Cell Biology
Mitophagy
Selective autophagy of damaged mitochondria — PINK1/Parkin, ubiquitination, LC3 recruitment
Mitophagy is the selective autophagy of damaged mitochondria — the quality-control program a cell uses to find a broken organelle, wrap it in a membrane, and digest it in the lysosome before it poisons the neighborhood. Selectivity comes from a molecular tag: in the canonical pathway the kinase PINK1 senses a mitochondrion that has lost its membrane potential, accumulates on its surface, and switches on the E3 ubiquitin ligase Parkin, which coats the organelle in ubiquitin chains that recruit LC3 and the engulfing autophagosome. PINK1 (PARK6) and Parkin (PARK2) were first mapped as recessive early-onset Parkinson's disease genes in 2004 and 1998 respectively; the mechanistic pathway was assembled largely by Richard Youle's lab from 2008 onward. Because neurons rarely divide, mitophagy is much of how they keep their mitochondrial population healthy across a human lifetime.
- SensorPINK1 kinase on damaged mito
- ExecutorParkin E3 ubiquitin ligase
- Switchphospho-ubiquitin at Ser65
- Diseaseearly-onset Parkinson's
- Parkin genePARK2, mapped 1998
- ReceptorsOPTN, NDP52, NIX, FUNDC1
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Why mitophagy matters
- It is the cell's mitochondrial recall program. Mitochondria leak reactive oxygen species, accumulate mutations in their own circular DNA, and can trigger cell death by releasing cytochrome c. A cell cannot fix a badly damaged mitochondrion in place, so it removes it whole. Mitophagy is that removal step — the sanitation service that keeps the energy plant from becoming a liability.
- Two of its core genes cause Parkinson's disease. Loss-of-function mutations in PARK2 (Parkin) and PARK6 (PINK1) are among the most common causes of autosomal-recessive early-onset Parkinson's, frequently before age 40. When damaged mitochondria are not cleared, the high-energy dopaminergic neurons of the substantia nigra are the first to fail.
- It runs on a schedule, not just on damage. Maturing red blood cells destroy every mitochondrion they own using the receptor NIX; NIX-knockout mice keep mitochondria in their erythrocytes and become anemic. Cardiac maturation, immune-cell reprogramming, and elimination of paternal mitochondria after fertilization all use programmed mitophagy.
- Baseline turnover is constant. Reporter mice (mt-Keima, mito-QC) revealed that heart, brain, and kidney continuously flux mitochondria through the lysosome even when nothing is obviously wrong, with organelle half-lives on the order of days to a few weeks — so mitophagy is a housekeeping tempo, not only an emergency.
- It is a drug target. Because boosting PINK1 or Parkin activity might slow neurodegeneration, companies are developing small-molecule PINK1 activators and Parkin activators, and a natural compound, urolithin A (from gut metabolism of pomegranate ellagitannins), induces mitophagy and has reached human muscle-function trials.
- It intersects with aging and metabolism. Declining mitophagy is a candidate driver of sarcopenia, insulin resistance, and the accumulation of dysfunctional mitochondria in aged tissue; exercise and caloric restriction both upregulate mitophagy, part of why they preserve mitochondrial quality.
How mitophagy works, step by step
The canonical route is the PINK1/Parkin pathway, and its logic is a switch that reads mitochondrial health. In a healthy organelle, the kinase PINK1 is imported across the outer membrane through the TOM complex and into the inner membrane through TIM23 — a step that depends on the inner-membrane potential of roughly −150 to −180 mV. Inside, PINK1 is cleaved by the mitochondrial processing peptidase (MPP) and the rhomboid protease PARL, retro-translocated to the cytosol, and destroyed by the proteasome under the N-end rule. The result is that healthy mitochondria keep PINK1 vanishingly low. This constant import-and-destroy cycle is the sensor.
When a mitochondrion loses membrane potential — from oxidative damage, an mtDNA mutation, or a protonophore like the drug CCCP used in the lab — TIM23 import stalls. PINK1 can no longer reach the protease that would cleave it, so full-length PINK1 accumulates on the outer membrane, clamped onto the TOM translocase. Levels rise within minutes. PINK1 then autophosphorylates, dimerizes, and phosphorylates ubiquitin at serine 65, and it phosphorylates the ubiquitin-like (Ubl) domain of Parkin at the equivalent Ser65.
Parkin, a RING-between-RING (RBR) E3 ligase, arrives from the cytosol in an autoinhibited closed conformation. Phospho-Ser65-ubiquitin binds Parkin as an allosteric activator, and phosphorylation of Parkin's own Ubl domain releases the catalytic RING2, opening the enzyme. Activated Parkin conjugates ubiquitin onto outer-membrane substrates — the fusion GTPases MFN1 and MFN2, the trafficking adaptor MIRO1, the channel VDAC, and dozens more. Crucially, PINK1 phosphorylates each newly added ubiquitin, and that phospho-ubiquitin recruits and activates still more Parkin. This feed-forward loop is why a sick mitochondrion goes from a bare surface to a dense phospho-ubiquitin coat in tens of minutes: the reaction bootstraps itself. Ubiquitinating MFN1/2 also blocks fusion, quarantining the damaged organelle so it cannot re-merge with the healthy network.
The phospho-ubiquitin chains — enriched in Lys63 and Lys6 linkages — are the "eat-me" mark. Autophagy receptors read it: optineurin (OPTN) and NDP52 (CALCOCO2) are the dominant ones, with TAX1BP1 and p62/SQSTM1 contributing. Each receptor grips ubiquitin with one domain and carries an LC3-interacting region (LIR) — a short W/F-x-x-L/I motif — that docks into a hydrophobic pocket on LC3/GABARAP proteins. OPTN and NDP52 also recruit the kinases ULK1 and TBK1 and the phagophore-nucleation machinery (WIPI2, DFCP1, ATG proteins), building a double-membrane autophagosome de novo around the tagged organelle. The completed autophagosome — now a mitophagosome — fuses with a lysosome, whose acidic hydrolases digest the mitochondrion back into amino acids, lipids, and nucleotides for reuse.
Alongside this ubiquitin-dependent route runs receptor-mediated mitophagy, which skips Parkin entirely. NIX/BNIP3L, BNIP3, and FUNDC1 sit in the outer membrane with their own LIR motifs and bind LC3 directly; FUNDC1 is switched on by hypoxia through dephosphorylation. A third route is lipid-mediated: cardiolipin, normally an inner-membrane lipid, externalizes after damage and is bound directly by LC3 in neurons. These pathways overlap and back each other up.
Common misconceptions
- "Mitophagy is just autophagy of mitochondria, nothing special." The downstream machinery is shared, but the defining feature is cargo selectivity. Without the ubiquitin tag (PINK1/Parkin) or a receptor's LIR motif (NIX, FUNDC1), the phagophore would engulf random cytosol. That recognition layer is what turns bulk recycling into targeted quality control.
- "PINK1 detects damage directly." PINK1 does not sense reactive oxygen species or damage as such — it senses failed import. Anything that collapses the inner-membrane potential, real damage or an experimental protonophore, stabilizes PINK1. That is why lab work often uses CCCP or valinomycin to trigger the pathway on demand.
- "Parkin is the enzyme, so it does the sensing." Parkin is the executor, not the sensor. It arrives closed and inactive; PINK1's phospho-ubiquitin signal is what opens it. In fact many cell types express little endogenous Parkin, and a lot of foundational data comes from Parkin-overexpressing lines — a caveat when generalizing to neurons.
- "Mitophagy only clears broken mitochondria." Programmed, damage-independent mitophagy remodels whole cell types on schedule — erasing every mitochondrion from maturing red cells via NIX, pruning networks during cardiac and immune-cell maturation, and clearing paternal mitochondria after fertilization.
- "LC3 binds the mitochondrion directly in the main pathway." In PINK1/Parkin mitophagy, LC3 is recruited indirectly through receptors (OPTN, NDP52) that bridge the ubiquitin coat to lipidated LC3. Direct LC3 engagement is the hallmark of the receptor pathways (NIX, BNIP3, FUNDC1) and of cardiolipin-mediated mitophagy, not the ubiquitin one.
- "Knock out PINK1 or Parkin and mice get Parkinson's." They largely do not — the knockouts show mild phenotypes, unmasked mostly under stress or with additional mutations. Redundant pathways (MUL1, ARIH1, NIX, FUNDC1, cardiolipin) compensate, which is a major reason modeling human mitophagy disease in mice has been so difficult.
Mitophagy vs bulk autophagy
| Feature | Mitophagy (selective) | Bulk macroautophagy |
|---|---|---|
| Cargo | A specific mitochondrion (usually damaged) | Random cytoplasm — proteins, ribosomes, organelles |
| Primary trigger | Loss of membrane potential; hypoxia; developmental cue | Nutrient/amino-acid starvation (mTORC1 off) |
| Recognition signal | Phospho-ubiquitin coat or receptor LIR motif | None — non-selective sequestration |
| Key sensor/tag | PINK1 + Parkin, or NIX/BNIP3/FUNDC1 | ULK1/mTOR nutrient sensing |
| Adaptors | OPTN, NDP52, TAX1BP1, p62 | Not required |
| Shared machinery | ATG proteins, LC3 lipidation, lysosomal fusion | ATG proteins, LC3 lipidation, lysosomal fusion |
| Main physiologic role | Organelle quality control, developmental remodeling | Nutrient recycling, general housekeeping |
| Disease link | Parkinson's (PINK1/Parkin) | Broad — cancer, neurodegeneration, infection |
Ubiquitin-dependent vs receptor-mediated mitophagy
| Property | PINK1/Parkin (ubiquitin-dependent) | Receptor-mediated (NIX / BNIP3 / FUNDC1) |
|---|---|---|
| Trigger | Depolarization, mtDNA damage, proteotoxic stress | Hypoxia, developmental/metabolic programs |
| Ubiquitin required | Yes — phospho-Ser65 chains built by Parkin | No — receptor carries its own LIR |
| Sensor | PINK1 kinase (reads failed import) | Receptor expression/phosphorylation state |
| LC3 engagement | Indirect, via OPTN/NDP52 adaptors | Direct, via receptor LIR motif |
| Amplification | Feed-forward phospho-ubiquitin loop | Graded with receptor level/dephosphorylation |
| Signature setting | Acute organelle damage, neurons | Reticulocyte maturation (NIX), low O₂ (FUNDC1) |
| Knockout phenotype | Mild in mice; Parkinson's in humans | NIX⁻/⁻: anemia, retained mito in red cells |
Famous experiments and history
- The genes come first (1998, 2004). Positional cloning of the PARK2 locus by Kitada, Shimizu, and colleagues in 1998 identified Parkin as the gene mutated in autosomal-recessive juvenile parkinsonism. In 2004, Valente and colleagues mapped PARK6 to PINK1, a mitochondrial kinase, in families with early-onset Parkinson's — placing mitochondria at the center of the disease before anyone knew the pathway.
- The fly puts them in one pathway (2006). Three landmark Drosophila papers (Clark, Park, and colleagues; Greene and Pallanck lab) showed PINK1 and Parkin mutants share a phenotype — swollen mitochondria, degenerating flight muscle and dopaminergic neurons — and that Parkin overexpression rescues PINK1 mutants but not vice versa. This ordered the genes: PINK1 upstream, Parkin downstream.
- Parkin translocates to damaged mitochondria (2008). Derek Narendra and Richard Youle showed with live imaging that GFP-Parkin, diffuse in the cytosol, is recruited specifically onto mitochondria depolarized by CCCP within minutes, and then drives their autophagic elimination. This single experiment turned the genetics into a visible, mechanistic pathway and launched the modern field.
- Phospho-ubiquitin is discovered (2014). Independent groups (Koyano and Matsuda; Kane; Kazlauskaite) showed PINK1 phosphorylates ubiquitin at Ser65, and that phospho-ubiquitin is the true activator of Parkin — solving how a mitochondrial kinase switches on a cytosolic ligase and explaining the feed-forward amplification.
- Receptors that read the tag (2015). Michael Lazarou, in the Youle lab, used receptor-pentuple-knockout cells to show that OPTN and NDP52 are the essential autophagy receptors for PINK1/Parkin mitophagy, and that they can nucleate autophagosomes on the cargo even in cells lacking the usual initiators — redefining where the phagophore is born.
- Reporters reveal baseline flux (2016–2017). The mt-Keima mouse (Sun, Finkel) and the mito-QC reporter (McWilliams, Ganley) let researchers quantify mitophagy in living tissue for the first time, revealing high constitutive turnover in heart, brain, and kidney — and, unexpectedly, that much of it proceeds independently of Parkin.
Frequently asked questions
How is mitophagy different from general autophagy?
General macroautophagy is bulk, non-selective self-digestion: the cell wraps a random patch of cytoplasm — soluble proteins, ribosomes, whatever is nearby — inside a double-membrane autophagosome and delivers it to the lysosome, most often as a starvation response to recycle nutrients. Mitophagy is a selective form of autophagy that targets one cargo: a mitochondrion, and usually a damaged one. Selectivity comes from a molecular tag. Either the organelle is coated in ubiquitin chains (in the PINK1/Parkin pathway) that autophagy receptors read, or it displays outer-membrane receptor proteins such as NIX/BNIP3L, BNIP3, or FUNDC1 that bind LC3 directly through a short LIR motif. That tag is what recruits the growing phagophore to the mitochondrion specifically rather than to bulk cytosol. So mitophagy shares the downstream machinery of autophagy — ATG proteins, LC3 lipidation, lysosomal fusion — but adds a cargo-recognition layer that makes it a quality-control tool rather than a nutrient-recycling one.
How does the PINK1/Parkin pathway sense a damaged mitochondrion?
The sensor is the kinase PINK1, and the signal is membrane potential. In a healthy mitochondrion, PINK1 is imported across both membranes through the TOM and TIM23 translocases, cleaved inside by the proteases MPP and PARL, and exported back to the cytosol to be destroyed by the proteasome via the N-end rule — so PINK1 never accumulates. When a mitochondrion loses its inner-membrane potential (the roughly -150 to -180 mV that TIM23 import depends on), PINK1 can no longer be imported and cleaved. It arrests on the outer membrane, clamped onto the TOM complex, and its levels rise within minutes. There it autophosphorylates, dimerizes, and phosphorylates both ubiquitin and the ubiquitin-like domain of Parkin at serine 65. Because PINK1 stabilization is a direct readout of failed import, only mitochondria that have actually lost function build up the signal — that is the switch that distinguishes a sick organelle from a healthy neighbor sitting a micron away.
What role does ubiquitin phosphorylation play in mitophagy?
Phospho-ubiquitin is the amplifier at the heart of the pathway. Parkin arrives from the cytosol in an autoinhibited, closed conformation with almost no ligase activity. Two events open it: PINK1 phosphorylates Parkin's own ubiquitin-like domain at serine 65, and PINK1 phosphorylates free ubiquitin at the equivalent serine 65. Phospho-ubiquitin binds Parkin with high affinity and acts as an allosteric activator, releasing the catalytic RING2 domain. Activated Parkin then conjugates more ubiquitin onto outer-membrane proteins such as MFN1, MFN2, MIRO1, and VDAC. PINK1 phosphorylates each newly added ubiquitin, which recruits and activates still more Parkin — a feed-forward loop that coats the whole organelle in phospho-ubiquitin chains within tens of minutes. This is why a single depolarized mitochondrion can accumulate a dense ubiquitin signal fast: the reaction bootstraps itself. The phospho-ubiquitin chains, especially Lys63- and Lys6-linked, are the actual 'eat-me' mark that autophagy receptors read.
How does LC3 get recruited to a mitochondrion marked for destruction?
LC3 is recruited indirectly, through autophagy receptors that bridge the ubiquitin coat to the forming autophagosome membrane. Once Parkin has decorated the mitochondrion with phospho-ubiquitin chains, cargo receptors — chiefly optineurin (OPTN) and NDP52 (CALCOCO2), with contributions from TAX1BP1 and p62/SQSTM1 — bind those chains through a ubiquitin-binding domain. Each receptor also carries an LC3-interacting region (LIR), a short W/F-x-x-L/I motif that docks into a hydrophobic pocket on LC3 and GABARAP proteins after they are lipidated onto the phagophore. Genetic dissection by the Youle lab showed OPTN and NDP52 are the dominant receptors for PINK1/Parkin mitophagy, and that they act partly upstream of LC3 by recruiting the ULK1 and TBK1 kinases and the WIPI/DFCP1 machinery to nucleate the membrane de novo around the cargo. In receptor-mediated mitophagy — NIX, BNIP3, FUNDC1 — the LIR sits directly in the outer-membrane protein, so LC3 is engaged without any ubiquitin step at all.
Why is mitophagy linked to Parkinson's disease?
Because the two central mitophagy genes are Parkinson's disease genes. Loss-of-function mutations in PARK2 (Parkin, mapped in 1998) and PARK6 (PINK1, mapped in 2004) cause autosomal-recessive early-onset Parkinson's, often striking before age 40. When PINK1 or Parkin is broken, damaged mitochondria are not efficiently cleared; dysfunctional organelles accumulate, reactive oxygen species and mutated mitochondrial DNA build up, and the substantia nigra dopaminergic neurons — cells with enormous, poorly myelinated axons and huge energy demands — are especially vulnerable to that mitochondrial debt. Fruit-fly work by the Guo and Chung labs in 2006 showed PINK1 and Parkin act in the same pathway with PINK1 upstream, and that Parkin overexpression rescues the PINK1 mutant flight-muscle and dopaminergic-neuron degeneration. The pathway also intersects with LRRK2 and alpha-synuclein biology, making impaired mitochondrial quality control a unifying theme across both familial and sporadic Parkinson's — and a target for drugs that would boost PINK1 or Parkin activity.
Does the body use mitophagy outside of clearing damage?
Yes — programmed, developmental mitophagy remodels whole cell types on schedule, independent of any damage signal. The clearest example is the red blood cell. As reticulocytes mature into erythrocytes they must destroy every mitochondrion, and they do it with the receptor NIX (BNIP3L): NIX-knockout mice retain mitochondria in their red cells and develop anemia with reduced red-cell lifespan. Programmed mitophagy also strips paternal mitochondria from the fertilized embryo in some species, helping enforce maternal inheritance of mitochondrial DNA, and it prunes the mitochondrial network during the metabolic reprogramming of activated immune cells, cardiac maturation, and retinal-ganglion development. The mito-QC and mt-Keima reporter mice built in the 2010s revealed a surprisingly high baseline: many tissues, including heart, brain, and kidney, turn over mitochondria constitutively even when nothing is obviously wrong, with organelle half-lives measured in days to a few weeks.
Is the PINK1/Parkin pathway the only way to do mitophagy?
No. PINK1/Parkin is the best-characterized, ubiquitin-dependent route, but it is one of several. Receptor-mediated mitophagy skips ubiquitin entirely: NIX/BNIP3L and BNIP3 on the outer membrane, and FUNDC1 (which is regulated by hypoxia and dephosphorylation), carry their own LIR motifs and bind LC3 directly. Lipid-mediated mitophagy uses cardiolipin, normally an inner-membrane lipid, which externalizes to the outer surface after damage and is bound directly by LC3 in neurons. There are also Parkin-independent ubiquitin routes driven by other E3 ligases such as MUL1, ARIH1, and Gp78, and ubiquitin-independent programs during specific developmental transitions. The pathways overlap and back each other up, which is one reason PINK1 or Parkin knockout mice show milder phenotypes than the human disease — redundancy masks the loss until the animal is stressed. Which route dominates depends on cell type, the nature of the insult, and oxygen tension.