraining mitofusin-2 (MFN2) and hence maintaining mitochondrial morphology and ER-mitochondria contacts in stressed mature neurons (
). While the determination of parkin substrate specificity is quite unclear, there is a cadence of consistent and functionally important ubiquitylations during the process of mitophagy (Fig. 2, B and C), some of which are highlighted in the following.
Mitochondrial Rho GTPase (MIRO) proteins are important factors for mitochondrial motility. In parkin-expressing HeLa and SH-SY5Y cells, they are rapidly poly-ubiquitylated. Interestingly, MIRO-2 is not completely eliminated for at least 90 min, and although the unmodified MIRO-1 band vanishes with a half-life time of ≈20 min (
). Moreover, the PINK1/parkin-induced burst of MFN2 phospho-ubiquitylation triggers p97-mediated MFN2 complex disassembly and thus reduces the tethering of mitochondria with ER, facilitating mitophagy (
). On the other hand, parkin has also been reported to have an impact on mitochondrial fission by actively ubiquitylating the dynamin-related protein 1 (DRP1) and targeting it for rapid proteasomal degradation (
). Indeed, proteomic studies identify strong parkin-dependent ubiquitylation sites of MFN1/2 after short times of mitochondrial depolarization, but hardly any consistent ubiquitylation events and no degradation of DRP1 seem to appear (
). Thus, parkin-mediated poly-ubiquitylations directly downregulate fusion factors and more indirectly attract fission factors, which promote the mitochondrial fragmentation necessary for efficient autophagic removal.
Porins comprise a class of parkin substrates that was recognized early on in experiments knocking down VDAC1 (
). In mammals, these most abundant MOM proteins exist in three isoforms: voltage-dependent anion selective channel VDAC1-3. While the relative contribution of each individual VDAC for distinct steps of mitophagy was somewhat controversial (
). Individual VDAC protein expression levels and ubiquitylation stoichiometry might differ with cell type and culture conditions, but VDACs are certainly targeted in parkin-mediated mitophagy, and it is possible that specific VDAC isoform PTMs affect selective aspects of mitochondrial quality control. For example, it was recently found that VDAC1 poly-ubiquitylations promoted mitophagy, whereas mono-ubiquitylation of VDAC1 at K274 rather conferred apoptosis (
Ubiquitin contains seven lysine residues (K6, K11, K27, K29, K33, K48, and K63). Depending on the lysine used as a linkage for attaching the next ubiquitin molecule, several distinct ubiquitin chains may form that will differ on their acquired structure. While some chains have clear physiological consequences such as K48 or K63-linkage chains, which trigger proteasomal degradation or autophagy activation, respectively, less is known about the other poly-ubiquitin chains (
Even though the detailed linkage types and structures of the ubiquitin chains built on parkin substrates are unknown to date, initial studies focused on ubiquitin chain types during mitophagy showed that parkin-dependent mitophagy triggers the elimination of mitochondrial substrates by building K48-linked, proteasome-targeting poly-ubiquitin chains (
). The timing of parkin target selection and the linkage specificity of ubiquitin chain elongation are likely regulated by PTMs and regulatory factors that need to be further elucidated. How parkin writes the ubiquitin code for the mitophagy process is a key question for future studies.
Role of DUBs
The MOM protein ubiquitylations formed during mitophagy are naturally subject to modulation by DUBs. The MOM residing DUB USP30 directly removes ubiquitin molecules attached by parkin on depolarized mitochondria, thus impeding parkin-dependent mitophagy progression (
). Moreover, parkin itself is deubiquitylated on the MOM by USP33, notably at an apparently regulatory site (K435), leading to altered ubiquitin linkage activities and thus interfering with parkin recruitment and mitophagy (
). Conversely, USP8 actually promotes parkin-dependent mitophagy by preferentially removing autocatalytically formed K6-linked ubiquitin conjugates from parkin. In the absence of USP8, persistent K6-linked poly-ubiqutin chains on parkin interfere with the translocation of parkin to depolarized mitochondria and the promotion of mitophagy, possibly by affecting the functional interactions with PINK1-phosphorylated ubiquitin and/or interactions with ubiquitin-binding autophagy receptors (
). Perhaps the dissociation of the DUB USP35 from depolarized mitochondria is sufficient to facilitate actions of the ubiquitin ligase parkin. It remains unknown whether USP35 additionally regulates components of the autophagy machinery as was shown, e.g., for USP33 (
). Instead, USP36 regulates the expression of ATG14L/beclin-1, possibly via epigenetic mechanisms.
Taken together, the process of parkin-mediated mitophagy is modulated at multiple steps: the direct parkin DUBs affect parkin activity and recruitment to damaged mitochondria in a facilitating (USP8) or repressive manner (USP33), the parkin-mediated MOM protein ubiquitylations can be subsequently removed (USP15, USP30) or left untouched (USP35), and specific effectors of the autophagy/mitophagy machinery can be indirectly regulated (USP36).
In addition to the PINK1/parkin-dependent mitophagy pathway described above, alternative parkin-independent pathways can recruit LC3 to mitochondria and promote their selective autophagic elimination (
). Two different parkin-independent mechanisms can be distinguished: (1) receptor-mediated or (2) ubiquitin-mediated mitophagy (Fig. 3). In response to low oxygen (hypoxia) or mitochondrial uncoupling, receptor-mediated mitophagy will take place. Autophagy receptors residing at the MOM, which are regulated by phosphorylation or dephosphorylation, directly interact with LC3 molecules attached to the autophagosome, thereby promoting mitochondrial elimination through autophagy (
). Up to the present day, the kinases responsible for BNIP3 and NIX activation remain unknown. FUNDC1 is another MOM autophagy receptor that is auto-inhibited in basal conditions. FUNDC1 is intrinsically phosphorylated at Y18 and S13 by Src kinase and casein kinase 2, respectively. It becomes active under conditions of severe hypoxia, where the phosphoglycerate mutase family member 5—a mitochondrial serine/threonine-protein phosphatase dephosphorylates S13, allowing the interaction of the LIR domain with LC3 (
). Like most autophagy receptors, AMBRA1 can be phosphorylated in the vicinity of the LIR motif (at S1014), which stimulates coupling to LC3B and promotes mitophagy. The inhibitor of nuclear factor κB (NF-κB) kinase-α was validated to phosphorylate its consensus sequence around S1014 -near the AMBRA1 LIR motif- upon mitophagy (
). How AMBRA1 and its associated factors accumulate on mitochondria for cargo-selective autophagic clearance remains to be further studied, as well as the potential regulatory role of Bcl-2 proteins interacting with AMBRA1 (
) is a straightforward part of the nutrient-sensing program, as all nutrients can be metabolized to acetyl-CoA under permissive cellular conditions. Specifically, general control of amino acid synthesis protein 5-like 1 (GCN5L1) is a positive regulator of protein acetylation in mitochondria (
). In oxidatively stressed endothelial cells, induction of SIRT3 promotes mitophagy via deacetylation of the forkhead box protein O3, promoting the stimulation of gene batteries regulating mitochondrial biogenesis, fission/fusion, and the mitophagy effectors BNIP3, NIX, and LC3 (
), both under normal conditions and in stressed cells with activated PINK1/parkin. However, the effects of SIRT3 are highly complex, and the molecular details for cargo-selective mitophagy in specific cell types remain to be resolved.
Mitochondria are rich in iron bound cytochromes and iron-sulfur clusters. Remarkably, the iron chelator deferiprone was a powerful inducer of PINK1/parkin-independent selective mitophagy even in PD patient fibroblasts (
). Likewise, deferiprone induces mitochondrial clearance in myotubes by hypoxia-inducible factor 1-α mediated transcriptional induction of the LC3-binding autophagy receptors BNIP3 and NIX but, interestingly, instead of autophagic degradation in this case, mitochondria were secreted as extracellular vesicles (
). Taken together, iron homeostasis is important for PINK1/parkin-independent mitochondrial clearance.
Cardiolipin is a phospholipid normally residing at the MIM, which can be externalized under mitochondrial stress conditions by phospholipid scramblase-3. Such relocalization to the MOM allows cardiolipin to directly bind to LC3 (
Apart from parkin, other E3 ubiquitin ligases are also capable of decorating the MOM with the purpose of targeting them for specific autophagy elimination. The ubiquitin ligase homologous to the E6-AP carboxyl terminus, ubiquitin-associated and WWE domain-containing protein 1 (HUWE1) acts as an alternative to parkin, binding AMBRA1 upon mitophagy. Similar to parkin, HUWE1 can be recruited to depolarized mitochondria where it ubiquitylates MFN2 for proteasomal degradation (
). Like parkin, gp78 ubiquitylates MFN proteins to target them for proteasomal degradation, thereby promoting mitochondrial fragmentation. Interestingly, gp78 engages LC3 to ER-associated mitochondrial sites and promotes mitophagy in an ATG5-dependent manner. Gp78 levels are regulated by self-ubiquitylation as well as ubiquitylation by the ubiquitin ligase mahogunin RING finger protein 1 (MGRN1). Expression of a disease-causing prion protein disrupts the Ca2+-dependent interaction of MGRN1 with gp78 (
), preventing the degradation of gp78 and thus initiating mitophagy. It is possible that calcium stress affects turnover of ER–mitochondria contact sites, involving the alternative ubiquitin ligase gp78, also in a PINK1-dependent manner (
The mitochondrial ubiquitin ligase MUL1 shares many substrates with parkin and is able to compensate for PINK1/parkin loss of function in the context of PD, in Drosophila and mouse neurons, ubiquitylating MFN2 and promoting its degradation through the proteasome system (
). MUL1 can act together with MFN2 to control mitochondria morphology and ER–mitochondria contracts, and thus, it is suggested that when MUL1-MFN2 pathway is disrupted, the PINK1/parkin mitophagy pathway will be activated (
In addition to completely PINK1/parkin-independent mitophagy pathways, the ariadne-1 homolog in humans (ARIH1) promotes mitophagy in cancer cells independently of parkin but in a PINK1-dependent manner (
). Both neurons and myocytes strongly depend on mitochondria. Thus, the complete autophagic elimination of mitochondria via the PINK1/parkin pathway commonly studied in cultured cell lines that satisfy their energy demands largely by glycolysis may not be reflected on all accounts in the brain, muscle, and heart in vivo. While DRP1-mediated mitochondrial fission is important for mitophagy in neurons and cardiomyocytes, parkin acts merely as a facilitating factor but not as essential regulator of mitophagy in neurons and cardiomyocytes (
), likely to pinch off damaged fragments from the mitochondrial network for piecemeal mitophagy. Mitochondrial fission and autophagic removal are coupled by parkin via pathways associated with dephosphorylation of DRP1 at S637 in its GTPase domain (
). The mitochondrial dynamics protein of 51 kDa is an important cofactor for parkin/DRP1-mediated mitochondrial fragmentation and autophagic removal, with mitochondrial dynamics protein of 49 kDa and the mitochondrial fission factor (MFF) also contributing (
) and may thus promote fission and mitophagy in cells suffering from energy crisis (Fig. 4A). Moreover, MFF expression is regulated by the RNA-binding translation repressor pumilio2 in an age-dependent manner (
). Moreover, MUL1 acts not only as a ubiquitin ligase but can also transfer the small ubiquitin-related modifier (SUMO) to DRP1. The mixed activities of MUL1 as a stabilizing DRP1-SUMO ligase and a destabilizing MFN2-ubiquitin ligase promote mitochondrial fission (
). Downregulation of mitochondrial fusion factors would shift the mitochondrial morphology dynamics toward fragmentation. A major factor driving fragmentation of damaged mitochondria is the parkin-mediated ubiquitylation of MFN, targeting for proteasomal degradation of a key mitochondrial fusion factor (see above). Interestingly, depletion of MFN1 may also contribute to mitochondrial quality control in the female germline (
). This process does not occur via the PINK1/parkin pathway. Rather the fragmented mutant mitochondria produced less ATP, which marked them for autophagic clearance involving Atg1 and BNIP3, but not Atg8a (
). An alternative macroautophagy pathway regulated by Atg1/ULK1 and beclin-1 was described in Atg5−/−/Atg7−/− double-knockout mouse embryonic fibroblasts, where autophagosomes seem to be assembled in a RAB9-dependent manner by the fusion of isolation membranes with vesicles derived from the trans-Golgi and late endosomes (
). In starved cardiomyocytes, (pS555)ULK1 phosphorylates RAB9 at S179, which leads to the formation of an (pS555)ULK1/(pS179)RAB9 complex on mitochondria, that recruits the receptor-interacting protein kinase 1, which in turn mediates the fission-promoting phosphorylation of DRP1 at S616 (
). This mitochondrial fission/mitophagy complex eventually associates also with the autophagolysosomal fusion SNARE syntaxin17. The regulation of mitochondrial cargo selectivity of alternative macroautophagy in developing and stressed cells remains to be studied in detail.
MFN1 is the major MOM fusion factor and OPA1 regulates MIM fusions (
). This shift is regulated by FUNDC1 dephosphorylation at S13 near the LIR motif. Thus, mitochondrial fission and autophagy engagement may be efficiently integrated cellular processes connected by multifunctional autophagy receptors, subject to regulation of PTMs.
Conversely, hyperfusion spares mitochondria from degradation upon starvation-induced autophagy (
). In this case, phosphorylation of DRP1 prevents the mitophagy-facilitating fragmentation of mitochondria. In addition, mitochondrial fusion is achieved by preventing cleavage of the fusion-promoting long isoform L-OPA1 by the endopeptidase overlapping with the mitochondrial matrix ATPase associated with diverse cellular activities (m-AAA) protease 1 (OMA1) (Fig. 4C). Regulated expression of OMA1 may modulate mitophagy via OPA1-mediated morphology dynamics (
). It remains to be further elucidated which protein deacetylations mediate sirtuin effects on mitochondrial metabolism and cargo-selective autophagy (see also above).
Bcl-2 family members
The MOM, which connects to the autophagy machinery during mitophagy, is strongly influenced by members of the Bcl-2 family. Bcl-2 proteins are key regulators of MOM permeability controlling the intrinsic, mitochondrial apoptosis pathway. It is becoming increasingly clear that Bcl-2 proteins have functions beyond their involvement in the regulation of apoptosis (
), including mitochondrial physiology and autophagy (Fig. 5). For example, interactions of anti-apoptotic Bcl-2 family members with beclin-1 repress autophagy (Fig. 5A). Specifically, beclin-1 contains a Bcl-2 homology 3 domain (BH3) and hence binds to antiapoptotic proteins such as Bcl-2, Bcl-W, Bcl-XL, and the induced myeloid leukemia cell differentiation protein Mcl-1 (
Bcl-2 family members are also affected in PINK1/parkin-dependent mitophagy. In CCCP-treated neurons, the upregulated PINK1 phosphorylates ubiquitin S65 and with a similar time course multiple serine residues of the Bcl-2 associated agonist of cell death, suppressing its proapoptotic translocation to mitochondria (
) and Bcl-2 homologous antagonist/killer (BAK), the resulting proteasomal degradation of these proapoptotic Bcl-2 family members would prevent apoptosis to allow cells to cope with the mitophagic stress (
). Thus, parkin may promote mitochondrial quality control by triggering mitophagy while suppressing apoptosis by ubiquitylation-mediated proteasomal targeting of proapoptotic Bcl-2 family members but when the damage cannot be relieved, parkin-mediated degradation of Mcl-1 may switch cell fate to apoptosis (
). Taken together, the cytoprotective enzymes PINK1 and parkin are distinctively interconnected with anti- and proapoptotic Bcl-2 family members and mitochondria at the crossroads of mitophagy and apoptosis (
). In the analyses of disease-related changes in the ubiquitylome, further development of existing methods enables the investigation of the ubiquitin chain architecture. For example, the “Ub-clipping” technique revealed that parkin mainly generates ubiquitylations that are mono-ubiquitylated or only consist of a short chain with a preference for distal ubiquitin phosphorylation (
The integration of several layers of -omics studies is essential to gain a more complete view of mitochondrial biology. Recently such integrative studies were done in the field of systems biology, e.g., for modeling biological processes involving mitochondria. For example, combination of transcriptomics, proteomics, and metabolomics data led to the identification of the activating transcription factor 4 as a key regulator of mitochondrial stress response in different mammalian cell lines (
). Combination of multiomics datasets with information on multiple cellular pathways allows even broader analysis of functional relationships. An example is the switch between mitophagy and cell death, which represents a complex and so far only poorly understood process (
) described parkin-dependent ubiquitylation events of multiple lysine residues on VDAC1 in a quantitative ubiquitylome analysis. VDAC1 serves as a poly-ubiquitin anchor during mitophagy but can become an essential part of the mitochondrial permeability transition pore in apoptosis. The choice of apoptosis or mitophagy may be driven by parkin-dependent mono- or poly-ubiquitylation on VDAC1 K274 (
). As mentioned above, the decision between mitophagy and apoptosis is also regulated by functional interactions of PINK1 and parkin with antiapoptotic and proapoptotic Bcl-2 family members. Large-scale mass spectrometric determinations of the phosphoproteome (
) help to appreciate the contribution of PTMs in the complex interplay between mitophagy and apoptosis. More advanced technologies will allow for even better integration of multiomics data to ultimately enhance the diagnosis and therapy of mitochondrial dysfunction in human diseases (
Autophagy receptors connecting to-be-degraded cargo to the autophagy machinery are logical regulatory effectors of selective autophagy. The tumor necrosis factor receptor–associated factor family member associated NF-κB activator-binding kinase 1 (TBK1) phosphorylates several autophagy receptor proteins on multiple, autophagy-relevant sites. When optineurin interacts with ubiquitin chains on MOM proteins during mitophagy, the constitutively optineurin-bound TBK1 is recruited to mitochondria and becomes activated (
), thus promoting PINK1/parkin-dependent mitophagy in a feed-forward manner (Fig. 2C). Moreover, TBK1-catalyzed phosphorylation of S177—near the optineurin LIR motif—stabilizes the binding of optineurin to ubiquitylated mitochondria (
). Rather, the LIR motif mediates secondary, ubiquitin-independent recruitment of optineurin and NDP52 to nascent autophagosomes, amplifying ULK1-regulated mitophagy. NDP52 tethering to peroxysomes and mitochondria, respectively, is sufficient to promote the recruitment of the ULK1/FIP200 complex facilitated by TBK1 (
Selective removal of mitochondria—mitophagy—is essential for cellular quality control during stress and disease as well as in development. Thus, impaired mitophagy situations underlie the pathogenesis of a number of chronic diseases such as cancer, cardiovascular diseases, and neurodegenerative diseases, with a particular link to PD via the recessive gene products PINK1 and parkin. As outlined in this review, there are multiple pathways capable of promoting mitochondria elimination. While they all differ on their activation and action mode, all of them depend on accurate PTM regulations for their proper progression and functionality. Because it is essential to understand the molecular mechanisms that trigger mitophagy in all its possible ways, here we gathered the most important PTMs that influence mitochondria targeting mechanisms as well as autophagy-related processes, culminating with mitochondrial elimination.
Enormous research efforts in the past two decades identified the core elements of autophagy (Fig. 1), stress-related PINK1/parkin-dependent mitophagy (Fig. 2), receptor-mediated mitophagy during development (Fig. 3), and the involvement of fission/fusion factors (Fig. 4) and Bcl-2 family members (Fig. 5). Now is an exciting time to build on this knowledge to unravel the regulatory mechanisms that govern the different mitophagy pathways. Recent advances and future improvements in large-scale multiomic analyses will help in this endeavor.
The AMPK-ULK1 axis for starvation-induced autophagy is well established. However, much of the PINK1/parkin mitophagy knowledge is derived from cancer cell lines treated with bulk uncoupling agents. Due to the Warburg effect, though, cancer cells treated with uncoupling agents do not experience the same degree of energy crisis as cells subjected to nutrient starvation. The autophagy machinery is clearly activated during PINK1/parkin-mediated autophagy, but are the upstream mechanisms exactly the same as in nutrient starvation models? The recent discovery of a novel phosphorylation site in ULK1 regulating the alternative, ATG5-independent autophagy (
) is a precedent showing that even long-known players can exert novel modes for distinct autophagy pathways. Moreover, abrupt complete breakdown of ΔΨm is obviously an idealized experimental condition with questionable relevance particularly to postmitotic neurons (
). The recent development of fluorescent reporter systems confirms the occurrence of mitophagy in neurons, but further research with more physiological systems and in vivo aging models is needed to clarify which mitophagy regulatory elements are global and which are more specific in distinct stressed organs.
A key question is how the various ubiquitin ligases establish the ubiquitin code on to-be-degraded mitochondria, and how the autophagy machinery deciphers this code. Some substrates appear straightforward, such as the proteasome-targeting K48-linked poly-ubiquitylation of MFNs. The differential dynamics of TOM ubiquitylations suggest a more regulatory impact of the mitochondrial protein import machinery for the insertion of PINK1 into the MOM and the execution of mitophagy. The complex ubiquitin linkages that occur throughout the time course of mitophagy (
) are instrumental in this regard. Another issue is the experimental modeling of ubiquitylated proteins. While protein serine/threonine phosphorylations can be mimicked faithfully in many cases by site-directed mutageneis to aspartate/glutamate and substitution with glutamine resembles acetyl-lysine to a certain degree, protein ubiquitin modifications cannot be simply mimicked by site-directed mutagenesis. Nevertheless, expanding the genetic code with amber suppression at a desired residue combined with sortase-mediated transpeptidation may provide a first step toward introducing specific ubiquitylated protein species into cells and study their functions (
Another issue to be fully resolved is the sequential dynamics of PTMs on the MOM and accessory factors. While the initiation phase is reasonably well understood (e.g., PINK1 phosphorylation of ubiquitin S65 and the activation of parkin, proteasome-targeting MFN ubiquitylations), it is less clear if there are later checkpoints. Are there more convergence points of PINK1/parkin labeled mitochondria with components of the autophagy machinery (for example, the ULK1 complex, ER–mitochondria contacts, phagophores, etc.), and how would those comprise checkpoints regulating mitophagy throughout the entire time course? Are priming phosphorylations and/or dephosphorylations required for subsequent ubiquitin modifications and substrate selection of the E3 ligase(s)? Would these be exclusively mediated by the currently known ubiquitin kinase PINK1 and phosphatases, respectively? Is the cascade of ubiquitin ligases MUL1-MITOL-parkin established or would there be more E3 ligases involved, also in feedback with distinct autophagy regulators? Does this have an impact on the types of (phospho)ubiquitin chains built on MOM substrates and the writing of a putative mitophagy ubiquitin code? And finally, are the regulatory PTMs restricted to the interface of the MOM with autophagy membranes? The nuclear USP36 is a strong regulator of mitophagy controlling the expression of ATG14L (
), affecting epigenetic and transcriptional regulators of stress-induced mitophagy?
Are protein phosphorylations, acetylations, and ubiquitin modifications the only PTMs regulating mitophagy? Recent evidence points to glycosylations. N-glycanase 1 (NGLY1) deficiency is a rare congenital disorder leading to global developmental delay and a multisystem syndrome including neuropathy. Interestingly, NGLY1 knockdown significantly impaired mitophagy in HeLa cells stably expressing parkin (
). While NGLY1 can regulate transcription of proteasomal genes via deglycosylation of the nuclear factor erythroid 2-like 1, a very recent study identified AMPKα as a novel NGLY1 target integrating energy sensing and mitochondrial homeostasis (
). It will be interesting to further study glycosylations in the regulation of autophagic turnover of mitochondria.
More questions arise when thinking of the cargo-selective specificity of (1) alternative macroautophagy in developing and stressed cells in general or (2) autophagy adaptors in particular, which may or not be driven by PTM modulations. More specifically, it still remains to be elucidated (1) which implications have sirtuin actions on mitochondrial metabolism and cargo-selective autophagy and (2) how do AMBRA1 and its associated factors accumulate on mitochondria for cargo-selective autophagy removal. Another interesting point to be addressed is how Bcl-2 like proteins may regulate autophagy adaptors (i.e., AMBRA1) or other E3-ligases (i.e., HUWE1, MITOL, MUL1). Increasingly sophisticated multiomic screens and future developments in biotechnological methodology combined with hypothesis-driven functional validations of PTM targets all along the mitophagy pathways will undoubtedly advance the comprehensive understanding of this vital process.
Conflict of interest
The authors declare that they have no conflict of interest with the contents of this article.
This work was supported by the German Research Foundation (DFG) Research Training Group GRK2364 “MOMbrane - multifaceted functions and dynamics of the mitochondrial outer membrane,” the German Center for Neurodegenerative Diseases (DZNE) within the Helmholtz Association, and the Hertie Foundation. We are grateful for funding from the Israel Science Foundation (Grant # 215/19 ), the Sagol Longevity Foundation , Joint NRF - ISF Research Fund (Grant # 3221/19 ), and the Yeda-Sela Center for Basic Research .
P. K. conceptualization; B. M. and P. K. funding acquisition; A. L. T., K. I. Z., B. M., M. F., Z. E., and P. K. writing–original draft; A. L. T. and P. K. writing–review and editing.
Funding and additional information
Z. E. is the incumbent of the Harold Korda Chair of Biology.